Human Myt1 Is a Cell Cycle-regulated Kinase That Inhibits Cdc2 but Not Cdk2 Activity*

Activation of the Cdc2·cyclin B kinase is a pivotal step of mitotic initiation. This step is mediated principally by the dephosphorylation of residues threonine 14 (Thr14) and tyrosine 15 (Tyr15) on the Cdc2 catalytic subunit. In several organisms homologs of the Wee1 kinase have been shown to be the major activity responsible for phosphorylating the Tyr15 inhibitory site. A membrane-bound kinase capable of phosphorylating residue Thr14, the Myt1 kinase, has been identified in the frog Xenopus laevis and more recently in human. In this study, we have examined the substrate specificity and cell cycle regulation of the human Myt1 kinase. We find that human Myt1 phosphorylates and inactivates Cdc2-containing cyclin complexes but not complexes containing Cdk2 or Cdk4. Analysis of endogenous Myt1 demonstrates that it remains membrane-bound throughout the cell cycle, but its kinase activity decreased during M phase arrest, when Myt1 became hyperphosphorylated. Further, Cdc2·cyclin B1 was capable of phosphorylating Myt1 in vitro, but this phosphorylation did not affect Myt1 kinase activity. These findings suggest that human Myt1 is negatively regulated by an M phase-activated kinase and that Myt1 inhibits mitosis due to its specificity for Cdc2·cyclin complexes.

The cyclin-dependent kinases (Cdks) 1 are a family of highly conserved serine/threonine kinases that mediate many of the cell cycle transitions that occur during duplication. Each of these Cdk catalytic subunits associates with a specific subset of regulatory subunits, termed cyclins, to produce a distinct Cdk⅐cyclin kinase complex that, in general, functions to execute a specific cell cycle event (1). The placement of a specific Cdk⅐cyclin complex function with a particular cell cycle transition has led to the identification of key downstream substrates as well as upstream regulatory mechanisms. Well characterized Cdk⅐cyclin-regulated cell cycle transitions that have been identified thus far include Cdk4⅐cyclin D and Cdk6⅐cyclin D complexes with the pRB-gated G 1 /S transition, Cdk2⅐cyclin E and Cdk2⅐cyclin A complexes with the initiation and progression of DNA replication and Cdc2⅐cyclin A and Cdc2⅐cyclin B complexes with the initiation of mitosis (reviewed in Refs. [2][3][4]. Activation of the Cdk⅐cyclin kinases during these transitions is controlled by a variety of regulatory mechanisms. For the Cdc2⅐cyclin B complex, inhibition of kinase activity during S phase and G 2 is accomplished by phosphorylation of Cdc2 residues Thr 14 and Tyr 15 , which are positioned within the ATPbinding cleft (5)(6)(7). Phosphorylation of Thr 14 and/or Tyr 15 is believed to suppress catalytic activity by disrupting the orientation of the ATP molecule present within this cleft (8,9). The abrupt dephosphorylation of these residues by the Cdc25 phosphatase results in the rapid activation of Cdc2⅐cyclin B kinase activity and downstream mitotic events (10). While phosphorylation/dephosphorylation of the conserved Tyr 15 site in Cdk2 likely plays an important role in regulating the G 1 /S transition, the role that Cdk2 Thr 14 phosphorylation plays is less clear (11,12). Phosphorylation of the corresponding inhibitory tyrosine residue in Cdk4 has also been observed (13,14). It has been proposed that Thr 14 /Tyr 15 phosphorylation functions to permit a cell to attain a critical concentration of inactive Cdk⅐cyclin complexes, which, upon activation, induces a rapid and complete cell cycle transition (15). There is evidence in mammalian cells that Thr 14 /Tyr 15 phosphorylation also functions, in part, to delay Cdk activation after DNA damage (7, 13, 16 -18).
The Schizosaccharomyces pombe wee1 gene product was the first kinase identified that is capable of phosphorylating Tyr 15 in Cdc2 (19). Homologs of the Wee1 kinase have been subsequently identified and biochemically characterized from a wide range of species including human, mouse, frog, Saccharomyces cerevisiae, and Drosophila (20 -26). In vertebrate systems, where Thr 14 in Cdc2 is also phosphorylated, the Wee1 kinase was capable of phosphorylating Cdc2 on Tyr 15 , but not Thr 14 , indicating that another kinase was responsible for Thr 14 phosphorylation (21,27). Direct evidence for the existence of a kinase activity that phosphorylated Cdc2 on Thr 14 in the membrane fractions of Xenopus egg extracts has been reported (28). The Xenopus gene encoding this membrane-associated kinase, the Myt1 kinase, has been isolated, and its gene product was shown to be capable of phosphorylating Thr 14 and, to a lessor extent, Tyr 15 in Cdc2 (29). A human Myt1 homolog displaying similar properties has been recently identified (30) (see "Results"). An apparently unrelated, non-membrane-associated Thr 14 kinase activity has also been identified in bovine thymus cytosol (31).
While Xenopus Myt1 clearly plays an important role in regulating the rapid early embryonic cell cycles, its role in cells that have distinct G 1 and G 2 phases as well as growth and checkpoint controls is not known. To investigate the role of Cdk Thr 14 phosphorylation in somatic cell cycle control, we have initiated a biochemical analysis of the human Myt1 kinase. The results presented reveal that Myt1, unlike Wee1, exhibits a restricted substrate specificity; it was capable of phosphorylating Thr 14 only on Cdc2⅐cyclin and not Cdk2⅐cyclin complexes. We also found that endogenous Myt1 kinase activity was reduced during a drug-induced M-phase arrest, concomitant with hyperphosphorylation of the Myt1 protein, implying that Myt1 may be negatively regulated by phosphorylation. Human Myt1 Cloning-A 325-base pair DNA fragment, corresponding to an internal region of EST clones 335241 and 336264, was polymerase chain reaction-generated using oligos 5Ј-AGCAGCCTCTCCAG-CAACTGG-3Ј and 5Ј-CAGAGAAGACCATGGAGTTCC-3Ј (5Ј and 3Ј primers, respectively) and the first strand cDNA synthesis of fetal brain RNA (Clonetech) as a template. This DNA fragment was 32 P-labeled by random priming (Pharmacia) and used as a probe to screen approximately 7 ϫ 10 5 clones of a human placental Lambda ZAPII cDNA library (Stratagene). Of nine positives isolated, clone 16 was found to contain the longest cDNA insert. The complete DNA sequence of this 1.98-kilobase pair insert was determined using an ABI sequencer.
Purification and in Vitro Labeling of Cdc2, Myt1, and Wee1 Kinases-The coding regions of Myt1, full-length Wee1 (clone 1E-12 encoding p98/100Wee1 (22), provided by Dr. Tony Hunter, Salk Institute), and cyclins were cloned into the baculovirus transfer vector pAcoG such that the expressed protein contained an N-terminal EE-epitope (32). The monomeric Cdc2 in Fig. 3C contained a C-terminal EE-tag and the cyclin B1 complexed with Cdc2-AF in Fig. 2A contained an N-terminal glutathione S-transferase fusion. All other Cdc2 and Cdk2 subunits used in this study contained a C-terminal HA tag and have been described previously (11,33). The coding region of Wee1⌬, which corresponds to the p49Wee1 truncated protein (20), was polymerase chain reaction-generated using the full-length Wee1 clone 1E-12 as a template. Cdk⅐cyclin complexes were obtained by mixing and incubating individual Cdk and cyclin lysates together prior to purification, except in the case of Cdk4⅐cyclin D1, which was obtained by co-infecting Sf9 cells with Cdk4 and cyclin D1-expressing baculoviruses. Additionally, Sf9 lysates containing Cdk-activating kinase (Cdk7 and cyclin H) (34) had been added to Cdc2-wild type and cyclin B1 lysates during the activation incubation. Baculovirus-expressed proteins were immunoaffinity purified from Sf9 cell lysates using anti-Glu-Glu (EE) antibodycoupled agarose beads (32) or affinity-purified using glutathione-agarose beads (Sigma).
In vitro 32 P-labeling reactions were performed by preincubating 1 l (1-7 g) of purified Cdk⅐cyclin complexes in 4 l of KAB (50 mM Tris, pH 7.5, 10 mM MgCl 2 , 10 mM DTT) containing 25 M ATP for 20 min at 25°C, followed by addition of 0.5 l p49Wee1 (0.78 mg/ml) or Myt1 (2.2 mg/ml) in 4.5 l of KAB containing 5 Ci of [␥-32 P]ATP (3000 Ci/mmol) and incubating at room temperature for 20 min. The reaction was stopped with sample buffer and resolved by 12% SDS-PAGE.
Mammalian Cell Transfection, Fractionation, and Cell Cycle Drug Arrests-Mammalian cell lines CEM, U2-0S, and C-33A used in this study were obtained from the American Type Culture Collection. Various Myt1-containing DNA fragments were cloned into a derivative of mammalian expression vector pcDNA3 (Invitrogen) that contained a 5Ј-HA epitope-coding sequence such that the expressed Myt1 proteins contained an N-terminal HA epitope tag. The expressed Myt1 proteins are designated as follows: Myt1-1b, residues 10 -499; Myt1-1A, residues 75-499; Myt1-K Ϫ , residues 10 -499 containing a Lys 139 to Ala 139 change that was created by site-directed mutagenesis; Myt1-⌬T.M., residues 10 -499 containing a deletion of the transmembrane domain (residues 379 -397) 2 (30). The S100 and P100 fractions were prepared from Myt1-expressing cell lines that had been lysed by Dounce homogenization in an ice-chilled hypotonic buffer (10 mM Hepes, pH 7.5, 5 mM KCl, 1.5 mM MgCl 2 , protease inhibitors). This lysate was ultracentrifuged for 30 min at 40,000 rpm in a TLA45 rotor at 4°C to obtain the S100 supernatant. The membrane-bound proteins in the P100 fractions were solubilized by extracting this fraction in the same buffer containing 1% Nonidet P-40 for 30 min at 4°C. The remaining insoluble material was removed by ultracentrifugation for 30 min at 40,000 rpm in a TLA45 rotor at 4°C. Synchronization of CEM cells by drug treatments was performed by culturing cells for 16 -23 h in the presence of either 5 mM hydroxyurea (Calbiochem), 1 mM mimosine (Sigma), or 50 ng/ml nocodazole (Calbiochem). U2-OS cells were treated with 125 ng/ml nocodazole.
Myt1 Immunoprecipitation and in Vitro Kinase Assays-Mammalian cells were lysed in ELB (50 mM Hepes, pH 7.4, 250 mM NaCl, 0.1% Nonidet-P40, 1 mM DTT) with protease inhibitors and clarified by microcentifugation at 8200 ϫ g for 15 min. Myt1 was immunoprecipitated by incubating 2 l of anti-Myt1 antibody serum in 0.2-0.4 ml of ELB containing 1 mg of lysate for 30 min at 4°C, followed by addition of 25 l of protein A-Sepharose (1:1 slurry, Pierce). After 1 h at 4°C, the beads were pelleted, washed three times with ELB, and once with KAB. To assess kinase activity, 10 l of Cdc2⅐cyclin B1 kinase reaction (1.5 l of Cdc2⅐cyclin B1 Sf9 lysates mix, see above, in 8.5 l of KAB containing 7.5 M ATP) was added to the beads and incubated for 45 min at room temperature with periodic mixing. Additionally, 10 M final concentration of the corresponding peptide was included during kinase assays of C-terminal antibody immunoprecipitates. This reaction was then assayed for histone H1 kinase activity by adding 5 l of a histone H1 kinase mix (50 mM Tris, pH 7.5, 10 mM MgCl 2 , 1 mM DTT, 5 g of histone H1, 1 Ci of [␥-32 P]ATP (3000 Ci/mmol)) and incubating for 15 min at 30°C with agitation. This reaction was stopped with SDS sample buffer, resolved by SDS-PAGE, and analyzed by autoradiography. For quantitation, the labeled bands were excised, and 32 P-incorporation was determined by Cerenkov counting. The mean of individual kinase assays performed on two or three aliquots of the same immunoprecipitation was determined.
Human Myt1 Analysis in Xenopus Egg Extracts-Analysis of Myt1 proteins in Xenopus interphase and mitotic egg extracts was performed essentially as described by Stukenberg et al. (35) except that the Myt1 proteins were produced from pcDNA3-containing Myt1 plasmids in a coupled transcription/translation system (TNT, Promega) in the absence of radiolabel.

RESULTS
Isolation of the Human MYT1 Gene-To identify a human homolog of the Xenopus MYT1 gene, we performed a TBLASTN search of the EST data base using the amino acid sequence encoded by Xenopus MYT1 as the query (29). As expected, the majority of high scoring matches were protein kinases, owing to the extensive conservation of residues within the catalytic domain of protein kinases (36). However, two nearly identical human ESTs (dbEST Id: 336264 and 335241) were found to match significantly amino acid tracts within the C-terminal (noncatalytic domain) 130 residues of the Xenopus Myt1 kinase. A full-length cDNA clone was isolated and found to encode the recently described human Myt1 kinase (30).
Myt1 inactivates Cdc2 but Not Cdk2 Kinase Activity-To analyze human Myt1 activity, we examined its ability to inactivate Cdk⅐cyclin complexes in a cell-free extract system. We used the baculovirus expression system to produce insect cell extracts that individually contained high levels of either human Cdk, cyclin, Wee1, or Myt1 proteins. It has been shown previously that Cdk⅐cyclin kinase activation can be reconstituted upon mixing individual Cdk and cyclin-containing lysates (33); thus, addition of an inhibitory activity would be expected to block this activation. We first assessed the ability of this assay system to detect Wee1 activity. Lysates containing wild-type Cdc2 were mixed with cyclin B1 lysates and with either mock-infected or Wee1-containing lysates. After a brief incubation, an aliquot of this lysate mixture was removed and Cdc2⅐cyclin B1 kinase activity was determined by performing a histone H1 kinase reaction. As shown in Fig. 1A, addition of Wee1 lysates effectively inhibited Cdc2⅐cyclin B1 kinase activation. The Wee1 inhibitory activity was dependent on the presence of the Tyr 15 residue in Cdc2, since both Cdc2-F15 (Tyr 15 replaced with phenylalanine) and Cdc2-AF (Thr 14 and Tyr 15 replaced with alanine and phenylalanine) proteins complexed to cyclin B1 retained full activity in the presence of Wee1 lysates.
To determine whether Myt1 can inactivate Cdk2⅐cyclin complexes, we measured the kinase activity of Cdk2 that had been activated in the presence of Myt1-containing lysates. We used the Cdk2-AF mutant as a reference to control for any non-Thr 14 /Tyr 15 phosphorylation inhibitory activity that may be present in the Myt1 lysate. As shown in Fig. 1C, an equivalent amount of histone H1 kinase activity was present in wild-type Cdk2 and Cdk2-AF lysates that had been mixed with either cyclin A or cyclin E lysates in the presence of Myt1, indicating that neither Cdk2⅐cyclin A nor Cdk2⅐cyclin E complexes were inactivated by Myt1. However, when Cdc2 was substituted for Cdk2 in the cyclin A complex, Myt1 significantly reduced histone H1 kinase activity, as compared with mock-treated or the Cdc2-AF control lysate (Fig. 1D). Thus, the Cdk subunit, not the cyclin subunit, is a key determinant for whether a particular Cdk⅐cyclin complex is recognized by the Myt1 kinase.
Human Myt1 Phosphorylates Cdc2 but Not Cdk2 in Vitro-To analyze Myt1 substrate specificity further, various combinations of affinity purified Myt1, Wee1, and Cdk⅐cyclin kinases were incubated together in an in vitro kinase reaction containing [␥-32 P]ATP followed by SDS-PAGE analysis of the 32 P-labeled proteins. As shown in Fig. 2A, both purified Myt1 and Wee1 were capable of phosphorylating wild-type Cdc2 as well as a kinase-deficient Cdc2 mutant (K Ϫ ), each of which was complexed with cyclin B1. The Cdc2-AF mutant was not 32 Plabeled by Myt1 or Wee1, indicating that these kinases were phosphorylating Cdc2 on residue Thr 14 and/or Tyr 15 . Direct phosphoamino acid analysis and immunoblot analysis using anti-Cdc2 phospho-specific antibodies showed that human Myt1, while having some Tyr 15 activity, preferentially phosphorylated Cdc2 on Thr 14 (data not shown), consistent with previous studies of human and Xenopus Myt1 (29,30).
When Cdk2⅐cyclin complexes were used as substrates in these kinase reactions, we found that Myt1 failed to phosphorylate Cdk2 that was complexed with cyclins A or E, as well as Cdk4 complexed with cyclin D1 (Fig. 2B). In contrast, Wee1 readily phosphorylated Cdk2 complexed with cyclins A or E, but not Cdk4 complexed with cyclin D1, as has been previously demonstrated (22). Thus, using either a crude lysate system or an in vitro kinase reaction with purified components, we consistently observed that Myt1 only phosphorylated and inactivated Cdk⅐cyclin complexes in which Cdc2 was the catalytic subunit.
To assess whether Myt1 phosphorylation of Cdc2 requires cyclin association, we performed a Myt1 kinase reaction using monomeric Cdc2 as substrate. As shown in Fig. 2C, Myt1 did not phosphorylate monomeric Cdc2. However, reconstitution of the Cdc2⅐cyclin B1 complex, by preincubating monomeric Cdc2 with monomeric cyclin B1, enabled Myt1 to phosphorylate Cdc2, indicating that Myt1 can only phosphorylate Cdc2 that is complexed with a cyclin subunit. The truncated Wee1 kinase was capable of 32 P-labeling monomeric Cdc2, but this phosphorylation was enhanced when the cyclin-bound form was reconstituted, which is consistent with previous studies (37).
Human Myt1 Is Hyperphosphorylated during Mitosis-To examine the regulation of endogenous Myt1, rabbit polyclonal antibodies were raised against three different peptides corresponding to Myt1 residues 10 -23, 472-487, and 486 -499. These antibodies are designated N2-1, C1-1, and C2-1, respectively. When used for immunoblotting of total cell lysates, cytosolic fractions, and detergent-solubilized membrane fractions prepared from asynchronously growing C-33A human cells, each of the anti-Myt1 antibodies recognized a 66-kDa membrane-associated protein (Fig. 3). This 66-kDa protein comigrated with baculovirus-expressed human Myt1 as well as Myt1 that was produced in an in vitro transcription/translation-coupled reaction (data not shown).
In Xenopus mitotic extracts, the electrophoretic mobility of Xe-Myt1 is drastically decreased, concomitant with Myt1 hyperphosphorylation and decreased kinase activity (29). To determine whether human Myt1 exhibits similar cell cycle regulation, we examined Myt1 in lysates prepared from unsynchronized human CEM cells as well as cells that had FIG. 1. Human Myt1 kinase activity in a Sf9 insect cell lysate system. As indicated, various combinations of lysates prepared from baculovirus-infected Sf9 cells were incubated together, and the kinase activity of the formed Cdk⅐cyclin complex was subsequently assessed by measuring the incorporation of 32 P into the exogenously added histone H1. The kinase reaction was stopped with SDS sample buffer, resolved by SDS-gel electrophoresis, and analyzed by autoradiography. The region of the autoradiographic film exposed by the 32 P-labeled histone H1 substrate is shown.
been arrested at G 1 , S, and M phases of the cell cycle by treatment with mimosine, hydroxyurea, and nocodazole, respectively. Immunoblot analysis using anti-Myt1 C-terminal peptide antibodies (C1-1 and C2-1) showed that Myt1 shifted to a slower migrating form in mitotic lysates (Fig. 4A). The main contribution to this modification is phosphorylation, since phosphatase treatment of the mitotic lysate increased the migration of Myt1 to nearly that observed in G 1 -and S-phase extracts (Fig. 4B). At least five laddering Myt1 forms were detectable in the phosphatase-treated mitotic lysate, indicating that multiple residues are phosphorylated during mitosis. Further, immunoblot analysis of cytosolic and detergent-extracted membrane fractions prepared from nocodazole-arrested CEM cells revealed that human Myt1 remained membrane-associated during M-phase arrest (Fig. 4C).
Interestingly, anti-Myt1 antibody N2-1 failed to recognize the slower migrating Myt1 protein in mitotic lysates but was capable of detecting Myt1 in lysates prepared from unsynchronized as well as mimosine, or hydroxyurea-arrested CEM cells (Fig. 4A). Phosphatase treatment of the mitotic lysate alleviated this Myt1 masking effect, suggesting that this antibody recognized only unphosphorylated Myt1 (data not shown). Consistent with the immunoblot results, this N-terminal antibody was incapable of immunoprecipitating mitotic Myt1 protein (Fig. 5A).
Since Myt1 likely spans the endoplasmic reticulum membrane 2 (30), it is possible that membrane association is required for Myt1 hyperphosphorylation observed during mitosis. To investigate this question, we established a U2-OS cell line that stably expressed a non-membrane-associated Myt1 mutant, Myt1⌬T.M., which lacks the 19-residue transmembrane domain (see "Experimental Procedures"). This Myt1⌬T.M.-expressing cell line was arrested in mitosis with nocodazole, and the soluble (S100) and detergent-solubilized particulate (P100) fractions were prepared. Immunoblot analysis of these fractions revealed that the S100-containing Myt1⌬T.M. protein exhibited a reduced mobility shift similar to that seen for the endogenous Myt1 protein (Fig. 4D) present in mitotic extracts. This result indicates that a kinase(s) exists in the cytosol that is capable of phosphorylating Myt1 during mitosis, independent of Myt1 association with endoplasmic reticulum membranes. We cannot, however, rule out the possibility that this unidentified kinase is also present in the endoplasmic reticulum lumen.
Human Myt1 Kinase Activity Is Reduced during Mitosis-To characterize the kinase activity of endogenous human Myt1, anti-Myt1 antibody immunoprecipitates from CEM cell extracts were tested for their ability to inactivate various Cdk⅐cyclin complexes. Fig. 5A shows that the C-terminal anti-Myt1 antibodies were capable of immunoprecipitating Myt1 from lysates prepared from nonsynchronized and mitotic cells, whereas, as described above, the N-terminal antibody failed to recognize mitotic Myt1. Consistent with our analysis of recombinant Myt1, anti-Myt1 immunoprecipitates dramatically reduced the histone H1 kinase activity of Cdc2⅐cyclin B1 and FIG. 2. In vitro phosphorylation of Cdk⅐cyclin complexes by Myt1. A, purified Myt1 or Wee1⌬ (p49Wee1) kinases were incubated either alone (lanes 10 and 11) or in combination with purified Cdc2⅐cyclin B1 (lanes 1-3), Cdc2-AF⅐cyclin B1 (lanes 4 -6), or Cdc2(K Ϫ )⅐cyclin B1 (lanes 7-9) complexes in an in vitro kinase reaction containing [␥-32 P]ATP. The reaction was subjected to SDS-PAGE and analyzed by autoradiography. The basis of the low level of 32 P-labeled Cdc2 in the untreated wild-type Cdc2⅐cyclin B1 kinase reaction (lane 1) is unknown, but a similar observation has been reported by Watanabe et al. (22). B, purified Cdk2⅐cyclin A (lanes 1-3), Cdk2⅐cyclin E (lanes 4 -6), and Cdk4⅐cyclin D1 (lanes 7 and 8) complexes were incubated either alone or in combination with purified Myt1 or Wee1⌬ (p49Wee1) in an in vitro kinase reaction and analyzed as in panel A. C, cyclin-dependent phosphorylation of Cdc2 by Myt1. Affinity-purified Cdc2⅐cyclin B1 complexes (lanes 1-3), monomeric Cdc2 (lanes 4 -6), monomeric Cdc2 plus monomeric cyclin B1 (lanes 7-9), or monomeric cyclin B1 (lanes 10 and 11) proteins were incubated either alone or in combination with purified Myt1 or Wee1⌬ (p49Wee1) in an in vitro kinase reaction and analyzed as in panel A. The faint band in lane 11 may represent an endogenous insect Cdk that co-purified with the human cyclin B1 protein.
To determine if the kinase activity of human Myt1 was cell cycle-dependent, we compared the kinase activity of Myt1 immunoprecipitates isolated from nonsynchronized and nocodazole-arrested CEM cells. Using two different C-terminal Myt1 peptide antibodies, Myt1 immunoprecipitated from mitotic extracts had approximately 40 -50% reduced kinase activity (Fig.  6A). Analysis of Myt1 immunoprecipitates using an N-terminal Myt1 peptide antibody showed that Myt1 was active in asynchronous, as well as mimosine-and hydroxyurea-arrested cells (Fig. 6C). As discussed above, this N-terminal peptide antibody failed to immunoprecipitate Myt1 from nocodazole-arrested cells. These results strongly suggest that Myt1 is active during G 1 , S, and G 2 but becomes inactive during M phase, concomitant with Myt1 hyperphosphorylation.
Phosphorylation of Human Myt1 in Xenopus Extracts and by Cdc2⅐cyclin B1-The similarity of cell cycle-dependent kinase activity and phosphorylation state for both human and Xenopus Myt1 suggested that Myt1 kinases may be regulated by a conserved mechanism. As an initial test of this possibility we compared the electrophoretic mobility of human Myt1 that had been incubated in Xenopus interphase and mitotic egg extracts. As shown in Fig. 7A, Xenopus mitotic extracts contained an activity that drastically reduced the electrophoretic mobility of each of the four human Myt1 proteins tested. The reduced mobility observed for Myt1 truncation mutants lacking the N-terminal 57 residues (-1A, lane 4) and C-terminal 206 residues 2 indicates that phosphorylated residues responsible for the Myt1 mobility shift are not contained exclusively within either of these regions.
Since the Xenopus mitotic extracts were derived from interphase egg extracts that had been induced into an M-phase state FIG. 4. Myt1 hyperphosphorylation during mitosis. A, asynchronous CEM cells and CEM cells arrested in G 1 , S, and M phase by respective mimosine, hydroxyurea, and nocodazole treatment were obtained. Cell lysates were prepared, and equal amounts (20 g) were resolved on a 4 -20% SDS-PAGE and immunoblotted with anti-Myt1 antibodies C1-1, N2-1, and C2-1, as indicated. B, lysates from nocodazole-and mimosine-treated cells from panel A were incubated either with or without lambda phosphatase. Equal amounts (20 g) of nocodazole (lane 1), mimosine (lane 4), and lambda phosphatase-treated nocodazole (lanes 2 and 3) lysates were subjected to gel electrophoresis and immunoblotted with anti-Myt1 C1-1 antibody. C, the cytosolic S100 (lanes 1 and 2) and detergent-solubilized P100 fractions (lanes 3 and 4) from asynchronously growing (lanes 1 and 3) and nocodazole-arrested (lanes 2 and 4) CEM cells were prepared, subjected to SDS-PAGE, and immunoblotted with anti-Myt1 antibody, C1-1. D, hyperphosphorylation of the cytosolic Myt1-⌬T.M. protein during mitosis. The S100 and detergent-extracted P100 fractions were prepared from asynchronously or nocodazole-arrested U2-OS cell lines that stably expressed HAtagged Myt1-1b or Myt1-⌬T.M. proteins. These fractions were subjected to SDS-PAGE and blotted with anti-HA monoclonal antibody (top) and anti-Myt1 C2-1 antibody (bottom).  lanes 3 and 7), and C2-2 (lanes 4 and 8). C2-2 is serum from a second rabbit that was immunized with the C2 peptide. The washed immunoprecipitates were resolved by SDS-PAGE and immunoblotted using a mouse anti-Myt1 polyclonal serum as probe. B, lysates were prepared from nonsynchronized CEM cells and subjected to immunoprecipitation with the N-terminal peptide anti-Myt1 antibody, N2-1. These immunoprecipitates were incubated with the indicated Cdc2⅐cyclin B1 complexes, and the resulting histone H1 kinase activity was determined as described under "Experimental Procedures." These data represent the mean of triplicate kinase assays. C, same as in panel B but anti-Myt1 immunoprecipitates were incubated with Cdc2⅐cyclin B1, Cdk2⅐cyclin A1, and Cdk2⅐cyclin E complexes. by addition of exogenous cyclin B1, it is possible that Cdc2⅐cyclin B1 may phosphorylate Myt1 directly, perhaps acting as an autocatalytic feedback mechanism such as has been proposed for Xenopus Cdc25 and Wee1 (24,38). To address this possibility, we first determined whether Cdc2⅐cyclin B1 could affect the electrophoretic mobility of a kinase-defective Myt1 (Myt1-K Ϫ ). Sf9 lysates containing Cdc2⅐cyclin B1 were incubated together with a Myt1-K Ϫ lysate in the presence of an ATP regeneration system. Immunoblot analysis of this reaction showed that the Myt1-K Ϫ protein migrated with a reduced mobility after Cdc2⅐cyclin B1 treatment (Fig. 7B), albeit to a lesser degree than observed for Myt1 in mitotic extracts derived from human CEM cells (data not shown). Interestingly, Fig. 7B also shows that the Myt1 N2-1 antibody exhibited a reduced ability to recognize Myt1 that had been phosphorylated by the Cdc2⅐cyclin B1 kinase. The direct phosphorylation of Myt1 by Cdc2⅐cyclin B1 was confirmed by including [␥-32 P]ATP in an in vitro kinase reaction that contained immunoprecipitated Myt1 and purified Cdc2⅐cyclin B1 (Fig. 7C). Phosphoamino acid analysis showed that Cdc2⅐cyclin B1 phosphorylated Myt1 on both threonine and serine residues (data not shown).
Using an Sf9 lysate system, we examined whether the activity of Myt1 was affected by Cdc2⅐cyclin B1 phosphorylation. A Myt1-containing lysate was first mixed with a Cdc2-AF⅐cyclin B1 lysate, followed by subsequent addition of wild-type Cdc2⅐cyclin B1. The latter Cdc2⅐cyclin B1 complex was isolated by immunoprecipitation and its kinase activity was determined by performing a histone H1 kinase assay. We observed that phosphorylation of Myt1 by Cdc2-AF⅐cyclin B1 had no affect on its ability to inactivate Cdc2 kinase activity (data not shown). Hence, phosphorylation by Cdc2 does not appear to directly block Myt1 catalytic activity. DISCUSSION In this study we report several biochemical properties of the human Myt1 kinase. We found that human Myt1 was capable of phosphorylating and inactivating Cdc2 associated with cyclin A or cyclin B1. In contrast, however, Myt1 was unable to phosphorylate or inactivate Cdk⅐cyclin complexes that function earlier in the cell cycle. In particular, Myt1 did not phosphorylate Cdk4⅐cyclin D1, Cdk2⅐cyclin E, or Cdk2⅐cyclin A complexes. The inability of Myt1 to phosphorylate Cdk4 was not unexpected since Cdk4 contains an alanine residue at the corresponding Thr 14 position. Since Myt1 readily phosphorylated Cdc2⅐cyclin A complexes, Myt1 specificity appears to be determined principally by the Cdk subunit. In vivo, Cdk2 Tyr 15 is phosphorylated to much greater extent than Thr 14 (11,12), consistent with our in vitro results that Myt1 does not phosphorylate Cdk2 on Thr 14 . It is possible, however, that Myt1 can phosphorylate Cdk2 that has been initially phosphorylated on Tyr 15 , as has been previously suggested (11). Alternatively, additional kinases may be responsible for phosphorylating Thr 14 in Cdk2, such as a non-membrane-associated Thr 14 kinase activity that is present in bovine thymus cytosol (31). In either case, the significantly greater extent of Tyr 15 phosphorylation on endogenous Cdk2 suggests that Wee1, or another Wee1-like kinase, is the major kinase activity that suppresses Cdk2⅐cyclin kinase activity during the G 1 /S transition.
The kinase activity of human Myt1 was decreased during M-phase arrest. This decreased activity correlated with hyperphosphorylated and electrophoretically slower forms of Myt1. Indeed, phosphorylation of the high number of TP and SP amino acid doublets in both human and Xenopus Myt1, 10 and 12, respectively, could account for the dramatic Myt1 hyperphosphorylation observed during M phase. Many of these TP/SP doublets conform to the Cdc2⅐cyclin B1 phosphorylation FIG. 6. Human Myt1 kinase activity during the cell cycle. A, lysates from nonsynchronized and nocodazole-arrested CEM cells were prepared and subjected to immunoprecipitation with anti-Myt1 antibodies C1-1 and C2-1. The washed anti-Myt1 immunoprecipitates were assayed for Myt1 kinase activity by incubating with Cdc2⅐cyclin B1 complexes and then measuring histone H1 kinase activity. These results represent the mean of duplicate kinase assays. B, a portion of the anti-Myt1 immunoprecipitates from panel A was subjected to immunoblotting using a mouse anti-Myt1 serum as probe. C, kinase assays of Myt1 immunoprecipitated from lysates prepared from nonsynchronized and mimosine-, hydroxyurea-, and nocodazole-arrested CEM cells. Myt1 was immunoprecipitated using the N-terminal anti-Myt1 antibody, N2-1, and assayed for kinase activity as described Fig. 5B. These results represent the mean of duplicate kinase assays. D, a portion of the anti-Myt1 immunoprecipitates from panel C was subjected to immunoblotting using a mouse anti-Myt1 polyclonal serum as probe. consensus site (39), consistent with our observation that Cdc2⅐cyclin B1 readily phosphorylated Myt1 in vitro. Our finding that an anti-Myt1 N-terminal peptide antibody recognized interphase but not M-phase Myt1, or Myt1 phosphorylated by Cdc2⅐cyclin B1 in vitro, strongly suggests that a residue near the N terminus is phosphorylated during mitosis.
Which kinase(s) phosphorylates Myt1 during mitosis? Our in vitro data demonstrate that Cdc2⅐cyclin B1 can phosphorylate Myt1 on sites that reduce its mobility on SDS-polyacrylamide gels. While this phosphorylation did not affect Myt1 kinase activity, we cannot rule out the possibility that, in vivo, Cdc2 phosphorylation of Myt1 may be a requisite for further modifications that directly block Myt1 catalytic activity. Additionally, the partial reduction in Myt1 gel mobility induced by Cdc2⅐cyclin B1 phosphorylation suggests that an additional kinase(s) may phosphorylate Myt1. Further experiments will be required to discern the mechanism of Myt1 regulation during the cell cycle.