The C-terminal domain of the Cdc2 inhibitory kinase Myt1 interacts with Cdc2 complexes and is required for inhibition of G₂/M progression

In all eukaryotic cells, progression through the cell cycle is regulated by the sequential activation of the cyclin-dependent protein kinases (CDKs) (Norbury and Nurse, 1992; Morgan, 1997). Activation of the p34^cdc2 (Cdc2) CDK is the universal event controlling the onset of mitosis. The ability of Cdc2 to induce M phase is dependent on its association with a cyclin partner, yet is also further regulated in a positive and negative fashion by phosphorylation. Specifically, phosphorylation at Thr161 by the CDK activating kinase (CAK) is necessary for activation of the CDK/cyclin complex, whereas phosphorylation of Thr14 and Tyr15 by Wee1 and Myt1 kinases maintains the complex in an inactive state. Indeed, in higher eukaryotes there is a gradual cytoplasmic accumulation of potentially active cyclin B/Cdc2 heterodimers during S and G₂, that are phosphorylated at Thr161 but maintained in the inactive state by phosphorylation at Thr14 and Tyr15 (Norbury et al., 1991). Dephosphorylation of Thr14 and Tyr15 upon activation of the dual-specificity phosphatase, Cdc25, coupled with inactivation of the Thr14 and Tyr15 kinases (McGowan and Russell, 1995; Parker et al., 1995; Watanabe et al., 1995; Mueller et al., 1995; Liu et al., 1997; Booher et al., 1997) results in the precipitate activation of cyclin B/Cdc2 which triggers M phase. In S. pombe, phosphorylation of Tyr15 is catalyzed by the Wee1 and Mik1 kinases; the additional level of inhibitory phosphorylation, at least under normal conditions, at Thr14 exists only in higher eukaryotes. Phosphorylation of Thr14/Tyr15 has been implicated in the G₂/M checkpoint that blocks entry into mitosis in the presence of damaged DNA (Jin et al., 1996; Poon et al., 1997) or incomplete DNA replication, although a cyclin B/Cdc2 inhibitory factor may also be important (Kumagai and Dunphy, 1995).

A dual-specificity, membrane-associated protein kinase activity, able to phosphorylate Cdc2 on both Thr14 and Tyr15 was identified in Xenopus and HeLa cell extracts (Kornbluth et al., 1994; Atherton et al., 1994) and the Thr14/Tyr15 kinase, Myt1, was subsequently cloned (Mueller et al., 1995; Liu et al., 1997). In contrast to Wee1, which is localized in the nucleus (McGowan and Russell, 1995), Myt1 is localized to the endoplasmic reticulum (ER) and Golgi complex by a membrane-targeting domain on the C-terminal side of the catalytic domain (Liu et al., 1997). Specific subcellular localization of protein kinases is of significant importance; indeed the compartmentalization modulated by a nuclear export sequence of cyclin B1 regulates the physiological activity of Cdc2/cyclin B1 (Li et al., 1997; Hagting et al., 1998; Yang et al., 1998). Myt1 exhibits a more restricted substrate specificity than Wee1, in that it phosphorylates Cdc2/cyclin complexes but not Cdk2/cyclin complexes (Booher et al., 1997). This observation strongly suggests that Myt1 specifically regulates G₂/M phase transition through the inhibitory phosphorylation of Cdc2. In turn, Myt1 itself is hyperphosphorylated during mitosis, which is coincident with its inactivation (Mueller et al., 1995; Booher et al., 1997). Interestingly studies on Xenopus Myt1 have identified the kinase as an MPM-2 epitope containing protein (Mueller et al., 1995). The MPM-2 monoclonal antibody epitope is a mitotic phosphorylation dependent motif, minimally phospho-Ser/Thr-Pro (Westendorf et al., 1994), and is recognized by a monoclonal antibody raised by immunization with mitotic HeLa cell extracts (Davis et al., 1983).

Recently a novel mitotic regulator, Pin1, a highly conserved peptidyl-prolyl isomerase (PPIase) has been described, that also recognizes the MPM-2 epitope (Ranganathan et al., 1997; Yaffe et al., 1997). Pin1 inhibits entry into mitosis when overexpressed in HeLa cells and microinjected into Xenopus two-cell stage embryos, yet is also required for progression through mitosis (Lu et al., 1996; Crenshaw et al., 1998; Shen et al., 1998). PPIases catalyze rotation around the peptide bond preceding a proline residue and may regulate protein folding and intracellular trafficking (for review see Schmid, 1995). Phosphorylation of a Ser/Thr-Pro motif catalyzed by Cdc2/cyclin B could create a Pin1 substrate, enabling conformational changes driven by prolyl isomerization. The conformational change may alter enzymatic activity, or modify the sensitivity to dephosphorylation by phosphatases creating a timed phosphorylation induced activity/ modification. Thus Pin1, as well as Myt1, may play an important role in the activation of Cdc2, pivotal for the initiation of mitosis.

In this paper we examined the function and regulation of the biological activity of Myt1 and its interaction with the peptidyl-prolyl isomerase, Pin1. We demonstrate that overexpression of Myt1 inhibits G₂/M progression in the eukaryotic cell cycle. Myt1 expression was restricted to proliferating cells, and furthermore there was no apparent decrease in abundance following entry into mitosis, coincident with loss of Myt1 activity. We also localized the cell cycle specific MPM-2 epitopes within Myt1 to the C-terminal domain and show that Pin1 interacts with this domain in a phosphorylation-dependent manner. We demonstrated that the G₂/M arrest induced upon overexpression of Myt1 requires its C-terminal domain. We ascribe this to the ability of the C-terminal domain to associate with its substrate, Cdc2/cyclin B1, which is necessary for Myt1 to phosphorylate Cdc2, and also anchors Cdc2 within the cytoplasmic compartment. Because this domain is phosphorylated by and interacts with Cdc2/cyclin B1, we deduce that it must lie in the cytoplasm. In conclusion, we provide evidence that the C-terminal domain is required in combination with the catalytic domain for the biological activity of Myt1 in inhibiting G₂/M progression.

The 9E-10 α-Myc and 12CA5 αHA monoclonal antibodies were utilized for immunoprecipitations and immunoblotting of Myc- and HA-tagged proteins respectively. Rabbit polyclonal antibodies to human Pin1 were used in the GST pull-down assay, as described (Lu et al., 1996). Immunoblotting of human Cdc2 was carried out with rabbit polyclonal serum (antiserum 5517) as described (Watanabe et al., 1995). Antibodies C1-1 and R5084 against Myt1 were kindly provided by Drs Booher and Piwnica-Worms, respectively (Booher et al., 1997; Liu et al., 1997). The monoclonal antibody MPM-2, was purchased from Upstate Biotechnology.

The human Myt1 expression vector, pcDNAmycMyt1, was a kind gift from Dr Piwnica-Worms (Liu et al., 1997). The Myt1 truncation series, in the mammalian expression vector pCS, was derived by PCR amplification of the human Myt1 cDNA and cloned into the XbaI site. The specific primers used were: Myt1 (full length) 5′ primer AGAGAGTCTAGAGATGCTAGAACGGCCTCCTGC and 3′ primer AGAGAGTCTAGATCAGGTTGGGTCTAGGGTGTC, ΔN-Myt1 5′?primer AGAGAGTCTAGAGTCCTTCTTCCAGCAGA and 3′ primer as for full length Myt1, ΔC-Myt1 5′ primer as for full length Myt1 and 3′ primer AGAGAGTCTAGATCAGCCCAGGGGCTGTAGCC-AGCTG and ΔNC-Myt1 5′ primer as for ΔN-Myt1 and 3′ primer as for ΔC-Myt1. pEGFPF utilized in cotransfections of U2-OS cells was previously described (Jiang and Hunter, 1998). For the GST pull-down assay the Myt1 C-terminal domain was cloned to the pGEX-KG vector, Myt1 COOH-GST, at the BamHI and XhoI sites by PCR amplification using the 5′ primer AGAGAGGGATCCGCCAGCT-GGCTACAGCCCCTG and 3′ primer AGAGAGCTCGAGTCA-GGTTGGGTCTAGGGTGTC. Pin1 was cloned by PCR amplification into pEVRF0 (Janknecht, 1996), a mammalian expression vector that contains an N-terminal HA tag, at the XmaI and Xba1 sites with the 5′ primer AGAGAGCCCGGGATGGCGGACGAGGAGAAGCTG and 3′ primer AGAGAGTCTAGATCACTCAGTGCGGAGGATG-ATG. Mutagenesis of CDK consensus phosphorylation sites with the C-terminal domain of Myt1 (described in the text) was carried out using the QuikChange site directed mutagenesis method (Stratagene) according to the manufacturer’s protocol. Constructs and mutations were analyzed by automated DNA sequencing using an ABI Prism 377XL.

The S. pombe strain 1058 (h⁺, leu1-32, ura4-D18, ade6-210, his7-366) was transformed by electroporation with pSLF273-Myt1, pSLF273-Myt1(D/A) and pSLF273 alone (Forsburg and Sherman, 1997), containing the mid strength nmt41 promoter, under repressed conditions (5 µg/ml thiamine). Transformed cells were washed five times with thiamine free medium, inoculated into medium with and without thiamine and incubated for a further 24 hours. Cells were then harvested for examination by phase contrast microscopy and flow cytometric analysis.

Human embryonal kidney 293T or human osteogenic sarcoma U2-OS cells were grown to 25% confluence and then transfected by the calcium phosphate coprecipitation method. Cells were harvested 36-48 hours later for flow cytometric analysis, immunoblotting or immunoprecipitation.

HSF8 cells were synchronized by serum starvation. Briefly, proliferating HSF8s were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, glutamine and antibiotics in a humidified atmosphere containing 10% CO₂ at 37°C. Cells were washed three times in phosphate buffered saline (PBS), and incubated for at least 30 hours in serum free supplemented DMEM. Cells were released from the G₀ block by readdition of DMEM containing 10% fetal bovine serum. Upon release samples were taken at the noted times. Ninety percent of the cells were harvested for lysis and subsequent immunoblotting. The remaining cells from each time point were subjected to flow cytometric analysis as follows: cells were fixed in 70% ethanol, rehydrated and then treated with RNase A (100 µg/ml final concentration) and propidium iodide (40 µg/ml final concentration) in PBS for 30 minutes at 37°C. Cell-cycle distribution of 10⁴ cells was then determined on a Becton Dickinson FACScan. For cell synchronization studies with HeLa and U2-OS cells, cultures were either released from a G₁/S block, obtained by a double thymidine arrest as described (Heintz et al., 1983) or released from an M phase block, obtained from a thymidine/nocodazole block as described (Kanemitsu et al., 1998).

Cells were lysed in either Nonidet P-40 lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Nonidet P-40, 10% glycerol), or RIPA buffer. Buffers were supplemented with 1 mM phenylmethylsulfonyl fluoride, 0.2 mM DTT, 10 units/ml aprotinin, 20 µg/ml leupeptin, 0.1 mM p-nitrophenylphosphate, 10 nM microcystin, 1 mM sodium fluoride and sodium orthovanadate and 0.1 mM sn-glycerol 2-phosphate. Cell lysates were clarified by centrifugation and protein concentrations were determined using the Bio-Rad protein assay, as detailed in the manufacturer’s protocol. Immunoprecipitations were either performed with 12CA5 (αHA) or 9E-10 (αMyc) and Protein A beads (Repligen). The immune complexes or proteins were separated by discontinous SDS/polyacrylamide electrophoresis, transferred to Immobilon-P membrane (Millipore) and then blotted with primary and secondary antibodies and detected by enhanced chemiluminescence (Amersham) as detailed in the manufacturer’s protocol.

E. coli BL21 cells transformed with Myt1 GST-COOH were induced by IPTG (0.1 mM final concentration) for 4 hours at 18°C. Induced cultures were harvested by centrifugation, resuspended in lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM EDTA, 1 mM EGTA, 10% glycerol, 0.1% Nonidet P-40, 20 mM β-mercaptoethanol, 1 mM PMSF, 1 mM benzamidine, 10 units/ml aprotinin and 20 µg/ml leupeptin), lysed by sonication and clarified by centrifugation. GST fusion proteins were bound to glutathione-agarose by a batch mixing method, washed in lysis buffer and subsequently eluted from a column with 100 mM Tris-HCl (pH 8.0), 5 mM EDTA, 20 mM reduced glutathione and 20 mM β-mercaptoethanol. Eluted proteins were then dialyzed against 20 mM Tris-HCl (pH 8.0), 100 mM NaCl, 20% glycerol, 20 mM β-mercaptoethanol. GST fusion proteins were phosphorylated with recombinant baculovirus purified Cdc2/cyclin B1 in kinase buffer (50 mM Hepes (pH 7.5), 20 mM MgCl₂, 1 mM DTT, and either 10 µM ATP and 10 µCi [γ-³²P]ATP or 100 µM ATP alone, at 30°C for 30 minutes. Following phosphorylation GST fusion proteins were bound to glutathione-agarose, washed three time in binding buffer (20 mM Tris-HCl (pH 8.0), 10% glycerol, 75 mM NaCl, 2 mM DTT and 0.025% Tween-20), and incubated with recombinant Pin1 (Ranganathan et al., 1997). The bound proteins were washed three times and analyzed by SDS-PAGE, following addition of SDS-sample buffer.

Transfected 293T cells were lysed in 50 mM Hepes (pH 7.5), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl₂, 1 mM EGTA 1 mM DTT and also containing the phosphatase and protease inhibitors described above, and immunoprecipitations carried out with the 9E-10 antibody. The immunoprecipitations were washed in 0.05% Tween-20, 25 mM Tris-HCl (pH 7.5), 150 mM NaCl before resuspension in the kinase buffer 50 mM Tris-HCl (pH 7.5), 10 mM MgCl₂ and 1 mM DTT. The kinase reaction was carried out as above with recombinant baculovirus purified catalytically inactive Cdc2 (K/R)/cyclin B, resolved by SDS-PAGE and analyzed by autoradiography.

Our initial aim was to investigate whether Myt1 overexpression in a human osteosarcoma cell line, U2-OS, would affect cell cycle progression. To assess the effect on the cell cycle distribution, cells were transiently cotransfected with expression vectors for Myc-tagged Myt1 and a membrane-targeted form of GFP, EGFPF (Jiang and Hunter, 1998), as a marker of transfection. Cells were harvested 48 hours later, and the DNA content of cells both negative and positive for GFP, within the same population, was then assessed by flow cytometric analysis. Expression of Myt1 caused a significant increase in the G₂/M population relative to either untransfected cells or cells transfected with GFP alone (Fig. 1a). The proportion of cells arrested was dependent on the level of Myt1 expression (data not shown, and Table 1). However, it was unclear whether Myt1 overexpression caused an increase in the G₂/M phase population by a block in cell cycle progression, or as a consequence of acceleration of G₁ and S phases. To address this we treated Myt1 transfectants with thymidine, to synchronize cells at the beginning of S-phase, 12 hours prior to harvesting for flow cytometric analysis. If cells transfected with Myt1 are already blocked in G₂/M phase, the drug should have little or no effect. As shown in Fig. 1b, cells transfected with Myt1 and subsequently treated with thymidine still had significant numbers of cells in G₂/M relative to cells transfected with GFP alone. Hence Myt1 overexpression causes a G₂/M phase arrest and not an acceleration of G₁ or S phases.

Interestingly, transfection with catalytically inactive Myt1 (Asp251Ala) (shown to lack activity by an in vitro kinase assay, see Fig. 6c), also induced an accumulation of cells within the G₂/M population, albeit less efficiently than wild-type Myt1 (Table 1), indicating that Myt1 kinase activity is not essential for the G₂/M arrest phenotype at higher expression levels.

Next we examined whether Myt1 functions in a fashion analogous to Wee1 when overexpressed in the fission yeast S. pombe. Although a homologue of Myt1 has not yet been described in S. pombe, Thr14 is conserved in S. pombe Cdc2 and phosphorylation of Thr14 is observed under some circumstances (Den Haese et al., 1995). Wild-type and catalytically inactive Myt1 were cloned into a fission yeast expression vector (pSLF273) that uses the mid-strength nmt promoter (Forsburg and Sherman, 1997). The resulting plasmids and vector alone were used to transform a wild-type S. pombe strain 1058, and the effect of Myt1 expression following thiamine depletion, which derepresses transcription from the nmt promoter, was assessed. Expression of Myt1 generated very elongated cells with a single diffuse nucleus (Fig. 2), indicative of an arrest at the G₂/M transition. Flow cytometric analysis was also consistent with such a conclusion (data not shown). Expression of catalytically inactive Myt1 (Asp251Ala) failed to arrest cell proliferation (data not shown). Hence, Myt1 kinase activity is required for the G₂ arrest when overexpressed in wild-type S. pombe. Continued growth of the cell following DNA replication, rather than entry into mitosis is also a hallmark of overexpression of the Wee1Hu catalytic domain in S. pombe (Igarashi et al., 1991). We conclude that Myt1, like Wee1, causes a G₂ cell cycle arrest following overexpression in S. pombe. These studies suggest that negative regulation of Myt1 is necessary for normal cell cycle progression through G₂/M.

To assess whether Myt1 is regulated at the protein level we examined the cell cycle phase-specific expression profile of Myt1. First, the cell cycle expression profile of Myt1 in the HSF strain (HSF8) of non-transformed, non-immortalized human diploid foreskin fibroblasts (W. Jiang, unpublished observations) was investigated. Cells were serum starved for 48 hours to induce synchronization/quiescence and released from the G₀ block by readdition of fetal bovine serum. Cells were subsequently harvested at intervals for cell cycle analysis by flow cytometry (Fig. 3a) and western blotting analysis with αMyt1 and αCdc2 antisera. Quiescence, induced by either serum starvation (Fig. 3a, upper panel), or contact inhibition at confluence (data not shown), was accompanied by a marked decrease in Myt1 levels. During the transition from G₀ into the cell cycle Myt1 protein was first detected during early S-phase, similar to the kinetics previously reported for its primary substrate Cdc2 (Fig. 3a, lower panel) (Pagano et al., 1993). As expected, a shift to a faster mobility form of Cdc2, due to Thr14/Tyr15 dephosphorylation during M-phase (Solomon et al., 1992), was observed at 27-30 hours following release from serum starvation. Myt1 protein was readily detectable at this time implying that degradation of Myt1 does not precede the dephosphorylation and therefore the activation of Cdc2.

Next we synchronized U2-OS cells at the beginning of S-phase by treatment with thymidine, then washed, released and promptly blocked the cells in metaphase by addition of the microtubule-depolymerizing drug nocodazole. Cells were subsequently released from the drug-induced block and samples taken at time intervals for flow cytometric analysis (Fig. 3b) and, in parallel, a portion was retained for the preparation of cell extracts. Western blotting analysis indicated that Myt1 protein levels did not fluctuate dramatically during the cell cycle in proliferating U2-OS cells (Fig. 3b). A small decrease in the level of the lower Myt1 protein band was observed in G₂/M phase extracts, although not dramatic and may indeed result from a reduction in mobility on SDS-PAGE following hyperphosphorylation in mitosis (see below). Similar results were also obtained when U2-OS cells and HeLa cells that were synchronized at the beginning of S-phase by a double thymidine block and released, enter mitosis (data not shown).

These data suggest that Myt1 is a proliferation marker that is downregulated upon quiescence, consistent with a role in the regulation of the G₂/M phase transition, but is not regulated at the protein level in proliferating cells. Although western blotting did not detect a significant decrease in Myt1 as cells enter mitosis, the protein had a dramatically slower mobility on SDS-PAGE in extracts prepared from cells blocked in metaphase (Fig. 3b). Previous studies have determined that this shift is due to hyperphosphorylation at mitosis (Mueller et al., 1995; Liu et al., 1997; Booher et al., 1997), which is coincident with Myt1 inactivation. Hence, we next examined whether phosphorylation of Myt1 was of importance for the regulation of the Cdc2 inhibitory activity of Myt1.

The monoclonal antibody MPM-2, which recognizes mitotic phosphoepitopes, has been an extremely useful tool for dissecting the function of mitotic phosphorylation of many proteins (Davis et al., 1983). To determine whether mitotic hyperphosphorylated Myt1Hu contains MPM-2 epitopes, transfected Myc-tagged Myt1Hu was immunoprecipitated with αMyc antibody from mitotic cellular extracts, prepared from 293T cells arrested in metaphase with nocodazole, and analyzed by western blotting with the MPM-2 antibody. The full length protein was recognized by the MPM-2 antibody (Fig. 4b, left panel), consistent with the results for Xenopus Myt1 (Mueller et al., 1995). To determine the region(s) in which the epitope(s) are localized, we prepared Myt1 expression constructs with either N- or C-terminal deletions, or both (Fig. 4a), and expressed these transiently in 293T cells. Proteins were immunoprecipitated with αMyc from mitotic cell extracts, and blotted with the MPM-2 antibody. Deletion of the N terminus (ΔN-Myt1) up to and including residue Pro102 had little effect on the MPM-2 reactivity. In contrast, truncation of the C-terminal domain at residue Gly408 (ΔC-Myt1), up to but not including the putative transmembrane domain, abolished reactivity with the MPM-2 antibody (Fig. 4b, left panel). The blot was stripped and reprobed with αMyc antibody to demonstrate that all the Myc-tagged Myt1 fusion proteins were immunoprecipitated (Fig. 4b, right panel). Liu et al. (1997) have previously demonstrated that truncation of the C-terminal 98 residues, leaving the putative transmembrane domain intact, has no effect on subcellular localization of Myt1, and thus the ΔC-Myt1 protein should have been present in the same cellular environment as wild-type Myt1, and was therefore in a position to be phosphorylated. In conclusion, the major MPM-2 reactivity is localized to the C-terminal domain of Myt1, which contains 5 putative CDK phosphorylation sites and therefore putative MPM-2 epitopes.

Consistent with the deletion studies, mutation of all 5 CDK consensus phosphoacceptor residues, Thr412, Ser416, Ser441, Thr455 and Thr461, prevented recognition of Myt1 by the MPM-2 antibody (data not shown). To determine which if any of these CDK consensus phosphorylation sites, within the C-terminal domain, are targets for CDK phosphorylation, a GST fusion with the C-terminal domain of Myt1 was constructed (residues Ala401-COOH) (Fig. 4a). An in vitro kinase assay demonstrated that the C-terminal domain is a good substrate for Cdc2/cyclin B1 (Fig. 4c). Interestingly phosphorylation of the C-terminal domain dramatically reduced its SDS-PAGE mobility (Fig. 4c) as observed for the wild-type protein during mitosis (Fig. 3b) (Mueller et al., 1995; Booher et al., 1997). Furthermore, a range of decreased electrophoretic mobilities of GST-COOH was observed by SDS-PAGE following phosphorylation, indicative of phosphorylation at multiple sites.

Pin1, a petidyl-prolyl isomerase (PPIase), may regulate the entry into and progression through mitosis (Lu et al., 1996). Structural analysis of the PPIase active site led to the hypothesis that Pin1 recognizes its substrates with acidic residues prior to the isomerized N-terminal proline peptide bond (Ranganathan et al., 1997). Recently, Yaffe et al. (1997) demonstrated that indeed Pin1 preferentially binds to and isomerizes proline bonds immediately preceded by a phospho-Thr or phospho-Ser, similar to the minimal MPM-2 epitope. Pin1 may therefore regulate mitosis by phosphorylation specific isomerization of the substrates of the MPM-2 kinases. Subsequent work has shown that this binding specificity is largely a property of the Pin1 WW domain (Lu et al., 1999).

To investigate whether Myt1 and Pin1 interact, 293T cells were transiently cotransfected with HA-tagged Pin1 and Myc-tagged Myt1 expression constructs, and cell extracts were prepared. Immunoprecipitation of either Myt1 or Pin1, via the epitope tags, and subsequent western analysis demonstrated that Pin1 and Myt1 did indeed interact. Immunoprecipitation of Myt1 and the subsequent detection of coimmunoprecipitated Pin1 is shown in Fig. 5a. Analysis of Pin1 coexpressed with either ΔN-Myt1 and ΔC-Myt1 (Fig. 4a) demonstrated that this interaction was dependent on the C-terminal domain of Myt1 (Fig. 5a). This suggests that the MPM-2 epitopes in the C-terminal domain of Myt1 participate in this interaction.

Two approaches were pursued to examine the dependence of the interaction on phosphorylation. First, mutation of the five consensus CDK phosphorylation sites, Thr412, Ser416, Ser441, Thr455 and Thr461 (Ser/Thr to Ala), in the Myt1 C terminus (Fig. 4a) significantly reduced the coimmunoprecipitation of wild-type Pin1 with Myt1 (data not shown). However, mutation of these phosphoacceptor sites did not abrogate its ability to induce a G₂/M arrest in U2-OS cells (data not shown). Second, utilizing the GST-COOH fusion protein (Fig. 4a and 4c) in a pull down assay, we demonstrated that Pin1 only interacted with the C-terminal domain following Cdc2/cyclin B1 phosphorylation (Fig. 5b). Together these results demonstrate that Pin1 interacts directly with the C-terminal domain of Myt1 in a phosphorylation-dependent manner. However, we have been unable to detect any modulation of in vitro Myt1 catalytic activity through its association with Pin1 (T. Tokusumi and N. Watanabe, unpublished observations).

The fact that Myt1 mitotic hyperphosphorylation is coincident with its inactivation, coupled with our localization of the mitotic MPM-2 epitopes to the C-terminal domain, suggests that phosphorylation of these sites could regulate the function of the C-terminal domain. To determine whether this domain is required for Myt1 biological activity in mammalian cells, ΔC-Myt1 (Fig. 4a) and a catalytically inactive variant of ΔC-Myt1 (Asp251Ala), were transfected into U2-OS cells and the effect on cell cycle progression was examined by flow cytometric analysis. Surprisingly, transient transfection studies in U2-OS cells demonstrated that ΔC-Myt1 (Fig. 6a) or ΔC-Myt1 (Asp251Ala) (data not shown) were unable to induce the G₂/M arrest phenotype observed for the wild-type control. Western blotting analysis indicated that the truncation mutant was expressed as well or better than wild-type Myt1 (Fig. 6b).

Next we examined the ability of ΔC-Myt1 to phosphorylate Cdc2 compared to that of wild-type Myt1. We carried out in vitro kinase assays on αMyc-immunoprecipitates of Myc-tagged wild-type Myt1 and ΔC-Myt1 from asynchronous transfected 293-T cells cell extracts using the inactive kinase Cdc2(KR)/cyclin B1 as a substrate. Consistent with the observation that ΔC-Myt1 does not cause a G₂/M arrest in U2-OS cells, deletion of the C-terminal domain significantly reduced Myt1 phosphorylation of Cdc2 (Fig. 6c). Interestingly, ΔC-Myt1 underwent autophosphorylation to the same extent as full length Myt1 (data not shown), and hence is still enzymatically active regardless of the C-terminal truncation. In contrast, as expected the Myt1 (Asp251Ala) mutant lacked both Cdc2 phosphorylating activity and autophosphorylating activity. Analysis by western blotting indicated that equal amounts of the Myt1 proteins, utilized in the kinase assays, were present in the cell extracts (Fig. 6c, right panel). These results imply that the C-terminal region plays some role in the interaction of Myt1 with its substrate Cdc2/cyclin B1.

To determine whether the C-terminal domain is required for the recognition and recruitment of its substrate, Cdc2/cyclin B1, we immunoprecipitated wild-type Myt1 and ΔC-Myt1 from 293T cell extracts and resolved the coimmunoprecipitating proteins by SDS-PAGE. Western blotting analysis with αCdc2 antibodies demonstrated that endogenous Cdc2 coimmunoprecipitated with wild-type Myt1 but not with ΔC-Myt1 (Fig. 6d). The blot was stripped and reprobed with αMyc antibody to demonstrate that ΔC-Myt1 was indeed immunoprecipitated. Interestingly, the majority of the associated Cdc2 was in the slowest mobility form, representing the doubly-phosphorylated and hence inactive Cdc2. Previous studies have demonstrated that Myt1 will only recognize and therefore phosphorylate its substrate, Cdc2, in a cyclin B1-dependent manner (Booher et al., 1997). Since we observed only doubly-phosphorylated Cdc2 in association with Myt1 (Fig. 6d), we deduce that Cdc2/cyclin B complexes were coimmunoprecipitated with Myt1. In contrast, both doubly-phosphorylated, inactive Cdc2 and hypophosphorylated, active Cdc2, were coimmunoprecipitated with the catalytically inactive Myt1 (Asp251Ala) protein (Fig. 6d); however, the level of the hypophosphorylated, active Cdc2 brought down was proportionately much less than the level of this form in the whole cell lysate. The presence of some doubly-phosphorylated Cdc2 associated with catalytically inactive Myt1 (Asp251Ala) could be due to its phosphorylation by endogenous Myt1.

We have demonstrated that inappropriate expression/activity of Myt1 prevents cells of both lower and higher eukaryotes from completing cell division after DNA replication. This is consistent with the fact that overexpression of a dominant negative Cdc2, to date the unique substrate of Myt1, causes cells to arrest at the G₂ to M transition (van den Heuvel and Harlow, 1993). Our studies also indicate that the C-terminal domain is essential for the biological activity of Myt1, as indicated by its requirement for Myt1 to cause a G₂/M arrest, and its role in recruitment and phosphorylation of Cdc2/cyclin B1 in vitro.

Expression studies in mammalian cells of a non-phosphorylatable Cdc2 (Thr14/Ala-Tyr15/Phe) (Cdc2AF) indicate that inhibitory phosphorylation is not solely responsible for repressing premature mitosis (Heald et al., 1993; Jin et al., 1996). The initiation of mitosis is accompanied by the dephosphorylation/activation of Cdc2 and translocation to the nucleus (Pines and Hunter, 1991, 1994). Heald et al. (1993) demonstrated that dephosphorylated/active Cdc2, which is restricted to the cytoplasm, is unable to catalyze entry into M phase. Regulated subcellular localization of Cdc2 may also explain the poor efficiency with which expression of Cdc2AF induces premature entry into mitosis (Heald et al., 1993; Jin et al., 1996, 1998). This is consistent with our observation that overexpression of a catalytically inactive Myt1 (Asp251Ala) mutant was also able to delay cell cycle progression in mammalian cells, which we suggest is due to its ability to bind Cdc2/cyclin B1 and sequester this complex in the cytoplasm. Sequestration of the mitotic kinase within the cytoplasmic compartment, thereby preventing its translocation to the nucleus, would effectively inhibit mitotic entry. Thus, Myt1 can inhibit Cdc2 and hence entry into mitosis by two independent mechanisms. First, by binding and sequestration of Cdc2/cyclin B1 from the nucleus, and second following recruitment of Cdc2/cyclin B1, by inhibitory phosphorylation of Thr14 and Tyr15. This model has received independent support from Liu et al. (1999), who have also shown that the C-terminal domain of Myt1 interacts with Cdc2/cyclin B1 via an RXL cyclin-binding motif, RNL, located within the C-terminal 63 residues of Myt1, and that this interaction is essential for kinase-inactive Myt1 to arrest mammalian cells in G₂. In addition, Liu et al. have also shown that the Myt1 C-terminal domain sequesters cyclin B1 in the cytoplasm of cells treated with leptomycin B, where it normally accumulates in the nucleus, thus providing compelling evidence that Myt1 can negatively regulate Cdc2/cyclin B1 by preventing its entry into the nucleus.

In contrast to its inhibitory effect in human cells, catalytically-inactive Myt1Hu did not induce a cell cycle arrest in S. pombe. This difference may be due to differing expression levels or a lower affinity of Myt1Hu for S. pombe Cdc2/Cdc13 (Cdc13 is the S. pombe cyclin B1 homologue). Transformation of Myt1Hu, under the full strength nmt promoter, into a humanized Cdc2 S. pombe strain (h⁻, leu1-32, his3-237, cdc2::CDC2Hu) under repressed conditions prevented colony formation, whereas colonies were obtained following a similar transformation of a wild-type S. pombe strain (N. J. Wells, unpublished observations). Thus, the low level of Myt1Hu expressed in repressed cells was sufficient to arrest S. pombe cells relying on human Cdc2, but insufficient to arrest cells with endogenous Cdc2. This suggests that Myt1Hu does have a higher affinity for human Cdc2 than S. pombe Cdc2. This could be because human Cdc2 is intrinsically a better substrate, but it is also possible that the S. pombe Cdc13 cyclin has a lower affinity for the RNL motif in Myt1 than human cyclin B1. Furthermore, S. pombe Cdc2/Cdc13 is reportedly restricted to the nuclear compartment (Alfa et al., 1989), and this may in part protect the Cdc2 kinase from sequestration by Myt1 in the cytoplasm, although a more recent study indicates that the translocation of Cdc2/Cdc13 into the nuclear compartment may even occur in S. pombe (Audit et al., 1996). We have been unable to suppress the wee1-50 mutant phenotype by expression of Myt1Hu from the nmt promoter (N. J. Wells, unpublished observations), which may also in part be accounted for by different subcellular locations of Myt1 and Wee1, which is nuclear.

Western blotting analyses indicate that Myt1 protein is dramatically downregulated upon cell contact inhibition or serum starvation in normal human fibroblasts. We did not observe a significant decrease in expression following transit through mitosis in either U2-OS or HeLa cells. This is consistent with the observation that when cells are arrested in metaphase, by treatment with nocodazole, Myt1 protein is still present (Fig. 3) (Booher et al., 1997). Indeed, cells blocked in metaphase by nocodazole and subsequently treated with DNA damaging agents, lose Cdc2/cyclin B activity concomitant with the appearance of phosphorylation of Thr14 and Tyr15 on Cdc2 (Poon et al., 1997). This implies that a Thr14 kinase is activated or at least derepressed during DNA damage in the metaphase arrest; this Thr14 kinase is probably Myt1, which is the major Thr14 kinase in Xenopus eggs (Mueller et al., 1995). In contrast to Myt1Hu, Wee1Hu protein levels are decreased at M/G₁ phases, highlighting a further difference in the regulation of the two higher eukaryotic Cdc2 inhibitory kinases (Watanabe et al., 1995; McGowan and Russell, 1995).

Myt1 inactivation is coincident with its mitotic hyperphosphorylation (Mueller et al., 1995; Booher et al., 1997). Utilizing deletion mutants we have localized the MPM-2 epitopes, and hence mitotic specific phosphorylation sites within Myt1 to the C-terminal domain, which contains 5 putative CDK Ser/Thr-Pro phosphorylation sites. We found that the Myt1 C-terminal domain is a good substrate for Cdc2/cyclin B1 complex in vitro, and that Cdc2/cyclin B1 phosphorylation also led to a dramatic reduction in mobility on SDS-PAGE, as is observed for Myt1 in mitotic cells (Mueller et al., 1995; Booher et al., 1997). Cell cycle phase specific phosphorylation of Myt1 at the N terminus has also been predicted, based on the observation that an αMyt1 N-terminal peptide antibody is able to recognize interphase but not mitotic Myt1 (Booher et al., 1997). The N terminus also contains consensus CDK phosphorylation sites, but our results suggest that the MPM-2 epitope monoclonal antibody is unable to recognize these phosphorylation sites, if they are indeed modified, and that Pin1 is also unable to interact with this domain.

Pin1 associates with epitopes that overlap with those of the mitotic phosphoepitope specific MPM-2 monoclonal antibody (Yaffe et al., 1997). Consistent with that study, we have demonstrated that Pin1 interacts with Myt1 in vivo and in vitro in a phosphorylation dependent manner. Whilst this work was in preparation Shen et al. (1998) also demonstrated that the mitotic hyperphosphorylated form of Myt1 was specifically isolated in a Pin1-GST pull down assay. We extended this preliminary observation by demonstrating that it is a direct, phosphorylation-dependent interaction, and requires one or more of the 5 CDK consensus sites in the C-terminal domain of Myt1. Interestingly, Pin1 also interacts with two other Cdc2 regulatory proteins, Wee1 and Cdc25C (Crenshaw et al., 1998; Shen et al., 1998). As the phosphorylation of Thr14/Tyr15, and hence the activity of Cdc2, is tightly regulated by feedback loops (reviewed by Coleman and Dunphy, 1994), the ability of Pin1 to interact with Cdc25, Wee1 and Myt1 provides the basis for modulation of these feedback pathways. To date direct changes in enzymatic activity due to conformational control catalyzed by the PPIase domain of Pin1 are still debatable. We examined the effect of Pin1 association with Myt1 in vitro and did not detect a significant change in the kinase activity of solubilized Myt1 in an immunoprecipitate (N.T. and N.W., unpublished observations). Future studies are needed with membrane-associated Myt1, where it may adopt a different conformation, to study the regulatory role of modifications including phosphorylation and Pin1 interaction on enzyme activity.

Initial fractionation studies in Xenopus egg and HeLa cell extracts showed that the Cdc2 Thr14 kinase was membrane associated and that its membrane association was not disrupted by high salt treatment (Kornbluth et al., 1994; Atherton et al., 1994). Subsequently, a membrane-targeting domain was identified in the Myt1 sequence downstream of the catalytic domain, and was shown by immunofluorescence staining to be required for the localization of Myt1 to the ER and Golgi complex (Liu et al., 1997). From these data it was predicted that the kinase homology domain is oriented toward the cytosolic face of these membrane compartments, but it was unclear whether the C-terminal tail is in the lumen of the ER (i.e. a type II transmembrane protein), or whether it is also in the cytosol. From our work showing that the Myt1 C-terminal domain binds Cdc2/cyclin B1 in interphase and is phosphorylated in mitosis, we infer that the Myt1 C-terminal domain is exposed in the cytoplasm, where it is able to recruit Cdc2/cyclin B1, an event necessary for phosphorylation of Thr14 and Tyr15.

A previous study indicates that phosphorylation of Myt1 by Cdc2 in vitro does not inhibit its Thr14/Tyr15 kinase activity (Booher et al., 1997). Perhaps some regulatory factor that associates with phosphorylated Myt1 is absent and/or another protein kinase(s) is also required for inhibition. One possibility is that mitotic phosphorylation of the Myt1 C-terminal domain could decrease the binding of cyclin B1/Cdc2, and recent evidence suggests that this is indeed the case (Liu et al., 1999). Palmer et al. (1998) have recently demonstrated that p90^Rsk can also bind and to phosphorylate the Myt1 C-terminal domain. Their data suggest that phosphorylation of the C-terminal domain by p90^Rsk inhibits the ability of Myt1 to phosphorylate and inhibit Cdc2. It will be interesting to examine whether phosphorylation of the C-terminal domain by p90^Rsk might also inhibit Myt1 activity by decreasing the affinity of the Myt1 C-terminal domain for Cdc2/cyclin B complexes. This study also demonstrates that the C-terminal domain of Myt1 is sufficient for interaction with Cdc2. Thus, phosphorylation of the C-terminal domain by both p90^Rsk and Cdc2/cyclin B and its phosphorylation-dependent interaction with Pin1 could all modulate the activity of Myt1.

In summary, our studies highlight the functional importance of the non-catalytic Myt1 C-terminal domain in the regulation of Cdc2/cyclin B1 activity, through its interaction with and phosphorylation by Cdc2/cyclin B1.

We thank H. Piwnica-Worms for providing the Myt1 cDNA and for sharing unpublished information with us, H. Piwnica-Worms and R. N. Booher for providing anti-Myt1 sera, S.L. Forsburg and J. Millar for S. pombe strains and expression constructs and help with the S. pombe experiments, D. Chambers for help with flow cytometric analysis, S. Tournier for help with photomicroscopy, and J. Noel, R. Janknecht and R. Shah for helpful discussions. This work was supported by fellowships from the Wellcome Trust #046847 (N.J.W.) and USPHS grants CA14195 and CA39780 (T.H.). T.H. is a Frank and Else Schilling American Cancer Society Research Professor.