Biphasic activation of two mitogen-activated protein kinases during the cell cycle in mammalian cells.

We studied mitogen-activated protein kinase (MAPK) activities during the cell cycle of Chinese hamster ovary (CHO) cells using site-specific antibodies against extracellular signal-regulated kinase-1, a 44-kDa MAPK (Boulton, T.G., Yancopoulos, G.D., Gregory, J.S., Slauer, C., Moomaw, C., Hsu, J., and Cobb, M.H. (1990) Science 249, 64-67). These antibodies detected two distinct MAPKs (44- and 42-kDa MAPKs) in CHO cells. CHO cells were arrested at metaphase in the M phase by treatment with nocodazole, and activities of MAPKs were analyzed at specific time points after release from arrest. Immune complex kinase assay and renaturation and phosphorylation assay in substrate-containing gel revealed that both 44- and 42-kDa MAPKs had activities in the G1 through S and G2/M phases and were activated biphasically, in the G1 phase and around the M phase. MAPKs were inactivated in metaphase-arrested cells. The amount of MAPKs did not change significantly in the cell cycle. In the G1, S, and G2/M phases, MAPKs were phosphorylated on both tyrosine and threonine residues and dephosphorylated in metaphase-arrested cells. Our data suggest that MAPKs may play some role in the cell cycle other than G0/G1 transition.

detected two distinct MAPKs (44-and 42-kDa MAPKs) in CHO cells. CHO cells were arrested at metaphase in the M phase by treatment with nocodazole, and activities of MAPKs were analyzed at specific time points after release from arrest. Immune complex kinase assay and renaturation and phosphorylation assay in substrate-containing gel revealed that both 44-and 42-kDa MAPKs had activities in the GI through S and G2/M phases and were activated biphasically, in the G1 phase and around the M phase. MAPKs were inactivated in metaphase-arrested cells. The amount of MAPKs did not change significantly in the cell cycle. In the G1, S, and G2/M phases, MAPKs were phosphorylated on both tyrosine and threonine residues and dephosphorylated in metaphase-arrested cells. Our data suggest that MAPKs may play some role in the cell cycle other than Go/G1 transition.
Mitogen-activated protein kinase (MAPK)' is a kinase that is activated by various mitogens (1)(2)(3)(4)(5)(6) and plays an important role in the kinase cascade that results in Go/G1 transition (7-9). It is a unique serine/threonine kinase that is activated through phosphorylation on its tyrosine and threonine residues (10, 11) and is believed to be the 42-kDa protein that is phosphorylated on stimulation with mitogens (12). Recently, a partial cDNA of the putative MAPK (or extracellular signalregulated kinase-1 (ERK-1)) has been cloned from rat brain cDNA library (13). Subsequently, two related kinases have been cloned (14). Interestingly, there is a considerable homology between amino acid sequences of the ERK-1 and * This work was supported by Juvenile Diabetes Foundation International Grant 190831 (to T. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: MAPK, mitogen-activated protein kinase: ERK-1, extracellular signal-regulated kinase-1; CHO cell, Chinese hamster ovary cell; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate; RT, room temperature. FUS3 or KSS1, which is a yeast gene involved in cell cycle control. In starfish and Xenopus oocytes, homologous kinase is activated in the M phase (15) and phosphorylated on tyrosine residue (16-19) (for review, see Refs. 20 and 21). Therefore, we were interested in whether MAPKs of mammals are controlled in a cell cycle-dependent manner. In the present study, MAPK activity was measured in the cell cycle in proliferating Chinese hamster ovary (CHO) cells. To this end, we used specific antisera against two synthetic peptides corresponding to the regions of the predicted amino acid sequence of ERK-1, a putative MAPK (13). To synchronize the cells, reagents that arrest cells in specific points in the cell cycle were used in combination with a collection of metaphase cells with mechanical shakeoff (22). Two reagents were used in this study. Nocodazole, a microtubule disrupting reagent, is known to arrest cells in the M phase (23), and aphidicholin, an inhibitor of the synthesis of DNA, is known to arrest cells in the S phase (24). Since synchronization by nocodazole seemed to be more effective, this reagent was used in most of the experiments in this study.
In this study, we have found that both 44-and 42-kDa MAPKs have activities in the G1 through S and G2/M phases and are activated biphasically: in GI and around the M phase.
We have also shown that activation of MAPKs in the cell cycle correlates with phosphorylation on both tyrosine and threonine residues as in stimulation with mitogens. Our data suggest that MAPKs may play some role in the cell cycle other than Go/G1 transition. Cell Culture and Synchronization-CHO cells were grown in aminimal essential medium (GIBCO) supplemented with 10% fetal calf serum. To avoid contact inhibition, cells were plated in a 9-cm dish at a density between 5 X lo5 and 2 X 106/dish. CHO cells were arrested at the metaphase/anaphase transition of the M phase by treating cells for 12-16 h with 0.5 pg/ml nocodazole. M phase-arrested cells were collected by gentle pipetting. Over 90% cells collected by this mechanical shakeoff method were in metaphase of the M phase (22). The cells were centrifuged at 1,000 rpm for 3 min, resuspended in fresh medium, and plated again. Replated cells were collected at the indicated time after replating. Cells were washed once with serumfree medium and were quickly frozen in liquid nitrogen and stored at -70 "C until assay. In S phase arrest, cells released from M phase arrest were further treated with aphidicholin. 4 h after release, 1 pg/ ml aphidicholin was added, and the cells were treated for 12 h. At the end of treatment, the cells were washed with fresh medium once and cultured further. Cells were washed and frozen as above after the indicated time from release.

Materials
Site-specific Antibodies to MAPK and cdc2 Kinuse-Antibodies were raised against keyhole limpet hemocyanin-conjugated synthetic peptides of either C-terminal peptide of cdc2 kinase or parts of MAPK reported as ERK-1. The characterization of the anti-MAPK antibodies is described elsewhere (25). Anti-MAPK antibodies raised against triple tyrosine-containing peptide (Y91:ITVEEALAHPYLEQYYD-DTDE) and C-terminal peptide (C92:ELIFQETARFQPGAPEAP) were used in this study and designated as aY91 and aC92, respectively. Anti-cdc2 kinase antibody was raised against peptide Y21: CLDNQIKKM and designated as aY21.
Immunoprecipitation and Immune Complex Kinase Assay-The immunoprecipitation procedure was essentially as described (25). Cells from each time point were solubilized in 900 p1 of lysis buffer containing 50 mM HEPES pH 7.4, 100 mM NaCI, 1 mM EDTA, 20 mM NaF, 20 mM sodium pyrophosphate, 1 mM sodium vanadate, 1% Triton X-100, 100 units/ml aprotinin, and 1 mM PMSF. The cell lysates were centrifuged at 14,000 rpm for 15 min, and the supernatants were recovered as crude lysates. The protein concentration of the samples was determined by Bradford method (26) and adjusted before assay. To 450 p1 of lysate, 3 p1 of anti-cdc2 (aY21) or anti-MAPK (aC92) serum was added. When antiserum aY91 was used for immunoprecipitation, 0.1-0.15% SDS was added to the lysis buffer. Without SDS, antiserum aY91 immunoprecipitated MAPK very poorly. After a 1-h incubation on ice, 40 pl of a 50% (v/v) suspension of protein A-Sepharose was added and incubated for 3 h at 4 "C with continuous agitation. Immunoprecipitated complex was washed twice with lysis buffer, once with 50 mM Tris, pH 7.4, 500 mM LiCl, twice with 50 mM Tris, pH 7.4, 10 mM MgCl,. Kinase assay was initiated by adding 30 pl of kinase buffer containing 50 mM Tris-HC1, pH 7.4, 10 mM MgCl,, 1 mM dithiothreitol, 20 mM ATP, 2 pCi of [r-"P] ATP, and substrate (final 0.2 mg/ml) to immune complex. As substrate, histone H1 and myelin basic protein were used for cdc2 kinase and MAPK, respectively. After a 15-min incubation at RT, the reaction was stopped by adding 15 p1 of 5 X concentrated Laemmli's sample buffer (27) and boiling for 5 min.
Thymidine Uptake-For thymidine uptake, cells arrested with nocodazole were collected and replated into six-well dishes (1 X lo5 cells/well). At the indicated times after replating, cells were washed once with serum-free medium and incubated in the same medium containing 5-6 X lo6 cpm of [methyl-3H]thymidine for 1 h a t 37 "C.
At the end of incubation, cells were washed three times with ice-cold phosphate-buffered saline and then 10% trichloroacetic acid and solubilized in 0.4 N NaOH. The count incorporated into the trichloroacetic acid-insoluble fraction was counted in Aquasol.
Western Blotting-Part of cell lysates were separated for Western blotting when the protein concentration was adjusted before immunoprecipitation. The lysates were mixed with 0.25 volume of 5 X concentrated Laemmli's sample buffer and boiled for 5 min. Samples were resolved through 10% polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore) with semidry blotting apparatus at 2.5 mA/cm2 constant current for 1 h.
The membrane was blocked with 3% albumin (Sigma fraction V) in phosphate-buffered saline for 30 min a t RT. Blocked membrane was incubated with 1/100 diluted antiserum in Tris-buffered saline for 1 h at RT or overnight at 4 "C. After washing in Tris-buffered saline for 30 min at RT, the membrane was incubated with '251-protein A in Tris-buffered saline for 1 h at RT. Then membrane was washed in Tris-buffered saline containing 1 M NaCl for 1 h at RT, and after drying up, autoradiographed at -70 "C with an intensifying screen.
In Viuo Labeling-Cells were treated with nocodazole for 8-12 h, collected with gentle pipetting and centrifugation, resuspended in fresh medium, and replated in a 6-cm dish at the density 4 X 105-8 X 1O5/dish. After the indicated time, the medium was aspirated and replaced with 4 ml of labeling medium (phosphate-free RPMI buffered with 20 mM HEPES pH 7.4 and supplemented with 10% dialyzed fetal calf serum). Cells were labeled with [32P]orthophosphate (1-2 mCi/dish) for 2-3 h a t 37 "C. At the end of the incubation, the medium was aspirated, and the cells were quickly frozen in liquid nitrogen and then solubilized on ice in 160 pl of lysis buffer. After 10 min on ice, the lysate was centrifuged a t 15,000 rpm for 15 min at 4 "C. The supernatant was mixed with 7.5 p1 of 10% SDS, vortexed, and then adjusted to 500 pl. The lysate was incubated with 5 pl of immunoaffinity-purified anti-MAPK antibody (aY91) for 2 h on ice.
The immune complex was precipitated by incubation with 70 pl of Pansorbin for 30 min on ice. The precipitated immune complex was washed five times with 500 p1 of wash buffer (lysis buffer containing 0.1% SDS). The 32P-labeled proteins were eluted by boiling in Laemmli's sample buffer and resolved through two-dimensional gel electrophoresis followed by autoradiography.
Renaturation and Phosphorylation in Substrate-containing Gel-Renaturation of the kinase after SDS-polyacrylamide gel electrophoresis was essentially as described (28). In brief, samples were resolved through a minigel containing 0.3-0.5 mg/ml myelin basic protein.
After electrophoresis, the gel was immersed in 50 mM Tris, pH 8.0, 20% isopropyl alcohol for 1 h at RT to remove SDS and then incubated in 50 mM Tris, pH 8.0, 5 mM P-mercaptoethanol for 1 h at RT. Subsequently, proteins in the gel were denatured completely in 6 M guanidine HC1 for 1 h at RT. The denatured proteins were renatured slowly by incubating the gel in 50 mM Tris, pH 8.0, 5 mM P-mercaptoethanol, and 0.04% Tween 40 for 16 h at 4 "C. The buffer was changed five times during this time. The renatured gel was preincubated in 50 mM HEPES pH 7.4, 2 mM dithiothreitol, and 10 mM MgCL for 30 min at RT. The kinase assay was done by adding kinase buffer containing 50 mM HEPES pH 7.4, 2 mM dithiothreitol, 10 mM MgCl,, 100 p M ATP, and 10 pCi/ml [T-~'P]ATP for 1 h at RT. After phosphorylation, the gel was washed extensively in 7% acetic acid, with at least five 500-ml changes.

RESULTS
Synchronization of the Cell Cycle-The synthesis of DNA assayed as [nethyl-3H]thymidine uptake began at 6-8 h after the M phase, and the maximum level was attained at 12-14 h (Fig. 1). Therefore, the S phase is thought to be 6-12 h from M phase arrest. At the point when nocodazole-arrested cells were collected, immunoprecipitated cdc2 kinase activity was very high, and this point was indicated as time 0. 16-18 h after release from M phase arrest by nocodazole, histone H1 kinase activity of immunoprecipitated cdc2 kinase was reactivated. This second peak of histone H1 kinase activity ranged from 14 to 20 h with its maximum at 16 h (Fig. 2B). Therefore, the second M phase is around 16 h from the release after M phase arrest.
Renaturation and Gel Phosphorylation Assay-When crude lysate was used, several bands were detected. Only the 44and 42-kDa bands seemed to change their intensity during the cell cycle, apparently in a coordinated fashion. The peaks of kinase activities of the 44-and 42-kDa bands were detected at about 2-6 h and 14-20 h after release from the M phase ( Fig. 2 A , upper panel). The second peak was more apparent, and the first peak was variable in several experiments. When samples were immunoprecipitated with antiserum aY91, the

R, cells remaining on the dish.
result was almost the same although kinase activities a t time 0 were not detectable ( Fig. 2A, lower panel). Since the band at time 0 in the crude lysate gel showed slightly faster mobility in duplicated experiments, this band may represent a distinct kinase. These results suggest that the 44-and 42-kDa kinase activities in crude lysate are the polypeptides recognized by this antiserum.

Immune Complex Kinase Assay-MAPKActivity in Comparison with cdc2
Kinase Activity-The histone H1 kinase activity of immune complex precipitated with anti-MAPK Cterminal antiserum (aC92) was assayed at time 0 to check the contaminating cdc2 kinase activity. Although cdc2 kinase activity was very high at time 0, histone H1 kinase activity of the immune complex by anti-MAPK (aC92) was negligible (Fig. 2B), indicating that contamination with cdc2 kinase is very little, if any. Since aC92 precipitates the 44 kDa predominantly, the kinase activity detected in this assay is that of the 44-kDa MAPK. The MAPK activity had two broad peaks during the cell cycle; the first peak appeared a t 2-6 h after release from the M phase, and the second peak appeared at 14-20 h after release (Fig. 2B). A summary of three independent experiments has revealed a biphasic activation of MAPK during the cell cycle (Fig.  2C). When cells were cellected after release from S phase arrest, a peak of cdc2 kinase activity appeared at 5-7 h after release, and the MAPK activity seemed to be activated slightly earlier (data not shown). In contrast to cdc2 kinase, which declined rapidly to a negligible level after release, MAPK activity was detected through the cell cycle. The only exception was at time 0 when the activity of MAPK was at the least level.
It seems paradoxical that MAPK activities are very low in M phase-arrested cells and activated at around the second M phase. This phenomenon can be interpreted in two ways. By one hypothesis, MAPK is activated in the late GP phase or in the early part of the M phase and then rapidly deactivated at the metaphase/anaphase transition of the M phase, at which point nocodazole-treated cells are arrested. An alternative hypothesis is that nocodazole per se interferes with MAPK. To clarify which is the case, M phase cells were collected by mechanical shakeoff from CHO cells in the second M phase without arresting cells with nocodazole. After centrifugation, collected cells were solubilized as above and divided into two parts for immunoprecipitation with either antiserum aC92 or antiserum against cdc2 kinase (aY21). Immunoprecipitated cdc2 kinase exhibited high histone H1 kinase activity; however, MAPK activity was very low (Fig. 20). Since no nocodazole treatment was done in this experiment, the low activity of MAPK in nocodazole-arrested cells is not an artifact caused by the reagent.
Amount of MAPKs during the Cell Cycle-The amount of MAPK during the cell cycle was determined by Western blotting (Fig. 3). The protein concentrations of the cell lysates from each time points were adjusted and blotted with antiserum aC92 or aY91. No significant change in the intensity of the bands was observed during the cell cycle. Phosphorylation State of MAPK in the Cell Cycle-The phosphorylation state of the MAPKs was analyzed by immunoprecipitation of MAPKs from cells labeled with 32P in vivo. By two-dimensional gel electrophoresis, both 44-and 42-kDa phosphoproteins were resolved into two close spots (Fig. 4). The acidic spot contains phosphotyrosine and phosphothreonine, and the basic spot contains phosphotyrosine only (29). All four spots of MAPKs were detected in the G1, S , and GZ/M phases. Only in nocodazole-arrested cells, phopshorylation of MAPKs was not detected (Fig. 4). Phosphopeptide mapping of the 44-and 42-kDa protein revealed that the phosphorylation pattern of these proteins in the M phase was very similar to that of the serum-stimulated proteins (data not shown).

DISCUSSION
In this study, two MAPK activities of 44 and 42 kDa were shown to be activated coordinatedly during the cell cycle in the gel phosphorylation assay. Although there are denatura-tion and renaturation procedures, this assay is at least semiquantitative (30). The same pattern of activation was confirmed by an immune complex kinase assay at least for the 44-kDa MAPK. The two MAPKs were activated biphasically; the first peak was 2-6 h from release from the M phase, and the second was 14-20 h. Compared with the pattern of thymidine uptake and histone H1 kinase activity of cdc2 kinase, the first peak was in the G1 phase, and the second peak was around the M phase. Although activities of MAPK were more apparent in the second peak, the level was lower compared with that of growth factor-stimulated MAPKs. The same pattern of activation was observed in the immune complex kinase assay. In contrast to cdc2 kinase whose activity is almost negligible during the G1 to Gz phase, MAPK activities of both 44-and 42-kDa proteins were clearly detected in the G1 and S phases, although the levels were lower compared with those in the G2 to M phase. In vivo labeling and twodimensional electrophoresis revealed that phosphothreonine and phosphotyrosine are contained in both 44-and 42-kDa proteins throughout the cell cycle except for time 0 when the cells are arrested at metaphase by nocodazole treatment. Since phosphorylation of both threonine and tyrosine residues is believed to be required for the activation of MAPK, this result is consistent with the presence of activities of the kinases in the cell cycle except for time 0. In Xenopus oocytes, MAPKs are phosphorylated in meiosis (16)(17)(18)(19). However, in nocodazole-arrested mammalian cells phosphorylation of the 42-kDa protein is not detectable (17). Our results are consistent with those results since the content of phosphotyrosine was least in the nocodazole-arrested cells. This phenomenon is not the result of a toxic effect of nocodazole because a similar result was obtained when metaphase cells were collected without nocodazole treatment. However, at 16 h after release from M phase arrest, clear phosphorylation of the tyrosine residue(s) is detectable. This result indicates that MAPKs are tyrosine-phosphorylated and activated at the G2/M transition or in the early part of the M phase. It seems likely that the mechanism for activation of MAPKs is almost the same in growth factor stimulation and in the M phase. In Xenopus oocytes, the time course of activation of histone H1 kinase (cdc2 kinase) and myelin basic protein kinase show periodic changes in a similar pattern, and the activation of cdc2 kinase appeared to precede the activation of myelin basic protein kinase (19). It is suggested from these data that MAPK is under control of cdc2 kinase. In our result, low but distinct MAPK activities are detected in the G1 and S phases, when cdc2 kinase activity is almost negligible. This result suggests that there could be an alternative mechanismb) that controls activities of MAPKs during the cell cycle.
In cdc2 kinase, specific sites are phosphorylated in a cell cycle-dependent manner. It may be related to the complex formation of cdc2 with p13 and cyclinB, which is the prerequisite for the activation of cdc2 kinase in the M phase. In contrast to cdc2 kinase, MAPKs seems to have only two states: activated and deactivated. It seems that only the fractions of activated MAPKs to total MAPKs are changing in the cell cycle, and the MAPKs may be controlled under the dynamic balance of activator(s) and phosphatase(s). Activation of MAPK by okadaic acid (4) supports this hypothesis. Recently, Posada and Cooper (31) showed that kinase(s) that phosphorylate MAPK is (are) activated in the M phase of Xenopus oocytes. The identity of the activator(s) in the M phase remains to be clarified.
Recently MAPK in the Xenopus oocyte was reported to be activated in the M phase (18, 19) and to control microtubule reorganization in the cell cycle through phosphorylation of the m~c r o t u~u~e -a s s o c~a t e~ protein (19). M A P K~ in the fibro-5. Gotoh, y., Nishida, E., Yamashita, T., Hoshi, M., Kawakami, M., and blasts of the mammals are activated when quiescent cells are in the G, to G~ transition (.,-9). this study, M A P K~ were 8. Ahn, N. G., S e w , R., Bratlien, R. L., D i k c., Tanks, N. K., and Krebs, shown to be activated not only in the Go/GI transition but 9. Ahn, N. G., and Krebs, E. G. (1990)