Cdc2-Cyclin B Phosphorylates p70 S6 Kinase on Ser411at Mitosis*

The carboxyl terminus of p70 S6 kinase (p70s6k) has a set of Ser and Thr residues (Ser411, Ser418, Ser424, and Thr421) phosphorylated in vivo by an unidentified kinase(s). These Ser/Thr sites are immediately followed by proline, a motif that is commonly seen in the substrates of cyclin-dependent kinases (Cdk) and mitogen-activated protein kinases. A previous study has shown that Cdc2 (Cdk1) indeed phosphorylates these p70s6k Ser/Thr residues in vitro. Here, we demonstrate that Cdc2-cyclin B complex phosphorylates Ser411 in the KIRSPRR sequence, whereas other Cdk-cyclin complexes including those containing Cdk2, Cdk4, or Cdk6 do not. Additionally, Ser411 phosphorylationin vivo was increased at mitosis in parallel with Cdc2 activation, and it was suppressed by a dominant negative form of Cdc2. These data indicate that p70s6k is a physiological substrate of Cdc2-cyclin B in mitosis. Since the activity of p70s6k is low during mitosis, Cdc2-cyclin B may play a role in inactivating p70s6k during mitosis, where protein synthesis is suppressed.

The p70 S6 kinase (p70 s6k ) is a Ser/Thr kinase that is activated by mitogenic stimulation at the G 0 /G 1 transition of the cell cycle in mammalian cells (1)(2)(3). It phosphorylates five Ser residues in ribosomal protein S6 in vitro (4) and is the major S6 kinase in vivo in mammalian cells (5,6). The role of the p70 s6k activation pathway in cell proliferation is not fully understood; the pathway may regulate translation of mRNAs encoding ribosomal proteins that have a conserved sequence at their 5Ј-end (pyrimidine tract followed by GC-rich regions) (7)(8)(9). Thus, the p70 s6k pathway may up-regulate ribosome biogenesis, which is critical for quiescent cells to enter and proceed through the cell division cycle (10). Indeed, the microinjection of quiescent rat fibroblasts with polyclonal antibodies against p70 s6k abolished serum-induced entry into S phase of the cell cycle (11). In addition, inhibition of the kinase pathway by rapamycin either prolongs G 1 progression or blocks the G 1 /S transition of the cycle (10,12). p70 s6k has two distinct sets of Ser/Thr residues phosphorylated in vivo. One set includes Thr 229 , Thr 389 , and Ser 404 (13,14), and the other set includes Ser 411 , Ser 418 , Ser 424 , and Thr 421 at the carboxyl terminus (15). Phosphorylation of the former set, especially Thr 229 and Thr 389 , plays a critical role in activation of the kinase, since replacement of these Thr residues by Ala abrogates kinase activity (14,16,17). Additionally, phosphorylation of these Thr and Ser residues is sensitive to rapamycin, which inactivates the kinase indirectly (13), and it is believed that phosphatidylinositol 3-kinase and the structurally related enzyme, mTOR (also termed FRAP or RAFT), are involved in the regulation of phosphorylation/dephosphorylation of these sites (16,18). Recently, it was shown that Akt (also termed protein kinase B) is involved in activation of p70 s6k by phosphatidylinositol 3-kinase (19), but direct regulators of p70 s6k are still not elucidated. In contrast to these phosphorylation sites, the role of phosphorylation at the carboxyl terminus (Ser 411 , Ser 418 , Ser 424 , Thr 421 ) is still obscure. The collection of studies examining mutations in these residues revealed that these phosphorylation sites are not essential for rapamycin-sensitive regulation of the kinase activity (20,21). Of interest, all of these Ser and Thr are within a conserved sequence (Ser/Pro or Thr/Pro) seen in substrates of cyclin-dependent kinase (Cdk) 1 families or mitogen-activated protein kinase families (22). A previous study demonstrated that Cdc2 phosphorylated these Ser/Thr residues in vitro when purified p70 s6k was used as a substrate, while the effect of mitogenactivated protein kinase was less clear (23). Indeed, of these Ser/Thr phosphorylation sites, Ser 411 is within a perfect consensus sequence (K/R-S-P-R/P-R/K/H) as a substrate for Cdc2 (24). However, Cdc2, which associates with either cyclin A or cyclin B, is activated exclusively in S to G 2 /M phases of the cell cycle (25), whereas the identified regulations of p70 s6k activity occur in earlier phases of the cell cycle (G 0 /G 1 to G 1 /S), as described above. Therefore, it was postulated that a Cdc2 homolog that is activated at an earlier phase of the cell cycle, such as Cdk2, Cdk4, or Cdk6, might phosphorylate p70 s6k in vivo, or alternatively that phosphorylation of p70 s6k by Cdc2 was solely an in vitro event. Here, we demonstrate that Cdc2-cyclin B has the highest activity among Cdk family members for phosphorylation of the carboxyl terminus of p70 s6k in vitro, and moreover Cdc2 phosphorylates the carboxyl terminus of p70 s6k during mitosis in vivo.

EXPERIMENTAL PROCEDURES
Reagents-Phorbol 12,13-dibutyrate (PDBu, Sigma) and calcium ionophore, ionomycin (Calbiochem, San Diego, CA) were dissolved in dimethyl sulfoxide (Me 2 SO). Rapamycin was obtained from the Drug Synthesis and Chemistry Branch, NCI, and dissolved in ethanol to give a 1 mg/ml stock solution. Aphidicolin (Sigma) was prepared as a 5 mg/ml stock solution in ethanol. Nocodazole (methyl-[5-(2-thienylcarbonyl)-1H-benzimidazol-2-yl]carbamate, Sigma) was prepared as a 1 mg/ml stock solution in Me 2 SO. Antibodies used in the study were as * This work was supported in part by Grants CA-64685 (to N. T.), DK-28353 (to J. M.), and GM-26743 (to J. M.) from the National Institutes of Health. 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.
Preparation of Recombinant p70 s6k -Recombinant p70 s6k was made using a baculovirus expression system (BAC-TO-BAC baculovirus expression system, Life Technologies, Inc.). Briefly, HA-tagged wild type-p70 s6k cDNA fragments prepared previously (17), were introduced into pFASTBAcHTb donor plasmid using the unique SalI site of the vector. Then, the recombinant vector was transformed into DH10BAC competent cells, which contain the bacmid with a mini-attTn7 target site and the helper plasmid for efficient transposition. The p70 s6k recombinant bacmids identified by disruption of the lacZ␣ gene, were then transfected into Sf9 cells. Similarly, the recombinant bacmid containing HA tagged-mutant p70 s6k (T229A or ⌬CT104) was produced and transfected into Sf9 cells. The recombinant p70 s6k was recovered by a metal affinity resin (TALON, CLONTECH, San Diego, CA), and confirmed in immunoblot analysis by reactivity with several different antibodies raised against p70 s6k peptides. The recombinant wild type p70 s6k (recWT-p70) had a high kinase activity for phosphorylation of ribosomal S6 peptide as described previously (29). In contrast, recT229A-p70 had little or no activity, and rec⌬CT-p70 had approximately one-fifth of the activity compared with the wild type. The recombinant viruses were amplified by serial infection and the supernatants were used for further expression of the proteins.
In Vitro Phosphorylation of Recombinant p70 s6k -Recombinant p70 s6k (approximately 100 ng of the protein) purified by TALON beads were washed once with a kinase buffer (50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol), and the beads were suspended in 50 l of the kinase buffer containing 100 M unlabeled ATP, 200 Ci/ml [␥-32 P]ATP, and 1 l of a purified Xenopus Cdc2-cyclin B or Cdk2-cyclin E complex (activities: ϳ1 pmol/l/min to transfer phosphate to whole histone H1 peptide). The reaction was allowed to proceed for 30 min at 30°C. As a control reaction, histone H1 (1 g) was added in 50 l of reaction mixture instead of rec-p70 resin. The reaction was terminated by adding a protein loading buffer. After boiling for 5 min, labeled proteins were separated by 10% SDS-polyacrylamide gel.
Phosphopeptide Mapping-Recombinant p70 s6k was cut out from SDS-polyacrylamide gel, and homogenized in 50 mM ammonium bicarbonate (pH 7.4) containing 1% SDS and ␤-mercaptoethanol. Extracted proteins were precipitated by trichloroacetic acid in the presence of 10 g of a carrier protein (RNase), washed with ethanol/ethyl ether (1:1), and treated with hydrogen peroxide/formic acid (1:10) solution. After drying, proteins were subjected to serial chymotrypsin and trypsin digestion. Samples were then dissolved in an electrophoresis buffer (2.25% formic acid, 7.75% acetic acid), loaded to TLC-cellulose plates (Merck5716 cellulose, 20 ϫ 20 cm, EM Science), and separated by electrophoresis at 1200 V for 30 min. The peptides were further separated by ascending chromatography for 16 h using a buffer containing 65% isobutyric acid, 5% pyridine, 3% acetic acid, and 2% butanol. The phosphopeptides were visualized by autoradiography. In some experiments, the Ser 411 peptide phosphorylated in vitro was purified using C-18 Sep-Pak cartridge (Millipore, Milford, MA), digested by chymotrypsin/trypsin and subjected to thin layer electrophoresis/chromatography as described above.
In Vitro Kinase Assay-Specific activities of Cdk-cyclin complexes were determined by 32 P incorporation into small peptides. For immune complex assay, cells (5 ϫ 10 6 ) were washed with phosphate-buffered saline and lysed at 4°C in 1 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 25 mM NaCl, 40 g/ml PMSF, 50 g/ml aprotinin, 50 M leupeptin, and 0.1% Nonidet P-40). For Cdk4 and Cdk6 activity, cells were lysed in a buffer containing 50 mM HEPES, pH 7.5, 1 mM EDTA, 2.5 mM EGTA, 150 mM NaCl, 1 mM dithiothreitol, 40 g/ml PMSF, 50 g/ml aprotinin, 50 M leupeptin, 100 M sodium orthovanadate, 10 mM ␤-glycerophosphate, and 0.1% (v/v) Tween 20. The extract (500 l) was incubated for 1 h at 4°C with anti-Cdc2, Cdk2, Cdk4, Cdk6, cyclin A, or cyclin B antibody. The immune complex was absorbed to Protein Gcoupled Sepharose beads for 1 h. Alternatively, the extract was incubated for 1 h at 4°C directly with p13 suc -agarose beads. The beads were washed three times with the lysis buffer, and once with kinase buffer (50 mM HEPES, pH 7.2, 10 mM MgCl 2 , 5 mM MnCl 2 , 1 mM dithiothreitol for Cdk4 and Cdk6 assay; 50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol for the rest). Following the final wash, the immune complexes were suspended in 50 l of the kinase buffer containing 100 M unlabeled ATP, 200 Ci/ml [␥-32 P]ATP, 10 Ϫ4 M peptide. The reaction was allowed to proceed for 15 min at 30°C and terminated by the addition of 10 l of 500 mM EDTA. Following a brief centrifugation, the supernatant (20 l) was applied to phosphocellulose paper and radioactivity was determined using a liquid scintillation counter. For the purified kinase complexes, Xenopus Cdc2-cyclin B complex or Xenopus Cdk2-cyclin E complex were used instead of immune complexes. Kinetic constants were determined using 10 Ϫ8 to 10 Ϫ3 M peptide. The activity of p70 s6k was measured as described previously (12).
Immunofluorescent Staining of Cells-Cells were plated and cultured on cover glass (0.1 mm thickness) in regular culture medium in CO 2incubator at 37°C for 16 h, fixed with 3% paraformaldehyde, permeabilized with 0.2% Triton X-100, and incubated at room temperature in a blocking buffer (Dulbecco's modified Eagle's medium and 10% fetal calf serum) for 30 min. Cells were then incubated for 1 h in blocking buffer containing 0.05 g/ml anti-phospho-Ser 411 antibody. After washing the cells with phosphate-buffered saline six times, cells were incubated for 1 h in blocking solution containing 0.05 g/ml secondary antibody (Cy3-conjugated donkey anti-rabbit IgG, Jackson ImmunoResearch, West Grove, PA) and with 10 g/ml Hoechst 33258 dye (Calbiochem, La Jolla, CA). After washing six times with phosphate-buffered saline, 100 l of a mounting solution containing 2 mg/ml O-phenylen-diamine-diHCl (Sigma) and 90% glycerol was applied to the glass slide and covered. After sealing the cover glass with clear nail polish, the samples were examined under fluorescence microscopy. Images of fluorescence staining were taken using Ektachrome Elite 400 (Eastman Kodak Co.).

Cdc2-Cyclin B but Not Cdk2-Cyclin E Phosphorylates
Recombinant p70 s6k on Ser 411 in Vitro-Although Cdc2 was demonstrated to phosphorylate the carboxyl terminus of p70 s6k in vitro in a previous study (23), the specificity for Cdc2 and the physiological relevance of the event has not been clarified. In an attempt to explore the specificity of the event among Cdk family members, we initially compared Cdk2 and Cdc2 in their abilities to phosphorylate recombinant p70 s6k . Cdk2 is the closest kinase to Cdc2 by homology in amino acid sequence and also in terms of substrate specificity (31). Cdk2 also has established roles in the G 1 phases of the cell cycle, when major regulation of p70 s6k is known to occur. Xenopus Cdc2-cyclin B and Cdk2cyclin E complexes were purified as described previously (28). For substrates, either wild type p70 s6k (recWT-p70), the kinase-dead T229A mutant (recT229A-p70) or a deletion mutant lacking 104 amino acids at the carboxyl terminus (rec⌬CT-p70) was prepared using the Sf9 cell baculovirus system. The recombinant proteins tagged by six repetitive histidine residues were then purified using nickel metal affinity resin. Here, we compared the ability of the Cdc2 and Cdk2 complexes to phosphorylate these recombinant substrates using histone H1 as a control. As demonstrated in Fig. 1, both Cdc2-cyclin B and Cdk2-cyclin E complexes phosphorylated histone H1 similarly in the conditions used. The Cdc2-cyclin B complex increased the phosphorylation of recT229A-p70 but not rec⌬CT-p70, which lacks the Cdk consensus sequence sites. In contrast to Cdc2-cyclin B, the Cdk2-cyclin E complex caused little increase in the phosphorylation of these recombinant p70 s6k proteins. The recT229A-p70 was weakly phosphorylated without addition of Cdk2-cyclin E or Cdc2-cyclin B (Control in the middle panel of Fig. 1). It is unknown whether this comes from autophosphorylation by a residual activity of the mutant kinase, or from an unidentified insect kinase in the preparation of nickel resin. The recWT-p70 was highly phosphorylated by itself, which was likely due to autophosphorylation of the kinase, but Cdc2-cyclin B complex still increased the phosphorylation of recWT-p70 by about 3-fold (data not shown). It should also be noted that there was an unidentified phosphorylated band below recT229A-p70 (Fig. 1, middle panel), which had a similar mobility to rec⌬CT-p70.
The in vitro phosphorylated recT229A-p70 was then cut out from the one dimensional SDS-polyacrylamide gel, digested with trypsin and chymotrypsin, and subjected to two-dimensional peptide electrophoresis/chromatography. As shown in Fig. 2 (A-C), Cdc2-cyclin B but not Cdk2-cyclin E increased the phosphorylation of several new spots in the peptide map. Based on the comparison of in vivo phosphorylation of WT-p70 and ⌬CT-p70 (Fig. 2, E and F), the three spots circled correspond to the phosphorylated peptides from the carboxyl-terminal phosphorylation sites (Ser 411 , Ser 418 , Thr 421 , Ser 424 ) of p70 s6k . In addition, previous studies demonstrated that spot A, B, and C correspond to the peptides including Ser 418 , Ser 424 , and Ser 411 , respectively (13,14). The spot C, corresponding to the peptide including Ser 411 , was the most highly phosphorylated by the Cdc2-cyclin B complex. To confirm this, a synthetic peptide corresponding to the sequence including Ser 411 (EPKIR- FIG. 1. Cdc2-cyclin B, but not Cdk2-cyclin E, phosphorylates a recombinant p70 s6k . Recombinant p70 s6k (recT229A-p70; Thr 229 is replaced by Ala, rec‚CT-p70; the carboxyl-terminal 109 amino acids are deleted) were prepared by a baculovirus protein expression system. In vitro kinase reaction by purified Xenopus Cdc2-cyclin B or Cdk2-cyclin E complex were performed using [␥-32 P]ATP. The radiolabeled substrates were separated by 10% SDS-polyacrylamide gels and visualized by autoradiography. The exposure times were 2 h and 1 h for histone H1 phosphorylation and recp70 phosphorylation, respectively.
FIG. 2. Cdc2-cyclin B, but not Cdk2-cyclin E, phosphorylates the carboxyl terminus of p70 s6k , especially at Ser 411 . A-C, the recT229A-p70 bands in the gel shown in Fig. 1 were cut out and subjected to trypsin/chymotrypsin digestion. The phosphopeptides were analyzed by two-dimensional thin layer electrophoresis/chromatography (TLE/TLC). D, a synthetic peptide (EPKIRSPRRFIG, the Ser 411 peptide) representing residues 406 -417 of human p70 s6k , was phosphorylated in vitro by Cdc2-cyclin B, digested with trypsin/chymotrypsin, and separated by TLE/TLC as described above. E and F, cells expressing recWT-p70 (E) or rec‚CT-p70 (F) were labeled with [ 32 P]orthophosphate for 3 h. The rec-p70 s6k proteins were recovered and subjected to the same TLE/TLC procedure as described above. SPRRFIG) was phosphorylated by Cdc2-cyclin B, digested with trypsin/chymotrypsin, and analyzed by phosphopeptide mapping. The digested peptide containing phospho-Ser 411 localized to the identical spot as the most highly phosphorylated peptide from recT229A-p70 (Fig. 2D). Co-migration of the peptides were confirmed by applying mixture of the phosphopeptides used in Fig. 2 (C and D) (data not shown). These data indicate that Cdc2-cyclin B complex phosphorylates p70 s6k in vitro at Ser 411 .
Cdc2-Cyclin B Phosphorylates the Ser 411 Peptide Better than Cdk2-Cyclin E-Consistent with our data that Cdc2 phosphorylates p70 s6k at Ser 411 , the sequence surrounding Ser 411 has the best match among the carboxyl-terminal phosphorylation sites in p70 s6k with the consensus phosphorylation site sequence for Cdc2 (K/R-S-P-R/P-R/K/H) determined by a peptide selection approach (24). Next, we examined whether the differential phosphorylation by Cdc2 and Cdk2 is due to differential amino acid sequences surrounding the phosphorylation sites in p70 s6k and histone H1. The activities of Cdc2 and Cdk2 to phosphorylate the Ser 411 peptide (EPKIRSPRRFIG) or the histone H1 peptide (AVAAKKSPKKAKKPA) were compared. Initially, we measured the relative phosphorylation rates of the Ser 411 peptide (approximately 10 Ϫ8 to 10 Ϫ3 M) by a purified Xenopus Cdc2-cyclin B and a Cdk2-cyclin E complex. As shown in Fig. 3, Cdc2-cyclin B phosphorylates the Ser 411 peptide in vitro with a K m of about 50 M, whereas Cdk2-cyclin E poorly phosphorylates the Ser 411 peptide (K m Ͼ500 M). This was in contrast to the fact that both kinases phosphorylate the histone H1 peptide similarly (K m : for Cdc2, ϳ10 M, for Cdk2, ϳ20 M) (data not shown).
Cdc2, but Not Cdk2, Cdk4, or Cdk6, Immunoprecipitated from Mammalian Cells, Phosphorylates the Ser 411 Peptide-The kinase activities of other Cdks for phosphorylation of the Ser 411 peptide were then examined using immunoprecipitated Cdks from mammalian cells. Using specific antibodies against Cdks or cyclins, various Cdk complexes were precipitated from proliferating human T lymphocytes. The phosphorylation of the Ser 411 peptide by the immune complexes was measured, and relative phosphorylation rates compared with phosphorylation of the histone H1 peptide or a truncated recombinant Rb protein are summarized in Table I and Fig. 4. As shown in Table I, the immune complexes purified using anti-Cdc2 or anti-cyclin B antibody phosphorylate the Ser 411 peptide well, but the complexes purified using anti-Cdk2 or anti-cyclin A antibody poorly phosphorylate the Ser 411 peptide. The p13 suc precipitates that contain both Cdc2 and Cdk2 also phosphorylate the Ser 411 peptide well. Cdc2 forms complexes with both cyclin B and cyclin A, and Cdk2 forms complexes with both cyclin E and cyclin A in the cells. Therefore, the data here indicate that the Cdc2-cyclin B complex, but neither Cdk2cyclin A nor Cdk2-cyclin E, phosphorylates the Ser 411 peptide well in vitro. The Cdc2-cyclin A complex probably does not phosphorylate the Ser 411 peptide well, but this is not clear because the ratio of Cdc2 and Cdk2 in the cyclin A immune complexes was not determined. Cdk4 and Cdk6 complexes with cyclin Ds are known to have a differential substrate specificity from Cdc2 or Cdk2 (32). In vitro, they will phosphorylate the Rb protein but not histone H1. As predicted, Cdk4 or Cdk6 immune complexes poorly phosphorylated the Ser 411 peptide when compared with their phosphorylation of Rb (Fig. 4). In addition, neither Cdk4 or Cdk6 immune complexes phosphorylated recT229A-p70 (data not shown).
The Phosphorylation of p70 s6k on Ser 411 in Vivo Increases at Late Phases of the Cell Cycle-The results above indicate that Cdc2 is the best kinase among Cdks we investigated for phosphorylation of the carboxyl terminus of p70 s6k in vitro. To further explore the relevance of the finding in vivo, we initially examined the phosphorylation status of p70 s6k using mitogenactivated primary human T lymphocytes. Primary T lymphocytes provide a population of cells that are about 99% in the G 0 status of the cell cycle (33). Cells were stimulated with PDBu and ionomycin in the presence or absence of rapamycin (10 ng/ml) and harvested at the indicated times (up to 48 h). As reported previously, the activity of p70 s6k kinase was induced within 3 h after mitogenic activation of cells (Fig. 5B), and it was paralleled by the mobility shift of the protein in the immunoblot (Fig. 5A, upper panel). This mobility shift of p70 s6k is known to correspond to phosphorylation of the kinase at Thr 389 and Thr 289 (13,14). Activation and the mobility shift of the kinase were inhibited by addition of rapamycin. The cell lysates were also immunoblotted using an antibody raised against a synthetic phospho-Ser 411 peptide corresponding to residues 406 -417 (EKIRSPRRFIGC) of p70 s6k (anti-phospho-Ser 411 antibody, New England Biolabs) (Fig. 5A, lower panel). p70 s6k detected by anti-phospho-Ser 411 antibody was increased only at late phases of the cell cycle (Fig. 5A), where Cdc2 activity was high (Fig. 5C) and cells were beginning to enter S and G 2 /M phases (Fig. 5D). In contrast, total amount of p70 s6k protein detected by anti-carboxyl terminus antibody, increased   FIG. 3. Cdc2-  earlier (within 24 h) (Fig. 5A, upper panel). Anti-phospho-Ser 411 antibody-reacted species were not detectable within 3 h even when 3-fold amounts of samples were applied (data not shown). In a previous study, we have shown that rapamycin prolonged G 1 phase and delayed resting T cells from entering the S phase by about 9 h when T cells were stimulated with PDBu and ionomycin (10). Of interest, Ser 411 phosphorylation was seen in late cell cycle phases in rapamycin-treated samples as well. These results indicate that the in vivo phosphorylation of p70 s6k on Ser 411 is increased at late phases of the cell cycle, regardless of the activity of p70 s6k . To confirm the efficacy of the anti-phospho-Ser 411 antibody, COS7 cells were transfected with either wild type or S411A mutant p70 s6k expression vector and the transfected p70 s6k proteins were subjected to the same immunoblotting analysis. The antibody reactivity was markedly reduced with the S411A mutant as compared with wild type (data not shown). Nocodazole Treatment Enhances the Phosphorylation of p70 s6k on Ser 411 -In order to further investigate where in the cell cycle Ser 411 phosphorylation is enhanced, proliferating cells (COS7 cells) were treated for 16 h with aphidicolin (to arrest the cycle at the early S phase) or nocodazole (to arrest the cycle at the mitotic phase). Cell lysates were immunoblotted with either the anti-carboxyl terminus antibody or the anti-phospho-Ser 411 antibody as described above. As shown in Fig. 6A, the phospho-Ser 411 -containing species of p70 s6k was increased by addition of nocodazole but not by aphidicolin. Neither of the drugs alters the expression level of p70 s6k , confirmed by immunoblotting with the anti-carboxyl terminus antibody. In addition, effects of release of cells from nocodazoleblock were examined by washing out nocodazole and leaving cells in the regular medium for 3 h. As shown in Fig. 6A (right  panels), release from a nocodazole-block decreased Ser 411 phosphorylation. Nocodazole treatment also enhanced Cdc2 activity, due to arrest of the cell cycle at metaphase, when Cdc2 activity was high (Fig. 6B). Release from nocodazole-block decreased Cdc2 activity (Fig. 6B). These data indicate that the phosphorylation of p70 s6k on Ser 411 is enhanced during the metaphase block induced by nocodazole where Cdc2 activity is high. In addition, the p70 s6k activity was measured in the system. Of interest, the p70 s6k activity was low in nocodazoletreated cells and increased by release from the drug (Fig. 6B).
The p70 s6k with Phosphorylated Ser 411 Is Detected Only in Mitotic Cells-In order to further investigate the phosphorylation status of p70 s6k on Ser 411 , NIH3T3 cells in normal growing culture were examined by indirect immunofluorescence microscopy using the anti-phospho-Ser 411 antibody as a primary antibody and a Cy3-conjugated secondary antibody. The cells were simultaneously stained with Hoechst 33258 dye to visualize DNA and nuclei. As shown in Fig. 7, cells in the mitotic phase were stained with anti-phospho-Ser 411 antibody. Especially, cells in prophase and metaphase of mitosis were highly stained, and cells in telophase were weakly stained. In contrast, cells not in mitotic phase were not stained at all with the antibody. Cells treated with nocodazole were also highly stained with the antibody (data not shown). These data suggest that p70 s6k is phosphorylated on Ser 411 exclusively in mitosis.
A Dominant Negative Mutant of Cdc2 Suppresses the Phosphorylation of p70 s6k on Ser 411 -In order to investigate further whether Cdc2 is essential for the phosphorylation of p70 s6k on Ser 411 in vivo, a dominant negative form of Cdc2 (D146N) was utilized. This form of Cdc2 has been demonstrated to inhibit endogenous Cdc2 activity in vivo (30). COS7 cells were cotransfected with HA-tagged wild type p70 s6k expression vector and either wild type or D146N mutant Cdc2 expression vector. Cells were pulse labeled with [ 32 P]orthophosphate at 21-24 h after transfection, and transfected p70 s6k was recovered by anti-HA antibody. The ratio of plasmids was 20:1 (Cdc2 vector: p70 s6k vector), and it was confirmed that most (Ͼ80%) of the p70 s6k -overexpressing cells were also overexpressing Cdc2 by immunofluorescence staining (data not shown). Fig. 8A illustrates that D146N Cdc2 decreases total phosphorylation of p70 s6k compared with the control (wild type Cdc2). Immunoblotting using the whole lysates demonstrates that D146N Cdc2 decreases Ser 411 phosphorylation but not total p70 s6k protein amount (Fig. 8, B and C). These data indicate that Cdc2 phosphorylates p70 s6k on Ser 411 in vivo. DISCUSSION Here, we have demonstrated that Cdc2, but not Cdk2, Cdk4, or Cdk6, phosphorylates p70 s6k at Ser 411 in vitro. Additionally, p70 s6k is phosphorylated at Ser 411 in vivo at mitosis, when Cdc2 is highly activated. Finally, a dominant negative mutant of Cdc2 suppressed the carboxyl-terminal phosphorylation of p70 s6k . Based on these findings, we conclude that Cdc2 is an important physiological kinase for the phosphorylation of p70 s6k at Ser 411 , and p70 s6k is a relevant mitotic substrate for Cdc2. However, the study does not totally exclude the involvement of other kinases for phosphorylation of p70 s6k at Ser 411 . For example, some Ser 411 phosphorylation was detected in cells treated with PDBu/ionomycin and rapamycin for 39 h, where there was little substantial increase in Cdc2 activity (Fig. 5, A  and C). This may suggest that another kinase in late cell cycle phases also takes part in this event.
Although previous reports have claimed that the carboxylterminal phosphorylation of p70 s6k including Ser 411 residues FIG. 4. Cdc2, but not Cdk4 or Cdk6, phosphorylates the Ser 411 peptide. A and B, phosphorylation rates of the Ser 411 peptide by Cdc2, Cdk4 or Cdk6 immune complex, immunopurified from proliferating human T lymphocytes, were analyzed (A). A truncated Rb protein was used as a control substrate for the kinases (B). occurs soon after mitogenic stimulation (13,15,23), those studies were done using serum-starved cell lines that are not completely in resting (G 0 ) status. In fact, Mukhopadhyay et al. demonstrated insulin-induced Cdc2 activity in serum-starved H4 hepatoma cells within 30 min, in parallel to an increase in the activity to phosphorylate the carboxyl terminus of p70 s6k in vitro by the cell lysates (23). Taken together with the fact that Cdc2 is active exclusively in late cell cycle phases (25), the increase in Ser 411 phosphorylation soon after mitogenic stimulation in these cell lines may not represent an event occurring in early G 1 phase. It may come from a cell fraction already in late phases of the cell cycle. In contrast, primary T lymphocytes provide a population of cells that are about 99% in the G 0 phase of the cell cycle (33). In mitogen activated primary T cells, neither Cdc2 activity nor cells containing greater than 2 N DNA is detectable within 30 h after mitogenic stimulation (33,34).
Cdc2 but not Cdk2, Cdk4, or Cdk6, phosphorylates p70 s6k at the physiological phosphorylation sites on the carboxyl terminus, and the Ser 411 peptide as well. The site that Cdc2 phosphorylates especially well (Ser 411 in p70 s6k ) has a consensus motif for Cdc2/Cdk2 substrates (S/T-P-X-K/R) (31). The differ-ential substrate specificity between Cdc2/Cdk2 and Cdk4/Cdk6 is well known; Cdc2/Cdk2 phosphorylates histone H1 and Rb well in vitro, but Cdk4/Cdk6 prefers Rb as an in vitro substrate. Among the numerous phosphorylation sites in Rb protein, FIG. 5. The in vivo phosphorylation of p70 s6k at Ser 411 was increased at the late phases of the cell cycle. A, human primary T cells were stimulated with PDBu and ionomycin in the presence or absence of rapamycin (10 ng/ml) for the indicated times. p70 s6k was immunoblotted using either an antibody raised against an epitope corresponding to amino acids 485-502 of p70 s6k (anti-carboxyl terminus antibody) (upper panel), or an antibody raised against a synthetic phospho-Ser 411 peptide corresponding to amino acids 406 -417 (EKIR-SPRRFIGC) of p70 s6k (anti-phospho-Ser 411 antibody) (lower panel). B and C, the activity of p70 s6k and Cdc2 of the cells harvested at the indicated time was measured as described under "Experimental Procedures." D, cell cycle progression was determined by staining cells with propidium iodide and subsequent fluorescence-activated cell sorting analysis.
FIG. 6. Nocodazole treatment enhanced the phosphorylation of p70 s6k at Ser 411 . A, COS7 cells were treated with vehicle alone (0.1% ethanol, control), aphidicolin, or nocodazole for 16 h. In the right panels, COS7 cells were treated with nocodazole (for 16 h) and cells in metaphase block were collected by shaking off the plate (nocodazole). Cells were washed three times with the drug-free medium, and then reseeded in the drug-free medium for 3 h (release). Cells harvested were lysed, and the p70 s6k protein was immunoblotted using anti-phospho-Ser 411 antibody (lower panel) or anti-carboxyl terminus antibody (upper panel). B, COS7 cells were treated with nocodazole (for 16 h) and cells in metaphase block were collected by shaking off the plate (nocodazole). Cells were washed three times with the drug-free medium, and then reseeded in the drug-free medium for 3 h (release). The activity of Cdc2 and p70 s6k of the cells was measured as described under "Experimental Procedures." The activity of the kinases in the same number of cells growing in non-synchronized fashion (1 day after passage) is also shown (growing). Cdk4 phosphorylates Ser at the 779 LSPIP sequence, but Cdk2 does not (32). Therefore, it is not surprising that Cdk4/Cdk6 did not phosphorylate the p70 s6k sequence whereas Cdc2 does. The clear difference between Cdc2 and Cdk2 in their abilities to phosphorylate the carboxyl terminus of p70 s6k was more surprising. Since a similar difference was seen when the Ser 411 peptide was used, this discrepancy is likely due to the differential preferences of Cdc2 and Cdk2 for the amino acid sequence surrounding Ser 411 . When compared with the ability to phosphorylate histone H1-derived sequence (KSPKK), the rate for p70 s6k peptide (RSPRR) phosphorylation by Cdc2 was 45-60%, but only ϳ6% for Cdk2. Of interest, a recent study analyzed in detail the substrate specificity for Cdc2 and Cdk2 (35). According to the study, KSPRK is phosphorylated equally by Cdc2-cyclin B, Cdc2-cyclin A, Cdk2-cyclin A, and Cdk2-cyclin E. However, when the last lysine in the sequence is changed to arginine, Cdk2-cyclin A or E barely phosphorylate the peptide (ϳ5%), whereas Cdc2-cyclin B and Cdc2-cyclin A maintain about 60% and 20% of the activity to phosphorylate the original peptide, respectively. Given the similarity in the differential activities observed here, Arg 414 (RSPRR) in p70 s6k is most likely responsible for the preferred phosphorylation of p70 s6k by Cdc2-cyclin B.
The carboxyl-terminal phosphorylation site in p70 s6k was originally proposed to be an autoinhibitory site for the kinase (22). The sequence at the phosphorylation site has homology with that seen in its physiological substrate S6. Moreover, the peptide derived from the carboxyl-terminal phosphorylation site can block the kinase activity of p70 s6k to phosphorylate S6 in vitro (36). Therefore, it was proposed that the phosphorylation of the carboxyl terminus will release this autoinhibitory effect, thus regulating the activity of p70 s6k . However, this model has not been fully supported by later experimental re-sults. First, the carboxyl-terminal phosphorylation is in fact not essential for the major regulation of activity of p70 s6k . A mutant in which all four serine/threonine residue at the carboxyl-terminal site are replaced by alanine maintains the ability to phosphorylate ribosomal S6 protein. Moreover, the mutant was activated by growth factor stimulation and inactivated by rapamycin as well as wild type p70 s6k (20). In addition, another mutant of p70 s6k with a deleted carboxyl terminus has weaker activity in phosphorylating S6, and is also positively and negatively regulated by growth factors and rapamycin, respectively (21). Here, we demonstrated that the carboxyl-terminal phosphorylation occurs at late phases of the cell cycle, whereas activation of the kinase was induced rapidly after mitogenic stimulation. Taken together, these results suggest that the carboxyl-terminal phosphorylation may not be a factor involved in activation of the kinase.
The Cdc2-cyclin B complex plays a central role in mitosis. It phosphorylates, for instance, nuclear lamins thus leading to their disassembly, an important event in the initiation of nuclear envelope breakdown (37)(38)(39). The complex also phosphorylates histone H1, which is thought to promote chromosome condensation, an event that occurs at the onset of mitosis as well (40). In addition, nucleolin, RNA polymerase II, p53, elongation factors 1␤ and 1␥, Src, and Cdc25-C serve as substrates for the complex (in vivo and/or in vitro) (41). Here, p70 s6k was demonstrated to be a new candidate for a physiological mitotic substrate of Cdc2-cyclin B. A recent study (42) has demonstrated, and we confirmed here that the activity of p70 s6k is low at mitotic phase, and release of cells from mitosis reactivates p70 s6k . Taken together with the proposed role of p70 s6k in protein synthesis, it is reasonable that p70 s6k is inactivated during mitosis where protein synthesis is generally suppressed (43,44). Cdc2-cyclin B may phosphorylate and inactivate p70 s6k at mitosis. However, treatment of recombinant p70 s6k (wild type) with purified Xenopus Cdc2-cyclin B did not suppress (nor increase) the activity of p70 s6k (data not shown), suggesting that the phosphorylation of Ser 411 by itself does not inactivate the kinase. A phosphoserine-binding protein may bind p70 s6k at phospho-Ser 411 , and inactivate the kinase at mitosis. An example of this kind of regulation has been shown for the cell cycle control protein Cdc25 C by Peng et al. (45). This phosphatase is phosphorylated on Ser 216 throughout interphase and has been shown to bind 14-3-3 proteins at this site, thereby sequestering and negatively regulating its activity until entry into mitosis (45). Proteins such as 14-3-3, pin 1, etc., may also play a role negatively regulating p70 s6k by a similar mechanism via Ser 411 phosphorylation. Identification of such p70 s6k -binding factors are now in progress. Cells were pulse-labeled with [ 32 P]orthophosphate at 21-24 h after transfection, and transfected p70 s6k was recovered by anti-HA antibody. A, D146N Cdc2 decreased total phosphorylation of p70 s6k compared with the control (wild type Cdc2). B, immunoblotting using anti-CT antibody demonstrates a similar amount of p70 s6k expressed. C, immunoblotting using anti-phospho-Ser 411 antibody demonstrates that D146N Cdc2 decreases Ser 411 phsophorylation.