Mitotic regulation of ribosomal S6 kinase 1 involves Ser/Thr, Pro phosphorylation of consensus and non-consensus sites by Cdc2.

During mitosis, the cyclin-dependent kinase, Cdc2, signals the inactivation of major anabolic processes such as transcription, mRNA processing, translation, and ribosome biogenesis, thereby providing energy needed for the radical and energetically costly structural reorganization of the cell. This is accomplished by phosphorylation and inactivation of several key anabolic elements, including TFIIIB, TFIID, RNA polymerase II, poly(A) polymerase, and translation elongation factor 1gamma. We report here that ribosomal S6 kinase 1 (S6K1), a protein kinase linked to the translation of ribosomal protein mRNAs, is also subject to regulation by Cdc2 in mitosis. In mitotic HeLa cells, when the activity of Cdc2 is high, S6K1 is phosphorylated at multiple Ser/Thr, Pro (S/TP) sites, including Ser(371), Ser(411), Thr(421), and Ser(424). Concomitant with this, the phosphorylation of the hydrophobic motif site, Thr(389), is reduced resulting in a decrease in the specific activity of S6K1. The mitotic S/TP phosphorylation sites are readily phosphorylated by Cdc2.cyclin B in vitro. These proline-directed phosphorylations are sensitive to chemical inhibitors of Cdc2 but not to inhibitors of mammalian target of rapamycin, phosphatidylinositol 3-kinase, MEK1/2, or p38. In murine FT210 cells arrested in mitosis, conditional inactivation of Cdc2 reduces phosphorylation of S6K1 at S/TP sites while simultaneously increasing phosphorylation of Thr(389) and of the S6K1 substrate, RPS6. A physical interaction exists between Cdc2 and S6K1, and this interaction is enhanced in mitotic cells. These results suggest that Cdc2 provides a signal that triggers inactivation of S6K1 in mitosis, presumably serving to spare energy for costly mitotic processes at the expense of ribosomal protein synthesis.

During the mitotic phase of the cell cycle, the substructure of mammalian cells is reorganized through a tightly regulated series of events, which ultimately ensures that each of the two resulting daughter cells receives the appropriate complement of genetic material. During this period, anabolic processes in the cell are actively repressed (1)(2)(3), serving to spare energy required for costly mitotic events such as chromosome condensation, breakdown of the nuclear envelope, and formation of the mitotic spindle. Transcription, mRNA processing, and translation are among the metabolic activities to undergo mitotic silencing. The collective inactivation of these processes is due in large part to phosphorylation of key pathway components by the cyclin-dependent kinase, Cdc2/Cdk1. A number of mitotic Cdc2 targets involved in the silencing of gene expression have been identified, including RNA polymerase II, TFIID, TFIIIB (reviewed in Ref. 4), poly(A) polymerase (5), and elongation factor 1␥ (6).
Among the most energetically expensive processes in cycling cells is ribosome biogenesis, which, as a result of transcription, processing, translation, and assembly of ribosomal proteins, has been calculated to demand as much as 80% of the energy expenditure of a proliferating cell (reviewed in Ref. 7). Cdc2 may act at multiple levels during mitosis to repress ribosome biogenesis. Inhibition of rDNA transcription is achieved by disruption of the RNA polymerase I complex through Cdc2mediated phosphorylation of TATA-binding protein (8), the TATA-binding protein-associated factor TAF I 110 (8), and thyroid transcription factor-1 (8). Additionally, Cdc2 may also suppress the translation of ribosomal protein-encoding mRNAs through regulation of S6K1 1 (9).
Much of what is known of the function and regulation of S6K1 derives from studies conducted during the transition from the quiescent state (G 0 ) to G 1 . As cells re-enter the cell cycle from G 0 , S6K1 is coordinately phosphorylated on multiple residues, leading to a robust stimulation of kinase activity. In quiescent cells, it is postulated that S6K1 is rendered inactive by virtue of an interaction between the putative autoinhibitory C-terminal tail and the catalytic pocket. This serves both to hinder the interaction of S6K1 with its substrates as well as to conceal internally situated sites of phosphorylation essential for maximal enzyme activity (reviewed in Ref. 10). In response to serum, up to six S/TP sites (i.e. Ser 411 , Ser 418 , Thr 421 , Ser 424 , Ser 429 , and Thr 447 ) are phosphorylated (11)(12)(13), disabling the autoinhibitory domain. However, phosphorylation of S/TP sites within the S6K1 autoinhibitory domain is not sufficient for stimulation of the catalytic activity as deletion variants lacking the autoinhibitory domain (and the six S/TP sites residing therein) are activated in response to mitogens to an extent similar to the full-length protein (13,14). In addition, within the kinase extension, phosphorylation of Thr 389 increases the affinity of the requisite S6K1 kinase, phosphoinositide-dependent kinase 1 (PDK1) (15). PDK1 then phosphorylates Thr 229 within the T-loop of the catalytic domain (16,17) inducing maximal kinase activity. A requirement for phosphorylation of Ser 371 , an additional S/TP site removed from the S/TP cluster within the autoinhibitory domain, for kinase activity has been demonstrated (18), although its order in the hierarchy of serum-stimulated phosphorylations is unclear. Several candidate Thr 389 kinases have been suggested, including the mammalian target of rapamycin (mTOR) (19), PDK1 (20), and NEK6/7 (21), although the provenance of the Thr 389 kinase has yet to be unequivocally demonstrated.
The nature of the proline-directed kinase (or kinases) that catalyzes the serum-stimulated, multisite phosphorylation of S/TP sites is also unclear. In vitro, a number of proline-directed kinases have been shown to phosphorylate S6K1, including ERK1/2, JNK, and Cdc2 (22). Ser 411 lies within a strong consensus for Cdc2 phosphorylation (K/RpSPR/PR/K/H (23)) and is phosphorylated during mitosis when the activity of Cdc2 is augmented (9). Furthermore, by anion-exchange chromatography, significant H1 kinase activity coeluted with a prolinedirected kinase activity toward a peptide substrate (p70 S6 kinase autoinhibitory pseudosubstrate) encompassing the S6K1 C terminus (22). The H1 kinase activity was Suc1-precipitable (22), supporting the premise that the cofractionating S6K1 kinase was Cdc2.
The catalytic activity of Cdc2 is reliant on the formation of a Cdc2⅐cyclin B complex, also known as maturation promoting factor. The abundance of the Cdc2 protein is constant throughout the cell cycle (24), whereas cyclin B accumulates progressively throughout interphase until late in mitosis, at which point it is rapidly destroyed by the anaphase promoting complex (reviewed in Refs. 25 and 26). The destruction of cyclin B, therefore, serves as a mechanism of Cdc2 inactivation. Additionally, the Cdc2⅐cyclin B heterodimer is stabilized by phosphorylation of Cdc2 at Thr 161 and is maintained inactive by dual phosphorylation of Thr 14 and Tyr 15 (reviewed in Ref. 27). At the end of G 2 , Cdc2 is dephosphorylated at Thr 14 and Tyr 15 by the Cdc25 phosphatase, inducing the rapid activation of Cdc2 and triggering entry into mitosis (reviewed in Ref. 28).
Numerous mitotic Cdc2 substrates have been identified, all of which are phosphorylated at S/TP sites. The abundance of mitotic Cdc2 substrates is illustrated by the appearance of numerous phospho-epitopes recognized by the MPM-2 antibody upon entry into mitosis (29), although some of these epitopes may result from phosphorylation by kinases other than Cdc2.
The substrate specificity of Cdc2 shows an absolute requirement for proline in the ϩ1 position, a secondary requirement for Arg or Lys at ϩ3, and a preference for basic residues at ϩ2 or ϩ3 positions (23). Although peptide library screens of potential Cdc2⅐cyclin B substrates predict that HHHKpSPRRR represents the optimal sequence of Cdc2 phosphorylation (23), circumstances in which the consensus has been relaxed are not uncommon (30). Furthermore, many Cdc2 substrates possess multiple S/TP sites, not all of which conform to the consensus sequence of phosphorylation, e.g. lamins (31), Src (32), eIF4E binding protein 1 (33), and poly(A) polymerase (5). In the case of S6K1, eight S/TP sites exist, one of which, Ser 411 , lies within a strong consensus for Cdc2 phosphorylation (RpSPRR), whereas the contexts of the remaining seven diverge somewhat (Fig. 1). Given the abundance of S/TP sites within S6K1 and that Cdc2 phosphorylates the consensus site, Ser 411 , in vitro, we postulated that other non-consensus S/TP sites were likely to be regulated by Cdc2 during mitosis. Here, we present biochemical, pharmacological, and genetic evidence that Cdc2 phosphorylates S6K1 at multiple S/TP sites during mitosis. Furthermore, we demonstrate that Cdc2 is required for mitotic inactivation S6K1 in FT210 cells.
Cell Culture and Transient Transfection-HeLa cells were propagated in Dulbecco's modified Eagle's medium supplemented with fetal calf serum (10%, v/v) and the antibiotics penicillin and streptomycin at 37°C. For cell synchronization, cells were arrested at metaphase by nocodazole treatment (0.4 g/ml) for 24 h. Alternatively, cells were synchronized at the G 1 /S boundary by sequential thymidine (2 mM) and aphidicolin (1 g/ml) blocks. Cells were subsequently released into the cell cycle by removal of aphidicolin and addition of fresh medium for the times indicated. The murine carcinoma cell line, FT210, carrying a temperature-sensitive CDC2 allele, was kindly provided by Fumio Hanaoka (Osaka University) and maintained in RPMI supplemented with fetal calf serum (10%, v/v) and antibiotics at 32°C. To induce Cdc2 deactivation, FT210 cells were cultured at the non-permissive temperature of 39.5°C for the times indicated.
HEK293T cells were propagated in Dulbecco's modified Eagle's medium supplemented with fetal calf serum (10% v/v), penicillin, and streptomycin at 37°C. Cells were transfected with 20 g of plasmid DNA in 100-mm dishes using the calcium phosphate method. Cells were harvested 48 h post-transfection.
Cell Lysis and Immunoprecipitation-Lysates were prepared either from freshly isolated cells or from cell pellets frozen at the time of harvest. Cells were extracted in lysis buffer (50 mM Tris-HCl, pH 7.4, 1% Nonidet P-40 (v/v), 150 mM NaCl, 1.5 mM EDTA, 1 g/ml leupeptin, 1 mM Na 3 VO 4 , 500 M phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 1 mM NaF, 1 mM benzamidine, and 1 g/ml aprotinin. Lysates were rocked for 20 min at 4°C on an end-over-end rotator and then clarified by centrifugation at 15,000 ϫ g at 4°C for 25 min. The protein concentration of cell lysates was determined using the Bio-Rad DC Protein Assay kit. Lysates were either mixed with an equal volume of 2ϫ sample buffer and heated at 100°C for 5 min or subjected to immunoprecipitation.
For immunoprecipitation, lysates were incubated for 2 h with preformed antibody complexes immobilized on BioMag magnetic particles. S6K1 was immunoprecipitated with rabbit polyclonal S6K1 antibody (Santa Cruz Biotechnology, catalog no. sc-230), Cdc2 with mouse monoclonal anti-Cdc2 antibody (Santa Cruz Biotechnology, catalog no. sc-54), and ectopically expressed HA-S6K1 with 12CA5 mouse monoclonal antibody. Immune complexes were then washed three times with lysis buffer and three times with high salt buffer (identical to lysis buffer except NaCl concentration is 500 mM instead of 150 mM). Immune complexes were either resuspended in 1ϫ sample buffer and heated at 100°C for 5 min or subjected to in vitro kinase assay.
Assay of S6 Kinase Activity-The activity of S6K1 was assayed exactly as outlined elsewhere (35). Briefly, anti-S6K1 immune complexes were incubated with a peptide substrate (AKRRRLSSLRA), and 32 P incorporation into the substrate was monitored by liquid scintillation counting.
Assay of Cdc2 Kinase Activity-The activity of Cdc2 was determined in Cdc2 immunoprecipitates using either purified histone H1 or immunoprecipitated HA-S6K1 as substrate. Immune complexes were washed as described above, washed once in reaction buffer (25 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 ), and incubated at 30°C in reaction buffer in the presence of 2 g of purified histone H1 (Calbiochem, catalog no. 382150) and 1 Ci of [␥-32 P]ATP. Phosphorylated histone H1 or HA-S6K1 was analyzed by SDS-PAGE and autoradiography.
Cell Cycle Analysis-Cell cycle analysis was performed by flow cytometry. Briefly, cells were trypsinized, washed in phosphate-buffered saline, fixed in ice-cold 70% ethanol, and treated simultaneously with propidium iodide (40 g/ml) and RNase A (200 g/ml) for 1 h at 37°C. Cells were subsequently analyzed on a FACScan with CellQuest software.

RESULTS
Although the specific activity of S6K1 is reduced in nocodazole-arrested Cos7 cells compared with cells in interphase (9), a similar treatment has no apparent effect in Swiss 3T3 fibroblasts (36). Furthermore, phosphorylation of the consensus Cdc2 S/TP site, Ser 411 , is induced both in nocodazolearrested Cos7 cells and in primary human T cells during G 2 /M (9). We therefore sought to determine whether S6K1 prepared from nocodazole-arrested HeLa cells displayed reduced kinase activity and whether or not this was associated with changes in site-specific phosphorylation. Initially, FACS analysis was performed on growing HeLa cells incubated for increasing periods with nocodazole ( Fig. 2A). These results indicate that 24 h of nocodazole exposure induced a nearcomplete M phase arrest. We therefore chose a 24 h incubation for all subsequent nocodazole treatments. This M phase arrest was associated with a prominent activation of Cdc2 as indicated by dephosphorylation of Tyr 15 (Fig. 2B) and by an increase in H1 kinase activity detected in anti-Cdc2 immunoprecipitates (Fig. 2C).
We observed an interesting profile of phosphorylation for mitotic S6K1. Using anti-phosphopeptide antibodies, four S/TP sites, Ser 371 , Ser 411 , Thr 421 , and Ser 424 , showed reproducible increases in phosphorylation in response to nocodazole, although to differing extents (Fig. 2D). The fold increases in site-specific, mitotic phosphorylation were as follows: Ser 371 , ϳ50%; Ser 411 , ϳ100%; and Thr 421 and Ser 424 , ϳ1000%. This was in stark contrast to Thr 389 , which showed substantial dephosphorylation upon mitotic arrest (Fig. 2D). Because stimulation of the kinase activity of S6K1 is absolutely reliant upon phosphorylation of Thr 389 , the reduction in S6 kinase activity measured in anti-S6K1 immunoprecipitates prepared from mitotic cells was consistent (Fig. 2E). It is noteworthy, however, that multisite phosphorylation of S/TP sites is predicted to disengage the autoinhibitory function of the C-terminal tail, thereby facilitating (and indeed being associated with) the activation of S6K1. These findings, therefore, are suggestive of a novel function of mitotic S/TP phosphorylation distinct from the mitogen-stimulated activation of S6K1.
Previous studies have established that S6K1 is a substrate of Cdc2⅐cyclin B in vitro, particularly at Ser 411 (9). Given the presence of multiple, neighboring S/TP sites within the S6K1 protein, and the fact that these sites are phosphorylated in mitotic cells, we reasoned that these sites may be substrates for Cdc2⅐cyclin B as well. N-terminal HA-tagged S6K1 was transiently expressed in HEK293T cells and immunoprecipitated from asynchronous cells with anti-HA antibody. Active Cdc2⅐cyclin B complexes were expressed in and purified from baculovirus-infected Sf9 cells and used to phosphorylate immunopurified HA-S6K1 in vitro. Of the S/TP sites evaluated, Ser 411 , Thr 421 , and Ser 424 were heavily phosphorylated, whereas Ser 371 was only modestly phosphorylated. The nonproline-directed hydrophobic motif site, Thr 389 , was not phosphorylated in the reaction (Fig. 2F). Thus, phosphorylation of S6K1 by Cdc2⅐cyclin B is specific for Ser/Thr in S/TP motifs, although both consensus and non-consensus sites are efficiently phosphorylated.
To confirm that the observed effects on S6K1 were due to mitotic arrest rather than nonspecific effects secondary to microtubule destabilization, we monitored the pattern of sitespecific phosphorylation of S6K1 throughout the HeLa cell cycle by immunoblotting with anti-phosphopeptide antibodies. Cells were synchronized at the G 1 /S boundary by sequential thymidine/aphidicolin blocks and then released into the cell cycle. FACS analysis revealed that synchronization of these cells was achieved and maintained over the entire cell cycle (Fig. 3A). Generally, from early S phase to G 2 /M (0 -9 h), the phosphorylation of S6K1 at the S/TP sites, Ser 371 , Thr 421 , and Ser 424 , rose steadily (Fig. 3B). Upon entry into the new cycle, these sites were dephosphorylated, and then subsequently their phosphorylation increased as the latter half of the subsequent phase approached (13-27 h). The phosphorylation of ribosomal protein S6 (RPS6), a natural S6K1 substrate, was elevated from early S phase (3 h) to G 2 /M (9 h) then returned to basal levels as cells progressed into the new cycle (Fig. 3B,   FIG. 2. Mitotic arrest induces S6K1 inactivation despite increased S/TPspecific phosphorylation. A, HeLa cells were incubated with nocodazole (0.4 g/ml) for the times indicated. Histograms show cell cycle distribution at each time point. Whole cell lysates (WCL) were prepared from HeLa extracts of asynchronous populations (interphase, I) or after 24-h incubation with nocodazole (mitotic arrest, M). B, HeLa lysates were separated by SDS-PAGE then immunoblotted with antibodies specific for Cdc2 or Cdc2 phosphorylated on Tyr 15 . C, Cdc2 was immunoprecipitated from HeLa extracts and incubated in a kinase reaction using purified histone H1 as substrate. The kinase reactions were separated by SDS-PAGE, transferred to nitrocellulose membranes, and then subjected to autoradiography. A representative autoradiogram is presented. D, HeLa lysates were immunoblotted with antibodies specific for S6K1 or S6K1 phosphorylated on Ser 371 , Ser 411 , Thr 421 /Ser 424 , or Thr 389 . E, S6K1 was immunoprecipitated from HeLa extracts and incubated with a synthetic peptide substrate corresponding to the region of phosphorylation of RPS6. The reactions were spotted onto filters and counted by liquid scintillation counting. The data are representative of five independent experiments. F, HA-S6K1 was transiently expressed in and purified from asynchronous 293T cells and incubated with purified, baculovirus-expressed Cdc2⅐cyclin B in a Cdc2 kinase reaction. The reactions were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and either subjected to autoradiography or immunoblotted with anti-phosphopeptide antisera as indicated. The data are representative of three independent experiments. histogram). It is important to point out that, although populations are enriched in the G 2 /M phase of the cell cycle at 9 h after release from the G 1 /S block, FACS analysis cannot distinguish between G 2 and M phase cells. Dephosphorylation of Cdc2 at Tyr 15 by Cdc25 is an event that triggers the activation of Cdc2 and entry into mitosis (reviewed in Ref. 28). Hence, Tyr 15 phosphorylation and dephosphorylation is a reliable marker of G 2 and M phases, respectively. In view of this, the presence of Tyr 15 phosphorylation at 9 h (Fig. 3B) indicates that most of these cells had yet to enter mitosis and thus were synchronized in G 2 . Furthermore, although by 13 h of release, cells had accumulated in G 1 , a small percentage remained in the G 2 /M phase; these are likely to have been in M phase rather than G 2 (see below).
To precisely monitor the site-specific phosphorylation of S6K1 during mitosis, we performed a narrower time course corresponding to the interval between 8.5 and 15 h after release from G 1 /S block. In these experiments, FACS analysis was carried out in conjunction with a visual assay of Hoeschtstained HeLa cells to identify those times at which the percentage of mitotic cells was greatest. The cell cycle distribution at each time point is shown in Fig. 4A. Based on this analysis, the period of 8.5-10 h comprised the end of G 2 , whereas that of 13.5-15 h represented the beginning of G 1 . The period from 10.5-12.5 h displayed the greatest enrichment in mitotic cells as determined by Hoescht staining. It is important to point out that only ϳ50% of Hoescht-stained cells were discernibly mitotic at 11.5 h, although this time point displayed the greatest mitotic index (data not shown). The C-terminal S/TP sites of S6K1, Ser 411 , Thr 421 , and Ser 424 , all displayed peak phosphorylation during mitosis (Fig. 4B). Conversely, RPS6 was increasingly dephosphorylated as cells entered mitosis, concomitant with dephosphorylation of Cdc2 at Tyr 15 (Fig. 4, compare  B-D).
Given that Cdc2 phosphorylates S6K1 at multiple S/TP sites in vitro (Ref. 9 and Fig. 2F) and that Cdc2 phosphorylation of non-consensus S/TP sites in other Cdc2 substrates is not uncommon (5), we reasoned that the non-consensus S/TP sites found in S6K1 (e.g. Ser 371 , Thr 421 , and Ser 424 ) may also be under Cdc2 control in vivo. It would then be reasonable to predict that pharmacological inhibition of Cdc2 during mitosis would block the appearance of phospho-epitopes at putative consensus and non-consensus Cdc2 sites. To test this hypothesis, HeLa cells were arrested in mitosis with nocodazole followed by the addition of various phosphatase and kinase inhibitors. The addition of sodium orthovanadate, a tyrosine phosphatase inhibitor, has been classically used to inhibit tyrosine dephosphorylation of Cdc2 by the dual-specificity phosphatase, Cdc25 (37), thereby maintaining Cdc2 inactivity. Because the Thr 14 and Tyr 15 kinases, Wee1 and Myt1, remain active (or, in fact, may be activated as a result of inhibition of Cdc2), Cdc2 is rephosphorylated after vanadate-induced inhibition of Cdc25. Addition of this inhibitor resulted in dephosphorylation of S6K1 at both consensus (Ser 411 ) and non-consensus (Thr 421 and Ser 424 ) Cdc2 sites (Fig. 5A). Similar results were obtained with the cyclin-dependent kinase inhibitors, roscovitine and purvalanol A, insofar as S/TP site phosphorylation is concerned. Proline-directed phosphorylation of mitotic S6K1 was both rapamycin-resistant (Fig. 5A) and wortmanninresistant (Fig. 5B), precluding the possibility that either mTOR or phosphatidylinositol 3-kinase controls the S/TP-specific phosphorylation of S6K1 during mitosis. However, the residual phosphorylation of Thr 389 , a rapamycin-sensitive, non-S/TP phosphorylation site within the S6K1 kinase extension, which persists in mitotic cells, was further dephosphorylated in response to rapamycin (data not shown). This indicates that  15 . Site-specific S/TP phosphorylation was calculated as the ratio of phospho-specific signal to total S6K1 signal for individual time points. These data are presented in the histogram above the immunoblots. These data represent means Ϯ S.E. for five independent determinations. mTOR retains the ability to signal to S6K1 during mitosis. Other proline-directed kinases reportedly active in mitotic cells (38,39) were also evaluated as candidate regulators of S6K1.
However, inhibitors of the MEK/ERK pathway, U0126 and PD098059 (Fig. 5A), and of the mitogen-activated protein kinase, p38, SB202190 (Fig. 5B), had little effect on the phospho- rylation of S/TP sites. Collectively, the results of these inhibitor studies provide compelling evidence that Cdc2 is likely to be the primary mitotic S6K1 kinase.
If Cdc2 represents the major mitotic kinase regulating S/TP site phosphorylation of S6K1, then genetic inactivation of Cdc2 should prevent such phosphorylation. In mammalian cells, a temperature-sensitive CDC2 allele has been identified at the genomic locus of the FT210 murine carcinoma cell line (40). The CDC2 gene product in these cells harbors two point mutations (I54V and P272S) that promote its rapid inactivation and degradation at the restrictive temperature of 39°C (40). As a result, FT210 cells are unable to enter mitosis and accumulate in G 2 when cultured at 39.5°C (Ref. 40 and Fig. 6A). To test whether active Cdc2 kinase is necessary for mitotic S/TP-specific phosphorylation of S6K1, mitotic arrest was induced with nocodazole at the permissive temperature (32°C), and cells were then transferred to the restrictive temperature to inactivate Cdc2. Based on the results of FACS analysis, this treatment induces a complete M phase arrest (Fig. 6A). We then switched the nocodazole-arrested cells to the restrictive temperature for increasing lengths of time and assayed site-specific phosphorylation of S6K1 by immunoblotting. Consistent with a critical role for Cdc2 in the regulation of S/TP-specific phosphorylation, S6K1 was increasingly dephosphorylated at Ser 411 , Thr 421 , and Ser 424 , when cells were incubated at the non-permissive temperature (Fig. 6B). These changes temporally paralleled the loss of Cdc2 protein observed at the restrictive temperature. Interestingly, after 6 h at the restrictive temperature, when the abundance of Cdc2 protein was substantially reduced, RPS6 was rephosphorylated, as was S6K1 at Thr 389 (Fig. 6B). These findings demonstrate not only that Cdc2 is necessary for S/TP-specific phosphorylation of S6K1 during mitosis but also that Cdc2 is necessary for inhibition of S6K1 in mitotic cells.
It has been reported that a significant H1 kinase activity copurifies with an S/TP kinase activity specific for S6K1 C terminus (amino acids 400 -436; also called p70 S6 kinase autoinhibitory pseudosubstrate peptide) prepared from H4 hepatoma cells (22). Because Cdc2-mediated regulation of S6K1 is likely to reflect direct phosphorylation of S/TP sites most notably during M phase, we reasoned that a complex of S6K1 and Cdc2 may be detectable by immunoprecipitation. We were able to detect Cdc2 in anti-S6K1 immunoprecipitates from interphase HeLa extracts (Fig. 7). Importantly, the formation of the S6K1⅐Cdc2 complex was enhanced ϳ3-fold during mitosis We have been unable to detect site-specific differences in phosphorylation of S6K1 between that pool bound to Cdc2 and that found in a whole cell extract (data not shown) sug- gesting that phosphorylation of S6K1 by Cdc2 does not affect the association of the two proteins.

DISCUSSION
In the present study, we have demonstrated that S6K1 is phosphorylated on both consensus and non-consensus S/TP residues in vitro by Cdc2⅐cyclin B and in vivo in a mitosisspecific fashion. Mitotic phosphorylation of S6K1 is associated with its inactivation as evidenced by decreased specific kinase activity and dephosphorylation of its substrate, RPS6. We have provided both pharmacological and genetic evidence suggesting that Cdc2 represents the major mitotic regulator of S6K1. Our results demonstrate that Cdc2 is required for mitotic phosphorylation of S/TP sites as well as mitotic inhibition of S6K1. Finally, Cdc2 and S6K1 form an immunoprecipitable complex whose stability is enhanced in mitotic cells.
An essential role of S6K1 in the G 1 /S phase transition has been inferred from two lines of evidence. First, Lane et al. (41) demonstrated that microinjection of neutralizing anti-S6K1 antibodies prevented serum-induced S phase entry in fibroblasts. Second, rapamycin, which rapidly inactivates S6K1 (42)(43)(44)(45)(46), also delays or inhibits cell cycle progression, specifically in G 1 (42,43,47). In view of this body of evidence, much effort has been focused on understanding cell cycle regulation of S6K1 in the context of the G 0 /G 1 transition. The addition of serum to quiescent cells induces phosphorylation of multiple S/TP sites within the autoinhibitory tail as well as within the kinase domain at Thr 229 and Thr 389 . Collectively, this series of events leads to a rapid increase in the kinase activity of S6K1 (48). Using synchronous populations of cells released from G 1 /S arrest, we have observed that phosphorylation of S6K1 at S/TP sites and of RPS6 is enhanced during late phases of the cell cycle. However, upon mitotic entry, S/TP site phosphorylation of S6K1 is further stimulated, whereas RPS6 is dephosphorylated. This indicates that, despite enhancd phosphorylation of S6K1 at multiple S/TP sites, the S6K1/RPS6 pathway is inactivated during mitosis. In synchronized HeLa cells, the incorporation of ribosomes into larger polysomes is greatest during G 2 (49), consistent with the G 2 phase representing a period of high translational activity. We found that the phosphorylation of RPS6 was not significantly stimulated as cells entered G 1 from M phase. Similar results were obtained from HeLa cells released from nocodazole arrest (data not shown). Collectively, these data argue that, in the absence of G 0 arrest, the machinery responsible for the translation of ribosomal protein mRNAs is engaged in the latter half of the cell cycle.
The findings that mitotic phosphorylation of the S/TP sites, Ser 411 , Thr 421 , and Ser 424 , is abrogated by Cdc2 inhibitors and is reduced in cells in which Cdc2 has been conditionally inactivated suggests that Cdc2 is likely to be the primary mitotic S6K1 regulator. This hypothesis is further supported by the fact that Ser 411 (Ref. 9 and Fig. 2F) as well as Ser 371 , Thr 421 , and Ser 424 ( Fig. 2A) are directly phosphorylated by Cdc2 in vitro and that the association of Cdc2 and S6K1 is induced in mitotic cells (Fig. 7). Eight S/TP sites are included in the S6K1 protein: two in the kinase extension (i.e. Thr 367 and Ser 371 ) and six in the C-terminal autoinhibitory tail (i.e. Ser 411 , Ser 418 , Thr 421 , Ser 424 , Ser 429 , and Thr 447 ). It should be noted that, although the phosphorylation of S6K1 at Thr 367 , Ser 418 , Ser 429 , and Thr 447 were not directly measured in this study, these residues might represent additional sites phosphorylated by Cdc2 in mitotic cells.
Among the S/TP sites, only Ser 411 conforms to the consensus for Cdc2 phosphorylation (23,30). Cdc2 substrates often possess a combination of consensus and non-consensus S/TP sites. The consensus sites of one such substrate, poly(A) polymerase, respond to lower levels of Cdc2 kinase activity than do nonconsensus sites, although phosphorylation of all sites is necessary for mitosis-specific regulation of the catalytic activity of poly(A) polymerase (5). This led to the interesting postulation that particular Cdc2 substrates may, by virtue of the presence of particular combinations of consensus and non-consensus sites, respond to varying levels of Cdc2 activity. In the case of S6K1, we observed increased phosphorylation of the Cdc2 consensus site, Ser 411 , during G 2 , whereas phosphorylation of Thr 421 and Ser 424 was increased only upon entry into mitosis (Fig. 4B). Therefore, Cdc2-mediated regulation of S6K1 may rely upon the cumulative effect of multiple S/TP phosphorylations that can only be achieved during mitosis, when the activity of Cdc2 is highest. It should be emphasized that phosphorylation of S6K1 by Cdc2 may serve some regulatory function without influencing S6K1 activity (see below). The increase in phosphorylation of S6K1 at Ser 411 may in fact derive from phosphorylation by a Cdk⅐cyclin A complex, such as Cdc2⅐cyclin A, which is active in G 2 . It is likely, however, that Cdc2⅐cyclin B phosphorylates S6K1 at the G 2 /M transition and throughout M phase.
The observation that loss of the Cdc2 protein in mitotically arrested cells temporally coincides with increased phosphorylation of RPS6 (Fig. 6B) implies that one function of active Cdc2 is to signal mitotic inactivation of S6K1. Initially this might seem paradoxical, because phosphorylation of multiple S/TP sites within the C-terminal tail is predicted to disengage the autoinhibitory mechanism, thereby "priming" S6K1 for further activation. In light of the observation that multisite, prolinedirected phosphorylation is reduced upon entry into G 1 (Fig. 4,  A and B), it is unlikely that mitotic phosphorylation would serve to prime S6K1 for activation in G 1 , especially because RPS6 is not phosphorylated during this period. What purpose might be served by multisite S/TP phosphorylation of S6K1 in mitotic cells? Potentially, these phosphorylations could alter the binding preference of S6K1 for mitosis-specific interaction partners. In fact, phosphorylation of the Cdc2 consensus site, Ser 411 , is predicted to create an optimal binding surface for the peptidyl-prolyl cis/trans isomerase, Pin1 (50), which has been shown to interact with S6K1 in a mitosis-specific fashion (50). Given that the S/TP-rich C-terminal tail is postulated to regulate S6K1 activity via a phosphorylation-dependent change in conformation, Pin1 may play an active role in this process during mitosis. Pin1 has been shown to facilitate dephosphorylation of S/TP sites by PP2A (51,52), potentially explaining the rapid dephosphorylation of S6K1 at these sites at the point of mitotic exit. Studies are currently in progress to address these possibilities.
The data presented in Figs. 2 and 6 suggest that Cdc2 signals the inactivation of S6K1 in mitotic cells by inducing dephosphorylation of Thr 389 . Culturing mitotically arrested FT210 FIG. 7. The association of S6K1 and Cdc2 is increased in mitotic cells. HeLa cells were either left untreated or treated with nocodazole (0.4 g/ml) for 24 h. Immunoprecipitations were performed using anti-S6K1 antibody or normal rabbit serum (NRS), and coprecipitating proteins were resolved by SDS-PAGE. Immunoblots were performed using antibodies selective for S6K1 or Cdc2. cells at the restrictive temperature results in rephosphorylation of Thr 389 , an effect that tightly parallels disappearance of the Cdc2 protein (Fig. 6). Cdc2 is therefore necessary for mitotic regulation of S6K1 by reducing the extent to which Thr 389 is phosphorylated. Given that Cdc2 is highly active in mitotic cells and that Thr 389 does not fall within a preferred S/TP context for phosphorylation by Cdc2, the regulation of Thr 389 phosphorylation by Cdc2 is likely to be indirect. Naturally, the reduction in phosphorylation of Thr 389 could derive from inhibition of the Thr 389 kinase or stimulation of the Thr 389 phosphatase. Reconciliation of this issue is complicated, however, by the fact that neither the Thr 389 kinase nor the Thr 389 phosphatase has been unequivocally described. Nevertheless, mTOR represents an attractive intermediary target in the regulation of Thr 389 phosphorylation by Cdc2. The phosphorylation of Thr 389 is sensitive to the mTOR inhibitor, rapamycin, which leads to the rapid inactivation of S6K1 in the presence of the compound (53). In vitro studies have demonstrated that mTOR can phosphorylate Thr 389 (19,54); however, several rapamycin-resistant S6K1 constructs have now been described in which Thr 389 is phosphorylated under mTOR-inactivating conditions (e.g. rapamycin treatment (55)(56)(57) and amino acid withdrawal (55)). Thus, the most plausible explanation of these observations is that mTOR indirectly controls Thr 389 phosphorylation of S6K1. It has been reported elsewhere that nocodazole arrest stimulates the kinase activity of mTOR in DOHH2 cells (58). Nevertheless, we cannot exclude the possibility that mTOR is inactivated in mitotic HeLa and FT210 cells. Collectively, our data suggest a bimodal regulation of S6K1 by Cdc2 (Fig. 8): in mitotic cells, Cdc2 directly phosphorylates S6K1 at multiple S/TP motifs and concomitantly inhibits the pathway affecting the phosphorylation of Thr 389 . This latter effect is indirect and accounts for the reduced activity of mitotic S6K1 compared with interphase S6K1.
The importance of S6K1 in cell cycle progression is underscored by the observation that the S6K1 gene is amplified or overactive in primary breast cancers (59,60), ovarian cancers (61), and in phosphatase and tensin homolog deleted on chromosome 10-null tumors (62). Additionally, not only is ribosome biogenesis a strong prognostic indicator of tumorigenicity (63), but significant, positive correlations have been established between the up-regulation of ribosomal protein-encoding mRNAs and the growth rate of transformed cells (64). Additionally, overexpression of wild-type S6K1, but not an inactive variant, in NIH3T3 cells appears to interfere with cytokinesis as evidenced by the multinucleation and polyploidization of transfected cells (65). This suggests that failure to inactivate S6K1 during mitosis would deleteriously affect the integrity of the genome as well as the growth characteristics of the cell. In mitosis, active Cdc2⅐cyclin B directly phosphorylates (solid arrows) S/TP sites within the kinase domain and C terminus of S6K1. In addition, Cdc2 negatively regulates the kinase activity of S6K1 by reducing phosphorylation of Thr 389 (hatched arrow). This negative regulation is likely to be indirect.