Cell cycle regulation of p70 S6 kinase and p42/p44 mitogen-activated protein kinases in Swiss mouse 3T3 fibroblasts.

We show here using synchronized Swiss mouse 3T3 fibroblasts that p70 S6 kinase (p70S6k) and mitogen-activated protein kinases (p42mapk/p44mapk) are not only activated at the G0/G1 boundary, but also in cells progressing from M into G1. p70S6k activity increases 20-fold in G1 cells released from G0. Throughout G1, S, and G2 it decreases constantly, so that during M phase low kinase activity is measured. The kinase is reactivated 10-fold when cells released from a nocodazole-induced metaphase block enter G1 of the next cell cycle. p42mapk/p44mapk in G0 cells are activated transiently early in G1 and are reactivated late in mitosis after nocodazole release. p70S6kactivity is dependent on permanent signaling from growth factors at all stages of the cell cycle. Immunofluorescence studies showed that p70S6k and its isoform p85S6k become concentrated in localized spots in the nucleus at certain stages in the cell cycle. Cell cycle-dependent changes in p70S6k activity are associated with alterations in the phosphorylation state of the protein. However, examination of the regulation of a p70S6kmutant in which the four carboxyl-terminal phosphorylation sites are changed to acidic amino acids suggests that a mechanism independent of these phosphorylation sites controls the activity of the enzyme during the cell cycle.

Signaling pathways that operate through tightly controlled protein phosphorylation cascades transduce extracellular signals to various intracellular targets. Some of these targets regulate the transcriptional and translational machinery to ensure proper cell cycle progression, cell growth, or differentiation. Phosphorylation of the S6 protein of 40 S ribosomal subunits is a highly conserved response of animal cells to treatment with growth factors, steroid hormones, phorbol esters, and oncogenes (1). Inhibition of S6 phosphorylation by exposure of cells to the immunosuppressant rapamycin selectively suppresses the translation of certain mRNAs that contain a polypyrimidine tract at the 5Ј end (2). These mRNAs encode ribosomal proteins and protein synthesis elongation factors, whose production is required for efficient transit through the G 1 phase of the cell cycle.
Two families of mitogen-stimulated S6 kinases have been identified: the rsk-encoded M r 85,000-92,000 S6 kinases (Ref. 3, referred to as p90 rsk ) 1 and the M r 70,000 and 85,000 S6 kinases (Refs. 4 and 5, referred to as p70 S6k and p85 S6k ). A variety of evidence indicates that p90 rsk and p70 S6k /p85 S6k lie on different signaling pathways (6 -10). Unlike p70 S6k (8), p90 rsk is phosphorylated and activated by the erk-encoded M r 42,000 and 44,000 mitogen-activated protein (MAP) kinases (Ref. 6, referred to as p42 mapk and p44 mapk ) in response to signals transmitted through p21 ras , p74 raf , and p47 mek (11). Once activated, p90 rsk and p42 mapk /p44 mapk can be translocated to the nucleus (12,13), where they are thought to phosphorylate nuclear transcription factors, thus promoting the transcription of genes required for the growth response.
p70 S6k is the physiological S6 kinase activity in mammalian cells (14). p85 S6k is a minor species that is identical to p70 S6k except for the presence of a 23-amino acid extension at the amino terminus that carries features of a nuclear targeting sequence (15,16). Indeed, p85 S6k is localized in the nucleus (17,18), where it might phosphorylate a nuclear pool of S6 protein (19) or the cAMP-response element modulator (CREM), a transcription factor which was recently identified as a substrate of p70 S6k (20). Activation of p70 S6k in response to mitogens is associated with phosphorylation of three serines and one threonine located at the carboxyl terminus of the kinase (21). p70 S6k also contains additional phosphate groups that become dephosphorylated upon rapamycin treatment, leading to inactivation of p70 S6k (22). Each set of phosphorylation sites might be modified by distinct kinases; however, direct activators of p70 S6k /p85 S6k are so far unknown. Recent experiments based on the use of rapamycin (23,24), specific phosphatidylinositol 3-kinase inhibitors such as wortmannin (25,26) and plateletderived growth factor receptor mutants (27) have suggested that phosphatidylinositol 3-kinase and the structurally related enzyme RAFT/FRAP are involved in upstream signaling to p70 S6k /p85 S6k .
Two lines of evidence have suggested that the function of p70 S6k /p85 S6k during G 1 is important for cell cycle progression. First, inhibition of p70 S6k /p85 S6k by treatment of cells with rapamycin leads to cell cycle arrest in G 1 or a delay of entry into S phase, depending on the cell type (9,14). Second, microinjection of rat embryo fibroblasts with antibodies that inhibit p70 S6k /p85 S6k abolishes the serum-induced entry into S phase (17,28). Activation of p42 mapk /p44 mapk is also thought to be essential for triggering the proliferative response in fibroblasts (29). To gain further insight into how p70 S6k /p85 S6k and p42 mapk /p44 mapk activity is regulated and what role the enzymes might have in cell growth and cell cycle control, we have examined the behavior of these enzymes during the cell cycle. We show here that p70 S6k and p42 mapk /p44 mapk activities are regulated in a cell cycle-dependent manner. Furthermore, we present evidence that the cell cycle regulation of p70 S6k /p85 S6k activity might involve compartmentation and a regulatory mechanism that is independent of the four carboxyl-terminal phosphorylation sites. Finally, our observations of the behavior of p70 S6k and p42 mapk /p44 mapk during the cell cycle suggest that there is cross-talk between these signaling molecules and the cell cycle machinery.

MATERIALS AND METHODS
Synchronization of Cells-Swiss mouse 3T3 fibroblasts were seeded in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.) containing 10% fetal calf serum (FCS, Life Technologies, Inc.) at 1.3 ϫ 10 5 cells per 10-cm plate or 3.0 ϫ 10 5 cells per 15-cm plate and allowed to grow for 3 days. To synchronize cells in G 0 , cells were serum-starved in DMEM plus 0.5% FCS for 24 -48 h. Then 5 nM epidermal growth factor (EGF), 1 nM insulin, and 6% FCS were added to induce re-entry into the cell cycle. Alternatively, 20% FCS was used as a mitogen.
To arrest cells in metaphase, cells were first presynchronized in G 0 by serum starvation and then stimulated with EGF, insulin, and FCS as described above. Nocodazole (0.4 g/ml, Sigma) was added 20 h after release from G 0 , before cells entered M phase. Mitotic cells were collected 4 h later by gentle pipetting and were reseeded into DMEM plus 10% FCS at 0.5 ϫ 10 6 cells per 10-cm plate. These conditions have been shown to result in an efficient and reversible M phase block that prevents cells from entering a polyploid state (30). Mitotic cells were isolated without nocodazole by multiple rounds of mechanical shake-off alone (31) from subconfluent plates of cells which were synchronized in G 0 and stimulated for 22 h with mitogens as described above. Cells collected from early rounds were stored in DMEM plus 10% FCS at 4°C during the collection process.
Cell Extraction, Protein Determination and Kinase Assays-Plates (10 cm) were placed on ice and the cells were washed twice with cold phosphate-buffered saline (PBS, 137 mM NaCl, 3 mM KCl and 12 mM P i , pH 7.4). Cells were scraped into 400 l of cold extraction buffer (120 mM NaCl, 20 mM NaF, 1 mM benzamidine, 5 mM EGTA, 30 mM sodium PP i , 0.1% Triton X-100, 30 mM 4-nitrophenyl phosphate, 0.5 mM dithiothreitol, 50 mM Tris, and 0.1 mM phenylmethylsulfonyl fluoride, pH 7.2), homogenized with 6 strokes in a Teflon-glass homogenizer, and centrifuged at 4°C for 10 min at 8000 ϫ g. The supernatants were stored at Ϫ70°C. Protein concentration in cell extract supernatants was determined by a Lowry method (32).
S6 kinase activity was measured using 40 S ribosomal subunits as substrate as described earlier (26). The reactions were stopped and the samples were subjected to polyacrylamide gel electrophoresis and scintillation counting as described previously (33). A unit is defined as the amount of kinase incorporating 1 pmol of P i into S6 per min.
Polyclonal antibodies to a His-tagged fragment of rat p70 S6k (amino acids 258 -469; Ref. 4) were produced and affinity purified as described previously (34). These purified antibodies show no cross-reaction with p90 rsk on Western blots or in immunoprecipitation kinase assays (data not shown). For immunocomplex S6 kinase assays, Triton X-100 (to 1%), phenylmethylsulfonyl fluoride (to 0.1 mM), and purified S6 kinase antibody (1 l) were added to 50 l of cell extract supernatant and kept on ice for 3 h. Then 25 l of 50% (v/v) protein A-agarose (preincubated with 1% bovine serum albumin in extraction buffer plus 1% Triton X-100) was added and incubated for 1 h at 4°C. The beads were washed twice with extraction buffer plus 1% Triton X-100 and twice with S6 kinase assay buffer without dithiothreitol. The kinase assays were performed essentially as described above. Immunocomplex S6 kinase assays with hemagglutinin (HA) antibody 12CA5 (35) were performed essentially as described above except that cell extract supernatants containing 800 g of protein were used per sample and more extensive washes of the beads were done.
For MAP kinase immunocomplex assays, immunoprecipitations with antibody 122 against p42 mapk were performed essentially as described above, except that the last two washes were done with MAP kinase assay buffer (30 mM Tris, pH 8, 20 mM MgCl 2 , 2 mM MnCl 2 , 0.1% Triton X-100, and 0.1 mM dithiothreitol). MAP kinase assays were initiated by adding 15 l of MAP kinase assay buffer containing 10 M ATP, 2 M of the peptide inhibitor of cAMP-dependent protein kinase (Sigma), 10 g of myelin basic protein (Sigma), and 0.33 l [␥-32 P]ATP. After 30 min at 37°C the reactions were stopped and the samples were subjected to electrophoresis on SDS-20% polyacrylamide gels, autoradiography and scintillation counting.
Flow Cytometry-Fibroblasts in 10-cm plates were trypsinized and collected by centrifugation. The cells were washed twice with PBS and fixed overnight in 70% ethanol. Before analysis, cells were washed twice with PBS and treated with RNase A at a final concentration of 1 mg/ml for 20 min. Cells were suspended in 500 l of propidium iodide solution (40 g/ml in PBS) and kept on ice until use. Data were collected and analyzed with a Becton Dickson FACScan and Lysis II software (Becton Dickson). Immunofluorescence-Cells were fixed directly on tissue culture dishes for 20 min at room temperature with 3% paraformaldehyde (w/v) in PBS (pH 8.0) and permeabilized with 0.1% Triton X-100 in PBS for 20 min. Mitotic cells were allowed to adhere to adhesion slides (Bio-Rad; 5 min) and fixed as described above. Incubation with primary antibodies (affinity-purified anti-p70 S6k antibody diluted 1:50 in PBS or anti-tubulin monoclonal YOL1/34 antibody diluted 1:500 in PBS) was performed overnight at room temperature. The secondary fluoresceinconjugated goat F(abЈ) 2 anti-rabbit Ig (GϩL) (Tago, Inc.) was diluted 1:50 in PBS and incubated 4 h at room temperature in the dark. The secondary rhodamine-conjugated goat anti-rat IgG (HϩL) (Jackson ImmunoResearch) was diluted 1:500 in PBS. Nuclei were counterstained with 0.5 g/ml 4Ј,6Ј-diamidino-2-phenylindole (DAPI). Fluorescence was visualized with a Zeiss Axiophot microscope equipped with a CCD camera. Images were processed with Gene Join and Photoshop programs.
Fibroblasts Stably Expressing p70 S6k Constructs-Mutant p70 S6k was generated by replacing Ser 3 Asp (residues 411, 418, and 424) and Thr 3 Glu (residue 421) by the polymerase chain reaction. 2 Wild-type or mutant p70 S6k cDNAs (nucleotides 137-1705; Ref. 4) carrying the HA tag 5Ј to the p70 S6k sequence were inserted into the mammaliam expression vector pMV-7 (36). The wild-type and mutant inserts were verified by nucleotide sequencing. pMV7ϩ wild-type S6 kinase, pMV7ϩ mutant S6 kinase, or vector alone were transfected into the packaging and virus-producing cell line GPϩE 86 (37) using calcium phosphate precipitation. Neo ϩ clones were selected by growth in medium containing 1 mg/ml G418 (Geneticin, Life Technologies, Inc.); clones were visible 8 -10 days after selection. Preparation of virus stocks and viral infection procedures were carried out essentially as described earlier (38). Briefly, Swiss mouse 3T3 fibroblasts were infected with culture fluid from the virus-producing cells in the presence of 8 g/ml Polybrene. The medium was changed the following day and two days after infection cells were trypsinized and replated at a 1:3 dilution in G418 medium. Stable transfectants were obtained 1-2 weeks after replating.

S6 Kinase and MAP Kinase Activity in Cells
Released from G 0 -Most studies examining the regulation of p70 S6k activity have focused on the first few hours after stimulation of G 0arrested cells. We examined the behavior of p70 S6k over the course of one complete cell cycle to determine whether the enzyme is active at other times besides G 1 and to compare its activity in G 0 /G 1 and M/G 1 cells. Subconfluent Swiss mouse 3T3 fibroblasts were synchronized in G 0 by serum starvation and then induced to enter the cell cycle by adding FCS plus insulin and EGF. Under these conditions the cells traversed the first cell cycle with good synchrony (Fig. 1A). Analysis of cellular DNA content by flow cytometry showed that cells started to enter S phase synchronously after 16 h of stimulation and 4 -6 h later 95% of the cells had doubled their DNA content. After 26 -28 h of stimulation most of the cells had gone through M phase and reappeared as G 1 cells with a 2 N content of DNA (Fig. 1A). Similar results were obtained using 20% FCS as a mitogen (data not shown).
To measure p70 S6k activity in extracts of synchronized cells, immunocomplex kinase assays were performed using 40 S ribosomal subunits as a substrate. In these experiments an antibody that recognizes both p70 S6k and p85 S6k was used. However, since p85 S6k is much less abundant than p70 S6k in these cells (17), most of the activity measured in the immunocomplexes is contributed by p70 S6k . p70 S6k activity measured 20 min after mitogen stimulation was more than 20 times higher than the activity measured in serum-starved cells (Fig.  1B, closed circles). The kinase lost 40% of its activity by the end of G 1 (Fig. 1B, 16 h), consistent with published data (28). During S and G 2 the activity continued to decrease, so that during M phase a relatively low level of activity was measured ( Fig. 1B). It was shown earlier that p70 S6k is activated at the first G 2 /M boundary during meiotic maturation of Xenopus laevis oocytes (39). However, we saw no increase in p70 S6k activity in fibroblasts cycling from G 2 into M (Fig. 1B). As the nuclear transcription factor CREM has been shown to be phosphorylated by p70 S6k (20), we also performed immunocomplex kinase assays with recombinant CREM protein as a substrate. The pattern of p70 S6k activity toward CREM during the cell cycle was virtually identical to that seen toward S6 (data not shown). All evidence obtained so far indicates that p70 S6k is activated by phosphorylation (21,40). To determine if the gradual decrease in p70 S6k activity observed in Fig. 1B could be due to dephosphorylation, the phosphorylation state of p70 S6k was assessed by its appearance as multiple bands on Western blots. The active, phosphorylated form of p70 S6k migrates more slowly in SDS-polyacrylamide gels than the dephosphorylated, inactive species (40). After 20 min of stimulation the most highly phosphorylated form of p70 S6k was detected (Fig. 1C, upper panel), which correlated to the highest S6 kinase activity ( Fig. 1B, closed circles). The hyperphosphorylated form was seen throughout G 1 , and by the end of G 1 partially dephosphorylated forms of the kinase appeared (Fig. 1C, upper panel). As p70 S6k activity continued to decrease, the hypophosphorylated forms of the kinase became predominant. The Western analysis also showed that the amount of p70 S6k protein remained constant during the cell cycle (Fig. 1C). Together, these data demonstrate that p70 S6k activity is cell cycle regulated and that the changes in activity appear to be mediated by phosphorylation-dephosphorylation.
In parallel to p70 S6k , we also examined the cell cycle regulation of MAP kinases. Immunocomplex kinase assays showed that p42 mapk was strongly but only transiently activated early in G 0 /G 1 (Fig. 1B, open squares). During S and G 2 , p42 mapk activity was at basal levels and during G 2 /M the kinase activity increased to a small extent. MAP kinases were also visualized on Western blots that were probed with an antibody that recognizes both p42 mapk and p44 mapk (Fig. 1C, lower panel). The mitogen-dependent activation of p42 mapk /p44 mapk was clearly seen as a shift of the proteins to the phosphorylated, more slowly migrating forms after 20 min of stimulation (Fig. 1C, lower panel). Phosphorylated species of p42 mapk /p44 mapk were also seen at 4 h but thereafter the dephosphorylated forms were predominant. Because p42 mapk gives a relatively faint signal on Western blots, no up-shifted band that would represent active p42 mapk was seen in G 2 /M extracts (Fig. 1C, lower  panel). However, a minor upper band representing active p44 mapk was visible at later stages in the cell cycle (Fig. 1C, lower panel).
S6 Kinase and MAP Kinase Activity in Cells Released from Metaphase-Because fibroblasts released from G 0 were not well synchronized after completing the first cell cycle (Fig. 1A), a different synchronization method had to be used in order to study p70 S6k in cells exiting mitosis and entering G 1 of the next cell cycle. We explored several methods to obtain enriched populations of cells in G 1 /S or G 2 /M and the best results were obtained using a nocodazole-induced metaphase block (see "Materials and Methods"). The mitotic cells were collected and reseeded into drug-free medium to allow them to re-enter the cell cycle. Analysis by flow cytometry showed that 95% of the collected cells were in M phase ( Fig. 2A, t ϭ 0 h). Within 2 h after reseeding, 90% of the cells had gone through mitosis and entered G 1 and after 12 h the cells started to enter S phase ( Fig.  2A). In agreement with the data in Fig. 1B, immunocomplex kinase assays showed that the metaphase-arrested cells exhibited relatively low p70 S6k activity (Fig. 2B, closed circles). When FIG. 1. Behavior of p70 S6k and p42 mapk /p44 mapk in synchronized cells released from G 0 . Subconfluent fibroblasts were serum starved and released into the cell cycle by adding mitogens at t ϭ 0 h (see "Materials and Methods"). A, synchrony of the cells was followed by analyzing their DNA content by flow cytometry. B, at the indicated times after release, extract supernatants were prepared from a parallel set of cells and immunocomplex kinase assays were performed to measure p70 S6k (q) and p42 mapk (Ⅺ) activity. C, proteins in cell extract supernatants were separated by polyacrylamide gel electrophoresis and transferred onto nitrocellulose. p70 S6k (upper panel) and MAP kinases (lower panel) were detected with polyclonal antibodies (see "Materials and Methods"). the cells progressed from M into G 1 the activity increased more than 10-fold (Fig. 2B). Western analysis showed that the increase in activity was associated with an increase in phosphorylation of p70 S6k and that the amount of expressed protein did not change (Fig. 2C, upper panel).
p42 mapk also became activated in G 1 cells after release from nocodazole but the activity returned very rapidly to near-basal levels (Fig. 2B, open squares). Western analysis showed that p42 mapk and p44 mapk were phosphorylated 1.5 h after release from nocodazole and extensively dephosphorylated 3 h after release (Fig. 2C, lower panel). These results show that p70 S6k and p42 mapk /p44 mapk are not only activated at the G 0 /G 1 boundary, but also in fibroblasts cycling from M into G 1 after release from a nocodazole block.
One could argue that the low p70 S6k activity measured in metaphase-arrested cells might be an artifact of nocodazole treatment. To exclude this possibility, we compared the amount of p70 S6k activity in mitotic cells that were collected by shake-off with or without nocodazole treatment and found that there was no significant difference (Fig. 3A). In addition, there was no difference in kinase activity in the drug-treated or non-treated cells that were left on the plates after mitotic shake-off (Fig. 3A). We also tested whether nocodazole inhibits the stimulation of p70 S6k in resting cells upon addition of EGF. Pretreatment of cells for 30 min with different concentrations of nocodazole did not significantly inhibit the EGF-induced activation of p70 S6k (Fig. 3B). Thus, these control experiments show that the low p70 S6k activity measured in nocodazoletreated cells in Fig. 2B is not an artifact of drug treatment.
Additional control experiments were performed to determine whether kinase activation might be due to withdrawal of nocodazole rather than to a specific cell cycle change. Removal of FIG. 2. p70 S6k and p42 mapk /p44 mapk in cells progressing from M phase into G 1 . Mitotic cells were collected after nocodazole treatment as described under "Materials and Methods" and reseeded into drugfree medium at t ϭ 0 h. A, entry into G 1 was followed by analyzing cellular DNA content by flow cytometry. B, at the indicated times, cell extract supernatants were made and p70 S6k (E) and p42 mapk (Ⅺ) activity was measured in immunocomplex kinase assays. C, the amount of expressed protein and the phosphorylation state of p70 S6k (upper panel) and MAP kinases (lower panel) were determined by Western analysis.
FIG. 3. Effect of nocodazole on S6 kinase activity. A, mitotic cells were collected by shake-off after treatment with (ϩ) or without (Ϫ) nocodazole (see "Materials and Methods"). Total S6 kinase activity was measured in extract supernatants prepared from mitotic cells and from cells that were left on the plates after the shake-off (non-mitotic cells). B, confluent cells were treated with (ϩ) or without (Ϫ) different concentrations of nocodazole for 30 min. Then the cells were treated with or without 5 nM EGF for 20 min. Total S6 kinase activity was measured. nocodazole from cycling cells, resting cells, or cells in S phase after 4 h of treatment did not significantly increase p70 S6k or p42 mapk /p44 mapk activity (data not shown). Furthermore, release of cells from a hydroxyurea-induced S phase arrest did not lead to kinase activation (data not shown). Therefore, the activation of p70 S6k and p42 mapk /p44 mapk seen in Fig. 2B is probably not a general phenomenon seen whenever cells are released from a cell cycle block.
To further pinpoint when during M or G 1 p70 S6k and p42 mapk /44 mapk become activated, shorter time points after release from the metaphase block were examined (Fig. 4A). Tubulin staining (red) and DAPI staining of DNA (blue) were done to follow the cells through the different stages of mitosis and to determine when cells entered G 1 . The immunofluorescence pictures show that the metaphase-arrested cells had condensed DNA but no mitotic spindles (Fig. 4A, t ϭ 0 min). Fifteen min after release from the nocodazole block the mitotic spindles had reformed (Fig. 4A). Thirty min after release the sister chromatids started to separate and 15 min later the cells were in anaphase. One hour after release the cells had gone through telophase and cytokinesis and had entered G 1 .
Examination of p70 S6k on Western blots showed that a minor fraction of p70 S6k was phosphorylated during anaphase (Fig.  4B, t ϭ 45 min), but most of the enzyme became hyperphosphorylated after cells had gone through cytokinesis and entered G 1 (Fig. 4B, t ϭ 60 min). In contrast to p70 S6k , a substantial fraction of p42 mapk /p44 mapk was already highly phosphorylated during anaphase of mitosis (Fig. 4B, t ϭ 45  min). Together, these analyses show that p70 S6k and p42 mapk / p44 mapk display different kinetics of activation in cells progressing from M into G 1 . p70 S6k is activated very early in G 1 and remains active throughout G 1 , whereas p42 mapk /p44 mapk are activated at the end of mitosis and remain active during very early G 1 (Figs. 2 and 4).
Localization of p70 S6k /p85 S6k during the Cell Cycle-p70 S6k and p85 S6k have been reported to be differentially distributed between the nucleus and cytoplasm in cycling cells (17,18). To determine if the localization of p70 S6k /p85 S6k changes during the cell cycle, the enzymes were examined by indirect immunofluorescence. In resting cells a faint staining of both cytoplasm and nucleus was seen (Fig. 5A, green). This distribution remained unchanged in cells stimulated for 45 min with 20% serum and in cells late in G 1 . During S phase the cytoplasm was still faintly stained but dots of p70 S6k /p85 S6k appeared in the nucleus in a pattern similar to that seen with DAPI DNA staining (Fig. 5A, blue). The speckled appearance of p70 S6k / p85 S6k staining in the nucleus was even more striking in G 2 cells (Fig. 5A, green). As the nuclear signal became more intense and less exposure time was required to produce the photographs, staining of the cytoplasm seemed to fade; however, it was still almost the same in G 2 cells as in resting cells. During mitosis most of the p70 S6k /p85 S6k signal overlapped with DAPI staining but the cytoplasm was still faintly stained (Fig. 5A).
We also examined the distribution of p70 S6k /p85 S6k in cells released from a nocodazole block. As was seen in Fig. 5A, the kinase colocalized with DNA during mitosis (Fig. 5B). In very early G 1 , after cytokinesis had occurred and the DNA had decondensed, the speckled S6 kinase pattern appeared in the nucleus (Fig. 5B, t ϭ 2 h) and then disappeared 1 h later. Thus, p70 S6k /p85 S6k is localized in the cytoplasm at all times during the cell cycle but becomes enriched at certain locations in the nucleus during S/G 2 phase and early G 1 in cells released from a metaphase block.
Constitutive Signaling to p70 S6k during the Cell Cycle-p70 S6k remains active to various extents during the entire cell cycle (Fig. 1B). One explanation for this might be that the activated kinase is stable and remains active for hours. Alternatively, the activity of the kinase might depend on constitutive signaling from extracellular growth factors. To discriminate between these two possibilities, mitogens were removed from cells in different stages of the cell cycle and the effect on p70 S6k activity was examined. Serum-starved cells were induced to enter the cell cycle synchronously as described before. As a control, one set of cells was left without any further treatment. A second set of cells was incubated at various times without FCS for 10 min, while a third set was incubated first without FCS and then with FCS for 20 min. Removal of serum after 20 min of mitogen stimulation did not lead to a decrease in p70 S6k activity (Fig. 6A, compare light gray with dark gray bars). However, withdrawal of growth factors at later times or from asynchronously cycling cells led to a significant reduction in p70 S6k activity (Fig. 6A). Western analysis showed that the decrease in p70 S6k activity correlated with a loss of the most highly phosphorylated form of the kinase (Fig. 6B). The kinase could be reactivated at any time by readdition of FCS; however, it was only activated to the level that is characteristic for a particular cell cycle stage, and never to the high level seen in early G 1 (Fig. 6A, compare dark gray and open bars). These data show that signaling to p70 S6k is constitutively on throughout the cell cycle. The reactivation data suggest that a negative regulatory mechanism is present during later parts of the cell cycle or that a component of the S6 kinase signaling pathway becomes limiting.
Involvement of Carboxyl-terminal Phosphorylation Sites in Regulating p70 S6k Activity during the Cell Cycle-p70 S6k was activated to a lower extent at the M/G 1 transition as compared to G 0 /G 1 (Figs. 1B and 2B), suggesting the possibility that different phosphorylation sites might be involved in activating Nocodazole-arrested cells were induced to enter G 1 (see "Materials and Methods"). A, at the indicated times after release, cells were fixed with paraformaldehyde and microtubules were stained with a monoclonal antibody against tubulin (red; see "Materials and Methods"). DNA was stained with DAPI (blue). B, cell extract supernatants were prepared at the indicated times and p70 S6k (upper panel) and MAP kinases (lower panel) were examined on Western blots (see "Materials and Methods"). the kinase during these two stages of cell cycle. As the carboxylterminal phosphorylation sites display a Ser/Thr-Pro motif that is recognized by cell cycle-regulated kinases (21), we examined the contribution of these phosphorylation sites to the cell cycle regulation of p70 S6k activity. A mutant p70 S6k was constructed in which the three serines were mutated to aspartic acid and the threonine to glutamic acid (Fig. 7A). These changes were introduced to mimic phosphorylation. To distinguish between endogenous and exogenous kinase, an HA epitope tag was added to the amino terminus of the protein (Fig. 7A). Constructs encoding the tagged wild-type and mutant p70 S6k were used to produce fibroblasts stably expressing these proteins. Sequential immunoprecipitation of cell extracts with antibodies to the HA tag and then to p70 S6k , followed by S6 kinase assays of the immunoprecipitates, suggested that the recombinant proteins were present in low amounts as compared to the endogenous p70 S6k (see legend to Fig. 7B). In addition, the exogenous kinases were not detectable on Western blots probed with HA antibodies (data not shown).
If the four carboxyl-terminal phosphorylation sites are responsible for regulating the activity of p70 S6k during the cell cycle, the mutant kinase should display the same level of activity at all times. To test this prediction, mitotic and G 1 populations of fibroblasts were collected and the tagged wild-type and mutant p70 S6k were assayed in HA immunoprecipitates. As expected, the activity of recombinant wild-type p70 S6k was low in mitotic cells and increased when cells entered G 1 (Fig.  7B). However, the mutant enzyme also became more active as cells moved from M phase into G 1 . No p70 S6k activity was immunoprecipitated from cells transfected with the empty vector (Fig. 7B). To further characterize the behavior of the mutant p70 S6k , we examined its ability to respond to mitogens in G 0 cells. Confluent fibroblasts were treated with or without EGF and the tagged kinases were assayed in HA immunoprecipitates. Both the wild-type and mutant kinases were activated upon addition of EGF (Fig. 7B). The degree of EGFinduced activation of the HA-tagged kinases (2.8 -4.0 fold; Fig.  7B) was significantly lower than that seen with endogenous p70 S6k in the parental cells (Fig. 3B). We therefore measured the activity of endogenous p70 S6k in the transfected cells by adding p70 S6k antibodies to supernatants that had been precleared with HA antibody. Assay of these immunoprecipitates showed that activation of the endogenous p70 S6k in pMV7transfected cells was also reduced (2.2-5.1 fold; see legend to Fig. 7B).
Having established that the mutant p70 S6k could still be activated during M/G 1 and G 0 /G 1 , we asked if the mutant enzyme was also sensitive to negative regulators of the p70 S6k pathway such as wortmannin (27) and rapamycin (14). Indeed, treatment of transfected fibroblasts with wortmannin or rapamycin completely abolished the EGF-induced activation of both mutant and wild-type p70 S6k (Fig. 7B). In addition, similar to the results obtained with endogenous p70 S6k in the parental cells (Fig. 6), withdrawal of FCS from transfected cells in S phase led to a rapid decline in S6 kinase activity of the mutant protein (Fig. 7B). Thus, the apparently normal regulation of the p70 S6k mutant suggests that the carboxyl-terminal phospho- rylation sites alone do not regulate p70 S6k activity at the G 0 /G 1 boundary or at other stages during the cell cycle.

DISCUSSION
Cell Cycle Regulation of p70 S6k and p42 mapk /p44 mapk Activity-We show here using highly synchronized populations of Swiss mouse 3T3 fibroblasts that the activities of p70 S6k and p42 mapk /p44 mapk are regulated in a cell cycle-dependent manner. The kinases are activated not only at the G 0 /G 1 boundary, but also in cells progressing from M into G 1 after release from a metaphase block (Figs. 1 and 2). p70 S6k was activated 20-fold when cells entered G 1 from a quiescent state and the activity remained relatively high throughout G 1 . During S and G 2 p70 S6k activity decreased slowly so that during M phase a low level of S6 kinase activity remained (Fig. 1B, closed circles). The kinase was reactivated 10-fold when mitotic cells entered G 1 of the next cell cycle (Fig. 2B, closed circles). We have also found that p70 S6k is activated at the M/G 1 transition in FDCP-1 cells synchronized by centrifugal elutriation (data not shown). MAP kinases in G 0 fibroblasts were activated transiently in early G 1 and then were reactivated at the end of mitosis ( Fig. 1B and 2B, open squares).
Our data suggest that there is a difference in the way that p70 S6k and p42 mapk /p44 mapk are activated in G 0 /G 1 cells as compared to M/G 1 cells. The kinases in G 0 are able to respond immediately to signals generated by external growth factors. By contrast, the enzymes in mitotic cells are maintained in an inactive state despite the presence of abundant growth factors in the medium, implying the existence of a negative regulatory mechanism. We have found that when metaphase-arrested cells are released into serum-free medium, mitosis is completed but p70 S6k is not activated (data not shown). Thus, activation of p70 S6k at M/G 1 is triggered by the presence of external growth factors and not by the completion of mitosis per se. Completion of mitosis appears to suppress the negative regulatory mechanism, thus creating an environment that is permissive for kinase activation. This negative regulation of p70 S6k and p42 mapk /p44 mapk might be mediated by a component of the cell cycle machinery such as a cyclin, which accumulates until the end of M phase and is then rapidly degraded.
p70 S6k and p42 mapk /p44 mapk were activated more strongly during the G 0 /G 1 transition (Fig. 1B) than during M/G 1 (Fig.  2B). This difference is probably linked to the differential protein synthesis requirement of these cells. Quiescent cells contain fewer ribosomes and synthesize proteins at a lower rate than cycling cells (41). A sustained increase in protein synthesis is required for resting cells to synthesize new ribosomes and other proteins essential for entry into S phase (42), and as a result the G 1 phase is approximately 4 h longer than G 1 of cycling cells. Since S6 phosphorylation enhances the synthesis of certain proteins involved in translation (2), it seems consistent that cells moving from a quiescent state into S phase would require higher levels of S6 kinase activity than M/G 1 cells. In addition, an essential function of MAP kinases in G 1 might be to phosphorylate an inhibitory subunit of translation initiation factor eIF-4E, thereby stimulating the overall rate of protein synthesis in response to mitogens (43).
Similar to our results, Tamemoto et al. (44) found that p42 mapk /p44 mapk activity is low in nocodazole-arrested Chinese hamster ovary cells and that the enzymes become active upon entry into G 1 . Furthermore, p42 mapk /p44 mapk were re-activated at around M phase of the next cell cycle (44). The interpretation of these results was that p42 mapk /p44 mapk are activated biphasically, first in G 1 and then in G 2 /M before the nocodazole arrest point. However, loss of synchrony by the end of the first cell cycle would make it impossible to determine whether activation occurred in G 2 /M or M/G 1 . It has been proposed that activation of MAP kinases might be a one-time event involved in releasing cells from an arrested state, rather than a recurring event required for progression through each cell cycle (45). This hypothesis is based mainly on experiments done with oocytes and early embryos. However, MAP kinase activation at M/G 1 in fibroblasts (Figs. 2 and 4) and Chinese hamster ovary cells (44) suggests that the requirement for MAP kinase activity during G 1 might be different in rapidly dividing embryonic cells and in established cell lines.
Nuclear Localization of p70 S6k /p85 S6k -Compartmentation is known to be an important mechanism that can regulate protein function. The presence of a nuclear localization signal in p85 S6k but not p70 S6k has prompted a number of studies examining the intracellular distribution of these enzymes (17,18,20). A summary of the results obtained is that while p70 S6k is mainly cytoplasmic and p85 S6k is mainly nuclear, their presence in the alternative compartment cannot be ruled out. Our examination of the distribution of p70 S6k /p85 S6k during the cell cycle revealed that bright speckles of p70 S6k /p85 S6k staining appeared in the nucleus of S phase cells, yielding an appearance similar to that provided by DAPI DNA staining (Fig. 5A). Staining of the cytoplasm remained about the same, so the increased signal in the nucleus might be due to concentration of nuclear p70 S6k /p85 S6k into localized spots, as opposed to an influx of cytoplasmic enzyme. Since the antibody used in these experiments detects both p70 S6k and p85 S6k , no statement can be made about which isoform is responsible for the increased nuclear signal. However, we have detected a major band corresponding to p70 S6k on Western blots of cytoplasmic and nuclear fractions prepared from cells in different phases of the cell cycle (data not shown). During mitosis p70 S6k /p85 S6k colocalized with chromosomes (Fig. 5). Components of small nuclear ribonucleoprotein particles show a speckled staining pattern FIG. 6. Effect of serum withdrawal on S6 kinase activity. A, fibroblasts were synchronized in G 0 by serum starvation as described under "Materials and Methods." At t ϭ 0 h 20% FCS was added to induce cell cycle entry. Synchronized or asynchronously cycling fibroblasts received no further treatment (control), or at various times were incubated for 10 min without FCS or for 10 min without FCS and then 20 min with FCS. Total S6 kinase activity was determined in cell extract supernatants. B, Western blot analysis of p70 S6k in cell extract supernatants prepared in A. Numbers refer to treatments in A. similar to the one seen here, but the speckles do not overlap with DAPI staining and the small nuclear ribonucleoprotein proteins do not migrate with DNA during mitosis (46). Therefore, p70 S6k /p85 S6k do not colocalize with RNA splicing centers but rather with heterochromatin, which is stained by DAPI. Interestingly, the speckly distribution of p70 S6k /p85 S6k in the nucleus did not seem to correlate with S6 kinase activity or with the overall phosphorylation state of p70 S6k , since the nuclear speckles were seen in both G 2 and early G 1 cells (Figs. 1, 2, and 5).
Ribosomal protein S6 is present in both the cytoplasm and nucleus, where it is found in nucleoli and in association with chromatin (19). The nucleolar pool of S6 has been shown to become phosphorylated in response to treatment of cells with phorbol esters (19). Since protein synthesis does not take place in the nucleus, the function of phosphorylated S6 protein in this compartment remains obscure. Although it has been shown that p85 S6k activity in the nucleus is required for entry into S phase (17), it is not known whether the essential function of this enzyme is to phosphorylate S6 or another substrate.
Constitutive Signaling to p70 S6k -p70 S6k activity depends on permanent signaling from extracellular growth factors. Removal of mitogens at any time during the cell cycle leads to a rapid inactivation of the enzyme that can be reversed by adding back growth factors (Fig. 6). Thus, the components of at least one signaling pathway leading to p70 S6k are always present and capable of transducing signals. However, the kinase cannot be reactivated to the high level measured in G 1 cells, indicating that signaling is attenuated during later stages of the cell cycle. Similar results were obtained with p42 mapk /p44 mapk (data not shown). Cell cycle-dependent attenuation of the kinase activation pathways could be accomplished by several mechanisms.
First, the activity of growth factor receptors might be regulated during the cell cycle. It has been shown in receptor overexpression experiments that there is a strong correlation between insulin receptor kinase activity and the activities of kinases lying downstream in the signaling cascade (47). Receptor activity could be down-regulated by a decrease in receptor number mediated by ligand-induced receptor internalization and degradation (48) or by a growth-dependent repression of receptor transcription (49). Alternatively, the activity of growth factor receptors might be reduced by production of inhibitors (50) or by post-translational modification (51). Another explanation for the attenuation of signaling to p70 S6k might be that a component in the signaling cascade downstream of the receptor becomes rate-limiting. Finally, a negative regulator of p70 S6k such as a phosphatase or an inhibitory subunit might become synthesized or activated as cells proceed through the cell cycle.
Involvement of Carboxyl-terminal Phosphorylation Sites in the Cell Cycle Regulation of p70 S6k -Phosphorylation-dephosphorylation is the only mechanism known at this time to regulate p70 S6k /p85 S6k activity (21,40). It was proposed that phosphorylation of four amino acids at the carboxyl terminus of p70 S6k might be responsible for mitogen-induced enzyme activation at G 0 /G 1 (21). However, a recent publication (52) that appeared after completion of this work showed that a p85 S6k mutant with 104 amino acids deleted from the carboxyl terminus could still be activated by FCS in G 0 cells and inhibited by rapamycin and wortmannin. Furthermore, it was proposed that kinase activation might result from the phosphorylation of Thr-252 in response to a signal generated by phosphatidylinositol 3-kinase (53). Our results using a p70 S6k molecule with more subtle point mutations confirm that p70 S6k can be activated independently of the carboxyl-terminal phosphorylation FIG. 7. Regulation of mutant p70 S6k during the cell cycle. A, structure of expressed S6 kinase molecules. Mutant S6 kinase contains three Ser 3 Asp and one Thr 3 Glu changes in the carboxyl-terminal phosphorylation sites. Both wild-type and mutant proteins have an HA epitope at the amino terminus. B, S6 kinase activity in extract supernatants from fibroblasts transfected with wild-type p70 S6k (pMV7ϩ wt S6K), mutant p70 S6k (pMV7ϩ mutant S6K), or the pMV7 vector alone (pMV7) was measured in HA immunoprecipitates (see "Materials and Methods"). Mitotic cells (M) were collected after nocodazole treatment as described under "Materials and Methods" and G 1 cells were obtained 2 h after reseeding mitotic cells into drug-free medium. Contact-inhibited cells were treated for 15 min without (G 0 ) or with (EGF) 5 nM EGF. Contact-inhibited cells were also pretreated for 15 min with either 400 nM wortmannin (EGFϩ WM) or 100 nM rapamycin (EGFϩ Rapa) before addition of EGF. S phase cells were collected 16 h after addition of mitogens to serum-starved cells as described under "Materials and Methods" (Sϩ FCS) and were incubated for 10 min with DMEM in the absence of FCS (Sϩ DMEM). The autoradiograph shows incorporation of 32 P i into S6 during the kinase assay. Similar results were obtained in several independent experiments (data not shown). The 32 P i in S6 was quantitated by scintillation counting; results from two experiments were averaged and values for pMV7 were subtracted as background. For pMV7ϩ wt S6K: 486 cpm (M), 1720 cpm (G 1 ), 635 cpm (G 0 ), 2519 cpm (EGF), 62 cpm (EGFϩ WM), 76 cpm (EGFϩ Rapa), 6948 cpm (Sϩ FCS) and 1910 cpm (S ϩ DMEM). For pMV7ϩ mutant S6K: 225 cpm (M), 469 cpm (G 1 ), 175 cpm (G 0 ), 482 cpm (EGF), 10 cpm (EGF ϩ WM), 128 cpm (EGFϩ Rapa), 855 cpm (S ϩ FCS), and 154 cpm (S ϩ DMEM). A portion (one-fifth volume) of the supernatants that had been subjected to precipitation with HA antibodies were then immunoprecipitated with antibodies to p70 S6k . Endogenous S6 kinase was assayed in the p70 S6k immunoprecipitates (see "Material and Methods") and quantitated by scintillation counting. For pMV7ϩ wt S6K: 2010 cpm (G 0 ) and 10,350 cpm (EGF). For pMV7ϩ mutant S6K: 5745 cpm (G 0 ) and 12,525 cpm (EGF). For pMV7: 4000 cpm (G 0 ) and 12,540 cpm (EGF). sites in G 0 /G 1 . In addition, our examination of the behavior of this p70 S6k mutant at M/G 1 and during S phase suggests that these phosphorylation sites are also not involved in controlling kinase activity during other phases of the cell cycle (Fig. 7B). The specific mechanisms that contribute to the cell cycle-dependent regulation of p70 S6k activity remain to be determined.
We have shown that the behavior of p70 S6k /p85 S6k and p42 mapk /p44 mapk is cell cycle-regulated. Because the activities of these enzymes are sensitive to changes in the growth factor supply, one might imagine that the kinases are part of a sensory system that evaluates growth conditions and makes the decision to exit or remain in the cell cycle. A tightly controlled interplay between these signal transduction molecules and the cell cycle machinery might be important for ensuring proper cell growth and proliferation.