Phosphorylation at Serine 10, a Major Phosphorylation Site of p27 Kip1 , Increases Its Protein Stability*

The association of the p27 Kip1 protein with cyclin and cyclin-dependent kinase complexes inhibits their kinase activities and contributes to the control of cell proliferation. The p27 Kip1 protein has now been shown to be phosphorylated in vivo, and this phosphorylation reduces the electrophoretic mobility of the protein. Substitution of Ser10 with Ala (S10A) markedly reduced the extent of p27 Kip1 phosphorylation and prevented the shift in electrophoretic mobility. Phosphopeptide mapping and phosphoamino acid analysis revealed that phosphorylation at Ser10 accounted for ∼70% of the total phosphorylation of p27 Kip1 , and the extent of phosphorylation at this site was ∼25- and 75-fold greater than that at Ser178 and Thr187, respectively. The phosphorylation of p27 Kip1 was markedly reduced when the positions of Ser10 and Pro11 were reversed, suggesting that a proline-directed kinase is responsible for the phosphorylation of Ser10. The extent of Ser10phosphorylation was markedly increased in cells in the G0-G1 phase of the cell cycle compared with that apparent for cells in S or M phase. The p27 Kip1 protein phosphorylated at Ser10 was significantly more stable than the unphosphorylated form. Furthermore, a mutant p27 Kip1 in which Ser10 was replaced with glutamic acid in order to mimic the effect of Ser10 phosphorylation exhibited a marked increase in stability both in vivo and in vitro compared with the wild-type or S10A mutant proteins. These results suggest that Ser10 is the major site of phosphorylation of p27 Kip1 and that phosphorylation at this site, like that at Thr187, contributes to regulation of p27 Kip1 stability.

The association of the p27 Kip1 protein with cyclin and cyclin-dependent kinase complexes inhibits their kinase activities and contributes to the control of cell proliferation. The p27 Kip1 protein has now been shown to be phosphorylated in vivo, and this phosphorylation reduces the electrophoretic mobility of the protein. Substitution of Ser 10 with Ala (S10A) markedly reduced the extent of p27 Kip1 phosphorylation and prevented the shift in electrophoretic mobility. Phosphopeptide mapping and phosphoamino acid analysis revealed that phosphorylation at Ser 10 accounted for ϳ70% of the total phosphorylation of p27 Kip1 , and the extent of phosphorylation at this site was ϳ25and 75-fold greater than that at Ser 178 and Thr 187 , respectively. The phosphorylation of p27 Kip1 was markedly reduced when the positions of Ser 10 and Pro 11 were reversed, suggesting that a proline-directed kinase is responsible for the phosphorylation of Ser 10 . The extent of Ser 10 phosphorylation was markedly increased in cells in the G 0 -G 1 phase of the cell cycle compared with that apparent for cells in S or M phase. The p27 Kip1 protein phosphorylated at Ser 10 was significantly more stable than the unphosphorylated form. Furthermore, a mutant p27 Kip1 in which Ser 10 was replaced with glutamic acid in order to mimic the effect of Ser 10 phosphorylation exhibited a marked increase in stability both in vivo and in vitro compared with the wild-type or S10A mutant proteins. These results suggest that Ser 10 is the major site of phosphorylation of p27 Kip1 and that phosphorylation at this site, like that at Thr 187 , contributes to regulation of p27 Kip1 stability.
Progression of the cell cycle in all eukaryotic cells depends on the activity of a series of kinase complexes composed of cyclins and cyclin-dependent kinases (CDKs). 1 The activity of cyclin⅐CDK complexes is regulated by various mechanisms, including association of the kinase subunit with the regulatory cyclin subunit, phosphorylation-dephosphorylation of the kinase subunit, and association of the complex with a group of CDK inhibitors (CKIs) (1,2). The interaction of CKIs with cyclin⅐CDK complexes is triggered by a variety of antimitogenic signals and results in inhibition of the catalytic activity of the complexes and consequent restraint of cell cycle progression. CKIs are classified into two families on the basis of their amino acid sequence similarity and putative targets (3,4). The Cip or Kip family comprises p21 Cip1 (also known as Waf1, Sdi1, and CAP20), p27 Kip1 , and p57 Kip2 , each of which possesses a conserved domain, termed the CDK-binding inhibitory domain, at its NH 2 terminus. The Ink4 family consists of p16 Ink4A , p15 Ink4B , p18 Ink4C , and p19 Ink4D , and its members each contain four tandem repeats of an ankyrin motif. Whereas members of the Ink4 family inhibit the activity of CDK4 or CDK6 specifically, members of the Cip-Kip family show a broad spectrum of inhibitory effects on cyclin⅐CDK complexes.
The p27 Kip1 protein plays a pivotal role in the control of cell proliferation (5,6). Transition from G 1 phase to S phase of the cell cycle is promoted by G 1 cyclin⅐CDK complexes, and p27 Kip1 inhibits the activities of these complexes directly by binding to them. In normal cells, the amount of p27 Kip1 is high during G 0 -G 1 phase, but it rapidly decreases on reentry into S phase triggered by specific mitogenic factors (7,8). Forced expression of p27 Kip1 results in cell cycle arrest in G 1 phase (5,6), and conversely, inhibition of p27 Kip1 expression by antisense oligonucleotides increases the number of cells in S phase (9). Moreover, mice with a homozygous deletion of the p27 Kip1 gene are larger than normal mice and exhibit multiple organ hyperplasia and a predisposition to spontaneous and radiation-or chemical-induced tumors (10 -13).
The concentration of p27 Kip1 is thought to be regulated predominantly by posttranslational mechanisms (14,15). We recently showed that p27 Kip1 is degraded by both the ubiquitinproteasome pathway and ubiquitin-independent proteolytic cleavage (16). Regulation of ubiquitin-mediated proteolysis is often achieved by phosphorylation of the target protein, which renders it more susceptible to degradation (17)(18)(19)(20)(21). Such may also be the case with p27 Kip1 , given that its down-regulation is promoted by its phosphorylation on Thr 187 by the cyclin E⅐CDK2 complex (22)(23)(24). Recent data have also suggested that Fbl1 (also known as Skp2), an F-box protein that is thought to function as the receptor component of an SCF(Skp1/Cul1/F-box protein) ubiquitin ligase complex, binds to p27 Kip1 only when Thr 187 is phosphorylated; such binding then results in the ubiquitination and degradation of p27 Kip1 (25)(26)(27).
Various kinases, such as mitogen-activated protein kinases (MAPKs) and CDKs, may trigger the degradation of p27 Kip1 in response to different upstream signaling pathways. For example, activation of members of the MAPK family is mediated through Ras (28), whereas rapid activation of cyclin E-CDK2 results from the induction of Myc (29,30). Kaposi's sarcoma herpes virus also destabilizes p27 Kip1 through phosphorylation of Thr 187 by the complex of the virus cyclin (K-cyclin) and CDK6 (31,32). These observations indicate that phosphorylation of p27 Kip1 controls its stability. However, because most studies have focused on the role of phosphorylation of Thr 187 in p27 Kip1 stability, little is known about the potential roles of other phosphorylation sites of this protein.
We now show that p27 Kip1 is phosphorylated on many sites, including Thr 187 , in vivo, with the predominant phosphorylation site being Ser 10 . Phosphorylation of Ser 10 is regulated in a cell cycle-dependent manner and may function to stabilize p27 Kip1 . Given that the level of phosphorylation of Ser 10 is substantially greater than that apparent at other phosphorylation sites, phosphorylation-dephosphorylation of p27 Kip1 at Ser 10 may be critical for regulation of cell cycle progression from the resting state to proliferation.

EXPERIMENTAL PROCEDURES
Cell Culture and Synchronization-293T, COS-7, and HeLa cells were cultured at 37°C and under an atmosphere of 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM) (Life Technologies, Inc.) supplemented with 10% (v/v) fetal bovine serum (FBS) (Life Technologies, Inc.). NIH 3T3 cells were cultured in DMEM supplemented with 10% (v/v) calf serum (Life Technologies, Inc.). For analysis of synchronized cells, HeLa or NIH 3T3 cells were arrested at G 0 -G 1 phase by subjecting them to contact inhibition during culture to confluence and to serum deprivation with medium supplemented with 0.1% FBS or calf serum, respectively. Cells were arrested in S phase by exposure to aphidicolin (1 g/ml) as described by Fang et al. (33). For analysis of cells in M phase, HeLa cells were arrested in aphidicolin-containing medium for 16 h, washed with phosphate-buffered saline, and then incubated in aphidicolin-free medium for 3 h. They were subsequently incubated with nocodazole (100 ng/ml) for 12-15 h to induce arrest at M phase, after which culture dishes were shaken and floating cells were harvested for recovery of only those cells in M phase.
Construction of Plasmids and Site-directed Mutagenesis-Complementary DNAs encoding all p27 Kip1 derivatives were prepared from the human p27 Kip1 cDNA, kindly provided by M. Nakanishi. The p27 Kip1 mutants were generated by replacing Ser 10 , Ser 178 , or Thr 187 with Ala (S10A, S178A, and T187A, respectively), or replacing Ser 10 with Glu (S10E) with the use of a QuickChange site-directed mutagenesis kit (Stratagene, La Jolla). Proteins tagged at their NH 2 termini with the FLAG epitope were generated with the use of the polymerase chain reaction as performed with the high fidelity thermostable DNA polymerase KOD (Toyobo, Tokyo, Japan). The sequences of all mutant cDNAs were confirmed in their entirety. The cDNAs encoding the various p27 Kip1 proteins, with or without the FLAG epitope tag, were then subcloned into pcDNA3 (Invitrogen, Carlsbad, CA) for transfection experiments or into pGEX6P (Amersham Pharmacia Biotech) for production in bacteria of glutathione S-transferase (GST) fusion proteins.
Alkaline Phosphatase Treatment of p27 Kip1 -Immunoprecipitates containing p27 Kip1 were washed thoroughly three times with ice-cold lysis buffer and once with lysis buffer without phosphatase inhibitors. They were then incubated for 5 h at 37°C in a final volume of 30 l containing 40 units of calf intestinal alkaline phosphatase (CIAP) (Takara), 50 mM Tris-HCl (pH 9.0), and 1 mM MgCl 2 . The reaction mixture was then subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblot analysis with antibodies to (anti-) p27 Kip1 . 32 P i Labeling of p27 Kip1 -Transfected 293T cells were incubated for 2 h in phosphate-free DMEM supplemented with 10% dialyzed FBS and then metabolically labeled for 4 h at 37°C with [ 32 P]P i (Amersham Pharmacia Biotech) at a concentration of 1 mCi/ml in the same medium. After extensive washing of the cells in isotope-free medium, they were then lysed and subjected to immunoprecipitation with anti-FLAG or anti-p27 Kip1 . The immunoprecipitates were fractionated by SDS-PAGE and subjected to autoradiography and quantitative analysis with a BAS-2000 image analyzer (Fuji Film, Kanagawa, Japan).
Phosphopeptide Mapping and Phosphoamino Acid Analysis-32 P-Labeled proteins were prepared for phosphopeptide mapping as described (23). Dried samples were treated with 10 g of trypsin (Roche Molecular Biochemicals) for at least 8 h at 37°C. The reaction mixtures were then lyophilized twice in 0.4-ml volumes of water and finally resuspended in 10 l of pH 1.9 buffer (20 ml of formic acid and 156 ml of glacial acetic acid per 1794 ml of water) prior to application to TLC plates. Electrophoresis and ascending chromatography were performed as described (35) with minor modifications; phosphochromatography buffer (750 ml of n-butanol, 500 ml of pyridine, and 150 ml of glacial acetic acid per 600 ml of water) was used. Plates were air-dried and then subjected to quantitative analysis with a BAS-2000 image analyzer. Phosphoamino acid analysis of tryptic phosphopeptides derived from p27 Kip1 was performed as described (35), with the exception that Multiphor II (Amersham Pharmacia Biotech) was used.
Two-dimensional Gel Electrophoresis and Immunoblot Analysis-Two-dimensional gel electrophoresis (two-dimensional PAGE) with separation in the first dimension by nonequilibrium pH gradient electrophoresis (NEPHGE) was performed as described by O'Farrell et al. (36). Cell lysate containing 0.15 to 0.5 mg of total protein was applied to a NEPHGE tube (130 ϫ 3 mm, inside diameter), gel (4% (w/v) acrylamide, 9.2 M urea, 2% (v/v) Ampholytes (Bio-Lyte, pH 3-10; Bio-Rad), 2% (v/v) Nonidet P-40) and electrophoresis was performed for 5-8 h at 400 V. The separated proteins were then resolved in the second dimension by standard PAGE on a 10% gel, which was subsequently subjected to immunoblot analysis with anti-p27 Kip1 .
In Vitro Degradation Assay-NIH 3T3 cell extracts (S100) were prepared as described (16). Human recombinant p27 Kip1 proteins (0.1 g) or lysate (2 g) of transfected 293T cells were incubated at 37°C for the indicated times in 20 l of a degradation mixture containing 50 mM Tris-HCl (pH 8.3), 5 mM MgCl 2 , 2 mM dithiothreitol, 10 mM ATP, 1 mM phosphocreatine, phosphocreatine kinase (500 units/ml) with or without 2 M okadaic acid, and 10 g of NIH 3T3 cell lysate proteins. The mixture was then subjected to SDS-PAGE on a 12% gel and immunoblot analysis with anti-p27 Kip1 .

Pulse-Chase Experiments-Transfected NIH 3T3 cells were metabolically labeled with [ 35 S]methionine and [ 35 S]cysteine (L-[ 35 S]in vitro
Cell Labeling Mix; Amersham Pharmacia Biotech) at a concentration of 80 Ci/ml for 1 h, and then incubated in isotope-free medium for 0, 3, 6, or 12 h. Cell lysates were prepared and subjected to immunoprecipitation with anti-p27 Kip1 , and the resulting precipitates were subjected to SDS-PAGE on a 12% gel, autoradiography, and quantitative analysis with a BAS-2000 image analyzer.

RESULTS
Phosphorylation of p27 Kip1 in Vivo-The p27 Kip1 protein contains three serine or threonine residues, at positions 10, 178, and 187 (Ser 10 , Ser 178 , and Thr 187 ), that are immediately upstream of proline residues ( Fig. 1A and Table I). We focused on the potential roles of these sites in determining the stability of p27 Kip1 because members of a group of kinases (known as proline-directed kinases) that require a proline immediately downstream of the target serine or threonine residue, and which include MAPKs and CDKs, contribute to mitogenic signaling pathways. We generated cDNAs that encode mutant human p27 Kip1 proteins in which each of the three residues Ser 10 , Ser 178 , and Thr 187 was replaced individually (S10A, S178A, and T187A) or together (S10A/S178A/T187A) with Ala (Fig. 1A). The phosphorylation status of these three sites of p27 Kip1 in vivo was investigated by transiently expressing the FLAG epitope-tagged wild-type and mutant proteins in 293T human embryonic kidney epithelial cells and metabolically labeling the cells with 32 P i . The p27 Kip1 proteins were then immunoprecipitated with anti-FLAG, and the extent of 32 P incorporation was evaluated by autoradiography and image analysis and normalized by the amount of p27 Kip1 protein estimated by immunoblot analysis of the immunoprecipitates with anti-p27 Kip1 (Fig. 1, B and C). The amount of 32 P incorporated by the S10A mutant or by the S10A/S178A/T187A triple mutant was ϳ30% that incorporated by wild-type p27 Kip1 , whereas that incorporated by the S178A or T187A mutants was virtually identical to that incorporated by the wild-type protein. These results indicated that Ser 10 is the major phosphorylation site of p27 Kip1 (accounting for ϳ70% of the total extent of p27 Kip1 phosphorylation).
Immunoblot analysis with anti-p27 Kip1 of wild-type p27 Kip1 expressed in cultured cells revealed that these antibodies recognized two bands, suggesting that the lower mobility band might correspond to phosphorylated p27 Kip1 (Fig. 1B and Fig.  2A; the two bands are more evident in the latter as a result of a difference in composition of the acrylamide gel). This electrophoretic mobility shift was apparent for p27 Kip1 expressed not only in 293T cells but also in HeLa (human cervical cancer), COS-7 (monkey kidney epithelial), and NIH 3T3 (mouse fibroblast) cells ( Fig. 2A). For all cells tested, mutation of Ser 10 of p27 Kip1 to Ala resulted in the disappearance of the more slowly migrating band. To confirm that the observed mobility shift was attributable to phosphorylation of p27 Kip1 , we expressed wild-type p27 Kip1 or the S10A mutant in 293T cells, immunoprecipitated the recombinant protein, and treated it with CIAP. Treatment with CIAP resulted in the disappearance of the lower mobility form of wild-type p27 Kip1 , but it had virtually no effect on the mobility of the S10A mutant (Fig. 2B). These results thus suggested that phosphorylation at Ser 10 was responsible for the observed shift in the electrophoretic mobility of p27 Kip1 and that the kinase or kinases that catalyze this reaction are present in cells from various tissues and species.
Cell Cycle-dependent Phosphorylation of p27 Kip1 on Ser 10 -To investigate the biological role of phosphorylation of p27 Kip1 on Ser 10 , we examined whether the phosphorylation status of this residue is dependent on phase of the cell cycle. Asynchronous NIH 3T3 cells were transfected with an expression plasmid encoding FLAG-tagged wild-type p27 Kip1 or its S10A mutant, and cell lysates were subjected to two-dimensional PAGE and immunoblot analysis with anti-p27 Kip1 in order to quantify the extent of phosphorylation at Ser 10  The cyclin binding domain, CDK binding domain, and nuclear localization signal (NLS) are indicated. B and C, FLAG-tagged wild-type (WT) p27 Kip1 or S10A, S178A, T187A, or S10A/S178A/T187A (triple) mutants of p27 Kip1 were transiently expressed in 293T cells and metabolically labeled by incubation of cells with 32 P i . Cell lysates (3 mg of protein) were then subjected to immunoprecipitation (IP) with anti-FLAG (␣-FLAG), and the resulting precipitates were subjected to autoradiography (upper panel) or to immunoblot analysis (IB) with anti-p27 Kip1 (␣-p27) (lower panel) (B). The extent of 32 P incorporation into wild-type and mutant p27 Kip1 proteins was then quantified with a BAS-2000 image analyzer and normalized by the abundance of p27 Kip1 revealed by immunoblot analysis (C). The normalized incorporation of 32 P into the wild-type protein is defined as 100%. Data are from an experiment that was repeated three times with similar results.  10 . A, 293T, HeLa, COS-7, or NIH 3T3 cells were transfected with empty expression plasmid alone (mock) or plasmids encoding either FLAG-tagged wild-type (WT) p27 Kip1 or its S10A mutant. Cell lysates (from 10 to 125 g of protein) were subjected to immunoblot analysis with anti-FLAG. Bands corresponding to FLAGtagged unphosphorylated and phosphorylated p27 Kip1 are indicated by FLAG-p27 and FLAG-pp27, respectively. B, FLAG-tagged wild-type p27 Kip1 and its S10A mutant were immunoprecipitated from transfected 293T cells with anti-FLAG, and the resulting immunoprecipitates were incubated for 5 h at 37°C in the absence (Ϫ) or presence (ϩ) of CIAP. The samples were then subjected to immunoblot analysis with anti-p27 Kip1 . 3A). Wild-type p27 Kip1 yielded two immunoreactive spots, the upper, corresponding to the form of the protein phosphorylated on Ser 10 , migrated in a more acidic position on NEPHGE in the first dimension because of the negative charge of the phosphate group; this spot was not detected with the S10A mutant. Endogenous p27 Kip1 exhibited a pattern similar to that of the recombinant wild-type protein, suggesting that phosphorylation at Ser 10 is not an artifact of overexpression.
Immunoblot analysis of synchronized HeLa or NIH 3T3 cells with anti-p27 Kip1 revealed that endogenous p27 Kip1 was abundant in G 0 -G 1 phase of the cell cycle but was present in markedly smaller amounts during S and M phases (Fig. 3B), similar to results previously obtained with many other cell types (7)(8)(9). Two-dimensional PAGE and immunoblot analysis with anti-p27 Kip1 of synchronized HeLa cells revealed that ϳ80% of endogenous p27 Kip1 was phosphorylated at Ser 10 during G 0 -G 1 phase, whereas the amount of this form of the protein was reduced to virtually zero (0.1%) during S phase. In M phase, although the abundance of p27 Kip1 was minimal, a small proportion (16.0%) of the total p27 Kip1 protein was phosphorylated at Ser 10 . Similar results were obtained with NIH 3T3 cells, although the phosphorylation state of p27 Kip1 in M phase could not be estimated because of the "mitotic slippage" apparent in rodent cell lines (37). These observations suggested that phosphorylation of p27 Kip1 on Ser 10 is cell cycle-dependent and that phosphorylation at this site might contribute to regulation of the stability of this protein.
Phosphopeptide Analysis of Phosphorylated p27 Kip1 -To characterize further the phosphorylation status of p27 Kip1 , we performed two-dimensional phosphopeptide mapping of wildtype and mutant p27 Kip1 proteins labeled with 32 P in vivo. The expected length and sequence of tryptic peptides of p27 Kip1 that contain serine or threonine are shown in Table I. Ser 10 is contained in a peptide composed of 10 amino acids, whereas Ser 178 and Thr 187 are both contained in the same 20-residue peptide. We compared the phosphopeptide maps of wild-type p27 Kip1 (with or without the FLAG tag) and its S10A, S178A, T187A, and S10A/S178A/T187A mutants after their immunoprecipitation from transfected 293T cells (Fig. 4A). No differences were detected between the phosphopeptide map of FLAG-tagged wild-type p27 Kip1 and that of the untagged protein. Six or seven radioactive spots were reproducibly detected, four of which (spots 3-6) appeared common to all maps. Two intensely labeled peptides (spots 1 and 2), however, were detected only in the maps of wild-type p27 Kip1 and those of its  10 . A, FLAG-tagged wild-type (WT) p27 Kip1 (upper panel) or its S10A mutant (lower panel) was expressed in NIH 3T3 cells, and cell lysates (200 g of protein) were subjected to two-dimensional PAGE and immunoblot analysis with anti-p27 Kip1 . The directions of electrophoresis (arrows) as well as the positions corresponding to transfected (exo) and endogenous (endo) p27 Kip1 are indicated. B, lysates (50 g of protein) of HeLa or NIH 3T3 cells synchronized in G 0 -G 1 , S, or M phases of the cell cycle were subjected to immunoblot analysis with either anti-p27 Kip1 (upper panel) or anti-␣-tubulin (lower panel). C, lysates of HeLa or NIH 3T3 cells synchronized in G 0 -G 1 (upper panels), S (middle panels), or M (lower panel) phase were subjected to two-dimensional-PAGE and immunoblot analysis with anti-p27 Kip1 . The amount of lysate protein analyzed was varied from 150 to 500 g in order to ensure that the amounts of endogenous p27 Kip1 were similar at the different phases of the cell cycle. The blots of lysates from cells in S or M phases were overexposed. The positions corresponding to unphosphorylated and phosphorylated p27 Kip1 are indicated, as are the amounts of each of these two forms of the protein expressed as a percentage of total p27 Kip1 (determined by image analysis with NIH Image software).

FIG. 4. Two-dimensional tryptic phosphopeptide mapping of wild-type and various mutant p27 Kip1 proteins.
A, wild-type (tagged or not with the FLAG epitope) or S10A, S178A, T187A, or S10A/S178A/T187A mutants of p27 Kip1 were expressed in 293T cells and metabolically labeled with 32 P i . The recombinant proteins were immunoprecipitated with anti-FLAG or anti-p27 Kip1 , and the resulting immunoprecipitates were subjected to two-dimensional tryptic phosphopeptide mapping. Major phosphopeptides are numbered 1-6. Phosphopeptides containing Ser 10 are indicated by open arrowheads, and those containing Ser 178 and Thr 187 are indicated by filled arrowheads. The origin of migration is indicated by an asterisk, and the directions of separation by TLC and electrophoresis are shown by arrows. B, the relative incorporation of 32 P by Ser 10 , Ser 178 , and Thr 187 of p27 Kip1 was estimated by image analysis of autoradiographs of phosphopeptide maps. The extent of 32 P incorporation by Ser 10 was defined as 100%. Because Ser 178 and Thr 187 are both present in the same tryptic peptide, the incorporation of 32 P at each site was calculated from the difference in incorporation into spot 6 (filled arrowheads in A) between wild-type and either S178A or T187A, respectively. Data are from an experiment that was repeated twice with similar results. S178A and T187A mutants and not in those of the S10A or triple mutants. These results suggested that the extent of phosphorylation of p27 Kip1 at Ser 10 in vivo was markedly greater than the extent of phosphorylation at other sites, including Ser 178 and Thr 187 . The observation that the phosphopeptide containing Ser 10 yielded two spots is likely attributable to treatment with performic acid during sample preparation. We also showed that Ser 178 and Thr 187 were contained in spot 6 by phosphopeptide analysis of recombinant p27 Kip1 phosphorylated in vitro by cyclin E-CDK2 (data not shown).
Phosphorylation of p27 Kip1 at Thr 187 by cyclin E-CDK2 is required for its degradation by the ubiquitin-proteasome pathway (22)(23)(24)(25)(26)(27). To estimate the relative amount of 32 P incorporated into p27 Kip1 at Thr 187 , we compared the autoradiographic intensity of the phosphopeptides derived from wild-type p27 Kip1 and its mutants. The amount of radioactivity incorporated into the peptide containing Ser 10 was ϳ75 and 25 times that incorporated by Thr 187 and Ser 178 , respectively (Fig. 4B). This apparent high relative amount of Ser 10 phosphorylation relative to Thr 187 phosphorylation is unlikely to reflect the fraction of p27 Kip1 that becomes phosphorylated at this site because the form phosphorylated on Thr 187 is thought be rapidly degraded.
Phosphoamino Acid Analysis of p27 Kip1 -To identify the phosphorylation sites of p27 Kip1 in vivo, and to confirm the phosphorylation at Ser 10 , Ser 178 , and Thr 187 , we performed phosphoamino acid analysis of seven major phosphopeptides of wild-type p27 Kip1 phosphorylated in 293T cells. The results revealed that peptides 1 and 2, which include Ser 10 , contained only phosphoserine, whereas peptide 6, which includes Ser 178 and Thr 187 , contained phosphoserine and, to a lesser extent, phosphothreonine (Fig. 5). The analysis also revealed that peptide 5 was phosphorylated on serine and to a lesser extent on threonine, whereas peptide 3 was phosphorylated on threonine and to a lesser extent on serine. Peptides 4 and 7 contained exclusively phosphoserine (the phosphorylation of peptide 7 was not detected in Fig. 4A, probably due to experimental variation among culture condition of the cells).
Phosphorylation of p27 Kip1 at Ser 10 by a Proline-directed Kinase-The Ser 10 residue of p27 Kip1 is located immediately upstream of a proline residue (Table I) and is therefore a potential target for proline-directed kinases such as MAPKs or CDKs. Proline possesses a fixed, rigid conformation and serves to reduce the flexibility of proteins at sites of its incorporation. We therefore constructed a p27 Kip1 mutant (S10P/P11S or PS) in which the positions of Ser 10 and Pro 11 were reversed, in order to investigate whether the kinase responsible for phosphorylation of Ser 10 is a proline-directed kinase while minimizing any introduced conformational change. Expression and metabolic labeling with 32 P of the PS mutant in 293T cells revealed that the extent of its phosphorylation was about onesixth of that of the wild-type protein (Fig. 6A). Phosphopeptide mapping also revealed that the extent of phosphorylation of the peptides corresponding to Ser 10 (or Ser 11 in the case of the mutant) was markedly greater for wild-type p27 Kip1 than for the PS mutant (Fig. 6B). Of the proline-directed kinases important in cell cycle control, MAPKs appeared more likely than did CDKs to be responsible for phosphorylation of Ser 10 of p27 Kip1 because CDKs usually require a basic amino acid immediately downstream of the Ser(Thr)-Pro sequence (38,39). Indeed, p27 Kip1 was phosphorylated by p42 MAPK (ERK2) in vitro, and the phosphopeptide map of the protein so phosphorylated was similar to that of p27 Kip1 phosphorylated in vivo (Fig. 6C). In contrast, p27 Kip1 was poorly phosphorylated at Ser 10 by recombinant cyclin E-CDK2 in vitro; rather, it was preferentially phosphorylated on Thr 187 by this kinase complex (data not shown). These data suggested that a proline-directed kinase, possibly a member of the MAPK family, phosphorylates p27 Kip1 on Ser 10 .
We thus investigated the effect on p27 Kip1 phosphorylation in vivo of PD98059 (40), a specific inhibitor of MEK1 and MEK2, which phosphorylate and thereby activate the MAPKs p44 (ERK1) and p42 (ERK2). Immunoblot analysis with anti-p27 Kip1 of 293T cells expressing wild-type p27 Kip1 revealed that the lower mobility band of the p27 Kip1 doublet, which corresponds to the form of the protein phosphorylated on Ser 10 , was detected at similar intensities with cells cultured with either dimethyl sulfoxide (Me 2 SO) (vehicle control) or PD98059 (Fig.  6D). In contrast, the phosphorylated forms of p42 and p44 MAPKs were detected in the cells treated with Me 2 SO but not in those treated with PD98059. These results indicated that the MAPK isoforms p44 (ERK1) and p42 (ERK2) do not phosphorylate p27 Kip1 on Ser 10 in vivo. It remains possible that other MAPKs, such as ERK5, stress-activated protein kinase (or c-Jun NH 2 -terminal kinase), or p38 MAPK, may mediate the phosphorylation of p27 Kip1 on Ser 10 in intact cells. Butyrolactone I (41), a potent inhibitor of CDK1, CDK2, and CDK5, also did not affect the phosphorylation of p27 Kip1 on Ser 10 in vivo (data not shown).
Effect of Mutation of Ser 10 of p27 Kip1 on CDK Inhibitory Activity-We next examined whether mutation of Ser 10 of p27 Kip1 affects the CDK inhibitory activity of the protein. A mutant p27 Kip1 in which Ser 10 was replaced with glutamic acid (S10E), which mimics the negative charge of phosphate (42), was generated. Bacterially expressed wild-type p27 Kip1 and its S10A and S10E mutants were subjected to an in vitro kinase assay either with cyclin E-CDK2 and its substrate histone H1 (Fig. 7A) or with cyclin D2-CDK4 and its substrate Rb protein (Fig. 7B). Each of the three p27 Kip1 proteins inhibited the kinase activity of cyclin E-CDK2 or cyclin D2-CDK4 to similar extents, suggesting that phosphorylation of p27 Kip1 on Ser 10 does not affect the CDK inhibitory function of the protein.
Effect of Phosphorylation of Ser 10 on the Stability of p27 Kip1 in Vitro and in Vivo-Given that the p27 Kip1 protein that accumulates in resting cells is highly phosphorylated on Ser 10 (Fig. 3), we compared the stability of the phosphorylated and unphosphorylated form of p27 Kip1 . Wild-type p27 Kip1 and its S10A mutant were expressed in 293T cells, and the lysates that contained both phosphorylated and unphosphorylated forms of p27 Kip1 protein were subjected to in vitro degradation assay as described under "Experimental Procedures." Phosphorylated p27 Kip1 was relatively stable compared with the unphosphorylated form, whose kinetics of degradation was similar to that of the S10A mutant (Fig. 8). The half-life of the phosphorylated p27 Kip1 was thus increased about 2-fold relative to that of the unphosphorylated form or of the S10A mutant.
Furthermore, we examined the stability of wild-type p27 Kip1 and its S10A and S10E mutants in vitro and in vivo. We previously showed that p27 Kip1 is degraded in NIH 3T3 cell lysates in vitro and in vivo, by both ubiquitination-dependent and -independent pathways, degradation by the latter pathway being apparent by the generation of a 22-kDa intermediate (p27⌬22k) (16). The stability of the S10E mutant in this in vitro degradation assay was markedly increased compared with those of the wild-type protein and the S10A mutant (Fig. 9). However, the observation that both S10A and S10E mutants underwent ubiquitination-independent cleavage suggests that phosphorylation of p27 Kip1 on Ser 10 does not affect such cleavage. We also examined the stability of wild-type p27 Kip1 and its S10A and S10E mutants in intact transfected NIH 3T3 cells. Consistent with the in vitro results, the stability of the S10E mutant was markedly greater than that of either the wild-type protein or the S10A mutant (Fig. 10); the half-life of the S10E mutant was thus increased more than 2-fold relative to that of the wild-type protein. The S10A mutant appeared to be unstable compared with the wild-type protein. The order of stability (S10E Ͼ wild-type Ͼ S10A) might be explained by the possibility that the phosphorylated wild-type protein might be rapidly dephosphorylated in cycling cells. Collectively, these data suggest that phosphorylation of p27 Kip1 on Ser 10 contributes to regulation of the stability of this protein. DISCUSSION Regulation of the cell cycle at the G 1 -S boundary is thought to be important for the control of cell proliferation. Kinase activity associated with two G 1 cyclins, cyclins D and E, is essential for this transition, predominantly because of the requirement for phosphorylation of Rb and the consequent termination of its inhibition of cell cycle progression (1,2). Among the mechanisms responsible for regulation of G 1 cyclin-associated kinase activity, control of the abundance of p27 Kip1 by external mitogenic signals appears important (3,4). The amount of p27 Kip1 is regulated predominantly by posttranslational modification, which affects protein stability, rather than by transcriptional control (14,15). The stability of p27 Kip1 has thus been shown to be affected by ubiquitin-dependent ( and Jab1-dependent (45) degradation.
The phosphorylation state of many proteins affects their stability, and phosphorylation of p27 Kip1 on Thr 187 has been shown to be essential for binding of Fbl1, an F-box protein component of an SCF(Skp1/Cul1/F-box protein) ubiquitin ligase complex (25)(26)(27). Thus, phosphorylation of Thr 187 has been thought to be a central mechanism in control of the stability of p27 Kip1 by ubiquitin-mediated degradation. However, we have now shown that the extent of phosphorylation of p27 Kip1 on Thr 187 represents only ϳ1% of the total extent of phosphorylation of this protein in vivo. In contrast, phosphorylation of Ser 10 accounts for ϳ70% of the total extent of phosphorylation of p27 Kip1 . Furthermore, the extent of phosphorylation at this site is increased in resting cells, and Ser 10 phosphorylation both affects protein stability and was apparent in various types of cells from several species. These data suggest that phosphorylation of Ser 10 may represent another important mechanism by which the stability of p27 Kip1 is regulated. It is of note that the extent of phosphorylation of Thr 187 is almost certainly underestimated since this residue is phosphorylated during a limited period of the cell cycle or if p27 Kip1 phosphorylated at this site is too unstable to be effectively detected by immunoblot analysis or labeling with 32 P. The observation that the abundance of p27 Kip1 is increased in cells of Fbl1-deficient mice (46) suggests that phosphorylation of Thr 187 is indeed an important determinant of this parameter. Although the degradation of p27 Kip1 was slower in Fbl1-deficient cells than in wild-type cells, the observation that a substantial extent of p27 Kip1 degradation was still apparent in these cells 2 is consistent with the existence of other pathways for p27 Kip1 degradation.
The increased stability of the Ser 10 -phosphorylated form of p27 Kip1 (Fig. 8) and the S10E mutant, which mimics the Ser 10phosphorylated form of the protein (Figs. 9 and 10), suggests that dephosphorylation of p27 Kip1 at Ser 10 might play an important role in progression of the cell cycle from G 0 -G 1 to S phase. However, both the kinase and phosphatase responsible for the phosphorylation and dephosphorylation at Ser 10 , respectively, as well as the mechanism by which phosphorylation of Ser 10 stabilizes p27 Kip1 , remain to be identified. It will also be important to determine whether such regulation of p27 Kip1 stability is linked to external mitogenic signals. The stability of the protein IB␣ is regulated by two independent mechanisms as follows: phosphorylation at sites near the NH 2 terminus, which is induced by external signals, and phosphorylation at sites near the COOH terminus, which controls the basal turnover rate (47). The signal-induced phosphorylation of IB␣ results in its targeting by the F-box protein Fbw1 (also known as FWD1 or ␤-TrCP) and its consequent ubiquitination-dependent degradation (20, 48 -50). The stability of p27 Kip1 thus might also be subjected to dual regulation by signal-induced phosphorylation at Thr 187 , which recruits the F-box protein Fbl1 and results in ubiquitination-dependent degradation, and by phosphorylation at Ser 10 .
The biochemical activity of p27 Kip1 suggests that the protein functions as a tumor suppressor. Indeed, mice lacking p27 Kip1 are prone to spontaneous tumorigenesis (10 -12). Furthermore, FIG. 8. Effect of phosphorylation of Ser 10 of the stability of p27 Kip1 . A, FLAG-tagged wild-type p27 Kip1 and its S10A mutants were expressed in 293T cells, and their lysates were subjected to an in vitro degradation assay for the indicated times. Subsequently, the reaction mixtures were subjected to immunoblot analysis with anti-p27 Kip1 . The positions corresponding to FLAG-tagged unphosphorylated and phosphorylated p27 Kip1 are indicated as FLAG-p27 and FLAG-pp27, respectively. B, the intensities of the bands corresponding to phosphorylated wild-type p27 Kip1 (open diamonds) and unphosphorylated wild-type p27 Kip1 (filled squares) and S10A mutant (filled circles) in the immunoblots shown in A were quantified and expressed as a percentage of the corresponding value at time 0. Data are from an experiment that was repeated two times with similar results.
FIG. 9. Effect of mutation of Ser 10 on the stability of p27 Kip1 in vitro. A, wild-type p27 Kip1 and its S10A and S10E mutants were expressed in and purified from bacteria and then subjected for the indicated times to an in vitro degradation assay with NIH 3T3 cell lysate. The reaction mixtures were analyzed by immunoblotting with anti-p27 Kip1 . The positions corresponding to unphosphorylated and phosphorylated p27 Kip1 as well as to the p27⌬22k cleavage product are indicated. B, the intensities of the bands corresponding to full-length wildtype p27 Kip1 (open diamonds) and its S10A (filled squares) and S10E (filled circles) mutants in the immunoblots shown in A were quantified and expressed as a percentage of the corresponding value at time 0. Data are from an experiment that was repeated three times with similar results. mice that possess one normal allele of the p27 Kip1 gene develop tumors at a greatly increased frequency (compared with wildtype animals) after exposure to chemical carcinogens or x-rays, without loss of the functional p27 Kip1 allele in the tumor cells (13). Although numerous clinical studies have attempted to identify mutations within the p27 Kip1 locus in individuals with cancer, such mutations have proved to be extremely rare (51)(52)(53)(54)(55)(56)(57)(58)(59). Reduced expression of p27 Kip1 has nevertheless been correlated with poor prognosis in cohorts of individuals with breast, colorectal, or stomach carcinoma (60 -66). Loda et al. (62) showed that tumors with low levels of p27 Kip1 expression exhibited relatively high rates of p27 Kip1 degradation (and vice versa). It is unlikely that this increased degradation of p27 Kip1 was due to nonspecific enhancement of general protein degradation, because degradation of neither p21 Cip1 nor cyclin A was affected in the same cancer patients. The mechanisms that control the stability of p27 Kip1 thus appear important in cancer development. Characterization of these mechanisms should shed light on fundamental issues such as how cell cycle regulation is linked to developmental control and how the disturbance of cell cycle regulation results in carcinogenesis (and may lead to the development of anti-cancer drugs with new modes of action).
FIG. 10. Pulse-chase analysis of the stability of Ser 10 mutants of p27 Kip1 in vivo. A, NIH 3T3 cells transfected with vectors encoding wild-type p27 Kip1 or its S10A or S10E mutants were pulse-labeled with [ 35 S]methionine and [ 35 S]cysteine and then incubated in the absence of isotope for the indicated chase periods. Cell lysates were then subjected to immunoprecipitation with anti-p27 Kip1 , and the resulting precipitates were subjected to SDS-PAGE, autoradiography, and scanning densitometry. B, the intensities of the bands corresponding to wild-type p27 Kip1 (open diamonds) and its S10A (filled squares) and S10E (filled circles) mutants in the autoradiograms shown in A were quantified and expressed as a percentage of the corresponding value for the beginning of the chase period (time 0). Data are from an experiment that was repeated twice with similar results.