Phosphorylation-independent Stabilization of p27 kip1 by the Phosphoinositide 3-Kinase Pathway in Glioblastoma Cells*

The PTEN tumor suppressor gene is a frequent target of somatic mutation, particularly in glioblastoma multiform and prostate cancer. The expression of PTEN in PTEN-mutant glioblastoma cells leads to a cell cycle arrest in G 0 /G 1 that is mediated at least partially by increased p27 kip1 levels. Here we show that p27 kip1 is not regulated by transcriptional control but that p27 kip1 protein shows increased stability after inhibition of the phosphoinositide (PI) 3-kinase pathway. Because p27 kip1 protein stability is known to be regulated by phosphorylation, we have examined modifications in the phosphorylation pattern after PI 3-kinase inhibition. Biochemical evidence sug-gests that p27 kip1 is phosphorylated on several serine residues, including Ser-10 and Ser-178, but that phosphorylation is unaltered by PI 3-kinase activity. This is further confirmed by the inducible expression of p27 kip1 phosphorylation site mutants, suggesting that p27 kip1 is destabilized in a phosphorylation-independent manner by the PI 3-kinase pathway at the G 1 /S transition. One of the most common mutations seen in glioblastoma multiform, the most frequent primary brain tumor, occurs in the tumor suppressor gene phosphatase and tensin homologue deleted on chromosome 10 ( PTEN ) (1). PTEN regulates the activity of the phosphoinositide (PI) 1 3-kinase pathway, a sig-naling Twenty single cell-derived independent drug-resistant colonies of each form of p27 kip1 were cloned and screened for each gene. Exogenous expression levels of each of these genes was monitored by Western blotting, and the highest induced gene expression cell line was chosen for further study. Different ranges of ponas- terone A (PonA) concentration were used to induce p27 kip1 to different degrees in the different cell lines. The amounts varied from 0.3 to 10 (cid:1) M but were different in the cell lines in a manner that was independent of the generation of phosphorylation sites and were probably more related to clonal variability.

The PTEN tumor suppressor gene is a frequent target of somatic mutation, particularly in glioblastoma multiform and prostate cancer. The expression of PTEN in PTENmutant glioblastoma cells leads to a cell cycle arrest in G 0 /G 1 that is mediated at least partially by increased p27 kip1 levels. Here we show that p27 kip1 is not regulated by transcriptional control but that p27 kip1 protein shows increased stability after inhibition of the phosphoinositide (PI) 3-kinase pathway. Because p27 kip1 protein stability is known to be regulated by phosphorylation, we have examined modifications in the phosphorylation pattern after PI 3-kinase inhibition. Biochemical evidence suggests that p27 kip1 is phosphorylated on several serine residues, including Ser-10 and Ser-178, but that phosphorylation is unaltered by PI 3-kinase activity. This is further confirmed by the inducible expression of p27 kip1 phosphorylation site mutants, suggesting that p27 kip1 is destabilized in a phosphorylation-independent manner by the PI 3-kinase pathway at the G 1 /S transition.
One of the most common mutations seen in glioblastoma multiform, the most frequent primary brain tumor, occurs in the tumor suppressor gene phosphatase and tensin homologue deleted on chromosome 10 (PTEN) (1). PTEN regulates the activity of the phosphoinositide (PI) 1 3-kinase pathway, a signaling cascade implicated in cancer development. PTEN dephosphorylates phosphatidylinositol 3,4,5-trisphosphate and phosphatidylinositol 3,4-bisphosphate, the lipid products of PI 3-kinase activity (2). We and other investigators have shown previously (3)(4)(5) that the introduction of wild type (WT) PTEN into PTEN-mutant glioblastoma cell lines decreases PI 3-kinase lipid products and the activity of a PI 3-kinase downstream target, AKT (also known as protein kinase B). The expression of PTEN also causes a cell cycle arrest in G 0 /G 1 , which is most closely associated with increased protein levels of the CDK inhibitor p27 kip1 (4,6,7). A reduction of p27 kip1 expression using antisense oligonucleotides against p27 kip1 prevents the PTEN-induced G 0 /G 1 arrest, implicating p27 kip1 as a key mediator of the PI 3-kinase-mediated cell cycle progression (7). The inhibition of PI 3-kinase in p27 kip1 null cells is also less effective at causing G 1 arrest (8). However, the exact mechanisms underlying the regulation of p27 kip1 in glioblastoma cells after the introduction of PTEN remain unclear.
Posttranslational modifications of p27 kip1 have been studied extensively. Not only is p27 kip1 an inhibitor of the Cyclin-Cdk2 complex, it can also act as a substrate. Cdk2-Cyclin E phosphorylates p27 kip1 on Thr-187, which leads to the recruitment of the Skp2-containing Skp1-Cul1-Fbox protein complex (a ubiquitin-protein isopeptide ligase) resulting in p27 kip1 ubiquitination and degradation by the proteasome (9 -11). Therefore, only a small fraction of steady state p27 kip1 would predictably be phosphorylated on Thr-187. The major phosphorylation site on p27 kip1 is Ser-10, which leads to CRM1-dependent nuclear export (12) and increased protein stability in the cytosol (12,13).
Recently, AKT has been shown to directly phosphorylate p27 kip1 on Thr-157 in breast cancer cell lines, preventing nuclear import (14 -16). It is not clear whether this also applies to other tumor types with high AKT activity or whether this regulation is specific for breast cancer. AKT may also phosphorylate Thr-198, promoting 14-3-3 association and cytoplasmic sequestration (17).
Finally, it has been demonstrated that Forkhead transcription factors, which are inactivated after the phosphorylation by AKT, can regulate p27 kip1 transcription (8,18). Nakamura et al. (19) have also proposed that Forkhead transcription factors increase p27 kip1 levels by affecting primarily protein stability.
The importance of p27 kip1 as an important mediator of PI 3-kinase-induced cell cycle progression is well established, but clearly the mechanism of regulation can be different in different cell types. We therefore explored the mechanisms underlying p27 kip1 induction following inhibition of PI 3-kinase in PTEN-mutant glioblastoma cells.

EXPERIMENTAL PROCEDURES
Cell Culture-U87-MG glioblastoma cells were maintained in Dul-beccoЈs modified EagleЈs medium (Invitrogen) containing 10% fetal bovine serum and supplemented with antibiotics. When cells were to be infected with adenovirus, the medium was changed to 2% fetal bovine serum 1 h prior to infection.
Inducible Cell Lines-Ecdysone-inducible cell lines were established using the ecdysone-inducible mammalian expression system (Invitrogen). Hygromycin-resistant pIND expression vectors containing Myctagged p27 kip1 and phosphorylation site mutants were transfected using FuGENE 6 (Roche Applied Science) into U87-MG cell lines expressing the ecdysone response receptor. Twenty single cell-derived independent drug-resistant colonies of each form of p27 kip1 were cloned and screened for each gene. Exogenous expression levels of each of these genes was monitored by Western blotting, and the highest induced gene expression cell line was chosen for further study. Different ranges of ponasterone A (PonA) concentration were used to induce p27 kip1 to different degrees in the different cell lines. The amounts varied from 0.3 to 10 M but were different in the cell lines in a manner that was independent of the generation of phosphorylation sites and were probably more related to clonal variability. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Control and WT PTEN-expressing adenoviruses were made as described previously (7). U87-MG cells were infected routinely at a multiplicity of infection of 10, and cells were harvested 48 h after infection. The infection efficiency achieved by this procedure was Ͼ98% (data not shown).
Quantitative Real Time PCR-RNA was purified by the RNAeasy protocol (Qiagen), and first strand cDNA was synthesized using reverse transcriptase. p27 kip1 RNA levels were analyzed by quantitative PCR (TaqMan) using the oligonucleotides 5Ј-GGTTAGCGGAGCAATGCG-3Ј and 5Ј-TCCACAGAACCGGCATTTG-3Ј to amplify the p27 kip1 cDNA and the fluorescently labeled internal oligonucleotide 5Ј-TGCAAC-CGACGATTCTTCTACTCAAAACAAA-3Ј for fluorescence detection, using an ABI 4700 analyzer. The quantification of p27 kip1 was performed relative to the control gene hGUS by subtracting the cycle threshold (C t ) of the hGUS gene from the C t of p27 kip1 . The resulting difference in cycle number (⌬C t ) is the exponent of the base 2 (because of the doubling function of PCR), representing the -fold difference of the template for these two genes (20).
Two-dimensional Isoelectrical Focusing-SDS-Protein Analysis Gel Electrophoresis-Equal amounts of material (as normalized by cell number) were lysed in 8 M urea, 4% CHAPS, 10 mM dithiothreitol, and 0.2% Biolytes (pH 3-10). Readystrips (pH 3-10, Bio-Rad) were actively rehydrated with 300 l of sample. The isoelectrical focusing was performed in a Bio-Rad chamber to separate proteins by charge, according to the manufacturer's instructions. After equilibration in appropriate buffers, 12% gels were run to separate proteins by mass in the second dimension. Proteins were transferred to Immobilon-P membranes and blotted using an anti-p27 kip1 antibody (BD Biosciences).
Immunoprecipitation and in Vitro Cdk2 Kinase Assay-Subconfluent monolayers of U87-MG cells were lysed as described for immunoblot analysis. Endogenous Cdk2 was immunoprecipitated by adding 1 g of anti-Cdk2 antibody (M2, Santa Cruz Biotechnology) and kept on ice for 1 h. Protein A beads (Amersham Biosciences) were added, and samples were collected after 1 h at 4°C. The beads were washed three times in lysis buffer. One-half of the immunoprecipitate was used for immunoblotting. The other half was used to measure kinase activity using 2 g of histone H1 (Upstate Biotechnology) as a substrate in a reaction mixture containing 20 mM Tris (pH 7.5), 75 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 20 M ATP, and 5 Ci of [␥-32 P]ATP (EasyTide, Amersham Biosciences). The kinase reaction was allowed to proceed at 30°C for 15 min and then stopped with 4ϫ Laemmli sample buffer. The samples were run on 12% SDS-polyacrylamide gels, and radioactivity was quantified using a STORM PhosphorImager (Molecular Dynamics).
Cycloheximide Experiments-U87-MG cells were treated with cycloheximide (200 g/ml) at the indicated times. Equal amounts of lysate were separated by SDS-PAGE and analyzed by Western blotting for p27 kip1 , p21 cip1 , Cyclin D3, and ␤-actin.
Pulse-Chase Experiments-To measure the turnover of p27 kip1 protein, subconfluent U87-MG cells were plated and induced with the indicated amounts of PonA as necessary. After preincubation for 1 h with methionine-and cysteine-free medium, cells were pulsed with 500 Ci of EasyTag EXPRE 35 S protein-labeling mixture (containing 35 Slabeled methionine and cysteine, catalog no. NEG772, PerkinElmer) and incubated for 4 h at 37°C. After they were washed twice with phosphate-buffered saline, the cells were replaced with medium containing methionine and cysteine and lysed after the indicated times in buffer containing 0.1% Nonidet P-40, 250 mM NaCl, 5 mM EDTA, 50 mM Hepes (pH 7.0), 25 mM NaF, 1 mM NaVO 4 , 1 mM dithiothreitol, and a protease inhibitor mixture (Roche Applied Science) at 4°C. After preclearing with protein G beads, p27 kip1 was immunoprecipitated with 1.5 g of anti-p27 kip1 antibody (catalog no. SC-528, Santa Cruz Biotechnology) and 20 l of protein G beads. The precipitate was washed six times in lysis buffer and separated on a 12% SDS gel. After transfer to an Immobilon-P membrane, p27 kip1 was quantified using the STORM PhosphorImager. Equal loading was confirmed by immunoblotting for p27 kip1 .
Antisense Oligonucleotides, [ 32 P]Orthophosphate Labeling, Phosphoamino Analysis, and Phosphopeptide Mapping-Cells were treated with p27 kip1 antisense oligonucleotides as described previously (7). Subconfluent U87-MG cells were preincubated in phosphate-free medium for 1 h and labeled with 1 mCi/ml [ 32 P]phosphoric acid for 6 h. Cells were lysed, and p27 kip1 was immunoprecipitated as described for pulsechase analyses.
One-third of the sample was separated on a 12% SDS gel, transferred to an Immobilon-P membrane, and immunoblotted for p27 kip1 . The membrane was exposed to the STORM PhosphorImager, and radiolabeled p27 kip1 was quantified. The membrane containing p27 kip1 was excised and subjected to partial hydrolysis in 6 N HCl for 60 min at 110°C. Two-dimensional phosphoamino analysis was performed as described previously (22).
Two-thirds of the sample were separated on a 12% SDS gel and exposed to the STORM PhosphorImager. The band corresponding to p27 kip1 was excised and digested with trypsin, and the phosphopeptides were separated by electrophoresis and thin layer chromatography (22).
Cell Cycle Analysis-U87-MG cells were trypsinized following 48 h of treatment with 20 M LY294002 and fixed in 70% ethanol at 4°C overnight. The cells were resuspended in phosphate-buffered saline containing 10 g/ml propidium iodide (Roche Applied Sciences) and 1 g/ml RNase A, incubated for 30 min at room temperature, and analyzed by fluorescence-activated cell sorting. To aid in the visualization of the effects on the cell cycle, one-half of the samples were treated with nocodazole (70 ng/ml) for 18 h prior to trypsinization to induce an arrest in G 2 , as described previously (23).

Inhibition of the PI 3-Kinase Pathway by PTEN or LY294002
Leads to p27 kip1 Stabilization-The introduction of PTEN leads to a G 1 arrest that is mediated by p27 kip1 (7). Using an adenovirus expression system Ͼ98% of cells were green fluorescent protein-positive by fluorescence microscopy and fluorescenceactivated cell sorting, indicating adenoviral gene transduction into Ͼ98% of cells (data not shown). We (7) and other investigators (4, 6, 24) have shown previously that p27 kip1 levels accumulate in U87-MG cells after the expression of WT PTEN or treatment with LY294002 (Fig. 1A). This accumulation was not caused by transcriptional regulation because quantitative PCR (Fig. 1A) or Northern blotting (data not shown) showed no significant differences in p27 kip1 mRNA levels. We therefore examined whether changes in p27 kip1 stability could underlie the increased levels seen following the inhibition of PI 3-kinase activity. As depicted in Fig. 1B, adding cycloheximide to prevent protein synthesis resulted in the decay of p27 kip1 levels in untreated cells, with a half-life of ϳ2 h. In contrast, when PI 3-kinase was inhibited by either infection with PTEN adenovirus or treatment with LY294002, cycloheximide addition did not result in a rapid loss of p27 kip1 protein. In the inhibited cells, the p27 kip1 half-life was greater than the duration of the experiment, i.e. Ͼ8 h. In contrast, the stability of two additional cell cycle proteins, p21 cip1 and Cyclin D3, was not affected by PI 3-kinase inhibition (Fig. 1B). p27 kip1 stabilization by PI 3-kinase inhibition was independent of AKT and mTOR, because the overexpression of dominant-negative (kinase-dead) AKT or treatment with rapamycin did not dramatically affect p27 kip1 stability (Fig. 1C). Increased p27 kip1 stability following the inhibition of PI 3-kinase was further confirmed by [ 35 S]Met/ Cys pulse-chase experiments (Fig. 1D). The half-life of endogenous p27 kip1 determined by these experiments was 1.5-2 h, which increased to 4 -6 h in the presence of LY294002 or PTEN adenovirus. This is slightly less that the half-life determined by cycloheximide treatment, suggesting that proteins required for p27 kip1 degradation may also have a short half-life. p27 kip1 Phosphorylation Pattern or Overall Charge Is Unaltered by PTEN Expression or Treatment with LY294002-Because p27 kip1 stability could be regulated by phosphorylation, we tested whether the inhibition of PI 3-kinase altered p27 kip1 stability by changing its phosphorylation state. Immunoprecipitation from [ 32 P]orthophosphate-labeled U87-MG cells demonstrated that endogenous p27 kip1 was a phosphoprotein under these conditions ( Fig. 2A). Endogenous phosphorylated Cdk2, the identity of which was confirmed by Western blot analysis (data not shown), was also co-immunoprecipitated under these circumstances. However, the stoichiometry of later, the medium was replaced with Met/Cys-containing medium with no label, and cells were harvested at the indicated times. Following preclearing of the lysates, p27 kip1 was immunoprecipitated and subjected to SDS-PAGE and Western blotting, and incorporated radioactivity was measured by phosphorimaging and ImageQuant analysis. Where indicated, cells were pretreated with 50 M N-acetyl-Leu-Leu-norleucinal (ALLN), a proteasome inhibitor, 1 h prior to adding [ 35 S]Cys/Met. In the adenovirus-treated cells, p27 kip1 appears as a doublet in which the lower of the two bands is p27 kip1 , and the upper band is green fluorescent protein that co-immunoprecipitates. Densitometric analysis was performed and plotted as relative levels. The radioactivity incorporated into p27 kip1 is displayed graphically in the lower panel.
p27 kip1 phosphorylation was not altered dramatically on PTEN expression or LY294002 treatment ( Fig. 2A). Phosphoamino analysis of total p27 kip1 showed that p27 kip1 was phosphorylated exclusively on serine residues and that threonine and tyrosine phosphorylation was not observed (Fig. 2B) even after long exposures. To determine whether individual phosphorylation sites might be affected by PI 3-kinase activity, we digested the phosphorylated protein with trypsin and separated the released phosphopeptides by two-dimensional electrophoresis. This analysis showed that four major phosphopeptides were consistently apparent (Fig. 2B). The two minor phosphopeptides (Fig. 2B, asterisks) were not consistently seen across several experiments. After the inhibition of PI 3-kinase with LY294002 (Fig. 2B) or PTEN (data not shown), this pattern of phosphopeptides was almost identical, suggesting that phosphorylation of individual p27 kip1 phosphorylation sites was not affected significantly by PI 3-kinase activity. In a complementary approach that was applied to analyze the phosphorylation status of p27 kip1 , as well as to analyze additional post-translational modifications, we performed two-dimensional gel and Western blot analyses. After treatment of U87-MG cells with LY294002 or PTEN, protein lysates were separated by isoelectrical focusing followed by SDS-PAGE and blotted for p27 kip1 . This analysis revealed two major and three minor differentially charged forms of p27 kip1 , which also did not consistently change following PI 3-kinase inhibition (Fig. 2C).
Inducible Expression of WT p27 kip1 and Phosphorylation Site Mutants Are Stabilized after Inhibition of the PI 3-Kinase Pathway-Even though the phosphorylation state of p27 kip1 did not change following PI 3-kinase inhibition, we reasoned that  ). 24 h later, the cells were incubated for 1 h in phosphate-free medium followed by an incubation of 6 h in phosphate-free medium containing 1 mCi of orthophosphate. Cells were then harvested and incubated with protein G beads (precleared (pc), odd lanes), and the supernatant of this incubation was immunoprecipitated with p27 kip1 antibody and protein G beads (p27, even lanes). The immunoprecipitates were resolved by SDS-PAGE, and the incorporated radioactivity was measured by phosphorimaging analysis. Arrows indicate p27 kip1 and the associated Cdk2 present in the immunoprecipitate. 10% of the immunoprecipitate was run on a separate SDS gel and Western blotted for p27 kip1 . B, p27 kip1 from [ 32 P]orthophosphate-labeled Me 2 SO-or from LY294002-treated cells was recovered as in A, excised from the gel, and digested with trypsin, and peptides were resolved in two dimensions on thin layer chromatography plates. Longer exposures of the p27 kip1 phosphopeptides from the Me 2 SO-treated cells were used to generate similar visual intensities. Shown on the top left is a schematic map of the resolved phosphopeptides. The bottom left panel shows p27 kip1 recovered from [ 32 P]orthophosphate-labeled LY294002-treated cells and subjected to phosphoamino acid analysis (paa). The other bottom panels are densitometric analyses of spots 1-4. C, U87-MG cells were treated with either Me 2 SO or LY294002 or infected with control or PTEN adenoviruses. Cells were subjected to isoelectric focusing followed by SDS-PAGE, as described under "Experimental Procedures," and Western blotted using a p27 kip1 antibody. Longer film exposure times were used for Me 2 SO-and control adenovirus-treated cells than for LY294002-or PTEN adenovirus-treated cells.
the regulation of Thr-187 phosphorylation could still be controlling p27 kip1 stability but that its phosphorylation would cause immediate degradation and therefore would not be detected. We therefore generated U87-MG cell lines that expressed inducible forms of WT Myc-tagged p27 kip1 and phosphorylation site mutant derivatives. As depicted in Fig.  3A, left panel, increasing doses of PonA resulted in a dosedependent increase in WT p27 kip1 levels. The inhibition of PI 3-kinase caused increased levels of these exogenously expressed proteins, consistent with their stabilization following the inhibition of this pathway (Fig. 3A, right panel). The p27 kip1 proteins were functional, as judged by their ability to inhibit Cdk2 kinase activity (Fig. 3A, bottom left panel), and caused a cell cycle arrest in G 1 (Fig. 3B). The expression of exogenous p27 kip1 also increased the protein levels of endogenous p27 kip1 , consistent with the notion that p27 kip1 is an inhibitor of Cyclin-Cdk2 complexes, thereby decreasing the phosphorylation of endogenous p27 kip1 on its regulatory phosphorylation site Thr-187. However, the mutation of the Cdk2 phosphorylation site (T187A) in p27 kip1 did not prevent an increased stabilization of p27 kip1 resulting from PI 3-kinase inhibition (Fig. 3C). The mutation of S10A, another phosphorylation site implicated in p27 kip1 stabilization (12,13), also did not prevent increased levels of p27 kip1 following PI 3-kinase inhibition (Fig. 3C). Furthermore, a form of p27 kip1 containing mutations in both Thr-187 and Ser-10 and a third phosphorylation site implicated in p27 kip1 regulation, Ser-178, was also stabilized by inhibiting PI 3-kinase to the same extent as WT p27 kip1 (Fig. 3C). Slight differences in the induction of the endogenous p27 kip1 levels after expression of the various exogenous p27 kip1 mutants were not observed to be consistent and varied among several clones. To confirm that the increased levels of exogenous p27 kip1 following PI 3-kinase inhibition were caused by protein stabilization, we performed pulse-chase analysis. Fig. 4A demonstrates that exogenous p27 kip1 had a half-life of approximately 1-2 h under these conditions, which increased to more than 7 h in cells subjected to the inhibition of PI 3-kinase, similar to the data shown in Fig. 1 for endogenous p27 kip1 .
As shown in Fig. 3A, the inducible expression of p27 kip1 affected endogenous levels of p27 kip1 . We suspected that the exogenous expression of p27 kip1 would inhibit Cdk2 (as shown in Fig. 3A), leading to a perturbation of the balance between p27 kip1 on one hand and Cyclin-Cdk2 on the other. To confirm   FIG. 3. Increased p27 kip1 in response to LY294002 is independent of phosphorylation on Thr-187, Ser-10, and Ser-178. A, Me 2 SO (DMSO)-and LY294002-treated U87-MG cells expressing Myc-tagged wild type p27 kip1 were subjected to increasing concentrations of PonA, and endogenous and exogenous p27 kip1 was detected by Western blotting using the p27 kip1 antibody. The middle panel shows Cdk2 activity against histone H1 after immunoprecipitation from the same lysates. Western blotting for ␤-actin served as a loading control. B, vector or inducible WT p27 kip1 -expressing U87-MG cells were treated with 20 M LY294002 or 10 M PonA. After an incubation of 12 h, 70 ng/ml nocodazole (Noc) was added, and cells were harvested for fluorescence-activated cell sorting analysis 18 h later. The percentage of cells with a 2n DNA content is indicted. C, cells expressing inducible phosphorylation site mutants of p27 kip1 were treated with either Me 2 SO or LY294002, and after 6 h, the cells subjected to increasing concentrations of PonA. p27 kip1 and ␤-actin were analyzed by Western blotting, and Cdk2 activity was assessed by its ability to phosphorylate histone H1. As in A, Cdk2 activity levels from LY294002-treated cells were always very low to undetectable, irrespective of p27 kip1 induction (data not shown).
that the effects of LY294002 on exogenous p27 kip1 were direct and did not result from decreased Cdk2 activity following increased levels of endogenous p27 kip1 , we utilized p27 kip1 antisense oligonucleotides. These can eliminate endogenous p27 kip1 but not affect the levels of exogenous p27 kip1 because of the introduction of silent mutations in the oligonucleotide binding site. As can be seen in Fig. 4B, the exogenous p27 kip1 was stabilized independently of the presence of endogenous p27 kip1 after treatment with LY294002. This showed that the titration of inducible WT p27 kip1 at physiological levels behaved similarly to endogenous p27 kip1 .
To gain further insight into the phosphopeptide maps of the endogenous p27 kip1 shown in Fig. 2 and to identify which spots corresponded to which sites, we performed two-dimensional phosphopeptide mapping experiments on the exogenous p27 kip1 phosphorylation site mutants. Fig. 5A shows that the phosphorylation of exogenous p27 kip1 was almost identical to that seen for the endogenous p27 kip1 , with a prominence of spots 1-4 in both cases. Treatment with LY294002 also did not dramatically change the pattern of phosphorylation of the exogenous p27 kip1 (Fig. 5A), confirming the results seen with endogenous p27 kip1 . The phosphorylation site represented by spot 3 did increase slightly following treatment with LY294002. However, because this difference was not apparent in endogenous p27 kip1 (Fig. 2B), we are unsure of its importance. Use of the T187A mutant of p27 kip1 (Fig. 5B) showed that phosphorylation at Thr-187 was barely detectable, consistent with the results of the phosphoamino acid analysis of endogenous p27 kip1 . We do not believe that spot 5 represents this site, because its appearance was somewhat variable and seemed to depend on the running conditions of the thin layer chromatography plate. The use of the S10A mutant (Fig. 5, B and C) showed that most of the phosphorylation of p27 kip1 occurred on this site and that mutation at this site caused the disappearance of spots 1 and 2. Further mutation of S178A also caused the disappearance of spot 4, suggesting that this residue was also phosphorylated in U87-MG cells under these conditions. There was still one remaining spot present even in the S10A/S178A/T187A triple mutant, suggesting that a further phosphorylation site accounted for approximately 10 -15% of p27 kip1 phosphorylation. As revealed by phosphoamino acid analysis, this residue was serine-phosphorylated (data not shown) (Fig. 2B). DISCUSSION Our data showed that p27 kip1 protein accumulated after the re-expression of WT PTEN or treatment with LY294002 in PTEN-mutant U87-MG cells. We have shown previously that p27 kip1 is important for the PTEN-mediated G 1 arrest (7). Transcriptional control of p27 kip1 by the PI 3-kinase pathway has been described previously, indicating that activated AKT phosphorylates Foxo transcription factors, causing cytosolic accumulation and thereby preventing Foxo-dependent transcription of p27 kip1 mRNA (8). Although AKT was highly activated in U87-MG cells because of the mutation of PTEN, p27 kip1 mRNA was readily detectable, and mRNA levels were not affected by the inhibition of the pathway (Fig. 1A). The lack of transcriptional regulation in glioma cells has been described FIG. 4. Exogenous p27 kip1 is stabilized by LY294002 independently of endogenous p27 kip1 levels. A, U87-MG cells expressing inducible p27 kip1 were treated first with 10 M PonA to induce p27 kip1 and then with 20 M LY294002 (LY) as indicated for 24 h; they were then subjected to [ 35 S]methionine/cysteine pulse-chase analysis as described under "Experimental Procedures." PonA-treated cells were also incubated with 50 M N-acetyl-Leu-Leu-norleucinal (ALLN) for 1 h prior to pulse-chase analysis. The p27 kip1 was immunoprecipitated using the p27 kip1 antibody, resolved by SDS-PAGE, and transferred to polyvinylidene difluoride membranes, and the incorporated radioactivity was measured by phosphorimaging analysis and quantitated using ImageQuant software. pc, precleared (lysates incubated with protein G-Sepharose in the absence of the p27 kip1 antibody). Western blotting of the radioactive polyvinylidene difluoride membrane confirmed the immunoprecipitation of exogenous and endogenous p27 kip1 . B, U87-MG cells expressing inducible p27 kip1 were treated with Lipofectin alone or Lipofectin in the presence of 300 nM p27 kip1 antisense or mismatch oligonucleotides. These oligonucleotides eliminate endogenous (End) p27 kip1 but do not affect the levels of exogenous p27 kip1 because of the introduction of silent mutations in the oligonucleotide binding site. 24 h after transfection, the cells were treated with PonA, as indicated, together with LY294002 as indicated. Lysates were separated by SDS-PAGE, and endogenous and exogenous p27 kip1 were detected using the p27 kip1 antibody. previously by us (7) and other investigators (24). Also, p27 kip1 protein levels remained constant despite the expression of dominant-negative (kinase-dead) AKT in U87-MG cells (Fig. 1C) (data not shown). Furthermore, the p27 kip1 expressed from an exogenous promoter was stabilized after PI 3-kinase inhibition (Fig. 3A), showing the independence from the endogenous promoter. In summary, this suggested that p27 kip1 was not transcriptionally regulated by the PI 3-kinase pathway, at least in U87-MG cells.
Translational control of p27 kip1 has been observed in normal and tumor cells (25)(26)(27). p27 kip1 mRNA has also been shown to be translated in a cap-independent manner, accounting for the increased expression of p27 kip1 in cells exhibiting decreased cap-dependent translation (28). However, two pieces of evidence argue against a strong contribution of translational control in regulating p27 kip1 levels in U87-MG cells. First, rapamycin, which decreases cap-dependent translation, had little effect on endogenous or exogenous p27 kip1 levels ( Fig. 1C) (data not shown). Second, exogenously expressed p27 kip1 lacking the 5Ј-and 3Ј-untranslated regions that are normally responsible for translational control was stabilized by PI 3-kinase inhibition in our inducible cell system (Figs. 3A and 4A). How-ever, we cannot exclude some role for translational regulation of WT and phosphorylation site mutants of p27 kip1 in our experiments.
Rather, the stability of p27 kip1 protein was affected strongly by PI 3-kinase activity (Figs. 1, B and C, and 4A). Surprising to us, we demonstrated that stabilization does not alter any detectable phosphorylation changes or any other post-translational modification that might affect the overall charge of p27 kip1 protein (Fig. 2). Mamillapalli et al. (24) have proposed that the restoration of PTEN in PTEN-mutant cells leads to transcriptional down-regulation of Skp2, thereby preventing degradation of p27 kip1 , which accumulates as a result. Although it is an attractive hypothesis, such a mechanism would result in the accumulation of p27 kip1 phosphorylated on Thr-187, which was not observed in U87-MG cells (Fig. 2B). Moreover, the p27 kip1 T187A mutant was stabilized equally after inhibition of the pathway. Likewise, the major phosphorylation site on p27 kip1 , Ser-10, was unaffected by the inhibition of the PI 3-kinase pathway, and mutants at this site responded normally to PI 3-kinase inhibition.
Other investigators have proposed Thr-157 phosphorylation of p27 kip1 by AKT as a means of localizing p27 kip1 to the FIG. 5. Ser-10 is the major phosphorylation site in p27 kip1 . A, U87-MG cells expressing inducible WT p27 kip1 were treated with 10 M PonA for 24 h and then treated with 20 M LY294002 for an additional 24 h. The cells were then labeled with [ 32 P]orthophosphate as described under "Experimental Procedures." p27 kip1 was isolated by immunoprecipitation, separated by SDS-PAGE, and digested with trypsin, and the peptides were separated by electrophoresis and chromatography. A schematic map of the resolved phosphopeptides is depicted on the left. B, U87-MG cells expressing inducible phosphorylation site mutants of p27 kip1 were treated with 10 M PonA and analyzed as described in A. Different exposure times were used for the different phosphorylation site mutants to generate similar spot intensities. C, densitometric analysis was performed of the individual spots, 1-4, and plotted in arbitrary units of intensity. cytoplasm by preventing nuclear import (14 -16). This mechanism has been reported exclusively in breast cancer cells, and it is not known at this time whether this tissue displays a unique mechanism for p27 kip1 regulation by the PI 3-kinase pathway. In our studies, p27 kip1 was localized primarily to the cytosol, and this localization did not change in response to PI 3-kinase activity (data not shown). In addition, threonine phosphorylation would have been readily detectable in phosphate labeling and phosphoamino acid analysis experiments.
Recently, a second proteolytic pathway for controlling p27 kip1 has been described by generating an elegant p27-T187A mouse knock-in model (29). Whereas the S-phase degradation of p27 kip1 is Thr-187-and Skp2-dependent, the degradation of p27 kip1 at the G 1 restriction point is independent of these. Instead, it is controlled by mitogen stimulation, because the addition of serum leads to the degradation of both wild type p27 kip1 and the Thr-187 mutant (29,30). Given that serum and mitogens stimulate the PI 3-kinase pathway and that we observed a G 1 arrest in cells after the re-expression of PTEN in PTEN-mutant glioblastoma cells, it is tempting to speculate that this G 1 -specific proteolytic pathway of p27 kip1 would be dependent on PTEN, independent of AKT.