Mitotic and stress-induced phosphorylation of HsPI3K-C2alpha targets the protein for degradation.

Activation of the phosphoinositide 3-kinases (PI 3-kinases) has been implicated in multiple cellular responses such as proliferation and survival, membrane and cytoskeletal reorganization, and intracellular vesicular trafficking. The activities and subcellular localization of PI 3-kinases were shown to be regulated by phosphorylation. Previously we demonstrated that class II HsPIK3-C2alpha becomes phosphorylated upon inhibition of RNA pol II-dependent transcription (Didichenko, S. A., and Thelen, M. (2001) J. Biol. Chem. 276, 48135-48142). In this study we investigated cell cycle-dependent and genotoxic stress-induced phosphorylation of HsPIK3-C2alpha. We find that the kinase becomes phosphorylated upon exposure of cells to UV irradiation and in proliferating cells at the G2/M transition of the cell cycle. Stress-dependent and mitotic phosphorylation of HsPIK3-C2alpha occurs on the same serine residue (Ser259) within a recognition motif for proline-directed kinases. Mitotic phosphorylation of HsPIK3-C2alpha can be attributed to Cdc2 activity, and stress-induced phosphorylation of HsPIK3-C2alpha is mediated by JNK/SAPK. The protein level of HsPIK3-C2alpha is regulated by proteolysis in a cell cycle-dependent manner and in response of cells to stress. Phosphorylation appears to be a prerequisite for proteasome-dependent degradation of HsPIK3-C2alpha and may therefore contribute indirectly to the regulation of the activity of the kinase.

Phosphoinositide 3-kinases (PI 3-kinases) 1 regulate diverse cellular processes, which include cell signaling, intracellular protein sorting, cell cycle progression, cell survival, and apoptosis (2). PI 3-kinases phosphorylate the D3 hydroxyl group on the inositol ring leading to 3-phosphoinositides which act as membrane-embedded second messengers mediating the activation of downstream effectors (3). Class I PI 3-kinases are heterodimeric enzymes consisting of a catalytic and a regulatory subunit. They are involved primarily in growth factor and chemotactic agonist-mediated signal transduction (4,5). In vitro, these PI 3-kinases are able to utilize phosphatidylinositol (PtdIns), PtdIns(4)P, and PtdIns(4,5)P 2 as substrates, but most likely produce PtdIns(3,4,5)P 3 in vivo (6). Following activation of resting cells, these kinases are recruited rapidly from the cytosol to the plasma membrane where they generate PtdIns(3,4,5)P 3 . Class II PI 3-kinases are monomeric proteins. Three human isozymes, HsPI3K-C2␣, HsPI3K-C2␤, and HsPI3K-C2␥ (7)(8)(9)(10), and their homologs in rodents have been characterized (11)(12)(13). PI 3-kinases of this class have been found also in Drosophila melanogaster (14) and in Caenorhabditis elegans (15), but not in yeast. Members of class II are distinguished from other PI 3-kinases by the presence of two tandem domains at their carboxyl terminus, a phox homology domain and a C2 domain, a module that is known to confer Ca 2ϩ -dependent phospholipid binding (16). However, the C2 domains of class II PI 3-kinases lack a critical Asp residue in the calcium binding loop (17), which is consistent with the finding that they do not bind to membranes in a calcium-dependent manner (14,18). In vitro, all class II PI 3-kinases phosphorylate PtdIns and PtdIns(4)P, but their in vivo substrate remains to be determined (2). Both HsPI3K-C2␣ and HsPI3K-C2␤ are implicated in signaling downstream of epidermal growth factor and platelet-derived growth factor receptors (18). HsPI3K-C2␣ was shown to concentrate in the trans-Golgi network and in clathrin-coated pits (19), whereas PI3K-C2␤ was found in the nuclei of rat liver cells (20). In general, the role of class II PI 3-kinases in signal transduction and mode of activation is poorly understood, and specific downstream targets have not been characterized.
Class I PI 3-kinases activities were shown to be regulated by phosphorylation (21)(22)(23). Phosphorylation of the regulatory subunit p85 by the catalytic subunit p110␣ of class I PI 3-kinase (p110/p85 heterodimer) down-regulates lipid kinase activity of the complex (21,22). Phosphorylation of class II PI 3-kinases was demonstrated; however, the physiological role of this phosphorylation remained unclear. Increased phosphorylation of class II PI 3-kinase C2␣ was found to correlate with a moderately elevated enzyme activity in insulin-stimulated cells (24). In contrast, our data demonstrated that the phosphorylation status neither changes the lipid kinase activity of PI3K-C2␣ nor affects the substrate specificity, but influences the intranuclear localization (1).
In this study we investigated the phosphorylation of HsPI3K-C2␣ induced by genotoxic stress and during the cell cycle. We show that the kinase becomes phosphorylated upon exposure of cells to UV irradiation and in proliferating cells at the G 2 /M transition of cell cycle. Stress-dependent and mitotic phosphorylation of HsPI3K-C2␣ occurs on the same serine residue (Ser 259 ) within a recognition motif (serine-proline sequence) for proline-directed kinases, such as mitogen-activated protein (MAP) kinases and cyclin-dependent protein kinases (Cdk). By using different selective inhibitors of MAP kinases and Cdks in in vitro and in vivo assays, we found that Cdc2 mediates mitotic phosphorylation, whereas JNK/SAPK is responsible for stress-induced phosphorylation of HsPI3K-C2␣. In either case phosphorylation provides a signal for proteasomedependent degradation of the protein.
Cell Culture, Synchronization, Transient and Stable Expressions-HeLa (ATCC), MCF7, COS-7, and HEK-293 cells were cultured at 37°C in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supplemented with 10% fetal calf serum and antibiotics. The following protocols were used to obtain HeLa cells arrested at specific stages of the cell cycle. Cells, enriched in G 1 , were obtained by treatment with mimosine (Sigma) at 400 M for 20 h. For early S phase arrest, subconfluent cultures were blocked by serum deprivation for 48 h followed by the addition of 5 g/ml aphidicolin (Sigma) for 24 h (26). To obtain a cell population enriched in G 2 phase, cells were presynchronized in S phase as described above and then released by transferring into fresh medium for 8 h. For M phase synchronization, cells were treated with nocodazole (Sigma) at 400 ng/ml for 16 h before collecting mitotic cells by shake off. Cell cycle distributions were confirmed by flow cytometry (fluorescenceactivated cell sorter). For fluorescence-activated cell sorter analysis of DNA content, cells were washed twice in phosphate-buffered saline and fixed in 90% methanol at Ϫ20°C for 15 min. After an additional wash in phosphate-buffered saline cells were resuspended in 4 mM sodiumcitrate, 0.1% Triton X-100 and treated with 10 g/ml RNase A in the presence of 50 g/ml propidium iodide for 10 min at 37°C. To inhibit proteasome activity, MG132 (Calbiochem) was added to cells at a concentration of 20 M.
Gel Electrophoresis, Immunoprecipitation, and Western Blot Analysis-Proteins were separated on 8 or 6% SDS-polyacrylamide gels prepared from the stock (33.5% acrylamide, 0.3% bisacrylamide) and blotted onto Immobilon-P (Millipore). Membranes were blocked with 5% non-fat milk in Tris-buffered saline containing 0.1% Triton X-100 and probed with specific antibodies. Immunoreactive bands were decorated with horseradish peroxidase-labeled secondary antibodies and visualized by enhanced chemiluminescence (Pierce).
Pulse-Chase Experiments-Subconfluent cultures of HeLa cells were labeled overnight with 50 Ci of [ 35 S]methionine/cysteine (Amersham Biosciences)/ml in methionine-free DMEM (Invitrogen) supplemented with 10% dialyzed fetal calf serum. After labeling cells were washed in phosphate-buffered saline, replated (1:3 dilution), and chased with complete DMEM containing 10% fetal calf serum for 48 h. To obtain mitotic cells, 400 ng/ml nocodazole was added to the medium for the last 36 h of the chase. Labeled mitotic cells were collected by shake off, washed three times in prewarmed DMEM, and released into fresh complete medium for 3 h. For metabolic labeling of cells at M/G 1 transition of cell cycle, nocodazole-treated mitotic HeLa cells were released into the labeling DMEM in the presence of 50 Ci of [ 35 S]methionine/cysteine for 1 or 3 h. Cells were subsequently subjected to immunoprecipitation analysis with anti-HsPI3K-C2␣ antibody (AXXIII) as described above. Immunoprecipitated proteins were resolved by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and visualized by autoradiography and immunoblotting.
Kinase and Protease Inhibitor Treatments-Roscovitine (Calbiochem) was used to inhibit Cdk activity. HeLa cells at late S phase (6 h after release from aphidicolin block) were treated or not with 30 M roscovitine for 2 h. 400 ng/ml nocodazole was added and treatment continued for 15 h. Nonadhering mitotic and adhering G 2 cells were collected by mechanical shock and trypsin treatment, respectively. Nocodazole-arrested mitotic HeLa cells were treated with 75 M roscovitine for 15, 45, and 90 min. For okadaic acid treatment nocodazolearrested mitotic HeLa cells were treated with 0.5 M okadaic acid for 30 min, then 75 M roscovitine was added, and treatment continued for 30 min. SP600125 (Tocris) was used to inhibit JNK activation, and PD98059 and SB202190 (both from Alexis) were used to inhibit ERK and p38 activation, respectively. HeLa cells were preatreated with the inhibitors at concentrations indicated for 30 min before UV irradiation. Irradiated cells were cultured for 90 min in the presence of the inhibitors before harvesting.
The specific protease inhibitors MG132, ALLM, and lactacystin were from Calbiochem. HEK-293 cells were UV irradiated as described above. After 2 h of recovery in fresh medium, cells were treated with protease inhibitors (20 M MG132, 100 M ALLM, or 50 M lactacystin) for the indicated times prior to Western blot analysis.
In Vitro Kinase Assay-HsPI3K-C2␣ was immunoprecipitated from mimosine-treated HeLa cells using affinity-purified anti-HsPI3K-C2␣ antibody AXIX (1). The antibody-antigen complexes were collected with GammaBind Plus-Sepharose and used as substrate for in vitro phosphorylation by cellular extracts. To obtain cellular extracts, pellets of interphase or mitotic HeLa cells were resuspended in 2 pellet volumes of ice-cold hypotonic buffer (50 mM Tris-HCl (pH 7.5), 1 mM dithiothreitol, 40 mM NaF, 0.5 mM sodium orthovanadate, 40 mM ␤-glycerophosphate, 5 mM sodium pyrophosphate, and a mixture of protease inhibitors (Complete, Roche)) and disrupted by brief sonication. The resulting homogenates were centrifuged at 400,000 ϫ g for 15 min at 4°C. Supernatants (ϳ10 mg of protein/ml) were supplemented with 150 mM NaCl and 10 mM MgCl 2 and used as a source of kinases. HsPI3K-C2␣ immunoprecipitates were mixed with supernatants in a final volume of 100 l, and phosphorylation assays were initiated by adding 1 mM ATP. Assays were carried out at 30°C for 1 h and terminated by the addition of ice-cold Tris-buffered saline containing 0.1% of Triton X-100. HsPI3K-C2␣ immunoprecipitates were collected by centrifugation, washed twice with Tris-buffered saline, and analyzed by Western blotting as described above.
For in vitro phosphorylation GST-⌬HsPI3K-C2␣ fusion proteins were expressed in Escherichia coli strain (BL21) and purified by absorption to glutathione-Sepharose beads (Amersham Biosciences). Fusion proteins were left attached to the beads, and phosphorylation reactions with cellular extracts were carried out as described above in the presence of 50 Ci of [␥-32 P]ATP and 1 mM ATP. In vitro phosphorylation of GST-⌬HsPI3K-C2␣ and GFP-HsPI3K-C2␣ fusion proteins by 10 units of purified recombinant human Cdc2-cyclin B (Calbiochem) was performed in Cdc2-kinase buffer (50 mM Tris-HCl (pH 7.5), 2 mM dithiothreitol, 10 mM MgCl 2 , 1 mM EGTA) in the presence of 10 Ci of [␥-32 P]ATP and 100 M ATP. Soluble GST-⌬HsPI3K-C2␣ was used as substrate in phosphorylation assays with immunoprecipitated Cdc2. Cdc2 was immunoprecipitated from cytosols of HeLa cells as described above, immunocomplexes bound to GammaBind Plus-Sepharose beads were additionally washed twice in the kinase buffer, and the reaction was initiated by addition of purified GST-⌬HsPI3K-C2␣ and ATP (100 M ATP, 10 Ci of [␥-32 P] ATP).
Immunofluorescence-Immunofluorescence experiments were performed as described previously (1) using the methanol fixation protocol.

UV-induced and Cell Cycle-dependent Phosphorylation of
HsPI3K-C2␣-Our previous immunofluorescence studies revealed that in interphase HeLa cells HsPI3K-C2␣ is localized to nuclear speckles together with the components of the splicing apparatus (1). Inhibition of transcription by actinomycin D or ␣-amanitin causes subnuclear relocation of the kinase and is accompanied by phosphorylation of the protein, which can be measured as mobility shift of HsPI3K-C2␣ during SDS-PAGE.
To investigate whether down-regulation of transcription caused by different types of genotoxic stress results in phosphorylation of HsPI3K-C2␣, we exposed cells to DNA-damaging treatment such as UV light or ionizing radiation. Exposure of HeLa and MCF7 cells to UV irradiation induced a collapse of nuclear HsPI3K-C2␣-positive speckles (Fig. 1A) similar to that observed in actinomycin D-treated cells: speckles lose their irregular shape, become round, and fuse into larger clusters (1). This effect was associated with the increased phosphorylation of HsPI3K-C2␣, as measured by its mobility shift on SDSpolyacrylamide gels (Fig. 1B). When cell extracts were treated with -phosphatase the appearance of the slower migrating band was abolished (not shown). In contrast to UV-treated cells, exposure of cells to ␥-irradiation neither changed subnuclear localization of HsPI3K-C2␣ (not shown) nor induced its phosphorylation (Fig. 1B). These results suggest that HsPI3K-C2␣ specifically participates in UV-induced damage response.
The observation that phosphorylation of HsPI3K-C2␣ correlates with changes in its subnuclear localization let us to speculate that the phosphorylation status of the kinase may also be cell cycle-dependent, because in mitotic cells HsPI3K-C2␣-positive speckles dissolve, and the kinase becomes equally distributed over the cytoplasm (Fig. 1A). We used HeLa cells to examine whether HsPI3K-C2␣ demonstrates different phosphorylation patterns during the cell cycle. Cells were synchronized in different stages of the cell cycle as follows: at late G 1 with mimosine, at M with nocodazole, in early S phase by serum deprivation followed by an aphidicolin block, and cells enriched in G 2 phase were obtained 8 h after release from aphidicolin block (26). Proteins from corresponding cell lysates were fractioned by SDS-PAGE, and HsPI3K-C2␣ was analyzed by immunoblotting ( Fig. 2A). In mimosine-treated cells HsPI3K-C2␣ was detected as a single band, in cells blocked in S phase a second slower migrating band became visible. Two bands, a faster and a slower migrating, of equal intensity were apparent in cells enriched in G 2 phase. A single slower migrating band was found in prometaphase-blocked mitotic cells. To confirm that altered gel mobility of HsPI3K-C2␣ was the result of phosphorylation, protein extracts from synchronized HeLa cells were treated with -phosphatase (Fig. 2). Phosphatase treatment resulted in the collapse of the slower migrating band of the kinase, indicating that indeed retarded mobility of HsPI3K-C2␣ during SDS-PAGE is a consequence of phosphorylation. These results demonstrate that HsPI3K-C2␣ undergoes a cell cycle-regulated phosphorylation that reaches its maximum in mitosis. Asynchronously growing HeLa cells were UV irradiated (UV) or left untreated (control). After incubation in fresh medium for 2 h, cells were fixed with methanol and immunostained with anti-HsPI3K-C2␣ antibody AXIX. Mitotic cells are indicated by arrows. B, asynchronously growing HeLa and MCF7 cells were exposed either to 300 J/m 2 UV light (UV) or ionizing radiation (100 Gy) (IR), or left untreated (Ϫ), and harvested 90 min later. Cell extracts were analyzed by SDS-PAGE followed by immunoblotting with anti-HsPI3K-C2␣ antibody AXXIII. Upon UV irradiation the band with slower electrophoretic mobility corresponding to the phosphorylated form of HsPI3K-C2␣ becomes more prominent.

HsPI3K-C2␣ Can Be Phosphorylated in Vitro by Kinases Present in HeLa Cell Extracts-
The initial strategy to identify residues on which HsPI3K-C2␣ becomes phosphorylated upon UV irradiation and during cell cycle was to label HeLa cells metabolically in the presence of 32 P i . In several attempts we did not succeed to obtain sufficient amounts of in vivo 32 P-labeled HsPI3K-C2␣ by immunoprecipitation to perform phosphopeptide mapping analysis. To overcome this problem, we developed an in vitro phosphorylation assay that allowed the identification of potential phosphorylation sites. As a substrate for phosphorylation we used HsPI3K-C2␣ immunoprecipitated from mimosine-treated HeLa cells. Concentrated high speed supernatants (S100) prepared from HeLa cells (asynchronously growing or mitotic) were used as sources of kinases. The supernatants are devoid of HsPI3K-C2␣ as shown previously (1), which excluded any additional input of the already phosphorylated substrate into the assay. Phosphorylation of HsPI3K-C2␣ was measured by mobility shift of the protein during SDS-PAGE. Fig. 2B illustrates that HsPI3K-C2␣ can be phosphorylated successfully in vitro by kinases present in cell ex-tracts, and it is not caused by autophosphorylation. Almost complete phosphorylation of HsPI3K-C2␣ was achieved in vitro by kinases from mitotic extracts, whereas the protein was less efficiently phosphorylated by kinases present in extracts from asynchronously cycling cells. The phosphorylation profile and the efficiency of phosphorylation of HsPI3K-C2␣ in vitro appeared to be remarkably similar to that seen in vivo (compare Fig. 2, A and B, upper panel), suggesting that in vitro phosphorylation assay accurately mimics the in vivo situation.
Ser 259 Is the Site of Mitotic as Well as UV-induced Phosphorylation-We expressed different domains covering the entire sequence of HsPI3K-C2␣ (1) as GFP fusion proteins in HeLa cells and found that only fusion proteins containing the NH 2terminal domain (amino acids 1-482) exhibited an electrophoretic mobility shift (not shown). This observation suggested that phosphorylation site(s) is (are) localized within this domain and that shorter fragments of the kinase can be used as reporters of phosphorylation. Therefore we expressed different segments of the NH 2 -terminal domain of the kinase as GST fusion proteins and used these proteins as substrates for in vitro phosphorylation. The shortest domain (GST-⌬HsPI3K-C2␣) that showed retarded electrophoretic mobility upon phosphorylation in vitro comprised amino acids 240 -275 (Fig. 2B).
Analysis of the amino acid sequence of HsPI3K-C2␣ between residues 240 and 275 revealed seven potential phosphorylation sites (Fig. 3A). Among them, Ser 254 conforms to the consensus motif for phosphorylation by casein kinase I (S 251 PKVS 254 ), Thr 243 is within a recognition motif for phosphorylation by casein kinase II (T 243 DLE), Ser 259 followed by Pro is a potential target for proline-directed protein kinases. To map the phosphorylation site(s) within GST-⌬HsPI3K-C2␣, we mutated Ser 254 , Ser 259 , Ser 262 , and Ser 266 by a single substitution to alanine and subjected the resulting GST fusion proteins to in vitro phosphorylation assays (Fig. 3B). Incorporation of 32 P as well as the electrophoretic mobility shift of the corresponding fusion protein was completely abolished when Ser 259 was mutated to alanine (S259A), indicating that this residue is a prime target of phosphorylation. The mutation of Ser 254 (S254A) did not affect phosphorylation but resulted in the loss of the electrophoretic mobility shift of phosphorylated protein, whereas mutations of either Ser 262 (S262A) or Ser 266 (S266A) resulted in a complete mobility shift of phosphorylated proteins. GST-⌬HsPI3K-C2␣ incorporated 32 P into several slower and faster migrating forms. To rule out the possibility that phosphorylation on Ser 259 might be a primary event necessary for subsequent phosphorylation of other residues, we mutated this serine to either aspartic acid (S259D) or glutamic acid (S259E) to mimic its phosphorylation status. However, despite the fact that the corresponding mutant proteins did change their electrophoretic mobility they failed to incorporate 32 P. To confirm further Ser 259 as phosphorylation site, 32 P-labeled forms of GST-⌬HsPI3K-C2␣ were analyzed by mass spectrometry. MALDI-TOF measurements of tryptic fragments of GST- Cells enriched in late G 1 were obtained by treatment with mimosine (G1); serum starvation followed by aphidicolin block yielded cells in S phase (S); cells enriched in G 2 phase were obtained 8 h after release from aphidicolin block (G2); mitotic cells were obtained after nocodazole arrest (M). Cellular extracts were treated or not with -phosphatase ( PPase), separated on SDS-6% PAGE, and Western blots were probed with anti-HsPI3K-C2␣ antibody AXXIII. B, in vitro phosphorylation of HsPI3K-C2␣ with cellular extracts. Immunoprecipitated HsPI3K-C2␣ immobilized on GammaBind Plus-Sepharose was incubated with extracts prepared from either asynchronously grown (A) or mitotic (M) HeLa cells in the presence of 1 mM ATP at 30°C for 1 h. As a control for autophosphorylation, immunoprecipitated HsPI3K-C2␣ was incubated under the same conditions without cell extract. When indicated, HsPI3K-C2␣ immunoprecipitates were treated with -phosphatase ( PPase) after termination of the phosphorylation assay. The immunoprecipitates were analyzed by SDS-PAGE followed by immunoblotting with anti-HsIP3K-C2␣ antibody AXIX (upper panel). GST-⌬HsPI3K-C2␣ fusion protein, which included the region between amino acids 240 and 275 of HsPI3K-C2, was used for in vitro phosphorylation. Recombinant proteins were bound to glutathione-Sepharose, and the phosphorylation assay was carried out as described above. Proteins were separated on SDS-11% polyacrylamide gels and visualized by Coomassie Blue staining (lower panel).
⌬HsPI3K-C2␣ (amino acids 240 -275) as GFP fusion proteins in HeLa cells and measured their phosphorylation status in mitosis and after UV irradiation. As shown in Fig. 4A, only mutations of Ser 259 (S259A and S259D) completely blocked the M phase-dependent as well as UV-induced electrophoretic mobility shift of the corresponding GFP fusion proteins. Similar results were obtained when full-length wild-type and mutated (S259A and S259D) forms of HsPI3K-C2␣ were transiently expressed as GFP fusion proteins in COS-7 cells. In asynchronously growing cells GFP-HsPI3K-C2␣ showed a marginal level of phosphorylation, and UV irradiation further induced the phosphorylation of this fusion protein, as judged by the increase in the level of the slower migrating form (Fig. 4B). In contrast, the mutants GFP-HsPI3K-C2␣/S259A and GFP-HsPI3K-C2␣/S259D were detected as single bands on SDSpolyacrylamide gels and did not show any change in their electrophoretic mobility upon UV irradiation. Analysis of no-codazole-arrested HEK-293 cells stably expressing HsPI3K-C2␣-GFP and HsPI3K-C2␣/S259A-GFP (Fig. 4C) confirmed that Ser 259 is also the site of mitotic phosphorylation.
Taken together, these data demonstrate that Ser 259 is a common site of the mitosis phase-dependent and the UV-induced phosphorylation of HsPI3K-C2␣. Moreover, the phosphorylation of this site induces a conformational change in HsPI3K-C2␣ which alters its mobility during SDS-PAGE. The latter conclusion is strengthened by observations that mutation of Ser 259 to aspartic acid or glutamic acid, which should mimic phosphorylation on this residue, results in mutants with electrophoretic mobility comparable to that of the phosphorylated protein. However, it can not be totally excluded that HsPIK3-C2␣ may also become phosphorylated at other sites.
In Vivo Phosphorylation of HsPI3K-C2␣ by Cdc2-The G 2 /M transition of the cell cycle is triggered by activation of a protein kinase cascade. The major kinase, p34 cdc2 , required for promo-
tion of mitosis, belongs to a family of proline-directed kinases. Because the identified site of mitotic phosphorylation in HsPI3K-C2␣ resembles the consensus motif (S-P-X-R/K) for phosphorylation by Cdc2, we tested whether in vivo phosphorylation of HsPI3K-C2␣ in mitosis is mediated by Cdc2. We used roscovitine, a highly selective inhibitor of cyclin-dependent kinases (27,28), to block Cdc2 activity in mitotically synchronized cells. HeLa cells were synchronized in S phase, released, and then allowed to proceed to mitosis under nocodazole block in the presence or absence of roscovitine. After exposure to nocodazole for 15 h approximately 98% of control cells were arrested in mitosis. In contrast, only approximately 10% of roscovitine-treated cells were nonadherent mitotic cells, whereas the majority remained adherent because of the arrest in G 2 phase. Immunoblot analysis revealed that roscovitine treatment led to a significant inhibition of phosphorylation of HsPI3K-C2␣ in mitotic and in G 2 phase-arrested cells (Fig. 5A).
To test whether Cdc2 activity is indispensable for maintaining phosphorylation of HsPI3K-C2␣ in mitosis, we treated the nocodazole-blocked mitotic HeLa cells with roscovitine for short periods of time and analyzed the phosphorylation state of HsPI3K-C2␣ by immunoblotting. As shown in Fig. 5B, the amount of phosphorylated HsPI3K-C2␣ present in the slower migrating form decreased with the duration of roscovitine treatment, whereas simultaneously an increase in the amount of the faster migrating form was observed. We speculated that the dephosphorylation of HsPI3K-C2␣ caused by the inhibition of Cdc2 might involve an okadaic acid-sensitive protein phosphatase. Indeed pretreatment of mitotic cells with okadaic acid markedly attenuated the dephosphorylation of HsPI3K-C2␣ observed after roscovitine treatment (Fig. 5C). Taken together, these findings suggest that in vivo Cdc2 is involved in the phosphorylation of HsPI3K-C2␣ during the cell cycle.
In Vitro Phosphorylation of HsPI3K-C2␣ by Cdc2-To test whether Cdc2 can directly phosphorylate HsPI3K-C2␣, we performed in vitro kinase assays using Cdc2 immunoprecipitated from HeLa cells (asynchronously growing, enriched in G 2 phase or mitotic) and the GST-⌬HsPI3K-C2␣ as substrate. Fig. 6A shows that GST-⌬HsPI3K-C2␣ is phosphorylated by immunoprecipitated Cdc2 in a roscovitine-sensitive manner (top panel). The efficiency of phosphorylation correlates with cell cycle-dependent activation of Cdc2 as detected by the loss of the inactive slower migrating hyperphosphorylated form of p34 cdc2 (bottom panel).
To confirm that Ser 259 is a target of Cdc2, we determined whether purified activated Cdc2/cyclinB could phosphorylate wild-type and mutated GST-⌬HsPI3K-C2␣ fusion proteins. Fig. 6B shows that mutation of Ser 259 to either alanine or aspartic acid abolished Cdc2-mediated phosphorylation. The same result was obtained when full-length native and mutated GFP-HsPI3K-C2␣-fusion proteins expressed in COS-7 cells were used as substrates for in vitro phosphorylation by purified activated Cdc2/cyclinB (Fig. 6C).
Phosphorylation Controls the Turnover of HsPI3K-C2␣ during the Cell Cycle-A dramatic change in the phosphorylation state of HsPI3K-C2␣ occurs at the M to G 1 phase transition of the cell cycle (Fig. 2A, compare G 1 and M). We reasoned that phosphorylation could regulate the stability of HsPI3K-C2␣ and potentially provide a signal for degradation of the kinase during mitosis. Therefore we performed pulse-chase experiments to determine whether HsPI3K-C2␣, which was phosphorylated in mitosis, becomes dephosphorylated or degraded upon reentry of cells into subsequent G 1 phase. Fig. 7A (left  panel) shows that HsPI3K-C2␣ metabolically labeled with [ 35 S]methionine during interphase becomes fully phosphorylated and remains stable in prometaphase-arrested cells. However, upon release of mitotic cells from the nocodazole block, the level of the 35 S-labeled phosphorylated HsPI3K-C2␣ decreases. Concomitantly unphosphorylated kinase appears, which is not labeled with [ 35 S]methionine, and therefore represents de novo synthesized protein (Fig. 7A, right panel). To confirm that HsPI3K-C2␣ is indeed newly synthesized at the M/G 1 transition of the cell cycle, unlabeled mitotic cells were released from nocodazole block in the presence of [ 35 S]methionine (Fig. 7B). Progression into G 1 phase resulted in the appearance of the unphosphorylated [ 35 S]methionine-labeled form of HsPI3K-C2␣.
To determine whether mitotic destruction HsPI3K-C2␣ is caused by the proteasome activity, the proteasome inhibitor MG132 (29) was added to mitotic cells during the release from nocodazole block. Fig. 7C shows that addition of MG132 caused significant stabilization of the phosphorylated form of the kinase as well as stabilization of cyclin B1, a known target of proteasome degradation in late mitosis. These results suggest that M/G 1 transition of the cell cycle triggers a proteasome-dependent degradation of phosphorylated HsPI3K-C2␣.
UV-induced Phosphorylation of HsPI3K-C2␣ at Ser 259 Is Mediated by JNKs-Cell cycle responses to DNA damage induced by UV irradiation lead to G 1 and G 2 phase delays as a result of the inhibition of cyclin-dependent kinases (30,31). In contrast, UV irradiation causes the activation of MAP kinases, including ERKs, JNKs, and p38 kinase (32)(33)(34). To identify the possible role of MAP kinases in mediating UV-induced phosphorylation of HsPI3K-C2␣ and to facilitate the identification of the relevant pathway, we examined the influence of specific chemical inhibitors on the phosphorylation status of HsPI3K-C2␣. Pretreatment of HeLa cells with either SB202190, a specific inhibitor of p38 (35,36), or PD98059 a specific inhibitor of MEK1 (37) failed to abolish UV-induced phosphorylation of HsPI3K-C2␣ (Fig. 8A). In contrast, UV-induced phosphorylation of the kinase was markedly attenuated by pretreatment of cells with SP600125 (38), an inhibitor of JNK catalytic activity (Fig. 8B). Analysis of UV-irradiated HEK-293 cells stably expressing HsPI3K-C2␣-GFP and HsPI3K-C2␣/S259A-GFP (Fig. 8C) gave similar results, showing that UV-induced phosphorylation of HsPI3K-C2␣-GFP at Ser 259 is sensitive only to SP600125.
To examine further whether JNKs are implicated in the phosphorylation of HsPI3K-C2␣, we tested whether the transient expression of JNKK2/MKK7, a specific JNK-activating MAPKK, influences the steady-state level of the phosphorylation of HsPI3K-C2␣. Transfection of JNKK2 into HEK-293 cells stably expressing HsPI3K-C2␣-GFP or HsPI3K-C2␣/S259A-GFP resulted in a significant increase of the steady-state level of phosphorylated HsPI3K-C2␣-GFP, which was close to that seen in UV-irradiated cells. As expected, the mutant HsPI3K-C2␣/S259A-GFP failed to show any change in the electrophoretic mobility upon UV irradiation (Fig. 9, upper panel). The efficiency of JNK activation caused by transfection of JNKK2 was similar in both cell lines as judged by the increase in c-Jun phosphorylation at Ser 73 (Fig. 9, lower panel).
UV Irradiation Induces Phosphorylation-dependent Degradation of HsPI3K-C2␣-Our observation that mitotic phosphorylation of HsPI3K-C2␣ was followed by its degradation at the M to G 1 transition of the cell cycle led us to speculate that phosphorylation at Ser 259 might be a common signal required to activate HsPI3K-C2␣ proteolysis. Therefore, we examined the rate of disappearance of phosphorylated HsPI3K-C2␣ after UV irradiation. HEK-293 cells expressing HsPI3K-C2␣-GFP or HsPI3K-C2␣/S259A-GFP were exposed to UV light and allowed to recover for 2-24 h. Whole cell extracts from equivalent numbers of cells were prepared and analyzed on Western blots. Fig. 10A illustrates that the slower migrating phosphorylated forms of HsPI3K-C2␣ and HsPI3K-C2␣-GFP are induced within 2 h after UV irradiation and diminish thereafter until they become barely detectable after 24 h (compare recovery time between 2 and 24 h). In contrast, the levels of faster migrating unphosphorylated forms of these proteins remain unaffected during the recovery period. Over the time course after UV irradiation the levels of both HsPI3K-C2␣ and HsPI3K-C2␣-GFP diminish by about 50% compared with those detected in unirradiated cells. Phosphorylation-deficient HsPI3K-C2␣/S259A-GFP remains stable after UV irradiation (Fig. 10A), suggesting that mutation of Ser 259 to alanine protects the protein from proteolysis. We used the reversible inhibitor MG132 to demonstrate that the reduction of the levels of phosphorylated HsPI3K-C2␣ and HsPI3K-C2␣-GFP is caused by subsequent degradation by the proteasome. MG132 is a potent inhibitor of the proteasome but also inhibits calpains. We therefore examined the effects of additional protease inhibitors on the HsPI3K-C2␣ turnover. As shown in Fig. 10B, treatment of cells with MG132 (29) or with lactacystin, an irreversible inhibitor specific for proteasome (39,40), for 16 h after UV irradiation resulted in a significant stabilization of the phosphorylated forms of HsPI3K-C2␣ and HsPI3K-C2␣-GFP. Conversely, ALLM, a peptide aldehyde that is more selective for calpain than to the proteasome (41), caused only marginal effects. Neither of the inhibitors affected stability of HsPI3K-C2␣/S259A-GFP. (S259A) and HsPI3K-C2␣/S259D (S259D) were transiently expressed as GFP fusion proteins in COS-7 cells. GFP fusion proteins were immunoprecipitated with anti-GFP antibody from cell extracts and used as substrates for in vitro kinase assays with purified Cdc2-cyclin B as described above. After SDS-PAGE and electrotransfer, phosphorylated proteins were visualized by autoradiography ( 32 P) and immunoblotting with anti-HsPI3K-C2␣ antibody (WB).

DISCUSSION
It has been reported that class II PI 3-kinases can be phosphorylated (1,11,24). However, identities of phosphorylation site(s) and the kinase(s) involved remained elusive. It has been also not clear how phosphorylation modulates the properties of these enzymes. Here we investigated cell cycle-dependent and genotoxic stress-induced phosphorylation of HsPI3K-C2␣ and demonstrated that HsPI3K-C2␣ becomes phosphorylated on Ser 259 in cellular response to UV irradiation and in dividing cells during mitosis. Phosphorylation on Ser 259 , which is located next to a proline, induces a conformational change in HsPI3K-C2␣ that leads to the decrease in its electrophoretic mobility on SDS-polyacrylamide gels.
The phosphorylation status of HsPI3K-C2␣ during the cell cycle is under the direct control of Cdc2. Several lines of evidence support this conclusion. First, marginal phosphorylation of HsPI3K-C2␣ is detected during interphase, whereas the phosphorylated form of the kinase appears at the G 2 /M transition phase and reaches a maximum in mitosis, which coincides with the timing of Cdc2 activation. Second, mitotic phosphorylation of HsPI3K-C2␣ in both in vivo and in vitro assays was sensitive to roscovitine, a highly selective inhibitor of cyclindependent kinases (27,28). Several cyclin-dependent kinases, including Cdc2-cyclin B and Cdk2-cyclin A or E, are sensitive to roscovitine. However, the main target of roscovitine in mitosis is Cdc2-cyclin B because Cdk2-cyclin E and Cdk2-cyclin A kinases are active at the G 1 /S transition and during S phase, respectively. Furthermore, purified activated Cdc2/cyclin B was able to phosphorylate HsPI3K-C2␣ fusion proteins on Ser 259 in vitro. Mutations of Ser 259 to either alanine or aspartate abolished Cdc2-mediated phosphorylation of HsPI3K-C2␣ , and then cultured in the presence of the inhibitors for 90 min. Cell extracts were analyzed by SDS-PAGE followed by immunoblotting with anti-HsPI3K-C2␣ antibody AXXIII and anti-c-Jun (pSer 63 ) antibody. C, HEK-293 cells that stably expressed full-length wild-type HsPI3K-C2␣ or mutant HsPI3K-C2␣/S259A as GFP fusion proteins were pretreated with the indicated inhibitors for 30 min, UV irradiated, and analyzed as described above. in vitro. In addition, Cdc2 activity appears to be indispensable for maintaining phosphorylation of HsPI3K-C2␣ during mitosis, probably by antagonizing an okadaic acid-sensitive phosphatase.
In interphase cells HsPI3K-C2␣ is concentrated in nuclear speckles (1), which represent a subnuclear compartment enriched in small nuclear ribonucleoprotein particles and other splicing factors. In mitosis, at the end of prophase the breakdown of the nuclear envelope is accompanied by the disassembly and dispersal of all major nuclear structures and downregulation of transcription and splicing. Following the changes in the nuclear structure, at the metaphase-anaphase transition HsPI3K-C2␣ is found to disperse throughout the cytoplasm concomitantly with its complete phosphorylation. A mechanism for mitotic repression of transcription is phosphorylation, which leads to inactivation of transcription factors and causes their release from mitotic chromatin (42,43). Whether phosphorylation of HsPI3K-C2␣ in early mitosis plays a similar role and controls the enzyme activity in vivo is not clear. The finding that in vitro lipid kinase activity of mitotic, fully phosphorylated HsPI3K-C2␣ is similar to that isolated from interphase cells suggests that phosphorylation per se does not alter the activity of the enzyme. In addition, the activity of phosphorylation-deficient mutants (S259A or S259D) of HsPI3K-C2␣ is similar to that of the wild-type enzyme. 2 Therefore, it is conceivable that phosphorylation of HsPI3K-C2␣ induces modifications in protein-protein or protein-nucleic acid interactions which affect localization of HsPI3K-C2␣ and perhaps prevent the kinase from reaching potential substrates.
Analysis of the protein turnover suggests that mitotic phosphorylation on Ser 259 indirectly controls the activity of HsPI3K-C2␣ by facilitating its degradation at the M/G 1 transition of cell cycle. Several examples of regulated proteolysis have been characterized. In many cases, modification of the substrate by phosphorylation provides a recognition signal for specific E3 ubiquitin-protein ligases, followed by subsequent digestion by the proteasome (44). The ability to inhibit degradation of the phosphorylated form of HsPI3K-C2␣ with MG132 supports the conclusion that degradation takes place in the proteasome. Cell cycle-regulated proteolysis in anaphase depends on anaphase-promoting complex, a multisubunit ubiquitin ligase (E3) (45). Targets of the anaphase-promoting complex contain destruction boxes necessary for ubiquitinationmediated proteolysis (46). Sequences which could represent putative destruction boxes (RXXL) are found in HsPI3K-C2␣, one of them in close vicinity to the phosphorylation site Ser 259 . In addition, a computer-assisted sequence analysis revealed a putative PEST motif (amino acids 519 -531). PEST sequences, which are found in numerous short lived proteins, are assumed to target proteins for degradation via the proteasome, although the exact mechanism is unclear (47). We are currently investigating whether HsPI3K-C2␣ is a substrate for ubiquitination and which domains of the protein are important for degradation.
Further studies are necessary to resolve the concise function of HsPI3K-C2␣ in cell cycle progression. Our findings suggest that elimination of the kinase at the exit from mitosis could be necessary to ensure a proper entry into subsequent G 1 phase. This view is consistent with our observation that overexpression of wild-type HsPI3K-C2␣ and its phosphorylation-deficient mutants (S259A and S259D) in COS-7 and Chinese hamster ovary cells leads to mitotic defects such as multipolar spindle assembly, which results in aberrant cytokinesis and formation of multinucleated cells. 2 10. UV irradiation induces phosphorylationdependent degradation of HsPI3K-C2␣. A, HEK-293 cells stably expressing full-length wild-type HsPI3K-C2␣ (Wt) or mutant HsPI3K-C2␣/S259A (S259A) as GFP fusion proteins were UV irradiated and then let recover for the indicated times. Whole cell extracts were subjected to SDS-PAGE and immunoblot analysis with anti-HsPI3K-C2␣ antibody AXXIII. B, inhibitors of the proteasome block UV-induced degradation of phosphorylated form of HsPI3K-C2␣. HEK-293 cells and the stable clones expressing full-length wildtype HsPI3K-C2␣ or mutant HsPI3K-C2␣/S259〈 as GFP fusion proteins were UV irradiated and allowed to recover in medium for the times shown. When indicated, 2 h after UV irradiation the following inhibitors were added to the cells: 100 M ALLM, 20 M MG132, or 50 M lactacystin for 14 h. Whole cell extracts were subjected to SDS-PAGE and to immunoblot analysis with anti-HsPI3K-C2␣ antibody AXXIII. cytokinesis was observed in cells expressing constitutively active class I PI 3-kinase (p110CAAX) (48). Thus, it is tempting to speculate that down-regulation of PI 3-kinase activity could be an essential step for execution of the mitotic program.
Cellular responses to UV irradiation include the activation of MAP kinase signaling pathways. By using selective inhibitors of different MAP kinases, we demonstrate that UV-induced phosphorylation of HsPI3K-C2␣ on Ser 259 appears to be dependent on the activation of JNK signaling pathway and does not involve ERKs and p38. Accordingly, ectopic expression of the upstream JNKK2/MKK7 increased the steady-state level of HsPIK3-C2␣ phosphorylated on Ser 259 , corroborating the involvement of JNKs in UV irradiation-induced phosphorylation. UV responses meditated by JNK include transcriptional output in the nucleus (49,50) and antiapoptotic signaling events in the cytoplasm (51,52). Prominent changes in the subnuclear localization of HsPI3K-C2␣ after UV irradiation and the observation that only a fraction of HsPI3K-C2␣ is accessible to phosphorylation suggest that JNK might target only nuclear HsPI3K-C2␣.
Similar to mitosis, UV-induced phosphorylation serves as a signal for activation of proteasome-dependent HsPI3K-C2␣ proteolysis. This conclusion is supported by the following observations: (i) there is an apparent preferential disappearance of phosphorylated form of HsPI3K-C2␣ over time after UV irradiation; (ii) treatment of UV-irradiated cells with proteasome inhibitors resulted in a significant stabilization of the phosphorylated form; (iii) the level of phosphorylation-deficient mutant GFP-HsPI3K-C2␣/S259A remained unaffected after UV irradiation.
It is well known that RNA synthesis is down-regulated in response to DNA damage to allow transcription-coupled repair. Inhibition of transcription caused by DNA-damaging agents, such as ␣-amanitin, actinomycin D, cisplatin, and UV irradiation, leads to the degradation of the polymerase II LS (53,54). The irreversible disassembly of transcription complexes as consequence of the degradation of polymerase II LS has been proposed as a mechanism for down-regulation of transcription (55). UV irradiation, similar to ␣-amanitin and actinomycin D treatment, leads to morphological changes of HsPI3K-C2␣positive speckles and induces its phosphorylation, suggesting that these events are linked to transcriptional repression. The fact that HsPI3K-C2␣ follows the fate of the polymerase II LS and is degraded upon cell exposure to UV irradiation could therefore reflect the disassembly of transcriptional complexes.
In summary, our findings suggest that the phosphorylation of HsPI3K-C2␣ on Ser 259 is critical for its subnuclear localization and turnover and that two distinct signaling pathways phosphorylate Ser 259 depending on the physiological state of the cell. Identification of proteins that associate with of HsPI3K-C2␣ and analysis of the regulation of these interactions by phosphorylation are necessary to determine the physiological role of serine phosphorylation in HsPI3K-C2␣ function.