Identification of a PKB/Akt hydrophobic motif Ser-473 kinase as DNA-dependent protein kinase.

Full activation of protein kinase B (PKB)/Akt requires phosphorylation on Thr-308 and Ser-473 by 3-phosphoinositide-dependent kinase-1 (PDK1) and Ser-473 kinase (S473K), respectively. Although PDK1 has been well characterized, the identification of the S473K remains controversial. A major PKB Ser-473 kinase activity was purified from the membrane fraction of HEK293 cells and found to be DNA-dependent protein kinase (DNA-PK). DNA-PK co-localized and associated with PKB at the plasma membrane. In vitro, DNA-PK phosphorylated PKB on Ser-473, resulting in a approximately 10-fold enhancement of PKB activity. Knockdown of DNA-PK by small interfering RNA inhibited Ser-473 phosphorylation induced by insulin and pervanadate. DNA-PK-deficient glioblastoma cells did not respond to insulin at the level of Ser-473 phosphorylation; this effect was restored by complementation with the human PRKDC gene. We conclude that DNA-PK is a long sought after kinase responsible for the Ser-473 phosphorylation step in the activation of PKB.

The signaling pathway centered on protein kinase B (PKB, 1 also called Akt) has emerged as a critical mediator of diverse cellular processes including metabolism, gene expression, migration, angiogenesis, proliferation, and cell survival (1,2). PKB is tightly controlled and the consequences of its deregulation have been implicated in the development of cancers and diabetes (1,2). The activity of PKB is markedly stimulated in a phosphatidylinositol 3-kinase (PI3K)-dependent manner. Upon stimulation, PKB is recruited to the plasma membrane through the binding of its N-terminal pleckstrin homology (PH) domain to phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ), a lipid product of PI3K. PKB␣ is then activated by phosphorylation on two residues: Thr-308 in the activation loop and Ser-473 in the hydrophobic motif of the C-terminal tail (3). There is convincing evidence that Thr-308 is phosphorylated by 3-phosphoinositol-dependent kinase 1 (PDK1) (4,5). In embryonic stem cells in which the PDK1 gene has been genetically disrupted, PKB is resistant to growth factor stimulation as consequence of loss of Thr-308 phosphorylation (5), but phosphorylation on Ser-473 still occurs. Like PKB, PDK1 also contains a PH domain that binds to PIP 3 (4,6). Phosphorylation of Thr-308 in vivo is dependent on PI3K activity, but it is unclear if this requirement is necessary for the unfolding of PKB to allow access of PDK1 to Thr-308 site or direct activation of PDK1 through its PH domain (6,7). Other results indicate that PI3K is important for PKB on Ser-473 because analysis of knock-in embryonic stem cells expressing PDK1 with a mutation in its PH domain revealed that PKB is not activated by insulin-like growth factor-1 (IGF-1), whereas ribosomal S6 kinase (RSK) is activated normally, indicating the importance of colocalization of PKB with PDK1 at the plasma membrane (8).
Identification of the kinase responsible for phosphorylating Ser-473 has been a major challenge for a number of years but remains elusive. Several kinases have been reported to possess Ser-473 phosphorylating activity, including mitogen-activated protein kinase-activated kinase-2 (MAPKAPK-2) (3,9), integrin-linked kinase (10,70), PDK1 (11), and PKB itself (12). However, there is evidence that these kinases are not the physiological PKB Ser-473 kinase (S473K) (3,5,(13)(14)(15)69). Activation of MAPKAPK-2 is PI3K-independent, whereas PKB Ser-473 phosphorylation is sensitive to PI 3-kinase inhibitors (3). PDK1-null cells undergo Ser-473 phosphorylation, suggesting that PDK1 is not required for Ser-473 phosphorylation (5). Furthermore, insulin-stimulated PKB Ser-473 phosphorylation does not require activation of PDK1 or PKB, as Ser-473 phosphorylation is not sensitive to staurosporine treatment, which inhibits both PDK1 and PKB (13). Overexpression of certain kinase inactive mutants can mimic wild type ILK in inducing Ser-473 phosphorylation, suggested that ILK influences PKB phosphorylation indirectly (69). Phosphorylation of PKB/Akt was unaffected in ILK-deficient fibroblasts (16) and chondrocytes (17), indicating that ILK is not required for the phosphorylation of PKB on Ser-473. Moreover, a physiological role of ILK in regulating PKB phosphorylation has been questioned, since ILK knock-out in Drosophila melangaster shows a phenotype more similar to the integrin knock-out than to the PKB knock-out (14). However, ILK knock-out in mouse macrophages resulted in substantial inhibition of Ser-473 phosphorylation (70). We and other groups (13,15,18,19) provided data that strongly argue against autophosphorylation mechanism. The recently solved crystal structure of PKB␤ reveals the importance of Ser-473 phosphorylation for PKB activation and strongly argues that autophosphorylation cannot be the physiological mechanism of Ser-473 phosphorylation (20).
We have previously identified S473K activity present in lipid rafts of plasma membrane from HEK293 cells (15). In this paper we report the purification and identification of the Ser-473 kinase(s) from HEK293 cell membrane extracts. We show here that DNA-dependent protein kinase (DNA-PK) is a dominant upstream kinase of PKB that specifically targets PKB phosphorylation on Ser-473 in vitro and in vivo.
Preparation of Plasmids and Proteins-pGEX2T-PKB␣ 419 -480 was prepared by inserting a PCR product using human PKB␣ cDNA as template into the pGEX2T vector (Amersham Biosciences). GST fusion proteins were expressed in Escherichia coli BL21 strain and purified on glutathione-Sepharose 4B (Amersham Biosciences) according to the manufacturer's instructions. Baculovirus-expressed ⌬PH-PKB␤ T309P was prepared as described in Ref. 20.
Purification of PKB Ser-473 Kinase-All steps were performed at 4°C. Four-hundred plates (Ø 150 cm) of HEK293 cells were harvested in ice-cold phosphate-buffered saline and homogenized in ice-cold buffer A (50 mM Tris-HCl, pH 7.4, 300 mM sucrose, 1 mM dithiothreitol, 10 mM sodium fluoride, 0.1 mM sodium orthovanadate, 20 nM okadiac acid, 1 mM phenylmethanesulfonyl fluoride, and 1 mM benzamidine) using a Polytron homogenizer. The nuclear fraction was removed by centrifuging at 1000 ϫ g for 10 min, and the resulting supernatant was further centrifuged at 100,000 ϫ g for 60 min. The crude membrane pellets were suspended in buffer B (20 mM Tris-HCl, pH 7.4, 0.5 M NaCl, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, and 1 mM benzamidine). Following a 30-min incubation at 4°C, insoluble proteins were pelleted by centrifugation at 100,000 ϫ g for 20 min, and the supernatant was dialyzed against buffer C (20 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 mM benzamidine, and 5% glycerol) for 2 h. After a brief centrifugation, the supernatant was loaded on to a Q-Sepharose Fast Flow column (2.6 ϫ 8 cm) pre-equilibrated with buffer C. The column was then extensively washed with buffer C plus 0.1 M NaCl and developed with a continuous gradient of NaCl (0.1-0.6 M) in buffer C. The fractions containing S473K activity were pooled and dialyzed against buffer C for 2 h prior to loading onto a Mono S (HR10/10) column. After washing with 10 volumes of buffer C, proteins were eluted with a continuous NaCl gradient (0.1-0.6 M) in buffer C. The kinase activity eluted between 0.34 and 0.38 M NaCl. Fractions were pooled and dialyzed against buffer C. The dialyzed samples were applied onto a Mono Q (HR5/5) column. The column was washed with 10 volumes of buffer C, and the proteins were eluted in a continuous NaCl gradient (0.1-0.5 M) in buffer C. All columns were attached to an fast performance liquid chromatography system (Amersham Biosciences).
Gel Filtration Analysis-Gel filtration was performed using a Superdex-200 HR10/30 column attached to an fast performance liquid chromatography system (Amersham Biosciences) as described previously (15).
Protein Kinase Assays-S473K activity was assayed (15) using the peptide RRPHFPQFSYSASSTA corresponding to the C terminus of PKB␣ (FSY peptide). A second peptide in which Ser-473 was changed to alanine (FAY peptide) was used to measure phosphorylation on other residues as described previously (15). In parallel, the S473K activity was also monitored by an alternative kinase assay method using GST-PKB␣ 419 -480 as substrate followed by Western blotting with a Ser-473 phospho-specific antibody. Briefly, assays were performed in 30-l reactions containing 30 mM Tris-HCl, pH 7.4, 1 mM dithiothreitol, 10 MgCl 2 , 1 M PKI, 1 M microcystin-LR, 150 M ATP, 1 g of GST-PKB␣ 419 -480 , and enzyme. After incubating for 30 min at 30°C, reactions were stopped by adding SDS sample buffer, boiled for 3 min at 95°C, and then resolved by SDS-PAGE followed by Western blotting with phospho-specific Ser-473 antibody. The in vitro PKB kinase assay was as described previously using the specific peptide RPRAATF (R7Ftide) as substrate (26).
Cell Culture and Transfections-HEK293 and 3T3L1 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin. For RNA interference, HEK293 cells were transfected with the DNA-PKcs small interfering RNA (siRNA) or a 21-nucleotide irrelevant RNA duplex as a control using oligofectamine (Invitrogen). The targeted sequence of human DNA-PKcs siRNA selected was 5Ј-AGGGC-CAAGCTGTCACTCT-3Ј (sense sequence nucleotides 5065-5083, Gen-Bank TM accession number NM_006904), and the 21-nucleotide synthetic siRNA duplex was synthesized by Qiagen. M059J and M059K were from ATCC and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1% non-essential amino acids, 4 mg/ml glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. M059J/Fus1 cells were grown in the same medium containing 250 g/ml Geneticin. For analytical experiments, cells were starved overnight prior to treatment and the cells were lysed in Nonidet P-40 lysis buffer (21).
Immunofluoresence-For immunofluorescence, cells were fixed with 3.7% formaldehyde and permeabilized with 0.2% Triton X-100. After blocking, cells were incubated with monoclonal anti-DNA-PK antibody (G4, 2 g/ml diluted in phosphate-buffered saline) for 2 h, followed by polyclonal anti-PKB (Ab10) antibody (2 g/ml diluted in phosphatebuffered saline) for a further 2h. Cells were then extensively washed and incubated with fluorescein isothiocyanate-conjugated anti-mouse IgG and TR-conjugated anti-rabbit IgG (both 1:100 dilution). After extensive washing, the antibodies were visualized using confocal microscopy.
Immunoprecipitation-Extracts (500 g) were incubated in co-immunoprecipitation buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM benzamidine, 1 mM phenylmethanesulfonyl fluoride) with monoclonal DNA-PK (G4, 2 g) or monoclonal PKB (A4D6, 2 g) antibodies and protein G beads (20 l, Amersham Biosciences). After 6 h continuous gentle agitation at 4°C, the beads were collected by centrifugation and washed three times in co-immunoprecipitation buffer, after they were resuspended in SDS sample buffer and heated at 95°C for 3 min before analysis by 6% SDS-PAGE followed by Western blotting.

Purification and Identification of DNA-PKcs as a Ser-473
Kinase-We previously (15) identified a S473K activity in lipid rafts of plasma membrane and now report an efficient method for purification and identification of this kinase from HEK293 cells. The purification procedure was monitored using a peptide derived from the hydrophobic motif of PKB␣ containing Ser-473 (FSY) and a mutant peptide with Ser-473 mutated to Ala (FAY) as substrates as described in Ref. 15. In parallel, specific S473K activity was also monitored using GST-PKB␣ 419 -480 (GST fusion protein containing the hydrophobic motif site) as substrate, followed by Western blotting with a Ser-473 phospho-specific antibody as described under "Experimental Procedures." This combined assay approach allowed the identification of the enzymes specifically targeting Ser-473 of PKB. The procedures for purification of S473K, including membrane fractionation, chromatography on Q-Sepharose, Mono S, and Mono Q columns are described under "Experimental Procedures." The elution profile from the final Mono Q step ( Fig. 1) shows two distinct kinase activities. S473K1 activity in fraction 27 specifically phosphorylated recombinant GST-PKB␣ 419 -480 , with only poor activity toward the peptides (inset in Fig. 1A). The second activity (S473K2) peaking at fraction 30 phosphorylated both recombinant PKB␣ 419 -480 and Ser-473 peptides (Fig. 1A). SDS-PAGE analysis of Mono Q fractions revealed one major band of apparent molecular mass ϳ350-kDa whose elution profile paralleled S473K1 activity (Fig. 1, A and B). Furthermore, the 350-kDa bands co-eluted with S473K1 activity on gel filtration chromatography (Fig. 2). The band corresponding to the 350-kDa proteins was analyzed by mass spectrome-try, and the sequences of 65 peptides obtained all showed a perfect match to the human DNA-PKcs (data not shown). Western blot analysis of Mono Q fractions revealed that the 350-kDa bands corresponded to DNA-PKcs, which exactly paralleled S473K1 activity (Fig. 1C). DNA-PKcs is a member of the phosphatidylinositol 3-kinase-related kinase (PIKK) family that includes mammalian target of rapamycin (mTOR), the ataxia telangiectasia gene product (ATM), and ATM-and RAD-3related kinase (ATR) (27). The holoenzyme of DNA-PK consists of a 465-kDa catalytic subunit (DNA-PKcs) and the Ku antigen complex Ku70/Ku80 (27)(28)(29) that has been reported to be required for V(D)J recombination, DNA repair, and transcriptional regulation (27). Interestingly, the peaks of two other DNA-PK components, Ku70/Ku80, were partially resolved from the DNA-PK peak (Fig. 1C). This suggests that the partially purified DNA-PKcs can significantly phosphorylate PKB in the absence of Ku70/Ku80 subunits. Gel filtration analysis of Mono Q fraction 27 showed that the S473K1 activity eluted at ϳ350 kDa and closely paralleled DNA-PKcs (Fig. 2). These results suggest that DNA-PKcs is not always complexed with the Ku70/Ku80 subunit, consistent with several earlier observations (27)(28)(29). As reported previously DNA-PK can be acti-vated by DNA ends (29); we therefore tested the DNA dependence of S473K1 by assaying the kinase activity of Mono Q fractions in the presence of linear double-stranded fetal calf thymus DNA. As expected, the S473K1 activity toward the substrate FSY peptide was robustly stimulated by DNA and peaked at fraction 27, corresponding exactly to DNA-PKcs (Fig.  1, A-C). It is worth noting that DNA also stimulates S473K2 activity probably due to contamination of S473K2 fractions with DNA-PKcs (Fig. 1, A and B). We are currently characterizing the protein kinase activities in the S473K2 and will be reported elsewhere.
To test whether Ser-473 phosphorylation catalyzed by purified S473K1 is sufficient for activation of PKB activity, a monophosphorylated form of recombinant PKB (⌬PH-PKB␤ T309P prepared by in vitro phosphorylation with PDK1 (20)) was first phosphorylated by S473K1 and PKB assayed using RPRAATF (R7Ftide) as substrate (26). As shown in Fig. 3A, ⌬PH-PKB␤ T309P activity toward R7Ftide was increased 10-fold after incubation with S473K1. Phosphorylation of ⌬PH-PKB␤ T309P on Ser-474 (identical to Ser-473 in PKB␣) was also increased in a time-dependent manner (Fig. 3A). Significantly we carried out in vitro phosphorylation reactions of DNA-PK with ⌬PH-PKB␤ T309P in the presence of 5 M staurosporine; this revealed that Ser-474 phosphorylation was unaffected by the inhibitor. This rules out the possibility that DNA-PK promotes PKB autophosphorylation because under these conditions PKB is fully inhibited by the inhibitor.
We further characterized the S473K1 fraction using PI3K inhibitors known to be effective inhibitors of DNA-PK activity (27,28). As shown in Fig. 3, B and C, S473K1 was potently inhibited by 5 M LY294002 and 2 M wortmannin suggesting that DNA-PKcs is a major active component of S473K1. This result is consistent with previous cell culture data that showed Ser-473 phosphorylation is sensitive to these two inhibitors (3). Our previous studies revealed that S473K is staurosporine insensitive (13). Consistent with this, S473K1 activity was not affected by 5 M staurosporine (Fig. 3, B and C). Based on these results, we conclude that DNA-PK is S473K1. In contrast, the S473K2 fraction is sensitive to staurosporine (data not shown), suggesting that S473K2 is distinct from S473K1.
DNA-PK Activity Is Required for Ser-473 Phosphorylation and Activation of PKB-To confirm that DNA-PK is S473K1, purified DNA-PK (from HeLa cells and human placenta) was tested for their ability to phosphorylate PKB on Ser-473. As shown in Fig. 4, A and B, both the FSY peptide and GST-PKB␣ 419 -480 were efficiently phosphorylated by purified DNA-PK in the presence of DNA. Strikingly, DNA-PK significantly phosphorylated GST-PKB␣ 419 -480 in the absence of added DNA but did not significantly phosphorylate FSY peptide in the absence of DNA (Fig. 4B). This result is consistent with S473K1 activity observed in assays of chromatographic fractions (Fig. 1A). In addition, the phosphorylation of PKB catalyzed by purified DNA-PK was inhibited by LY294002 and wortmannin but was resistant to staurosporine (Fig. 4, A and  B). Next, we tested the effect of phosphorylation of ⌬PH-PKB␤ T309P by purified DNA-PK under the conditions described above. As shown in Fig. 4C, PKB activity toward R7Ftide was enhanced more than 8-fold with increasing Ser-474 phosphorylation in a time-dependent manner, similar to that observed with our partially purified S473K1 fraction (Fig. 3A). The phosphorylation of ⌬PH-PKB␤ T309P on Ser-474 was also not affected by staurosporine (data not shown). Incubation of ⌬PH-PKB␤ T309P with Mg 2ϩ /ATP in the absence of DNA-PK failed to increase both Ser-474 phosphorylation and PKB activity, thus ruling out the possibility of autophosphorylation (data not shown). These results suggest that Ser-474 phosphorylation is due to DNA-PK. The stoichiometry of GST-PKB␣ 419 -480 phosphorylation was ϳ0.3 mol of phosphate/mol of protein, comparable with phosphorylation of p53 by DNA-PK (Fig. 4D). Thus, the properties of nearly homogenous DNA-PK purified from either Hela cells or placenta coincide perfectly with the properties of our partially purified S473K1 activity, further supporting the concept that DNA-PK is a S473K1.
Impairment of Ser-473 Phosphorylation in DNA-PKcs siRNA-treated Cells and in DNA-PKcs-deficient Cells-To con-firm that the DNA-PK can function as a S473K in vivo we transfected HEK293 cells with specific siRNA leading to a marked loss of DNA-PKcs protein (Fig. 5A, upper panel). Stimulation of siRNA-transfected cells with insulin or pervanadate markedly reduced PKB phosphorylation on Ser-473 phosphorylation but did not markedly affect Thr-308 phosphorylation (Fig. 5A). Similar results were also obtained by knockdown of DNA-PKcs in human fetal lung MRC5 fibroblasts (data not shown). Thus, DNA-PK again appears to be required for Ser-473 phosphorylation. The human glioblastoma cell line M059J lacks DNA-PKcs and DNA-PK activity, while the related cell line M059K contains wild type DNA-PK (30). Treatment of cells with insulin increased phosphorylation of both Ser-473 and Thr-308 residues in M059K cells, but there was no significant increase in M059J cells (Fig. 5B). Surprisingly the level of phosphorylation of both sites was higher in starved M059J cells suggesting further alterations in PKB regulation in this transformed cell. To further determine whether the insulin-resistant phenotype of PKB phosphorylation is due to a defect in DNA-PK in M059J cells, rescue experiments were performed with M059J/Fus1 cells. M059J/Fus1 cells were generated from M059J by cell fusion leading to DNA-PK expression because a fragment of chromosome 8 with the PRKDC gene is integrated into these cells (31). The results in Fig. 5B clearly show that M059J/Fus1 cells regained an insulin-sensitive phenotype for Ser-473 phosphorylation similar to that of M059K. Thus Ser-473 phosphorylation can be restored by complementation of DNA-PKcs-defective cells with a fragment of chromosome 8 containing a copy of the PRKDC gene, again implicating DNA-PK as a determinant kinase responsible for Ser-473 phosphorylation. This persistent but low Ser-473 phosphorylation in DNA-PKcs-deficient cells (Fig. 5, A and B) indicates the existence of additional kinase(s) that can phosphorylate the hydrophobic motif site Ser-473. This could correspond to the S473K2 activity identified in the Mono Q fraction (Fig. 1A). Interestingly in the absence of DNA-PK, or because of cell transformation, this activity appears to be constitutively up-regulated.
DNA-PK Localizes at the Plasma Membrane and Interacts with PKB-It has been shown that efficient DNA repair re- quires growth factor signaling (32) and that this effect may be due to the physical association of DNA-PK with epidermal growth factor receptor (33). DNA-PK can be detected in both the nucleus and cytoplasm (34,35). Significantly inositol hexakisphosphate (IP 6 ) was reported to bind to DNA-PK (36 -38), suggesting that IP 6 might play a pivotal role in modulating the localization and/or biological properties of DNA-PK. More recent data indicate a novel localization of the DNA-PK complex in lipid rafts (39) in line with our finding that PKB S473K1 is associated with plasma membrane lipid rafts (15). Subcellular fractionation of HEK293 cells revealed substantial DNA-PKcs in the membrane fraction as shown in Fig. 6A. Both the cytosolic marker protein ␤-tubulin and nuclear marker protein lamin A/C were absent in the membrane fraction indicating that no significant contamination of the membrane fraction from cytosol and nuclear fractions. Similarly the Ku subunits were also found in both membrane and nuclear fractions (data not shown).
We also tested whether DNA-PKcs and PKB co-localized in unstimulated 3T3L1, M059K, and M059J cells (Fig. 6B). DNA-PKcs was present mainly in the nucleus but with a substantial amount also in the cytosol and at the membrane, whereas PKB . Aliquots (40 g) of cell extracts were analyzed by Western blotting using phospho-specific antibodies pS473, pT308, and total protein PKB using Ab10 and DNA-PKcs protein with monoclonal G4 antibody. B, M059K, M059J, and Fus1 cells were starved for 24 h and the cells treated with insulin (Ins) for different times before harvest. Aliquots (30 g) of cell extracts were analyzed by Western blotting using phopho-specific antibodies pS473, pT308 and total PKB protein with Ab10.
was mainly cytosolic with a fraction associated with the membrane in 3T3Ll and M059K cells. Significantly DNA-PK colocalized with PKB at the plasma membrane in these cells. No DNA-PKcs staining was observed in the membrane and nucleus in the DNA-PKcs-deficient M059J cells (Fig. 6B). To determine whether DNA-PKcs associates with PKB, HEK293 and M059K cells were starved and treated with insulin for 30 min. Cells were lysed in Nonidet P-40 lysis buffer and immunoprecipitated with DNA-PK or PKB monoclonal specific antibodies (Fig. 6C). The results reveal that PKB is associated with DNA-PKcs and following insulin treatment was phosphorylated on Ser-473. DISCUSSION Here we report for the first time the identification of the DNA-PK as a PKB hydrophobic motif (Ser-473) kinase. We predicted, from a number of different experimental approaches, that the Ser-473 kinase should be located at the membrane. Following the development of appropriate assays (15) we have now identified the predicted kinase as DNA-PK, a member of the PIKK subfamily of protein kinases. This is a surprising result and probably explains why it has been so difficult to identify this kinase. This study now opens up many significant possibilities in terms of fully understanding PKB regulation and also extends the roles of the DNA-PK. Significantly mTOR, a close relative of DNA-PK in the PIKK family, phosphorylates p70S6 kinase on Thr-389 of hydrophobic motif (40). In addition, mTOR also plays a role in controlling phosphorylation of novel PKC␦ and novel PKC⑀ on the hydrophobic C-terminal site (41).
Our results would appear to rule out a number of earlier proposed mechanisms for the regulation of PKB Ser-473 hydrophobic motif phosphorylation. First, we found no evidence for autophosphorylation on this site using purified PKB (⌬PH-PKB␤ T309P ) prepared for crystallographic studies (19). 2 Also DNA-PK phosphorylated Ser-473 under conditions where PKB activity is completely inhibited by staurosporine, thus ruling out the possibility that DNA-PK promotes autophosphorylation. 2 A number of other kinases have been proposed (see Introduction) to phosphorylate the hydrophobic motif of PKB. None of these candidates were identified in the Mono Q fraction (Fig. 1) we extensively studied in this paper (all protein bands in the purified S473K fraction were analyzed by mass spectroscopy yielding a total of about 150 proteins).
We found that a fraction of DNA-PK localized to the membrane, consistent with a previous observation that DNA-PK complex could be recovered from lipid rafts (39). The mechanism of how DNA-PK localizes to the membrane is not clear but DNA-PK has been reported to be associated with epidermal growth factor receptor (33). Recent findings revealed that the inositol phosphates, especially IP 6 , serve as a potent co-factor for DNA-PK activity in non-homologous end-joining via binding to the Ku subunits (36 -38). IP 6 is an abundant molecule in cells, and it is reported that IP 6 can enhance Ku mobility (42,43). It is likely that IP 6 might play a significant role in modulating the localization of DNA-PK. DNA-PKcs is a large protein of about 4100 amino acids with the kinase domain located at the C-terminal of the polypeptide. A recent analysis predicts that DNA-PKcs contains several domains located N-terminally to the kinase domain, including two different type helical repeat motifs that specifically interact with some known and other as yet unidentified proteins (44,45). These helical regions are thought to interact with Ku70/ Ku80 subunits, c-Abl, and other factors to modulate DNA-PK function. It will be important in the future to investigate how this large multifunctional enzyme is acutely regulated at membrane. It is established that DNA-PKcs is a phosphoprotein, and indeed DNA-PK activity is regulated by phosphorylation (27). Several of the phosphorylation sites are conserved in vertebrate species (46,47) and become phosphorylated in vivo following okadaic acid treatment (46). Phosphorylation on Thr-2638/Thr-2647 is essential for radioresistance conferred by DNA-PKcs (48). Significantly, protein phosphatase 5 interacts with DNA-PKcs and preferentially dephosphorylates Thr-2609 and a lesser extent to Ser-2056 (49). Overall it will be necessary to reinvestigate the phosphorylation of DNA-PKcs in response 2 J. Feng, J. Park, and B. A. Hemmings, unpublished data. to growth factors and DNA damage to establish the role of this modification in regulating kinase activity, localization, and interaction with other proteins. A recent study (50) showed that the ATM protein kinase is a dimer with the kinase domain bound to Ser-1981 contained in the FAT domain. Following cellular irradiation Ser-1981 becomes autophosphorylated and thus promotes the dissociation of the complex. Apparently this event leads to an activation of ATM activity. It has been reported that c-Abl and PKC␦ interacts and phosphorylates DNA-PKcs, which leads to inhibition of the ability of DNA-PK to form a complex with DNA (51,52). The fact that DNA can robustly activate DNA-PKcs suggests there could be several mechanisms for activating the kinase. It is possible that alternative mechanisms are utilized in different subcellular locations.
Although DNA was originally shown to be a potent activator of DNA-PKcs, recent studies (53)(54)(55) indicate that the multiple proteins are recruited by the Ku subunits that may also serve as activators to stimulate DNA-PK activity. For instance, several homeodomain proteins, including Oct-1, have recently been shown (53) to interact with Ku70 and enhance DNA-PK phosphorylation. Consistent with these findings, thyroid hormone receptor-binding protein interacts with Ku70 subunit, resulting in the stimulation of DNA-PK activity in the absence of DNA (55). The DNA-independent stimulation of DNA-PK by protein factors was also observed with other proteins such as C1D DNA-binding protein (56,57). In addition, an anti-DNA-PKcs antibody synergistically activates DNA-PK together with thyroid hormone receptor-binding protein (55), strongly indicating that DNA-PK can be fully activated in the absence of DNA. It appears that although DNA-PK can be stimulated by DNA ends (27), or by other kinases (51,52), the large DNA-PK protein may have a scaffold function and the protein-protein interactions via the heterodimeric Ku subunits may be another important but previously less appreciated mechanism for DNA-PK activation.
PKB belongs to the AGC family of protein kinases that possess a highly conserved activation loop phosphorylation site in the central kinase domain and a hydrophobic motif phosphorylation site in the C terminus (reviewed in Ref. 58). The hydrophobic motif of most AGC kinases is characterized by a conserved motif: F-X-X-F-S/T-Y/F (the S/T residue is equivalent to Ser-473 of PKB␣) (58). DNA-PK can phosphorylates many protein substrates on Ser/Thr residues followed by glutamine, i.e. the "S/TQ" motif (59,60). However, DNA-PK also phosphorylates proteins at so-called "non-S/TQ" sites, with a preference for Ser/Thr residues followed by a hydrophobic amino acid (27,59,60). It is noteworthy that DNA-PK appears to have a predisposition for phosphorylation sites at the extreme termini of its substrates; this may indicate better accessibility of the substrate to the active site of this large kinase complex (60). The substrate specificity of DNA-PK warrants further investigation; our initial data indicate the PKB hydrophobic motif FSY peptide is about three times more effective than the p53 peptide (data not shown). We also tested DNA-PK with several other hydrophobic motif peptides from RSK1, RSK3, PKC␣, NDR2, S6K-1, and SGK-1 modeled on the PKB site and found that only those from the three PKB isoforms served as substrate (data not shown). Furthermore, our in vitro results with PI3K inhibitors on DNA-PK activity suggest that at high concentrations in vivo these compounds can directly inhibit PKB phosphorylation on Ser-473. Interpretation of results obtained with LY294002 and wortmannin need to be reevaluated because of their apparent direct inhibition of PKB phosporylation.
DNA-PK is activated upon DNA damage by UV irradiation, as is PKB (61). Induction of apoptosis by cisplatin was ex-plained by a decrease in DNA-PK activity through proteolytic degradation of DNA-PK (62); significantly, PKB activity and Ser-473 phosphorylation are also inhibited by cisplatin treatment (63). Mutation or deletion of the gene encoding DNA-PKcs is responsible for severe combined immune deficiency (SCID) (reviewed in Ref. 64) and is consistent with studies based on recently generated DNA-PKcs knock-out mice and cell lines, which display a similar phenotype to the mouse SCID cell line (65,66). Mouse and human cells deficient in DNA-PK are hypersensitive to ionizing radiation and to radiomimetic drugs (27,65,66); a similar phenotype can be observed in Akt1/PKB knock-out mice (67) and with mouse embryo fibroblasts derived from such mice (68). Our results using RNA interference on HEK293 cells, and studies of the M059J and K cells, and with M059J/Fus1 cells, provide compelling evidence that DNA-PK phosphorylates PKB in vivo. It will be necessary in future to investigate the role of DNA-PK in the activation of PKB using SCID or DNA-PKcs knock-out mice to obtain additional evidence for this novel regulatory mechanism.
The identification of DNA-PK as the elusive Ser-473 hydrophobic motif kinase is unexpected, and several important questions are posed by this discovery. DNA-PK pathway plays a crucial role in controlling transcription, the cell cycle progression, and apoptosis (27). Similarly PKB is also implicated in the regulation of many different cellular processes (1,2). It is now important to integrate these two fields and to understand precisely how the PI3K pathway influences DNA-PK. Our current view of how PKB is regulated by PDK1/DNA-PK is shown in Fig. 7 and also shows the parallels with p70S6K regulation by FIG. 7. Model for PKB activation by upstream kinases. Following stimulation by growth factors, PI3K is activated, which in turn generates the second messenger PIP 3 to recruit PKB and PDK1 to the membrane lipid rafts, where PKB is subsequently phosphorylated on Thr-308 by PDK1 and on the hydrophobic motif Ser-473 by DNA-PK. S6 kinase is also included to show the parallels for the regulation by activation loop and hydrophobic site phosphorylation by PDK1 and mTOR. PDGF, platelet-derived growth factor; EGF, epidermal growth factor; IGF-1, insulin-like growth factor-1. PDK1/mTOR. This leads to the suggestion that possibly other members of the PIKK family could function as hydrophobic motif kinases. It will of course be necessary to identify the kinase in the S473K2 faction to fully elucidate the regulation of PKB. However, the identification of a second protein kinase with potential to phosphorylate for Ser-473 leads to the speculation that possibly different stimuli (growth factors, chemokines, DNA damage) activate specific protein kinase to phosphorylate the key hydrophobic motif regulatory phosphorylation site. Furthermore our data (Fig. 5B) with the transformed glioblastoma cell line M059J appears to indicate that this kinase is deregulated in these cells leading to constitutive phosphorylation on the hydrophobic motif even in the absence of the DNA-PK.