mTOR·RICTOR Is the Ser473 Kinase for Akt/Protein Kinase B in 3T3-L1 Adipocytes*

The insulin-signaling pathway leading to the activation of Akt/protein kinase B has been well characterized except for a single step, the phosphorylation of Akt at Ser-473. Double-stranded DNA-dependent protein kinase (DNA-PK), ataxia telangiectasia mutated (ATM) gene product, integrin-linked kinase (ILK), protein kinase Cα (PKCα), and mammalian target of rapamycin (mTOR), when complexed to rapamycin-insensitive companion of mTOR (RICTOR), have all been identified as playing a critical role in Akt Ser-473 phosphorylation. However, the apparently disparate results reported in these studies are difficult to evaluate, given that different stimuli and cell types were examined and that all of the candidate proteins have never been systematically studied in a single system. Additionally, none of these studies were performed in a classical insulin-responsive cell type or tissue such as muscle or fat. We therefore examined each of these candidates in 3T3-L1 adipocytes. In vitro kinase assays, using different subcellular fractions of 3T3-L1 adipocytes, revealed that phosphatidylinositol 3,4,5-trisphosphate-stimulated Ser-473 phosphorylation correlated well with the amount of DNA-PK, mTOR, and RICTOR but did not correlate with levels of ATM, ILK, and PKCα. PKCα was completely absent from compartments with Ser-473 phosphorylation activity. Although purified DNA-PK could phosphorylate a peptide derived from Akt that contains amino acid Ser-473, it could not phosphorylate full-length Akt2. Vesicles immunoprecipitated from low density microsomes using antibodies directed against mTOR or RICTOR had phosphatidylinositol 3,4,5-trisphosphate-stimulated Ser-473 activity that was sensitive to wortmannin but not staurosporine. In contrast, immunopurified low density microsome vesicles containing ILK could not phosphorylate Akt on Ser-473 in vitro. Small interference RNA knockdown of RICTOR, but not DNA-PK, ATM, or ILK, suppressed insulin-activated Ser-473 phosphorylation and, to a lesser extent, Thr-308 phosphorylation in 3T3-L1 adipocytes. Based on our cell-free kinase and small interference RNA results, we conclude that mTOR complexed to RICTOR is the Ser-473 kinase in 3T3-L1 adipocytes.

Insulin increases glucose transport in muscle and fat by stimulating the translocation of the glucose transporter isoform Glut 4 from intracellular storage pools to the cell surface (1,2), a process that requires the activation of the serine/threonine kinase Akt (protein kinase B) (3)(4)(5)(6)(7)(8).
Although the molecular events that occur after Akt activation are poorly understood, the early insulin signaling pathway has been well characterized except for one step, the phosphorylation of Akt at Ser-473. The signaling cascade is initiated when insulin binds to specific receptors at the plasma membrane (PM), 2 triggering the autophosphorylation of several critical intracellular tyrosine residues and thereby activating an intrinsic tyrosine kinase that can phosphorylate cellular substrates, most notably the insulin receptor substrate (IRS) proteins (9 -11). Tyrosine-phosphorylated IRS proteins can recruit and activate phosphoinositide 3-kinase (PI 3-kinase), which generates phosphatidylinositol 3,4,5-trisphosphate (PIP3), using inositol-containing phospholipids in the PM as substrates (12). The activity of Akt is markedly stimulated in a PI 3-kinase-dependent manner (13). This phenomenon predominantly relies on the phosphorylation of Akt on two of its amino acid residues: 1) threonine 308 in the activation loop of the kinase catalytic domain and 2) serine 473 in the "hydrophobic motif " (HM) carboxylterminal domain (14). The phosphorylations of both of these regulatory sites occur after the recruitment of Akt to the PM through the binding of its pleckstrin homology (PH) domain to PIP3 (15) and are completely ablated in vivo by the PI 3-kinase inhibitor wortmannin (16). The protein kinase responsible for phosphorylating Akt on Thr-308 is phosphoinositide-dependent kinase 1 (17)(18)(19). The identity of the kinase responsible for phosphorylating Akt on Ser-473 is controversial.
Akt is part of the protein kinase AGC (cAMP-dependent, cGMP-dependent, and protein kinase C) subfamily whose members are activated in a manner similar to Akt (20). All members possess a phosphorylation site in the activation loop equivalent to Thr-308, whereas only some require phosphorylation at the HM site. Some AGC members have a negatively charged acidic residue instead of a Ser/Thr in the C-terminal hydrophobic domain (21). Phosphorylation of the HM site or substitution with an acidic residue provides a docking site to recruit phosphoinositide-dependent kinase 1 and stimulate the phosphorylation of Thr-308 (20). X-ray crystallographic studies have suggested that in the inactive state the activation loop of Akt adopts a disordered structure that prevents the binding of ATP and protein substrates (22). Phosphorylation of Ser-473 results in a disorder to order transition, allowing interaction between the HM domain and the N-terminal lobe leading to the activation of the kinase (23).
At least 10 kinases have been proposed to function as the hydrophobic motif domain protein kinase (HMD-PK) that phosphorylates Akt on Ser-473, including integrin-linked kinase (ILK), protein kinase C␣ (PKC␣), double-stranded DNA-dependent protein kinase (DNA-PK), * This work was supported in part by National Institute of Health Grant R01 DK067229 and a research grant from the American Diabetes Association. 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 U.S.C. Section 1734 solely to indicate this fact. 1 To whom all correspondence should be addressed: Dept ataxia telangiectasia mutated (ATM) gene product, and the mammalian target of rapamycin (mTOR) (24). It has also been suggested that Akt can undergo autophosphorylation at Ser-473 (25); however, this is most likely not the mechanism by which insulin stimulates Ser-473 phosphorylation (15,16,26,27). Of all of the proposed candidates, evidence supporting DNA-PK and mTOR is the most compelling. Based on the observation that membrane localization is sufficient to activate Akt (15), Hemmings and co-workers used a biochemical approach to purify a constitutively active membrane kinase from HEK 293 cells capable of phosphorylating Akt on Ser-473 (28). The kinase, DNA-PK, was found to associate and co-localize with Akt at the PM as well as phosphorylate and activate Akt 10-fold in vitro (29). These results were surprising, since DNA-PK has been described as a nuclear protein involved in DNA replication, gene transcription, and DNA repair (30). Nevertheless, knockdown studies with small interference RNA (siRNA) and experiments using DNA-PK-deficient glioblastoma cells revealed that insulinstimulated Ser-473 phosphorylation was greatly impaired in both cases (29). Like DNA-PK, mTOR is a member of the family of PI 3-kinaserelated kinases whose activities are strongly inhibited by wortmannin and LY294002 (31,32). mTOR can regulate cell growth and proliferation through its activation by growth factors and nutrients (33). It is also the best characterized HM kinase, capable of phosphorylating the hydrophobic motif of p70S6 kinase when mTOR is complexed with G␤L (34) and regulatory associated protein of mTOR (RAPTOR) (35). Phosphorylation at the HM site of p70S6K (32) but not Akt (36,37) is inhibited by rapamycin. It is for this reason that mTOR was initially ruled out as being the HMD-PK for Akt. Recently, however, Sabatini and co-workers (31,38) have shown that mTOR can associate with G␤L and another protein called rapamycin-insensitive companion of mTOR (RICTOR), forming a rapamycin-insensitive complex capable of phosphorylating Akt at Ser-473. RNA interference experiments in Drosophila and human cell lines as well as several in vitro studies provide intriguing evidence that the mTOR⅐RICTOR complex can phosphorylate the HM site of Akt (31). It is still not exactly clear which if any of the proposed candidate kinases are the HMD-PK for Akt in an insulin-responsive tissue such as muscle and fat. In fact, it has been suggested that phosphorylation of Ser-473 may be due to multiple kinases that are cell type-or signaling pathway-specific (24). Several years ago, we characterized the phosphorylation of Akt at Ser-473 using a cell-free insulin-signaling assay that we developed using subcellular fractions from 3T3-L1 adipocytes (27,39). We found that insulin-stimulated Akt phosphorylation at Ser-473 was PI 3-kinase-dependent and due to a kinase localized in both low density microsomes (LDM) and a subpopulation of the PM fraction enriched in proteins associated with the actin cytoskeleton (Ext-HiP) (27). Utilizing the cell-free system as well as in vivo siRNA knockdown studies, we addressed whether ILK, PKC␣, DNA-PK, ATM, or mTOR was involved in the insulin-stimulated phosphorylation of Akt at Ser-473 in 3T3-L1 adipocytes. Our results indicate that the mTOR⅐RICTOR complex is the HMD-PK for Akt in fat cells.

EXPERIMENTAL PROCEDURES
Cell Culture of 3T3-L1 Adipocytes-3T3-L1 preadipocytes obtained from the American Type Culture Collection were grown to confluence and 48 h later subjected to differentiation as described previously (40). 3T3-L1 adipocytes were used 10 -14 days after initiating differentiation except where otherwise noted.
In Vitro Assay-Mature 3T3-L1 adipocytes were serum-starved overnight and then fractionated in the basal state as described previously, except that 1 mM dithiothreitol was included in HES (50 mM Hepes, pH 7.4, 255 mM sucrose, 1 mM EDTA, and protease inhibitors) and IC (20 mM Hepes, pH 7.4, 140 mM potassium glutamate, 5 mM NaCl, 1 mM EGTA, and protease inhibitors) buffers (39,41). Protease inhibitors (Sigma) were as described (39). Ext-HiP and salt-washed plasma membranes were isolated in the manner described previously (27). In vitro reactions were prepared by mixing various combinations of cytosol (CYT), PM, nuclear (NUC) enriched fraction, LDM, and Ext-HiP with 16 nM Akt2 (Upstate Biotechnology, Inc., Lake Placid, NY) and 10 M PIP3 (Calbiochem), prepared in a sonicated mixture of 100 M phosphatidylcholine (Avanti Polar Lipids) and 100 M phosphatidylserine (Avanti Polar Lipids). Reactions (typically 100 l, final volume) were initiated with the addition of an ATP-regenerating system (final reaction concentrations, 1 mM ATP, 8 mM creatine phosphate, 30 units/ml creatine phosphokinase, and 5 mM MgCl 2 ) and then incubated with rotation at 37°C for the indicated period of time. Assays were quenched by the addition of an equal volume of buffer A (50 mM Hepes, pH 7.4, 150 mM NaCl, 1 mM sodium vanadate, 100 mM NaF, 10 mM sodium pyrophosphate, and protease inhibitors) containing 2% SDS.
Immunoblot Analysis-Protein samples were subjected to SDS-PAGE and transferred to nitrocellulose. Akt phosphospecific and PKC␣ (2056) antibodies were obtained from Cell Signaling Technology. DNA-PK cs (H-163) and ATM (5C2) antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Two mTOR antibodies were used for immunoblot analysis, a polyclonal antibody mTOR (mTab2) from Upstate Biotechnology and a monoclonal antibody RAFT1 (611132) from BD Biosciences. RAPTOR antibodies were either from Cell Signaling Technology (4978) or from Abcam (antibody 5454). RICTOR antibodies were a kind gift from Dr. David M. Sabatini (Whitehead Institute) and were also obtained from Bethyl Laboratories. Immunoblots visualized by ECL detection (Amersham Biosciences) were quantified using NIH Image 1.62f. 6 and 12 g of control proteins were run on each gel to establish the linearity of the autoradiogram.
Immunoprecipitation of LDM-500 g of LDM in 500 l of IC buffer (with 1 mM dithiothreitol) were precleared with 40 l of protein A-agarose for 30 min. After centrifugation for 10 min at 4°C in a microcentrifuge, the supernatant was rotated overnight with 4 g of primary antibody. Samples were then centrifuged for 5 min at 4°C in a microcentrifuge to remove nonspecific aggregates that formed overnight. 30 l of protein A-agarose were added to the supernatants for 2 h at 4°C. Pellets were washed six times with IC buffer, resuspended in IC buffer, and then assayed in vitro for HMD-PK activity. ILK (3862) polyclonal antibodies (Cell Signaling Technology), mTOR (mTab1) polyclonal antibodies (Upstate Biotechnology), and RICTOR antibodies (D. M. Sabatini) were used in the immunoprecipitation studies.
HMD-PK Peptide Assay-HMD-PK activity was assayed using a modified version of Hill et al. (28). 50-l reactions were incubated for 20 min at 30°C in a buffer containing IC buffer, 1 mM dithiothreitol, 10 mM MgCl 2 , 1 M protein kinase A inhibitor peptide, 100 M ATP, 1 Ci of [␥-32 P]ATP, 0.5 mg/ml substrate (RRPHFPQFSYSASSTA), or control (RRPHFPQFAYSASSTA) peptide. Peptides were phosphorylated using 25 units DNA-PK (Promega), 0.75 mg/ml LDM, or 0.4 mg/ml Ext-HiP. 10 g/ml salmon sperm DNA was added in some cases. Reactions were quenched by adding 5 l of 100% (w/v) trichloroacetic acid and then centrifuged for 10 min in a microcentrifuge at room temperature. 35 l of the supernatant were spotted on P81 paper (Whatman), washed extensively in 0.1% phosphoric acid, and then analyzed by scintillation counting. Specific counts were calculated from the difference between the Ser-473 peptide and the control peptide.
siRNA Electroporation of 3T3-L1 Adipocytes-3T3-L1 adipocytes were electroporated according to the procedure of Jiang et al. (6). Five days after differentiation, 3T3-L1 adipocytes were detached from cell culture plates with 1 mg/ml collagenase and 0.05% trypsin, 0.02% EDTA. Cells were washed three times with ice-cold PBS and resuspended in PBS prior to electroporation. Approximately 5 million cells (half of the cells on a 150-mm dish) were resuspended in 0.5 ml of PBS, mixed with 20 nmol of siRNA duplexes, and electroporated at the setting of 0.18 kV and 950 microfarads using the Bio-Rad Gene Pulser Xcell electroporation system. Cells were immediately incubated with fresh media for 10 min before reseeding. Experiments were conducted 72 h after electroporation.

Subcellular Localizations of DNA-PK and mTOR⅐RICTOR Correlate
with HMD-PK Activity-ILK, PKC␣, DNA-PK, ATM, and mTOR have all been proposed to play a major role in Akt Ser-473 phosphorylation based on experiments carried out on a variety of different cell types and stimuli (24). We have decided to directly compare the functions of these five proteins in terms of insulin-stimulated Akt Ser-473 phosphorylation in a single cell type, 3T3-L1 adipocytes. Previously, we have shown that LDM and Ext-HiP, purified from 3T3-L1 adipocytes, are greatly enriched in insulin-stimulated HMD-PK activity (27). To determine whether any of these proteins are enriched in LDM and Ext-HiP, we carried out immunoblot analyses on subcellular fractions prepared from serum-starved 3T3-L1 adipocytes ( Fig. 1). DNA-PK was found in the NUC fraction as expected due to its well documented nuclear functions (30). It was, however, more greatly enriched in LDM and Ext-HiP but absent (or below detection) in high density microsomes (HDM), mitochondrial, and CYT fractions. HDM is enriched in endoplasmic reticulum, and LDM contains Golgi markers (42). ATM, another member of the PI 3-kinase-related kinase family, had the same distribution as DNA-PK but was more abundant in NUC than LDM and Ext-HiP as compared with DNA-PK. ILK had a ubiquitous distribution but was slightly more enriched in NUC, LDM, and Ext-HiP fractions. PKC␣ was found almost exclusively in CYT in the basal state. PKCs are cytosolic proteins that translocate to membranes upon hormone stimulation (43). mTOR was greatly enriched in Ext-HiP but was also found in NUC, HDM, and LDM fractions. mTOR has been previously reported to be associated with the endoplasmic reticulum (44) and Golgi (44) and in the nucleus (45). RICTOR, the binding protein of mTOR that is necessary for Ser-473 phosphorylation (31), was very abundant in LDM and less abundant in Ext-HiP but was also present in NUC and HDM fractions.
Next, we wanted to directly correlate HMD-PK activity with the amount of each candidate protein using the cell-free assay. We have previously shown that PIP3-dependent Akt Ser-473 phosphorylation in  In vitro reactions containing 3 mg/ml CYT, 1 mg/ml NUC, 1 mg/ml PM, or 0.37 mg/ml LDM were premixed on ice with 16 nM Akt2 and 10 M PIP3 (sonicated in a mixture of 100 M phosphatidylcholine and 100 M phosphatidylserine). Reactions were initiated with ATP, incubated for 10 min at 37°C, and then subjected to immunoblot analysis using the designated antibodies. our in vitro assay was entirely dependent on exogenous ATP and the time of reaction incubation (27,39). In vitro reactions were performed by combining CYT, NUC, PM, and LDM fractions with purified recombinant Akt2 and ATP in the absence and presence of PIP3. The amount of the particular fraction and the time of incubation were such that the reaction activities were in the linear range (data not shown). Assays were quenched and examined by immunoblot analyses (Fig. 2). The PIP3-stimulated Ser-473 activity in these particular reactions was highest in LDM, followed by NUC and then PM. As we previously observed, no HMD-PK activity was found in CYT (27). The Ser-473 activity correlated well with the amount of DNA-PK, mTOR, and RICTOR but did not correlate with levels of ATM, ILK, and PKC␣. Based on the fact that PKC␣ was almost exclusively localized in CYT, which had no HMD-PK activity, we concluded that PKC␣ was not the HMD-PK in 3T3-L1 adipocytes.
As previously discussed, DNA-PK, ATM, and mTOR are members of the PI 3-kinase-related family of kinases (30). If one of these proteins is indeed HMD-PK in 3T3-L1 adipocytes, then our in vitro Ser-473 phosphorylation activity should be directly sensitive to PI 3-kinase inhibitors. As shown in Fig. 3, PIP3-stimulated Ser-473 activities in both Ext-HiP and LDM were inhibited by wortmannin and LY294002.
Purified DNA-PK Does Not Phosphorylate Full-length Akt2 in a PIP3dependent Manner-Feng et al. (29) has shown that purified DNA-PK (Promega) can phosphorylate a C-terminal Akt peptide containing Ser-473 and two truncated versions of Akt (GST-AKT1-(419 -480), and DPH-Akt2 T309P ) lacking the PH domain. Only the peptide phosphorylation was DNA-dependent. Since it was never shown, we wanted to verify that purified DNA-PK could phosphorylate full-length Akt2 in a PIP3-dependent manner. In vitro reactions, conducted using various combinations of DNA-PK, Akt2, PIP3, and double-stranded DNA, were analyzed by immunoblot analysis (Fig. 4). Control reactions were also performed containing Ext-HiP and LDM. Full-length purified Akt2 was phosphorylated on Ser-473 in a PIP3-dependent manner with either Ext-HiP (Fig. 4A) or LDM (Fig. 4B). The phosphorylation was not stim- , and 10 g/ml salmon sperm DNA were mixed on ice. A, Ext-HiP reactions were initiated with ATP and incubated for 15 min at 37°C. 3ϫ-Ext-HiP refers to reactions in which three times the amount of Ext-HiP (0.6 mg/ml) was used along with Akt2 and PIP3. B, LDM reactions were initiated with ATP and incubated for 5 min at 37°C. 2ϫ-LDM refers to reactions in which twice the amount of LDM (1.5 mg/ml) was used along with Akt2 and PIP3. Samples in A and B were quenched and subjected to immunoblot analysis using Akt Ser-473 phosphospecific or DNA-PK antibodies. C, HMD-PK peptide assays were carried out as described under "Experimental Procedures." Results were quantified as the mean Ϯ S.E. of three independent experiments. The white bars represent results from basal cells, and hatched bars denote results of insulin-stimulated cells. DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 ulated by DNA. No Ser-473 phosphorylation was observed with Akt2 or Ext-HiP alone. Reactions containing LDM, Akt2, and PIP3 generated two Ser-473 bands of different molecular weights. We have found that the lower band corresponds to endogenous Akt present in LDM, and the upper band is due to the His-tagged recombinant Akt2 (data not shown). In the absence of exogenous Akt2, only the lower Ser-473 band was observed in reactions containing LDM and PIP3. Surprisingly, purified DNA-PK could not phosphorylate full-length Akt2 to any discernible degree in either the presence or absence of PIP3 and DNA. When reactions were analyzed for DNA-PK content, the amount of exogenous DNA-PK greatly exceeded the amount found in Ext-HiP or LDM. To verify that the commercial DNA-PK was active, assays were carried out with the same Ser-473 peptide used by Feng et al. (29) (Fig. 4C). As was reported previously, purified DNA-PK phosphorylated the Ser-473 peptide in a DNA-dependent manner (29). Ext-HiP and LDM could also phosphorylate the peptide; however, the presence of DNA had no statistical effect on the peptide phosphorylation. These results indicate that purified DNA-PK by itself cannot phosphorylate full-length Akt2. It is possible that DNA-PK requires an accessory protein, found in Ext-HiP or LDM, to phosphorylate full-length Akt. The holoenzyme of DNA-PK consists of a 465-kDa catalytic subunit and two Ku antigen subunits (Ku70/Ku80) (30). All three subunits, however, were found in purified DNA-PK, Ext-HiP, and LDM (data not shown). If an accessory protein is required, it is unlikely to be Ku70/80. To determine whether an accessory protein was present in Ext-HiP and LDM and was in excess relative to DNA-PK, we added purified DNA-PK to in vitro reactions containing Ext-HiP/Akt2 (Fig. 4A) or LDM/Akt2 (Fig. 4B) and compared the measured PIP3-stimulated Ser-473 activity with reactions containing Ext-HiP/Akt2 or LDM/Akt2 without exogenous DNA-PK. We found that adding exogenous DNA-PK had no effect. Control reactions containing 3 times the amount of Ext-HiP (3ϫ-Ext-HiP) or twice the amount of LDM (2ϫ-LDM) verified that the reactions were not limited in substrate. Since purified DNA-PK could not phosphorylate full-length Akt2 in a PIP3-dependent manner, either DNA-PK is not HMD-PK or an accessory protein is required but is limiting in LDM or Ext-HiP relative to the amount of DNA-PK.

Phosphorylation of Akt/Protein Kinase B Ser-473 by mTOR⅐RICTOR
HMD-PK Activity Is Not Associated with ILK in LDM-Since Glut 4 vesicles can be easily immunoisolated from LDM of 3T3-L1 adipocytes (46), we attempted to immunoprecipitate DNA-PK, ATM, or ILK from LDM and measure PIP3-stimulated Ser-473 activity in the immunoprecipitates using an in vitro assay. We tested five antibodies for both DNA-PK and ATM and failed to find one that recognized the mouse form of the proteins and that could immunoprecipitate. ILK, however, was successfully immunoisolated from LDM. In vitro assays were then conducted using supernatants and pellets from ILK or control immunoprecipitation reactions. Although we brought down ϳ50% of ILK from LDM, we observed no change in the Ser-473 activity remaining in the supernatant, nor could we measure PIP3-stimulated Ser-473 activity associated with ILK in the pellets (Fig. 5).
siRNA-mediated Depletion of DNA-PK, ATM, or ILK Had No Effect on Insulin-stimulated Akt Ser-473 Phosphorylation in 3T3-L1 Adipocytes-Next we wanted to determine whether suppressing the expression of DNA-PK, ATM, or ILK by siRNA had any effect on insulin-activated Akt Ser-473 phosphorylation in intact 3T3-L1 adipocytes. Using the electroporation method described by Czech and co-workers (6), we could substantially reduce the expressions of DNA-PK (88.7 Ϯ 6.6%), ATM (87 Ϯ 3.5%), and ILK (64 Ϯ 2.6%) (Fig. 6A). Serum-starved cells were then stimulated with insulin and tested for Akt phosphorylation using phosphospecific antibodies. We observed no change in Akt phosphorylation at either Ser-473 or Thr-308 in cells depleted of DNA-PK or ATM compared with cells tested with a negative control siRNA (Fig. 6, B and C). In ILK-depleted cells, we consistently observed a 25% increase in insulin-induced Ser-473 phosphorylation but saw no change in insulin-stimulated Thr-308 phosphorylation (Fig. 6, B and C). The observed increase in Ser-473 phosphorylation was surprising, since we had expected a decrease in insulin-stimulated Ser-473 phosphorylation if ILK was indeed the HMD-PK for Akt. Therefore, based on these results in intact adipocytes and in conjunction with our cell-free experiments, we conclude that DNA-PK, ATM, ILK, and PKC␣ are most likely not the HMD-PK in 3T3-L1 adipocytes.
PIP3-stimulated HMD-PK Activity Was Associated with mTOR⅐RICTOR Complexes from LDM-Next, we focused on mTOR⅐RICTOR by immunoisolating mTOR complexes from LDM and measuring Ser-473 activity with an in vitro assay. PIP3-stimulated Ser-473 activity was found in mTOR complexes immunoprecipitated with antibodies directed against mTOR or RICTOR (Fig. 7A). mTOR, RIC-TOR, and RAPTOR were found in the pellets using the mTOR antibody, whereas the RICTOR antibody brought down mTOR and RICTOR but not RAPTOR. We found no commercial RAPTOR antibody that could successfully immunoprecipitate RAPTOR. Nevertheless, our results indicate that complexes containing mTOR and RICTOR are capable of phosphorylating Akt on Ser-473 in a PIP3-dependent manner. The in vitro reactions were also probed for the presence of DNA-PK, ATM, and ILK. We found no detectable amounts of these three proteins in the mTOR or RICTOR immunoprecipitation pellets, although all three were present in total LDM (Fig. 7B).
To help rule out the possibility that Akt was phosphorylated on Ser-473 by a kinase other than mTOR that happens to co-immunoprecipitate in our system, we immunoprecipitated mTOR from LDM and then measured HMD-PK activity in the presence of PI 3-kinase inhibitors and staurosporine (Fig. 8). PIP3-stimulated Ser-473 activity was inhibited with wortmannin and LY294002 but not staurosporine. Our results are consistent with those of Sabatini and co-workers for mTOR (31).
siRNA-mediated Depletion of RICTOR Inhibited Insulin-stimulated Akt Ser-473 Phosphorylation in 3T3-L1 Adipocytes-Unfortunately, we were unsuccessful at silencing mTOR expression in 3T3-L1 adipocytes using six different commercially available siRNAs. Regardless, it has been reported that almost complete knockdown of mTOR is required to observe any effect on insulin-stimulated Akt Ser-473 phosphorylation (31), a result most likely due to the feedback inhibition of insulin signaling caused by serine phosphorylation of IRS-1 by mTOR/RAPTOR (36). To circumvent this problem, we attempted to suppress RICTOR expression in order to affect insulin-stimulated Akt Ser-473 phosphorylation. We were able to substantially silence by electroporated siRNA (85% Ϯ 3.5%) RICTOR expression with no effect on mTOR levels (Fig.  9A). Serum-starved adipocytes were stimulated with insulin and then analyzed for Akt phosphorylation using phosphospecific antibodies (Fig. 9, B and C). Insulin-activated Akt phosphorylation of Ser-473 and Thr-308 were suppressed 87 Ϯ 3.2 and 55 Ϯ 9.35%, respectively, when RICTOR expression was reduced with siRNA. These results were also confirmed using a pool of two independent RICTOR siRNAs (target regions 470 -490 and 903-923; Qiagen) whose sequences were differ-ent from the ones used above (results not shown). The tight correlation between RICTOR expression levels and the amount of insulin-stimulated Akt Ser-473 phosphorylation was further demonstrated by a RIC-TOR siRNA dose response experiment (Fig. 9D). Based on our cell-free and siRNA results, we conclude mTOR complexed with RICTOR is the HMD-PK in 3T3-L1 adipocytes.

DISCUSSION
DNA-PK (29), ATM, (47) ILK (48,49), PKC␣ (50), and mTOR (31) have all been recently identified as candidates for the HMD-PK for Akt. These studies, however, were all conducted in different cell types using a variety of stimuli. Our report is the first to directly address all five candidates in a single cell type, 3T3-L1 adipocytes, activated using a single stimulus, insulin. We selected 3T3-L1 adipocytes because these FIGURE 6. siRNA-mediated depletion of DNA-PK, ATM, or ILK has no effect on insulin-stimulated Akt Ser-473 phosphorylation in 3T3-L1 adipocytes. Control, DNA-PK, ATM, and ILK siRNAs were electroporated in differentiated 3T3-L1 adipocytes (day 5) as described under "Experimental Procedures." 72 h after electroporation, cells were serum-starved for 2 h, activated or not with 100 nM insulin for 1.5 min, and then harvested in Buffer A containing 1% SDS. Equal amounts of lysates were subjected to immunoblot analysis using the designated primary antibodies. A, immunoblot analysis using DNA-PK-, ATM-, and ILK-specific antibodies. B, immunoblot analysis using Akt Ser-473 and Akt-Thr-308 phosphospecific antibodies. C, quantification of the mean Ϯ S.E. of three independent experiments. White bars represent results from basal cells, and hatched bars denote results of insulin-stimulated cells. Phosphorylations were normalized to ϩinsulin controls. *, phosphorylations differed significantly (p Ͻ 0.05) compared with insulin-stimulated control samples. An unpaired Student's t test was used in all statistical analyses. DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 cells are very responsive in terms of insulin-stimulated glucose transport (51), a process in which Akt activation is a critical step (3)(4)(5)8).

Phosphorylation of Akt/Protein Kinase B Ser-473 by mTOR⅐RICTOR
The original identification of DNA-PK as the Ser-473 kinase was very surprising, since DNA-PK was already well documented to possess several DNA-related functions that take place in the nucleus (30). To carry out these functions, double-stranded DNA is required to co-localize substrates with DNA-PK through their interactions with Ku proteins and to activate DNA-PK. It is possible that our apparent enrichment of DNA-PK in LDM and Ext-HiP is entirely due to the disruption of nuclei that may occur during the homogenization process (Fig. 1). What argues against this is that Feng et al. (29) have shown by confocal microscopy that a significant fraction of DNA-PK is found in the cytosol and at the membrane in addition to the nucleus in 3T3-L1 fibroblasts and M059K cells. The strongest data that supported the idea that DNA-PK can function as a Ser-473 kinase were siRNA knockdown experiments in HEK293 cells, the lack of significant insulin-induced Ser-473 phosphorylation in DNA-PK-deficient M059J glioblastoma cells, and rescue experiments with the M059J/Fus1 cells (29). The residual amount of insulin-activated Ser-473 phosphorylation in M059J cells has been recently shown to be due to mTOR⅐RICTOR (31). Nevertheless, mTOR⅐RICTOR cannot explain why Ser-473 phosphorylation is so low in the absence of DNA-PK in M059J cells. It is possible that in certain cell types under certain conditions, DNA-PK does function as the primary HMD-PK for Akt. Recently, DNA-PK has been implicated in the activation of Akt by CpG-DNA (52). Stimulation of Ser-473 phosphorylation by CpG-DNA is markedly reduced in bone marrow-derived macrophages from mice lacking DNA-PK. Our results, however, indicated that DNA-PK could not phosphorylate full-length Akt2, although FIGURE 7. PIP3-stimulated HMD-PK activity associates with mTOR⅐RICTOR complexes from LDM. 500 g of LDM were immunoprecipitated (IP) with control (CON)-, mTOR-, or RICTOR-specific polyclonal antibodies as described under "Experimental Procedures." In vitro reactions containing pellets from the immunoprecipitates, Akt2, and with or without 10 M PIP3 (sonicated in a mixture of 100 M phosphatidylcholine and 100 M phosphatidylserine) were mixed with ATP and incubated for 10 min at 37°C. A, reactions were quenched and subjected to immunoblot analysis using the designated antibodies. B, reactions from A were also subjected to immunoblot analysis using DNA-PK-, ATM-, and ILK-specific antibodies. Total LDM provides a positive control for the DNA-PK, ATM, and ILK antibodies. it could phosphorylate truncated versions of Akt (29) (Fig. 4). Therefore, if DNA-PK can really phosphorylate Ser-473 in cells, it most likely would require an accessory factor. The necessity of an accessory factor for DNA-PK to phosphorylate Akt would not be unexpected, since its related family member mTOR must form a complex with either RAP-TOR or RICTOR to phosphorylate p70S6 kinase (35) or Akt (31), respectively. In the case of DNA-PK, this accessory factor would be necessary to shift its phosphorylation preference from (S/T)Q motifs (53) to the FXXF(S/T)(Y/F) found in Akt (20). In fact, it was recently shown that purified DNA-PK could phosphorylate Akt2 in the presence FIGURE 9. siRNA-mediated depletion of RICTOR inhibits insulin-stimulated Akt Ser-473 phosphorylation in 3T3-L1 adipocytes. Control (20 nmol) and RICTOR (50 nmol) siRNAs were electroporated in differentiated 3T3-L1 adipocytes (day 5) as described under "Experimental Procedures." 72 h after electroporation, cells were serum-starved for 2 h, activated or not with 100 nM insulin for 1.5 min, and then harvested in Buffer A containing 1% SDS. Equal amounts of lysates were subjected to immunoblot analysis using the designated primary antibodies. A, immunoblot analysis using RICTOR-, mTOR-, and Akt-specific antibodies. B, immunoblot analysis using Akt Ser-473 and Akt-Thr-308 phosphospecific antibodies. C, results in panel B were quantified as the mean Ϯ S.E. of three independent experiments. White bars represent results from basal cells, and hatched bars denote results of insulin-stimulated cells. Phosphorylations were normalized to ϩinsulin controls. *, phosphorylations differed significantly (p Ͻ 0.05) compared with insulin-stimulated control samples. An unpaired Student's t test was used in all statistical analyses. D, RICTOR siRNA dose response upon insulin-stimulated Akt Ser-473 phosphorylation. The amounts (in nmoles) of control and RICTOR siRNAs are indicated in parentheses. Immunoblot analyses using RICTOR antibodies and Akt Ser-473 phosphospecific antibodies were quantified as the mean Ϯ S.E. of three independent experiments. of a synthetic CpG oligonucleotide (CpG-OGN) but not in its absence (52). It is clear, however, from the lack of an effect on insulin-stimulated Akt Ser-473 phosphorylation when DNA-PK was knocked down by almost 90% with siRNA in 3T3-L1 adipocytes, that DNA-PK does not significantly contribute to HMD-PK activity in fat cells in response to insulin (Fig. 6). Our results, however, do not address the function of DNA-PK in other cell types.
ATM, like DNA-PK, is a kinase involved in DNA repair (54) that has also been implicated in Akt Ser-473 phosphorylation (47). It has been reported that overexpression of ATM robustly increased Ser-473 phosphorylation, whereas knockdown by siRNA significantly decreased it. Thr-308 phosphorylation was not affected in either case. ATM, however, cannot directly phosphorylate Akt on Ser-473 in vitro, suggesting that it may play an indirect role in Ser-473 phosphorylation. Our results in 3T3-L1 adipocytes indicated that ATM most likely has little if any role in Ser-473 phosphorylation, direct or indirect. ATM was not very abundant in LDM or Ext-HiP, the two fractions that possess the highest Ser-473 activity ( Figs. 1 and 2). In addition, a greater than 85% reduction in the expression of ATM by siRNA had no effect on the phosphorylation of Ser-473 in response to insulin (Fig. 6).
Dedhar and co-workers (48,49,55) have published multiple reports providing evidence that ILK can function as the Akt Ser-473 kinase. Recombinant ILK can phosphorylate Akt at Ser-473 in vitro (55). A kinase-dead ILK has been shown to act as a dominant negative suppressing endogenous Ser-473 phosphorylation activity (49). Knockdown experiments by siRNA and knockouts using the Cre-LoxP system markedly reduced Ser-473 phosphorylation (48). Despite these data, there are other studies that indicate ILK may not be the physiological Akt Ser-473 kinase. It was reported that in ILK-deficient fibroblasts (56) and chondrocytes (57), Ser-473 phosphorylation was normal. Other groups have not been able to phosphorylate Akt at the HM site in vitro (28,58,59). There is even uncertainty that ILK is a functional kinase, since several critical residues normally found in the catalytic domain of protein kinases are not conserved (59). Thus, what role ILK has in terms of the phosphorylation of Akt on Ser-473 may be indirect. In 3T3-L1 adipocytes, our data indicated that ILK was not the HMD-PK for Akt. We saw no change in the level of PIP3-stimulated Akt Ser-473 phospho-rylation activity in LDM after removing more than half the ILK and saw no corresponding appearance of activity in the pellets after immunoprecipitation (Fig. 5). Similarly, after reducing the expression of ILK by ϳ65% with siRNA, we actually observed a 25% increase rather than a decrease in insulin-activated Akt Ser-473 phosphorylation (Fig. 6). Further study is necessary to determine whether the 25% increase in Ser-473 phosphorylation is of physiological significance.
PKC␣ is activated by diacylglycerol and calcium (43) and has been implicated in Ser-473 phosphorylation (50). This PKC isoform can phosphorylate Akt on Ser-473 in vitro in the presence of lipid and calcium as well as undergo IGF-1-stimulated translocation to membranes in endothelial cells. Overexpression of PKC␣ increased Ser-473 phosphorylation, whereas dominant-negative PKC␣ constructs or reducing the expression by siRNA decreased Ser-473 phosphorylation. We found that in 3T3-L1 adipocytes, PKC␣ was almost entirely cytosolic and absent from compartments with Ser-473 phosphorylation activity (Figs. 1 and 2). In addition, our cell-free assays were all conducted in the absence of calcium, since 1 mM EGTA was included in the reaction buffer. Our conclusion that PKC␣ is not involved in the Ser-473 phosphorylation of Akt in adipocytes is in agreement with an earlier report that Akt phosphorylation at Ser-473 actually increased when PKC␣ was knocked out in fat and muscle (60). The authors of the knock-out study concluded that PKC␣ inhibits insulin signaling by phosphorylating IRS proteins on serine residues resulting in the suppression of insulin-stimulated tyrosine phosphorylation of IRS proteins by the insulin receptor. Interesting with regard to PKC␣, Sarbassov et al. (38) reported that mTOR⅐RICTOR complex can regulate the organization of the actin cytoskeleton and they have proposed that this regulation is mediated via the phosphorylation of PKC␣. PKC␣ does not, however, form a stable complex with mTOR⅐RICTOR in detergent-solubilized cells (31). In addition, PKC␣ immunoprecipitates prepared from detergent-solubilized cells did not phosphorylate Akt Ser-473 in an in vitro kinase assay.
Based on our results, we have concluded that mTOR⅐RICTOR is the HMD-PK for Akt in 3T3-L1 adipocytes. mTOR⅐RICTOR was greatly enriched in fractions that contain the highest Ser-473 activity (Figs. 1  and 2). mTOR⅐RICTOR complexes, purified by immunoprecipitation of LDM with mTOR or RICTOR antibodies, had PIP3-stimulated Ser-473

Phosphorylation of Akt/Protein Kinase B Ser-473 by mTOR⅐RICTOR
phosphorylation activity (Fig. 7A) and did not contain DNA-PK, ATM, or ILK (Fig. 7B). In addition, the Ser-473 activity associated with mTOR⅐RICTOR complexes was sensitive to PI 3-kinase inhibitors but not staurosporine (Fig. 8), the same drug sensitivity has been shown for insulin-stimulated Akt Ser-473 phosphorylation (16,61). Decreasing the expression of RICTOR by siRNA substantially inhibited insulinstimulated Akt Ser-473 and to a lesser extent Thr-308 phosphorylation in intact adipocytes (Fig. 9). Effects on both Ser-473 and Thr-308 phosphorylations are in agreement with the model that the phosphorylation of Ser-473 provides a docking site to recruit phosphoinositide-dependent kinase 1 in order to phosphorylate Thr-308 (20). It should also be pointed out that the basal activity we observed in the Ser-473 peptide assays conducted with Ext-HiP or LDM was most likely due to mTOR⅐RICTOR complexes in these fractions and not due to DNA-PK (Fig. 4C). All of our results are entirely consistent with those published by Sabatini and co-workers (31). It has also been recently reported that treating 3T3-L1 adipocytes with rapamycin actually increases insulinstimulated Akt Ser-473 phosphorylation (36). This result can be fully explained in terms of feedback inhibition of insulin signaling caused by serine phosphorylation of IRS-1. Rapamycin inhibited insulin activation of p70S6 kinase, decreased Ser 636/639 IRS-1 phosphorylation, increased PIP3 levels, and resulted in an increase in Akt Ser-473 phosphorylation (36).
The model proposed by Sabatini and co-workers (31) also explains how mTOR can be found both upstream and downstream of Akt in the insulin-stimulated PI 3-kinase pathway (Fig. 10). mTOR, when complexed to RICTOR and G␤L acts as HMD-PK activating Akt through the phosphorylation of Ser-473. Akt also activates mTOR/RAPTOR, either directly or through the phosphorylation of the tumor suppressor, TSC2, leading to the stimulation of the p70S6 kinase/4EBP1 pathway (62).
Interestingly, Sabatini and co-workers (31) have also reported that the Ser-473 kinase activity of mTOR⅐RICTOR was markedly stimulated by serum in vivo. Kinase assays conducted on immunoprecipitates isolated from 0.3% CHAPS-solubilized serum-treated HeLa cells were significantly higher than those isolated from basal cells. Activation of HMD-PK by serum would disagree with the assumption by Hemminngs that the Ser-473 kinase is constitutively active at the plasma membrane (15,63,64). It is thought that binding of Akt to PIP3, formed in PM in response to insulin, has two effects, it co-localizes Akt with a constitutively active HMD-PK (15,63) and also induces a conformational change in Akt that allows for Ser-473 and Thr-308 phosphorylation (65) (Fig. 10). Therefore, insulin enhances Ser-473 phosphorylation through effects on Akt itself as opposed to effects on HMD-PK. We, in fact, have observed no difference in HMD-PK activity in cell-free assays conducted using Ext-HiP (or LDM) purified from basal and insulin-treated adipocytes, 3 which tend to support the idea that insulin does not activate HMD-PK directly. One explanation for the apparent discrepancy is that the observed serum-induced increase in HMD-PK activity reported by Sabatini and co-workers (31) is actually due to the presence of PIP3 in the immunoprecipitates when mTOR complexes were isolated from serum-treated cells. 0.3% CHAPS is not a high enough detergent concentration to completely solubilize membranes. In fact, these same authors have reported that the mTOR⅐RICTOR complex was not stable under conditions (1% Triton X-100) typically used to solubilize membranes.
In summary, our results indicate that mTOR⅐RICTOR, is responsible for insulin-induced Akt Ser-473 phosphorylation in 3T3-L1 adipocytes.
Our results do not address whether DNA-PK, ATM, ILK, or PKC␣ can function as HMD-PK for Akt in other cell types under certain conditions.