Inhibition of Insulin-induced Activation of Akt by a Kinase-deficient Mutant of the ε Isozyme of Protein Kinase C*

Akt, also known as protein kinase B, is a protein-serine/threonine kinase that is activated by growth factors in a phosphoinositide (PI) 3-kinase-dependent manner. Although Akt mediates a variety of biological activities, the mechanisms by which its activity is regulated remain unclear. The potential role of the ε isozyme of protein kinase C (PKC) in the activation of Akt induced by insulin has now been examined. Expression of a kinase-deficient mutant of PKCε (εKD), but not that of wild-type PKCε or of kinase-deficient mutants of PKCα or PKCλ, with the use of adenovirus-mediated gene transfer inhibited the phosphorylation and activation of Akt induced by insulin in Chinese hamster ovary cells or L6 myotubes. Whereas the εKD mutant did not affect insulin stimulation of PI 3-kinase activity, the phosphorylation and activation of Akt induced by a constitutively active mutant of PI 3-kinase were inhibited by εKD, suggesting that εKD affects insulin signaling downstream of PI 3-kinase. PDK1 (3′-phosphoinositide-dependent kinase 1) is thought to participate in Akt activation. Overexpression of PDK1 with the use of an adenovirus vector induced the phosphorylation and activation of Akt; εKD inhibited, whereas wild-type PKCε had no effect on, these actions of PDK1. These results suggest that εKD inhibits the insulin-induced phosphorylation and activation of Akt by interfering with the ability of PDK1 to phosphorylate Akt.

Akt, also known as protein kinase B, is activated by growth factors such as platelet-derived growth factor (PDGF) 1 and insulin (1)(2)(3), by cytokines (4), by ligands of G protein-coupled receptors (5), and by cellular stresses such as hyperosmolarity, heat shock, fluid shear stress, and hydrogen peroxide-induced oxidative stress (6 -8). The activation of Akt by these stimuli contributes to a variety of their biological effects, including promotion of cell survival and protection from apoptosis (4,9), induction of meiosis in oocytes (10), regulation of vascular contractility (8,11), activation of the transcription factor NF-B (12,13), and various metabolic actions of insulin (14 -17).
Activation of Akt is mediated by phosphorylation of a threonine residue in the kinase activation loop (Thr 308 in Akt1) and a serine residue in the COOH-terminal region (Ser 473 in Akt1) (18). Akt mutants in which these residues are replaced by neutral amino acids are not activated in cells (15,18). The phosphorylation and activation of Akt induced by growth factors are blocked by pharmacological or molecular biological inhibitors of phosphoinositide (PI) 3-kinase (1)(2)(3)15), indicating that PI 3-kinase is required for the phosphorylation of Thr 308 and Ser 473 and activation of Akt in response to growth factors.
PDK1 (3Ј-phosphoinositide-dependent kinase 1), originally identified as a kinase that selectively phosphorylates Thr 308 of Akt in vitro, is thought to contribute to Akt activation (19 -21). However, the mechanism by which PDK1 phosphorylates Akt in intact cells remains unclear. Although the phosphorylation of Akt by PDK1 is stimulated in the presence of 3Ј-phosphoinositides in vitro (19 -21), the kinase activity of PDK1 immunoprecipitated from cells is not affected by prior treatment of the cells with either growth factors or pharmacological inhibitors of PI 3-kinase (22), suggesting that PDK1 is constitutively active in cells. Given that Akt translocates from the cytosol to the membrane fraction of cells in response to growth factors (23), and that membrane-targeted mutants of Akt are constitutively active in quiescent cells (23,24), it is thought that membrane-associated Akt is phosphorylated by PDK1.
Mutational analysis has revealed that phosphorylation of both Thr 308 and Ser 473 is required for Akt activation (18). Although a kinase that phosphorylates Ser 473 of Akt has been tentatively designated PDK2, its nature remains unclear. Balendran et al. (25) recently showed that PDK1 phosphorylates Ser 473 of Akt in vitro only in the presence of a glutathione S-transferase (GST) fusion protein that contains the COOHterminal portion of protein kinase C (PKC)-related kinase 2 (PRK2) or of synthetic peptides corresponding to this region of PRK2. In contrast, the phosphorylation of p70 S6 kinase by PDK1 was inhibited in the presence of the same GST fusion protein or peptides (26). Moreover, when overexpressed in intact cells, this same region of PRK2 prevented ligand-induced activation of p70 S6 kinase (26). These observations suggest that the interactions of PDK1 with its substrates, at least those with Akt and p70 S6 kinase, are regulated by another kinase.
PKC⑀ is a member of the novel subfamily of PKC isozymes (27,28). When expressed in fibroblasts also expressing various mutant PDGF receptors, PKC⑀ contributed to the transactivation of the TPA (12-O-tetradecanoylphorbol 13-acetate)-responsive element induced by PDGF in a PI 3-kinase-dependent manner (29), suggesting that PKC⑀ participates in signaling downstream of PI 3-kinase. We have therefore now investigated whether PKC⑀ plays a role in activation of Akt. We examined the effects of overexpression of wild-type or a kinasedeficient mutant of PKC⑀ on Akt activation induced by insulin, heat shock, or hydrogen peroxide. Expression of the kinasedeficient mutant of PKC⑀ inhibited the phosphorylation and activation of Akt in response to all three stimuli, probably by affecting the ability of PDK1 to phosphorylate Akt.

EXPERIMENTAL PROCEDURES
Cells and Antibodies-L6 myoblasts were maintained and induced to differentiate into myotubes as described previously (30). We amplified a full-length mouse serum-and glucocorticoid-regulated protein kinase (SGK) 2 cDNA (31) by the polymerase chain reaction (PCR) with cDNA synthesized from RNA extracted from mouse liver as template. The amplified SGK2 cDNA was tagged with the HA epitope at its NH 2 terminus with the use of PCR. To establish CHO-IR cells that express (in addition to insulin receptors) FLAG epitope-tagged Akt1 or HA epitope-tagged SGK2 (CHO-IR/Akt cells and CHO-IR/SGK2 cells, respectively), we transfected CHO-IR cells with pSV40-hgh (which confers resistance to hygromycin) and a PECE vector encoding FLAG-tagged rat Akt1 (RAC-PK␣) (6) or a pCMV4 vector encoding HA-tagged mouse SGK2. Transfected cells were selected and cloned as described previously (32). CHO-IR cells that express phosphodiesterase 3B (PDE3B), designated CHO-IR/ PDE3B-WT cells, were previously described (16). Polyclonal antibodies to Akt, to mitogen-activated protein (MAP) kinase, and to PDE3B were generated as described previously (15,16,33). Polyclonal antibodies to PKC⑀ as well as monoclonal antibodies to PKC␣ and to PKC were obtained from Santa Cruz Biotechnology and Transduction Laboratories, respectively. Monoclonal antibodies to the HA epitope tag (12CA5) or to the FLAG epitope tag were obtained from Roche Molecular Biochemicals. Polyclonal antibodies specific for phospho-Thr 308 or phospho-Ser 473 forms of Akt or for the phosphorylated form of MAP kinase were obtained from New England BioLabs.
Construction of and Infection with Adenovirus Vectors-Adenovirus vectors encoding a kinase-deficient mutant of PKC in which Lys 273 is replaced by glutamate (AxCAKD) (32), or a constitutively active mutant of PI 3-kinase (AxCAMyr-p110) (16,34) were described previously. Complementary DNAs for wild-type PKC⑀ (⑀WT, Ref. 27) and a kinasedeficient mutant of PKC⑀ in which Lys 437 is replaced by methionine (⑀KD, Ref. 35) were modified with the use of PCR to encode the T7 or FLAG epitope tags at the NH 2 termini of the respective proteins. A kinase-deficient mutant of PKC⑀ in which Thr 566 is replaced by alanine (⑀T566A) was as described (36). A constitutively active mutant of PKC⑀ in which Ala 159 is replaced by glutamate (⑀A159E) was constructed from the cDNA that encodes FLAG epitope-tagged ⑀WT with the use of a QuickChange site-directed mutagenesis kit (Stratagene). Complementary DNAs encoding T7-tagged ⑀WT, the FLAG-tagged ⑀KD, ⑀T566A, FLAG-tagged ⑀A159E, Myc epitope-tagged PDK1 (21), kindly provided by L. Stephens (The Babraham Institute), or a kinase-deficient mutant of PKC␣ in which Lys 368 is replaced with arginine (37) were subcloned into pAxCAwt (38), and adenoviral vectors containing these cDNAs were generated with the use of an adenovirus expression kit (Takara, Tokyo, Japan) as described previously (15,16). The resultant adenovirus vectors were termed AxCA⑀WT, AxCA⑀KD, AxCA⑀T566A, AxCA⑀A159E, AxCAPDK1, and AxCA␣KD, respectively. CHO-IR/Akt cells or differentiated L6 myotubes were infected with adenovirus vectors at the indicated multiplicity of infection (MOI), expressed in plaque-forming units (PFU) per cell, as described previously (33). The cells were subjected to experiments 24 -48 h after infection.
Kinase Assays-L6 myotubes or CHO-IR/Akt cells were deprived of serum for 16 -20 h, incubated in the absence or presence of 100 nM insulin for the indicated time, and then immediately frozen with liquid nitrogen. The assay for MAP kinase activity was performed with immunoprecipitates prepared with antibodies to MAP kinase, as described previously (33). For assay of PI 3-kinase activity, cells were lysed and subjected to immunoprecipitation with antibodies to phosphotyrosine (PY20; Transduction Laboratories); the resulting immunoprecipitates were washed and PI 3-kinase activity in the washed precipitates was assayed as described previously (33). For assay of Akt activity, cells were lysed and subjected to immunoprecipitation with polyclonal antibodies to Akt as described (15). The immunoprecipitates were then mixed with 30 l of kinase reaction mixture containing 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM of the specific peptide inhibitor of cAMP-dependent protein kinase (PKI), 5 M nonradioactive ATP, 2 Ci of [␥-32 P]ATP (4000 Ci/mmol), and 5 M Crosstide peptide (GRPRTSSFAEG) (39) and then incubated for 30 min at 30°C. Assay of SGK activity was performed essentially as described (40) with the following modifications. Cells were lysed in a solution containing 50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 1% Triton X-100, 1 mM benzamidine, 25 mM NaF, 40 mM ␤-glycerophosphate, 1 mM phenylmethylsulfonyl fluoride, and leupeptin (10 g/ml). The lysate was centrifuged, and the resulting supernatant was subjected to immunoprecipitation with antibodies to HA. The immunoprecipitates were washed once with the lysis buffer containing 500 mM NaCl and with the lysis buffer and then twice with 50 mM Tris-HCl, pH 7.5 containing 0.1% (v/v) of ␤-mercaptoethanol. The immunoprecipitates were then mixed with 30 l of kinase reaction mixture containing 60 mM Tris-HCl, pH 7.5, 12 mM MgCl 2 , 0.12 mM EDTA, 0.12% (v/v) ␤-mercaptoethanol, 3 g/ml of PKI, 5 M nonradioactive ATP, 4 Ci of [␥-32 P]ATP (4000 Ci/mmol), and 36 M Crosstide peptide, and then incubated for 60 min at 30°C. The kinase recation mixture for Akt or SGK was spotted onto a P81 phosphocellulose filter (Whatman), the filters were washed three times with 0.5% (w/v) orthophosphoric acid, and the radioactivity remaining on the filters was measured.
In Vivo Phosphorylation of PDE3B-The in vivo phosphorylation of PDE3B was assayed as described previously (16). In brief, CHO-IR/ PDE3B-WT cells, previously infected (or not) with AxCA⑀KD, were labeled with [ 32 P]orthophosphate, incubated in the absence or presence of 100 nM insulin for 15 min, and lysed. Cell lysates were subjected to immunoprecipitation with polyclonal antibodies to PDE3B, the resulting precipitates were separated by SDS-polyacrylamide gel electrophoresis on a 7% gel, and the incorporation of radioactivity into PDE3B was visualized and quantitated with a Fuji BAS2000 image analyzer.

Effects of Wild-type and a Kinase-deficient Mutant of PKC⑀
on Insulin-induced Activation of Akt-Infection of CHO-IR/Akt cells, which stably express both human insulin receptors and rat Akt1, with an adenovirus vector that encodes a kinasedeficient mutant of PKC⑀ (AxCA⑀KD) resulted in a dose-dependent increase in the amount of PKC⑀ protein as assessed by immunoblot analysis; the amount of PKC⑀ protein in cells infected at an MOI of 10 PFU/cell was ϳ10 -20 times that of endogenous PKC⑀ (Fig. 1A). Exposure of noninfected CHO-IR/ Akt cells to insulin resulted in a Ͼ30-fold increase in the amount of Akt activity measured in immunoprecipitates prepared with antibodies to Akt (Fig. 1A). Expression of the kinase-deficient mutant of PKC⑀ (⑀KD) in the cells inhibited insulin-induced activation of Akt in a dose-dependent manner. Infection of CHO-IR/Akt cells with AxCA⑀KD also inhibited insulin-induced phosphorylation of both Thr 308 and Ser 473 of Akt, as assessed by immunoblot analysis with antibodies specific for either phospho-Thr 308 or phospho-Ser 473 forms of Akt (Fig. 1A). Expression of a structurally distinct kinase deficient mutant of PKC⑀ (⑀T566A) also exerted similar inhibitory effects on insulin-induced activity (Fig. 1B) and phosphorylation (data not shown) of Akt in CHO-IR/Akt cells.
We next investigated whether ⑀KD exerted a similar inhibitory effect on Akt activation in physiological target cells of insulin. Insulin induced an approximately 6-fold increase in Akt activity in L6 myotubes (Fig. 1C). Expression of ⑀KD inhibited insulin-induced phosphorylation and activation of Akt in a dose-dependent manner (Fig. 1C). In contrast, overexpression of wild-type PKC⑀ in L6 myotubes did not affect insulininduced phosphorylation and activation of Akt (Fig. 1D). Expression of wild-type PKC⑀ also had no effect on the activation of Akt in response to insulin in CHO-IR/Akt cells, and expres- The cell lysates were either subjected to immunoprecipitation with antibodies to Akt and thereby assayed for Akt activity (upper panels), or subjected to immunoblot analysis with antibodies to PKC⑀, to the phospho-Thr 308 (pT308), or the phospho-Ser 473 (pS473) forms of Akt, or to total Akt (lower panels). Quantitative data are means Ϯ S.E. from three experiments. Immunoblot data are representative of three independent experiments. sion of either wild-type PKC⑀ or a constitutively active mutant of PKC⑀ (⑀A159E), of which kinase activity was ϳ10-fold higher than that of wild-type PKC⑀ (data not shown), inconsistent with a previous report (41), did not induce Akt activation in the absence of insulin in CHO/IR Akt cells or in L6 myotubes (data not shown). These results suggest that ⑀KD inhibits insulininduced activation of Akt by preventing the phosphorylation of Thr 308 and Ser 473 , and that signal mediated through PKC⑀ alone is not sufficient to activate Akt in both CHO-IR/Akt cells and L6 myotubes.
Effects of Kinase-deficient Mutants of Various PKC Isozymes on Insulin-induced Activation of Akt-We investigated whether kinase-deficient mutants of other PKC isozymes also affected the activation of Akt by insulin. Infection of L6 myotubes with adenovirus vectors encoding ⑀KD, wild-type PKC⑀, or kinasedeficient mutants of PKC␣ (␣KD) or PKC (KD) resulted in marked expression of the encoded proteins ( Fig. 2A); at an MOI of 20 PFU/cell; the amount of each recombinant protein was at least 10 times that of the corresponding endogenous PKC isoform. Whereas expression of ⑀KD inhibited insulin-induced activation of Akt, ␣KD had no effect on this action of insulin ( Fig. 2A). The expression of KD resulted in a slight enhancement of the effect of insulin on Akt activity, consistent with previous observations (42).

Effects of a Kinase-deficient Mutant of PKC⑀ on Akt Activation Induced by Heat Shock or Hydrogen
Peroxide-Akt is also activated by heat shock and by hydrogen peroxide (6, 7). These stimuli induced ϳ10and 20-fold increases in Akt activity, respectively, in CHO-IR/Akt cells (Fig. 2B). Expression of ⑀KD inhibited the activation of Akt in response to either heat shock or hydrogen peroxide. The phosphorylation of Akt on Thr 308 and Ser 473 induced by these stimuli was also inhibited by

FIG. 2. Effects of kinase-deficient mutants of various PKC isozymes on insulin-induced activation of Akt (A) and effects of ⑀KD on Akt activation induced by heat shock or hydrogen peroxide (B).
A, L6 mytotubes that had been infected (or not) with either AxCA⑀KD (⑀KD), AxCA⑀WT (⑀WT), AxCA␣KD (␣KD), or AxCAKD (KD) at an MOI of 20 PFU/cell were incubated in the absence or presence of 100 nM insulin for 10 min and then lysed. Cell lysates were either subjected to immunoprecipitation with antibodies to Akt and thereby assayed for Akt activity (upper panel), or subjected to immunoblot analysis with antibodies to PKC⑀, to PKC␣, or to PKC (lower panel). B, CHO-IR/Akt cells that had been infected (or not) with AxCA⑀KD at an MOI of 10 PFU/cell were either incubated for 10 min in the absence or presence of 8.8 mM hydrogen peroxide or subjected to heat shock for 10 min at 45°C. Cell lysates were then prepared and either subjected to immunoprecipitation with antibodies to Akt and thereby assayed for Akt activity (upper panel), or subjected to immunoblot analysis with antibodies to PKC⑀ or to phospho-Thr 308 or to phospho-Ser 473 forms of Akt. Quantitative data are means Ϯ S.E. from three experiments. Immunoblot data are representative of three independent experiments.

FIG. 3. Effect of ⑀KD on insulin-induced phosphorylation of PDE3B. CHO-IR/PDE3B-WT cells that had been infected (or not) with
AxCA⑀KD at an MOI of 3 PFU/cell were labeled with [ 32 P]orthophosphate, incubated in the absence or presence of 100 nM insulin for 15 min, and lysed. Cell lysates were subjected to immunoprecipitation with antibodies to PDE3B, the resulting precipitates were subjected to electrophoresis, and 32 P incorporation into PDE3B was visualized (lower panel) or quantitated (upper panel) with an image analyzer. Quantitative data are means of two experiments, and the image shown is representative of two independent experiments. expression of ⑀KD (Fig. 2B).
Effect of a Kinase-deficient Mutant of PKC⑀ on Insulin-induced Phosphorylation of PDE3B-PDE3B was recently identified as a direct substrate of Akt (16). In CHO-IR cells expressing PDE3B, insulin induced an approximately 2-fold increase in the extent of phosphorylation of PDE3B (Fig. 3). Infection of the cells with AxCA⑀KD at an MOI of 3 PFU/cell, a virus dose that inhibited insulin-induced activation of Akt by ϳ50% in CHO-IR/Akt cells (Fig. 1A), resulted in ϳ50% inhibition of insulin-induced phosphorylation of PDE3B. This observation suggests that inhibition by ⑀KD of Akt activation attenuates signaling downstream of Akt.

Effect of a Kinase-deficient Mutant of PKC⑀ on Signaling
Downstream of PI 3-Kinase-We further attempted to identify the step of the insulin signaling pathway leading to activation of Akt that is affected by ⑀KD. Overexpression of ⑀KD had no effect on the insulin-induced increase in the activity of PI 3-kinase immunoprecipitated from L6 myotubes with antibodies to phosphotyrosine (Fig. 4A), suggesting that ⑀KD affects insulin-induced activation of Akt at a step downstream of PI 3-kinase. To verify this conclusion, we examined the effect of ⑀KD on the activation of Akt by a constitutively active mutant of PI 3-kinase, Myr-p110, which comprises the catalytic subunit of PI 3-kinase ligated to a myristoylation signal sequence -p110  (B). A and C, L6 mytotubes that had been infected with AxCA⑀KD at the indicated MOI (PFU/cell) were incubated in the absence or presence of 100 nM insulin for 10 min and then lysed. The lysates were subjected to immunoprecipitation with antibodies to phoshotyrosine (A) or to MAP kinase (C), and the resulting precipitates were assayed for PI 3-kinase and MAP kinase activities, respectively. Alternatively, cell lysates from (C) were subjected to immunoblot analysis with antibodies to phospho-MAP kinase or to total MAP kinase (MAPK, lower panel). B, L6 mytotubes were infected (or not) with AxCAMyr-p110 at an MOI of 10 PFU/cell and, after 12 h, with AxCA⑀KD at the indicated MOI (PFU/cell). After an additional 36 h, the cells were lysed and the lysates were either subjected to immunoprecipitation with antibodies to Akt and thereby assayed for Akt activity (upper panel) or subjected to immunoblot analysis with antibodies to PKC⑀, to the phospho-Ser 473 form of Akt, or to total Akt (lower panel). Quantitative data are means Ϯ S.E. from three experiments. Immunoblot data are representative of three independent experiments. at its NH 2 terminus (16,34). Infection of L6 myotubes with an adenovirus vector encoding Myr-p110 (AxCAMyr-p110) induced both the phosphorylation and activation of Akt (Fig. 4B). Coexpression of ⑀KD inhibited in a dose-dependent manner the phosphorylation and activation of Akt induced by Myr-p110 (Fig. 4B); expression of ⑀KD did not affect the amount Myr-p110 protein as assessed by immunoblot analysis (data not shown). These observations are thus consistent with the notion that ⑀KD inhibits insulin-induced activation of Akt by affecting signaling downstream of PI 3-kinase. Expression of ⑀KD did not affect insulin-induced phosphorylation and activation of MAP kinase (Fig. 4C), indicating that early events of insulin signaling, such as activation of the insulin receptor kinase and phosphorylation of its substrates, are not prevented by ⑀KD.

FIG. 4. Effects of ⑀KD on insulin-induced activation of PI 3-kinase (A) and MAP kinase (C) as well as on Akt activation by Myr
Effect of a Kinase-deficient Mutant of PKC⑀ on PDK1-induced Activation of Akt-PDK1 is a serine/threonine kinase that phosphorylates and activates Akt in vitro (19 -21). We thus investigated the effect of ⑀KD on PDK1-induced activation of Akt. Infection of CHO-IR/Akt cells with an adenovirus vector encoding PDK1 (AxCAPDK1) induced the phosphorylation of Akt on both Thr 308 and Ser 473 as well as the activation of this enzyme (Fig. 5). Coexpression of ⑀KD with PDK1 resulted in inhibition of the PDK1-induced phosphorylation and activation of Akt, with no effect on the amount of PDK1 protein. In contrast, expression of wild-type PKC⑀ had no effect on the phosphorylation and activation of Akt induced by PDK1. These results suggest that ⑀KD inhibited the ability of PDK1 to phosphorylate and activate Akt.
Effect of a Kinase-deficient Mutant of PKC⑀ on Insulin-induced Activation of SGK-PDK1 has recently been shown to contribute to the phosphorylation and the activation of SGK (31,40,43). We finally investigated the effect of ⑀KD on insulininduced activation of SGK. CHO-IR/SGK2 cells, which stably express both human insulin receptors and HA-tagged mouse SGK2, were incubated in the absence or presence of insulin for 10 min, lysed, and immunoprecipitated with antibodies to HA, and then SGK kinase activity toward Crosstide was assayed in the immunoprecipitates. Insulin induced ϳ6-fold increase in the activity of SGK, and overexpression of ⑀KD in the cells inhibited insulin-induced activation of SGK in a dose-dependent manner (Fig. 6). DISCUSSION We have shown that overexpression of kinase-deficient mutants of PKC⑀ (⑀KD or ⑀T566A) with the use of adenovirusmediated gene transfer inhibited the phosphorylation and activation of Akt induced by insulin. An adenovirus vector encoding wild-type PKC⑀ had no effect on the insulin-induced increase in Akt activity, and the virus encoding ⑀KD had no effect on insulin-induced activation of either MAP kinase or PI 3-kinase, suggesting that the inhibition of Akt activation by ⑀KD is due neither to nonspecific effects of viral infection nor to general inhibition of insulin signaling. The phosphorylation and activation of Akt induced by heat shock or hydrogen peroxide were also inhibited by ⑀KD, suggesting that this mutant protein affects a common component of the Akt activation pathways triggered by various extracellular stimuli. PI 3-kinase acts as an upstream mediator of Akt activation induced by several stimuli (1-3, 5, 7). However, our observations that ⑀KD both did not inhibit insulin-induced activation of PI 3-kinase and prevented Akt activation by a constitutively active mutant of PI 3-kinase indicate that ⑀KD affects a signaling component that acts downstream of PI 3-kinase.
PDK1 was originally identified as a kinase that phosphorylates Thr 308 of Akt in vitro (20,21), and an unidentified kinase that phosphorylates Ser 473 of Akt has been termed PDK2. We have now shown that overexpression of PDK1 induced the phosphorylation of Akt on both Thr 308 and Ser 473 in intact cells. PDK1 has been shown to phosphorylate Ser 473 of Akt in vitro in the presence of a GST fusion protein containing the COOHterminal region of PRK2 or of synthetic peptides that encompass this region of PRK2 (25). It is thus possible that, in intact cells, PDK1 phosphorylates both Thr 308 and Ser 473 of Akt, which would explain why overexpression of PDK1 alone resulted in the activation of Akt. We cannot, however, exclude the possibility that expression of PDK1 resulted in the activation of PDK2, and that PDK1 and PDK2 coordinately phosphorylate and activate Akt.
We have shown that ⑀KD inhibited PDK1-induced phosphorylation and activation of Akt, suggesting that ⑀KD affects the insulin signaling pathway at a step downstream of PDK1 action. PDK1 has been shown, at least in vitro, to phosphorylate not only Akt but various other kinases including p70 S6 kinase (44), SGK (31,40,43), p90 RSK (45), cAMP-dependent protein kinase (46), and PKC isozymes including PKC and PKC␦ (47), all of which belong to the AGC family of protein kinases. Thus, the observation that ⑀KD inhibited insulin-induced activity of SGK is also consistent with this hypothesis that ⑀KD interfere with the insulin signaling at a step downstream of PDK1. Given that PKC⑀ is a member of this family of kinases, it also might serve as a substrate for PDK1. However, it is not likely that ⑀KD inhibits insulin-induced activation of Akt simply by competing with Akt for PDK1, because expression of wild-type PKC⑀ did not inhibit this effect of insulin.
The mechanism by which ⑀KD inhibits the ability of PDK1 to activate Akt remains unclear. It is possible that endogenous PKC⑀ phosphorylates an unidentified substrate that is important for the interaction between PDK1 and Akt, and that ⑀KD exerts a dominant-negative effect on endogenous PKC⑀. Our observation that overexpression of wild-type PKC⑀ or a constitutively active form of PKC⑀ alone did not increase Akt activity in the absence of insulin may be explained if such a substrate is constitutively phosphorylated in cells, so that overexpression of the wild-type or a constitutively active enzyme would not alter the phosphorylation state of the substrate.
The COOH-terminal region of PRK2, which includes a consensus sequence for a PDK2 phosphorylation site similar to that present in Akt with the exception that the residue equivalent to Ser 473 is aspartic acid, modulates PDK1 activity (25,26). Analysis of the three-dimensional structure of PDK1 suggests the presence in the kinase domain of a hydrophobic pocket that interacts with this region of PRK2 (48). Given that PKC⑀ also contains a sequence similar to the COOH-terminal region of PRK2, it is possible that ⑀KD affects the ability of PDK1 to phosphorylate Akt by interacting with the hydrophobic pocket of PDK1. Whereas serine and threonine residues in the COOHterminal region of wild-type PKC␤II are phosphorylated in intact cells, those of a kinase-deficient mutant of the enzyme are not (49). It is thus possible that the phosphorylation status of wildtype PKC⑀ and ⑀KD differ and that the difference in the abilities of these molecules to affect Akt activation might be because of such a difference in phosphorylation status.