Substitution of the Autophosphorylation Site Thr516with a Negatively Charged Residue Confers Constitutive Activity to Mouse 3-Phosphoinositide-dependent Protein Kinase-1 in Cells*

3-Phosphoinositide-dependent protein kinase-1 (PDK-1)is a serine/threonine kinase that has been found to phosphorylate and activate several members of the AGC protein kinase family including protein kinase B (Akt), p70 S6 kinase, and protein kinase Cζ. However, the mechanism(s) by which PDK-1 is regulated remains unclear. Here we show that mouse PDK-1 (mPDK-1) undergoes autophosphorylation in vitro on both serine and threonine residues. In addition, we have identified Ser399and Thr516 as the major mPDK-1 autophosphorylation sitesin vitro. Furthermore, we have found that these two residues, as well as Ser244 in the activation loop, are phosphorylated in cells and demonstrated that Ser244 is a major in vivo phosphorylation site. Abolishment of phosphorylation at Ser244, but not at Ser399 or Thr516, led to a significant decrease of mPDK-1 autophosphorylation and kinase activity in vitro, indicating that autophosphorylation at Ser399 or Thr516 is not essential for mPDK-1 autokinase activity. However, overexpression of mPDK-1T516E, but not of mPDK-1S244E or mPDK-1S399D, in Chinese hamster ovary and HEK293 cells was sufficient to induce Akt phosphorylation at Thr308 to a level similar to that of insulin stimulation. Furthermore, this increase in phosphorylation was independent of the Pleckstrin homology domain of Akt. Taken together, our results suggest that mPDK-1 undergoes autophosphorylation at multiple sites and that this phosphorylation may be essential for PDK-1 to interact with and phosphorylate its downstream substrates in vivo.

3-Phosphoinositide-dependent protein kinase-1 (PDK-1) 1 is a recently identified protein kinase that functions downstream from PI 3-kinase and upstream of protein kinase B (Akt) in receptor tyrosine kinase signal transduction pathways (1). PDK-1 phosphorylates Thr 308 in the activation loop of Akt and contributes to the activation of the enzyme (2,3). In addition to Akt, PDK-1 also phosphorylates and activates several other downstream effectors of PI 3-kinase including p70 S6 kinase (4,5), protein kinase C isoforms (6 -9), serum and glucocorticoidinducible kinase (10 -12), and protein kinase C-related kinases 1 and 2 (13,14). These kinases play versatile roles in numerous cellular events; thus, the activation state of PDK-1 toward these substrates may play an important physiological role in the regulation of a variety of biological activities including cell proliferation, differentiation, and apoptosis.
Although the activation of several PDK-1 substrates has been characterized, the mechanism by which PDK-1 activity is regulated in cells is largely unknown. Phosphorylation of protein kinase C and protein kinase C␦ by PDK-1 in vivo has been shown to be inhibited by the PI 3-kinase inhibitor LY294002 (6), suggesting that PDK-1 activity is mediated by a PI 3-kinasedependent mechanism. Consistent with this finding, it has recently been found that insulin stimulates the activity of PDK-1 toward Akt and that this activation is blocked by wortmannin, another inhibitor of PI 3-kinase (15). However, others have found PDK-1 to be constitutively activated (4,8), and PDK-1-mediated phosphorylation of p70 S6 kinase in cells has been shown to be unaffected by wortmannin treatment (5).
Although activation of PI 3-kinase may be important for the activity of PDK-1 toward some of its downstream substrates, differential phosphorylation of PDK-1 itself may also play an important role in regulating the kinase activity of the enzyme. A recent study identified Ser 25 , Ser 241 , Ser 393 , Ser 396 , and Ser 410 on human PDK-1 (hPDK-1) as phosphorylated sites in cells (17). Of these sites, phosphorylation at Ser 241 was found to be essential for hPDK-1 activity in vitro (17). However, the functional role(s) of PDK-1 phosphorylation at the other sites is unknown.
Recently, we found that purified mPDK-1 underwent significant autophosphorylation in vitro (8,18). However, the identity of these sites and the extent to which phosphorylation of these residues played a role in PDK-1 function remained unclear. In the present study, we report the identification of two major in vitro autophosphorylation sites on mPDK-1: Thr 516 and Ser 399 . Interestingly, we found that Ser 244 in the activation loop of mPDK-1 is only marginally phosphorylated in vitro. On the other hand, we demonstrate that Ser 244 , Ser 399 , and Thr 516 of mPDK-1 are phosphorylated in cells and that Ser 244 is a major in vivo phosphorylation site. Overexpression of mPDK-1 T516E , but not mPDK-1 S244E , mPDK-1 S399D , or mPDK-1 T516A , resulted in Akt Thr 308 phosphorylation to a level similar to that of insulin stimulation. Furthermore, this increase in phosphorylation is independent of the PH domain of Akt. Taken together, our results suggest that phosphorylation of PDK-1 at multiple sites plays distinct and important roles in the regula-tion of PDK-1 function in vivo under various cellular conditions.
Site-directed Mutagenesis-All mPDK-1 mutants were generated using single-stranded site-directed mutagenesis according to the protocol described by Kunkel et al. (19) using customized primers. All constructs were verified by restriction mapping and DNA sequencing.
Cell Culture, Transfection, Immunoprecipitation, and Western Blot-Chinese hamster ovary cells overexpressing the insulin receptor (CHO/ IR), and human embryonic kidney (HEK293) cells were used in all experiments. Transfections were performed using LipofectAMINE (Invitrogen) according to the manufacturer's protocol. The cells were lysed in Buffer B. The cell lysates were centrifuged at 12,000 ϫ g for 10 min at 4°C, and the clarified supernatant was used for either immunoprecipitation or Western blot analysis. The proteins in cell lysates were immunoprecipitated by incubation with the primary antibody conjugated to protein G-Sepharose beads for 6 -18 h at 4°C. The immunoprecipitates were washed three times with ice-cold Buffer A. For immunoblot analysis, the cell lysates or immunoprecipitates were separated by SDS-PAGE using 10 or 15% polyacrylamide gels. Following electrophoresis, the proteins were transferred onto nitrocellulose membranes, and the bound proteins were detected by blotting with primary antibody, followed by horseradish peroxidase-or alkaline phosphatase-conjugated secondary antibodies (Promega, Madison, WI).
In Vitro Phosphorylation Assays-CHO/IR cells transiently expressing Myc-tagged wild type or mutant PDK-1 were washed once with phosphate-buffered saline followed by serum starvation for 1 h at 37°C. The cells were lysed in Buffer B, and the proteins were immunoprecipitated using antibody to the tag as described above. The bound protein was washed twice with ice-cold Buffer A and once with Buffer C. Phosphorylation was initiated with the addition of 30 l of Buffer C plus 2 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences) and incubated for 20 min at 30°C. The bound protein was eluted by heating at 95°C for 3 min in SDS-PAGE sample buffer. The proteins were separated by SDS-PAGE and blotted to nitrocellulose or polyvinylidene difluoride membrane. Phosphorylation of PDK-1 was visualized by autoradiography, and the bands corresponding to PDK-1 were excised and used in various assays.
mPDK-1 in Vitro Kinase Activity Assay-CHO/IR cells overexpressing wild type or mutant mPDK-1 were washed with phosphate-buffered saline and serum starved for 1 h. The cells were lysed in Buffer B, and Myc-tagged PDK-1 was immunoprecipitated using antibody to the tag. Immunoprecipitates were washed twice with ice-cold Buffer A and once with Buffer C. The activity of mPDK-1 was determined by incubation of the immunoprecipitates with 40 l of Buffer C containing 2 Ci of [␥-32 P]ATP and PDK1 substrate peptide corresponding to the sequence surrounding Thr 308 of Akt-1 (KTFCGTPEYLAPEVRR) for 30 min at 30°C. Phosphorylated peptide substrate was precipitated with trichloroacetic acid, spotted on p81 Whatman paper, and washed three times in 1% H 3 PO 4 . Peptide phosphorylation by PDK-1 was determined by scintillation counting.
In Vivo Metabolic 32 P Labeling of mPDK-1-CHO/IR cells overexpressing Myc-tagged wild type or mutant PDK-1 were washed once and incubated with warmed Krebs-Ringer bicarbonate buffer for 30 min at 37°C. The cells were labeled with 400 -800 Ci of [ 32 P]orthophosphate (PerkinElmer Life Sciences) for 4 h and then left untreated or treated with 10 nM insulin for 5 min at 37°C. The cells were lysed in Buffer B and immunoprecipitated as described above. The bound protein was washed two times with high and low salt wash buffers (20 mM Na 2 HPO 4 , pH 8.6, 0.5% Triton X-100, 0.1% SDS, 0.02% NaN 3 , and either 1 or 0.15 M NaCl) and eluted by heating at 95°C for 3 min in SDS-PAGE sample buffer. The proteins were separated by SDS-PAGE, blotted to nitrocellulose or polyvinylidene difluoride membrane, and utilized as described above.
Phosphopeptide and Phosphoamino Acid Analysis-In vitro or metabolically 32 P-labeled proteins were separated by SDS-PAGE, transferred to nitrocellulose (phosphopeptide mapping) or polyvinylidene difluoride (phosphoamino acid analysis) membranes, and excised as described above. The bound protein was blocked with 0.5% polyvinylpyrrolidone-360 (Sigma) in 100 mM acetic acid for 30 min at 37°C. The protein was washed twice with double distilled H 2 O, washed once with 50 mM ammonium bicarbonate, and subjected to trypsin digest for 18 h at 37°C. The digests were collected and lyophilized, and the samples were desalted using C 18 ZipTips (Millipore, Bedford, MA) according to a modified protocol by the manufacturer. The samples were loaded on TLC cellulose plates and subjected to electrophoresis and liquid chromatography as described previously (20). Phosphopeptides were visualized by autoradiography using x-ray film. Phosphoamino acid analysis was carried out using a protocol described previously (21).

mPDK-1 Undergoes Autophosphorylation on Both Serine and
Threonine Residues in Vitro-We (8,18) and others (22) have previously demonstrated that PDK-1 undergoes significant autophosphorylation in vitro. In an attempt to identify PDK-1 autophosphorylation sites, we carried out phosphoamino acid analysis and two-dimensional phosphopeptide mapping studies on in vitro phosphorylated wild type mPDK-1. Our results showed that mPDK-1 autophosphorylated in vitro on both serine and threonine residues at approximately a 1:1 ratio (Fig.  1A, left panel). No in vitro phosphorylation was observed for a kinase-inactive mutant of mPDK-1 (mPDK-1 K114G ) (data not shown). Phosphopeptide analysis of wild type mPDK-1 revealed three major (Fig. 1A, right panel, 1-3) and nine minor (Fig. 1A, right panel, a-i) phosphopeptides, suggesting that PDK-1 autophosphorylates multiple sites in vitro. Phosphoamino acid analysis of phosphopeptides 1-3 revealed that phosphopeptides 1 and 2 contained exclusively phosphothreonine, whereas phosphopeptide 3 contained only phosphoserine (data not shown).
We have recently shown that Ser 244 of mPDK-1 is important for PDK-1 function in cells (15). To determine whether Ser 244 is an in vitro autophosphorylation site, we examined the phosphorylation of wild type and Ser 244 mutants of mPDK-1 (Fig.  1B). We found that wild type mPDK-1 underwent significant autophosphorylation in vitro. Replacing Ser 244 of mPDK-1 with alanine resulted in a dramatic decrease in the in vitro phosphorylation of the protein. This decrease in phosphorylation was partially recovered by replacing Ser 244 with a negatively charged glutamate. Interestingly, substitution of Ser 244 with a phosphorylatable threonine residue resulted in an overall increase in autophosphorylation compared with wild type. Replacing Ser 244 of mPDK-1 with alanine also led to a significant decrease of the in vitro kinase activity of the enzyme, which was partially recovered by substitution of this residue with either glutamate or threonine (Fig. 1C). These findings further support Ser 244 as an autophosphorylation site and demonstrate that phosphorylation at this site is important for mPDK-1 autokinase activity.
To further characterize the role of Ser 244 in mPDK-1 in vitro autophosphorylation, we performed phosphoamino acid analysis and two-dimensional phosphopeptide mapping studies on in vitro autophosphorylated mPDK-1 S244A . Our results showed that although overall autophosphorylation of the mutant PDK-1 was significantly decreased compared with wild type (Fig. 1B), there was no significant change in the overall serine: threonine ratio ( Fig. 1, D, left panel versus A, left panel), suggesting that Ser 244 is not a major autophosphorylation site in vitro. Consistent with this, two-dimensional phosphopeptide mapping studies revealed that despite the significant decrease in overall phosphorylation of this mutant (Fig. 1B), mutating Ser 244 to alanine did not abolish the in vitro autophosphorylation of the three major peptides (peptides 1-3) (Fig. 1

, D, right panel versus A, right panel).
On the other hand, replacing Ser 244 of mPDK-1 with threonine led to a notable increase in threonine:serine autophosphorylation compared with the wild type enzyme (Fig. 1

, E, left panel versus A, left panel)
. Twodimensional phosphopeptide mapping studies of the mPDK-1 S244T mutant showed a significant increase in autophosphorylation of phosphopeptide e compared with wild type protein ( Fig. 1

, E, right panel versus A, right panel).
Phosphoamino acid analysis of phosphopeptide e revealed that this peptide contained only phosphoserine for the wild type and exclusively phosphothreonine for the S244T mutant of mPDK-1 (data not shown). Taken together, these results identified phosphopeptide e of wild type mPDK-1 as a Ser 244 -containing peptide.
Identification of Ser 399 as a Major Site of mPDK-1 Autophosphorylation in Vitro-Because Ser 244 of mPDK-1 was found to be only marginally phosphorylated in vitro, we attempted to identify the major mPDK-1 in vitro autophosphorylation sites. At least five serine residues in hPDK-1, including Ser 25 , Ser 241 , Ser 393 , Ser 396 , and Ser 410 were found phosphorylated in cells (17). To test whether the corresponding sites in mPDK-1 are phosphorylated in vitro, we generated mutants of mPDK-1 in which the corresponding sites, Ser 25  . On the other hand, replacing Ser 399 with glycine (mPDK-1 S399G ) significantly decreased phosphoserine content compared with that of wild type (Fig. 2, left panel versus Fig.  1A, left panel). Phosphopeptide mapping studies revealed that mutation at this site resulted in the loss of phosphopeptide 3 (Fig. 2, right panel), implicating Ser 399 of mPDK-1 as a major in vitro autophosphorylation site.
Identification of Thr 516 as a Major mPDK-1 in Vitro Autophosphorylation Site-In an attempt to localize potential threonine autophosphorylation site(s), we examined autophosphorylation of a mPDK-1 mutant lacking the carboxyl-terminal PH domain (mPDK-1 ⌬PH , residues 1-459). Phosphoamino acid analysis revealed that deletion of the PH domain resulted in a complete abolishment of threonine autophosphorylation (Fig.  3A, left panel). Phosphopeptide mapping studies of mPDK-1 ⌬PH revealed a loss of both phosphopeptides 1 and 2 (Fig. 3A, right   panel), suggesting that the major mPDK-1 threonine autophosphorylation site(s) might be localized in the PH domain.
Sequence analysis of mPDK-1 revealed the presence of four threonine residues in the PH domain: Thr 482 , Thr 516 , Thr 521 , and Thr 525 (Fig. 3B). To identify potential mPDK-1 threonine autophosphorylation sites, we mutated these threonine residues individually to alanine. The mPDK-1 threonine mutants were then expressed in CHO/IR cells, purified by immunoprecipitation, and autophosphorylated in vitro in the presence of [␥-32 P]ATP. We found that mutations at Thr 482 , Thr 521 , and Thr 525 had no significant effect on mPDK-1 autophosphorylation in vitro (data not shown). On the other hand, replacing Thr 516 with alanine significantly decreased phosphothreonine content compared with that of the wild type enzyme (Fig. 3C, left panel versus Fig. 1A, left panel). Interestingly, mutating Thr 516 to alanine resulted in the loss of both phosphothreoninecontaining peptides 1 and 2 in the two-dimensional phosphopeptide map of this mutant (Fig. 3C, right panel). One possible explanation for this observation may be that two phosphopeptides were generated because of incomplete digestion of the Thr 516 -containing tryptic peptide. Consistent with this idea, Thr 516 is immediately flanked on the amino terminus by a lysine residue (Lys 515 ). Phosphorylation at Thr 516 could potentially impair the ability of trypsin to cleave the peptide bond between Lys 515 and Thr 516 , resulting in two Thr 516 -containing phosphopeptides. To confirm this, we mutated Thr 516 of mPDK-1 to serine. mPDK-1 T516S was expressed in CHO/IR cells, immunoprecipitated, autophosphorylated in vitro, and subjected to two-dimensional phosphopeptide mapping analysis. Phosphopeptide mapping showed a significant decrease in phosphorylation at phosphopeptides 1 and 2. Subsequent phosphoamino acid analysis revealed that both phosphopeptides 1 and 2 contained exclusively phosphoserine (data not shown). These results provide further evidence that phosphopeptides 1 and 2 are due to incomplete tryptic digestion of the Thr 516containing peptide.
Ser 244 , Ser 399 , and Thr 516 of mPDK-1 Are Phosphorylated in Cells-To determine whether Ser 244 , Ser 399 , and Thr 516 are phosphorylated in cells, phosphopeptide mapping studies were performed on wild type and point-mutated mPDK-1 overexpressed and metabolically labeled in CHO/IR cells. Two-dimensional phosphopeptide mapping of the metabolically labeled wild type mPDK-1 revealed the presence of phosphopeptides 1-3 (Fig. 4, panel a), indicating that both Ser 399 and Thr 516 are phosphorylated in intact cells. Consistent with the results from in vitro autophosphorylation and two-dimensional phosphopeptide mapping studies, mutation of Thr 516 to alanine resulted in a loss of both phosphopeptides 1 and 2 for the metabolically labeled mPDK-1 (Fig. 4, panel b). Similarly, mutating Ser 399 to glycine resulted in the loss of phosphopeptide 3 (Fig. 4, panel c). Our results also showed phosphopeptide e to be  heavily phosphorylated in vivo, indicating that Ser 244 is one of the major phosphorylation sites in cells (Fig. 4). Phosphoamino acid analysis of phosphopeptide e recovered from the maps of metabolically labeled wild type or the S244T mutant of mPDK-1 indicated that this peptide contained solely phosphoserine or phosphothreonine, respectively (data not shown). Interestingly, phosphoamino acid analysis of phosphopeptides f and i recovered from the maps of metabolically labeled wild type or mPDK-1 S244T indicated that these two peptides also contained only phosphoserine or phosphothreonine, respectively (data not shown). These findings implicate Ser 244 as a major phosphorylation site of mPDK-1 in cells and suggest that phosphopeptides e, f, and i are most likely due to incomplete digestion of Ser 244 -containing peptide. Interestingly, we found that the in vivo phosphorylation pattern of kinase-inactive PDK-1 K114G is similar to that of the wild type enzyme, suggesting that PDK-1 may undergo transphosphorylation in vivo. 2 Mutation at Ser 244 and Thr 516 of mPDK-1 Affects Phosphorylation of mPDK-1 in CHO/IR Cells-We have recently found that insulin stimulates mPDK-1 phosphorylation in NIH3T3 cells and rat adipocytes (15). To test whether a mutation at Ser 399 or Thr 516 affects the overall phosphorylation of mPDK-1 in vivo, we examined basal and insulin-stimulated phosphorylation of wild type and mPDK-1 phosphorylation site mutants in CHO/IR cells. Consistent with our recent findings (15), insulin treatment resulted in a significant increase in mPDK-1 phosphorylation (Fig. 5A, lane 2 versus lane 1). Interestingly, we found that despite an overall decrease in phosphorylation, the mPDK-1 S244A mutant was still phosphorylated in cells, and the phosphorylation was enhanced in response to insulin stimulation (Fig. 5A, lanes 3 and 4 versus lanes 1 and 2). These findings suggest that insulin stimulates the phosphorylation at a site(s) in addition to or other than Ser 244 in CHO/IR cells. Although mutating Ser 399 to glycine or aspartate had no significant effect on mPDK-1 phosphorylation in vivo (data not shown), replacing Thr 516 of mPDK-1 with glutamate (Fig. 5A, lane 7 versus lane 2), but not alanine (Fig. 5A, lanes 5 versus lane 2), resulted in an increase in the basal phosphorylation of mPDK-1 to the same level as that of the insulin-stimulated wild type protein. Furthermore, this phosphorylation was not further increased following insulin treatment (Fig. 5A, lane 8 versus lane 7). Interestingly, the recently identified constitutively active mutant of mPDK-1, mPDK-1 A280V (18), also showed an enhanced and insulin-independent increase in basal phosphorylation (Fig. 5A, lanes 9 and 10 versus lanes 1 and 2). These findings suggest that phosphorylation of Thr 516 may result in a similar conformational change induced by the alanine to valine mutation at position 280, which subsequently leads to enhanced phosphorylation of mPDK-1.
Mutating Thr 516 of mPDK-1 to Glutamate Constitutively Activates mPDK-1 in Cells-To determine whether autophosphorylation plays a role in the function of mPDK-1, we examined the kinase activity of mPDK-1, both in vitro and in cells. In vitro, we found that mutation of Ser 399 of mPDK-1 to glycine or aspartate, or Thr 516 to alanine or glutamate, had no significant effect of mPDK-1 kinase activity (data not shown). To examine the effect of PDK-1 autophosphorylation on its kinase activity in cells, phosphorylation of the PDK-1 substrate Akt by wild type and autophosphorylation mutants was examined. As expected, insulin treatment resulted in a significant increase in Akt phosphorylation at Thr 308 (Fig. 5B, lane 7 versus lane 1). Although co-expression of mPDK-1 S244E or mPDK-1 S399D mutants had no effect on Akt phosphorylation at Thr 308 under basal conditions (Fig. 5B, lanes 2 and 3 versus lane 1), co-expression of mPDK-1 T516E resulted in a significant increase in Akt phosphorylation to a level similar to that induced by insulin stimulation (Fig. 5B, lane 5 versus lane 7). Under this condition, only a small increase in Akt phosphorylation at Thr 308 was observed in cells co-expressing mPDK-1 T516A (Fig.  5B, lane 4 versus lane 1), suggesting that a negatively charged residue at this site plays a critical role in promoting PDK-1 to phosphorylate Akt in cells. Furthermore, insulin stimulation did not increase the elevated Akt Thr 308 phosphorylation in cells co-expressing mPDK-1 T516E (Fig. 5B, lane 11 versus lane  5). Similar results were obtained for mPDK-1 T516E co-expressed with Akt in CHO and HEK-293 cells (data not shown). 2 Wick, M. J., and Liu, F., manuscript in preparation.

FIG. 5. Phosphorylation at Thr affects mPDK-1 phosphorylation and activity in cells.
A, replacing Thr 516 with glutamate leads to an increase in mPDK-1 phosphorylation in vivo. CHO/IR cells transiently expressing Myc-tagged wild type (WT) or various mutants of mPDK-1 were metabolically labeled with [ 32 P]orthophosphate and then treated with (ϩ) or without (Ϫ) insulin. 32 P-Labeled mPDK-1 proteins were immunoprecipitated using antibody to the tag, separated by SDS-PAGE, and detected by autoradiography (upper panel). Expression of mPDK-1 proteins was determined by Western blot (WB) analysis using antibody to the tag (lower panel). B, overexpression of PDK-1 T516E stimulates Akt phosphorylation at Thr 308 . CHO/IR co-expressing Akt and wild type (lanes 1 and 7) or mutant (lanes 3-6 and 8 -12) mPDK-1 were serum-starved and then left untreated (lanes 1-6) or treated with insulin (lanes 7-12). The cell lysates were resolved by SDS-PAGE and transferred to nitrocellulose, and Akt Thr 308 phosphorylation was determined by Western blot using a phospho-specific antibody to Thr 308 (top panel). The expression of Akt (middle panel) and PDK-1 (bottom panel) was determined by Western blot analysis using antibodies specific to these proteins. C, mPDK-1 T516E -mediated phosphorylation of Thr 308 does not require the PH domain of Akt. FLAG-tagged full-length (lanes 1-4) or PH domain-deleted mutant (lanes 5-8) of Akt were transiently expressed in CHO/IR cells together with either wild type or mutants of mPDK-1. The cells were serum-starved and treated with (ϩ) or without (Ϫ) insulin. Akt phosphorylation at Thr 308 was determined by Western blot using a phospho-specific antibody. The expression of FLAG-tagged Akt, ⌬PH-Akt, and Myc-tagged PDK-1 was determined by Western blot using antibodies to FLAG (middle panel) or PDK-1 (bottom panel), respectively.
In agreement with our recent study (18), co-expression of mPDK-1 A280V with Akt also resulted in a similar increase in basal Akt phosphorylation at Thr 308 (Fig. 5B, lane 6).
Previous studies have suggested that translocation of Akt to the plasma membrane is required for full phosphorylation and activation of the enzyme (23,24). Studies have also shown that this translocation requires an intact PH domain (25). To better understand the mechanism by which autophosphorylation of PDK-1 regulates its activation of downstream substrates, we co-expressed FLAG-tagged wild type Akt-1 or the PH domain deletion mutant (Akt-1 ⌬PH ) with either wild type or mutant mPDK-1 in CHO/IR cells. Insulin stimulation (Fig. 5C, lane 2 versus lane 1) or co-expression of either mPDK-1 T516E (Fig. 5C, lane 3 versus lane 1) or mPDK-1 A280V (Fig. 5C, lane 4 versus lane 1) resulted in significant Akt phosphorylation at Thr 308 . The PH domain-deleted mutant of Akt was not phosphorylated on Thr 308 in response to insulin stimulation (Fig. 5C, lane 6). However, co-expression of PDK-1 T516E was sufficient to restore Akt-1 ⌬PH phosphorylation at Thr 308 to the same level as that of insulin-stimulated, full-length Akt (Fig. 5C, lane 7 versus lane  2). It is of interest to note that the insulin-independent and Akt PH domain-independent phosphorylation of Akt at Thr 308 can be mediated by both PDK-1 T516E and PDK-1 A280V mutants (Fig. 5, B, lanes 5 and 6 and C, lanes 7 and 8). These data suggest that substituting Val 280 with alanine or Thr 516 with glutamate may lead to a similar conformational change of mPDK-1, which is essential for interaction and phosphorylation of downstream substrates such as Akt in cells. DISCUSSION The mechanism(s) by which PDK-1 activity is regulated has remained, for a large part, unclear. Subcellular localization and substrate conformation have been hypothesized as the primary means for regulating PDK-1 activity enzyme (22,26,27); however, some studies have indicated that phosphorylation of PDK-1 may also play a role (8,15). PDK-1 is phosphorylated both in vitro and in cells (2,17), and treatment of PDK-1expressing cells with insulin or hydrogen peroxide has been shown to increase phosphorylation of PDK-1 on specific residues (15,28). Several studies have focused on Ser 244 (mPDK-1) or Ser 241 (hPDK-1) in the activation loop as a critical phosphorylation site, because phosphorylation upon this residue is essential for PDK-1 kinase activity (15,17). Some studies have indicated that phosphorylation at Ser 241 is not regulated by growth factor stimulation (17). On the other hand, we have found that in certain cell types, mPDK-1 phosphorylation and activity increase following insulin stimulation in a Ser 244 -dependent manner (15). However, in CHO/IR cells, we found that the mPDK-1 S244A mutant still underwent insulin-stimulated phosphorylation (Fig. 5A). These findings suggest that insulinstimulated phosphorylation of mPDK-1 occurs at sites other than or in addition to Ser 244 , and this phosphorylation may function in conjunction with Ser 244 to regulate mPDK-1 activity in cells.
In the present study, we identify Ser 399 and Thr 516 on mPDK-1 as two major in vitro autophosphorylation sites, which can also be phosphorylated in cells. This finding is in agreement with a recent study by Park et al. (28), who showed the Thr 516 -corresponding site in hPDK-1 to be phosphorylated in vitro. However, our results also show that phosphorylation at these sites may play different roles in PDK-1 function in cells. Indeed, substitution of Thr 516 with glutamate increases basal PDK-1 phosphorylation in cells to that seen with insulin stimulation, whereas replacing Ser 399 with aspartate does not significantly affect either basal or insulin-stimulated phosphorylation of the enzyme. In addition, overexpression of mPDK-1 T516E , but not mPDK-1 S399D , led to Akt phosphorylation at Thr 308 to the same level as that induced by insulin treatment. Taken together, these findings suggest that autophosphorylation at Thr 516 , but not Ser 399 , regulates PDK-1 phosphorylation upon additional residues in cells and subsequently functions to regulate the activity of PDK-1 toward its substrates.
Previous studies have shown that hPDK-1 isolated from cells is constitutively active in vitro, presumably as a result of autophosphorylation upon Ser 241 (4,8,22). In agreement with this, we have found that a mutation of the corresponding Ser 244 on mPDK-1 renders the kinase inactive in vitro (Fig. 1C) and in cells (Ref. 15 and data not shown). Interestingly, although we found the kinase activity of mPDK-1 to be significantly reduced by a mutation at Ser 244 , we found the ratio of serine:threonine autophosphorylation unchanged in its absence. This suggests that Ser 244 is not a major site for in vitro autophosphorylation. Low autophosphorylation at Ser 244 may result from phosphorylation at this site in cells that remains stable in vitro. In support of this, phosphorylation at Ser 244 is less stable if the residue is replaced with threonine, resulting in an increase in autophosphorylation at this site in vitro (Fig. 1B). Although Ser 244 is a minor phosphorylation site, phosphopeptide maps of metabolically labeled wild type and S244T mutant of mPDK-1 indicated that Ser 244 is highly phosphorylated in CHO/IR cells, which is in agreement with our recent studies showing that Ser 244 is a major phosphorylation site in NIH-3T3 cells (15). In contrast, autophosphorylation upon the major in vitro sites, Ser 399 and Thr 516 , is less stable in cells, which indicates that phosphorylation upon these residues is reversible and may be tightly regulated in cells.
Insulin stimulation has been shown to increase PDK-1 phosphorylation in both NIH-3T3 and rat adipose cells (15). Consistent with this, we observed insulin-stimulated phosphorylation of mPDK-1 when overexpressed in CHO/IR cells (Fig. 5A); however, we found that this insulin-stimulated increase could occur in the absence of Ser 244 phosphorylation. Mutations at Ser 399 and Thr 516 did not significantly affect insulin-stimulated mPDK-1 phosphorylation (Fig. 5A and data not shown), suggesting the presence of other insulin-stimulated phosphorylation site(s). On the other hand, replacement of Thr 516 with a negatively charged glutamate residue resulted in a significant increase in basal phosphorylation. The increase in basal mPDK-1 T516E phosphorylation also correlates with an increase in mPDK-1 T516E -mediated phosphorylation of Akt at Thr 308 (Fig. 5B), which is in agreement with our recent finding that insulin-stimulated phosphorylation of mPDK-1 correlates with an increase in activity (15).
It has been previously shown that growth factor-mediated phosphorylation and activation of Akt requires a translocation event via an intact PH domain (23)(24)(25)29). Recently, we (18) and others (22,30) have shown that phosphorylation at Thr 308 of Akt by PDK-1 is facilitated by an intact PDK-1-PH domain, whereas others have suggested that translocation of PDK-1 to the membrane may play a role in the phosphorylation and activation of the enzyme (31). In the present study, we have found that mutating Thr 516 to glutamate allows PDK-1 to phosphorylate Thr 308 not only in a growth factor-independent fashion but also without the requirement for the PH domain of Akt (Fig. 5C). These data suggest that the T516E mutation of mPDK-1 bypasses the normal requirement for translocation and allows it to interact with and subsequently phosphorylate Akt at Thr 308 . One possible explanation for this observation may be that the introduction of a negatively charged residue at Thr 516 results in a conformational change in mPDK-1, which normally occurs after growth factor-mediated translocation of the enzyme to the plasma membrane. This resulting conformation may result in an increase in autophosphorylation at Ser 244 and a subsequent increase in mPDK-1 activity. Consistent with this, a mPDK-1 S244A/T516E mutant was unable to increase phosphorylation of mPDK-1 or Thr 308 of Akt in cells (data not shown), suggesting that phosphorylation at Thr 516 modifies activity of mPDK-1 through a Ser 244 phosphorylation-dependent mechanism.
In Caenorhabditis elegans, a natural point mutation (A277V) allows the mammalian PDK-1 homolog to bypass the need for PI 3-kinase for its activation (16). We recently reported that a corresponding mutation in mPDK1 (A280V) resulted in an increase in the in vitro autophosphorylation of mPDK-1 and constitutive activation of the kinase toward Akt in cells (18). Whereas this mutation has not been found to occur naturally in higher species, it is interesting to note the similarity between mPDK-1 A280V and mPDK-1 T516E . We found that although neither had any significant effect on mPDK-1 activity in vitro (data not shown), both elevated mPDK-1 basal phosphorylation in cells (Fig. 5A). Furthermore, overexpression of either mPDK-1 A280V or mPDK-1 T516E was sufficient to elevate Akt phosphorylation at Thr 308 in a growth factor-and Akt PH domainindependent manner (Fig. 5, B and Fig. C). These findings suggest that each mutation may result in a conformational change and/or cellular relocalization of mPDK1, which allows the enzyme to interact with and phosphorylate its downstream substrates such as Akt.
In conclusion, our results suggest that phosphorylation at Thr 516 plays an important role in the regulation of mPDK-1 in vivo activity toward its substrates. Although it remains to be established whether phosphorylation upon Thr 516 is directly regulated by insulin stimulation, autophosphorylation at this site may specifically direct the extent of phosphorylation at other site(s) and subsequently the activity of mPDK-1 in response to insulin or other growth factors. Autophosphorylation at Thr 516 may be specifically regulated in cells by various cellular events, such as conformational changes of the kinase following growth factor stimulation. Alternatively, Thr 516 may be constitutively autophosphorylated in cells, and this phosphorylation may be tightly controlled by a serine/threonine phosphatase, allowing dynamic regulation of this pivotal kinase in response to growth factor stimulation. Further studies will be needed to test these possibilities.