Involvement of 3-phosphoinositide-dependent protein kinase-1 in the MEK/MAPK signal transduction pathway.

The phosphatidylinositide-3-OH kinase/3-phospho-inositide-dependent protein kinase-1 (PDK1)/Akt and the Raf/mitogen-activated protein kinase (MAPK/ERK) kinase (MEK)/mitogen-activated protein kinase (MAPK) pathways have central roles in the regulation of cell survival and proliferation. Despite their importance, however, the cross-talk between these two pathways has not been fully understood. Here we report that PDK1 promotes MAPK activation in a MEK-dependent manner. In vitro kinase assay revealed that the direct targets of PDK1 in the MAPK pathway were the upstream MAPK kinases MEK1 and MEK2. The identified PDK1 phosphorylation sites in MEK1 and MEK2 are Ser222 and Ser226, respectively, and are known to be essential for full activation. To date, these sites are thought to be phosphorylated by Raf kinases. However, PDK1 gene silencing using small interference RNA demonstrates that PDK1 is associated with maintaining the steady-state phosphorylated MEK level and cell growth. The small interference RNA-mediated down-regulation of PDK1 attenuated maximum MEK and MAPK activities but could not prolong MAPK signaling duration. Stable and transient expression of constitutively active MEK1 overcame these effects. Our results suggest a novel cross-talk between the phosphatidylinositide-3-OH kinase/PDK1/Akt pathway and the Raf/MEK/MAPK pathway.

Many growth factors and cytokines have been reported to promote cell survival. Interaction between these factors and their specific receptors trigger the activation of phosphatidylinositide-3-OH kinase (PI3K) 1 and Ras (1). Activated PI3K generates the phospholipid second messenger molecules phosphatidylinositol 3,4,5-trisphosphate and phosphatidylinositol 3,4-bisphosphate (2)(3)(4). These lipids, then, induce the activation of several members of the protein kinase A, G, and C (AGC) family of protein kinases including Akt/protein kinase B, p70 ribosomal protein S6 kinase (p70 S6K ), protein kinase C isoforms (PKCs), and serum-and glucocorticoid-inducible kinases (SGKs). Activated Ras stimulates Raf translocation from cytosol to the cell membrane by the direct interaction (5,6), and the membrane-translocated Raf is phosphorylated and activated by multiple kinases. Activated Raf catalyzes the phosphorylation of its downstream kinases, MEK1 and MEK2 (MEK1/2), through phosphorylation at Ser 218 and Ser 222 of MEK1 and at Ser 222 and Ser 226 of MEK2 in their activation loops (7,8). Then activated phospho-MEK1/2 triggers MAPK signaling pathways (9,10). Activated kinases including Akt and MAPK1/2 mediate survival signal transduction and cell cycle progression by phosphorylating downstream key regulatory proteins (11)(12)(13). Because it has been reported that the activity of Akt or MAPK or both are elevated in many cancer cells, these molecules are thought to be suitable as molecular targets of anticancer drugs (1, 14 -16).
PDK1 was originally identified as a kinase that could phosphorylate Akt on its activation loop (residue Thr 308 ) (17)(18)(19). Later works, however, have shown that PDK1 is not only an Akt kinase but also a kinase responsible for phosphorylating members of the AGC family of protein kinases: p70 S6K , SGKs, PKCs, and p90 ribosomal protein S6 kinases (RSKs) at the equivalent residues of Thr 308 of Akt (reviewed in Ref. 13). PDK1 itself is also a member of the AGC subfamily of protein kinases and is phosphorylated at the Ser 241 residue (equivalent to Thr 308 of Akt) in the activation loop. Since PDK1 expressed in bacteria was active and was phosphorylated at Ser 241 , it is thought that PDK1 phosphorylates itself at this site (20).
During our analysis of the PDK1-mediated signal transduction pathway, we discovered an increase in the phosphorylated MAPK level of PDK1-transfected cells and saw that MAPK activation depended upon MEK activation. Sequence comparison of the residues around the PDK1-mediated phosphorylation sites in AGC kinases and MEK1/2 revealed that MEK1/2 possessed the PDK1-mediated phosphorylation sites. In vitro and in vivo analysis revealed that PDK1 could directly phosphorylate the sites (Ser 222 of MEK1 and Ser 226 of MEK2). To date, these sites were thought to be phosphorylated by Raf kinases. Because silencing of the PDK1 gene by siRNA decreased the phospho-MEK and phospho-MAPK levels and suppressed cell proliferation, PDK1 regulated the MEK/MAPK pathway by directly phosphorylating MEK1/2. Our results indicate the importance of PDK1 as a MAPK kinase kinase in MEK/MAPK signaling pathways.
Transient and Stable Transfection, Immunoprecipitation, and Western Blot Analysis-Cells were transfected with the appropriate plasmids using Superfect transfection reagent (Qiagen, Chatsworth, CA) or LipofectAMINE 2000 reagent (Invitrogen), according to the manufacturers' instructions. Stable mock, WT-MEK1, or DD-MEK1 transfectants were established by transfecting pUSEamp encoding nothing, WT-MEK1, or DD-MEK1 into HT1080 cells. The stable transfectants were selected by cultivating them in medium containing 400 g/ml Geneticin (Invitrogen).
Immunostaining-Cells were transfected with pUSEamp-WT-MEK2 and pcDNA3-EGFP together with a pFLAG-CMV-2 vector encoding nothing (Mock) or ⌬N51-PDK1. 24 h after transfection, cells were plated onto collagen-coated culture dishes. After incubation for a further 24 h, cells were washed with phosphate-buffered saline and fixed in 3.7% formaldehyde for 15 min. Permeabilization was carried out in 0.2% Triton X-100 for 5 min. After incubation for 1 h in phosphate-buffered saline supplemented with 1% bovine serum albumin, the labeling was carried out by incubation overnight with a rabbit polyclonal anti-phospho-MAPK (Thr 202 /Tyr 204 ) antibody (Cell Signaling Technology) followed by a 90-min incubation with AlexaFluor 568-conjugated goat anti-rabbit IgG (Molecular Probes, Inc., Eugene, OR). After washing the cells, we visualized them using a fluorescence microscope (Olympus IX-70; Olympus, Tokyo, Japan) equipped with a CCD camera.
In Vitro Kinase Assay-To perform the in vitro kinase assay, recombinant, active Raf-1 or immunopurified, FLAG-tagged ⌬N51-PDK1 was incubated with recombinant inactive MEK1 (0.3 g/assay), recombinant inactive MAPK2 (1 g/assay), or both in 50 l (final volume) of kinase reaction buffer for PDK1 ( Reactions were spotted onto P81 phosphocellulose paper, washed three times with 0.75% phosphoric acid, air-dried, and subjected to Cerenkov counting. PDK1-mediated MEK phosphorylation was also estimated by incubating immunoprecipitated, FLAG-tagged ⌬N51-PDK1, or the kinase-dead form of ⌬N51-PDK1 (S241A) with recombinant inactive MEK1 (0.8 g/assay) in kinase reaction buffer for PDK1 for 1 h at 30°C. Appropriate amounts of recombinant ⌬N51-PDK1 and the kinase-dead form of ⌬N51-PDK1 (K111A/D223A) were also incubated with recombinant inactive MEK1 (1 g/assay) in kinase reaction buffer for PDK1 for 2 h at 37°C. Reactions were electrophoresed and immunoblotted with antibodies to phospho-MEK, MEK1, phospho-PDK1 (Ser 241 ) (Cell Signaling Technology), and PDK1. In some experiments, WT or point-mutated MEK1 proteins were immunoprecipitated and incubated with recombinant, inactive MAPK2 (0.5 g/assay) in kinase reaction buffer for Raf-1 for 30 min at 30°C, following immunoblot analysis with antibodies to phospho-MAPK and MAPK. In other experiments, MAPK activity was measured using a MAPK immunoprecipitation kinase assay kit (Upstate Biotechnology), according to the manufacturer's instructions.

Activation of MEK/MAPK Signal Transduction
Pathway by PDK1-To analyze the MEK/MAPK signaling pathway, 293T cells were transfected with pUSEamp-MEK2 plasmid together with EGFP-expressing plasmid to detect the transfected cells. Overexpression of MEK2 alone could not induce MAPK phosphorylation (Fig. 1A, yellow arrowheads). Interestingly, co-expression of ⌬N51-PDK1, which possesses almost the same activity as full-length WT-PDK1 (data not shown), resulted in the increased phospho-MAPK amount in the transfected cells (Fig.  1A, blue arrowheads). We then examined whether or not PDK1 directly activates MAPK. In vitro incubation of PDK1 with recombinant, inactive MAPK2 did not induce MAPK2 activation (Fig. 1B). We observed PDK1-induced MAPK activation in vitro only in the presence of both inactive MEK1 and inactive MAPK2, as we found with Raf-1 (Fig. 1B). Therefore, PDK1 cannot directly activate MAPK; it needs MEK for in vitro activation. The fact that the immunoprecipitated, kinase-dead form of PDK1 (S241A) failed to phosphorylate and activate MEK1 indicates that PDK1-dependent MEK1 activation is not mediated by other co-precipitated kinases (Fig. 1C) (data not shown). Moreover, in vitro incubation of recombinant inactive MEK1 with purified recombinant ⌬N51-PDK1, but not the kinase-dead form of ⌬N51-PDK1, increased the phospho-MEK1 levels in a dose-dependent manner (Fig. 1D). The results suggest that the target of PDK1 in the MAPK signaling pathway is not MAPK itself but rather its upstream kinase MEK.
PDK1 was originally isolated as a kinase responsible for the phosphorylation of Akt on its activation loop (Thr 308 residue) (17)(18)(19). Later studies have shown that PDK1 is not only an Akt kinase, but it also regulates multiple kinases, which belong to the AGC family of protein kinases through phosphorylating Ser or Thr residues equivalent to Thr 308 of Akt (13,29). Se-quence comparison of the residues around Ser 222 in human MEK1 and Ser 226 in human MEK2 in their activation loops showed that these residues had homology to the previously reported PDK1 phosphorylation sites in human Akt1, SGK1, and PAK1 (Fig. 1E) (13,29). To prove that MEK phosphorylation was PDK1-dependent, we examined the change in phospho-MEK1/2 levels with an anti-phospho-MEK antibody. The Cell Signaling antibody used recognized MEK1/2 only when MEK1/2 were phosphorylated at Ser 218 and/or Ser 222 in MEK1 and at Ser 222 and/or Ser 226 in MEK2. Immunoblot analysis showed that co-transfection of PDK1 increased the phospho-MEK1/2 levels in cells (Fig. 1F). Similar results were obtained using a Santa Cruz Biotechnology antibody (sc-7995-R) that could recognize MEK1/2 when MEK1/2 were phosphorylated at Ser 218 and/or Ser 222 in MEK1 and at Ser 222 and/or Ser 226 in MEK2 (data not shown). We also found increased phospho-MEK levels when we used a BIOSOURCE antibody that specifically recognized phosphorylated MEK1 (Ser 222 ) or MEK2 (Ser 226 ) (data not shown). To exclude the possibility that MEK phosphorylation was accomplished by autophosphorylation, we generated a kinase-dead form of MEK1 by mutating Lys 97 to Ala. As shown in Fig. 1G, PDK1 also increased the phospho-MEK1 (Ser 218 /Ser 222 ) levels in KD-MEK1. Of note, the antiphospho-MEK1/2 antibody used did not recognize DD-MEK in which both Ser 218 and Ser 222 residues had been substituted for the phosphomimic Asp residues. These results indicate that PDK1 activates the MEK/MAPK signaling pathway by directly phosphorylating MEK1/2.
Because PDK1 is known to phosphorylate and activate multiple downstream kinases (13,29), we investigated the role of PDK1 downstream kinases on MEK phosphorylation. Akt and SGK were known to act as main mediators of PI3K/PDK1regulated biological responses (12,13). The increase in phospho-MEK levels was not observed in 293T cells that had been transfected with the active form of Akt1 or SGK1 cDNA although transfection of PDK1 or the active form of Raf-1 cDNA induced MEKl phosphorylation (Fig. 3A). Thus, Akt and SGK might not be involved in MEK phosphorylation and MAPK activation. PDK1 is also known to phosphorylate and activate PKCs (13, 18, 29), especially PKC, which was reported to activate Raf-1 following MAPK activation in neuronal cells (30,31). In 293T cells, however, transfection of WT-PKC cDNA did not increase phospho-MEK and phospho-MAPK levels (Fig.  3B). Moreover, overexpression of the dominant negative form of PKC did not interfere with PDK1-dependent MEK phosphorylation (Fig. 3B). We further examined the effects of PKC inhibitors on Raf/MEK signaling. The PKC␦ inhibitor rottlerin and synthetic myristoylated PKC and PKC pseudosubstrate inhibitor peptides had minimal effects on PDK1-dependent MEK phosphorylation in 293T and HT1080 cells (Fig. 3C). Therefore, PKCs may only be involved in PDK1-dependent MEK phosphorylation in neuronal cells.
PDK1 Preferentially Phosphorylates MEK1 at the Ser 222 Residue-Although MEK has been reported to be phosphorylated by several kinases, such as Tpl-2, Mos, and MEKK-1 (32-34), the predominant MEK activator in most cell types is reported to be Raf-1 (35). To identify the PDK1-mediated phosphorylation sites in MEK, we generated several MEK1 point mutants and transfected them into 293T cells together with plasmids containing WT-PDK1 or the active form of Raf-1 cDNA. Consistent with previous reports (7,8), Raf-1 phosphorylated MEK1 at both Ser 218 and Ser 222 residues, because MEK phosphorylation was observed whenever either site was converted to Ala or Asp. In contrast, PDK1 phosphorylated MEK1 only at the Ser 222 residue, because the anti-phospho-MEK1 antibody did not detect the increases in levels of the phosphorylated form of S222A-or S222D-MEK1, even when cells were cotransfected with WT-PDK1 (Fig. 4A). An in vitro kinase assay revealed that PDK1 transfection increased the kinase activity of S218D-MEK1 but not of S222D-MEK1 (Fig. 4A). Raf-1 overexpression increased MEK kinase activity whenever either Ser 218 or Ser 222 were mutated to Asp. To confirm the result, we incubated the immunoprecipitated and PP2A-treated WT-, S218A-, and S222A-MEK1 with immunoprecipitated PDK1 in vitro. Immunoblot analysis revealed that PDK1 could phospho- These results indicate that PDK1 preferentially phosphorylates MEK1 at the Ser 222 residue, whereas Raf-1 can phosphorylate both Ser 218 and Ser 222 residues, as reported previously (7,8).
Interestingly, S218D-MEK1 was constitutively phosphorylated at the Ser 222 residue in cells (Fig. 4A). To analyze the mechanisms of increased Ser 222 phosphorylation, we examined the change of PDK1 binding to point-mutated MEK1. The S218D mutation increased its association with PDK1 (Fig. 4B). When point-mutated MEK1 was phosphorylated by overex-pressing the active form of Raf-1, PDK1 bound to S218A-MEK1 and S222A-MEK1 almost equally (Fig. 4C). Moreover, PDK1 bound to phosphorylated S218D-MEK1 and S222D-MEK1 (Fig.  4C), and the amounts of co-immunoprecipitated S218D-or S222D-MEK1 were larger than that of co-immunoprecipitated S218A-or S222A-MEK1. Raf-1 binding to MEK1 seemed to be unaffected by point mutation at either the Ser 218 or Ser 222 residue (Fig. 4D). These results suggest that phosphorylation of either the Ser 218 or Ser 222 residue in MEK1 increases its association with PDK1 and that the phosphorylation of both sites in MEK1 further increased its binding ability to PDK1. Because the Ser 222 residue in S218D-MEK1 was constitutively phosphorylated in cells (Fig. 4A), S218D-MEK1 exhibited high affinity to PDK1 in cells (Fig. 4B).
PDK1 Gene Silencing Attenuates the MEK/MAPK Signal Transduction Pathway-To confirm the role of PDK1 in MEK/ MAPK signaling pathway in vivo, we tried to knock down PDK1 expression using siRNA. We designed five siRNAs from the human PDK1 gene sequence. Although there was a difference in silencing ability, all of the siRNAs were able to suppress endogenous PDK1 protein expression in 293T and A549 cells (Fig. 5A). As reported previously (17)(18)(19), PDK1 is essential for Akt phosphorylation at the Thr 308 site. Transfection of PDK1-4 siRNA into 293T cells reduced the phospho-Akt (Thr 308 ) level without affecting the Akt protein amount (Fig. 5B). Therefore, PDK1-4 siRNA was useful to investigate the phosphorylation status of PDK1 substrates in cells.
When 293T and HT1080 cells were transfected with PDK1-4 siRNA, endogenous PDK1 expression was significantly suppressed (Fig. 5C). In this condition, we observed the decrease in steady-state levels of phospho-MEK and phospho-MAPK in 293T and HT1080 cells without affecting the MEK and MAPK protein amount. The inhibitory effects on MEK/MAPK signaling were also found in cells transfected with siRNA to c-Raf-1, whereas siRNA to Akt1 and Akt2 (Aktc siRNA) had no effect. These results clearly showed the connection between PDK1 and MEK/MAPK signaling in cells. For validation of siRNA data, we generated the pCMV3-PDK1-4 Res plasmid in which three bases of PDK1-4 siRNA-targeting sequence were mutated without affecting amino acid sequence of PDK1 protein.
As shown in Fig. 5D, PDK1 protein expression was observed in PDK1-4 siRNA-treated cells when the cells were transfected with pCMV3-PDK1-4 Res . In contrast, no PDK1 protein was expressed in PDK1-4 siRNA-treated cells even when cells were co-transfected with pCMV3-WT-PDK1. The PDK1-4 siRNAmediated decrease in phospho-MEK levels was rescued by cotransfection with pCMV3-PDK1-4 Res but not with pCMV3-WT-PDK1. These results indicate that the effect of PDK1-4 siRNA on lowering the phospho-MEK level is not an artifact.
To exclude the possibility that PDK1 gene silencing affected the association between Raf-1 and MEK, MEK was immunoprecipitated from control siRNA-or PDK1-4 siRNA-transfected cells. As shown in Fig. 5E, the amount of co-immunoprecipitated c-Raf-1 from PDK1-4 siRNA-transfected 293T cells was almost the same as that from nonsilencing siRNA-transfected cells. When cells were treated with EGF or phorbol 12-myristate 13-acetate, the activation of MEK signaling was observed whether or not PDK1 expression was suppressed (Fig. 5, F and  G). However, PDK1 gene silencing attenuated the maximum MEK phosphorylation level and the functional read-out (phospho-RSK and phospho-MAPK). Because suppression of PDK1 expression could not prolong the duration time of MEK/MAPK/ RSK signaling, Raf-1 might compensate MEK phosphorylation in EGF-or phorbol 12-myristate 13-acetate-stimulated cells (Fig. 5, F and G). Therefore, PDK1 was involved in MEK/ MAPK signaling as a MAPK kinase kinase to increase the maximum MAPK activity that might lead to cell proliferation or differentiation.
Overcoming PDK1 Gene Silencing by Expressing Constitutively Active MEK-To confirm that PDK1 regulates MAPK signaling by directly phosphorylating MEK, we transfected PDK1-4 siRNA to 293T cells that had been transfected with WT-MEK1 or constitutively active DD-MEK1 cDNA. Transfection of PDK1-4 siRNA or c-Raf-1 siRNA decreased the phosphorylated forms of MAPK, MEK, and RSK in WT-MEK1 cDNA-transfected cells (Fig. 6A). In contrast, PDK1-4 siRNA or c-Raf-1 siRNA could not decrease the phospho-MAPK or phospho-RSK level in DD-MEK1 cDNA-transfected cells, but these same siRNAs down-regulated phospho-MEK levels (Fig. 6A). We generated HT1080 cell clones that stably transfected with mock (M1 and M2 cells), WT-MEK1 cDNA (WT29 cells), or DD-MEK1 cDNA (DD5 cells), and we examined the PDK1 gene silencing in these stable transfectants. Transfection of PDK1-4 siRNA suppressed MEK/MAPK/RSK or MEK/MAPK/c-Myc signaling in M1 and WT29 cells (Fig. 6B). In DD5 cells, PDK1-4 siRNA attenuated the phospho-MEK level but not the phospho-MAPK, phospho-RSK, or phospho-c-Myc levels (Fig. 6B). Measurement of MAPK activity confirmed that it was suppressed by PDK1-4 siRNA in mock transfectants (M2) but not in DD-MEK1 cDNA transfectants (DD5; Fig. 6C). When we examined the growth of stable transfectants, we found that PDK1 gene silencing by PDK1-4 siRNA attenuated M1 (filled circles) and WT29 (filled squares) cell growth but not DD5 (filled triangles) cell growth (Fig. 6D). These results strongly indicate that PDK1 plays an important role in MEK/MAPK signaling that is involved in cell proliferation by regulating MEK activity. DISCUSSION The MAPK cascades are evolutionarily conserved signaling pathways from yeast to humans, and they control such fundamental cellular processes as proliferation, differentiation, survival, and apoptosis. The cascades consist of three kinases: MAPK kinase kinase, MAPK kinase, and MAPK. There are four subgroups in the MAPK family: MAPK/ERK, p38, ERK5, and c-Jun N-terminal kinase/stress-activated protein kinase. Each MAPK has its own MAPK kinase kinase and MAPK kinase. Of the four MAPK cascades, the most extensively studied pathway is the Raf/MEK/MAPK cascade (35)(36)(37).
Upon growth factor stimulation, Raf is translocated to the membrane and activated in a Ras-dependent manner. Activated Raf stimulates dual specificity protein kinase MEK activation by phosphorylating two serine residues in their activation loops, which results in MAPK activation. Activated MAPK mediates survival and growth signals by phosphorylating RSKs and transcriptional factors such as cAMP response element-binding protein, Elk-1, and c-Jun (36). Growth factors also stimulate the activation of PI3K. The activated PI3K, then, generates phospholipid second messenger molecules phosphatidylinositol 3,4,5-trisphosphate and phosphatidylinositol 3,4-bisphosphate, which raise a diverse set of cellular responses. The major targets of phosphatidylinositol 3,4,5trisphosphate and phosphatidylinositol 3,4-bisphosphate are PH domain-containing proteins, such as Akt (also known as protein kinase B) (12,13). Akt mediates many PI3K-regulated biological responses including glucose uptake, protein synthesis, cell cycle progression, and apoptosis inhibition. For Akt activation, Akt needs to translocate from cytosol to plasma membrane and to be phosphorylated at two specific phosphorylation sites, one in the "T-loop" of the kinase domain (Thr 308 ) and the other in the COOH-terminal of the catalytic domain (Ser 473 ) in a region termed the hydrophobic motif (12,13). Phosphorylation of Akt at Thr 308 is catalyzed by the ubiquitously expressed PDK1, and the kinase responsible for phosphorylation of Akt at Ser 473 is called PDK2 (13,29). Although PDK1 was originally identified as an Akt kinase, later works revealed that PDK1 also participates in the activation of members of the AGC family of protein kinases by phosphorylating their equivalent residues of Thr 308 of Akt (13,29).
Although Raf/MEK/MAPK and PI3K/PDK1/Akt pathways have been thought to organize distinct cascades, recent reports suggested that a certain degree of cross-talk exists between these pathways. For example, Akt was reported to inactivate Raf-1 through phosphorylation at Ser 259 (38) although this inhibitory effect appears to depend on cell type and stage of differentiation (39). Moreover, MAPK and PDK1 coordinately activate RSK2, and the activated RSK2, in turn, activates PDK1 (40). In this report, we described a novel cross-talk between these two cascades. Four lines of evidence support the assumption that PDK1 is involved in MEK/MAPK signaling. First, overexpression of PDK1 activates MAPK in vivo and in vitro (Fig. 1); second, PDK1 directly phosphorylates MEK1/2 in the activation loop ( Figs. 1 and 2); third, PDK1 binds to MEK in vivo (Fig. 4); fourth, siRNA directed to PDK1 decreased phospho-MEK and phospho-MAPK levels in vivo (Figs. 5 and 6). Because PDK1 was reported to activate PKCs (13,18,29), it is possible that PDK1 activates PKCs, leading to activation of Raf-1. Previous reports have also suggested that PKC is associated with PDK1-dependent MAPK activation in neuronal cells (30,31). In our hands, PDK1-dependent MAPK activation was not dependent on PKCs, because DN-PKC transfection or PKC inhibitors had no effects on PDK1-mediated MEK phosphorylation (Fig. 3).
Although PDK1 gene silencing did not affect the association between Raf-1 and MEK (Fig. 4D), we could not exclude the possibility that PDK1-4 siRNA transfection decreased phospho-MEK levels by suppressing c-Raf-1 kinase activity. In fact, we found a slight decrease in c-Raf-1 kinase activity and phospho-c-Raf-1 level in PDK1-4 siRNA-transfected 293T cells (data not shown). Because the role of PDK1 in the regulation of c-Raf-1 kinase activity has not been reported yet, we have to clarify the mechanism in future. However, there is no doubt that PDK1 is associated with MEK phosphorylation, since PDK1 directly phosphorylates MEK in vitro (Fig. 1), and PDK1 binds to MEK in cells (Fig. 4). We observed nearly equal decrease in phospho-MEK and phospho-MAPK levels in PDK1-4 siRNA-and c-Raf-1 siRNA-transfected cells (Fig. 5C). It does not mean that PDK1 and c-Raf-1 contribute equally in the activation of MEK, because the gene silencing efficiency of PDK1-4 siRNA seemed to be stronger than that of c-Raf-1 siRNA. Thus, c-Raf-1 might play the main role in activating MEK/MAPK signaling; PDK1 might contribute to maintaining the steady-state phosphorylated MEK level.
Recently, Alessi and co-workers (41) reported that MEK/ MAPK activity was consistently about 2-fold higher in PDK1 Ϫ/Ϫ ES cells than in PDK1 ϩ/ϩ ES cells. They discussed in their report (41) that PDK1 might suppress the basal MEK/MAPK activity through Akt-mediated Raf inactivation. In our experiments, however, interference of PDK1 expression by siRNA attenuated the basal levels of phospho-MEK and phospho-MAPK (Fig. 5C). Moreover, transfection of PDK1 cDNA increased the MEK/ MAPK activation (Figs. 1 and 2). It was not clear why PDK1 Ϫ/Ϫ ES cells showed this increased MEK/MAPK activity. Because ES cells express ES cell-specific ras and other genes (42), the PDK1dependent MEK/MAPK activation seemed to be dependent on cell type and stage of differentiation.
All of the reported PDK1 phosphorylation sites in the AGC family of kinases possessed the conserved pTFCGT motif (where pT represents PDK1-targeting Thr) (Fig. 1E) (13,29). PDK1 could also phosphorylate p21-activated protein kinase-1 (PAK1) at the Thr 423 residue of the pTMVGT motif (where pT represents PDK1-targeting Thr) (43), suggesting that there is some redundancy in the PDK1 targeting site. Our identified PDK1 phosphorylation sites in MEK1 and MEK2 had homology to the motif, although the PDK1-phosphorylation site was After transfection for 24 h, cells were replated onto a 12-well plate at 1 ϫ 10 4 cells/well. 48, 72, and 96 h after transfection, a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay was performed as described under "Experimental Procedures." not the Thr but rather the Ser residue (Fig. 1E). Because PDK1 was known to phosphorylate by itself at the Ser 241 residue (Fig.  1E) (20) and Ser residues in RSK2 and MSK1/2 (40), it was no surprise to us that PDK1 phosphorylated MEK1/2 at the Ser residues in the activation loop. Since several proteins, other than the AGC family of protein kinases, were also phosphorylated by PDK1 in cells, 2 PDK1 might be involved in the regulation of many kinases in addition to the AGC kinases.