Phosphorylation and Activation of p70s6k by PDK1

See allHide authors and affiliations

Science  30 Jan 1998:
Vol. 279, Issue 5351, pp. 707-710
DOI: 10.1126/science.279.5351.707


Activation of the protein p70s6k by mitogens leads to increased translation of a family of messenger RNAs that encode essential components of the protein synthetic apparatus. Activation of the kinase requires hierarchical phosphorylation at multiple sites, culminating in the phosphorylation of the threonine in position 229 (Thr229), in the catalytic domain. The homologous site in protein kinase B (PKB), Thr308, has been shown to be phosphorylated by the phosphoinositide-dependent protein kinase PDK1. A regulatory link between p70s6k and PKB was demonstrated, as PDK1 was found to selectively phosphorylate p70s6k at Thr229. More importantly, PDK1 activated p70s6k in vitro and in vivo, whereas the catalytically inactive PDK1 blocked insulin-induced activation of p70s6k.

The p70 ribosomal protein S6 kinase, p70s6k, participates in the translational control of mRNA transcripts that contain a polypyrimidine tract at their transcriptional start site (1). Although these transcripts represent only 100 to 200 genes, they can encode up to 20% of the cell's mRNA (2), with most of these transcripts encoding components of the translational apparatus (2). The p70s6k has been tentatively identified as a downstream effector of the phosphoinositide 3-kinase (PI3K) signaling pathway, but the upstream kinases linking PI3K with p70s6k have not been identified (3). In addition, p70s6k activity is controlled by the mammalian target of rapamycin (mTOR), which appears to protect the kinase from a phosphatase (4). Activation of p70s6k is associated with increased phosphorylation of eight residues (Fig.1A) (4). Of these, three appear to be essential in regulating kinase activation, including Thr229 in the catalytic domain as well as Ser371 and Thr389 in the linker domain (Fig.1A) (4, 5). All three sites are conserved in many members of the AGC (protein kinases A, G, and C) family of Ser/Thr kinases, including PKB and PKC. The sequence surrounding Thr229 is also highly conserved in the Ca2+- and calmodulin-dependent protein kinase family (CaMK), the closest neighbor to that of the AGC family (Fig. 1A) (6). Recently, it was demonstrated that a newly described kinase, termed PDK1, phosphorylates the equivalent site to Thr229 in PKB, Thr308 (7). Further studies have indicated that although PDK1 is constitutively active, its ability to phosphorylate Thr308 is blocked by the NH2-terminal pleckstrin homology (PH) domain of PKB (7). Upon mitogen stimulation, PI3K increases the level of phosphatidylinositide-3,4,5-P3, which binds to the PH domain of PKB, presumably disrupting its interaction with Thr308 (8). The in vivo kinase that phosphorylates Thr229 in the catalytic domain of p70s6k is also thought to be constitutively active, but requires prior phosphorylation of Thr389, in the linker region of p70s6k (Fig. 1A), to bring about kinase activation (9).

Figure 1

Site-specific phosphorylation of p70s6k by PDK1. (A) Schematic diagram of p70s6k showing the catalytic domain (white), linker domain (hatched), and autoinhibitory domain (black). Known phosphorylation sites are indicated. The sequence (20) surrounding the activation loop site, Thr229, is compared with those from other kinases (6), and identities are boxed. The four (Ser/Thr)-Pro sites, substituted with acidics in the D3E series, are noted. (B) HA-ΔPHPKB (11), Myc-PDK1, and Myc-PDK1-KD (10) were independently expressed in 293 cells (12). Total cell extracts expressing ΔPHPKB (40 μg), PDK1 (2.5 μg), or PDK1-KD (2.5 μg) were mixed (12), and the immunoprecipitates were incubated with 10 μCi [γ32P]adenosine triphosphate (ATP) (25 μM, 45 min, 30°C). The 32P-labeled proteins were resolved by SDS–polyacrylamide gel electrophoresis (PAGE) and visualized by phosphorimagery. Arrowheads denote the positions of Myc-PDK1 and HA-ΔPHPKB. (C) Myc-p70s6k constructs were expressed in 293 cells, and extracts (40 μg) were mixed with PDK1 or PDK1-KD (2.5 μg) and treated as in (B) after immunoprecipitation. The amount of each kinase used was equal by immunoblot analysis (40 μg, upper); the position of phosphorylated p70s6kis indicated (lower). WT, wild type. (D) Myc-p70s6kE389D3E (left panel) or Myc-p70s6kS229,E389D3E (15) (left panel, inset) were immunoprecipitated with PDK1 and treated as in (B). The 32P-labeled p70s6k was digested in the gel, and the released phosphopeptides were mapped as in (9). The major phosphopeptide from each map was eluted from the cellulose matrix for phospho–amino acid analysis (right panel). The positions of the standards (pThr and pSer), free phosphate (Pi), and origin (Ori) are indicated by arrowheads, and the activation-loop residue for each construct is indicated under the panel.

To investigate whether PDK1 phosphorylates Thr229in p70s6k, Myc-tagged PDK1 (10) and Myc-p70s6k were expressed in 293 cells. As a control, PDK1 activity was assessed with an influenza hemagglutinin-tagged PKB variant lacking the PH domain, termed HA-ΔPHPKB (11), which is a substrate for PDK1 in the absence of phospholipids (8). Myc-PDK1 phosphorylated HA-ΔPHPKB, whereas the catalytically inactive variant of PDK1, Myc-PDK1-KD, did not (Fig. 1B) (12). Incubation of Myc-PDK1 with Myc-p70s6k led to only a small increase in the amount of phosphate incorporated into Myc-p70s6k (Fig. 1C). The phosphorylation of Thr229 appears to depend on phosphorylation of Thr389 (13), and this reaction is facilitated by converting four (Ser/Thr)-Pro phosphorylation sites in the autoinhibitory domain to acidic residues (Fig. 1A) (14). Therefore, we tested whether a p70s6k variant with acidic residues placed at Thr389 and at the four (Ser/Thr)-Pro sites (Myc-p70s6kE389D3E) (14) would serve as a better substrate for PDK1. The Myc-p70s6kE389D3E protein was more efficiently phosphorylated by PDK1 than the wild-type enzyme or the Myc-p70s6kD3E variant, in which only the (Ser/Thr)-Pro sites are converted to acidic residues (Fig. 1C). Myc-PDK1-KD did not phosphorylate Myc-p70s6kE389D3E (Fig. 1C). To verify the site of phosphorylation, the Myc-p70s6kE389D3E variant and a mutant in which Thr229 was converted to serine (15) were phosphorylated in vitro by Myc-PDK1 and subjected to two-dimensional phosphopeptide mapping (9). In both cases, a single major phosphopeptide was observed (Fig. 1D) that migrated at the identical position. Amino acid analysis of the two phosphopeptides revealed phosphothreonine in the peptide from the Myc-p70s6kE389D3E mutant and phosphoserine in the peptide from the Ser229 mutant (Fig. 1D). Thus, PDK1 selectively phosphorylates p70s6k at position Thr229.

Only the active form of PDK1 phosphorylates and activates HA-ΔPHPKB in vitro (Fig. 2A) (8). In the case of Myc-p70s6k and Myc-p70s6kD3E, the small increase in phosphorylation catalyzed by Myc-PDK1 had no detectable effect on activity (Fig. 2B). This finding is consistent with the inability of Myc-PDK1 to phosphorylate Thr389, which appears to be an absolute requirement for p70s6k activation (9, 14). Although Myc-p70s6kE389D3E has high basal kinase activity, Myc-PDK1 further increased the activity of this variant (Fig. 2B). The catalytically inactive Myc-PDK1-KD did not activate Myc-p70s6kE389D3E (Fig. 2B). These results support the hypothesis that phosphorylation of Thr389 is a prerequisite for phosphorylation of Thr229.

Figure 2

Activation of p70s6k and PKB by PDK1 in vitro. Equal amounts of either (A) HA-ΔPHPKB or (B) Myc-tagged variants of p70s6k were immunoprecipitated with either Myc-PDK1 or Myc–PDK1-KD as in Fig. 1B. The ΔPHPKB and p70s6k were activated by PDK1 by incubation with 200 μM MgATP. The PKB activity was measured with Crosstide (19) by filter paper assay (A), and p70s6k activity, with 40S ribosomal subunits (9) (B). Phosphorylated S6 was resolved by SDS-PAGE and visualized by phosphorimagery. Each p70s6kvariant used and phosphorylated S6 are indicated. The results are representative of two independent experiments.

The threonine residues in the catalytic domains of p70s6kand PKB appear to be selectively phosphorylated by PDK1. Members of CaMK family of protein kinases also have similar amino acid sequences in their catalytic domains (Fig. 1A) (6). The catalytic domain kinase for a member of the CaMK family, CaMK IV, has been cloned and termed CaMK kinase (16), allowing the possibility of testing the two kinases against alternative substrates. Purified Myc-PDK1 did not phosphorylate CaMK IV (Fig.3A). Similarly, CaMK kinase did not appreciably phosphorylate either purified glutathione S-transferase (GST)–tagged p70s6kE389D3E or HA-ΔPHPKB (Fig. 3A). Thus, the upstream kinases for the catalytic domains of the AGC and CaMK protein kinase families are apparently selective for their respective family members.

Figure 3

Specificity of catalytic domain kinases. (A) After immunoprecipitation, HA-ΔPHPKB and Myc-PDK1 were eluted from immune complexes with HA and Myc epitope peptides, respectively, in buffer B. Myc-p70s6kE389D3E-GST, from transfected 293 cells, was purified on glutathione-Sepharose and then eluted with 10 mM glutathione in buffer B (12). The purity and concentration of each kinase were verified by SDS-PAGE and Coomassie staining. Equal amounts (200 ng) of purified, soluble HA-ΔPHPKB, Myc-p70s6kE389D3E-GST, or CaMK IV were incubated with soluble Myc-PDK1 or CaMK kinase (18) and Mg[γ-32P]ATP as in Fig. 1B. Phosphorylated products were resolved by SDS-PAGE and visualized by phosphorimagery. Each substrate is indicated above and the phosphorylating kinase is indicated below the panel. The asterisk indicates the position of autophosphorylated Myc-PDK1. The results are representative of two independent experiments. (B) Myc-PDK1 was expressed in 293 cells and then either extracted immediately, stimulated with insulin (1 μM, 10 min) and extracted, or treated (30 min) with wortmannin (250 nM) or rapamycin (20 nM) and extracted. After immunoprecipitation, Myc-PDK1 (expression, top panel) activity was measured with purified Myc-p70s6kE389D3E-GST (200 ng) as in (A). The 32P-labeled Myc-p70s6kE389D3E-GST was visualized after SDS-PAGE and phosphorimagery (bottom panel). The results are representative of two independent experiments.

Earlier studies indicated that the Thr229 kinase is constitutively active in quiescent cells and resistant to wortmannin and rapamycin, agents known to block p70s6kphosphorylation and activation (9). Consistent with this result, the activity of PDK1 has been reported to be constitutive (8). We therefore immunoprecipitated Myc-PDK1 from quiescent or insulin-stimulated 293 cells treated with wortmannin or rapamycin. Myc-PDK1 was then tested for its ability to phosphorylate Myc-p70s6kE389D3E in vitro. The activity of PDK1 was unaffected by insulin stimulation, and treatment of cells with either wortmannin or rapamycin did not inhibit PDK1 activity (Fig. 3B). Thus, PDK1 is constitutively active and insensitive to agents known to block insulin-induced p70s6k activation.

The observations above suggest that PDK1 is the kinase responsible for regulating p70s6kE389D3E activation in vivo. Expression of Myc-PDK1 slightly enhanced basal activity of a Myc-p70s6k-GST reporter, but potently activated Myc-p70s6kE389D3E- GST, an effect that was not further augmented by insulin (Fig.4A). In contrast, expression of catalytically inactive Myc-PDK1-KD blocked insulin-induced activation of Myc-p70s6kE389D3E-GST (Fig. 4A) as well as Myc-p70s6k-GST (13). The specificity of PDK1 was further supported by the finding that co-expression of PDK1 with the mitogen-activated protein kinase (p44mapk) had no effect on either basal or 12-O-tetradecanoylphorbol 13-acetate (TPA)–induced p44mapk (Fig. 4B). In addition, Myc-PDK1 activation of Myc-p70s6kE389D3E was not inhibited by wortmannin or rapamycin (Fig. 4C). Co-expression of Myc-PDK1 also increased HA-PKB activity; however, the effect on ΔPHPKB activity was much larger (Fig. 4D). Thus, PDK1 can increase the activity of p70s6k and PKB in vivo, if the regulatory constraints that would normally block access to the catalytic domain phosphorylation site are removed.

Figure 4

PDK1 activation of p70s6k is resistant to treatment with wortmannin and rapamycin. The results are representative of at least two independent transfections. (A) Myc-p70s6k-GST or Myc-p70s6kE389D3E-GST activity (1) after co-expression with or without Myc-PDK1 (left panel) or Myc-PDK1-KD (right panel), after treatment with or without insulin (1 μM, 30 min). Expression of reporter p70s6k and PDK1 was determined by immunoblot (40 μg, upper panels), and reporter activity was assessed by phosphorylation of S6 after precipitation with glutathione-Sepharose (lower panels). (B) HA-p44mapk activity was measured as a function of Myc-PDK1 transfection and TPA (10 μM) treatment. Expression of p44mapk reporter and PDK1 was determined as in (A) (upper panel), and reporter activity was assessed by in vitro phosphorylation of myelin basic protein (MBP) after immunoprecipitation (lower panel). (C) Resting 293 cells expressing Myc-p70s6kE389D3E-GST alone or with Myc-PDK1 were either extracted directly or stimulated with insulin (1 μM) with or without pretreatment (30 min) with rapamycin (20 nM) or wortmannin (250 nM). Expression of Myc-p70s6kE389D3E-GST and Myc-PDK1 was determined (upper panel), and reporter p70s6k activity was assessed by S6 phosphorylation (lower panel) as in (A). (D) The effect of insulin on HA-PKB and HA-ΔPHPKB activity, with or without PDK1 co-expression, was measured with Crosstide (19); expression of each construct was observed to be equal by immunoblot analysis (data not shown). (E) A model of p70s6k activation predicts Thr229phosphorylation by PDK1 after prior phosphorylation of Thr389. Tentative placement of other potential components in the activation scheme are indicated with a dotted line. PPase, phosphatase.

The results presented here are consistent with PDK1 as the in vivo kinase responsible for mediating Thr229phosphorylation in the catalytic domain of p70s6k. The kinases responsible for mediating activation of p70s6k have been difficult to identify because of the multiple and hierarchical regulatory steps required to bring about its activation (4, 9, 14). Initially, it had been suggested that Thr229 phosphorylation was regulated by a kinase that was activated directly or indirectly by PI3K in a wortmannin-sensitive manner (17). However, our studies indicate that the Thr229 kinase is constitutively active, wortmannin-resistant (9), and dependent on prior phosphorylation of Thr389 to provide access to Thr229 (13). These latter requirements are fulfilled by PDK1 (Fig. 4E) (4, 9). Activation of p70s6k appears to be first mediated by phosphorylation of the (Ser/Thr)-Pro sites in the autoinhibitory domain, which facilitates phosphorylation at Thr389 by disrupting the interaction of the COOH- and NH2-termini of the kinase, thereby allowing phosphorylation of Thr229 (Fig. 4E) (4,9). A key step in this process is Thr389phosphorylation, which appears to be positively regulated by a wortmannin-sensitive, PI3K-dependent input, possibly through PKB, and is suppressed by a rapamycin-activated Thr389phosphatase (9), through inhibition of mTOR (Fig. 4E). Many members of the AGC family of Ser/Thr kinases share the same conserved catalytic domain of p70s6k and PKB (Fig. 1A) (6), suggesting that PDK1 may be a member of a family of kinases that mediate activation-loop phosphorylation of AGC protein kinases. Consistent with this possibility, we have identified a number of PDK1-like cDNAs (18).

  • * These authors contributed equally to this work.


Stay Connected to Science

Navigate This Article