Direct Regulation of the Akt Proto-Oncogene Product by Phosphatidylinositol-3,4-bisphosphate

See allHide authors and affiliations

Science  31 Jan 1997:
Vol. 275, Issue 5300, pp. 665-668
DOI: 10.1126/science.275.5300.665


The regulation of the serine-threonine kinase Akt by lipid products of phosphoinositide 3-kinase (PI 3-kinase) was investigated. Akt activity was found to correlate with the amount of phosphatidylinositol-3,4-bisphosphate (PtdIns-3,4-P2) in vivo, and synthetic PtdIns-3,4-P2 activated Akt both in vitro and in vivo. Binding of PtdIns-3,4-P2 occurred within the Akt pleckstrin homology (PH) domain and facilitated dimerization of Akt. Akt mutated in the PH domain was not activated by PI 3-kinase in vivo or by PtdIns-3,4-P2 in vitro, and it was impaired in binding to PtdIns-3,4-P2. Examination of the binding to other phosphoinositides revealed that they bound to the Akt PH domain with much lower affinity than did PtdIns-3,4-P2 and failed to increase Akt activity. Thus, Akt is apparently regulated by the direct interaction of PtdIns-3,4-P2 with the Akt PH domain.

Stimulation of cells by several growth factors activates PI 3-kinase (1). In vivo, the activation of PI 3-kinase increases the intracellular amounts of PtdIns-3,4-P2 and phosphatidylinositol-3,4,5-trisphosphate (PtdIns-3,4,5-P3) (1). Various downstream targets of PI 3-kinase have been identified, including the serine-threonine kinase Akt (15). Akt (also referred to as PKBα or Racα) is encoded by the Akt proto-oncogene and is defined by an NH2-terminal regulatory domain of protein-protein interaction [Akt homology (AH) domain)] that contains a PH domain (2). Akt participates in the activation of the p70 ribosomal protein S6 kinase (p70S6K) (2) and inhibits glycogen synthase kinase-3 (3), and it has a role in proliferative and anti-apoptotic cell responses (4, 6). Akt activation by growth factors requires PI 3-kinase activity (2), but there are other pathways that can also lead to Akt activation (7).

To examine whether PI 3-kinase is sufficient to stimulate Akt activity, we coexpressed hemagglutinin epitope-tagged Akt (HA-Akt) and activated PI 3-kinase (8). Activated PI 3-kinase (piSH2-p110·MT) induced the activity of HA-Akt in serum-starved COS-1 and NIH 3T3 cells (Fig. 1), and stimulation of HA-Akt by activated PI 3-kinase was blocked by the PI 3-kinase inhibitor wortmannin (9). Activation of HA-Akt by piSH2-p110·MT in NIH 3T3 cells was enhanced by platelet-derived growth factor (PDGF) (Fig. 1B). Mutant piSH2-p110(K227E)·MT [in which Lys (K) at position 227 is mutated to Glu (E)] that is deficient in Ras binding (10) stimulated the activity of HA-Akt in serum-starved cells, but it was less effective than piSH2-p110·MT at enhancing PDGF stimulation of HA-Akt (9). The activity of HA-Akt(R25C), which contains a point mutation of Arg (R) to Cys (C) in the Akt PH domain, was not significantly increased by PDGF treatment or coexpression of piSH2-p110·MT (Fig. 1B). The Akt PH domain was therefore important for Akt activation by PDGF and by PI 3-kinase.

Fig. 1.

Increased Akt activity in cells transfected with activated PI 3-kinase. (A) In vitro kinase assays of Akt immunoprecipitated from lysates of serum-starved COS-1 cells expressing the indicated constructs (8). Immunoprecipitations with monoclonal antibody to HA (anti-HA) (Boehringer) were followed by kinase assays with histone H2B (3) (panel 1). Expression of HA-Akt and HA-Akt(R25C) was determined by protein immunoblotting with anti-HA (panel 2). p110·MT (panel 3) and piSH2·MT (panel 4) were immunoprecipitated with monoclonal antibody to MT (4) and detected with polyclonal antibody to 9E10 (EQKLISEEDL) (20). (B) Histone H2B kinase activity in anti-HA immunoprecipitates from lysates of serum-starved or serum-starved and PDGF-stimulated [PDGF (50 ng/ml) for 5 min] NIH 3T3 cells (PDGF-stimulated HA-Akt activity equals 100% in cells transfected only with HA-Akt).

We next examined the relation of Akt activity to phosphoinositide amounts in vivo. After treatment of human platelets with thrombin receptor-activating peptide (TRAP), the amounts of the PI 3-kinase products PtdIns-3,4-P2 and PtdIns-3,4,5-P3 increase with distinct times (11). We measured Akt autophosphorylation (Fig. 2A), Akt phosphorylation of histone H2B (Fig. 2B), and phosphorylation of phosphoinositides (Fig. 2B) as a function of time after the addition of TRAP (12). The concentration of PtdIns-3,4,5-P3 peaked 25 s after TRAP addition, but full Akt activation did not occur until the concentration of PtdIns-3,4-P2 had peaked after 60 s (Fig. 2B). Activation of Akt was blocked by wortmannin at concentrations that blocked TRAP-dependent increases in the concentrations of PtdIns-3,4-P2 and PtdIns-3,4,5-P3 (Fig. 2).

Fig. 2.

Correlation of Akt activity with the amount of cellular PtdIns-3,4-P2. (A) Akt autophosphorylation activity in human platelets is induced by 25 μM TRAP. The autophosphorylation activity of Akt was determined by immunoprecipitation (IP) and kinase assays. In the bottom panel, Akt was detected by protein immunoblot with antibody to Akt-CT. The results shown were reproduced in three independent experiments. (B) Temporal coincidence of Akt activity with increased PtdIns-3,4-P2 concentrations. Human platelets were incubated with 25 μM TRAP for various times. Akt activity toward histone H2B (black symbols) was measured after immunoprecipitation and is shown as percentage of maximum TRAP-stimulated activity. Black circles, no wortmannin; black squares, pretreatment with wortmannin. Amounts of [32P]PtdIns-3,4-P2 (open circle) and [32P]PtdIns-3,4,5-P3 (open square) are presented in 103 dpm. D3-phospholipids have a phosphate at the D3 position.

The time course suggested that PtdIns-3,4-P2 would be critical for in vivo activation of Akt. It has been shown that a product of PI 3-kinase activates immunoprecipitates of Akt (4), so we investigated the effects of synthetic phosphoinositides on purified Akt (13). We found that DiC16PtdIns-3,4-P2 stimulated Akt activity in a dose-dependent manner with a more than fourfold increase in activation at 10 μM (9), but neither DiC16PtdIns-3-P nor PtdIns-4,5-P2 increased Akt activity (Fig. 3), nor did PtdIns (9). DiC16PtdIns-3,4,5-P3 even inhibited Akt basal activity (Fig. 3). The Akt PH domain mutation R25C abrogated the ability of DiC16PtdIns-3,4-P2 to induce activity of Akt (Fig. 3).

Fig. 3.

Activation of Akt by DiC16PtdIns-3,4-P2 in vitro. Purified Akt and Akt(R25C) were incubated with phosphoinositides (10 μM). Akt and Akt(R25C) activities were determined by phosphorylation of histone H2B and are shown compared with carrier alone (0%) and Akt stimulated by PtdIns-3,4-P2 (100%). PI-3-P, PtdIns-3-P; PI-3,4-P2, PtdIns-3,4-P2; PI-4,5-P2, PtdIns-4,5-P2; PIP3, PtdIns-3,4,5-P3.

DiC16PtdIns-3,4,5-P3 can induce membrane ruffling and chemotaxis when added to whole cells (14), suggesting that phospholipid vesicles are able to fuse with the plasma membrane and that they can reach the inner leaflet to trigger biologic responses. When added to intact NIH 3T3 cells at a concentration of 5 μM (15), DiC16PtdIns-3,4-P2 caused more than a threefold increase in the stimulation of Akt autophosphorylation activity (Fig. 4A) that was dose-dependent (9). Other phosphoinositides did not activate Akt, and DiC16PtdIns-3,4,5-P3 even reduced Akt autophosphorylation activity (Fig. 4A). PI 3-kinase activity was not required because Akt activation by DiC16PtdIns-3,4-P2 occurred even after treatment of cells with wortmannin (Fig. 4A). PtdIns-4,5-P2 did not increase Akt activity in vivo (Fig. 4A). Thus, the lipid had similar effects on Akt when added to intact cells compared with those observed with purified Akt.

Fig. 4.

Activation of Akt by DiC16PtdIns-3,4-P2 in vivo. (A) DiC16PtdIns-3,4-P2 induces Akt activity in intact cells. Phosphoinositides (5 μM) were added to serum-starved NIH 3T3 cells for 10 min. Cells were pretreated with 100 nM wortmannin where indicated. The activity of immunoprecipitated Akt was determined by autophosphorylation. (B) Activation of Akt by DiC16PtdIns-3,4-P2 depends on the PH domain. Chinese hamster ovary cells were transfected with HA-Akt or HA-Akt(Δ11-35) expression constructs. After serum-starvation, the cells were incubated with DiC16PtdIns-3,4-P2 (5 μM) or PDGF (50 ng/ml) for 10 min, and Akt activity was determined by phosphorylation of histone H2B (panel 1). Panel 2 shows expression of HA-Akt and HA-Akt(Δ11-35).

We also examined the activity of HA-Akt after stimulation with PDGF and PtdIns-3,4-P2 (Fig. 4B). HA-Akt activity increased when HA-Akt was immunoprecipitated from cells treated with PDGF or DiC16PtdIns-3,4-P2 (Fig. 4B). The integrity of the Akt PH domain was necessary for Akt activation by DiC16PtdIns-3,4-P2. Mutant HA-Akt(R25C) (9) or HA-Akt(Δ11-35) (containing a deletion of amino acids 11 through 35) (Fig. 4B) was not activated in cells treated with PDGF or synthetic DiC16PtdIns-3,4-P2.

To determine whether PtdIns-3,4-P2 directly interacts with the Akt PH domain, we examined the binding of phosphoinositides to glutathione-S-transferase (GST)-Akt PH domain fusion protein (16). 32P-labeled phosphoinositides were incubated with GST-Akt PH protein, and weakly bound lipids were washed away (5). Both PtdIns-3,4-P2 and PtdIns-3,4,5-P3 bound to the Akt PH domain as determined by high-pressure liquid chromatography analysis of the bound lipids (9). [32P]PtdIns-4,5-P2 did not significantly bind (9), indicating that binding was stereospecific and required phosphate at the D3 position (Fig. 5A).

Fig. 5.

(A) Binding of PtdIns-3,4-P2 to the Akt AH (PH) domain. GST-Akt PH domain fusion protein was incubated with 32P-labeled D3-phosphoinositides and unlabeled phosphoinositides. Bound [32P]PtdIns-3,4-P2 ([32P]PI-3,4-P2]) was determined by thin-layer chromatography (TLC) after washing. The remaining [32P]PtdIns-3,4-P2 was normalized to the amount of [32P]PtdIns-3,4-P2 used in the incubation. (B) PtdIns-3,4-P2-binding to Akt AH (PH) domain protein is abolished by the R25C mutation. 32P-labeled D3-phosphoinositides were incubated with GST-Akt AH, Akt PH, Akt AH(R25C), or PH(R25C) domain fusion protein. The amount of bound phosphoinositide was determined by TLC. (C) DiC16PtdIns-3,4-P2 enhances Akt dimer formation. GST-Akt AH domain and GST-Akt AH(R25C) domain fusion protein were incubated with phosphoinositides and Akt protein. After stringent washing, the Akt proteins were detected by protein immunoblotting with antibody to 13 amino acids (KRQEEETMDFRSG) (20) in the Akt AH domain. The results shown were reproduced at least three times.

To determine relative affinities, we measured the binding of 32P-labeled phosphoinositides to the GST-Akt PH domain (amino acids 1 through 106) fusion protein in the presence of various concentrations of unlabeled phosphoinositides (17). DiC16PtdIns-3,4-P2 displaced binding of [32P]PtdIns-3,4-P2 to the Akt PH domain with 50% maximal effect at ∼5 μM. DiC16PtdIns-3,4,5-P3 competed for binding at higher concentrations, but PtdIns-4,5-P2 did not compete at concentrations up to 100 μM (Fig. 5A). Similar results were observed with displacement of [32P]PtdIns-3,4,5-P3 (9). Inositol-1,3,4-trisphosphate competed for binding of [32P] PtdIns-3,4-P2 with a >50% maximal effect at 100 μM (9), whereas inositol-1,4,5-trisphosphate and inositol-1,3,4,5-tetrakisphosphate did not compete at this concentration. Inositol-1,3,4-trisphosphate failed to specifically stimulate Akt activity at these concentrations (9).

Full-length Akt AH domain (amino acids 1 through 147) protein bound somewhat better to [32P]PtdIns-3,4-P2 than did the PH domain (amino acids 1 through 106) (Fig. 5B); a single point mutation (R25C), however, prevented binding to PtdIns-3,4-P2 (Fig. 5B). Thus, the Akt AH (PH) domain residue that was critical for Akt activation by PtdIns-3,4-P2 was important for binding.

We also tested whether binding of PtdIns-3,4-P2 facilitates Akt dimerization. The GST-Akt AH domain and GST-Akt AH(R25C) domain fusion proteins were incubated with Akt protein from cell lysates in the presence of phosphoinositides. After washing, the bound Akt was revealed by protein immunoblotting (Fig. 5C). Akt dimerization with GST-Akt AH increased more than threefold in the presence of 10 μM DiC16PtdIns-3,4-P2 (Fig. 5C), but was not observed with GST-Akt AH(R25C) (Fig. 5C). PtdIns-4,5-P2 and DiC16PtdIns-3,4,5-P3 failed to cause dimerization.

In vivo, the binding of PtdIns-3,4-P2, and to a lesser extent of PtdIns-3,4,5-P3, to Akt may transiently mobilize cytosolic Akt to the plasma membrane where it is activated by PtdIns-3,4-P2. After stimulation, the presence of PtdIns-3,4-P2 is no longer required, and immunoprecipitates of Akt remain activated after detergent wash. This persistent activation may be explained by phosphorylation that “locks” Akt into an active conformation (18). There are, however, PI 3-kinase-independent pathways that lead to Akt activation, indicating the presence of kinase cascades that bypass the PtdIns-3,4-P2 dependence (7).

PI 3-kinase has been implicated in various cellular processes. The quantities of the three lipid products of this enzyme are regulated by a complex set of kinases and phosphatases (1, 19). Here we show that PtdIns-3,4-P2 is the only known phosphoinositide that can activate Akt. Although DiC16PtdIns-3,4,5-P3 at high concentrations bound to the Akt AH (PH) domain, Akt dimerization and Akt activity were not increased by this lipid. Thus, the additional phosphate on PtdIns-3,4,5-P3 appears to interfere with dimerization. Structural studies are needed to understand the basis for these differences in lipid effects. Our studies further indicate that distinct products of PI 3-kinase have distinct functions and that Akt is likely to be an important mediator of some PI 3-kinase responses.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
View Abstract

Navigate This Article