Protein Kinase B Kinases That Mediate Phosphatidylinositol 3,4,5-Trisphosphate-Dependent Activation of Protein Kinase B

See allHide authors and affiliations

Science  30 Jan 1998:
Vol. 279, Issue 5351, pp. 710-714
DOI: 10.1126/science.279.5351.710


Protein kinase B (PKB) is activated in response to phosphoinositide 3-kinases and their lipid products phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3] and PtdIns(3,4)P2in the signaling pathways used by a wide variety of growth factors, antigens, and inflammatory stimuli. PKB is a direct target of these lipids, but this regulation is complex. The lipids can bind to the pleckstrin homologous domain of PKB, causing its translocation to the membrane, and also enable upstream, Thr308-directed kinases to phosphorylate and activate PKB. Four isoforms of these PKB kinases were purified from sheep brain. They bound PtdIns(3,4,5)P3and associated with lipid vesicles containing it. These kinases contain an NH2-terminal catalytic domain and a COOH-terminal pleckstrin homologous domain, and their heterologous expression augments receptor activation of PKB, which suggests they are the primary signal transducers that enable PtdIns(3,4,5)P3 or PtdIns- (3,4)P2 to activate PKB and hence to control signaling pathways regulating cell survival, glucose uptake, and glycogen metabolism.

Phosphoinositide 3-kinases (PI3Ks) are a diverse family of enzymes capable of 3-phosphorylating inositol phospholipids (1). One subfamily can be activated by receptors through various signal transduction mechanisms. These enzymes phosphorylate PtdIns, PtdIns(4)P, and PtdIns(4,5)P2in vitro and apparently preferentially phosphorylate the latter in vivo, yielding PtdIns(3,4,5)P3 (2, 3). As a result many receptors can stimulate accumulation of PtdIns(3,4,5)P3. PtdIns- (3,4)P2 can often be detected under similar circumstances. The existence of PtdIns- (3,4,5)P3 5-phosphatases in many cells, combined with the (sometimes) relatively slower accumulation of PtdIns(3,4)P2than of PtdIns(3,4,5)P3, suggests that at least some of the PtdIns(3,4)P2 is derived from PtdIns(3,4,5)P3(4); however, some is probably generated directly by hormone-sensitive lipid kinases.

PtdIns(3,4,5)P3 and PtdIns(3,4)P2 are potential participants in intracellular signaling pathways but the primary target or targets of these lipids are not clear. However, several signal transducers are activated as a consequence of PI3K activity, including PKB (5, 6), protein kinase Cs (PKCs) (7-9), and the small GTP-hydrolyzing protein rac (10). Specific PtdIns(3,4,5)P3 and PtdIns(3,4)P2 binding proteins have been identified and a subgroup of pleckstrin homologous (PH) domains may be specialized for binding these lipids (11). However, binding of 3-phosphorylated lipids to the PH domain in PKB cannot alone account for regulation of PKB (12). PKB can be phosphorylated on Thr308 and Ser473 in a PI3K-dependent fashion, and one or more upstream kinases phosphorylate and contribute to the activation of PKB (13). A protein kinase or kinases that phosphorylate Thr308 only when PKB is bound to PtdIns(3,4,5)P2 or PtdIns(3,4)P2 has been partially purified (14, 15); the upstream kinases are themselves activated by PtdIns- (3,4,5)P3, implying a complex two-level regulation by PtdIns(3,4,5)P3(14). We describe experiments aimed at characterizing PtdIns(3,4,5)P3- and PtdIns(3,4)P2-sensitive PKB kinases and the mechanism by which they are regulated.

[32P]PtdIns(3,4,5)P3 binding protein or proteins copurify with PKB kinase activity and four distinct forms of PKB kinase can be resolved (Fig. 1). All four activities phosphorylate and activate phosphorylation of myelin basic protein (MBP) by PKB in the presence of the biological stereoisomers of PtdIns(3,4,5)P3 or PtdIns(3,4)P2 {for example, (1-stearoyl 2-arachidonyl)-sn-phosphatidyl-d-myo-inositol 3,4,5-trisphosphate [D-D-S/A-PtdIns(3,4,5)P3]} (16) (Fig. 2A). The purified kinases are (i) inactive against PKB in the presence of the enantiomers of these lipids or PtdIns(4,5)P2; (ii) active at lower concentrations of stearoyl arachidonyl than dipalmitoyl versions of these lipids; and (iii) equally effectively activated by D-D-S/A-PtdIns(3,4,5)P3 and its diastereoisomer differing only in the arrangement of the chiral center in the glycerol backbone [D-L-S/A-PtdIns(3,4,5)P3], which suggests that, although interaction depends on the nature of the fatty acids, it is not dictated by their precise stereochemistry with respect to the water-soluble headgroup (Fig. 2B).

Figure 1

Purification of PKB kinases from sheep brain. (A) Flow diagram for purification of PKB kinases A to D from sheep brain cytosol (27) and records for the quantity of protein carried through each step. The overall recovery of activity from the initial cytosol fractions was 16%. Further details can be found at the Science Website ( (B) Analysis of the active fractions eluting from the final SEC. The native sizes were estimated to be 58, 58, 68, and 54 kD and their SDS-denatured sizes were estimated to be 57, 57, 70, and 55 kD (the positions to which 220-, 97-, 69-, 46-, and 31-kD standards had migrated during SDS-PAGE are indicated). (C) Copurification of PKB kinase activity with a [32P]PtdIns(3,4,5)P3 binding protein. A partially purified preparation of PKB kinase was applied to a SEC (28) and fractions were analyzed for PtdIns(3,4,5)P3-dependent PKB kinase activity [bottom;32P counts per second (cps) in PKB] (25), [32P]PtdIns(3,4,5)P3 binding {middle, a renatured immunoblot was incubated with [32P]PtdIns- (3,4,5)P3} (29), or total proteins [by silver staining of an SDS-polyacrylamide gel (top)].

Figure 2

Characterization of PKB kinases A to D. (A) PtdIns(3,4,5)P3-dependent activation of PKB. Assays were run in two stages. The first stage was run with mixed lipid vesicles either with or without D-D-S/A-PtdIns(3,4,5)P3 (final concentration, 5 μM) and in the presence or absence of PKB kinase A (6 nM), wild-type (EE)-PKB (2.5 μM), or Thr308 to Ala–Ser473 to Ala-(EE)-PKB and [γ-32P]ATP (adenosine triphosphate) (50 μM). Percentage of PKB phosphorylated is indicated by stippled bars. The assays were stopped and PKB proteins were immunoprecipitated with anti-(EE) beads, washed, and then incubated with [γ-32P]ATP (10 μM) and MBP (7 μM) to determine the activity of the immobilized PKB (hatched bars). Results are from a single experiment and are represented as means ± SE (n = 3 to 5); four other experiments gave similar results. PKB kinases B, C, and D gave very similar results. (B) Phospholipid specificity of activation of PKB phosphorylation. Assay mixtures contained a constant concentration of mixed lipid vesicles (27) with the indicated concentrations of inositol phospholipids, (EE)-PKB (2.5 μM), PKB kinase A (5 nM), and [γ-32P]ATP (1 μM; total volume, 12 μl). The data shown are pooled from 12 separate experiments and are means (n = 3 to 6; average SE was 6%). Identical patterns of activation were observed for PKB kinases B, C, and D. Open circles, L-L-S/A-PtdIns(3,4,5)P3; open squares, L-D-S/A-PtdIns-(3,4,5)P3; solid squares, D-L-S/A-PtdIns(3,4,5)P3; solid circles, D-D-S/A-PtdIns(3,4,5)P3; open diamonds, L-L-P/P-PtdIns(3,4)P2; open triangles, PtdIns(4,5)P2 (brain); solid diamonds, D-D-P/P-PtdIns(3,4)P2; solid triangles, D-D-P/P-PtdIns(3,4,5)P3. The 32P incorporated into PKB in the presence of 4 μM D-D-S/A-PtdIns(3,4,5)P3was defined as 100%. (C) Association of PKB with lipid vesicles. (EE)-PKB (40 nM) was incubated with sucrose-loaded lipid vesicles (or their vehicle) under conditions similar to those in (B). After 4 min the vesicles were sedimented by centrifugation and the quantities of PKB in the supernatants and pellets were quantitated by immunoblotting with anti-(EE). Inset immunoblots show results of an experiment with D-D-S/A-PtdIns(3,4,5)P3 (maximum concentration in assay 1, 16 μM). The data shown are pooled from a total of seven independent experiments and represent means (n = 2 to 3; average range about those means was 11.0%). Solid squares, D-PtdIns(4,5)P2; open diamonds, L-L-P/P-PtdIns(3,4)P2; open circles, L-L-S/A-PtdIns(3,4,5)P3; solid diamonds, D-D-P/P-PtdIns(3,4)P2; solid triangles, D-D-P/P-PtdIns(3,4,5)P3; solid circles, D-D-S/A-PtdIns(3,4,5)P3. (D) Association of PKB kinases with lipid vesicles. PKB kinase A (5 nM) was mixed with sucrose-loaded lipid vesicles (or their vehicle) containing various concentrations of inositol phospholipids as shown (30). After 4 min at 30°C the vesicles were collected by centrifugation and portions of the supernatants were assayed for PKB kinase activity in the presence of 5 μM D-D-S/A-PtdIns(3,4,5)P3. A mean of 8.2% of the total activity was sedimented in the presence of lipid vesicles containing no added inositol phospholipids; the activity remaining in the supernatant after centrifugation of lipid vesicles containing no added inositol lipids defined the 100% value to which other treatments were compared. Data shown are means (n = 4 to 6, pooled from 16 separate experiments; average SE was 8%). Solid squares, D-PtdIns(4,5)P2; open diamonds, L-L-P/P-PtdIns(3,4)P2; open circles, L-L-S/A-PtdIns(3,4,5)-P3; solid diamonds, D-D-P/P-PtdIns(3,4)P2; solid triangles, D-D-P/P- PtdIns(3,4,5)-P3; solid circles, D-D-S/A-PtdIns(3,4,5)P3.

With the same assay conditions used for the PKB phosphorylation studies, we investigated the association of both the PKB kinases (Fig. 2D) and PKB itself (Fig. 2C) with the lipid vesicles in these assays. The PKB kinases associated with the lipid vesicles containing very low molar percentages of the PtdIns(3,4,5)P3 stereoisomers [0.003% for D-D-S/A-PtdIns(3,4,5)P3] and D-D-P/P-PtdIns(3,4)P2 but not PtdIns(4,5)P2 or L-L-P/P-PtdIns(3,4)P2. This is consistent with the observations that the PKB kinase or kinases can bind [32P]PtdIns(3,4,5)P3 and that phosphorylation of a water-soluble, 30-mer peptide, based on the sequence of PKB around Thr308, was inhibited by PtdIns-(3,4,5)P3 or PtdIns(3,4)P2(17).

Under the same assay conditions PKB also associated with the lipid vesicles (Fig. 2C). However, substantially higher concentrations of 3-phosphorylated lipids were required to detect translocation of PKB; the lipid specificity of this event was distinct from that for translocation of the PKB kinases but very similar to that for activation of phosphorylation of PKB (Fig. 2, B and C). Thus, the lipid binding properties of the two kinases are quite distinct and the specificity of the phosphorylation is largely dictated by the recruitment of PKB (that is, the affinity and specificity of the PH domain of PKB). Probably only a relatively small proportion of the PKB is associated with the vesicles at concentrations of active lipids that have caused most of the PKB kinase to associate. Hence, the biggest effects of the translocation of PKB kinase on the phosphorylation of PKB will occur at very low concentrations of PtdIns(3,4,5)P3, whereas at higher concentrations the translocation of PKB may become the major factor. These properties explain why membrane-targeting effectively activates PKB (18); presumably there is adequate basal turnover of 3-phosphorylated lipid in these cells to ensure there is some PKB kinase already in place.

A preparation of PKB kinase A was blotted onto nitrocellulose and treated with trypsin in situ. The liberated peptides were subjected to analysis by NH2-terminal sequencing and mass spectrometry (19). Four peptides were defined and used to search the databases and a family of human expressed sequence tag (EST) sequences were identified. Combined use of the peptide sequences (to fix the reading frame) and information from further sequencing of the EST clones and of cDNAs isolated from a human U937 cell cDNA library (20) defined a cDNA with a potential open reading frame (ORF) containing all four peptides (allowing for species differences) and encoding a protein with a predicted molecular size of 63 kD (Fig.3).This predicted protein contains an NH2-terminal protein kinase domain and a COOH-terminal PH domain (Fig. 3). We have also isolated cDNAs from a rat brain library that encode a protein with 94% identity at the amino acid level of this ORF (EMBL accession number Y15748) and have identified a closely related sequence in the database from Drosophila(EMBL accession number Y07908) (21).

Figure 3

Primary structure of a PKB kinase (31). A potential ORF defined by cDNAs isolated from our human U937 cell library is shown (EMBL accession number Y15056). The four peptide sequences derived from the 57-kD sheep brain PKB kinase A are shown in boldface above the sequence. The area of similarity to other protein kinase catalytic domains is boxed in a solid line; the area of similarity to other PH domains is boxed in a dashed line.

Genomic sequence information in the databases regarding the PKB kinase gene allowed us to define a precise chromosomal localization (human chromosome 16p13.3) (22) and some, but not all, intron-exon boundaries. It is clear from this information and from sequencing a number of cDNAs that this locus gives rise to a complex pattern of alternatively and apparently incompletely spliced transcripts. One human transcript is precisely equivalent to the ORF defined in Fig. 3 except that it is missing the exon encoding the substrate recognition motif of the protein kinase domain (residues 238 to 263; peptide sequence data showed that our purified enzyme contained this motif) (Fig. 3). We used this cDNA to generate a catalytically inactive version of the PKB kinase.

We have constructed mammalian expression vectors encoding NH2-terminally tagged PKB kinase and its catalytically inactive variant (23). EE-tagged versions of these proteins were expressed in and purified from COS-7 cells (19) (Fig.4). The protein with the intact protein kinase domain phosphorylated PKB in a PtdIns(3,4,5)P3-sensitive manner, indicating that this activity resided in the ORF defined in Fig. 3.

Figure 4

Expression of PKB kinases in COS-7 and aortic endothelial cells. (A) Mammalian expression vectors containing NH2-terminal EE-tagged versions of the PKB kinase (23) [(EE)-II refers to the complete ORF and (EE)-I refers to the kinase-inactive splice variant] were transiently expressed in COS-7 cells (24). Proteins were purified via their (EE)-tags (antibodies to myc were used as controls) and were detected by immunoblotting. Polyvinylidene difluoride filters, probed with anti-(EE) (detection by enhanced chemiluminescence; right) were then stained with Coomassie blue [left; similar results were obtained with (EE)-I]. Samples were assayed for PKB kinase activity in the presence of lipid vesicles either with or without D-D-S/A-PtdIns(3,4,5)P3 and [γ-32P]ATP (300 nM; 3 μM and 1 μM final concentrations, respectively). Autoradiogram shows 32P in PKB. IP, immunoprecipitated with; PIP3, PtdIns(3,4,5)P3. (B) PAE cells were cotransfected with (EE)-PKB and either (myc)-PKB kinase (23) or an irrelevant DNA. After 12 hours the cells were transferred into serum-free medium and after a further 12 hours some were stimulated with PDGF for 45 s. The activity of PKB in anti-(EE) immunoprecipitates prepared from lysates of the transfected cells was quantitated with MBP as a substrate (6); data are presented from a single experiment (means ± SE; n = 4) representative of four.

Coexpression of (EE)-PKB with (myc)-ERK-2 or (myc)-PKB kinase in porcine aortic endothelial (PAE) cells resulted in an increase in both the basal and platelet-derived growth factor (PDGF)–stimulated activity of PKB but had no effect on the activity of ERK-2 (24), despite the fact that ERK-2 is activated by PDGF in a wortmannin-sensitive fashion in these cells (25). In the presence of the transfected upstream kinase, the fold increase in PKB activity in response to PDGF fell; however, the difference between control and PDGF-stimulated PKB activities increased by about threefold (Fig. 4). This suggests that the PKB kinase we have purified and cloned can participate in the pathway by which PDGF activates PKB. The increase in the basal PKB activity in the presence of PKB kinase is most simply explained by our observation that the upstream kinase appears to be very sensitive to PtdIns(3,4,5)P3. Presumably, under the conditions of our experiments the cells are still sufficiently basally activated that during long periods of exposure (the 24-hour transfection protocol) to very high levels of PKB kinase, significant amounts of the heterologous PKB can become phosphorylated and hence active.

Our results indicate that PtdIns(3,4,5)P3 and PtdIns(3,4)P2 activate PKB by causing a translocation to the membrane of PKB itself and an upstream Thr308-directed protein kinase. These results confirm the function of PH domains in propagation of the PI3K signal in cells (11, 26) and suggest that positional information is of particular importance in this pathway.

  • * To whom correspondence should be addressed. E-mail: len.stephens{at}


Stay Connected to Science

Navigate This Article