Mitogenic Activation, Phosphorylation, and Nuclear Translocation of Protein Kinase Bβ*

Protein kinase B (PKB) is a member of the second messenger-dependent family of serine/threonine kinases that has been implicated in signaling pathways downstream of growth factor receptor tyrosine kinases and phosphatidylinositol 3-kinase. Here we report the characterization of the human β-isoform of PKB (PKBβ). PKBβ is ubiquitously expressed in a number of human tissues, with mRNA and protein levels elevated in heart, liver, skeletal muscle, and kidney. After transfection into HEK-293 or COS-1 cells, PKBβ is activated 2- to 12-fold by mitogens and survival factors. Activation was due to phosphorylation on Thr-309 and Ser-474, which correspond to Thr-308 and Ser-473 implicated in the regulation of PKBα. Both phosphorylation and activation were prevented by the phosphatidylinositol 3-kinase inhibitor wortmannin. Moreover, membrane-targeted PKBβ was constitutively activated when overexpressed in HEK-293 cells. Although the specific activity of PKBβ was lower than that of PKBα toward Crosstide as a substrate (23 nmol/min/mg compared with 178 nmol/min/mg for PKBα), both enzymes showed similar substrate specificities. Using confocal microscopy, we show that activation of PKBβ results in its nuclear translocation within 20 to 30 min after stimulation. These observations provide evidence that PKBβ undergoes nuclear translocation upon mitogenic activation and support a role for PKB in signaling from receptor tyrosine kinases to the nucleus through phosphatidylinositol 3-kinase.

Two further isoforms termed PKB␤ and -␥ (13)(14)(15), which display an overall structure similar to that of PKB␣, have also been cloned. Both isoforms consist of an amino-terminal PH domain (16) adjacent to the kinase catalytic domain and a short carboxyl-terminal Ser/Thr-rich regulatory domain. PKB␣ and -␤ are amplified and overexpressed in a number of ovarian (14), gastric (17), pancreatic (18) and breast carcinomas (19). Although the mechanism by which PKB contributes to neoplastic transformation and proliferation is still unclear, PKB␣ has been implicated in a number of cellular responses. In particular, PKB␣ phosphorylates and inhibits GSK-3 (6), lies upstream of p70 ribosomal S6 protein kinase (2) and stimulates glucose uptake in adipocytes by promoting GLUT4 translocation to the plasma membrane (20). Furthermore, PKB␣ is involved in the promotion of cell survival through inhibition of apoptosis (21), plays a role in PI3K-mediated neuronal cell survival (22), suppression of c-Myc induced apoptosis (23) and IL-2 signaling (24). Its ability to promote cell survival was shown to be directly proportional to kinase activity that modulates Ced3/ICE-like protease activity in fibroblasts (25).
Here we report the characterization of human PKB␤. We show that it is ubiquitously expressed, activated by mitogens and survival factors in a PI3K-dependent manner, and that activation is dependent upon phosphorylation on Thr-309 and Ser-474 (the corresponding residues in PKB␣ are Thr-308 and Ser-473). Furthermore, we show that PKB␤ and -␣ display similar substrate specificities and present evidence that activation of PKB␤ results in its translocation to the nucleus.
Construction of Expression Vectors-The cytomegalovirus-based expression construct encoding the human myristoylation/palmitoylation (m/p)-HA-PKB␣ has been described (38). Wild-type HA-PKB␤ was amplified by PCR, either from a human MCF-7 cell cDNA (13) or placenta cDNA 2 and subcloned as a KpnI/XbaI fragment into the pECE or pCMV4 expression vectors. The primers were designed according to the human PKB␤/Akt2 sequence (14). PKB␤ from human placenta carried a polymorphism (Thr-20 to Ala substitution) that did not affect its kinase activity compared with the MCF-7 cell line-derived PKB␤ (data not shown). The pCMV4 constructs encoding HA-T309A-PKB␤ and HA-T474A-PKB␤, respectively, were made by two-stage PCR (28) and subcloned as KpnI/XbaI fragments into pCMV4. The membrane targeting pECE-m/p-HA-PKB␤ construct was made by PCR to add 12 amino acids derived from the NH 2 terminus of the Lck tyrosine kinase 2 to the NH 2 -terminus of HA-PKB␤. The PCR fragment was subcloned as a KpnI/XbaI fragment into pECE. The correctness of the constructs was confirmed by restriction analysis and automated DNA sequencing. The sequences of all oligonucleotides used for modifications of PKB␤ are available on request.
Antibody Purification and Immunoblot Analysis-Synthetic peptides corresponding to the C-terminal domain of human PKB␣ (DQDDS-MECVDSERR) or -␤ (DRYDSLGLLELDQRTHF) were coupled to keyhole limpet hemocyanin and used to immunize rabbits (HTI, Bio-Prod-ucts, California). Antisera were collected and affinity-purified on 10 mg/ml of peptide coupled to Affi-Gel 15 (Bio-Rad) and eluted with 100 mM glycine-HCl, pH 2.5. Fractions were neutralized with 1 M Tris-HCl, pH 8.0, to give a final concentration of 50 mM Tris-HCl, pH 8.0. Human tissue extracts (CLONTECH), total cell extracts, and immunoprecipitates were resolved by 10% SDS-PAGE, and immunoblotted as described previously (8).
Immunoprecipitation and in Vitro Kinase Assays-Transiently transfected cells were lysed as described previously (7). HA-PKB␤ protein was immunoprecipitated with the monoclonal antibody 12CA5 coupled to protein A-Sepharose (Pharmacia Biotech Inc.). Immunocomplexes were washed three times in lysis buffer containing 0.5 M NaCl, once with lysis buffer, and once with kinase buffer (50 mM Tris-HCl, pH 7.4, 1 mM benzamidine, 0.5 mM phenylmethylsulfonyl fluoride). In vitro kinase assays were performed as described (8) using Crosstide (GRPRTSSFAEG) or other peptides (as stated in the text) as substrate. PKB␤ expression levels were quantified using FITC-labeled anti-mouse IgG (Sigma) as secondary antibody. Protein bands were visualized by chemifluorescence using a Storm 840/860 phosphorimager and quantified using ImageQuant software (Molecular Dynamics). 32 P-Labeling of HEK-293 Cells Transfected with HA-PKB␤-Transfected HEK-293 cells were cultured on 10-cm dishes, transfected with wild-type HA-PKB␤, washed with phosphate-free DMEM (ICN), incubated with [ 32 P]-orthophosphate (1 mCi/ml), stimulated with insulin or IGF-1 and lysed, then PKB␤ was immunoprecipitated and assayed as described previously (8). The 32 P-labeled HA-PKB␤ immunoprecipitates were washed, alkylated with 4-vinylpyridine (Sigma), electrophoresed, and digested with trypsin as described (8).  Immunofluorescence Staining and Laser Confocal Scanning Microscopy-HEK-293 cells were grown on glass coverslips to 80% confluency, washed with PBS, and fixed in 3.7% formaldehyde in PBS for 10 min at room temperature. Permeabilization of the cells was achieved through incubation with PBS containing 0.2% Triton X-100 for 10 min followed by a PBS wash. For detection of HA-PKB␤, cells were incubated for 1 h at 37°C with the 12CA5 antibody diluted 1:10 in PBS. After washing with PBS, cells were incubated for 30 min with goat anti-rabbit IgG-FITC conjugate (Sigma) diluted 1:50. The cells were washed with PBS followed by water and mounted in Moviol supplemented with 0.1% p-phenylenediamine. Laser confocal scanning images were obtained using a TCS 4D confocal system (Leica Instruments).

Expression of PKB␤ in Human
Tissue-To study the tissue distribution of PKB at the protein level, we developed peptidespecific antisera to both the PKB␤ and -␣ isoforms (Abs C511 and C520, respectively, see Fig. 1A). Specificity of the antibodies was established by expressing epitope-tagged cDNAs corresponding to the ␤ and ␣ isoforms of PKB. HA-PKB␤ and -␣ were transiently expressed in HEK-293 cells, immunoprecipitated with the HA-specific mAb 12CA5 and analyzed by 10% SDS-PAGE. Immunoblots probed with either Ab C511 or Ab C520 revealed a high degree of specificity for PKB␤ and -␣, respectively (Fig. 1B). The Ab C511 (␤-specific) detected a polypeptide with an apparent molecular mass of 60 kDa on immunoblots derived from HEK-293 total cell extracts transfected with HA-PKB␤ (Fig. 1C). This size was similar to that of the in vitro translated human PKB␤ (14). Western blot analysis of PKB␤ expression in human tissues (Fig. 1D) was consistent with the Northern blot data (data not shown): PKB␤ and -␣ were expressed in all tissues analyzed but were most abundant in heart, liver, skeletal muscle, and kidney. Expression of PKB␤ in brain, placenta, lung, and pancreas was significantly lower.
Regulation of PKB␤ Activity-To investigate the regulation of PKB␤, we determined its activity in response to different stimuli. For the experiments outlined below, we used a PKB␤ cDNA, isolated from a placenta cDNA library, which corresponds to the human PKB␤ sequence described previously (14). Transiently transfected and serum starved HEK-293 or COS-1 cells were treated with insulin, IGF-1, FCS, pervanadate, PDGF-BB, EGF, or TPA. HA-PKB␤ was immunoprecipitated and tested for in vitro kinase activity using Crosstide as substrate, which is a peptide derived from the NH 2 terminus of GSK-3 (6). These experiments (Fig. 2, A and B) revealed increased PKB␤ activity in response to every stimuli tested except TPA treatment. Pervanadate was the most potent activator (12-fold activation) in both cells lines. PDGF and EGF each caused a 6-fold activation of PKB␤ in COS-1 cells, while 5-and 7-fold activation occurred in response to insulin and IGF-1, respectively. FCS treatment caused a 2-fold increase above basal activity. To test whether PKB␤ activation was PI3K-de- pendent, we treated transiently transfected HEK-293 cells with 100 nM wortmannin for 15 min prior to stimulation. Similar to its effects on PKB␣, wortmannin abolished the activation of PKB␤ in response to insulin or IGF-1 stimulation (Fig. 2C).
Endogenous PKB␤ from pervanadate-treated HEK-293 cells recovered by immunoprecipitation with Ab C511 showed a 3-fold activation compared with unstimulated PKB␤, and recovery of PKB␤ was blocked by addition of the antigenic peptide (data not shown). Similarly, using a different PKB␤ antibody (raised against the carboxyl-terminal peptide CRYDSLGSLELDQRTH), IGF-1 stimulation of HEK-293 cells caused a 2-3-fold wortmannin-sensitive activation of PKB␤. The analysis of total cell extracts from quiescent and pervanadate-stimulated HEK-293 cells by Western blotting revealed a substantial decrease in the electrophoretic mobility of endogenous PKB␤. Thus, the protein appeared as a single major band in quiescent cells and formed a second, slower migrating band after stimulation with pervanadate (Fig. 2D).
To compare the specific activities of PKB␤ and -␣, we measured the activity of each enzyme toward various peptide substrates. Transiently expressed wild-type HA-PKB␤ and -␣ were immunoprecipitated with the 12CA5 antibody from pervanadate-stimulated HEK-293 cells and assayed using peptides related to the sequence surrounding the phosphorylation site on GSK-3. These experiments revealed no significant difference in the substrate specificity of PKB␤ compared with PKB␣. Thus, both isoforms phosphorylated Crosstide (GRPRTSSFAEG) and RPRAATF most efficiently, followed by KKRNRTLSV and KKLNRRLSVA. KKLRRTLSVA and KKLNRTLSVA proved to be poor substrates for both, PKB␤ and -␣ (data not shown). Although we did not observe major differences in substrate specificity, the specific activities of the two isoforms differed significantly. Using Crosstide, the specific activity of PKB␤ was about 8-fold lower (23 nmol/min/mg) compared with that of PKB␣ (178 nmol/min/mg) when HEK-293 cells were stimulated with pervanadate (see "Discussion").

Activation of HA-PKB␤ by Insulin and IGF-1 in HEK-293 Cells Is Accompanied by a PI3K-dependent Phosphorylation of Thr-309 and Ser-474 -HEK-293 cells transiently expressing
HA-PKB␤ were 32 P-labeled and treated with and without 100 nM wortmannin for 15 min and then stimulated with buffer, insulin, or IGF-1 for 10 min prior to cell lysis. In the absence of wortmannin, HA-PKB␤ was activated about 7-fold with either insulin or IGF-1, whereas in the presence of wortmannin, no activation occurred in response to insulin or IGF-1 (data not shown). The 32 P-labeled HA-PKB␤ was immunoprecipitated from cell lysates and digested with trypsin, and the resulting peptides were analyzed by C18 chromatography. Three prominent 32 P-labeled peptides were present in unstimulated HEK- 293 cells, termed peptides A, C, and E (Fig. 3A). Peptide A eluted at 20% acetonitrile, as did the peptide derived from PKB␣ that is phosphorylated at Ser-124. Peptide C and E eluted at 27 and 33% acetonitrile, respectively. Stimulation of cells with insulin (Fig. 3B) or IGF-1(data not shown) did not affect the 32 P-labeling of peptides A, C, and E (Fig. 3, A and B) but induced the 32 P-labeling of two other peptides, termed B (23% acetonitrile) and D (28% acetonitrile), which eluted at the same acetonitrile concentrations as peptides derived from PKB␣ that are phosphorylated on Ser-473 and Thr-308 (8). Treatment of HEK-293 cells expressing HA-PKB␤ with 100 nM wortmannin prior to stimulation with insulin ( Fig. 3C) or IGF-1 (data not shown) prevented the phosphorylation of peptides B and D but had no effect on the 32 P-labeling of peptides A, C, and E (Fig. 3C). Peptide A contained phosphoserine (data not shown), and when this peptide was subjected to solid phase sequencing, 32 P-radioactivity was released after the third cycle of Edman degradation confirming that the phosphorylated residue corresponds to Ser-126 (Ser-124 in PKB␣) (Fig. 4, Peptide  A). Peptide B contained phosphoserine (data not shown), and 32 P-radioactivity was released after the seventh cycle of Edman degradation identifying the phosphorylated residue as Ser-474 (Ser-473 in PKB␣) (Fig. 4, Peptide B). Peptide D contained phosphothreonine (data not shown), and 32 P-radioactivity was released after the first cycle of Edman degradation indicating phosphorylation of Thr-309 (Thr-308 in PKB␣) (Fig. 4, Peptide  D). Peptides C and E contained phosphothreonine (data not shown). When these peptides were subjected to solid phase sequencing, 32 P-radioactivity was released after the fourteenth cycle of Edman degradation, suggesting that each is phosphorylated on Thr-451 (Thr-450 in PKB␣) (Fig. 4, Peptides C and E) where peptide C contains residues 439 -456 and peptide E contains residues 439 -467.
To confirm the in vivo labeling data, various PKB␤ mutants (HA-PKB␤, HA-PKB␤-T309A, and HA-PKB␤-S474A) were transiently transfected into HEK-293 cells and assayed following IGF-1 treatment (Fig. 5). IGF-1 activated wild-type PKB␤ about 7-fold but failed to activate either the T309A or the S474A mutant, thus demonstrating the importance of Thr-309 and Ser-474 as regulatory phosphorylation sites. Taken to-gether, our data indicate that PKB␤ is constitutively phosphorylated at both Ser-126 and Thr-451 in HEK-293 cells and that insulin and IGF-1 stimulation induces the phosphorylation on Thr-309 and Ser-474.
Membrane-targeting Promotes Constitutive Activation of PKB␤-The PH domain of PKB␣ is thought to mediate translocation of the enzyme to the plasma membrane where it becomes activated by phosphorylation (4,38). To test whether a similar mechanism exists for PKB␤, we fused the Lck-derived NH 2 -terminal motif that promotes myristoylation and palmi-toylation␤ and tested whether this motif would promote membrane localization of PKB␤. Wild-type HA-PKB␤ and m/p-HA-PKB␤ were transiently expressed in HEK-293 cells, and after serum starvation, cytoplasmic localization of wild-type HA-PKB␤ (data not shown) and membrane-localization of m/p-HA-PKB␤ was confirmed by immunofluorescence (Fig. 6A). Immunoprecipitation of wild-type HA-PKB␤ or m/p-HA-PKB␤ from cells treated with or without IGF-1 followed by in vitro kinase assays revealed that membrane association led to a 15-fold activation of m/p-HA-PKB␤, and the activity of m/p-HA-PKB␤ was not further increased by IGF-1 treatment (Fig. 6B).
Translocation of PKB␤ from the Cytoplasm to the Nucleus after Mitogenic Stimulation-Previous studies showed that PKB is primarily localized in the cytoplasm of unstimulated cells, but the oncogenic form of PKB␣ (v-Akt) is present at about equal levels in the plasma membrane, cytoplasm, and the nucleus (29,30). To test whether activation of PKB␤ changes its localization, we expressed HA-PKB␤ in HEK-293 or REF-52 cells and examined kinase localization by confocal laser microscopy in serum-starved IGF-1 (HEK-293) or FCS plus okadaic acid (REF-52)-stimulated cells (data not shown and Fig. 7). In unstimulated cells, PKB␤ was cytoplasmic in both cell lines. After mitogenic stimulation, the subcellular distribution changed in both HEK-293 and REF-52 cells within 20 -30 min with most of the PKB␤ being localized to the nucleus.

DISCUSSION
Overexpression of PKB␤ is thought to contribute to the malignant phenotype of some pancreatic, ovarian, and breast carcinomas (30). To investigate the physiological role of PKB␤ and how it might contribute to cellular transformation, we undertook a detailed characterization of the enzyme.
Recently, we identified Thr-308 and Ser-473 as positive regulatory phosphorylation sites in PKB␣ that cause activation of the kinase in response to insulin and IGF-1 (8). Since these regulatory phosphorylation sites are well conserved between the ␣and ␤-isoforms, we tested for a similar activation mechanism for PKB␤ mediated by mitogenic stimulation. Indeed, like PKB␣, PKB␤ was activated by mitogens, survival factors, and pervanadate but not by agents that activate protein kinase C (Fig. 2, A and B). Similarity of function between PKB␣ and -␤ is further supported by the following observations. First, the pattern of expression of PKB␤ and -␣ is similar in various tissues although the different levels of expression of the two isoforms suggest that they perform specific functions. Second, like PKB␣, PKB␤ activation is PI3K-dependent since pretreatment of cells with wortmannin abolished the activation of overexpressed and endogenous PKB␤ in response to insulin and IGF-1 treatment (Fig. 2C and data not shown). Wortmanninsensitive phosphorylation on Thr-309 and Ser-474 was essential for PKB␤ activation. In addition, PKB␤ was constitutively phosphorylated at Ser-126 and Thr-451, two residues that are also phosphorylated in PKB␣ (Ser-124 and Thr-450) (8). Third, membrane-targeted PKB␤ was constitutively activated in unstimulated cells (Fig. 6), indicating that membrane localization brings PKB␤ into close contact to its upstream activators. This is in good agreement with our previous work (38) and the work of others demonstrating that membrane localization coincides with phosphorylation of the regulatory residues (Thr-308 and Ser-473) and constitutive activation of PKB␣.
Although comparison of a panel of different substrate peptides (derived from the GSK-3 phosphorylation site) indicated that both isoforms have similar substrate specificities, the following suggests a major difference in the regulation of the isoforms. The specific activity of both endogenous and overexpressed PKB␤ after stimulation with mitogens and survival factors was significantly lower than that of PKB␣. In addition, IGF-1 stimulation of HEK-293 cells transiently expressing wild-type HA-PKB␣ resulted in about a 6-fold higher level of phosphorylation of Thr-309 and Ser-474 than that which occurred in cells transiently expressing wild-type HA-PKB␤ ( Fig.  3 and data not shown). This difference in the degree of phosphorylation and activation of PKB␣ and -␤ could be due to the presence of an alternative PI3K-independent activation mechanism (31, 32,) 3 or tight control of PKB␤ by phosphatases. It also remains to be elucidated whether PKB␤ is a substrate for PDK1, a 3-phosphoinositide-dependent Thr-308 kinase of PKB␣ (9), and it will be important to understand the mechanism by which Ser-474, the second regulatory phosphorylation site in PKB, becomes phosphorylated. Furthermore, differences in enzyme activities could arise from interactions of the ␣and ␤-isoforms with different phosphatidylinositol-phosphates via their PH domains (10,12).
Data obtained from two different cell lines suggest that a significant fraction of PKB␤ is translocated from the cytoplasm to the nucleus after activation of the enzyme (Fig. 7 and data not shown). We recently made a similar observation with PKB␣, 4 suggesting that both isoforms translocate to the nucleus. The finding that activated PKB␤, which lacks a nuclear localization signal, accumulates in the nucleus indicates a different mechanism by which translocation occurs. One possible mechanism to account for this result is the activation-dependent association of PKB with a second protein that provides the nuclear import signal. The existence of such a mechanism has been reported for stimulation-dependent MAP kinase translocation by MAP kinase kinase (33,34). Partial nuclear translocation has also been observed with the constitutively activated v-Akt (2), a protein carrying a viral Gag sequence (29). The presence of one myristoylation signal in the Gag sequence is not sufficient to stably anchor the protein to the membrane (35). This may explain why (activated) v-akt can translocate to the nucleus, but m/p-PKB␤, which contains two additional palmitoylation signals (36), is stably associated to the membrane and therefore cannot translocate (Fig. 6A).
In conclusion, the results presented here provide the basis for the understanding how PKB␤ contributes to cellular transformation. The fact that PKB␤ translocates to the nucleus following stimulation with mitogens or survival factors suggest that transcription factors could be important in this context. The identification of authentic substrates of the different isoforms of PKB will provide considerable insight into this signaling cascade. 3 M. Andjelković, unpublished observations. 4 N. Lamb, A. Fernandez, and B. A. Hemmings, unpublished data.
FIG. 6. Membrane targeting causes constitutive activation of PKB␤. A, HEK-293 cells were transiently transfected with m/p-HA-PKB␤. Cells were stained with mAb 12CA5, followed by a FITC-conjugated secondary antibody and analyzed by laser confocal scanning microscopy (see "Experimental Procedures"). B, HEK-293 cells, transiently transfected with wild-type or m/p-HA-PKB␤, were stimulated with 50 ng/ml IGF-1 for 10 min or left untreated. After cell lysis, PKB␤ was immunoprecipitated from total cell lysates using mAb 12CA5 and assayed for in vitro kinase activity as described under "Experimental Procedures." The results are shown in means Ϯ S.D. for two experiments performed in duplicate.
FIG. 7. Immunolocalization of PKB␤ in REF52 cells. Cells were transfected with HA-PKB␤ as described under "Experimental Procedures." Following starvation, cells were stimulated with 20% FCS supplemented with 1 M okadaic acid (FCS/OA) for 30 min as indicated or left untreated (Control). After stimulation, cells were fixed, permeabilized, stained with mAb anti-HA 12CA5 followed by a FITC-conjugated secondary antibody, and prepared for analysis by Laser Confocal Scanning Microscopy. Scale bars, 6 m.