A Human Protein Kinase Bγ with Regulatory Phosphorylation Sites in the Activation Loop and in the C-terminal Hydrophobic Domain*

We have cloned human protein kinase Bγ (PKBγ) and found that it contains two regulatory phosphorylation sites, Thr305 and Ser472, which correspond to Thr308 and Ser473 of PKBα. Thus it differs significantly from the previously published rat PKBγ. We have also isolated a similar clone from a mouse cDNA library. In human tissues, PKBγ is widely expressed as two transcripts. A mutational analysis of the two regulatory sites of human PKBγ showed that phosphorylation of both sites, occurring in a phosphoinositide 3-kinase-dependent manner, is required for full activity. Our results suggest that the two phosphorylation sites act in concert to produce full activation of PKBγ, similar to PKBα. This contrasts with rat PKBγ, which is thought to be regulated by 3-phosphoinositide-dependent protein kinase 1 alone.

Only a few studies have been reported on the third member of the PKB family, PKB␥, and these all involved a clone originating from a rat brain cDNA library (4,26). A major feature distinguishing rat PKB␥ from the otherwise very similar ␣ and ␤ isoforms is the C terminus, which is truncated by 23 amino acids and lacks Ser 473/474 , one of the two phosphorylation sites essential for activation of PKB␣ and ␤ (15,16). Consequently, it has been suggested that rat PKB␥ activation depends solely on the upstream kinase PDK1. We now report the cloning and characterization of human PKB␥. This isoform differs significantly from the rat enzyme in that it contains a C-terminal domain similar to PKB␣/␤, with a putative second regulatory phosphorylation site at Ser 472 .

EXPERIMENTAL PROCEDURES
Cloning of Human HA-PKB␥ and Mutant Isoforms-A 525-bp PCR product corresponding to nucleotides 815-1340 of the rat PKB␥ sequence (4) was amplified from mouse brain cDNA and used to screen several different human cDNA libraries. Twelve overlapping clones were isolated and assembled to a cDNA encoding amino acids 16 -479 of human PKB␥. This cDNA was repaired by PCR-mediated addition of a hemagglutinin (HA) tag and the missing N-terminal amino acids deduced from the rat PKB␥ sequence and ligated as a KpnI/XbaI fragment into the pCMV5 eucaryotic expression vector (27). Subsequently, the 5Ј end of PKB␥ was amplified by 5Ј-rapid amplification of cDNA ends from human brain cDNA. A mouse brain cDNA library screened with the same probe yielded 13 overlapping clones which could be assembled into a cDNA encoding the entire reading frame of mouse PKB␥. Mutations in HA-PKB␥ (HA-PKB␥T305A and HA-PKB␥T305D) were done by Quikchange (Stratagene) or with mutagenizing 3Ј primers (HA-PKB␥S472A and HA-PKB␥S472D). HA-⌬PHPKB␥ was obtained by PCR with a primer encoding the HA-tag and amino acids 119 -126. All PCR-cloned constructs were verified by DNA sequencing.
Northern Blot Analysis-Human adult and fetal multiple tissue Northern blots (CLONTECH) were hybridized with a 825-bp fragment encoding amino acids 110 -384 of PKB␥ according to the manufacturer's instructions.
Cell Culture, Immunoprecipitation, in Vitro Kinase Assays, and Immunoblot Analysis-Human embryonic kidney (HEK) 293 cells were maintained and transfected by a modified calcium phosphate method as described previously (15,28). Stimulation was for 5 min with 0.2 mM pervanadate (7) or for 15 min with 500 nM insulin (Boehringer Mannheim). Pretreatment with the PI3K inhibitor wortmannin (200 nM; gift of Dr. Markus Thelen, Theodor Kocher Institute, Bern, Switzerland) was for 15 min. Cells were extracted and HA-PKB␥ activity determined exactly as described in Ref. 28. Western blot analysis was performed as described before (15) and developed with the polyclonal phospho-specific Ser 473 antibody (1:1000, New England Biolabs), an alkaline phosphatase (AP)-coupled goat-anti mouse IgG secondary antibody (1:2000, Sigma), and alkaline phosphatase color development reagents (Boehringer Mannheim).

RESULTS AND DISCUSSION
Screening of several human cDNA libraries led to the isolation of 12 clones encoding partial and overlapping sections of the open reading frame of human PKB␥, and the cDNA was * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF124141 and AF124142.
‡ To whom correspondence should be addressed: Friedrich Miescher-Institut, P. O. Box 2543, CH-4002 Basel, Switzerland. Tel.: 41-61-697-40-46; Fax: 41-61-697-39-76; E-mail: hemmings@fmi.ch. 1 The abbreviations used are: PKB, protein kinase B; PH, pleckstrin homology; PI3K, phosphoinositide 3-kinase; PDK1, 3-phosphoinositidedependent protein kinase 1; HA, hemagglutinin; HEK, human embryonic kidney; bp, base pair; PCR, polymerase chain reaction. assembled as described under "Experimental Procedures." The 479-residue amino acid sequence of human PKB␥ is presented in Fig. 1, as an alignment with human PKB␣, PKB␤, mouse PKB␥ (see below), and with the C-terminal domain of rat PKB␥. Human PKB␥ is 83% identical to PKB␣, 78% identical to PKB␤, and 99% identical to rat PKB␥ (two changes in 451 amino acids and a different C terminus), indicating that we have isolated the authentic human PKB␥ isoform and not a PKB␣ or PKB␤ variant. Moreover, we cloned a similar PKB␥ from a mouse brain cDNA library, demonstrating that this isoform is not restricted to one species. Human and mouse PKB␥ were found to be more than 99% identical (2 amino acid changes in 479). The major characteristic distinguishing human and mouse PKB␥ from the rat isoform is the presence of a C-terminal domain similar to PKB␣ and ␤, containing a second putative regulatory phosphorylation site at Ser 472 (marked with an asterisk in Fig. 1). To ascertain whether the human PKB␥ cDNA corresponded to the major mRNA species, we performed 3Ј-rapid amplification of cDNA ends and found that all 15 clones sequenced, which spanned the C terminus of the protein contained the Ser 472 domain. However, human and rat PKB␥ diverge in amino acid sequence precisely at a site where an exon boundary has been mapped for the mouse PKB␣ gene (29). Thus, it is possible that the published rat cDNA sequence constitutes a minor splice variant of PKB␥ or a partially processed mRNA.
To assess the tissue distribution of transcripts encoding PKB␥, we used an isoform-specific radiolabeled cDNA fragment to probe two human multiple tissue Northern blots. Two equally expressed transcripts of 8.5 and 6.5 kilobases were detected in all tissues tested, with highest levels found in adult brain, lung, and kidney and very low levels in heart and liver ( Fig. 2A). Two transcripts of similar size were detected in fetal tissues (Fig. 2B), with high levels found in heart, brain, and liver, but none in the kidney. This observation, and the size of the transcripts, which are much larger than the 3.2-3.4-kilobases transcripts of PKB␣/␤, 2 indicated the presence of long untranslated regions, and thus the possibility of developmental regulation of expression or post-transcriptional modifications affecting mRNA stability.
In contrast to the rat enzyme, human PKB␥ contains two predicted regulatory phosphorylation sites, Thr 305 in the activation loop, and Ser 472 in the C-terminal domain, as does PKB␣/␤. To determine the importance of these two residues, we mutated them to alanine which cannot be phosphorylated, or to aspartate to mimic the phosphorylated state, and assayed HA-PKB␥ kinase activity following transient transfection and stimulation with the insulin mimetic compound pervanadate (Fig.  3). The results presented here show that wild type HA-PKB␥, which had a low basal activity, could be stimulated 67-fold by pervanadate treatment; furthermore, mutation of Thr 305 to alanine completely ablated activation. No activity above basal levels was observed for HA-PKB␥T305A, and the same was true for the double mutant HA-PKB␥T305A,S472A. On the other hand, mutation of the C-terminal regulatory site (HA-PKB␥S472A) reduced but did not abolish activation by pervanadate (10-fold). We also tested the effects of aspartate mutations, since in the case of PKB␣, a double aspartate mutant was constitutively active (15). However, PKB␥ was not active above basal levels upon mutation of the activation loop site to aspartate (HA-PKB␥T305D and HA-PKB␥T305D,S472D) and, furthermore, could not be stimulated by pervanadate treatment. Thus the aspartic acid moiety could not substitute for the phosphorylated threonine residue. Again, mutation of Ser 472 to aspartate (HA-PKB␥S472D) resulted in a protein that could still be activated by pervanadate (35-fold), albeit to a lesser 2 B. A. Hemmings, unpublished results.  2. Northern blot analysis of PKB␥ expression in human  tissues. A, adult and B, fetal multiple tissue Northern blots were probed for expression of PKB␥ with a [␣-32 P]dATP-random-prime-labeled probe derived from a 825-bp fragment of the human PKB␥ cDNA spanning amino acids 110 -384 and exposed for 3 days at Ϫ70°C. RNA molecular weight markers (in kilobases) are indicated. extent than the wild type. These results establish that the phosphorylation site Thr 305 in the activation loop is absolutely necessary for activation of PKB␥, with conformational constraints around the active site apparently so stringent that substitution by a negatively charged residue is not tolerated.
To determine the role of the C-terminal Ser 472 in regulation of human PKB␥, we tested whether activation of HA-PKB␥, HA-PKB␥S472A, and HA-PKB␥S472D by insulin was sensitive to inhibition of PI3K. Fig. 4A depicts the results of a representative experiment, which show that insulin activated HA-PKB␥, HA-PKB␥S472D, and, to a lesser extent, HA-PKB␥S472A. This stimulation was dependent on the activity of PI3K, since pretreatment of transfected cells with the PI3K inhibitor wortmannin inhibited activation by insulin. Furthermore, we subjected the immunoprecipitated proteins to Western blot analysis with an antibody generated specifically against the phosphorylated Ser 473 peptide of PKB␣ (Fig. 4A, inset). The antibody cross-reacted with HA-PKB␥ only upon stimulation with insulin, and phosphorylation of Ser 472 was prevented by wortmannin. Since Ser 472 was mutated in HA-PKB␥S472A and HA-PKB␥S472D and could not be phosphorylated, we concluded that the wortmannin-sensitive, insulin-stimulated activity of these proteins was entirely due to phosphorylation at Thr 305 , dependent on the presence of 3-phosphorylated phospholipids.
In this analysis, we also included PKB␥ constructs lacking the N-terminal PH domain. In the basal state, this domain is thought to restrict access to the phosphorylation site in the activation loop, thus leaving Thr 305 more accessible to phosphorylation by upstream kinases when it is removed. The proteins lacking the PH domain now presented a different picture (Fig. 4B): HA-⌬PHPKB␥ was maximally activated under basal conditions and could not be stimulated further by insulin treatment. This contrasts with results obtained for ⌬PHPKB␣, shown to be activated by insulin (6). Furthermore, pretreatment of the cells with wortmannin led to a reduction in activity of HA-⌬PHPKB␥, indicating that it was still a target for PI3Kdependent phosphorylation. HA-⌬PHPKB␥S472A activity was comparable with that of wortmannin-treated HA-⌬PHPKB␥ and was not responsive to insulin or wortmannin. In contrast, HA-⌬PHPKB␥S472D was again fully active in the absence of stimulation but, unlike HA-⌬PHPKB␥, was not inhibited by wortmannin. The Western blot signals with the phospho-specific Ser 473 antibody correlated with the activities observed ( Fig. 4B, inset); HA-⌬PHPKB␥ was strongly phosphorylated in extracts of unstimulated and insulin-stimulated cells, but pretreatment with wortmannin reduced the signal. We found previously that transiently transfected PDK1, the upstream kinase phosphorylating Thr 308 in PKB␣, is active in serumstarved HEK-293 cells. 3 The present results seem to indicate that removal of the PH domain causes a conformational change of PKB␥ favorable to phosphorylation at Thr 305 , so as to make it independent of PI3K activity. Basal activity of PI3K in unstimulated cells allowed phosphorylation of HA-⌬PHPKB␥ by the Ser 473 kinase, resulting in full activation. Conversely, inhibiting this basal PI3K activity by wortmannin treatment reduced HA-⌬PHPKB␥ activity, probably due to the rapid action of phosphatases on phosphorylated Ser 472 . Thus, HA-⌬PHPKB␥ is a model for studying phosphorylation of Ser 472 , the second regulatory site of human PKB␥, almost independent of Thr 305 .
In summary, we report the cloning and characterization of human PKB␥, a PKB isoform distinct from its rat counterpart in having two regulatory phosphorylation sites, Thr 305 and Ser 472 , both of which are required for full activation of the protein. Our results suggest markedly similar regulation mechanisms for PKB␥ and the ␣ and ␤ isoforms, with both upstream kinases phosphorylating the regulatory sites being sensitive to PI3K-derived signals. Furthermore, we found a high abundance of PKB␥ mRNAs encoding the C-terminal hydrophobic domain and have isolated a similar mouse PKB␥, showing that this isoform is not restricted to humans. Taken together, we conclude that the truncated rat PKB␥ used in all studies so far (26) probably constitutes a minor splice variant of endogenous PKB␥ protein. The crucial question now emerging is that of the specific roles of the three different PKB isoforms.