Two Splice Variants of Protein Kinase B g Have Different Regulatory Capacity Depending on the Presence or Absence of the Regulatory Phosphorylation Site Serine 472 in the Carboxyl-terminal Hydrophobic Domain*

We have reported previously the cloning and characterization of human and mouse protein kinase B g (PKB g ), the third member of the PKB family of second messenger-regulated serine/threonine kinases (Brod-beck, D., Cron, P., and Hemmings, B. A. (1999) J. Biol. Chem. 274, 9133–9136). Here we report the isolation of human and mouse PKB g 1, a splice variant lacking the second regulatory phosphorylation site Ser-472 in the hydrophobic C-terminal domain. Expression of PKB g 1 is low compared with PKB g , and it is regulated in different human tissues. We show that PKB g and PKB g 1 differ in their response to stimulation by insulin, pervanadate, peroxide, or okadaic acid. Activation of PKB g 1 requires phosphorylation at a single regulatory site Thr-305. In-terestingly, this site is phosphorylated to a higher extent in PKB g compared with PKB g 1 upon maximal stimulation by pervanadate, and this is reflected in the respective specific kinase activities. Furthermore, upon insulin stimulation of transfected cells, PKB g 1 translocates to the plasma membrane to a lesser extent than PKB g . Taken together, these results suggest that phosphorylation of the hydrophobic motif at the extreme C terminus of PKB g may facilitate translocation of the kinase to the membrane and/or its phosphorylation on the activation

Protein kinase B (PKB) 1 is implicated in a wide variety of cellular responses to insulin and growth factor signaling (for review, see Refs. 1 and 2). The three PKB isoforms identified so far, PKB␣, ␤, and ␥, constitute a subfamily among the second messenger-regulated serine/threonine protein kinases (3)(4)(5)(6)(7)(8)(9). The human enzymes are 73% identical in amino acid sequence, and all contain an N-terminal pleckstrin homology domain, a central kinase domain, and a C-terminal regulatory hydrophobic domain. Activation occurs in response to signaling via phosphoinositide 3-kinase (10 -13). PtdIns (3,4,5)P 3 , the membranebound active second messenger (for review, see Refs. 2 and 14), is thought to recruit PKB to the membrane, promoting a conformational change that allows phosphorylation on two regulatory sites by upstream kinases that are either located at the membrane or brought there through interaction with PtdIns(3,4,5)P 3 themselves (15)(16)(17). In support of this model, it has been shown that PtdIns(3,4,5)P 3 and its metabolite PtdIns(3,4)P 2 bind to the pleckstrin homology domain (18,19) and that PKB␣ containing an N-terminal membrane-targeting signal is constitutively active, independent of the presence of a pleckstrin homology domain (20,21). The two phosphorylation sites critical for regulation of the activity of all three enzymes have been identified (7,22,23); one is found in the activation loop of the kinase domain (Thr-308 in PKB␣, Thr-309 in PKB␤, and Thr-305 in PKB␥), and the other is in the C-terminal hydrophobic domain (Ser-473 in PKB␣, Ser-474 in PKB␤, and Ser-472 in PKB␥).
Of the upstream kinases, the one that phosphorylates Thr-308 in the activation loop of PKB␣ in a PtdIns(3,4,5)P 3 -dependent manner has been identified and termed 3-phosphoinositidedependent kinase-1 (24 -27). In addition to PKB␣, this kinase also phosphorylates equivalent sites in other second messenger-regulated kinases, such as p70S6 kinase (28,29), protein kinase A (30), protein kinase C (31), serum-and glucocorticoidregulated kinase (32,33), and p90 ribosomal S6 kinase-2 (34). For the second regulatory site in the hydrophobic C-terminal domain of PKB, several candidate upstream kinases have been identified. MAP kinase activated protein kinase-2 (MAPKAP-K2) is able to phosphorylate Ser-473 in vitro, but in vivo it is activated by stress and other stimuli that fail to activate PKB (22). The integrin-linked kinase ILK-1 is activated by phosphoinositide 3-kinase signaling and can phosphorylate Ser-473 in vitro (35). It has also been reported that 3-phosphoinositide-dependent kinase-1 might acquire the capability to phosphorylate Ser-473 by interaction with a region of protein kinase C-related kinase-2 (PRK2; 36), and a recent report claims that PKB␣ phosphorylated on Thr-308 is able to autophosphorylate on the Ser-473 site (37).
Activated PKB mediates a range of cellular responses to insulin and growth factors, such as phosphorylation and inactivation of glycogen synthase kinase-3 (38), up-regulation of protein translation through activation of the mammalian target of rapamycin mTOR (39), inactivation of the translational repressor 4E-BP1 (40), and stimulation of ribosomal p70S6 kinase, as well as activation of phosphofructokinase-2 (41). Constitutively active PKB stimulates glucose uptake through * This study was supported in part by the Schweizerische Krebsliga (to M. M. H. and B. A. H.). The Friedrich Miescher-Institut is part of the Novartis Research Foundation. 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) AY005799.
‡ To whom correspondence should be addressed. recruitment of glucose transporters to the plasma membrane of adipocytes (11). Furthermore, PKB is linked to phosphorylation and activation of endothelial nitric-oxide synthase induced by shear stress on endothelial cells in the vasculature (42,43). Stimulated PKB translocates to the nucleus in human embryonic kidney (HEK)-293 cells (21,23), activates cAMP-responsive element-binding protein by phosphorylation (44), and mediates insulin response sequence-specific effects on hepatic gene expression (45).
Many reports show that PKB is a critical regulator of cell survival (46). Phosphorylation by PKB inactivates both the apoptosis-promoting Bcl-2 family member BAD (47,48) and the protease caspase 9 (49,50). Nuclear targets of PKB promoting cell survival are the Forkhead family of transcription factors, where the human homologs FKHR (51), FKHRL1 (52,53), and AFX (54) contain three consensus sites for phosphorylation by PKB (55). Accordingly, PKB phosphorylates and inhibits FKHRL1 and FKHR, promoting their export from the nucleus or leading to retention in the cytoplasm, as part of an evolutionarily conserved mechanism by which insulin and growth factors regulate gene expression (56 -59).
In this paper, we report the cloning and characterization of a novel variant of human and mouse PKB␥. Human PKB␥1 is 14 amino acids shorter, and mouse PKB␥1 9 amino acids, than PKB␥, and both lack the second regulatory phosphorylation site, Ser-472. Both variants arise from the same gene through differential splicing of the C-terminal exons, and the respective mRNAs are expressed in a tissue-specific manner. Both kinases can be activated by a range of stimuli, but they vary considerably in the extent of phosphorylation on the regulatory site Thr-305 in their specific kinase activities and with respect to membrane localization.

EXPERIMENTAL PROCEDURES
Cloning and Construction of Expression Vectors HA-PKB␥, HA-PKB␥1, and Phosphorylation Site Mutants-The cloning of human and mouse HA-PKB␥ has been described previously (7). Two additional human PKB␥ cDNA clones were isolated (encoding amino acids 16 -451) which contained sequences for a further 14 amino acids and 3Ј-noncoding region. These sequences have been submitted to GenBank under the accession number AY005799. To generate HA-PKB␥1 in the pCMV5 eukaryotic expression vector (60), a C-terminal 153-base pair HindIII/XbaI fragment from a PCR product amplified with the 5Ј-primer 20091 (5Ј-GGACTATCTACATTCCGGAAAG-3Ј) and the 3Ј-primer 20429 (5Ј-GGGTCTAGATTATTATTTTTTCCAGTTACC-CAGCATGCCACAATCTGA-3Ј) was ligated into a digested HA-PKB␥ plasmid (7).
Site-directed mutagenesis of HA-PKB␥1 using the QuikChange kit (Stratagene) was used to generate the activation loop site mutations HA-PKB␥1T305A and HA-PKB␥1T305D and the Thr-447 site mutations HA-PKB␥1T447A and HA-PKB␥1T447D. All PCR-cloned constructs were verified by DNA sequencing.
Generation of a Mouse BAC Sublibrary and Screening-The fulllength mouse PKB␥ cDNA (7) was used to isolate a BAC clone (Incyte Genomics) containing the PKB␥ gene. This clone was digested with different restriction enzymes, and the fragments were ligated into pBluescript. Bacterial colonies were screened with a probe spanning exons 12 and 13 of mouse PKB␥, or about 400 base pairs of putative intronic sequence upstream of exon 14, by standard procedures.
Amplification of Large PCR Fragments-Long range PCR was performed with the Expand Long Template PCR System (Roche Molecular Biochemicals) according to the manufacturer's instructions, using 200 ng of human genomic DNA (Roche Molecular Biochemicals) as template. The PCR protocol consisted of 10 cycles of 94°C for 10 s, 55°C for 30 s, 68°C for 4 min. This was followed by 30 cycles of 94°C for 10 s, 55°C for 30 s, 68°C for 4 min, and an extension of the elongation step by 20 s/cycle. The reaction was reamplified if necessary, and the products were analyzed on a 0.6% agarose gel, extracted, and sequenced with the same primers used for amplification (primer 20090, 5Ј-CCT-CAAGTAACATCTGAGACAG-3Ј; primer 20429, see above; primer 26759, 5Ј-GGGTCTAGATTATTATTCTCGTCCACTTGCAGAGTAGG-AAAATTG-3Ј; and primer 28144, 5Ј-GCAGGGGCACCTTCC-GACATC-3Ј).
Isolation of Total RNA from Mouse Brain and Embryo-Mouse brain or embryos (E15-E20) frozen in liquid N 2 were extracted with Trizol reagent (Life Technologies, Inc) according to the manufacturer's instructions and the amount of total RNA quantitated by absorption at A 260 .
Reverse Transcription and Semiquantitative PCR-Reverse transcription was performed with the GeneAmp RNA PCR kit (PE Biosystems), essentially according to the manufacturer's instructions, using random primers and 1 g of total RNA from mouse brain or embryos as template in a total volume of 20 l, and a reaction protocol of 10 min at 25°C, 60 min at 42°C and 5 min at 99°C. 5 l of the reaction was used as template for PCR in a total volume of 20 l, with a protocol of 35 cycles of 95°C for 15 s, 54°C for 30 s, and 72°C for 30 s, and the reaction was reamplified if necessary. For amplification, the common sense primer 20091 (see above) was used in combination with the mouse splice variant-specific primers 32212 (5Ј-GGTGAAGACCCTTGGCTG-GTC-3Ј), resulting in a band of 798 base pairs, and 32210 (5Ј-GGGTCTAGATTACTTTTTATTATCATTTTTTTTCCAGTTAC-3Ј), resulting in a band of 636 base pairs. For semiquantitative RT-PCR of human tissue RNA, this protocol was modified as follows. To 1 g of total RNA (CLONTECH) 10 5 copies of synthetic pAW109 RNA were added as normalization control. Serial dilutions were made of the reverse transcription, and 10 l of four consecutive dilutions was employed as template for PCR. Primer 20091 was employed in combination with primers 26939 (5Ј-GTGGCGAGGGGTGAGGACCCTT-3Ј), resulting in a band of 809 base pairs, and 20429 (see above), resulting in a band of 621 base pairs, and annealing was raised to 56°C. As a control for the reverse transcription, a 308-base pair fragment of pAW109 RNA was amplified with the primers DM151 and DM152 supplied in the kit. The reactions were electrophoresed on a 1% agarose gel, photographed under UV light, and the bands quantitated using Molecular Dynamics software.
For immunofluorescence, the cells were washed once with phosphate-buffered saline, fixed in 4% p-formaldehyde (Merck) in phosphate-buffered saline for 30 min at room temperature, and permeabilized with 0.2% Triton X-100 (Sigma) for 5 min. They were then stained with the monoclonal anti-HA antibody 12CA5 and a fluorescein isothiocyanate-coupled anti-mouse IgG secondary antibody (1:500) and observed by confocal microscopy.
Immunoprecipitation, Kinase Assays, and Immunoblot Analysis-HA-PKB␥ protein was immunoprecipitated from 50 -100 g of cell lysate, with the 12CA5 monoclonal antibody bound to protein A-Sepharose beads. The immunoprecipitates were washed as described previously (21) and incubated in 50 l of in vitro kinase assay mix (50 mM Tris, pH 7.5, 15 mM ␤-mercaptoethanol, 10 mM MgCl 2 , 1 M protein kinase A inhibitor peptide (Bachem, Switzerland), 30 M Crosstide substrate (GRPRTSSFAEG; 38), and 50 M [␥-32 P]ATP (Amersham Pharmacia Biotech, 2500 cpm/pmol) for 30 -60 min at 30°C. The reaction was stopped and processed as described previously (21). Peptides utilized in the substrate specificity assay were obtained from Neosystems or synthesized at the institute facility.
Western blot analysis was performed as described (22) and developed with either the monoclonal anti-HA antibody 12CA5, a polyclonal antibody raised against a pan-phospho-Thr PKB peptide (DAATMKT-(P)FCGTP), or a polyclonal antibody raised against a pan-phospho-Ser PKB peptide (RPHFPQFS(P)YSAS). Alkaline phosphatase-coupled anti-mouse or anti-rabbit IgG was used as secondary antibody (1:2,000, Sigma). Color development reagents were from Bio-Rad.

RESULTS
Two Variants of PKB␥ Arise from a Single Gene through Differential Splicing of C-terminal Exons-In the process of cloning human PKB␥ (7), we isolated a cDNA clone, termed PKB␥1, which was identical to PKB␥ until amino acid 451 but differed in sequence at the C terminus. The human PKB␥1 sequence extended for a further 14 amino acids only, showed no homology to the C-terminal 28 residues of PKB␥, and did not contain the Ser-472 regulatory phosphorylation site (Fig. 1). With the novel amino acid sequence, we performed a search of expressed sequence tag (EST) data bases, which revealed the existence of a large number of human and several mouse and rat ESTs containing this sequence. Interestingly, all mouse ESTs encoded an additional 5 amino acids but did not contain a site homologous to the Ser-472 phosphorylation site, whereas the rat ESTs encoded an additional 8 amino acids (Fig. 1).
Significantly, the starting point of the sequence divergence coincided with the location where an exon boundary had been mapped for the mouse PKB␣ gene (61). Taking into account the high degree of conservation among individual PKB isoforms, we assumed that the exon boundaries would be conserved and that the diversity at the C terminus could be caused by alternative splicing. With the two cDNA clones encoding PKB␥ and PKB␥1 we screened the NCBI human genome data base and obtained homology with the same contig NT_004480, a chromosome 1 working draft sequence segment. The exon sizes, alignments with the contig, and analysis of the splice acceptor and splice donor sites are listed in Table I; the exons were numbered according to the mouse PKB␣ gene (61). The contig comprised exons 2-15 of the human PKB␥ gene. Exons 2-12 were shared by both splice variants, exon 13 encoded the PKB␥-specific C terminus and 3Ј-untranslated region, exon 14 encoded the PKB␥1-specific C terminus and 6 nucleotides past the stop codon, and exon 15 contained the remaining 3Ј-untranslated region of PKB␥1. A graphic depiction of the exon structure of the two human PKB␥ splice variants is given in Fig. 2A. The sequences bordering the exons fit well to the consensus for splice acceptor and splice donor sites (Table I). Because this contig was only a working draft sequence segment, the sizes of the intervening introns could not be derived reliably.
To determine the gene structure and intron size in the region where alternative splicing occurred, we designed antisense primers specific for the two C-terminal coding exons (exon 13-primer 26759 for PKB␥; exon 14-primer 20429 for PKB␥1) and a common sense primer (20090) located 63 base pairs upstream of the splice donor site in exon 12 and performed long range PCR on human genomic DNA (Fig. 2B). For the mouse PKB␣ gene the distance between exons 12 and 13 has been mapped to 348 base pairs. 2 Using human genomic DNA as a template and the PKB␥-specific primer pair 20090/26759, we amplified a fragment of about 7 kb (Fig. 2C, lane 1). With the primer pair 20090/20429 specific for PKB␥1, a fragment of 12 kb was amplified (Fig. 2C, lane 2), and using the sense primer 28144 located within the PKB␥-specific exon, together with the PKB␥1-specific primer 20429, resulted in a 5-kb fragment (Fig.  2C, lane 3). We partially sequenced all fragments to confirm their identity, and furthermore were able to derive PKB␥specific sequences from the 12-kb fragment, thus confirming the co-linearity of exon 12, exon 13 (PKB␥-specific), and exon 14 (PKB␥1-specific) and the sizes of the intervening introns on genomic DNA.
Mouse PKB␥1 Is Expressed in Adult Brain and in Embryos-Because this was the first evidence of differential splicing of a PKB isoform, we wanted to confirm this in other species. The sequences of the mouse PKB␥ variants are highly homologous to their human counterparts (7, and Fig. 1). Therefore we screened several sublibraries made of a BAC clone encoding the mouse PKB␥ gene with two different probes, one derived from exons 12 and 13 of the mouse PKB␥ cDNA, the other made by PCR from the intronic sequences adjacent to the putative exon 14. We obtained five partially overlapping clones containing exons 12, 13, and 14, separated by introns of about 7 and 6 kb, establishing that the mouse PKB␥ gene had a similar architecture (Fig. 3A).
To demonstrate the presence of transcripts corresponding to both splice variants in vivo, we used RT-PCR. Total RNA was isolated from adult mouse brain or whole mouse embryos (embryonic days E15-E20), reverse transcribed with oligo(dT) primers and used as template for PCRs with a common sense primer 20091 located in exon 8 and antisense primers specific for the individual isoforms (32212 for mPKB␥, 32210 for mPKB␥1). The products were purified over a 1% agarose gel and sequenced to confirm their identity (Fig. 3, B and C), establishing that both C-terminal splice variants of PKB␥ were expressed in the mouse.
Analysis of the Abundance of mRNA Transcripts of Human PKB␥ and PKB␥1-A more detailed analysis of the relative abundance of the two mRNA transcripts was done for the human PKB␥ splice variants. Among the 12 human PKB␥ cDNA clones isolated, 9 encoded the C terminus, and of these, two contained the PKB␥1-specific sequences. Furthermore, 15 clones obtained by performing 3Ј-RACE on brain cDNA-encoded PKB␥. Northern blot analysis had revealed the presence of two transcripts of 8.5 and 6.5 kb with comparable relative expression levels in several human adult and fetal tissues (7), but when the same blots were hybridized with a probe specific for PKB␥1 we obtained no signal. These results implied that PKB␥1 was less abundant than PKB␥, prompting us to quantitate relative abundance and possibly tissue-specific expression of PKB␥ and PKB␥1 transcripts by semiquantitative RT-PCR. As templates for the RT reaction, we used total RNA samples derived from human prostate, testis, uterus, kidney, mammary gland, skeletal muscle, brain, trachea, lung, heart, and liver. Four consecutive 2-fold dilutions of these reactions were used as templates for the PCRs, chosen to en-2 Z. Yang and B. A. Hemmings, unpublished results. FIG. 1. Sequence alignment of the C-terminal domains of human, mouse, and rat PKB␥ and PKB␥1. Shown here are the amino acid sequences of the C-terminal domains of human, mouse, and rat PKB␥ (amino acids 406 -479; see Refs. 6 and 7) and human, mouse, and rat PKB␥1, with the regulatory phosphorylation site Ser-472 indicated with an asterisk and the presumed constitutive phosphorylation site Thr-447 marked with a dot. Human and mouse PKB␥1 sequences were confirmed using a human cDNA clone and RT-PCR products. The rat PKB␥1 C-terminal sequences were obtained from an EST data base. The arrowhead indicates the site where the exon boundary had been mapped for PKB␣ (Ref. 61 and Footnote 2). The sequence of human PKB␥1 has been submitted to GenBank under the accession number AY005799.
sure that amplification was still in the linear range. The end products were analyzed on a 1% agarose gel and quantified in relation to the product amplified from the synthetic control pAW109 RNA included in the RT reaction. The results of these experiments are given in Table II and show that the expression of PKB␥ varied among different tissues, ranging from unde-tectable levels in liver and low levels in heart and lung to highest expression in the brain, testis, uterus, and prostate. PKB␥1 was also detected in all tissues except liver, heart, and lung, and its abundance varied considerably but at significantly lower levels. Highest levels were detected in prostate, testis, and mammary gland. Thus, in tissues of the genitourinary tract, such as prostate, testis, uterus, and kidney, in the mammary gland and in skeletal muscle, PKB␥1 accounted for 2-8% of total PKB␥ transcripts, whereas in trachea and brain, levels made up less than 1% (Table II).
Regulation of HA-PKB␥1 Activity-We have reported previously that mutation of Thr-305 of HA-PKB␥, the regulatory phosphorylation site in the activation loop, resulted in complete inactivation of the protein (7). Wild-type HA-PKB␥1, which lacks the second regulatory site in the hydrophobic domain, could be stimulated by pervanadate treatment when transiently expressed in HEK-293 cells (Fig. 4A). When Thr-305 of HA-PKB␥1 was mutated to an alanine that cannot be phosphorylated (HA-PKB␥1T305A), the activation was abolished (Fig. 4A). When we mutated Thr-305 to an aspartate (HA-PKB␥1T305D) to mimic the activated state, the activity of the kinase did not increase significantly above the levels of unstimulated wild-type protein and could not be activated by pervanadate treatment (Fig. 4A), again confirming the observations made with HA-PKB␥ (7). This is in contrast to PKB␣, where mutation of the two regulatory phosphorylation sites to acidic residues produced a constitutively active protein (22).
Because PKB␥1 lacked the Ser-472 residue, we tested whether the phosphorylation status of Thr-447, shown to be a site of constitutive phosphorylation in PKB␣ (22), influenced the activation of the kinase, in effect taking over the role of the second regulatory phosphorylation site. Mutants carrying either an alanine or an aspartate residue at Thr-447 (HA-PKB␥1T447A and HA-PKB␥1T447D) were transiently transfected and kinase activity measured in response to stimulation by pervanadate. We observed an activation profile comparable to the wild-type protein with both constructs, indicating that the phosphorylation status of Thr-447 had no influence on the activation of HA-PKB␥1 (Fig. 4A). Furthermore, a double mutant HAPKB␥1T305D/T447D was also not active above basal levels and could not be stimulated (data not shown).
Western blot analysis of cell lysates revealed that HA-PKB␥1 wild-type and mutated at Thr-305 migrated as a doublet (Fig.  4B). In contrast, the Thr-447 mutants ran as single bands, with HA-PKB␥1T447A co-migrating with the faster, and HA-  PKB␥1T447D with the slower conformation of HA-PKB␥1. Thus it was possible that the double band observed with HA-PKB␥1 was caused by differential phosphorylation at Thr-447.
Comparison of the Activities of HA-PKB␥ and HA-PKB␥1 Elicited by Different Stimuli-We compared kinase activities of transiently transfected HA-PKB␥ and HA-PKB␥ 1 under basal conditions and in response to a range of stimuli, including 500 nM insulin, 0.2 mM pervanadate, 10 mM peroxide, and 1 M okadaic acid. The results shown in Fig. 5A demonstrated a robust activation of HA-PKB␥ upon treatment of the cells with insulin or the insulin-mimetic compound pervanadate, but less potent stimulation of kinase activity when the cells were treated with peroxide or okadaic acid. In contrast, the most potent activation of HA-PKB␥1 was achieved in response to pervanadate stimulation, whereas only a weak activation was observed upon treatment with insulin or peroxide and none in response to okadaic acid treatment (Fig. 5A). Other stimuli tested included 12-O-tetradecanoylphorbol 13-acetate, forskolin, or the calcium ionophore A23187, none of which resulted in activation of either splice variant.
Overall, HA-PKB␥1 had lower basal and stimulated relative kinase activities compared with HA-PKB␥. The immunoprecipitates prepared for activity assays were analyzed by Western blot, showing that HA-PKB␥ was expressed at a slightly lower level than HA-PKB␥1 (Fig. 5B). Relative kinase activities of both splice variants were reflected in the phosphorylation status of Thr-305, the lower level of phosphorylation of HA-PKB␥1 correlating with lesser activation (Fig. 5C). Finally, by monitoring the phosphorylation status of Ser-472 of HA-PKB␥ under the different conditions we found that okadaic acid treatment led to phosphorylation at Ser-472 but not Thr-305 and to low levels of kinase activation (Fig. 5D).  3 and 4) total RNA was reverse transcribed with oligo(dT) primers, used as template for amplification with mPKB␥and mPKB␥1-specific antisense primers (indicated in panel A; sense primer 20091 located in exon 8), and the products were run on a 1% agarose gel. Molecular weight (MW) markers are indicated in kb. Panel C, the major bands were isolated from the gel, sequenced, and the amino acid sequence derived.

TABLE II
Quantitation of human PKB␥ and PKB␥1 transcripts 1 g of total RNA of human prostate, testis, uterus, kidney, mammary gland, skeletal muscle, brain, trachea, lung, heart, and liver and 10 5 copies of synthetic pAW109 RNA were reverse transcribed, appropriate serial dilutions amplified by PCR, analyzed on 1% agarose gels, and quantitated using Molecular Dynamics software. Indicated is the average (ϮS.D.) of two determinations of the abundance of PKB␥ and PKB␥1 mRNA transcripts, relative to amplification of the synthetic pAW109 RNA. Also indicated is the ratio of expression of PKB␥1 to PKB␥ (in %).  50 M C2-ceramide for 2 h did not have an effect on basal or stimulated activity of either splice variant. This was in contrast to PKB␣, where ceramide treatment led to inhibition of kinase activity concomitant with a reduction in Ser-473 phosphorylation (62). However, we observed a differential effect on kinase activation of HA-PKB␥ and HA-PKB␥1 under conditions of osmotic stress. Whereas pretreatment of cells with sorbitol prior to stimulation with insulin or pervanadate did not affect activation of HA-PKB␥, it did have an effect on kinase activity of HA-PKB␥1, inhibiting the stimulation by pervanadate, but not the slight activation elicited by insulin (Fig. 6A). This again was reflected by the phosphorylation status of Thr-305. HA-PKB␥1 was less phosphorylated on Thr-305 when stimulated with pervanadate following sorbitol pretreatment (Fig. 6, B and C), whereas no effect was observed on either Thr-305 or Ser-472 phosphorylation of HA-PKB␥ (Fig. 6, C and D).
Comparison of the Substrate Specificities of HA-PKB␥ and HA-PKB␥1-We compared specific kinase activities of pervanadate-stimulated HA-PKB␥ and HA-PKB␥1 assayed with 15 different peptides, which were derived from possible kinase substrates. For this experiment, equal amounts of HA-tagged protein were immunoprecipitated from lysates of transiently transfected cells; the results are shown in Table III, with specific kinase activity expressed in pmol/min. For both splice variants, the most activity could be measured using Crosstide as the substrate, followed by a number of peptides conforming to the consensus motif RXRXXS/T. A minimal peptide of the sequence RPRAATF elicited about half-maximal phosphorylation, but any variation in the position of the upstream arginines immediately led to a substantial decrease in activity. Surprisingly, a peptide with the motif of the phosphorylation site Ser-136 of BAD, a known target of PKB␣ (47), was also a rather weak substrate for both kinases. Likewise, peptides corresponding to the phosphorylation site Ser-259 of the Raf isoforms or to Ser-112 of BAD were not phosphorylated, even though they conformed to the consensus motif. The Raf peptides were also tested as substrates for HA-PKB␣ but elicited no kinase activity (data not shown). Specific kinase activity of HA-PKB␥ was higher than that of HA-PKB␥1 and was slightly more sensitive to variations in peptide sequence, but overall no major divergence in substrate preferences could be determined.
Localization of HA-PKB␥ and HA-PKB␥1 in Transiently Transfected Insulin-stimulated Cells-We have shown here that in contrast to HA-PKB␥, HA-PKB␥1 kinase activity was stimulated only weakly in response to insulin treatment of transiently transfected cells. This observation led us to investigate the localization of the two splice variants in response to insulin stimulation. It had been shown previously that insulinlike growth factor I or insulin treatment led to translocation of PKB␣ and PKB␤ to the membrane, where the activation by

TABLE III
Peptide substrate specificity of HA-PKB␥ and HA-PKB␥1 HEK-293 cells were transfected with constructs encoding HA-PKB␥ or HA-PKB␥1, serum-starved for 18 h, stimulated with 0.2 mM pervanadate for 5 min, and the extracts prepared. After normalizing the expression of HA-tagged proteins by Western blot, the kinase activity of equal amounts of immunoprecipitated HA-PKB␥ or HA-PKB␥1 was determined as described (21), except that the substrate Crosstide was replaced with the relevant peptide. The amino acid sequences of the peptides are indicated; the relevant residues of the consensus phosphorylation site are underlined, and the residue presumed to be phosphorylated indicated in bold. Specific kinase activity (pmol/min) is the average (ϮS.D.) of two experiments assayed in duplicate. The experiment was repeated twice more with comparable results.  (21,23). In Fig. 7A, typical examples of HA-PKB␥-and HA-PKB␥1-transfected cells are shown, in the absence (top) and the presence (bottom) of insulin stimulation. To quantitate the translocation of PKB␥ and PKB␥1 in response to insulin stimulation, cells were classified according to the intracellular localization of HA-tagged proteins. The data show (Fig. 7B) the fraction of cells where HA-tagged protein was localized in the cytosol, in cytosol and membrane, or at the membrane only, expressed as percentage of a total of about 1,200 cells counted for each condition. These data suggested that either HA-PKB␥1 translocated to the membrane less readily than HA-PKB␥ upon insulin stimulation or that it dissociated more rapidly from the membrane and then became quickly dephosphorylated. DISCUSSION Previous studies on the regulation of the three PKB isoforms ␣, ␤, and ␥ revealed the central importance of the two regulatory phosphorylation sites, Thr-308/309/305 and Ser-473/474/ 472, in controlling kinase activity in response to insulin or growth factor signaling via the phosphoinositide 3-kinase pathway, with both sites required for full activation (for review, see Refs. 1 and 2). Rat PKB␥ lacked the regulatory phosphorylation site in the C-terminal domain, and yet transiently transfected protein could be activated in vitro (6,63). The rat PKB␥ sequence differed significantly only at the C terminus; thus, it seemed likely to be a splice variant. We now provide compelling evidence that the human and mouse PKB␥ genes contain additional exons that are utilized in a splice variant, leading to translation of proteins with different C termini (Table I and Fig. 2A). This splice variant PKB␥1 also lacked the Ser-472 phosphorylation site, and yet kinase activity of human PKB␥1 could be stimulated through a single regulatory phosphorylation site in the activation loop, Thr-305 (Fig. 4).
We have characterized the expression pattern of the two splice variants PKB␥ and PKB␥1 by RT-PCR, showing that both are expressed in adult mouse brain and in E15-E20 embryos (Fig. 3). Furthermore, an extensive analysis of several different human tissue RNA samples revealed co-expression of both transcripts. PKB␥1 was expressed at relatively high levels in RNA samples from prostate, testis, uterus, and the mammary gland; in these tissues, and also in kidney and in skeletal muscle, it accounted for more than 2-8% of total PKB␥ expressed (Table II). PKB␥, the more abundant message, additionally showed high expression in brain and trachea, reflecting the Northern blot results published previously (6,7). Also, neither of the splice variants were detected in liver and only very low levels of PKB␥ in heart and lung, whereas analysis of expression of PKB␣ and ␤ by human multiple tissue Northern blots revealed that both of the latter isoforms were highly expressed in these tissues. 3 These results indicated that expression and alternative splicing of the PKB␥ gene are under tissue-dependent regulation. Alternatively, the low abundance of mRNA encoding the minor variant PKB␥1 could be the result of high, or even exclusive, expression in only a small subset of cells of a given tissue.
Many examples of alternate splicing of protein kinases to generate multiple variants have been documented in the literature. However, only rarely did alternate C-terminal exon usage produce variants of the same kinase with different regulatory potential. Examples include the myotonic dystrophy protein kinase gene, where the primary mRNA transcripts were spliced to generate six major isoforms that differed in the presence or absence of an internal motif and in their C-terminal ends. Thus the mRNA transcript could be spliced to result in a protein with a strongly hydrophobic C-terminal domain, but usage of a cryptic splice acceptor site led to a frameshift and to translation of a less hydrophobic C terminus. This seemingly occurred as a stochastic event, whereas the skipping of two exons, resulting in a C-terminally truncated isoform, was cell type-dependent (64). In L6 skeletal muscle cells, stimulation by insulin shifted C-terminal alternate exon usage to enhanced inclusion of the PKC␤II-specific, rather than the PKC␤I-specific, exon into the mature mRNA transcript, a process likely regulated by phosphorylation of SR proteins (65). The two PKC␤ splice variants differed in their binding to F-actin, implicating PKC␤II in the process of insulin-stimulated actin rearrangements (66). The mechanism by which C-terminal splicing of the PKB␥ gene is regulated remains to be established.
When we compared the kinase activities of the two PKB␥ splice variants, we observed that PKB␥1 could not be activated to the same extent as PKB␥ with any of the stimuli tested. This was in part due to the absence of the Ser-472 phosphorylation site and because the activation loop site Thr-305 was also phosphorylated to a lesser extent in PKB␥1 compared with PKB␥ ( Figs. 5 and 6). Kinase activity and Thr-305 phosphorylation of PKB␥1 were more sensitive to osmotic stress (Fig. 6) and less sensitive to stimulation by insulin (Figs. 5 and 6). Furthermore, less PKB␥1 than PKB␥ was localized at the membrane of transfected cells after insulin stimulation (Fig. 7). Together, these results implied that the presence of the second phosphorylation site Ser-472 positively influenced the induction or the maintenance of phosphorylation at Thr-305 and thus the activity of the kinase. This may occur through interactions of the unique termini of PKB␥ and PKB␥1 with different modulators of PKB activity, which may regulate intracellular localization and facilitate translocation of the kinase to 3  the membrane, or the phosphorylation event of the kinase itself. Alternatively, an altered sensitivity to inactivation of the two splice variants may be dependent on differing interactions with phosphatases that down-regulate kinase activity. In support of this hypothesis, it has been suggested that phosphoinositide 3-kinase activates a signal transduction pathway promoting rapid dephosphorylation and inactivation of PKB and furthermore that membrane localization led to partial protection from inactivation (67).
The potential implications for downstream signaling offered by the presence of two splice variants of PKB␥ are manifold. Although PKB␥ and PKB␥1 possess identical kinase domains, the variable C-terminal sequences could impose different constraints on access of the substrate to the kinase pocket. To investigate this possibility, we examined the relative activities of the splice variants toward a range of peptides. However, even though specific kinase activities of the splice variants were considerably different, there were no great variations in substrate preference. Nevertheless, downstream signaling of the two splice variants may be controlled by intracellular colocalization, or cell type-specific co-expression, with particular substrates.
In summary, we have presented evidence for two splice variants of human and mouse PKB␥ differing in the last coding exons, which resulted in the presence or absence of a phosphorylation site and thus in a distinct regulatory potential. Activation of PKB␥1 relied on a single phosphorylation event on Thr-305. The two splice variants differed in terms of tissue distribution and abundance of mRNA transcripts and in their potential to be activated by a range of stimuli, which was reflected in the extent of phosphorylation at the regulatory sites. Most notably, PKB␥1 had a lower specific kinase activity and translocated to the membrane to a lesser extent than PKB␥ upon insulin stimulation, indicating a role for the Cterminal regulatory phosphorylation site Ser-472 in mediating the translocation of the kinase or the phosphorylation of Thr-305.