Molecular cloning, expression and characterization of the human serine/threonine kinase Akt-3


A. Richardson, Department of Biochemistry, Janssen Research Foundation, Turnhoutseweg 30, B-2340 Beerse, Belgium. Fax: + 32 14 606515, Tel.: + 32 14 605357, E-mail:


Akt (also known as PKB or RAC-PK) is an intracellular serine/threonine kinase involved in regulating cell survival. Although this makes it a promising target for the discovery of drugs to treat human cancer, a complicating factor may be the role played by Akt in insulin signalling. Two human isoforms, Akt-1 and Akt-2, have been described previously and a third isoform has been identified in rats (here termed Akt-3, but also called RAC-PK-γ or PKB-γ). We describe the identification of the corresponding human isoform of Akt-3. The gene encoding human Akt-3 was localized to chromosome 1q43–44. The predicted protein sequence is 83% identical to human Akt-1 and 78% identical to human Akt-2, and contains a pleckstrin homology domain and a kinase domain. In contrast to the published rat Akt-3 isoform, human and mouse Akt-3 also possess a C-terminal ‘tail’ that contains a phosphorylation site (Ser472) thought to be involved in the activation of Akt kinases. In addition to phosphorylation of Ser472, phosphorylation of Thr305 also appears to contribute to the activation of Akt-3 because mutation of both these residues to aspartate increased the catalytic activity of Akt-3, whereas mutation to alanine inhibited activation. Akt-3 activity could be inhibited by the broad spectrum kinase inhibitor staurosporine and by the PKC inhibitor Ro 31-8220, but not by other PKC or PKA inhibitors tested. Although Akt-3 is expressed widely, it is not highly expressed in liver or skeletal muscle, suggesting that its principle function may not be in regulating insulin signalling. These observations suggest that Akt-3 is a promising target for the discovery of novel chemotherapeutic agents which do not interfere with insulin signalling.


extracellular matrix


expressed sequence tag


fluorescent in situ hybridization


glyceraldehyde 3-phosphate dehydrogenase


glutathione S-transferase




insulin like growth factor-1


MAPK/ERK kinase


pleckstrin homology

PtdIns 3-kinase

phosphatidylinositol 3-kinase


protein kinase A


protein kinase B


protein kinase C


reverse transcription PCR.

A characteristic feature of many cancer cells is their ability to grow independently of adhesion. In contrast, when untransformed endothelial cells are prevented from adhering to the extracellular matrix (ECM), they undergo apoptosis [1,2]. The process by which normally adherent cells are triggered to undergo apoptosis when they are unable to adhere to ECM has been termed ‘anoikis’[3]. Changes in signalling by adhesion molecules can lead to resistance to anoikis [3] and this may contribute to the mechanism whereby cancer cells that grow independently of adhesion are able to avoid anoikis. Thus, the identification of signalling pathways which suppress anoikis in tumor cells may provide novel therapeutic avenues for the treatment of human cancer.

Akt [also known as protein kinase B (PKB) or ‘related to A and C protein kinase’ (RAC-PK)] is a serine/threonine kinase that has been implicated in regulating cell survival [4–9]. Akt can inhibit apoptosis induced by detachment from ECM [4], as well as by survival factor withdrawal [5–7,10–13] or irradiation [14]. In some cell types, dominant-negative variants of Akt promote apoptosis [4,5,12,13]. Akt may suppress apoptosis by catalysing the phosphorylation of BAD, a pro-apoptotic member of the Bcl-2 protein family [8,15], by phosphorylation of caspase-9 [16], or by inhibition of the Forkhead transcription factors [17,18]. Although Akt is itself over-expressed in certain human tumors [19–21], Akt can also be activated by expression of ras [4], and activation of Akt is antagonized by the tumor suppressor PTEN (reviewed in [22]). Significantly, a mutant of ras that is unable to activate the PtdIns 3-kinase/Akt pathway is unable to transform fibroblasts unless cotransfected with an activated variant of Akt [23]. These observations suggest that Akt may be an appropriate target for the development of compounds to treat cancer.

Akt comprises an N-terminal pleckstrin homology (PH) domain involved in lipid binding, a kinase domain and a C-terminal ‘tail’. Akt is thought to be activated by recruitment to the plasma membrane and subsequent phosphorylation by two upstream kinases, PDK-1 and PDK-2 [24,25]. The binding of 3-phosphoinositides, generated by phosphatidylinositol 3-kinase (PtdIns 3-kinase), to the PH domain of Akt is believed to promote translocation to the plasma membrane and to facilitate phosphorylation of Akt-1 by PDK-1 at Thr308 [24–28] or of Akt-2 at Thr309 [29]. In addition to phosphorylation of Thr308/Thr309, full activation requires phosphorylation of the COOH tail at Ser473 in Akt-1 [26] or at Ser474 in Akt-2 [29]. Two human isoforms of Akt have been described to date, Akt-1 and Akt-2 [20,30,31]. A third isoform, here referred to as Akt-3, has been described in the rat [32]. As rat Akt-3 possesses an apparently truncated tail and thereby lacks Ser473, its regulation may differ from that of Akt-1 and Akt-2. Both Akt-1 and Akt-2 are expressed widely, although the expression of Akt-2 is most prominent in insulin-responsive tissues, such as liver and skeletal muscle [33,34]. Akt-1 and Akt-2 are activated by insulin in rat adipocytes, hepatocytes and skeletal muscle. In contrast, Akt-3 does not appear to be strongly activated by insulin in these tissues [35]. The role of the various Akt isoforms in insulin signalling may limit the utility of compounds that inhibit Akt-1 or Akt-2 activity as such agents may induce symptoms observed in patients with diabetes. We hypothesized that this problem may be avoided by using selective inhibitors of Akt-3 and this prompted us to identify the human analogue of rat Akt-3. Here, we report the molecular cloning and characterization of a cDNA that encodes the human isoform of Akt-3. Significantly, human Akt-3 possesses a C-terminal tail that contains an amino acid residue analogous to Ser473/Ser474 previously implicated in the activation of Akt-1/Akt-2, but absent in the rat Akt-3 protein. We also show that human Akt-3 appears to be activated in a manner comparable to Akt-1 and Akt-2. However, its expression profile in human tissue suggests that it may not play a major role in insulin signalling. This makes Akt-3 an attractive target for the discovery of drugs to treat human cancer.

Materials and methods

Oligonucleotide synthesis and DNA sequence determination

All primers were obtained from Eurogentec, Seraing, Belgium. Sequencing reactions were performed using BigDye™ Terminator Cycle Sequencing Ready Reaction kits (Perkin Elmer) and were run on an Applied Biosystems 377 DNA sequencer (Perkin Elmer).

Molecular cloning of human Akt-3.

Using the rat RAC-PKγ sequence ([32]; GenBank accession number D49836) as a query sequence, a blast (Basic Local Alignment Search Tool) search was carried out in the WashU Merck expressed sequence tag (EST) database and in the proprietary LifeSeq™ human EST database (Incyte Pharmaceuticals Inc, Palo Alto, CA, USA). Several human EST clones with high similarity to the rat RAC-PKγ were identified. One EST sequence (Incyte accession number 2573448) derived from a hippocampal cDNA library, contained part of the coding sequence including the putative methionine start codon (ATG). Based on this 239 bp sequence, oligonucleotide sense primers were synthesized for 3′ RACE experiments: Akt-3sp1 = 5′-ACCATTTCTCCAAGTTGGGGGCTCAG-3′ and Akt-3sp2 = 5′-GGGAGTCATCATGAGCGATGTTACC-3′. 3′ RACE experiments were performed on human fetal brain or human cerebellum Marathon-Ready™ cDNA (Clontech) according to manufacturer’s instructions using Akt-3sp1/race-ap1 as primers in the primary PCR and Akt-3sp2/race-ap2 in the nested PCR. This extended the Akt-3 coding sequence by 916 bp, but the novel sequence did not include an in-frame stop codon. A second round of 3′ RACE amplification was performed on human brain Marathon Ready™ cDNA using sense primers based on the sequence obtained in the first round (Akt-3sp3 = 5′-CACTCCAGAATATCTGGCACCAGAGG-3′ and Akt-3sp4 = 5′-CTATGGCCGAGCAGTAGACTGGTGG-3′) in combination with race-ap1 and race-ap2, respectively. The sequence obtained included an in-frame stop codon and the 3′ untranslated sequence up to the poly(A) tail. Antisense primers were designed based on the 3′ untranslated region (Akt-3ap4 = 5′-TGCCCCTGCTATGTGTAAGAGCTAGG-3′ and Akt-3ap5 = 5′-AAGAGCTAGGACTGGTGATGTCCAGG-3′) and the complete Akt-3 coding sequence was amplified from human hippocampal cDNA using Akt-3sp1/Akt-3ap4 (primary PCR) and Akt-3sp2/Akt-3ap5 (nested PCR) as primers. The resulting 1200 bp PCR fragment was then cloned in the vector pCR2.1 (Invitrogen) and the inserts of several clones were completely sequenced. The required insert was subcloned into the mammalian expression vector pcDNA-3 (Invitrogen). An haemagglutinin (HA) tag (YPYDVPDYA) was introduced by PCR after amino acid 479 of Akt-3.

Comparison of human and rat Akt-3 sequences

To investigate whether the difference between the human and the published rat Akt-3 cDNA sequence (this paper and [32]) is real or due to a cloning artefact, a forward primer common to the human and rat Akt-3 cDNA sequence (either Akt-3sp5 = 5′-TCCTCGAACACTCTCTTCAGATGC-3′ or Akt-3sp6 = 5′-GCTTTCAGGGCTCTTGATAAAGGATCC-3′) was combined with a reverse primer based on the published rat sequence (Akt-3rat-ap1 = 5′-CTTCCTGAAATTGAACCAGATTGGTCC-3′) or a degenerate reverse primer based on the human Akt-3 sequence (either Akt-3deg-ap1 = 5′-GTAGGARAAYTGNGGRAARTGNGG-3′ or Akt-3deg-ap2 = 5′-TGTGGCCGCCKYTCRTTRTCCAT-3′). PCR experiments were performed on human or rat genomic DNA or brain cDNA (Clontech) using different combinations of these primers. PCR bands of the expected size were purified from agarose gels and directly sequenced using the same primers used for their amplification as sequencing primers.

Constructs and mutants for Escherichia coli expression of Akt-3

To express the human Akt-3 protein in E. coli, the complete Akt-3 coding sequence was subcloned into pGEX-4T-3 (Amersham Pharmacia Biotech). Mutants of this construct were made using the Quickchange site-directed mutagenesis kit (Stratagene) according to the manufacturer’s instructions. The T305D mutant was created by mutating ACA at position 923–925 to GAC, resulting in a Thr305 to Asp mutation in the resulting protein. The S472D mutant was created by changing TC at position 1404–1405 to GA using PCR with an antisense primer incorporating the change, resulting in a Ser472 to Asp mutation in the resulting protein. A double mutant was also constructed by site-directed mutagenesis on the S472D construct. The inserts of all resulting constructs were confirmed by complete sequence analysis. The fusion proteins resulting from expression of these constructs in E. coli contain a GST moiety coupled to the N-terminus of the human Akt-3 sequence.

Expression and assay of wild-type and mutant Akt-3.

The pGEX expression constructs were transformed into E. coli strain BL21 DE3 and GST-fusion proteins of wild-type and mutated Akt-3 were purified on glutathione sepharose according to the manufacturer’s instructions (Amersham Pharmacia Biotech). The protein eluted from the beads was stored in 50% glycerol at −20 °C. Akt activity was assessed by incubating 0.8 µg of the purified enzyme for 30 min at room temperature (unless otherwise indicated) in a buffer containing 10 mm Hepes, 10 mm MgCl2, 1 mm dithiothreitol, 0.1 mg·mL−1 histone H2B at pH 7.0, in a total volume of 25 µL and containing 10 µCi [γ-32P]-ATP (6000 Ci·mmol−1). Initial experiments indicated that the reaction was linear with time for at least 45 min. The reaction was stopped by the addition of 25 µL sample buffer for SDS/PAGE. The results were quantified on a phosphorimager following SDS/PAGE on a 15% (w/v) acrylamide gel.

For expression in HEK-293 cells, cells were transfected with pCDNA-3 Akt-3 constructs as described previously [26]. After stimulation with IGF, the cells were lysed [26] and HA-Akt immunoprecipitated with antibody 3F10 (Roche Molecular Biochemicals). Akt activity was assessed in immune complexes by measuring phosphorylation of a peptide substrate (Crosstide) in the presence of 1 µm PKI (PKA inhibitor) and 1 µm GF109302X (PKC inhibitor) as described [26].

Chromosomal mapping studies

Chromosomal mapping studies were carried out by SeeDNA Biotech Inc, Toronto, Canada using fluorescent in situ hybridization (FISH) analysis essentially as described [36,37] using a cDNA probe corresponding to full length Akt-3.

Northern blot analysis.

Northern blots containing 2 µg of poly(A)-rich RNA derived from different human tissues (Clontech) were hybridized according to the manufacturer’s instructions with a α-[32P]-dCTP random-priming labelled (HighPrime kit, Roche Diagnostics) 454 bp NotI–XbaI Akt-3 fragment (nucleotides 1404–1857) corresponding to part of the 3′ untranslated sequence.

Reverse transcription (RT)-PCR analysis

Oligonucleotide primers were designed for the specific PCR amplification of fragments from Akt-1, Akt-2 and Akt-3, respectively. The primers were first tested to ensure specific amplification of Akt-1, Akt-2 or Akt-3 without any cross-reactivity (data not shown). Experiments were performed on human multiple tissue cDNA (MTC™) panels normalized to the mRNA expression levels of six different housekeeping genes (Clontech). PCR amplifications for human glyceraldehyde 3-phosphate dehydrogenase (G3PDH) were performed on the same cDNA samples as positive controls using G3PDH specific primers. Images of the ethidium bromide stained gels were obtained using the Eagle Eye II Video system (Stratagene) and PCR bands analysed using the eaglesight software.


Molecular cloning of human Akt-3.

Similarity searching of the LifeSeq™ and EMBL databases using the rat Akt-3 sequence as a query sequence yielded several human EST sequences which encoded part of the human homologue of rat Akt-3. Using the DNA sequence information in the databases, we were able in subsequent 3′ RACE experiments to deduce the complete cDNA sequence for the human Akt-3 (Fig. 1). The obtained cDNA sequence encoded a protein of 479 amino acid residues (calculated molecular mass of 55 770 Da) comprising a PH domain, a kinase domain and a C-terminal ‘tail’. The predicted Akt-3 protein shows significant similarity with Akt-1 ([31]; 83.6% identity) and with Akt-2 ([20]; 78% identity) (Fig. 1). The first 451 amino acids of the human Akt-3 protein contain only two differences to the rat Akt-3 sequence [32], Asp10 (rat) versus Gly10 (human) and Pro396 (rat) versus Ala396 (human); alignment of all the previously described Akt-1 and Akt-2 sequences from other species demonstrates Gly and Ala at these positions. Further evidence that we have identified the Akt-3 isoform comes from the presence of isotype-specific sequences represented by human Akt-3 residues 47–49 (LPY), 118–122 (NCSPT) and 139–141 (HHK). For each isotype, these residues are conserved between species, but differ between the isotypes.

Figure 1.

Figure 1.

    Alignment of the deduced amino acid sequences for human Akt-1, Akt-2 and Akt-3. The sequences were aligned using the clustalw alignment program (EMBL, Heidelberg, Germany). Amino acid residues conserved between all three proteins are included in the black areas. Residues conserved between only two of the sequences are shaded in grey. Amino acid residues are numbered in the right hand column. The conserved Thr and Ser residues that are presumed to be phosphorylated upon activation are marked with an asterisk above the sequence.

    The C-terminal ‘tail’ has been observed in both human and rat Akt-1 and Akt-2 proteins, but it is apparently truncated in the rat Akt-3 sequence [32]. The ‘tail’ in human Akt-3 comprises 28 amino acid residues that replace three amino acid residues in the rat sequence and includes a serine residue at position 472 that corresponds to Ser473 in Akt-1 or Ser474 in Akt-2. Phosphorylation of Ser473 and Ser474 has previously been implicated in the activation of Akt-1 and Akt-2, respectively [26,29]. To further investigate the discrepancy between the reported rat Akt-3 sequence and the human sequence reported here, we tried to identify by PCR both the truncated rat and human C-terminal sequences using rat brain and human brain cDNA as templates. The identity of PCR fragments obtained were confirmed by direct sequencing of the purified PCR bands. Using degenerate primers based on the human sequence and rat brain cDNA as template, we identified two independent PCR products both of which encoded the C-terminal domain identified in human Akt-3. Using a reverse primer based on the published rat sequence, no Akt-3 fragments could be amplified from human or rat brain cDNA or from genomic DNA. In addition, a mouse EST has been identified (EMBL database, accession number. AA929481) that also includes the putative Ser472 phosphorylation site.

    Characterization of Akt-3 activity

    To characterize the enzymatic activity of Akt-3, we expressed and purified the enzyme as a GST fusion protein. Analysis of the purified product by SDS/PAGE indicated the protein was apparently > 90% pure. The purified enzyme was able to phosphorylate histone H2B (Fig. 2), and no phosphorylation was observed using recombinant GST alone. Previously, the enzymatic activity of Akt-1 has been shown to be increased by phosphorylation of Thr308 and Ser473, and mutation of both these residues to Asp (to mimic phosphorylation) synergistically activates Akt-1 [26]. To investigate whether Akt-3 is similarly regulated, GST-fusion proteins in which either Thr305 or Ser472 (corresponding to Thr308 and Ser473 in Akt-1) or both Thr305 and Ser472 had been mutated to Asp were expressed and assayed in comparison to the wild-type enzyme. Mutation of Thr305 to Asp (‘T305D’) resulted in a 2.0-fold increase in the initial rate of phosphorylation of histone H2B, whereas mutation of Ser472 to Asp (‘S472D’) increased the initial rate only 1.4-fold (Fig. 2A). When both Thr305 and Ser472 (‘T305D/S472D’) were mutated to Asp, a 3.2-fold increase in the initial phosphorylation rate was observed.

    Figure 2.

    Figure 2.

      Expression and activation of Akt-3 variants. (A) Akt-3 was expressed as a GST fusion protein in E. coli. To assess Akt-3 activity, Histone H2B was incubated with GST-Akt-3 and GST-Akt-3 variants for the indicated time and the extent of phosphorylation assessed after SDS/PAGE. The variants of Akt-3 are designated: W.T., wild type; T305D, Thr305 mutated to Asp; S472D, Ser472 mutated to Asp; T305D/S472D, both Thr305 and Ser472 mutated to Asp. No significant phosphorylation was observed when GST was used in place of GST-Akt. The results are the mean (± SEM; n = 3–6) and are expressed relative to the extent of phosphorylation of H2B catalysed by T305D/S472D hAkt-3 after 45 min. Insert: The purity of the purified GST (lane1), wild-type Akt-3 (lane 2), T305D Akt-3 (lane 3), S472D Akt-3 (lane 4) or T305D/S472D Akt-3 (lane 5) was assessed by SDS/PAGE and by Coomassie blue staining. (B) HEK-293 cells were transfected with either vector (lanes 1 and 2) or Akt-3 (lanes 3 and 4), Akt–3 T305A (lanes 5 and 6) or Akt-3 S472A (lanes 7 and 8) and either treated with buffer (lanes 1,3,5,7) or IGF-1 (50 ng·mL−1; lanes 2,4,6,8). Akt-3 was immunoprecipitated with antibody 3F10 (anti-HA tag). Samples were analysed by blotting for the HA-tag (upper panel) or with a phospho-specific antibody which recognizes phosphorylated Ser472 (lower panel). (C) Akt activity in HA-immunoprecipitates from samples prepared as described above was assessed by measuring phosphorylation of a peptide substrate (Crosstide). The results are expressed as the increase in activity compared to unstimulated cells transfected with empty vector (mean ± SEM, n = 7).

      To confirm that extracellular stimuli can activate Akt-3 in mammalian cells, HEK293 cells were transfected with a cDNA encoding Akt-3 fused to a HA epitope tag. Upon treatment with IGF, Akt-3 activity in anti-HA immunoprecipitates (Fig. 2B) was increased almost 60-fold above that in untransfected cells (Fig. 2C). Akt variants in which Thr305 and Ser472 were mutated to alanine were refractory to activation by IGF. Consistent with this, Western blotting with a Ser472 phospho-specific antibody of HA immunoprecipitates from cells stimulated with IGF demonstrated that Ser472 was phosphorylated following stimulation with IGF (Fig. 2B). In addition, activation of Akt-3 was inhibited by prior treatment with LY294002 (100 µm, 94% inhibition, data not shown).

      To characterize human Akt-3 further, we investigated the ability of a range of Ser/Thr kinase inhibitors to inhibit Akt-3. These included Go 6976, GF-109203X [both protein kinase C (PKC) inhibitors]; H-85, H-88, H-89 and KT5720 [protein kinase A (PKA) inhibitors], KN-62 (Ca2+/calmodulin dependent kinase inhibitor) and PD 98059 (MEK inhibitor). When tested at a concentration of 10 µm these compounds had no significant effect on the activity of the T305D/S472D variant of Akt-3. However, the broad spectrum kinase inhibitor staurosporine (IC50 = 2.0 ± 0.3 µm) and the PKC inhibitor Ro 31-8220 (IC50 = 3.2 ± 1.0 µm) inhibited theT305D/S472D variant of Akt-3 (Fig. 3).

      Figure 3.

      Figure 3.

        Inhibition of Akt-3 by staurosporine and Ro 31-8220. Histone H2B was treated with Akt-3 (T305D/S472D variant) in the presence of the indicated concentrations of either staurosporine (upper graph) or Ro 31-8220 (lower graph). After 30 min, the reaction was terminated and the extent of H2B phosphorylation quantified on a phosphorimager following SDS/PAGE. The results (mean ± SEM, n = 3) are expressed as percentage of the phosphorylation observed in the presence of solvent (control, ‘C’).

        Chromosomal localization of Akt-3

        The complete coding sequence of Akt-3 was used as a probe for FISH analysis. Akt-3 was located on human chromosome 1, region q43-q44 (Fig. 4). There was no additional locus identified by FISH detection.

        Figure 4.

        Figure 4.

          Chromosomal localization (A) and FISH mapping (B) of human Akt-3. (A) Diagram of FISH mapping results for Akt-3. Each dot represents the double FISH signals detected on human chromosome 1, region q43-q44. (B) Example of FISH mapping of Akt-3. The left panel shows the FISH signals on chromosome 1. The right panel shows the same mitotic figure stained with 4′,6-diamidino-2-phenylindole to identify chromosome 1.

          Tissue distribution of Akt-3 mRNA

          Northern blot analysis was performed on mRNA derived from different human tissues. Akt-3 mRNA was detected as two transcripts of approximately 4.5 kb and 7.5 kb, showing similar patterns of expression (Fig. 5A). Akt-3 mRNA was expressed in a range of tissues, most prominently in brain. Similarly, rat Akt-3 was detected as multiple transcripts most highly expressed in brain [32]. The weakest expression of Akt-3 was observed in two insulin-responsive tissues, skeletal muscle and liver. Akt-3 was also expressed in a number of cancer cell lines including SW480 colorectal adenocarcinoma, A549 lung carcinoma and G361 melanoma (data not shown).

          Figure 5.

          Figure 5.

            Expression of Akt-3 in different humantissues. (A) Northern blot analysis of tissue expression of Akt-3. The expression of Akt-3 mRNA in different human tissues was assessed using a probe corresponding to the 3′ untranslated region of Akt-3 to analyse a blot of human poly(A) rich RNA (‘Multiple Tissue Northern’). Human β-actin was used as a control to confirm equal loading of the lanes (data not shown). (B) RT-PCR analysis of tissue expression of Akt-1, Akt-2 and Akt-3. RT-PCR analyses were performed on cDNA from different human tissues using primers specific for human Akt-1, Akt-2, Akt-3 or G3PDH (control) for the indicated number of PCR cycles. Bands of the expected size are visible on the gels. The images from the ethidium bromide stained 1.2% agarose gels were inverted for clarity using the eaglesight software (Stratagene). The results from similar PCR reactions performed for 25, 30 or 35 cycles are not shown but indicated that the results from this figure are in the linear range of amplification.

            To confirm the Northern blot analysis and to compare the expression of Akt-3 with other Akt isoforms, reverse transcription PCR analysis was conducted with Akt-1, Akt-2 and Akt-3. The relatively high homology between Akt-1, Akt-2 and Akt-3 precluded the design of suitable probes for a Northern analysis. PCR reactions were performed with Akt specific and G3PDH-specific (internal control) primers on cDNAs derived from different human tissues (Fig. 5B). The Akt-3 message was detected in every tissue tested, as a specific 425 bp fragment was amplified in every cDNA after 30 cycles of PCR. Akt-3 mRNA expression was highest in placenta, ovary and spleen and in confirmation of the results obtained by Northern analysis. Significantly, expression was lowest in liver, lung and skeletal muscle. For comparison purposes we investigated the expression of Akt-1 and Akt-2, and these were widely expressed (although expression of Akt-1 was low in skeletal muscle and brain), consistent with reported data [33].


            Akt-1 and Akt-2 have been identified in several species. Human [30,31] mouse [38] and bovine [30] Akt-1 clones have been reported, whereas human [20] mouse [34] and rat [33] clones of Akt-2 have been identified. However, Akt-3 has only been previously identified in rat [32]. We have identified the human isoform of Akt-3. Although human Akt-3 shows considerable similarity to human Akt-1 and Akt-2, the discovery of human Akt-3 is particularly significant because the cDNA sequence encodes a C-terminal ‘tail’ which includes a phosphorylation site implicated in the activation of Akt-1 and Akt-2 [26,29]. This tail is absent from the published rat amino acid sequence. We have shown here that this is probably due to a cloning artefact and that human, rat and mouse Akt-3 all contain the Ser472 phosphorylation site. Akt-3 appears to be activated by phosphorylation in a similar fashion as Akt-1 and Akt-2. However, its expression profile suggests that the principal function of this enzyme is not in regulating responses to insulin.

            The human Akt-3 cDNA sequence was predicted to encode a N-terminal pleckstrin homology (PH) domain [39] and a C-terminal kinase domain [40]. A striking difference between the human and rat Akt-3 protein sequence [32] is the presence of a C-terminal ‘tail’ comprising 74 residues after the kinase domain. The last 28 amino acid residues in human Akt-3 are absent from the rat Akt-3 sequence and encompass a stretch of 10 residues (residues 467–476 in human Akt-3) which are identical to the corresponding region of human Akt-1 and Akt-2. These residues include Ser472 (which corresponds to Ser473 in Akt-1), a residue critical for the regulation of Akt activity [26]. This suggests that the sequence we have identified is authentic and that human Akt-3 is potentially regulated in a manner similar to Akt-1 and Akt-2. Using rat brain cDNA as a template, we were able to identify rat Akt-3 clones which possessed a C-terminal sequence analogous to that observed in human Akt-3. When we used primers corresponding to the published rat Akt-3 sequence, we were unable to obtain a PCR product. These observations suggest that the C-terminal truncation of the rat Akt-3 sequence most probably represents a cloning artefact.

            To verify that the predicted kinase domain was catalytically active, we expressed Akt-3 as a GST fusion protein in E. coli. The purified protein was able to phosphorylate an exogenous substrate, whereas no catalytic activity was observed using GST in place of GST-Akt-3. To confirm that Akt-3 is indeed regulated in a manner akin to Akt-1 and Akt-2, we mutated Thr305 and Ser473, either separately or jointly, to Asp. This strategy has previously been shown to faithfully mimic the effect of phosphorylation of these residues in Akt-1 [26]. Mutation of either of these residues resulted in increased activity, although the increase was less than that observed with Akt-1 [26]. Additionally, we did not observe a synergistic activation of Akt-3 by mutation of both Thr305 and Ser473. In contrast, when both the corresponding residues were simultaneously mutated to Asp in Akt-1, synergistic activation was observed [26]. The apparent quantitative differences between Akt-1 and Akt-3 may reflect true differences in the regulation of these two isoforms, or it may be due to other factors such as the different expression system used. Nevertheless, our results demonstrate that Akt-3 is qualitatively regulated in a fashion similar to Akt-1. The importance of phosphorylation of Thr305 and Ser472 was confirmed by expression of Akt-3 in HEK-293 cells. When these residues were mutated to Ala, activation of Akt-3 by IGF was blocked. Previous work has also shown that activation of Akt is dependent upon PtdIns 3-kinase to generate 3-phosphoinositides that bind the PH domain of Akt, promote translocation of Akt to the plasma membrane and facilitate the phosphorylation of Akt by upstream kinases (reviewed in [24,25]). Our observation that the T305D/S472D mutant of Akt-3 is more active than the wild type enzyme (Fig. 3), when measured in the absence of 3-phosphoinositides, suggests that after phosphorylation Akt-3 becomes at least partially independent of phosphoinositide binding.

            The structure of the catalytic domain of Akt is closely related to protein kinase A and protein kinase C. This prompted us to investigate whether existing inhibitors of PKA or PKC, as well as other serine/threonine kinase inhibitors, could be used as inhibitors of Akt-3. Of the compounds tested, only staurosporine and the structurally related compound Ro 31-8220 potently inhibited Akt-3. Staurosporine is a nonselective kinase inhibitor, whereas Ro 31-8220 is a more selective PKC inhibitor [41]. Although Ro 31-8220 is an approximately 100-fold more potent (IC50 ≈ 10 nm[41]) inhibitor of PKC than of Akt-3, this observation cautions that experiments using high concentrations of Ro 31–8820 may affect Akt-3. In contrast to staurosporine and Ro 31-8220, two other PKC inhibitors and three other PKA inhibitors did not inhibit Akt-3. This suggests that although Akt-3 is closely related in sequence to PKC, it may be possible to find selective inhibitors of Akt.

            The observation that Akt-3 is activated by IGF-1 suggests that Akt-3 may play a role in regulating cell survival. One concern in using Akt as a target for drug development in cancer is that Akt plays a role in insulin signalling. Thus, inhibitors of Akt may induce the symptoms associated with diabetes. One solution that has been proposed is to develop selective inhibitors of Akt-2 [35]. This is based in part on the observation that Akt-1 is strongly activated by insulin in rat hepatocytes and skeletal muscle, whereas Akt-2 is only weakly activated by insulin in these tissues. However, rat Akt-3 appears to be even more weakly activated by insulin in these tissues [35], and in this study we have shown that Akt-3 mRNA is expressed only at low levels in human liver and skeletal muscle, which are insulin-responsive tissues. This suggests that selective inhibitors of Akt-3, rather than Akt-2, could have less potential to cause symptoms analogous to those observed in diabetes. The localization of human Akt-3 to human chromosome 1q43–44 is also interesting, as patients with haematological cancers have been reported with chromosomal abnormalities in this region [42]. Although the significance of the latter observation is debatable, as chromosomal abnormalities at numerous loci have been observed in patients with haematological cancers, the results presented here indicate that Akt-3 may prove to be an important target for the development of novel therapeutics for the treatment of cancer. While this paper was under review, clones encoding human Akt-3 were reported elsewhere [43,44].


            We wish to thank Natalie Delcroix and Petra De Wilde for assistance in DNA sequencing.