HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 275, Issue 46, 36108-36115, November 17, 2000

This Article Abstract Full Text (PDF) All Versions of this Article: 275/46/36108    most recent M005497200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Obata, T. Articles by Cantley, L. C. Search for Related Content PubMed PubMed Citation Articles by Obata, T. Articles by Cantley, L. C. Social Bookmarking                      What's this?

Peptide and Protein Library Screening Defines Optimal Substrate Motifs for AKT/PKB^*

Toshiyuki Obata^§¶**, Michael B. Yaffe^§§§, German G. Leparc^§, Elizabeth T. Piro^§, Hiroshi Maegawa^¶, Atsunori Kashiwagi^¶, Ryuichi Kikkawa^¶, and Lewis C. Cantley^§¶¶

From the Departments of  Medicine and  Surgery, Beth Israel Deaconess Medical Center and the ^§ Department of Cell Biology, Harvard Medical School, Boston, Massachusetts 02215 and the ^¶ Third Department of Medicine, Shiga University of Medical Science, Otsu, Shiga 520-2192, Japan

Received for publication, June 22, 2000

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

AKT was originally identified as a proto-oncogene with a pleckstrin homology and Ser/Thr protein kinase domains. Recent studies revealed that AKT regulates a variety of cellular functions including cell survival, cell growth, cell differentiation, cell cycle progression, transcription, translation, and cellular metabolism. To clarify the substrate specificity of AKT, we have used an oriented peptide library approach to determine optimal amino acids at positions N-terminal and C-terminal to the site of phosphorylation. The predicted optimal peptide substrate (Arg-Lys-Arg-Xaa-Arg-Thr-Tyr-Ser*-Phe-Gly where Ser* is the phosphorylation site) has similarities to but is distinct from optimal substrates that we previously defined for related basophilic protein kinases such as protein kinase A, Ser/Arg-rich kinases, and protein kinase C family members. The positions most important for high V_max/K_m ratio were Arg-3>Arg-5>Arg-7. The substrate specificity of AKT was further investigated by screening a GEX phage HeLa cell cDNA expression library. All of the substrates identified by this procedure contained Arg-Xaa-Arg-Xaa-Xaa-(Ser/Thr) motifs and were in close agreement with the motif identified by peptide library screening. The results of this study should help in prediction of likely AKT substrates from primary sequences.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The AKT protein kinase (also referred to as protein kinase B or Rac-protein kinase) was initially identified as an acute transforming component of the AKT8 virus isolated from a murine T cell lymphoma (1, 2). The catalytic domain of AKT is closely related to those of protein kinase C (PKC)¹ and protein kinase A (PKA) family members (3-5). The kinase activity of AKT is stimulated by a variety of growth factors (6), cytokines, chemokines, heat shock, hyperosmolarity, hypoxia, integrin engagement, and T cell receptor (7-15). The pleckstrin homology domain of AKT binds to the lipid products of phosphoinositide 3'-kinase (PI3K) and thereby mediates recruitment to the membrane in response to PI3K (6, 16-18). The translocation of AKT allows phosphorylation at Thr-308 by another Ser/Thr kinase, 3-phosphoinositide-dependent protein kinase 1 (PDK-1) (19-21). Full activation of AKT requires phosphorylation at Ser-473. Although there may be multiple mechanisms for phosphorylation at Ser-473, recent studies indicate that phosphorylation is enhanced by PDK-1-dependent phosphorylation of Thr-308 and is dependent on a kinase-active AKT supporting an autophosphorylation model (22-23).

Given the importance of AKT in a variety of cellular functions, numerous laboratories have sought in vivo substrates of AKT that could explain its various roles in signaling. However, identification of in vivo substrates of protein kinases is complicated by the existence of protein kinase cascades. Thus, even if in vivo phosphorylation of a specific site on a protein is blocked by disrupting the function of a single protein kinase, one cannot conclude that the kinase of interest directly phosphorylates the site. A downstream kinase could be responsible. Support for direct phosphorylation can be obtained by demonstrating that the site is preferentially phosphorylated by the kinase of interest in a pure in vitro assay. Knowledge of the optimal motif of a protein kinase can also accelerate discovery of targets by allowing prediction of sites from a global search of genome sequences or by a restricted search of candidate proteins. Ultimately, such predictions must be tested by both in vivo and in vitro experiments.

Recently, a number of in vivo and/or in vitro studies have suggested substrates for AKT, including BAD (24-25), glycogen synthase kinase 3 (GSK3) (26), 6-phosphofructo-2-kinase (27), caspase-9 (28), endothelial nitric-oxide synthase (29-30), IB kinase (31-32), phosphodiesterase 3B (33), rac1 (34), raf-1 protein kinase (35-36), mammalian target of rapamycin (37), breast cancer susceptibility gene 1 (BRCA1) (38), insulin receptor substrate 1 (39), and fork head transcription factors (40-41). The sites identified in these substrates define a minimal motif, Arg-Xaa-Arg-Xaa-Xaa-(Ser/Thr), that is similar to motifs identified for some of the PKC family members (42).

In this study, to better understand the substrate specificity of AKT and compare it to other basophilic protein kinases, we have used a degenerate peptide library approach that provides an unbiased evaluation of the importance of the residues surrounding the site of phosphorylation for an optimal substrate. To further test the specificity in a protein context and to identify candidate substrates, we performed phosphorylation screening of a  phage cDNA expression library for AKT using purified protein kinase and [-³²P]ATP, and searched for evidence of AKT phosphorylation motifs within these newly identified substrates.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials-- HA-tagged mouse AKT-1 (43) was kindly provided by Dr. A. Bellacosa (Fox Chase Cancer Center, PA). His₆-tagged AKT-1 was generated by a PCR reaction with use of the primers 5'-ATGAGAGGATCGCATCACCATCACCATCACTACCCATACGATGTTCCAGATTAC-3' and 5'-GTTGTAAAACGACGGCCAGT-3' and HA·AKT-1 subcloned in pBluescript-KS+ (Stratagene, La Jolla, CA) as a template. The amplified PCR products were subcloned into pCR2.1 (Invitrogen, San Diego, CA) using the TA-cloning kit according to manufacturer's instructions. GST·p110 baculovirus was generously provided by Dr. Andrea Musacchio (Harvard Medical School, Boston, MA). Aminoethylsulfonyl fluoride (AEBSF) was purchased from ICN Biomedicals (Aurora, OH), P81 paper was from Whatman (Rockland, MA), [-³²P]ATP was from Dupont NEN (Boston, MA), ferric iminodiacetic acid (ferric IDA) beads were from Pierce (Rockford, IL). The degenerate peptide library and substrate peptides were synthesized using N--FMOC-protected amino acids and standard 1-benzotriazolyloxy-trisdemethyl-amino-phosphoniumhexafluorophosphate (BOP/HOBt) N-hydroxybenzotriazole-coupling methods using a BioSynthesizer (Minipore Model) as described previously (44). All other reagents were of analytical grade from Sigma.

Isolation of the PDK-1 cDNA-- The partial nucleotide sequence containing a complete open reading frame of human PDK-1 cDNA (AF017995) was amplified from total RNA of HepG2 cells by reverse transcription-PCR reaction with the primers 5'-GCCCATGGCCAGGACCACCAGC-3' and 5'-TCACTGCACAAGCGCGTCCG-3', and the fragment was subcloned into pCR2.1 by TA cloning. The 1.6-kb fragment obtained by PCR was used as a probe for standard phage plaque hybridization of a human fetal brain cDNA library (Stratagene). Five ZAP II phages (Stratagene) containing full-length human PDK-1 cDNA were obtained and in vitro excision was performed using a helper phage (ExAssist^TM) (Stratagene). The full-length cDNA was subcloned into pBluescript, and a FLAG epitope (MDYLDDDDK) was added by PCR using the primers 5'-CATGGACTACAAGGACGACGATGACAAGGCCCATGGCCAGGACCACCAG-3' and 5'-TCACTGCACAGCGGCGTCCG-3'. The PCR product (PDK1 residues 2-556 with the N-terminal FLAG tag) was subcloned into pCR2.1, and confirmed by DNA sequence analysis.

Baculovirus Constructs-- Subcloned cDNA of FLAG·PDK-1 in pCR2.1 was digested with EcoRI and subcloned into baculovirus transfer vector pVL1393 (PharMingen, San Diego, CA). His-tagged mouse AKT-1 (His·AKT) was also subcloned into pVL1393 by the EcoRI site. These plasmids were co-transfected with Baculo-gold linearized DNA into Sf-9 insect cells according to the manufacturer's instruction (PharMingen).

Generation and Purification of Enzymatically Active AKT-- Highly enzymatically active AKT was produced in SF-9 insect cells by co-infection of 3 recombinant baculoviruses encoding GST·p110 of PI3K (GST-tagged catalytic subunits of PI3K), FLAG·PDK-1, and His·AKT at multiplicities of infection (MOI) of 3, 3, and 10, respectively. Infected SF-9 cells were harvested after 60 h, lysed with lysis buffer containing 50 mM Tris-HCl buffer, pH 7.5, 1% Nonidet P-40, 10% glycerol, 137 mM NaCl, 50 mM -glycerophosphate, 50 mM NaF, 0.5 mM EGTA, 4 µg/ml AEBSF, 4 µg/ml aprotinin, 4 µg/ml pepstatin A, and 4 µg/ml leupeptin. These lysates were kept on ice for 20 min and centrifuged at 15,000 × g for 10 min and the His·AKT was then purified with a Ni²⁺-chelating affinity column (Ni²⁺-nitrilotriacetic acid-agarose beads, Qiagen, Valencia, CA). Following extensive washing with a LiCl wash buffer (lysis buffer plus 20 mM imidazole and 0.5 M LiCl), active AKT was eluted using lysis buffer containing 250 mM imidazole. Neither FLAG·PDK-1 nor GST·p110 was detectable in the purified AKT preparation by immunoblotting with anti-FLAG (M2) or anti-GST antibodies (data not shown). The specific activity of purified AKT was enhanced by 6-10-fold in the AKT/PDK-1/p110 triply infected Sf-9 cells compared with infection of AKT alone (data not shown).

Peptide Library Screening-- Peptide library screening was used to determine the optimal AKT kinase substrate motif as described previously (42, 45). Briefly, a degenerate peptide library containing peptides of the general sequence MAXXXXRXXSXXXXAKKK, where X indicates all amino acids except Cys were used. In initial experiments, Ser, Thr, or Tyr were omitted from the degenerate positions to restrict phosphorylation solely to the fixed Ser residue (total library degeneracy = 1.1 × 10¹² distinct sequences). Subsequent motif refinement was performed using the same degenerate peptide library except that Ser, Thr, and Tyr were included in the degenerate positions (total library degeneracy = 6.1 × 10¹² distinct sequences). Peptides (~1 mg) were incubated with recombinant active AKT in the presence of 100 µM unlabeled ATP and a trace amount of [-³²P]ATP at 30 °C for 2 h until ~0.5-1% of the peptides were phosphorylated. The phosphopeptides were separated from the non-phosphorylated peptides using a ferric-IDA column, and the mixture of phosphopeptides was sequenced in batch by automated Edman degradation using an Applied Biosystems process 4-cartridge protein microsequencer (Applied Biosystems Co., Foster City, CA). Data analysis was performed as described previously (44). Relative preference values were calculated by comparing the abundance of each amino acid within each degenerate position in the recovered sample to that in the starting library.

AKT Assay-- AKT activity was assayed in vitro using synthetic peptides. The reaction mixture (30 µl) contained 20 mM HEPES, pH 7.4, 100 µM ATP with 3 µCi of [-³²P]ATP, 10 mM dithiothreitol, 10 mM MgCl₂, 0.5 mM EGTA, and indicated amounts of synthetic substrate peptide. Reactions were begun by adding 0.1 µg of purified recombinant AKT and incubated at 25 °C for the indicated times. Reactions were terminated by adding 10 µl of stop solution containing 8 N HCl and 1 mM ATP. Aliquots were spotted onto P81-phosphocellulose paper, washed four times in 0.5% phosphoric acid, and quantitated by scintillation counting. For each experimental condition, values for control reactions lacking substrate peptides were subtracted as blanks. For the determination of K_m and V_max, we verified that the reaction rates were linear with respect to time for all conditions of peptide employed, corresponding to less than 10% peptide substrate phosphorylation.

Phosphoamino Acid Analysis-- For analysis of phosphoamino acid content, 2 µl of a 5 mM substrate peptide solution (final 50 µM, either AKTide-2T or AKTide) was added to reaction mixtures (98 µl) containing 20 mM HEPES, pH 7.4, 100 µM ATP, 1 µCi of [-³²P]ATP, 10 mM dithiothreitol, 10 mM MgCl₂, 0.5 mM EGTA, and ~0.2 µg of recombinant AKT on Ni²⁺-NTA-agarose beads, followed by incubation at 25 °C for the indicated times. After a brief centrifugation to remove bead-bound AKT, the supernatants were quenched using 8 N HCl, and the peptides were separated from unincorporated ATP by gel filtration with Sephadex G-25 (Amersham Pharmacia Biotech) saturated with 25 mM ammonium bicarbonate, pH 8.0. Samples were dried in a Speed Vac apparatus (Savant, Holbrook, NY), resuspended in 50 µl of 6 N HCl and incubated at 110 °C for 90 min. The samples were again evaporated to dryness, dissolved in 5 µl of distilled water, and applied to a cellulose thin layer chromatography plate (Merck, Gibbstown, NJ) along with 2 µg each of phospho-Ser, phospho-Thr, and phospho-Tyr standards. The plate was saturated with buffer containing 5% acetic acid, 0.5% pyridine, and subjected to horizontal electrophoresis at 1,000 V for 60 min using the same buffer. Standards were visualized by ninhydrin staining, and the phosphoamino acid samples were identified by autoradiography.

Phosphorylation Screening-- Solid-phase phosphorylation screening as originally developed by Fukunaga and Hunter (46) was used to identify novel AKT substrates. Briefly, a human HeLa cell cDNA library in GEX5 phage, (with each encoded cDNA expressible as a GST fusion protein) was plated on Escherichia coli strain BB4 at a density of 1.5 × 10⁴ plaques per 150-mm agar plate. After incubation for 3.5 h at 37 °C, the plates were overlaid with BAS-85 nitrocellulose membrane filters (Schleicher & Schuell, Keene, NH) impregnated with 10 mM isopropyl-thiogalactopyranoside. After an additional 10-h incubation at 37 °C, the filters were immersed in blocking buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 3% bovine serum albumin and were gently agitated at room temperature for 60 min. The filters were washed three times for 20 min in wash buffer containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM EDTA, 1 mM EGTA, 0.5% Triton X-100, 1 mM dithiothreitol, 4 µg/ml AEBSF and were rinsed for 10 min in reaction buffer containing 20 mM HEPES-NaOH, pH 7.5, 10 mM MgCl₂, 10 mM -glycerophosphate, 5 mM NaF, 2 mM dithiothreitol, 0.1% Triton X-100. The filters were then incubated for 60 min at room temperature in reaction buffer containing 5 µM unlabeled ATP, washed for 10 min in the same buffer without ATP, and finally incubated for 60 min at room temperature with gentle shaking in a reaction mixture containing 5 µM unlabeled ATP, 5 µCi/ml [-³²P]ATP and 1 µg/ml purified AKT. The filters were then washed six times for at least 5 min/wash in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM EDTA, 1 mM EGTA, 20 mM NaF, 10 mM -glycerophosphate, and 0.1% Triton X-100 and then were exposed to x-ray film. After autoradiography, positive clones were plaque purified by secondary screening and the phage DNA was prepared by the plate lysate method. The phage DNA was digested with NotI and the cDNA-containing plasmid (pGEX-PUC-3T) (46) was recovered by self-ligation followed by transformation of E. coli XL-1 Blue MRF' (Stratagene). Plasmids recovered by this technique were subjected to DNA sequencing. In this screening, a GEX phage encoding the AKT phosphorylation motif of GSK3 (GRARTSSFAEP, phosphorylatable Ser denoted by an underline) was used as a positive control, and a phage encoding partial cDNA of human granulocyte colony-stimulating factor receptor was used as a negative control (46).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Determination of the Optimal Substrate Sequence for AKT-- To determine the optimal substrate sequence for AKT, we used an oriented degenerate peptide library technique (47). Initial experiments, performed using libraries containing a variety of different orienting amino acids in addition to the fixed Ser residue, revealed that efficient peptide phosphorylation by AKT required an Arg residue in the phospho-Ser-3 position (data not shown). Therefore, the initial screening was performed using a library containing the degenerate sequence XXXXRXXSXXXX, where X denotes all amino acids except Cys, Ser, Thr, and Tyr. The latter 3 amino acids were omitted to limit phosphorylation to a single fixed phosphorylation site (Ser). This screen revealed strong selection for additional Arg residues in the S5 and S7 positions. Because several known AKT substrates contain either a Ser or Thr in the 2 position, we performed a broader secondary screening of AKT using the same library orientation, but including Ser, Thr, and Tyr in each degenerate position (Table I). Data from this screen was essentially identical to that obtained in the initial screen but also revealed moderate selection for Thr in the 2 position and for aliphatic and aromatic residues such as Tyr and Phe in both the S1 and S+1 position (Table I). There was also a slight selection for residues that allow tight turns (Gly, Ser, Asn, Thr) at the S+2 position.

View this table: [in this window] [in a new window]   Table I Amino acids selected in a peptide library screen for AKT substrates Recombinant baculovirus-derived Akt-1, activated by co-infection of both PI3K and PDK-1, was used for peptide library screening as described under "Experimental Procedures." The experiment was performed three times, and results from a representative experiment are shown. Bold letters indicate residues strongly selected. Positions 3 and 0 contained fixed R and S residues, respectively. The first number in parenthesis is the selectivity value as previously defined (Songyang et al. (45)), while the second number indicates the relative change in selectivity value compared with the previous degenerate cycle. No ratio change is shown in the p-7 position since the previous cycles are not degenerate.

The results of the peptide library screen are consistent with the frequencies of amino acids within mapped AKT phosphorylation sites in previously identified substrates (Table II). All known substrates contain Arg in the p3 position and p5 position. Most have hydrophobic amino acids at the p+1 position, in excellent agreement with the peptide library results. In addition, approximately half of the sites have Ser or Thr at the S2 position. Also, the most frequent residues in the p+2 position in natural substrates are residues that form tight turns (Pro, Ser, Asn, Gly, and Asp), also in agreement with the library results. Thus, protein substrates of AKT appear to have evolved nearly optimal motifs surrounding the site of phosphorylation.

View this table: [in this window] [in a new window]   Table II Most frequent amino acids in previously mapped phosphorylation sites of AKT protein substrates AKT phosphorylation sites of the following proteins were used: human GSK3, human PFK2, mouse IRS-1, human Raf-1, human BAD, human caspase-9, human eNOS, human IKK, human FRAP/mTOR, mouse BRCA1, Daf-16/FKHR, mouse PDE3B, human Histone H2B. GSK-3, glycogen synthase kinase 3; PFK2, 6-phosphofructo-2-kinase; eNOS, endothelial nitric oxide synthase; IKK, IB kinase; PDE3B, phosphodiesterase 3B, FRAP; FKBP-12-rapamycin associated protein, mTOR; mammalian target or rapamycin, BRCA1; breast cancer susceptibility gene 1.

Screening of a HeLa Cell cDNA Expression Library for Proteins Phosphorylated by AKT in Vitro-- To further characterize the substrate specificity of AKT, we screened a GEX5 phage expression library containing human HeLa cell cDNA inserts subcloned downstream of GST. This library had been previously used by Fukunaga and co-workers (46, 48) to identify novel substrates of Erk1 and Cdk2.

An example of positively selected phage plaques that were phosphorylated on filters by AKT using this approach is shown in Fig. 1A. Positive (GST·GSK3) and negative (GST·GCSFR) controls are also presented (Fig. 1B). Phage-expressing GST fusion proteins that could be phosphorylated by AKT underwent rounds of plaque purification and repeated screening (Fig. 1B). Typically, tertiary screening was sufficient to obtain a pure phage population containing a single insert.

View larger version (50K): [in this window] [in a new window]   Fig. 1.   Phage plaques that express proteins phosphorylated by AKT. A, typical autoradiogram of a primary screen of a HeLa cell cDNA expression library (1.5 × 104 pfu/150-mm plate) is shown. The arrowheads indicate clones that were phosphorylated by AKT and confirmed by secondary screens. B, GEX phage that encodes the AKT phosphorylation motif of GSK3 (GRARTSSFAEP) was used as a positive control, and a phage encoding a partial cDNA of GCSF-R was used as a negative control (upper). Autoradiograms of secondary and tertiary screens are shown (middle and lower). The arrowheads also indicate positive clones that were confirmed after further analysis.

Of the 3 × 10⁵ clones screened, approximately 300 clones were isolated as putative in vitro AKT substrates, and those encoding inserts that extended more than ~5 kDa beyond the GST region (70 clones) were selected for further study. Phage inserts were excised and religated to generate plasmids encoding GST fusions of these proteins (46) and were evaluated by DNA sequence analysis. Following plasmid transformation into XL1blue MRF' bacterial cells, the GST fusion proteins were expressed and purified on glutathione agarose resin. All of the proteins identified in this screen could be reproducibly phosphorylated by recombinant AKT in vitro, whereas there was no phosphorylation of GST·GCSF-receptor (46) expressed from a control plasmid (data not shown). DNA sequencing and evaluation of open reading frames revealed that most of the positive clones encoded RXRXX(S/T) motifs as summarized in Table IV. Furthermore, frequencies of amino acids at positions surrounding the likely AKT phosphorylation sites were in good agreement with the results of the peptide library screening and the peptide substrates of AKT (Table III).

Most of the positive clones were found to correspond to spliceosomal or ribosomal proteins such as heterogeneous ribonucleoprotein A1 (hnRNP A1) or ribosomal L37 protein. One of the clones encoded a transcription factor (HMG box containing protein 1, HMG protein 1), and three clones corresponded to membrane proteins (neuronal axonal membrane protein 22, Pinin A, and Lamin A). Two clones encoded different sizes of an uncharacterized small G-protein, NGB (novel GTP-binding protein). Five clones appear to be completely novel, showing no significant homology to any previously characterized proteins in the GenPept database. Twenty-five clones were the result of out-of-frame fusions of known cDNAs with the GST coding sequence.

Table IV shows a sequence alignment of the most likely AKT phosphorylation motifs contained within these positive clones. Several clones were found to contain multiple AKT phosphorylation motifs located within 15 amino acids of each other, such as the protein Zis (SRRSRSRSRSSSSSQ, possible sites of AKT phosphorylation are underlined).

Comparison of the Phosphorylation Efficiency and Determination of K_m and V_max Values for Synthetic Peptides-- Synthetic peptides corresponding to the optimal consensus sequences for AKT phosphorylation were synthesized and defined as "AKTides" (Fig. 2A). The optimal peptide predicted on the basis of library screening contains a Thr in the S2 position and Arg residues in the S3, S5, and S7 positions, RXRXRTXS. Because of the Arg at 5 and 7, the Thr residue at 2 in this sequence is also a potential site for phosphorylation by AKT. We therefore replaced this Thr in the S2 position with an Ala to eliminate confounding phosphorylation events. Peptide library screening also revealed minor selection for Glu in the S4 position, though selection for acidic residues in phosphorylation motifs is complicated by an increased affinity of the ferric-IDA column used in phosphopeptide purification for non-phosphorylated peptides rich in acidic residues. Amino acids at other positions were chosen based on the residue most highly selected in the library analysis (Table I). His at S+3 and S+4 was weakly selected (data not shown), and Ala at S8 and S+5 was added because they were fixed in the peptide library. We therefore defined the optimal AKT substrate peptide ARKRERAYSFGHHA as our reference peptide (AKTide) and synthesized mutant versions for kinetic analysis (Fig. 2, A and B).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Kinetic analysis of the predicted optimal AKT/PKB substrate. A, synthetic peptides based on the predicted optimal substrate of AKT (AKTide) and variations involving single amino acid substitutions. B, phosphorylation of the various peptides by AKT was linear over a 15-min incubation time at 25 °C using conditions described under "Experimental Procedures." C, Km and Vmax values for each of the peptides were determined using conditions described above. Less than 10% of the peptides were phosphorylated in all reactions. The reactions were performed three times, and average values are shown. N.D., not determined because of low phosphorylation efficiency.

Various concentrations of wild-type and mutant versions of AKTide were phosphorylated in vitro using purified recombinant AKT to determine K_m and V_max values (see Fig. 2, B and C). Reaction rates were linear with respect to time for at least 15 min for all concentrations of peptides employed. As expected, neither the AKTide-SA (lacking a phosphorylatable Ser residue) nor the AKTide-3A mutant (lacking an Arg in the S3 position) peptides underwent significant AKT-directed phosphorylation (Fig. 2B). The AKTide-5A peptide, lacking an Arg in the S5 position, was also a poor AKT substrate, showing a nearly 30-fold decrease in the V_max/K_m ratio compared with AKTide (Fig. 2, B and C). Substitution of Ala for the Arg residue in the S7 position resulted in an ~2-fold drop in V_max/K_m, whereas substitution of Ala for residues at the S6 or S4 positions had only minimal effect. These data are in good agreement with the relative selectivity values from the peptide library screen (Table I). The order of importance of residues is Arg-3>Arg-5>Arg-7. Also in agreement with the peptide library results, AKTide-2T, containing the optimal motif selected by oriented peptide library screening, proved to be the best in vitro substrate for AKT (Fig. 2C). Interestingly, although AKTide-2T had the lowest K_m (3.9 µM) and highest V_max/K_m ratio, it did not have the highest V_max value. These data are in agreement with Nishikawa's report that our peptide library screening approach selects substrates on the basis of low K_m and/or high V_max/K_m ratios rather than on the basis of V_max alone (42).

Phosphoamino Acid Analysis of AKTide-2T-- Because the best optimal peptide, AKTide-2T (ARKRERTYSFGHHA) has two potential AKT phosphorylation sites (Ser or Thr), it is conceivable that its optimal kinetic behavior reflects tandem Ser and Thr phosphorylation within the same peptide. However, contrary to our expectation, phosphoamino acid analysis revealed that AKT only phosphorylated the Ser but not the Thr residue within AKTide-2T, exactly as observed with the wild-type AKTide lacking Thr in the S2 position (Fig. 3).

View larger version (37K): [in this window] [in a new window]   Fig. 3.   Phosphoamino acid analysis of AKTide. AKTide and AKTide-2T were phosphorylated by AKT as described in the legend to Fig. 2. Phosphoamino acid analysis was performed on the reaction products as described under "Experimental Procedures." The autoradiogram on the left reveals the exclusive presence of phospho-Ser in both peptides. The location of phospho-Ser, phospho-Thr, and phospho-Tyr internal standards is shown in the ninhydrin staining image (right).

Inhibition of AKT Kinase Activity by AKTides-- One general class of protein kinase inhibitors can be created by altering the phosphorylated residue in an optimal substrate to a non-phosphorylatable analog. This was first demonstrated with the PKA substrate inhibitor, PKI (49). We investigated whether a similar type of inhibition could be generated for AKT. In these experiments, we used a non-phosphorylatable analog of AKTide (AKTide-SA) to inhibit the phosphorylation of histone H2B by AKT. We also investigated the optimal substrate peptide (AKTide-2T) as an inhibitor (Fig. 4). Both peptides showed competitive inhibition of histone H2B phosphorylation as confirmed by a Lineweaver-Burk plot (data not shown), with K_i values of 12 and 56 µM for AKTide-2T and AKTide-SA, respectively.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Inhibition of AKT kinase activity by AKTides. In vitro kinase assays were performed as described in the legend to Fig. 2 using histone H2B as substrate in the presence of the indicated amounts of AKTide-2T (open square) or AKTide-SA (open circle). Competitive inhibition was confirmed by a Lineweaver-Burk plot. A linear regression fit to the data gave Ki values of 12 and 56 mM for AKTide-2T and AKTide-SA, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

In this study, we have determined the optimal phosphorylation motif for Akt by oriented peptide library screening and corroborated the result by identifying similar motifs in AKT protein substrates isolated by screening of a HeLa cell cDNA expression library. The optimal motif was confirmed by a detailed kinetic analysis of individual peptides containing Ala substitutions at positions within the motif. We constructed a non-phosphorylatable substrate analog, and showed that this peptide could function as a competitive inhibitor of AKT phosphorylation in vitro.

As shown in Table II, most known physiological substrates of AKT have Arg at both the 3 and 5 positions and hydrophobic amino acids (e.g. Phe) in the +1 position. However several other basophilic protein kinases have similar requirements for optimal substrates. For example, PKC, PKC, and PKC also select substrates with Arg at 3 and 5 and Phe at +1 (Table V). A distinguishing feature of optimal AKT substrate is selection for residues at S+2 that form tight turns (e.g. Gly), in contrast to optimal PKC substrates which select for basic or aromatic residues in the S+2 position. In addition, selection for Thr at S2 also distinguishes AKT from all other basophilic kinases investigated to date (Table I and Table V). Thus, the information provided by peptide library screening can help to distinguish AKT phosphorylation sites from sites likely to be phosphorylated by other related basophilic protein kinases. Our results are in good agreement with those of Alessi et al. (50) who showed that substitution of Ala for Arg in the S5 or S3 positions or Ala for Phe in the S+1 position of a peptide derived from the AKT phosphorylation site in GSK3 resulted in reduced phosphorylation. Our results also agree with those of Kobayashi and Cohen (51) who found a critical dependence on Arg in the S3 position. Mutation of this residue, even to Lys, another basic amino acid, severely compromised the ability of AKT to phosphorylate a peptide substrate. Similarly, Kobayashi and Cohen (51) reported that mutation of Phe in the S+1 position to the charged amino acids Lys or Glu decreased the efficiency of phosphorylation by 30 and 13%, respectively. In agreement with these data, Phe was the optimal residue in the S+1 position by peptide library screening, and as shown in Table I, most known substrates of Akt have hydrophobic amino acids (often Phe) in this position.

A Structural Model for AKT Motif Selection-- We have previously proposed a structural basis for substrate selectivity of basophilic protein kinases based on the crystal structure of the PKA/PKI complex and residues conserved between PKA and other basophilic protein kinases (42, 47, 52). To rationalize the substrate specificity we observed for AKT, we modeled the structure of the AKT kinase domain using the x-ray structure of the PKA·PKI·ATP ternary complex as a basis set. The resulting molecular surface of the modeled AKT protein, shaded by electrostatic potential, is shown in Fig. 5. All of the basophilic kinases in Table V preferentially phosphorylate substrates with Arg in the S3 position. These results can be rationalized by a weakly acidic pocket in PKA at the S3 position (Fig. 5, red) with contributions from both a highly conserved acidic residue (Glu-127 in PKA; Glu-234 in AKT) that salt bridges to the guanidino nitrogens and the proximity of the - and -phosphates of ATP to this residue. In addition, both PKC and AKT contain residues predicted to form an acidic patch at the S5 position (e.g. Asp-467, Glu-530, and Asp-503 of PKC and Glu-278, Glu-341, and Glu-314 in AKT that loosely map to the S2 pocket in PKA (Glu-170, Glu-230, and Glu-202)). Similarly, nearly all basophilic kinases contain a hydrophobic pocket at S+1 which is particularly deep in the case of AKT (e.g. Phe-309, Pro-313, and Leu-316) rationalizing selection for Phe in the S+1 position. As discussed above, the selection for peptides with residues at S+2 that tend to form tight turns is unique to AKT. This may be explained if the substrate binding cleft terminates at this position, forcing a tight turn to exit the pocket. Interestingly, several of the in vivo substrates of AKT have Pro at the S+2 position and are known to bind to 14-3-3 proteins after phosphorylation (53-55). The crystal structures of 14-3-3 protein·phosphopeptide complexes reveal the necessity for a tight turn residue at S+2 for exit from a deep binding cleft (53-54). In addition, tight binding to 14-3-3 protein is favored by Arg at S3 and Ser or Thr at S2 and is further facilitated by Arg at S5 (54) (Table VI). Thus, AKT appears to have evolved a substrate selection that is skewed toward motifs recognized by 14-3-3 proteins. Consistent with this idea, the sites on Forkhead-related proteins that are phosphorylated by AKT and bind to 14-3-3 are highly conserved from Caenorhabditis elegans to mammals (Table VI).

View larger version (49K): [in this window] [in a new window]   Fig. 5.   Structural model of the AKT catalytic cleft. The AKT kinase domain sequence was modeled with SWISS-MODEL (57-58) using the PKA·PKI·ATP ternary complex x-ray structure as a template (59). The resulting molecular surfaces for PKA (left) and AKT (right) shaded by electrostatic potential were calculated using GRASP (60). Areas of positive and negative potential are shaded blue and red, respectively. ATP and PKI are shown in stick representation with oxygen colored red, nitrogen blue, carbon gray, and phosphate yellow. The catalytic clefts of both kinases are shown in the identical orientation. The regions corresponding to the putative S5, S3, and S+1 pockets in AKT are indicated.

Identification of in Vitro AKT Substrates-- Several new in vitro AKT substrates were identified using solid-phase phosphorylation screening. Many of the positive clones we identified were spliceosomal proteins. This selection may result, in part, from cDNA bias, because spliceosomal proteins tend to be expressed in relatively large amounts within cells (56). Furthermore, many have Arg/Ser (RS)-rich domains that are the targets for SR kinase-mediated phosphorylation events. These Arg/Ser domains frequently contain multiple AKT phosphorylation motifs lying within narrow stretches of sequence, as described for Zis above. The splicing factor SC35, for example, has more than 20 possible AKT phosphorylation motifs in its protein sequence. When the entire GenPept data base was searched using a profile-based bioinformatics program and the optimal consensus motif determined for AKT,² 81 of the top 300 scoring sequences were spliceosomal proteins. Thus there is consistency between the peptide library screening/data base search approach and an independent cDNA expression library screening for AKT substrates. Intriguingly, this bioinformatics search also identified all known AKT substrates including BAD, and members of the Forkhead transcription factor family. An AKT phosphorylation site on mTOR, the mammalian target of rapamycin, at Ser-2448 was also identified, and this site has recently been reported by Nave et al. (37).

The relevance of our finding that many proteins involved in RNA processing and previously identified as substrates of SR kinases are also good substrates for AKT is not yet clear. This may reflect similarities between optimal substrates for SR kinases and AKT (Table V). One major difference between SR kinases and AKT is that SR kinases prefer substrates with Arg at the S+1 position (52), whereas AKT prefers hydrophobic residues. Thus, the substrates in Table IV with Arg at S+1 are more likely to be SR kinase substrates in vivo whereas those with hydrophobic residues at S+1 are good candidates for in vivo AKT substrates. Further work will be necessary to evaluate the in vivo relevance of these candidate substrates.

AKT is known to participate in a wide range of normal cell functions and up-regulation of its activity has been noted in a variety of neoplasms. Identification of the optimal substrate motif using oriented peptide library screening and protein library screening should facilitate the identification of additional in vivo AKT targets relevant to tumorigenesis and cancer progression.

	ACKNOWLEDGEMENTS

We thank Dr. Rikoro Fukunaga (Osaka University, Suita, Japan) for the human HeLa cell cDNA library in GEX phage and technical advice, Dr. Alfonso Bellacosa (Fox Chase Cancer Institute, Philadelphia, PA) for the cDNA of mouse AKT-1, Dr. Andrea Musacchio (Harvard Medical School, Boston, MA) for baculovirus GST-tagged p110 of PI3K, Dr. Colleen Sweeny (Harvard Medical School, Boston, MA) for phosphoamino acid analysis direction, and Drs. Masashi Suzaki and Ryohei Okamoto (Central Research Center, Shiga University of Medical Science, Otsu, Japan) for DNA sequencing and oligonucleotide synthesis. We would also like to thank members of the Division of Signal Transduction and Third Department of Medicine for support and helpful suggestions throughout the period of this work.

	FOOTNOTES

^* This research was supported in part by National Institutes of Health Grant GM56203 (to L. C. C.)The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

These authors contributed equally to this work.

^** Supported by the Kanae Foundation and the Uehara Memorial Foundation.

^§§ Supported by National Institutes of Health Grant HL03601 and the recipient of a Burroughs-Wellcome Career Development Award.

^¶¶ To whom correspondence should be addressed: Division of Signal Transduction, Harvard Institutes of Medicine, 10th Floor, 330 Brookline Ave., Boston, MA 02215. Tel.: 617-667-0947; Fax: 617-667-0957; E-mail: cantley@helix.mgh.harvard.edu.

Published, JBC Papers in Press, August 16, 2000, DOI 10.1074/jbc.M005497200

² M. B. Yaffe, G. G. Leparc, J. Lai, T. Obata, S. Volinia, and L. C. Cantley, submitted for publication.

	ABBREVIATIONS

The abbreviations used are: PKC, protein kinase C; PKA, cAMP-dependent protein kinase A; PI3K, phosphatidylinositol 3'-kinase; PDK-1, 3-phosphoinositide-dependent protein kinase 1; GSK3, glycogen synthase kinase 3; AEBSF, aminoethylsulfonyl fluoride; IPTG, isopropyl-thiogalactopyranoside; GST, glutathione S-transferase; IDA, iminodiacetic acid; HA, hemagglutinin; PCR, polymerase chain reaction; GCSF-R, granulocyte colony-stimulating factor receptor.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				M. H. Gao, T. Tang, T. Guo, A. Miyanohara, T. Yajima, K. Pestonjamasp, J. R. Feramisco, and H. K. Hammond Adenylyl Cyclase Type VI Increases Akt Activity and Phospholamban Phosphorylation in Cardiac Myocytes J. Biol. Chem., November 28, 2008; 283(48): 33527 - 33535. [Abstract] [Full Text] [PDF]

				A. K. Snabaitis, F. Cuello, and M. Avkiran Protein Kinase B/Akt Phosphorylates and Inhibits the Cardiac Na+/H+ Exchanger NHE1 Circ. Res., October 10, 2008; 103(8): 881 - 890. [Abstract] [Full Text] [PDF]

				M. L. Miller, L. J. Jensen, F. Diella, C. Jorgensen, M. Tinti, L. Li, M. Hsiung, S. A. Parker, J. Bordeaux, T. Sicheritz-Ponten, et al. Linear Motif Atlas for Phosphorylation-Dependent Signaling Sci. Signal., September 2, 2008; 1(35): ra2 - ra2. [Abstract] [Full Text] [PDF]