Inhibition of Akt Kinase Activity by a Peptide Spanning the βA Strand of the Proto-oncogene TCL1*

Akt plays a central role in the regulation of cellular anti-apoptosis underlying various human neoplastic diseases. We have demonstrated previously that TCL1 (a proto-oncogene underlying human T cell prolymphocytic leukemia) interacts with Akt and functions as an Akt kinase co-activator. With the aim to develop an Akt kinase inhibitor, we hypothesized that a peptide, which spans the Akt-binding site, binds to Akt and modulates Akt kinase activity and its downstream biological responses. Indeed, we demonstrated that a peptide, named “Akt-in” (Akt inhibitor, NH2-AVTDHPDRLWAWEKF-COOH, encompassing the βA strand of human TCL1), interacted with Akt and specifically inhibited its kinase activity. Nuclear magnetic resonance studies suggested that interaction of Akt-in with the pleckstrin homology domain (PH) of Akt caused conformational changes on the variable loop 1 of Akt, the locus mediating phosphoinositide binding. Consistently, interaction of Akt-in with the Akt PH domain prevented phosphoinositide binding and hence inhibited membrane translocation and activation of Akt. Moreover, Akt-in inhibited not only cellular proliferation and anti-apoptosis in vitro but also in vivo tumor growth without any adverse effect. The roles of Akt, which possesses a PH domain, in intracellular signaling were well established. Hence, Akt inhibitors create an attractive target for anticancer therapy. However, no effective inhibitors specific for Akt have been developed. Akt-in, which inhibits association of phosphatidylinositol with Akt, is the first molecule to demonstrate specific Akt kinase inhibition potency. This observation will facilitate the design of specific inhibitors for Akt, a core intracellular survival factor underlying various human neoplastic diseases.

emerged as a pivotal regulator of many cellular processes (1)(2)(3)(4). Three highly homologous Akt isoforms (Akt1, Akt2, and Akt3) exist in mammals. Akt is composed of three functionally distinct regions: an N-terminal pleckstrin homology (PH) domain, a central catalytic domain, and a C-terminal hydrophobic region. The PH domain is a small 100 -120-residue module found in many proteins involved in cell signaling or cytoskeletal rearrangement. The PH domain of Akt is similar to other proteins, and it consists of seven ␤ strands forming two orthogonal antiparallel ␤-sheets that are closed with the C-terminal ␣-helix (5)(6)(7)(8)(9).
Activation of Akt promotes cell survival (18); thus, it could be the underlying mechanism for numerous human neoplastic diseases including lung, ovarian, and prostate cancers (1,19). Activation of Akt is also induced in the mutation of PTEN (phosphatase and tensin homolog deleted on chromosome 10) tumor suppressor gene. PTEN antagonizes PI3K function by the reduction in the levels of both PtdIns(3,4,5)P 3 and PtdIns(3,4)P 2 . Mutations of PTEN are implicated in several tumor types, including glioblastoma, endometrial tumors, and Cowden's syndrome (20,21).
We have demonstrated that the proto-oncogene TCL1 is an Akt kinase co-activator (22)(23)(24)(25). TCL1 contains two distinct functional motifs responsible for Akt association and homodimerization. Both Akt association and homodimerization of TCL1 are required for the complete function of TCL1 to enhance Akt kinase activity. TCL1 binds to Akt and activates Akt via a transphosphorylation reaction (26,27). TCL1 oncogene was first implicated in human T cell prolymphocytic leukemia, a chronic adulthood leukemia (28). Under physiological conditions, TCL1 expression is limited to early developmental stages such as the immune system (24,28,29). Because the PI3K-Akt pathway is involved in various human neoplastic diseases, Akt represents an attractive target for drug development (19,30). A small peptide was proven to effectively modulate activity of kinases effectively (31)(32)(33). One class of Akt inhibitors under development is based on the cross-reactivity between known kinase inhibitors (e.g. cyclic AMP-dependent protein kinase (PKA) or PI3K) (34 -36); however, these drugs are not specific for Akt. With the goal to develop a putative Akt kinase inhibitor, we hypothesized that a peptide, which is spanning the Akt-binding site, binds to Akt and modulates Akt kinase activity along with its downstream biological responses.
Based on the binding domain of TCL1 with Akt, we identified and characterized a peptide that encompassed the ␤A strand of TCL1, interacted with Akt, and inhibited Akt kinase activity. Akt-in prevented PtdIns binding to Akt, and consequently it inhibited membrane translocation of Akt and its downstream biological responses. Given the pivotal role of Akt kinase as a core intracellular survival factor implicated in the molecular mechanisms of human neoplastic diseases, the results could help to design Akt kinase-specific inhibitors for therapeutic approaches.

EXPERIMENTAL PROCEDURES
Peptide Design-For the Akt-in peptides, the amino acid positions 10 -24 of human TCL1, NH 2 -AVTDHPDRLWAWEKF-COOH are used. For TAT-FLAG Akt-in, the sequence is YGRKKRRQRRRDYKDDDD-KAVTDHPDRLWAWEKF-COOH.
Co-immunoprecipitation Assay-Co-immunoprecipitation assays were performed as described previously (22). Briefly, Akt1, Akt2, or Akt3 in pCMV6 was transfected into 293 cells (ATCC). The cells were then harvested, lysed, and pre-cleaned with protein G/A-agarose mixture (50% v/v, Pro-G/A, Amersham Biosciences). FLAG-Akt-in or control peptides (␤C) at 400 M were added to the cell lysates, incubated at 4°C for 3 h, and incubated with Pro-G/A preconjugated with anti FLAG M2 antibody (Sigma). The resultant immune precipitants were washed and run on SDS-PAGE and immunoblotted with anti-HA antibody (3F10, Roche Applied Science).
GST Pull-down Assay-293T cells (ATCC) were transfected with 10 g of FLAG-tagged wild type Akt3, PH domain, or C-terminal Akt3 (27). The cell lysates were immunoprecipitated with anti-FLAG antibody (FLAG M2, Sigma) bound to Pro-G/A (27). Fifty ng of GST fusion proteins were incubated with 20 l of immobilized Akt3, PH domain, or C-terminal Akt. The samples were run on SDS gels and immunoblotted with anti-GST antibody (Amersham Biosciences). The results were consistent in at least three independent experiments. GST fusion Akt-in was generated by subcloning with the corresponding nucleotide into pGEX4T-2 vectors (Amersham Biosciences). All nucleotide sequences were verified before the experiments. GST Competition Assay-Recombinant GST-Akt-in fusion protein was generated by pGEX 4T-2 Vector (Amersham Biosciences) using oligonucleotide pairs (5Ј-aattcgcagtcaccgaccacccggaccgcctgtgggcctgggagaagttctagg-3Ј). 0.1 g of Akt (activated, Upstate Biotechnology, Inc.) was incubated with TAT-FLAG, TAT-Akt-in, or TAT-␤C at the concen-tration of 0, 50, 100, or 250 M in HEPES Binding Buffer (20 mM HEPES (pH 7.0), 150 mM NaCl, 0.5 g/l bovine serum albumin). 0.1 g of GST-Akt-in was then added and incubated for an additional 20 min at 4°C. Twenty l of immobilized Akt beads (Cell Signaling) were added to the sample, washed five times with HEPES Binding Buffer in the presence of 0.1% Nonidet P-40, resolved onto an SDS gel, and immunoblotted by anti-Akt (Cell Signaling) or anti-GST (Amersham Biosciences) antibodies using ECL (Amersham Biosciences).
Kinetics of Akt-in-The kinetics of Akt-in with the human Akt2-PH domain (amino acid 1-125 of human Akt2) was performed using the Applied Biosystems 8500 Affinity Chip Analyzer. Briefly, His fusion protein of human Akt2-PH domain was generated using pQE30 (Qiagen) by PCR. 1.25 pg of GST fusion proteins (Akt-in or wild type TCL1) were spotted onto the protein-A/G Affinity Chips preconjugated with anti-GST antibody (Sigma). Fifty M of His-Akt2-PH domain was applied, and the dissociation constant was calculated by using data analysis software (Applied Biosystems). The values (mean Ϯ S.D.) were calculated from the 80 measurements.
In Vitro Akt Kinase Assay-In vitro Akt kinase assays were performed using the Akt kinase assay kit (Cell Signaling) (22). Briefly, the immobilized Akt was incubated with 0, 200, or 400 M of indicated peptides for 2 h, and then an in vitro kinase assay reaction was performed for 4 min at 30°C. The samples were heat-denatured, separated on SDS-PAGE, and immunoblotted with anti-phospho-GSK or anti-Akt (Cell Signaling) using ECL (Amersham Biosciences).
PKA Kinase Assay-In vitro PKA kinase assays were performed using Peptag (Promega) as described previously (22). The indicated concentrations of peptides (Akt-in or TAT-FLAG control peptide) were incubated with 25 ng of PKA with 100 ng of bovine serum albumin for 1 h in the presence or absence of 2 M PKA inhibitor (Calbiochem catalog number 116805), followed by the kinase reaction for 20 min at 26°C, and then separated on 0.8% TBE-agarose gel.
Phosphorylation of Akt, BAD, FKHR, or p44/42 MAP Kinase in 293 Cells-293 cells (ATCC) were cultured in a 60-mm dish and transfected (or nontransfected in Fig. 3B) with 5 g of m-BAD (Fig. 3C, pEBG-mBad, Cell Signaling) using calcium phosphate transfection as described previously (22). 24 h after transfection, the cells were serumstarved (0.2% fetal bovine serum) and treated with either control (TAT-FLAG) or Akt-in (TAT-Akt-in) at 50 M for additional 12 h. The cells were stimulated with or without 20 ng/ml PDGF for 8 (in Fig. 3C) or 5 min (in Fig. 3B) and lysed with Brij lysis buffer in the presence of phosphatase inhibitors (22), and the resultant samples were resolved on 4 -20% SDS gel (Kyoto Daiichi Kagaku Co., Ltd. NMR Experiment-NMR spectra were recorded on 0.25-0.3-ml (Shigemi tubes pre-coated with a silicon solution (Sigma)) samples of 0.05 mM 15 N-labeled Akt2-PH dissolved in the conditioning buffer (10 mM Tris/H 2 O (pH 7.4), 300 mM NaCl, 0.1 mM benzamidine, 0.1 mM EDTA, with 5-10% 2 H 2 O for the lock), in the presence or absence of 20 mM Akt-in. NMR experiments were carried out at 10°C on a Bruker AVANCE 600 spectrometer equipped with 5-mm z-shielded gradient 1 H-13 C-15 N triple resonance cryogenic probe. 1 H chemical shifts were directly referenced to the resonance of 2,2-dimethyl-2-silapentane-5sulfonate sodium salt, and 15 N chemical shifts were indirectly referenced with the absolute frequency ratios ( 15 N/ 1 H) ϭ 0.101329118. In all experiments, the 1 H carrier was centered on the water resonance, and a WATERGATE sequence was incorporated to suppress the solvent resonance. All NMR spectra were acquired in the phase-sensitive mode with Digital Quadrature Detection in the F2 dimension and hypercomplex States-TPPI method in F1 dimension and processed using Gifa (version 4.22) software. 1 H, 15 N-HSQC spectra were recorded using a time domain data size of 64 t 1 ϫ 1 K t 2 complex points and 32 transients per complex t 1 increment.
PtdIns(3,4,5)P 3 Lipid-Protein Pull-down Assay-A lipid-protein pulldown assay was performed using PIP Beads (PtdIns(3,4,5)P 3 , Echelon Bioscience Inc.). Indicated peptides (Akt-in or ␤C control) were incubated with 50 ng of Akt kinase (unactivated, Upstate Biotechnology Inc., catalog number 14-279) with 400 ng/ml bovine serum albumin for 2 h with gentle agitation at 4°C. TAT-FLAG control was added to adjust the final peptide concentration to be equal throughout the samples. 25 l of PIP Beads were then added to each sample and incubated for an additional 16 h. The reactions were then washed four times with washing buffer (10 mM HEPES (pH 7.4), 0.25% Nonidet P-40, 140 mM NaCl), resolved on SDS gel, and immunoblotted by ECL (Amersham Biosciences).
Membrane Translocation Experiment-293 cells (ATCC) were grown on a poly-L-lysine-coated cover glass and were transfected with 1 g of HA-Akt1 or Akt-PH-GFP or Btk (Bruton tyrosine kinase)-PH-GFP in a mammalian expression vector (39) using FuGENE 6 (Roche Applied Science). Six hours after the transfection, 50 M Akt-in or TAT-FLAG control was added and incubated for 16 h. The cells were then serumstarved (0.5%) and then incubated for an additional 24 h. The cells were treated with or without 50 nM wortmannin for 20 min, stimulated with 50 ng/ml PDGF-AB (Sigma, 3226) for 10 min, fixed with 4% paraformaldehyde, stained with 10 ng/ml fluorescein isothiocyanate-conjugated anti-HA antibody (12CA5, MBL) or phospho-Ser-473 antibody (587-F11, Cell Signaling), and examined using a confocal microscope (Nikon).
Cell Death Assay and Mitochondrial Permeability Transition Assay-T4 cells (human T cell leukemia cells) were treated with the indicated concentrations of peptide (TAT-FLAG control or TAT-Akt-in) for 24 h. The cells were transfected with either myr-Akt (Upstate Biotechnology, Inc., catalog number 17-253) or a control. Cell death was assessed by staining with 2 g/ml propidium iodide. Mitochondrial permeability transition was verified by staining with rhodamine 123 (Molecular Probe) at 5 M for 15 min at 37°C (26) and analyzed using fluorescence-activated cell sorter (Cell Quest).

Peptide Design of Akt-in and the Structure of TCL1-TCL1
forms a closed symmetrical ␤-barrel structure, consisting of eight antiparallel ␤ strands (41) (Fig. 1A). In our previous studies, we showed that the surface composed of ␤A and ␤E strands of TCL1 mediated the interaction with Akt (22, 26, 42) (Fig. 1A, top surface). Both dimerization and Akt interaction are essential for the full function of TCL1 to activate Akt (26). We hypothesized that a peptide, which spans the Akt-binding sequences, can modulate Akt kinase activity and its downstream signals. We designed a peptide (named Akt-in, Akt inhibitor, positions 10 -24 of human TCL1, NH 2 -AVTDHP-DRLWAWEKF-COOH), which encompasses the ␤A strand of TCL1 for further study (Fig. 1B). For functional assays, the Akt-in peptide (amino acid positions 10 -24 of TCL1, Fig. 1B) was fused with TAT (YGRKKRRQRRR) and/or FLAG epitope (DYKDDDDK). The TAT fusion protein, which contains an N-terminal 11-amino acid protein transduction domain from FIG. 1. Structure-based alignment of TCL1 and Akt-in. A, Akt-in that encompasses the ␤A strand, which forms the interface for Akt interaction, is shown with TCL1. The structural study demonstrated that TCL1 forms a ␤-barrel structure consisting of eight anti-parallel ␤ strands (41). Previously, we demonstrated that the surface, which consists of both ␤A and ␤E strands, formed an interface mediating the Akt interaction (26). We hypothesized that a peptide spanning the Akt-binding site binds to Akt and modulates Akt kinase activity along with its downstream biological responses. In the current study, we generated a peptide, named Akt-in (NH 2 -AVTDHPDRL-WAWEKF-COOH, encompassing the ␤A strand of human TCL1) to test our hypothesis. B, amino acid sequence of Akt-in is shown with the alignment of the sequence of TCL1 (41). Akt-in consists of 15 amino acids (position 10 -24 of TCL1, NH 2 -AVTDHPDRLWAWEKF-COOH, shaded gray), which spans the crucial binding site for Akt interaction (Asp-16, boldface and underlined) (22,26). the human immunodeficiency virus, can be efficiently imported to the cytosol (37,43). By immunohistochemical staining, either TAT controls or TAT-Akt-in peptides was efficiently imported to the cytosol (data not shown).
Akt-in Specifically Interacts with Three Akt Isoforms through Pleckstrin Homology Domain-Because TCL1 interacts with all three Akt isoforms (22,27), we first investigated whether Akt-in can interact with all three isoforms of Akt. In pull-down assays, Akt-in did interact with them (Fig. 2, A-C, Akt1, Akt2, and Akt3, respectively).
Wild type TCL1 interacts with Akt through an N-terminal pleckstrin homology (PH) domain (22,23). Akt-in encompasses the ␤A strand of TCL1, the locus mediating TCL1-Akt interaction. Therefore, it is predicted that Akt-in interacts with Akt via a PH domain.
In pull-down experiments, Akt-in bound the full-length wild type Akt (Fig. 2D, lane 2) and the PH domain of Akt (Fig. 2E, lane 2) but not the C-terminal Akt (Fig. 2F, lane 2). Furthermore, control peptides (GST protein, or ␤C peptide) did not interact with either the full-length Akt or the PH domain confirming the specificity of Akt-in (Fig. 2, D-F, lanes 1 and 4, ␤C peptide and GST protein, respectively). In GST competition assays, specific binding of Akt-in with Akt was further demonstrated by the dose-dependent inhibition by Akt-in (Fig. 2G, lanes 5-8) but not by control peptides (Fig. 2G, lanes 1-4 or lanes 9 -12, TAT-FLAG or TAT-␤C, respectively).
The kinetic study of Akt-in interaction with the Akt-PH domain was performed by Spot Matrix SPR technology. The dissociation constant (K d ) of Akt-in with the Akt-PH domain was 18 Ϯ 4.8 M (Akt-in, mean Ϯ S.D.), and the K d of wild type TCL1 with the Akt-PH domain was 5.4 Ϯ 1.4 M.
Akt-in Specifically Inhibits Akt Kinase Activity-Akt is regulated by multiple site phosphorylation. Akt kinase assays performed in vitro showed that the addition of Akt-in compromised Akt kinase activity as assessed by the phosphorylation levels of GSK3␣ in a dose-dependent manner (Fig. 3A, top  panels).
To determine the specific inhibition of Akt-in on Akt activation in intact cells, 293 cells were treated with TAT-Akt-in and stimulated with PDGF. Akt-in treatment inhibited phosphorylation of PDGF-induced Akt activation at Thr-308 (Fig. 3B, top row, lanes 2-5) or Ser-473 (2nd row, lanes 2-5). However, Akt-in did not inhibit the levels of phosphorylation of P38 MAP kinase (Fig. 3B, 4th row, lanes 2-5). Control peptide (50 M TAT-FLAG) showed no inhibition in this experiment (Fig. 3,  lane 6 of each panel).
TAT-Akt-in treatment of the 293 cells inhibited the phosphorylation of Akt at Ser-473 (Fig. 3C, compare the top row, lane 6 with lane 4, TAT-FLAG control peptide versus TAT-Akt-in after PDGF stimulation). In the same experiment, TAT-Akt-in treatment inhibited the phosphorylation levels at serine 136 of BAD (Fig. 3C, 5th row) or serine at 256 of FKHR (Fig. 3C, 7th row) after PDGF stimulation. To determine the specificity of Akt-in on Akt phosphorylation, we have examined the phosphorylation levels of p44/42 MAP kinase (Fig. 3C, 3rd row). Akt-in treatment did not inhibit p44/42 MAP kinase, suggesting the specificity of Akt-in on Akt phosphorylation. Consistently, TAT-Akt-in treatment of QRsP-11 fibrosarcoma cells (40) inhibited PDGF-induced Akt phosphorylation on both Ser-473 and Thr-308 detected by immunoblotting (data not shown).  2-4, versus lanes 5-7, TAT-FLAG control versus Akt-in, respectively), because PDK1 is an upstream kinase that activates Akt at Thr-308. The results suggested that Akt-in directly inhibited Akt kinase activity, but not through the inhibition of PDK1 activation. E, to examine further the specificity of Akt-in to Akt, PKA kinase assays were performed using Peptag (Promega). Akt-in did not inhibit PKA kinase activity in in vitro PKA kinase assays as measured by phosphorylation levels of Kemptide (lanes 3-6 versus lanes 9 -12, Akt-in versus TAT-FLAG control, respectively). PKA inhibitor (Calbiochem catalog number 116805) inhibited the kinase reaction (lanes 2 and 8). The results (A-E) were consistent in at least two independent experiments.
Further demonstrating the specificity of Akt-in, phosphorylation of PDK1 was examined. PDK1 phosphorylates Thr-308 of Akt, which is a crucial step for activating Akt (14,15,44). By using recombinant PDK1, Akt-in did not inhibit PDK1 kinase activity as measured by the phosphorylation of PDK1 in a dose-dependent manner (Fig. 3D). Akt-in did not inhibit phosphorylation of PDK1 substrates (data not shown), consistent with the previous studies that demonstrated wild type TCL1 does not interact with PDK1 (22).
The catalytic domain of Akt is structurally similar to PKA (7). To determine further the specificity of Akt-in, an in vitro PKA kinase assay was carried out. The inhibitory effect of Akt-in was specific to Akt, because Akt-in did not inhibit PKA activity (Fig. 3E). In order to further support that Akt-in specifically inhibited Akt kinase activity, we examined the effect of Akt-in on protein kinase C (PKC) kinase activity using an in vitro kinase assay. Akt-in did not inhibit phosphorylation of PKC kinase activity in PKC kinase assay in vitro (data not shown).

Akt-in Inhibits Phosphoinositide (3,4,5)P 3 Binding to Akt and Consequently Inhibits Membrane Translocation and Activation of Akt-
The product of PI3K activation, PtdIns(3,4,5)P 3 or PtdIns(3,4)P 2 , triggers the activation by interacting with the Akt PH domain (1,8). PtdIns is abundant in the plasma membrane, and translocalization of Akt to the plasma membrane is one of the critical steps for the initiation of Akt activation (45,46). NMR mapping study suggested that Akt-in could induce the conformational change at the locus responsible for PtdIns binding. The results prompted us to investigate whether Akt-in can affect the interaction of PtdIns(3,4,5)P 3 with Akt.
Membrane translocation of Akt is mediated through the interaction of PtdIns(3,4,5)P3 with Akt, which triggers the activation of Akt (45,46). Therefore, we next examined whether TAT-Akt-in could inhibit the membrane translocation of Akt in the intact cells (Fig. 5B). As predicted by the result of lipidprotein pull-down assays, TAT-Akt-in treatment of the 293 cells (Fig. 5B, g-l) prevented PDGF-induced membrane translocalization of Akt from the cytosol (g) and inhibited phosphorylation of Akt (Ser-473) (Fig. 5B, h and i, P-Ser-473 Akt and the overlay view of Akt (HA) plus P-Ser-473 Akt, respectively). In contrast, control peptide (TAT-FLAG) treatment (Fig. 5B, m-r) did not inhibit PDGF-induced membranous translocation of Akt (m, green), phosphorylation of Akt at Ser-473 (n, red), and the overlay view of Akt (HA) plus P-Ser-473 Akt (o, yellow). Wortmannin treatment of the cells completely inhibited the Akt activation and membrane translocation in this experiment (Fig. 5B, a-c, Akt (HA), P-Ser-473, and the overlay view of Akt (HA) plus P-Ser-473 Akt, respectively).
To examine the specificity of the inhibitory effect on membrane translocation by Akt-in, we compared the PDGF-induced membrane translocation of Akt (Akt-PH-GFP) or Btk (Bruton tyrosine kinase, Btk-PH-GFP (39)), both of which contain a PH domain and are also known to be translocalized upon PDGF stimulation. Akt-in treatment of the cells potently inhibited translocation of the GFP-fused Akt (Akt-PH-GFP) from the cytosol to plasma membrane after PDGF stimulation (Fig. 5C,  compare b and d, control peptide, or Akt-in treated cells after the PDGF stimulation, respectively). However, Akt-in treatment of the cells did not inhibit translocation of Btk-PH-GFP from the cytosol to plasma membrane after PDGF stimulation. The results indicated Akt-in specifically inhibited the membrane translocation of Akt from the cytosol upon PDGF stimulation.
Akt-in Prevents the Biological Consequences of Akt Activation-We next examined whether Akt-in could inhibit the biological effects of Akt activation (17,47,48). Treatment of T4 cells by TAT-Akt-in not only prevented cellular proliferation (Fig. 6A) but also inhibited cell survival in a dose-dependent manner (Fig. 6B). The inhibitory effect was Akt-dependent, because introduction of myr-Akt (a constitutive active form of Akt) reversed inhibition of the proliferation assay as well as the cell death assay (Fig. 6B and data not shown). Similar enhancement of apoptosis was observed with DNA content analysis and annexin V staining (data not shown). Moreover, TAT-Akt-in could inhibit stabilization of mitochondrial outer membrane depolarization using rhodamine 123 staining (Fig. 6C).
Akt-in Inhibits Tumor Growth in Vivo-Finally, we investigated whether Akt-in inhibits tumor growth in vivo. QRsP-11 fibrosarcoma cells (40,49) were transplanted subcutaneously into syngeneic C57BL/6 mice, and TAT-Akt-in (or TAT-FLAG) was directly injected into the tumor (2 mol per mouse, three times per week starting from day 5). TAT-Akt-in treatment suppressed tumor growth when compared with the control groups (Fig. 7A). Most importantly, no obvious side effects were observed by the weight gain (Fig. 7A, inset). On day 19, the mean tumor volume from the TAT-Akt-in-treated mice was significantly smaller than the tumor from the control peptidetreated mice (240 Ϯ 223 mm 3 versus 1428 Ϯ 319 mm 3 , p Ͻ 0.01, TAT-Akt-in treated versus control peptide treated mice, respectively). Microscopically, the TAT-Akt-in-treated tumors resulted in an increased number of degenerated cells with apoptotic cells, which correlated with lower phosphorylation of P473 staining when compared with the control peptide-treated tumor (Fig. 7B, TAT-Akt-in or TAT-FLAG control peptidetreated tumors, the second row (H & E), the third row (TUNEL), and the fourth row (phospho-Ser-473 staining)). Moreover, TAT-Akt-in-treated mice significantly prolonged their mean survival time (48.3 Ϯ 9.4 versus 37.3 Ϯ 5.9 days, p Ͻ 0.05, Akt-in versus control peptide-treated mice, respectively). DISCUSSION We have demonstrated previously that the proto-oncogene TCL1 is an Akt kinase co-activator (22,23). In yeast two-hybrid assays and co-immunoprecipitation assays, we showed that TCL1 interacted with Akt through the PH domain and activated Akt (22, 26 -28). TCL1 forms a closed symmetrical ␤-bar-rel structure, consisting of eight antiparallel ␤ strands (41) (Fig. 1A). By mutational studies, we demonstrated that both Akt interaction and dimerization of TCL1 are required for complete function of TCL1 to enhance Akt kinase activity (26,42). We hypothesized that a peptide spanning the Akt-binding sequences of TCL1, which lacks the ability to form an oligo-  PtdIns(1,3,4,5)P 4 (shown in blue) as well as Akt-in (shown in green) are depicted as ball-and-stick representations. The position of PtdIns(1,3,4,5)P 4 was deduced from NMR (5) and x-ray (9) studies. The conformation of Akt-in is presented from the complex of Akt2-PH with wild type TCL1 (25). Colored residues indicate the footprint in the presence of Akt-in as determined by NMR mapping: red residue represented significant deviation in the chemical shift of the amide group, and orange residues represented a modest shift in the study. Unaffected residues are shown in gray. E, sequence alignment of human Akt-PH domain of human and mouse Akt1, Akt2, and Akt3 are shown. The amino acid residues responsible for binding to the phosphates of PtdIns are shaded blue with arrows (5,8,9,13). The amino acids responsible for PtdIns bindings are located at (and around) VL1 (the variable loop between ␤1 and ␤2 strands of PH domain). The footprints of Akt-in that are deduced from the NMR mapping study are shown by a red arrow with shading. The amino acid residues, which showed above the threshold levels of the resonance shift specific to Akt-in, are indicated by green arrow with shading. meric complex of Akt-TCL1 complex (22) but retains the ability for interaction with Akt, can down-modulate the Akt activation. We demonstrated that a peptide (named Akt-in, encompassing the ␤A strand of TCL1) specifically interacted with the Akt-PH domain and inhibited Akt kinase activity.
As demonstrated by GST pull-down experiments and NMR studies, Akt-in interacted with the Akt-PH domain. NMR mapping techniques further defined the molecular interaction of Akt-in with the Akt-PH domain at the amino acid level. The results indicated that similar to wild type TCL1, Akt-in, encompassing the ␤A strand of wild type TCL1, interacted with Akt-PH domain via the ␤5and ␤6 strands and the ␣-helix of Akt-PH domain (25). Moreover, interaction of Akt-in with the Akt-PH domain caused above the threshold levels of the resonance shifts of the amino acid residues on VL1, the ␤5 strand, ␤6 strand, and ␣-helix of Akt-PH domain. The significant resonance shifts were detected only in the presence of Akt-in, suggesting the specific interaction between Akt-in and Akt-PH domain, consistent with the observation that Akt-in specifically interacted with Akt-PH domain in co-immunoprecipitation assays. This conformational change is not induced by wild type TCL1, because wild type TCL1 interacted with the Akt-PH domain via both ␤A and ␤E strands of TCL1, which can help to stabilize the interaction. The deviations on the VL1 by Akt-in could also be induced by the indirect conformational change because of the very hydrophobic C-terminal end of Akt-in, which is protected from water by the ␤-barrel structure in wild type (native form) TCL1; hence it remains unexposed to the solvent. In lipid protein pull-down experiments using PtdIns beads, the interaction of Akt-in with the Akt PH domain dramatically decreased the affinity of Akt for PtdIns. Because binding of phosphatidylinositol 3,4,5-triphosphate to the PH domain of Akt induces a conformational change, a prerequisite for triggering the Akt activation (8), dissociation of PtdIns with the Akt PH domain by Akt-in could explain the inhibitory effects on Akt activation observed in vitro. In intact cells, disruption of PtdIns with the Akt PH domain resulted in the failure of translocalization of Akt to the plasma membrane and activation of Akt. The Akt kinase properties depend on its ability to bind PtdIns (45,46,50). Ser-473 phosphorylation as well as membrane anchoring are required to achieve full Thr-308 phosphorylation (10,15). Thus, Akt-in inhibited PtdIns binding to the Akt PH domain, and consequently it failed to activate Akt.
The N-terminal PH domain of Akt, which is common to over 100 signaling molecules, provides a lipid-binding module to direct Akt to PI3K-generated phosphoinositide PtdIns(3,4,5)P 3 or PtdIns(3,4)P 2 (1,2,51). PH domains are located in the N-terminal portion of Akt and, among the three Akt isoforms, over 90% are conserved at the amino acid level. Recently, the molecular interaction of PtdIns with the Akt PH domain has been clarified. Milburn et al. (8) reported by crystallography that the interaction of Akt with PtdIns resulted in a conformational change of VL1 (variable loop 1 between the ␤1 and the ␤2 strand of PH domain), which is involved in the PtdIns binding. By structural and functional studies, Thomas et al. (9) described that D4 phosphate interacted with Lys-14, D3 phosphate with Lys-14, Arg-23, and Arg-25, and D1 phosphate with Tyr-18, Ile-19, and Arg-23. It is noteworthy that the main interactions were mediated through binding to the D3 and D4 phosphates, whereas the D5 phosphate had no significant interactions (9). In conjunction with our findings, we demonstrated that PtdIns(1,3,4,5)P 4 (the polar head group of PtdIns(3,4,5)P 3 ) interacted with PH domain of Akt through the variable loops VL1 (5,8,9,13,25) (Fig. 4E).
Akt is one of the effector kinases that is activated by the product of PI3K activation, PtdIns(3,4,5)P 3 , and its immediate breakdown product PtdIns(3,4)P 2 (2, 10 -12). Association of PtdIns with the Akt PH domain induces conformational change, which is a prerequisite to the phosphorylation of Thr-308 by PDK1 (10,52). PDK1 phosphorylates Akt at Thr-308, which is required to initiate Akt activation (44,53). The findings that Akt-in did not inhibit kinase activity of PDK1 suggested that Akt-in could directly inhibit Akt kinase activity. The observation is consistent with one that Akt-in could not bind to PDK1 in the GST pull-down experiment (data not shown); therefore, in vitro PDK1 kinase assays did not demonstrate inhibition.
The full ranges of target molecules of Akt-in that harbor the PH domain remain to be elucidated. In co-immunoprecipitation assays, wild type TCL1 interacted solely with Akt but not with PDK1 or PKA (22). Consistently, Akt-in did not interact with y PDK1 in GST pull-down experiments (data not shown). NMR mapping studies indicated that Akt-in interacted with Akt-PH domain via ␤4 and ␤5 strands. It is notable that the amino acid sequences of the PH domain at (and around) the ␤4 and ␤5 strands (the locus mediating the Akt-in interaction) appear to be divergent among the PH domains (13,25,51). In addition, the study of the PH domain revealed that in contrast to the VL1 loop of the Akt-PH domain, which contains a short ␤1-2 typed loop, the VL1 of PDK1 has a long ␤1-2 typed loop (13). Although we could not completely exclude the possibility that Akt-in suppressed other kinases, Akt-in did not inhibit PKA, PKC, PDK1, p42/44 MAPK, or p38 MAPK. These observations further suggested that Akt-in interacts solely with Akt and specifically inhibits Akt kinase activity.
In vitro, Akt-in compromised Akt-dependent cellular proliferation, stabilization of mitochondrial permeability transition, and anti-apoptosis. In addition to T4 cells (human T cell leukemia cells), other cell lines such as P3HR-1 (Epstein-Barr virus-transformed human B cell line, HTB), 293T (human kidney CRL-11268), or WiDr cells (human colon adenocarcinoma cells, CCL-218) inhibited cellular proliferation following treatment with Akt-in. However, the inhibitory effect seemed stronger on TCL1-negative cell lines than on TCL1-positive cell lines becauseT4 cells, 293T, or WiDr cells are negative on TCL1b or MTCP1, other isoforms of TCL1 family proteins determined by Western blotting. Because of its considerably lower K d value (5.4 versus 18.0 M, wild type TCL1, versus Akt-in, respectively), Akt-in may not efficiently remove or replace native FIG. 7. Akt-in suppresses tumor growth in vivo. A, to verify the in vivo effect of Akt-in for the potential treatment for the neoplastic diseases, QRsP-11 fibrosarcoma cells were transplanted subcutaneously into C57BL/6 mice. The peptides (TAT-Akt-in, TAT-FLAG control peptide) were directly injected into the tumor on the days shown by the arrowheads. TAT-Akt-in inhibited tumor growth, whereas the control peptide (TAT-FLAG) or phosphate-buffered saline showed no inhibition. Eight mice in each group were examined. No adverse effects were observed based on body weight measurement (shown in inset). B, macroscopic appearances of QRsP-11 fibrosarcoma on day 9 after the transplantation are shown (1st row, gross appearance of TAT-Akt-in and the TAT-FLAG control peptide-treated tumors). High power views of the tumor from the TAT-Akt-in-treated mouse shows degenerated cells with apoptotic cells (indicated by arrows) are often present (H & E, 2nd row) compared with the tumor cells treated with the control peptide. The presence of apoptotic cells was further determined by the TUNEL method (3rd row). TAT-Akt-in treatment efficiently inhibited phosphorylation of Akt at Ser-473 compared with the control peptide treated tumor (4th row). Phosphate-buffered saline treatment did not affect tumor growth (data not shown). TCL1 molecules from Akt, which explains the decrease in kinase inhibition potency in TCL1-positive cell lines. Moreover, it is possible that once wild type TCL1 binds and enhances Akt kinase activity, a limited window appears for exogenous Akt-in to silence or inactivate the elevated level of Akt kinase activity.
Three isoforms of the TCL1 proto-oncogene exist, and they are 30% homologous at the amino acid level (28). Based on alignment of the peptide sequences of Akt-in (10 -24 of human TCL1), the sequences (ϪPro-15 to Leu-18 to Trp-19) appear to be conserved among the TCL1 family proteins. It is logical to speculate that the corresponding amino acid sequences in the other members of the TCL1 family (10 -22 (GVPPGRL-WIQRPG) in TCL1B and 7-19 (GAPPDHLWVHQEG) in MTCP1) may also inhibit Akt kinase activity. The roles of phosphoinositide binding to the PH domain in intracellular signaling are well established (1,2,51,54). In this regard, mutations of PTEN (20), which activates Akt by accumulating PtdIns, can be a promising target for potential therapy with Akt-in. Given the roles of the PH domain in cellular signaling (51,55), further modification of the Akt-in design may help to develop specific inhibitors for the signaling molecules that the PH domain.
Akt activation promotes cell survival, which affects tumor progression and resistance to chemotherapy and radiation in cancers (2,56,57). Hence, Akt inhibitor is believed to be an attractive target for anticancer therapy (19,30). Several attempts have been undertaken to develop Akt inhibitors (34 -36, 58, 59). Previous evidence has suggested that Akt kinase activity could be down-modulated by protein-protein interactions (Hsp90 or CTMP) (60 -62). Moreover, a small peptide was proved to effectively modulate kinase activity of AGC family kinases (32,33). However, no effective inhibitors specific for Akt have been developed.
We showed that Akt-in did not inhibit kinase activities of p44/42 MAPK, PDK, PKA, p38 MAP kinase, or PKC. Moreover, Akt-in did inhibit the membrane translocation of Akt, but not Btk. Both contain a PH domain and are also known to be translocalized upon PDGF stimulation. We cannot completely exclude the possibility that Akt-in may inhibit other PH-bearing kinases. However, our observations supported the notion that Akt-in binds specifically with Akt and therefore inhibits Akt kinase activity.
To develop a more powerful form of inhibitor, we created a dimer of Akt-in (2XAkt-in, repetitive sequences of 10 -23 of TCL1, NH 2-VTDHPDRLWAWEKGGGVTDHPDRLWAWEK-COOH). 2XAkt-in efficiently inhibited Akt kinase activity in a dose-dependent manner in an Akt kinase assay performed in vitro. Dissociation constant of 2XAkt-in was almost two times higher than the K d value of monomeric Akt-in (18 Ϯ 4.8 versus 10 Ϯ 2.7 M, Akt-in versus 2XAkt-in, mean Ϯ S.D., respectively) using Applied Biosystems 8500 Affinity Chip Analyzer. The results may facilitate to develop more efficient Akt inhibitors for therapeutic approaches.
Akt-in, which inhibits association of PtdIns with Akt, is the first molecule to demonstrate specific Akt kinase inhibition potency. Akt-in inhibited anti-apoptosis and tumor cell growth in vivo without any toxic effects. Given the pivotal role of Akt kinase as a core intracellular survival factor in the molecular mechanisms of human neoplastic diseases, the results should facilitate the design of Akt-specific inhibitors for human cancer therapy.