Specific Inhibition of the Akt1 Pleckstrin Homology Domain by D-3-Deoxy-Phosphatidyl-myo-Inositol Analogues¹

Introduction

The Ser/Thr protein kinase Akt, also called protein kinase B or related to A- and C-kinase, is a downstream target of PI-3 kinase³ (1, 2). PI-3 kinase phosphorylates the D-3-hydroxyl position of the myo-inositol ring of PtdIns (3) to generate the PtdIns-3-phosphates, PtdIns(3)P, PtdIns(3,4)P₂, and PtdIns(3,4,5)P₃ (4). PI-3 kinase is activated by many growth factor receptors and oncogenic protein tyrosine kinases (5–7), as well as by p21^Ras (8), leading to increased cell growth and inhibition of apoptosis (9, 10). PI-3 kinase expression is increased in ovarian cancer (11), and it is constitutively activated in human small cell lung cancer cell lines, where it leads to anchorage-independent growth and has been suggested to be a cause of metastasis (12). However, the major role for PI-3 kinase in cancer cell growth is its role in survival signaling mediated by Akt to prevent apoptosis (13).

Akt mediates a variety of biological responses, including the inhibition of apoptosis and promotion of cell survival (reviewed in Ref. 2). There are three mammalian isoforms: (a) Akt1/α; (b) Akt2/β; and (c) Akt3/γ (2). PtdIns-3-phosphates formed by PI-3 kinase (14) present in the inner leaflet of the plasma membrane bind to the PH domain of Akt (15, 16), which causes the translocation of Akt from the cytoplasm to the plasma membrane (17). Akt is then activated by phosphorylation on Ser⁴⁷³ and Thr³⁰⁸ by PDK-1 and integrin-linked kinase, respectively (18, 19). Phosphorylated Akt can then detach from the plasma membrane and move to the nucleus (17, 20). Activated Akt phosphorylates proteins, such as Bad, an inhibitor of apoptosis, FRAP, an activator of p70^S6k, which is required for cell cycle progression, caspase-9, forkhead transcription factors, and nuclear factor κB, thereby regulating cell proliferation and promoting cell survival (reviewed in Ref. 1). Akt1 is overexpressed in gastric adenocarcinoma (21), and Akt2 is overexpressed in breast cancer (22), ovarian cancer (22, 23), and pancreatic cancer (24).

Activation of Akt is negatively regulated by the tumor suppressor protein PTEN/MMAC, a tensin homologue detected in chromosome 10 and mutated in multiple advanced cancer/phosphatase (25). PTEN is a dual specificity tyrosine-threonine/lipid phosphatase that dephosphorylates the 3-position of PtdIns-3-phosphate (26, 27), thus inhibiting the PI-3 kinase/Akt signaling pathway (26, 28). In PTEN-deficient mouse embryo fibroblasts, constitutively elevated Akt activity together with decreased sensitivity to cell death in response to apoptotic stimuli demonstrate a role for PTEN as a negative regulator of cell survival (25). Somatic mutations of PTEN have been reported in several types of human tumors, including those from brain, breast, endometrium, kidney, and prostate (29, 30). In addition, germ-line mutations of the PTEN gene have been shown to be associated with Cowden’s disease, an autosomal dominant cancer predisposition syndrome associated with an elevated risk for tumors of the breast, thyroid, and skin (31).

Inhibition of Akt signaling thus offers the opportunity to inhibit a major survival signaling pathway found in many types of human cancers, and it further provides a unique opportunity to replace the activity loss of the tumor suppressor protein PTEN. We have used a novel strategy to inhibit Akt based on the use of DPIs that cannot be phosphorylated by PI-3 kinase. We demonstrate that the DPIs bind specifically to the PH domain of Akt, probably preventing Akt’s translocation to the plasma membrane and its activation by PDKs, thus inhibiting cell growth and inducing apoptosis.

Materials and Methods

Reagents.

DPI 1-[(R)-2,3-bis(hexadecanoyloxy)propyl hydrogen phosphate], its ether lipid analogue DPIEL, and its carbonate analogue DCIEL (Fig. 1) were synthesized as described previously (32, 33). L-α-PtdIns was obtained from Sigma Chemical Co. (St. Louis, MO). The cDNA encoding for myristylated HA-tagged human Akt1/α in the pCMV6 plasmid was a generous gift from Dr. L. Karnitz (Mayo Clinic, Rochester, MN). Human PDGF-BB homodimer was purchased from Genzyme (Cambridge, MA). Polyclonal antibodiesto phospho-Ser⁴⁷³-Akt, phospho-Thr³⁰⁸-Akt, phospho-Ser¹³⁶-BAD, pan-phospho-PKCs, and phosho-Ser²⁴¹-PDK-1 were all purchased from New England Biolabs-Cell Signaling (Beverly, MA); a goat polyclonal antibody to Akt1 and mouse monoclonal p-Erk (E-4) antibody were from Santa Cruz Biotechnology (Santa Cruz, CA). Recombinant Akt PH domain, polyclonal antibodies to Akt recognizing Akt1, 2 and 3 isoforms, a sheep polyclonal antibody against the PH domain (amino acids 1–149) of human Akt1, and antiphosphotyrosine antibodies (clone 4G10) coupled to agarose were purchased from Upstate Biotechnologies, Inc. (Lake Placid, NY). Anti-HA antibody (clone 12CA5) was purchased from Exalpha Biological, Inc. (Boston, MA). Donkey antirabbit IgG conjugated to peroxidase was purchased from Amersham Pharmacia Biotech (Piscataway, NJ), and rabbit antisheep IgG conjugated to peroxidase was obtained from Chemicon (Temecula, CA). The Renaissance chemiluminescence kit used for Western Blot detection was obtained from NEN Life Science Products, Inc. (Boston, MA). Akt/GSK, a peptide substrate related to the phosphorylation site of GSK-3, was obtained from Upstate Biotechnolgies. The LipoTAXI system used for transient transfection was purchased from Stratagene Cloning Systems (La Jolla, CA).

Cells.

Mouse NIH3T3 embryo-derived fibroblastic cells, human MCF-7 breast cancer cells, human HT-29 colorectal adenocarcinoma cancer cells, human DU-145, and LNCaP prostate cancer cells were obtained from the American Tissue Type Culture Collection (Rockville, MD). All cell lines were maintained in bulk culture in DMEM supplemented with 10% FBS and passaged using 0.25% trypsin and 0.02% EDTA.

Measurement of Phospho-Akt1 in NIH3T3, MCF-7, DU-145, and LNCaP Cells.

Cells were grown in 35-mm culture dishes in DMEM with 10% heat-inactivated FBS, 4.5 grams/liter glucose, 100 units/ml penicillin, and 100 μg/ml streptomycin in a 5% CO₂ atmosphere at 37°C to 75% confluence. Sixteen hours before the study, the medium was replaced by DMEM without FBS. The cells were incubated with the DPIs in DMEM for 4 h and stimulated with 50 ng/ml PDGF for 1 h (for NIH3T3 cells), with 100 ng/ml epidermal growth factor for 30 min (for MCF-7, DU-145, and LNCaP). Control cells were incubated with DMEM without growth factor. The culture media was aspirated, and the cells were lysed in 50 mm HEPES (pH 7.5), 50 mm NaCl, 1% NP40, 0.25% sodium deoxycholate, 1 mm EDTA, and 1 mm sodium orthovanadate. A measurement of 20 μg of total cell lysates was boiled for 5 min, and the samples were loaded on a 12% acrylamide/bisacrylamide gel and separated by electrophoresis at 160 V for 40 min (34). Proteins were electrophoretically transferred to polyvinylidene fluoride membranes, preincubated in blocking buffer (137 mm NaCl, 2.7 mm KCl, 897 μm CaCl₂, 491 μm MgCl₂, 3.4 mm Na₂HPO₄, 593 mm KH₂PO₄, and 5% BSA), and incubated with anti-phospho-Ser⁴⁷³-Akt polyclonal antibody. Immunoreactive bands were detected using donkey antirabbit IgG peroxidase-coupled secondary antibody and detected using the Renaissance chemiluminescence kit on Kodak X-Omat Blue XB films. Bands were quantified using Eagle Eye software (Stratagene). The detection of pPKCs, pPDK-1, and pERK was performed as described for pAkt using specific anti-phospho antibodies.

Akt1 Kinase Assay.

The kinase assay was performed according to the manufacturer’s instructions (UBI). Briefly, NIH3T3 cells were grown in T75-cm² flasks (∼5.1⁶cells/flasks), stimulated with 50 ng/ml PDGF for 10 min, and then lysed with 50 mm Tris-HCl (pH 7.4), 1% NP40, 0.25% sodium deoxycholate, 150 mm NaCl, 1 mm ethylene glycol bis(β-aminoethylether)N,N,N′,N′-tetraacetic acid, 1 mm p-toluenesulfonyl fluoride, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mm Na₃VO₄, 1 mm NaF, and 1 μm microcystin. The lysate was mixed with 4 μg of rabbit antihuman Akt1 polyclonal antibody coupled to protein G agarose beads (UBI) for 2 h at 4°C. After several washes in 20 mm 3-(N-morpholino) propane sulfonic acid (pH 7.2), 25 mm glycerol phosphate, 1 mm DTT, and 1 mm sodium orthovanadate, the immunoprecipitated Akt1 was preincubated for 1 h with the DPI analogue concentrations ranging from 10 to 20 μm at room temperature. An in vitro kinase assay was performed for 10 min at 30°C using as substrate 10 μl of a 0.4 mm solution of Akt/GSK and 10 μCi/sample of [γ-³²P]-ATP (1 mCi/100 μl, 3000 Ci/mmol; NEN Life Science Products). To each sample, 20 μl of 40% trichloroacetic acid were added, and after 5 min, 40 μl of the supernatant fraction were transferred to a P81 phosphocellulose paper square (Whatman International, Ltd., Kent, United Kingdom). The assay squares were washed three times with 0.75% phosphoric acid and once with acetone, and the radioactivity was determined by liquid scintillation counting.

Cell Growth Inhibition.

Mouse NIH3T3 cells were seeded at a density of 30,000 cells/24-mm diameter dish in DMEM with 10% FBS. After 16 h, the DPIs were added to the culture medium at concentrations ranging from 0.5 to 20 μm. Three days later, cells were counted using a hemocytometer and trypan blue exclusion to discriminate viable from dead cells. All measurements were made in triplicate.

Measurement of Apoptosis.

The annexin V apoptosis kit (Roche, Indianapolis, IN) was used according to manufacturer instructions with minor modifications. Briefly, prostate cancer cells were incubated with DPIEL for 24 h, collected, and washed in 137 mm NaCl, 2.7 mm KCl, 897 μm CaCl₂, 491 μm MgCl₂, 3.44 mm Na₂HPO₄, and 593 mm KH₂PO₄. Cells were resuspended in 20 μl of diluted FITC-annexin V/propidium iodide stain. Cells were incubated at 22°C for 20 min and then diluted with 500 μl of buffer supplied with the kit. The fluorescence emission of 10,000 cells was counted by flow cytometry for each condition using a FACScan (Becton Dickinson, San Jose, CA), and the results were analyzed using CELL-Quest software.

Myristoylated HA-Akt1 and GST-Akt1 Transfections in NIH3T3 Cells.

NIH3T3 cells were seeded into 100-mm diameter dishes, grown to 50% confluency in DMEM with 10% FBS, and transiently transfected using LipoTAXI (Stratagene Cloning Systems) with pCMV6 plasmid containing the cDNA insert for HA-tagged myr-Akt1 and pCMV5 plasmid containing the cDNA for GST-Akt1. After 24 h, the cells were reseeded into 24- and 40-mm diameter dishes for cell growth and Akt activity measurements, respectively. Expression was confirmed by Western blotting using anti-HA antibodies for the myr-Akt1 and monoclonal anti-GST antibodies for Akt (data not shown).

PI-3 Kinase Assay.

In vitro PI-3 kinase activity was measured as described previously (35). Briefly, NIH3T3 cells were stimulated with 100 ng/ml PDGF for 10 min and lysed, and PI-3 kinase was immunoprecipitated using antiphosphotyrosine antibody (clone 4G10) coupled to agarose beads. PI-3 kinase was eluted from the beads using 2 mm phenylphosphate, and PI-3 kinase activity was measured in the presence of 10 μm [γ-³²P]-ATP (1 mCi/100 μl, 3000 Ci/mmol; NEN Life Science Products) and 10 μl of a 10 mg/ml solution of L-α-PtdIns for 45 min at 37°C. Samples were quenched by the addition of 100 μl of 1N HCl, extracted with 400 μl of 1:1 chloroform/methanol, and centrifuged at 2000 × g for 1 min. For each sample, 25 μl of the lower organic phase were spotted onto the preabsorbent strip of an individual lane in a multichannel thin-layer chromatography plate (Whatman, Hillsboro, OR). Plates were developed in 65% n-propyl alcohol/35% 2 m acetic acid. [³²P]-labeled PtdIns product was quantitated by phosphorimager analysis (Molecular Dynamics, Sunnyvale, CA).

Translocation of Akt1 in NIH3T3 Cells.

NIH3T3 cells were grown to 50% confluency on glass coverslips in DMEM with 10% FBS. The cells were then incubated in DMEM without FBS for 2 h. DPIEL was added for 3 h, Wortmannin was added for 1 h, and cells were stimulated with 100 ng/ml PDGF for 60 min. After the stimulation, cells were fixed with 4% paraformaldehyde for 45 min and incubated for 1 h in a PBS buffer containing 137 mm NaCl, 2.7 mm KCl, 897 μm CaCl₂, 491 μm MgCl₂, 3.44 mm Na₂HPO₄, 593 mm KH₂PO₄ supplemented with 1% BSA, 0.1% Triton X-100, and 1% goat serum. Antibody against Akt was added overnight in the same buffer, and after several washes, slides were incubated with 10 μg/ml antigoat secondary antibody coupled to Alexa Fluor 568. Nuclear staining was performed using a solution of 1.5 ng/ml YOYO-1, according to the instructions from the manufacturer. Slides were examined under a Leica-TCS confocal microscope equipped with an Ar/Kr laser.

Lipid Binding to Akt-PH Domain.

Lipids, dissolved in chloroform at 1 mg/ml, were added to the wells of a polycarbonate 96-well plate at 100 μg/well, and the chloroform was allowed to evaporate. The wells were then blocked with 3% BSA fatty acid free for 1 h at room temperature. A measurement of 100 μl of a solution containing 0.5 μg/ml PH domain of Akt1 was added to each well overnight at 4°C. The plate was washed twice with PBS, and a sheep polyclonal antibody against the PH domain of Akt1 was added for 1 h at room temperature. The wells were washed twice in PBS, and bound PH domain was detected using a peroxidase-coupled secondary antisheep antibody, followed by a developing solution containing 2% 2,2′-azino-di-(3-ethylbenzethiazoline sulfonate) in 1 m sodium acetate and 0.03% H₂O₂. Plates were read at 405 nm using a SpectraMAX Plus spectrophotomer-ELISA reader (Molecular Devices).

Molecular Modeling and Docking Studies.

A homology model of the PH domain of human Akt 1 was built based on the crystal structure of Btk PH domain as described previously by Rong et al. (36). The model was used for docking studies of DPIs. To explore the interactions of DPIEL with the PH domain, a molecular model of the compound was built, energy was minimized performing the necessary replacements, hydrogen atoms were added using INSIGHT II (Version 2000; ACCELRYS, Inc., San Diego, CA)⁴, and refining was done by energy minimization (1000 cycles of steepest descent and 2 × 1000 cycles of conjugate gradient), while constraining the positions of the heavy atoms. The entire structure was then subjected to conjugate gradient minimization, and constraints were gradually removed until convergence was reached. After 100-ps molecular dynamic simulations at 300 K, energy minimization (2 × 1000 cycles of conjugate gradient minimization) was performed using DISCOVER-3 with Class II Force Fields.⁵ This structure subsequently served as the model structure for further energy refinement, docking, molecular dynamics, and complex formation. The low energy model was manually docked into the active site, and energies were computed using AFFINITY (37, 38).

To clarify the orientation of the ligand in the binding site, the electrostatic potentials at the Van der Waals surface of the active site pocket of the PH domain were determined using solvent surface calculations. Simulated annealing docking with 100 fs per stage duration (50 simulated annealing stages) was then performed to find the most favorable orientation. The orientation with the lowest intermolecular potential energy was obtained while moving the ligand and highly amended ligand-binding amino acid residues. The resulting ligand-PH domain complex trajectories were energy minimized using 1000 cycles of conjugate gradient minimizer, and the binding energies were calculated for each complex (39).

Results

DPIs Bind Specifically to the PH Domain of Akt1.

DPIs were synthesized based on the fact that these compounds are not able to be phosphorylated by PI-3 kinase on the 3-position of the myo-inositol ring (Fig. 1) but may act as competitors for Akt activation at the plasma membrane and, thus, behave as downstream inhibitors of the effects of PtdIns-3-phosphates. First, the ability of the DPIs to directly bind the PH domain of Akt was tested using an in vitro binding assay (Fig. 2). DAG and PtdIns used as control lipids bound only weakly to the PH domain of Akt. DPI and DPIEL bound to the PH domain of Akt to the same extent to PtdIns(3,4)P₂, whereas DCIEL showed a lesser binding.

To better understand these binding differences, molecular modeling and docking studies were used to calculate the relative binding affinity of DPIs for the PH domain of Akt (Table 1 and Fig. 3). The myo-inositol ring is involved in many interactions with Arg²⁵, Tyr³⁸, Arg⁴⁸, and Arg⁸⁶ (36). Docking analysis of the DPIs into the PH domain of Akt revealed that the myo-inositol moiety of DPIEL binds to the positively charged binding pocket of the PH domain with high energy (−109.3 kcal/mol) as compared with DPI or DCIEL (−59.4 and −56.6 kcal/mol, respectively). The 3-deoxy-myo-inositol ring of DPIEL is stabilized by hydrogen bond interactions with the terminal nitrogen atom of Arg²⁵, and the 1-phosphate group exhibits a strong interaction with the positively charged pocket using a network of hydrogen bonds: two bonds with 1-P=O ^… HN-Lys¹⁴ and the carbonyl oxygen atom of the side chain from the Arg²³ residue with a distance of 2.48 and 3.34 Å, respectively (Fig. 3). Two other hydrogen bonds are between the ether oxygen atoms of 1-phosphate and amine hydrogen atoms of Arg²³ and Arg²⁵ residues. The functional methoxy group exhibits a strong steric interaction between Lys¹⁴ and Thr²¹.

The comparison of the binding mode of DPIEL with DPI and DCIEL revealed that the 3-deoxy-myo-inositol ring of all three compounds exhibits a similar position. All hydroxyl groups of the DPI’s ring retained their interactions with Tyr³⁸ and Arg⁴⁸. In contrast, the interaction between the 1-phosphate group of DPI and Lys¹⁴ is lost. The 1-phosphate group position is shifted 1.2 Å toward Arg²⁵, and the hydroxyl group of the phosphate function displays a hydrogen bond interaction with the functional amine group of Arg²⁵ with a distance of 2.04 Å. For DCIEL, no hydrogen bond interactions were formed because of the absence of the 1-phosphate group. The position and orientation of 3-deoxy-myo-inositol ring is stabilized by participation of the 4,5-OH groups in hydrogen bond interactions with Arg²⁵ and Arg⁴⁸. DPIEL was calculated to bind much more strongly to the PH domain of Akt as compared with DPI and DCIEL. DPIEL docks into the PH domain of Akt and exhibits strong interactions with the amino acids thought to be involved into the binding of PtdIns-phosphates.

DPIs Inhibit Akt Kinase Activity in Vitro.

PtdIns phosphates are known to modulate Akt kinase activity (15). Thus, we studied the effects of DPIs and Wortmannin, a PI-3 kinase inhibitor (40, 41) on Akt1 kinase activity in an in vitro assay (Fig. 4). Akt1 kinase activity was increased in PDGF-stimulated cells. Akt activity was not inhibited by 5 μm Wortmannin but was inhibited by the DPIs with approximate IC₅₀s for DPI and DPIEL of 8 and 17.5 μm, respectively.

DPIs Inhibit Akt Activity in NIH3T3 Mouse Fibroblasts.

Akt1 is the major Akt isoform expressed in mouse NIH3T3 fibroblast cells where it represents 59% of the total Akt present. Akt2 and Akt3 are 23 and 18%, respectively (Fig. 5A). Akt1 activation measured by phosphorylation on Ser⁴⁷³, was increased ≤2-fold on PDGF stimulation of NIH3T3 cells (Fig. 5B, Lane 2 as compared with Lane 1 and Fig. 5C). In the presence of 1 μm Wortmannin, Akt1 phosphorylation was completely inhibited (Fig. 5B, Lane 3). DPIEL (Lane 6) and, to a lesser extent, DPI and DCIEL decreased Akt1 phosphorylation (Fig. 5B, Lanes 5 and 4, respectively). PtdIns was used as a control and did not affect the phosphorylation of Akt (Lane 7). Fig. 5C represents the quantification of phospho-Ser⁴⁷³-Akt from at least five different Western blots.

DPIs Inhibit Cell Growth and Induce Apoptosis in a Dose-dependent Manner.

Increasing concentrations of DPI, DPIEL, and DCIEL inhibited Akt1 phosphorylation in NIH3T3 cells in a dose-dependent manner (Table 2) with IC₅₀s (±SD, n = 4) of 27.4 ± 4.4 μm, 1.5 ± 0.3 μm, and 12.5 ± 2 μm, respectively. The DPIs also inhibited NIH3T3 cell growth with IC₅₀s for the DPI, DPIEL, and DCIEL of 17.6, 4.3, and 14.1 μm, respectively. Human cancer cell lines were also tested (36), and results are summarized in Table 2. Overall, DPIEL appeared to better inhibit the proliferation of NIH3T3 and MCF-7 cells. Interestingly, DPIEL and DCIEL had similar effects on HT29 cell proliferation. DPIEL inhibited Akt in DU-145 and LNCaP prostate cancer cells (Fig. 6A) and caused a dose-dependent increase in apoptosis (Fig. 6B).

DPIs Inhibit Bad Phosphorylation in a Dose-dependent Manner.

Akt phosphorylates the antiapoptotic protein Bad on Ser¹³⁶ (42, 43). Fig. 7, A and B shows the effects of the DPIs on Bad phosphorylation measured in cell lysates using a specific anti-phospho-Ser¹³⁶-Bad antibody (Panel A). Both DPI and DPIEL significantly inhibited Bad phosphorylation in PDGF-stimulated NIH3T3 cells. Panel B represents the quantification of phospho-Ser¹³⁶-Bad from at least five different Western blots.

DPIs Inhibit PI-3 Kinase Activity in Vitro.

The ability of the DPIs to inhibit PI-3 kinase was measured in an in vitro kinase assay (Fig. 8,A and B). DPI, DPIEL, and DCIEL (data not shown for DCIEL) inhibited PI-3 kinase activity in a dose-dependent manner with IC₅₀s (±SD) of 38.4 ± 4.1, 16.4 ± 2.9, and 15.5 ± 1.8 μm, respectively. A representative chromatogram is shown in Panel A.

DPIs Do Not Inhibit Other PH Domain-containing Proteins Involved in the Pathway.

PDK-1 is a PH domain containing serine kinase that phosphorylates Akt on Ser⁴⁷³ (18). It also phosphorylates conventional PKCs (44). To investigate the effects of DPIEL on PDK-1 activity, we have measured the phosphorylation of PKCs in MCF-7 breast cancer cells. DPIEL had no effect on PKC phosphorylation in MCF-7 cells (Fig. 9). The phosphorylation on Ser²⁴¹ of PDK-1 is necessary for its activity (45). We tested whether DPIEL can affect PDK-1 activation by using specific anti-Phosho-Ser²⁴¹-PDK-1 antibodies. No change in PDK-1 phosphorylation was observed in the presence of DPIEL in MCF-7 breast cancer cells (Fig. 9). Erk (p42/44) phosphorylation was also not affected by DPIEL. As shown in other cell lines (NIH3T3, Fig. 5B; DU145 and LNCaP; Fig. 6A), increasing concentrations of DPIEL inhibit Akt phosphorylation on Ser⁴⁷³ in MCF-7 cells.

Constitutive myr-Akt Activity Is Not Affected by DPIs.

To test the possibility that DPIs interact directly with the PH domain of Akt as suggested by the immunohistochemistry studies, NIH3T3 cells were transiently transfected with a myristylated (myr) Akt1 construct, which permanently localizes Akt at the plasma membrane where it is constitutively activated (46). The cells were incubated with DPI, DPIEL, or Wortmannin for 2 days and counted (Fig. 10A). The growth inhibitory effects of DPIEL were abolished in myr-Akt-transfected cells, and the growth inhibitory effects of DPI were considerably reduced. We also measured Akt1 phosphorylation in Akt1-transfected- and myr-Akt1-transfected NIH3T3 cells in the presence and absence of the DPIs (Fig. 10, B and C). Myr-Akt1 was phosphorylated in the absence of PDGF (46), and this phosphorylation was not affected by DPIEL (Fig. 10, B and C). DPI caused a small inhibition of myr-Akt phosphorylation. Unexpectedly, Wortmannin completely inhibited myr-Akt phosphorylation.

Discussion

Our work shows that the DPIs are inhibitors of the activation of Akt1 in NIH3T3 cells and that this inhibition is associated with increased apoptosis and the inhibition of the phosphorylation of at least one of the downstream targets of Akt, the cell survival protein, Bad (42, 43). The most active DPI was the ether lipid analogue DPIEL with an IC₅₀ for Akt1 inhibition in NIH3T3 cells of 1.5 μm. The DPIs were effective inhibitors of cell growth with IC₅₀s for DPI, DPIEL, and DCIEL of 18, 4, and 14 μm, respectively. The cell growth inhibitory effects of DPIEL on NIH3T3 cells could be prevented by expressing myristoylated-Akt1, which is a membrane-bound and constitutively active Akt (46). DPIEL had no effect on myristoylated-Akt1 phosphorylation. DPIs also inhibited Akt in mouse NIH3T3 cells, human MCF-7 breast cancer cells, and human LNCaP and DU-145 prostate cancer cells. Thus, the growth inhibitory effects of the DPIs appear to be associated with the inhibition of Akt.

The mechanism by which the DPIs could enter cells to inhibit Akt is not known, but related compounds with ether lipid groups, such as alkyl-lysophospholipids, i.e., edelfosine (or ET-18-OCH₃), miltefosine (HePC) rapidly enter cells and inhibit cell growth (47, 48). Edelfosine is able to inhibit Akt at higher concentrations than DPIEL (data not shown and Ref. 49). DPIEL appears to be a specific inhibitor of the PH domain of Akt. DPIEL had no effect on the activity of another PH domain containing protein in the Akt pathway, PDK-1. PDK-1 is constitutively located at the plasma membrane and requires phosphorylation on Ser²⁴¹ to be active (45). DPIEL did not affect the phosphorylation on Ser²⁴¹ and did not affect PDK-1 activity as measured by phosphorylation of PKCs, which are also substrates for PDK-1. DPIEL also did not significantly alter p42/44 Erk phosphorylation. DPIEL is able to inhibit PI-3 kinase (IC₅₀ 16 μm) and is a weak direct inhibitor of Akt1 kinase activity (IC₅₀ 17 μm). However, the inhibition of Akt1 activation and cell growth by DPIEL occurred at 10-fold lower concentrations. Thus, the predominant activity of DPIEL in cells at concentration caused cell death and apoptosis.

We studied two other DPI analogues, DPI itself and the carbonate analogue DCIEL that was synthesized to prevent possible breakdown by PtdIns-specific phospholipase C (50, 51). Both analogues were less potent than DPIEL at inhibiting Akt1 in cells and also less specific. DPI and DCIEL were inhibitors of PI-3 kinase at concentrations that inhibited Akt1 in cells. Although DPI is able to inhibit Akt kinase activity in vitro at a lower concentration than DPIEL (8 versus 17.5 μm, respectively), it was a weak inhibitor of Akt in cells, probably because DPI may be broken down by phospholipases. DPI, unlike DPIEL, inhibited myristoylated-Akt1 phosphorylation. In this respect, it was similar to the PI-3 kinase inhibitor Wortmannin (40, 41), which, surprisingly, almost completely blocked myristoylated-Akt1 phosphorylation in NIH3T3 cells. This observation suggests that PI-3 kinase, which has serine/threonine kinase activity (52), could be phosphorylating membrane-bound Akt1 directly. However, the Wortmannin concentration used in the cells was much higher than required to inhibit purified PI-3 kinase so that other mechanisms could be responsible for Wortmannin’s effects on myristoylated-Akt1 phosphorylation.

The PH domain is a phosphoinositide-binding motif found in a number of signal-transducing proteins, including Akt, that gives membrane-binding properties to the host proteins. The three-dimensional organization of individual PH domains gives different binding specificities (53) with the Group 1 PH domain (containing, e.g., IRS-1’s or Btk’s PH domains) recognizing PtdIns(3,4,5)P_3, the Group 2 PH domain recognizing PtdIns(4,5)P₂ (containing, e.g., the PH domains of mSos1 or βARK), and the Group 3 PH domain recognizing PtdIns(3,4)P₂ and PtdIns(3,4,5)P (containing, e.g., Akt; Ref. 54). We found that DPIs bind directly to the PH domain of Akt in an in vitro binding assay. However, the assay only shows that the DPIs bind to the PH domain of Akt and does not allow the calculation of relative binding affinities. Molecular modeling and docking analysis of DPIEL into the PtdIns-3-phosphate binding pocket of a variety of PH domains allowed us to demonstrate a high affinity for the binding of DPIEL to Group 3 PH domain of Akt (see Fig. 4 and data not shown). DPIEL bound strongly to the PH domain of Akt as compared with DPI and DCIEL. These results are consistent with the respective effects of each compound on Akt activity in cells. DPIEL appears to be the most potent Akt inhibitor in the cells because it binds the best to the PH domain of Akt. Moreover, we found that DPIEL bound with relatively low affinity to the Group 1 PH domains of the Group Btks and IRS-1, whereas no significant binding of DPIEL to the Group 2 PH domains of βARK and mSos1 PH was observed (data not shown). As noted previously we also found that DPIEL did not block the activity of the Group 3 PH domain protein PDK-1 in cells. Thus, the docking and activity studies suggest that DPIEL has also specificity for Akt among the PH domain containing proteins investigated.

There are a number of cellular mechanisms by which Akt inhibition by the DPIEL may occur (Fig. 11). We have found using immunohistochemistry that DPIEL blocks the translocation of Akt1 from the cytoplasm to the plasma membrane in NIH3T3 cells where it normally binds to PtdIns-3-phosphates in the inner leaflet of the membrane (data not shown). Taken together, a first mechanism that can be suggested is that DPIEL could block the formation of PtdIns-3-phosphates. However, inhibition of PI-3 kinase by DPIEL only occurs at concentrations of an order of magnitude higher than required to inhibit Akt. We have also not seen a decrease in the levels of tumor PtdIns(3,4,5)P₃ by DPIEL measured by immunohistochemistry.⁶ A second mechanism is that DPIEL, or a metabolite, in the cytoplasm binds the PH domain of Akt, thus trapping Akt in the cytoplasm. We found that DPIEL binds to the PH domain of Akt1 to the same extent as PtdIns(3,4)P₂. We do not know the localization of DPIEL in the cell or whether it is metabolized. We have observed that DPIEL treatment gives a more diffuse immunohistochemical staining of Akt1 in the cytoplasm than the one seen in nontreated cells (data not shown). Akt has been reported to oligomerize through a PH domain interaction, which could contribute to the activation of Akt (55). By binding to the PH domain, DPIEL, or a metabolite, may prevent the oligomerization of Akt. A third mechanism is that DPIEL is in the plasma membrane as described similarly for alkyl-lysophospholipids (48). Akt binds to DPIEL at the membrane and by preventing Akt phosphorylation, induces the release of Akt back into the cytosol. Such a mechanism could explain the diffuse immunohistochemical staining observed as Akt is shuttling from the cytosol to the plasma membrane without complete activation (data not shown). A fourth mechanism is a reorganization of the plasma membrane by DPIEL, which prevents the binding of Akt to PtdIns(3,4,5)P₃ without there being direct binding of DPIEL to Akt. Although we have no evidence to support or disprove such a mechanism, our finding that DPIEL binds to the PH domain of Akt and the docking studies that show a high affinity specific interaction with Akt’s PH domain compared with the PH domain of other proteins and compared with the other DPIs suggest that a mechanism with direct binding to the PH domain of Akt is most likely to occur in the cells.

In summary, we have shown that the DPIs bind to the PH domain of Akt and inhibit Akt activation in cells. DPIEL was the most active compound, and it prevented Akt translocation to the plasma membrane. DPIEL had no effect on myr-Akt phosphorylation and was much less effective at inhibiting the growth of these cells. The results suggest that the growth inhibitory effects of DPIEL are caused by inhibition of Akt activation probably after a block of Akt’s translocation from the cytoplasm to the plasma membrane. This is attributable, at least in part, to the binding of DPIEL or a metabolite to the PH domain of Akt. Thus, the DPIs may represent a new class of potential anticancer drugs with a novel mechanism of action causing specific inhibition of the PH domain-dependent translocation of Akt.