A Novel Inhibitor of AKT1–PDPK1 Interaction Efficiently Suppresses the Activity of AKT Pathway and Restricts Tumor Growth In Vivo Free

The serine/threonine kinase AKT (also known as protein kinase B, PKB) belongs to the AGC family of kinases and has a central role in regulating the survival and proliferation of normal and malignant cells (1, 2). Abnormal activation of the AKT pathway has been commonly described in prostate, breast, liver, and colorectal carcinomas (3–7). Furthermore, the constitutive activation of the PI3K/AKT pathway confers resistance to many chemotherapeutic drugs and is a poor prognostic marker for a number of cancer types (8); thus, targeting AKT pathway is a promising strategy in tumor therapy.

Correspondingly, a variety of AKT inhibitors have been developed to date (9). Most of these are the inhibitors of kinase activity and by binding to its kinase active site act as ATP competitors. Because the ATP binding pocket of AKT/PKB, PKA, and PKC is highly homologous, these inhibitors have an activity toward PKA and PKC (10). Phosphatidylinositol (PI) analogues block PI(3,4,5)P3 binding to AKT, prevent its translocation to plasma membrane and subsequent activation (11). The PI analogues are more selective to AKT, but they have a potential to interfere with the function of other pleckstrin homology (PH) domain-containing proteins (9). In addition, a few inhibitors of PDPK1, the critical upstream activator of AKT, have been developed, which suppress the kinase activity of PDPK1 either as ATP competitors or by allosteric mechanisms (12, 13). Nevertheless, the data regarding their usability in clinical setting are scarce.

Protein–protein interactions (PPI) are key events in vast majority of molecular signaling pathways and are thus attractive targets for intervention with small molecular compounds. Despite of the challenges that the researchers face when designing the PPI inhibitors, the number of small molecules that specifically interfere with PPIs, which are crucial for the integrity of the pathway of interest, is constantly growing (14). Although there exist a number of small molecule inhibitors of the AKT pathway, to our best knowledge, no true PPI inhibitors that target this pathway have been identified so far. However, the compounds, which lock the intramolecular interaction between the kinase active site and PH domain of the AKT molecule in the inactive state, can be viewed as a specific subclass of allosteric PPIs (15, 16). At least one such compound has entered successfully into the clinical trials underlining the potential of the small molecules with PPI-inhibitory properties in tumor therapy (17).

To identify AKT pathway inhibitors with principally novel mechanism of action, we decided to target the critical step in AKT pathway—the physical interaction of PDPK1 and AKT1, which results in phosphorylation of AKT at T308 and its activation (18). For the detection of the PDPK1–AKT1 interaction, we took advantage of the protein complementation assay (PCA) based on Renilla luciferase (Rluc; ref. 19). First, two specially crafted halves of Rluc, which do not possess the luciferase activity, are fused to the interaction partners. Next, the fusion proteins are expressed in cells and in the case these proteins interact the Rluc fragments are brought in close proximity, which results in reconstitution of the enzyme activity (Fig. 1A). Due to the reversible nature of the interaction between the Rluc fragments, the inhibition of protein–protein interaction unfolds the functional enzyme and destroys the Rluc activity. Using a small molecule library screen in live cells, we identified one compound—NSC156529—that interacted preferentially with PDPK1, inhibited AKT1 phosphorylation, and suppressed AKT-mediated signal transduction to several AKT1 substrates. Furthermore, the discovered compound efficiently decreased the proliferation of human cancer cells in vitro and inhibited tumor growth in a prostate tumor xenograft model in vivo. Interestingly, in addition to the reduction in the mitotic activity in NSC156529-treated tumors, the tumor cells exhibited a substantial increase in the expression of prostate differentiation markers, suggesting that the inhibition of tumor growth was achieved at least in part through enhanced differentiation of tumor cells toward prostate epithelium.

Low-passage cultures (p3-8) of cell lines (source; date received) were used in this study: human non–small cell lung carcinoma cells H1299 (ATCC; 03.2004), human embryonic kidney cells HEK293 (ATCC; 02.1993), human hepatoma cells Hep3B (ATCC; 2011), PC-3 (ATCC; 05.2006), and K07074 mouse primary liver tumor cell line (established in-house; 09.2013). H1299, Hep3B, HEK293, and K07074 cells were cultivated in IMDM medium (Lonza). The normal primary human fibroblasts and osteoblasts [a kind gift from Aare Märtson and Siim Suutre (Clinic of Traumatology and Orthopedics, Tartu University Clinics)] were cultivated in IMDM medium; for osteoblasts, IMDM was supplemented with 100 μg/mL ascorbic acid (Sigma). Human prostate cancer cells PC-3 were grown in RPMI medium (Gibco). All media were supplemented with 10% (v/v) FBS, 50 units/mL penicillin, 50 μg/mL streptomycin, and 1 mmol/L l-glutamine. Cells were maintained at 37°C in a humidified atmosphere supplemented with 5% (v/v) CO₂.

The identity of cell lines was routinely verified using species-specific primers and karyotyping.

NCI Diversity Set I (National Cancer Institute) containing 2,000 small molecular compounds was used to perform the screen. H1299 cells were transfected either with 6 μg pCNEO empty vector, 3 μg of p53-F1, HDM2-F2, Akt1-F1, and Pdpk1-F2, or 2 μg of full-length Renilla luciferase expression vector (pRluc). Forty-eight hours after the transfection, cells were incubated with DMSO, GSK 2334470, Nutlin-3, or with compounds originating from the chemical library for 2 hours at concentration 10 μmol/L. Next, luciferase activity and cell number were measured using ViviRen Live Cell Substrate (Promega), CellTiter-Glo Reagent (Promega), and TECAN InfiniteM200 PRO plate reader according to the manufacturer's instructions. The luciferase reads were normalized to the cell viability counts, and the average of 2 experiments for each data point was calculated. The testing of 74 initial hits was performed in 4 replicates as described above. In addition, H1299 cells transfected with pRLuc were used to evaluate the effect of the compounds on Renilla luciferase.

To assess the quality of the screen setup, the Z-factor was measured using the constructs encoding the p53-F1 and HDM2-F2 fusion proteins and the inhibitor of p53–HDM2 interaction—Nutlin-3 (10 μmol/L). The calculated value of the Z-factor for Rluc PCA was 0.5, demonstrating its suitability for detecting the inhibition of the protein–protein interactions. Retrospectively calculated Z-factor for AKT1–PDPK1 interaction when using 10 μmol/L AKT–PDPK1 inhibitor NSC156529 was 0.75.

In situ proximity ligation assay (PLA) allows a specific and sensitive detection of protein–protein interactions. PLA is based on the immunodetection of the interaction partners with specific primary and oligo-tagged secondary antibodies. If the proteins of interest interact, the in situ PCR generates specific DNA sequence, which is detected by the hybridization with fluorescently labeled oligos (20).

PC-3 cells were plated on cover slips, incubated with 20 μmol/L NSC156529, 20 μmol/L PDPK1 inhibitor, or 20 μmol/L DMSO for 4 hours, and fixed with 4% paraformaldehyde (AppliChem) for 10 minutes at room temperature. Subsequently, cells were permeabilized with methanol for 10 minutes at −20°C, followed by washing for 2 × 2 minutes in PBS, 0.05% Tween-20 at room temperature. Slides were blocked with in situ blocking buffer for 1 hour at room temperature, followed by the incubation with primary antibodies recognizing AKT1 (anti-mouse, 1:100; Cell Signaling) and PDPK1 (anti-rabbit, 1:100; Cell Signaling) overnight at 4°C. The negative control was performed using only one primary antibody. After washing the slides in washing buffer A for 2 × 5 minutes, the PLA probes were incubated in 40 μL in situ blocking buffer for 1 hour at +37°C. Next, the cells were washed for 2 × 5 minutes in washing buffer A followed by the incubation in 40 μL ligation solution for 1 hour at +37°C. The slides were washed for 2 × 2 minutes in wash buffer A and incubated in 40 μL amplification–polymerase solution for 100 minutes at +37°C. The slides were washed in 1× wash buffer B for 2 × 10 minutes and in 0.01× wash buffer B for 1 minute. The cells were left to dry, then mounted with mounting medium, incubated for 15 minutes, and imaged. The images were acquired with Olympus BX-71 fluorescent microscope using bundled Cell-R software. Five different fields per slide were acquired and analyzed using Imaris7.6.3 software (Bitplane AG). The number of PLA signals per cell was counted defining a true signal as a local intensity maximum above a background threshold. The same input parameters were used throughout all experiments. All the experiments were repeated at least 3 times.

The lysates obtained from PC-3 cells treated with 20 μmol/L NSC156529, 20 μmol/L PDPK1 inhibitor, and 20 μmol/L DMSO were separated by SDS-PAGE and blotted using antibodies listed in Supplementary Table S1. The signals were detected using Fujifilm LAS4000 image analysis system. The experiment was repeated 3 times.

Twenty-four hours after seeding, the PC-3 cells were incubated with 10 μmol/L small molecular compounds, 10 μmol/L PDPK1 inhibitor, and 10 μmol/L DMSO for 24, 48, 72, and/or 96 hours. Cell proliferation was assessed using the CellTiter-Glo luminescent cell viability assay (Promega) according to the manufacturer's protocol. Each experiment was carried out in triplicate, and at least 3 independent experiments were performed.

Xenograft experiments were performed as a service by vivarium of the Department of Gene Technology, Tallinn Technology University, Tallinn, Estonia, using proprietary protocols and cell line. Briefly, 5 × 10⁶ PC-3 human prostate carcinoma cells stably expressing EGFP were injected s.c. into the right flank of athymic nude (Foxn1^nu/nu) female mice. Tumors were grown 17 days until they reached a median size of 29 to 32 mm³. For injections, the compound was dissolved in vehicle (13.75% DMSO, 1.25% Tween80, 85% double-distilled sterile water). Animals were randomly divided into four groups: group A (n = 16) received 10 mg/kg of compound, group B (n = 11) received 5 mg/kg of compound, group C (n = 10) received 1 mg/kg of compound, and group D (n = 16) was treated with vehicle only. Subcutaneous injections were performed approximately 1 cm around the tumor every other day during 28-day treatment period. Tumor sizes were measured with external caliper, and tumor volumes were calculated as follows: width × width × length/2 (21). Simultaneously, the tumor size was measured by in vivo fluorescence imaging using IVIS Lumina I In Vivo Imaging System (Caliper Life Sciences) and Living Image 4.0 software (Caliper Life Sciences). Data were collected at different wavelengths (excitation: 430 nm and 465 nm; emission filter: 515–575 nm). During the measurements, mice were anesthetized using a vapor Isoflurane inhalation narcosis system Gas Anesthesia System for IVIS. All animal experiments were approved by local ethics authorities.

Body weight and the general condition of mice were assessed at least 3 times a week. Blood samples were obtained from mice prior sacrifice. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) values were measured using standard laboratory routines at the Central Laboratory of the Tartu University Clinics. Finally, mice were sacrificed, tumors were removed and weighed, and tumor samples were embedded in optimal cutting temperature and stored at −80°C for further use.

Microscale thermophoresis (MST) enables the detection of interactions between various molecules regardless of their size or origin. Measuring the changes in fluorescence, which reflect the motility of fluorescently tagged molecules in a microscopic temperature gradient caused by binding an unlabeled interaction partner, causes the shift in the motility of labeled molecules (22). Binding kinetics of AKT1–EYFP and PDPK1–EYFP to NSC156529 was measured by MST with a NanoTemperMonolith NT.115 instrument (NanoTemper Technologies). H1299 cells were transfected either with 5 μg AKT1–EYFP or PDPK1–EYFP expressing vector. The resulting cell lysates were diluted in reaction buffer (50 mmol/L Tris–HCl, pH 7.5, 150 mmol/L NaCl, 10 mmol/L MgCl₂, 0.05% Tween20, 0.05% Tween80). Serial dilutions of NSC156529 were prepared with the identical buffer (concentrations 200 μmol/L–6.1 nmol/L). For thermophoresis, each ligand dilution was mixed with one volume of labeled proteins. After 1-hour incubation at 4°C, approximately 4 μL of each solution was filled into Monolith NT standard-treated capillaries (NanoTemper Technologies GmbH). Thermophoresis was measured using a Monolith NT.115 instrument (NanoTemper Technologies GmbH). The signal values were analyzed using NT.Analysis software version 1.5.41 (NanoTemper Technologies).

RNA was isolated from human tumor xenografts using TRIzol (GIBCO) according the to manufacturer's instructions. Total RNA (1 μg) was reverse transcribed with a RevertAid First Strand cDNA Synthesis kit (Thermo Scientific) according to the manufacturer's instructions. All qPCR reactions were carried out with an ABI Prism 7900 HT Sequence Detection System (Applied Biosystems), and the data acquired were analyzed with ABI SDS software (version 2.1 from Applied Biosystems). The primer sequences used are listed in Supplementary Table S2. Target cDNAs were normalized to endogenous mRNA levels of the housekeeping reference gene Hprt1. The PCR reactions were performed at least 3 times for each sample. Statistical significance was assessed with the Student t test.

Sections (5-μm-thick) were treated with 4% paraformaldehyde (PFA; Applichem) and 0.1% Triton-X-100 for 10 to 15 minutes at room temperature. Sections were blocked with 4% normal goat serum (NGS) and incubated with primary antibodies (see antibody list in Supplementary Table S3) followed by incubation of fluorochrome-conjugated secondary antibodies (see Supplementary Table S3), both for 1 hour at room temperature. The images were acquired with Olympus BX-71 fluorescent microscope using bundled Cell-R software.

See the Supplementary Information page for further details.

To identify the inhibitors of AKT1 and PDPK1 interaction, we carried out a small chemical library screen using the NCI Diversity Set I small molecular compound library consisting of 2,000 compounds in live cells. We took advantage of the PCA based on split Renilla reniformis luciferase (Rluc) fragments (F1 and F2) fused to the proteins of interest (ref. 19; Fig. 1A).

The screening experiments were performed in H1299 cells, which are known to harbor an active AKT signaling (23, 24) and are well suited for experiments involving DNA transfection (Fig. 1B). To rule out unspecific modulation of the luciferase readouts, quadruple replicates of H1299 cells transfected with pCNEO cloning vector (see Supplementary Information); H1299 cells transfected with p53-F1 and HDM2-F2 treated with either DMSO or Nutlin-3; and H1299 cells transfected with AKT1-F1, PDPK1-F2 treated with DMSO were used (Supplementary Table S4). To eliminate the error caused by the well-to-well variation in cell numbers, the luciferase reads were normalized to the viability counts as described in “Materials and Methods.” In addition, the viability counts were used to identify and exclude toxic compounds. The cells treated with vehicle only (DMSO) were used to set the baseline values. The cells transfected with p53-F1 and HDM2-F2 fusions treated with Nutlin-3 were used as assay controls. At least 2-fold reduction in luciferase signal upon Nutlin-3 addition was considered to indicate a successful experiment.

The screen was performed in two steps. In the first step, the 2,000 chemicals were screened for their ability to inhibit the interaction of transfected AKT1-F1 and PDPK1-F2.

The overall selection criteria for selection of compounds were set as follows: (i) at least 50% reduction of normalized AKT1-F1–PDPK1-F2 interaction values when compared with the average of DMSO-treated AKT1-F1– and PDPK1-F2–transfected cells; (ii) 25% or less reduction in cell viability counts when compared with the average of DMSO-treated controls; (iii) less than 10% reduction in full-length Rluc signal when compared with the average reading of DMSO-treated full-length Rluc-transfected cells (Fig. 1C).

We identified 74 chemicals out of 2,000, which inhibited the AKT1-F1–PDPK1-F2 interaction at least 50% or more. The toxicity criterion was not applied at this stage. However, 412 chemicals from 2,000 elicited acute toxicity toward H1299 cells (downregulation of viability counts 25% or more). In the next step, the identified 74 chemicals were retested for AKT1–PDPK1 PCA inhibition; in parallel, the inhibition of full-length Rluc and cell toxicity were evaluated. All 74 chemicals inhibited the AKT1–PDPK1 PCA; however, 29 chemicals of 74 concomitantly inhibited the activity of full-length Rluc, and 13 chemicals were toxic for cells. Nine of 45 chemicals, which specifically inhibited the AKT1–PDPK1 PCA, reduced the viability counts 25% or more and thus were excluded from further analysis. As a result, 36 chemicals were considered for the next evaluation step (Supplementary Table S5).

The hallmark of AKT1 activation by PDPK1 is the phosphorylation of AKT1 at T308 (18, 25, 26). To evaluate the ability of the 36 chemicals identified in the previous step to inhibit the phosphorylation of T308, we performed a Western blot analysis of H1299 cells treated with these chemicals (Fig. 2A). To minimize possible indirect effects on the AKT1 phosphorylation, the treatment was carried out for 4 hours. GSK 2334470, which directly inhibits the kinase properties of PDPK1, was used as a positive control. By using an antibody, which specifically recognized phospho-T308 on AKT1, we identified 12 compounds of 36, which inhibited AKT1 protein phosphorylation to a various degree. None of these compounds changed significantly the basal levels of AKT1 and PDPK1 proteins. In parallel, the cultured cells treated with these compounds were carefully examined for signs of toxicity and 4 chemicals of 12, which did not cause notable changes in cell density and morphology during the 4-hour treatment, were chosen for further analysis (Fig. 2A and B; Supplementary Fig. S1A–S1D).

Next, we sought to find out whether the chemicals selected in the previous steps could inhibit the interaction of endogenous AKT1 and PDPK1 proteins. For this, we utilized in situ PLA in PC-3 prostate cancer cells known to harbor an activated AKT signaling pathway (27). The PC-3 cells, seeded to glass slides, were incubated with the 4 chemicals selected in the previous step and DMSO for 4 hours. A number of AKT1–PDPK1 interaction sites were readily detected in the DMSO-treated PC-3 cells (Fig. 2C; Supplementary Fig. S2A), whereas the count of AKT1–PDPK1 interaction sites was significally decreased in the cells treated with the NSC156529 (Fig. 2C; Supplementary Fig. S2G) compared with the AKT2–PDPK1 or AKT3–PDPK1 interactions (Supplementary Fig. S2H and S2I) indicating the preferred inhibition of AKT1–PDPK1 interaction by NSC156529. The number of AKT1–PDPK1 interaction sites in cells treated with NSC5113, NSC15784, and NSC292596 (Supplementary Fig. S2D–S2F and S2J) remained at the same level as in the control cells. Cells treated with PDPK inhibitor and cells with no primary antibody incubation were used as controls (Supplementary Fig. S2B and S2C). Based on this, experiment one compound, NSC156529, was selected for further analysis.

Next, we treated H1299 cells transfected with AKT-F1 and PDPK-F2 with increasing concentrations of NSC156529. The results of luciferase assay showed that NSC156529 dose-dependently inhibited AKT1–PDPK1 interaction (Fig. 3A) with calculated IC₅₀ of 3.862 μmol/L.

The experiments conducted so far showed that NSC156529 prevents the interaction between AKT1 and PDPK1. This might be caused by direct interaction of NSC156529 either with AKT1 or PDPK1. To test this hypothesis, we studied the potential interaction between NSC156529 and EYFP-tagged AKT1 or PDPK1 in cell lysates using MST. For MST analysis, either EYFP-tagged AKT1 or PDPK1 (AKT1–EYFP; PDPK1–EYFP) was transfected into H1299 cells, the cell lysates containing fluorescently labeled proteins were treated with serially diluted unlabeled NSC156529 (concentration range, 200 μmol/L–6.1nmol/L) and analyzed using a dedicated MST instrument as described in “Materials and Methods.” Although no stable binding of NSC156529 to AKT1–EYFP could be detected (Supplementary Fig. S3A), a binding event with the Kd of 981 ± 131 nmol/L was determined for the NSC156529 and PDPK1–EYFP interaction (Supplementary Fig. S3B), suggesting that NSC156529 binds directly to PDPK1.

Two well-defined docking sites have been identified on PDPK1 molecule: the ATP pocket and PIF pocket, which is required for the phosphorylation of selected PDPK1 targets (28). To examine the possibility that NSC156529 binds to these sites, docking calculations using Glide (Glide, v6.2, Schrödinger, LLC, 2014) with standard settings were performed using the previously published PDPK1 crystal structure [PDB ID–1H1W, (29)]. The results of the calculations demonstrated no reliable binding of NSC156529 to these two sites, suggesting that NSC156529 most likely interacts with PDPK1 at other less structured locations (U. Maran and A.T. García-Sosa; personal communication).

Previous experiments showed that NSC156529 inhibits the interaction between AKT1 and PDPK1 proteins and reduces phosphorylation of AKT1 protein at Thr308. To find out whether the treatment with NSC156529 decreases also the phosphorylation of AKT1 downstream targets, we studied the phosphorylation status of GSK3β, FOXO3a, BAD, and procaspase 9 by Western blot (30–36). PC-3 cells were incubated with NSC15652 and with DMSO or PDPK1 inhibitor as background and positive controls correspondingly for 4 hours. The incubation with NSC156529 resulted in a marked decrease in the phosphorylation of the studied AKT1 target proteins (Fig. 3B). The intracellular amount of these proteins did not change with the exception of FOXO3a where both NSC156529 and, to a lesser extent, the PDPK1 inhibitor reduced the overall FOXO3a protein level. Taken together, these results indicate that the treatment with NSC156529 results in the inhibition of the key biochemical activities of the AKT signaling pathway.

Our next goal was to determine the effect of NSC156529 on the growth of malignant and normal cells. PC-3 prostate tumor cells, normal primary human fibroblasts, and osteoblasts were treated with increasing concentrations of NSC156529 for 96 hours. The treatment with DMSO was used as reference. We found that NSC156529 inhibits dose-dependently the growth of PC-3 cells, fibroblasts, and osteoblasts (Fig. 3C). Notably, the growth of PC-3 cells was inhibited to a larger extent than that of normal cells. To verify that our observations were not cell line specific, other four immortalized and tumor cell lines, Hek293, H1299, K07074, and Hep3B, were treated with 10 μmol/L NSC156529, DMSO, and PDPK1 inhibitors for 24, 48, 72, and 96 hours (Supplementary Fig. S4A–S4D). The experiments showed that NSC156529 inhibited strongly the growth of these cell lines by 96 hours. Notably, NSC156529 was more efficient in inhibiting the cell growth than PDPK1 inhibitor (Supplementary Fig. S4C and S4D).

To examine the ability of the NSC156529 to suppress tumor growth in vivo, we took advantage of a tumor xenograft model (Fig. 4A). To establish tumors, nude mice were injected subcutaneously with PC-3 prostate cancer cells, which constitutively expressed EGFP. When tumors achieved the median size of 29 to 32 mm³, the mice were injected with NSC156529 subcutaneously 3 times a week at concentrations 1 mg/kg, 5 mg/kg, 10 mg/kg, or with vehicle only. The tumor size was measured externally by using a caliper, and in parallel the number of tumor cells was monitored by using an in vivo imaging device as described in “Materials and Methods.” During the 28-day treatment period, the suppression of tumor growth was observed for all concentrations used and NSC156529 inhibited growth of PC-3–derived tumors (Fig. 4B). Measurements obtained with external caliper were consistent with GFP measurements showing unequivocally that NSC156529 reduces the tumor growth in vivo (Fig. 4C).

During the treatment course, no adverse side effects, such as weight loss, ulcerations, or general non–well-being of the animals, were observed. To assess the toxicity of the compound, the ALT and AST levels, indicative of liver damage, were measured from sera collected from NSC156529- and vehicle-treated mice at the endpoint of the experiment (Supplementary Fig. S5). No considerable increase in the ALT and AST values was detected showing that in our experimental system NSC156529 does not exert hepatotoxicity.

To shed light on the mechanism by which NSC156529 inhibits the tumor growth in nude mice, we first studied the activity of AKT pathway in the NSC156529-treated and control tumor xenografts by evaluating the phosphorylation status of T308 in AKT1 and BAD using immunofluorescence analysis (Fig. 5A and B). We could confirm the inhibition of the AKT pathway as the number of cells positive for the pAKT(T308) and the phosphorylated form of BAD protein were noticeably reduced in NSC156529-treated PC-3-GFP xenografts. To study the mitotic activity present in the control and NSC156529-treated xenografts, we stained the tumor sections with an antibody recognizing phospho-histone H3 (pH3)—a marker for mitotic cells. The number of pH3-positive cell nuclei was decreased in NSC156529-treated xenografts, indicating that at least in part the inhibition of cell proliferation accounts was causing the inhibition of tumor growth (Fig. 5C and D).

A plausible mechanism by which NSC156529 could inhibit the tumor growth was the increase in the number of apoptotic cells in the treated tumors. To study this possibility, we detected the presence of apoptosis markers—fragmented DNA and cleaved caspase-3 (CC3) in tumor xenograft cryosections. To detect the presence of DNA fragments, we used terminal deoxynucleotidyl transferase nick end labeling (TUNEL) assay, and to measure the amount CC3, we utilized a specific antibody recognizing this form of caspase 3. We could detect the presence of a number of apoptotic cells positive for TUNEL and CC3 already in the untreated tumors (Supplementary Fig. S6A and S6B). The presence of cells, which spontaneously undergo apoptosis, is a characteristic feature of malignant neoplasms. The treatment with NSC156529 did not significantly increase the number of cells positive for TUNEL nor CC3, which indicates that the induction of apoptosis did not have a significant role in the growth reduction of mouse tumor xenografts induced by NSC156529.

Another possible mechanism for the tumor growth reduction is the induction of differentiation. Because PC-3 cells display properties of poorly differentiated prostate cancer (37), we hypothesized that downregulation of the AKT pathway induced the differentiation of grafted tumor cells. To test this possibility, we studied the expression of cytokeratins 15, 17, 18, and 8 in tumor xenografts. Cytokeratins 15 and 17 (CK15/17) are the markers for human prostate basal epithelial cells, and cytokeratins 8 and 18 (CK8/18) label the differentiated luminal cells (38). We found that the expression of CK15/CK17 (Fig. 6A, C, and D) and CK8/CK18 (Fig. 6B, E, and F) was increased in NSC156529-treated PC-3-GFP xenografts, indicating that NSC156529 limits the tumor growth at least in part by directing the cancer cells to differentiate.

The purpose of the present study was to discover small molecular compounds that would utilize a principally novel mechanism to inhibit the signal transduction along the PI3K/AKT pathway. We chose to target the interaction between AKT1 and PDPK1—a critical step in the AKT signaling cascade. To reach our goal, we took advantage of the reversible complementation of two Renilla luciferase fragments, which enabled us to detect the interaction of two proteins of interest in live cells (19). The screening and following experiments identified one chemical—NSC156529—that reduced the interaction of overexpressed and endogenous AKT1 and PDPK1 proteins. Concomitantly, this compound reduced AKT1 phosphorylation and inhibited tumor cell growth in vitro and in vivo tumor xenograft model. Although the detailed molecular mechanism of AKT1 inactivation by NSC156529 remains to be elucidated, our results suggest that it interfered with the AKT1–PDPK1 interaction by direct binding to PDPK1.

Prostate cancer is the second leading cause of death among men in Western world. Although this disease can have a relatively benign course as well-differentiated forms of this tumor remain indolent and are never lethal, the undifferentiated prostate cancer is a highly malignant disease, which requires radical intervention (39). Thus, the discovery of novel substances that would direct the tumor cells to differentiate rather than induce cell death is an emerging possibility for cancer treatment. As an example of a differentiation-based tumor therapy is the addition of retinoids in the acute promyelocytic leukemia (APML) treatment scheme, which dramatically improves patient survival (40). Furthermore, conventional chemotherapy is often associated with the development of drug resistance and systemic toxicity, thereby limiting therapeutic effectiveness (41). For this reason, using a combined therapy scheme where one component would be compound, which induces tumor cell differentiation, would facilitate the reduction of drug dosage, limit the occurrence of side effects, and reduce the occurrence of drug resistance (42, 43). It is known that malignant prostate cancer cells and cell lines including PC-3 harbor an increased level of AKT signaling (27). Consequently, our finding that NSC156529-treated xenografts expressed increased levels of differentiation markers suggests that manipulating the activity of AKT pathway, at least at the level of AKT1–PDPK1 interaction, might open up a new option for tumor treatment via inducing cell differentiation.

Hyperactivated AKT signaling pathway has also been found in many human tumor types (6, 44–46). In line with this, our in vitro cell growth analysis confirmed that the cell proliferation–inhibiting effect of NSC156529 extended to the cell lines of variable origin, thus the potential use of this chemical should not be limited with prostate cancer only. Indeed, data mining revealed that NSC156529 inhibited the growth of B16 melanoma cells and leukemia cell lines L1210 and P388 in tumor xenograft assays conducted by NCI within the framework of Developmental Therapeutics Program (47).

Although NSC156529 inhibited effectively the growth of normal human fibroblasts and osteoblasts in vitro, the compound was well-tolerated in vivo when using subcutaneous administration route. Because systemic administration of this compound to the blood stream was not tested, the relatively low general toxicity observed during our studies might be caused by a reduced exposure of the host organism to NSC156259. When the compound was administered intraperitoneally, severe symptoms of irritation were observed, warranting for potential local toxicity on mucous membranes and endothelium. Interestingly, we did not notice any severe symptoms of irritation at the injection sites in skin. Further toxicity studies should resolve this discrepancy.

Taken together, we have demonstrated that the small molecular compound NSC156529 is a potent inhibitor of AKT1 signaling pathway in tumor cells. NSC156529 reduces the AKT activity, inhibits the activation of AKT target proteins that are involved in regulating cell survival, proliferation, and metabolism, and inhibits the growth of tumor cells in vitro and in vivo. Notably, the treatment with NSC156529 increases the differentiation status of cancer cells, which presents this compound as a promising lead for the development of a novel class of tumor therapeutics.

No potential conflicts of interest were disclosed.

Conception and design: V. Jaks

Development of methodology: K. Mäemets-Allas, J. Viil, V. Jaks

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Mäemets-Allas, J. Viil

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Mäemets-Allas, J. Viil, V. Jaks

Writing, review, and/or revision of the manuscript: K. Mäemets-Allas, J. Viil, V. Jaks

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Mäemets-Allas

Study supervision: V. Jaks

The authors thank Illar Pata, Pille Pata, and Marina Skolnaja (Tallinn Technical University) for the help with animal experiments; Piotr Wardega, Emilia Danilowicz-Luebert, and Heide Marie Resch (Nanotemper Inc.) for the kind assistance with the MST experiments; and Uko Maran and Alfonso T. García-Sosa (Institute of Chemistry, University of Tartu) for performing the docking calculations.

V. Jaks, K. Mäemets-Allas, and J. Viil were supported by the Competence Center for Cancer Research grant from Enterprise Estonia, EMBO Installation Grant number 1819, and Personal Grant number 0004 from Estonian Research Agency.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.