Preclinical Pharmacology, Antitumor Activity, and Development of Pharmacodynamic Markers for the Novel, Potent AKT Inhibitor CCT128930

This article is featured in Highlights of This Issue, p. 219

The serine-threonine kinase AKT (protein kinase B) is a key component of the phosphatidylinositol 3-kinase (PI3K)-AKT-mammalian target of rapamycin (mTOR) signaling network, with activation resulting in increased cell survival, proliferation, and growth (1, 2). AKT is activated following the binding of its pleckstrin homology domain with cell membrane phosphatidylinositol (3,4,5)-trisphosphate (PIP₃), which is produced by PI3K (3). This leads to the phosphorylation of AKT residues threonine 308 by phosphoinositide-dependent kinase 1 (PDK1; ref. 4) and serine 473 by the mammalian target of rapamycin complex 2 (mTORC2; ref. 5). Activated AKT results in phosphorylation of important downstream proteins involved with cell proliferation, survival, and growth such as glycogen synthase kinase-3β (GSK3β; ref. 6), the forkhead family of transcription factors (7), proline-rich AKT substrate 40 (PRAS40), and S6 ribosomal protein (S6RP; ref. 8).

Hyperactivation of the PI3K-AKT-mTOR signaling network is involved in malignant transformation and chemoresistance and may occur through upstream stimulation by receptor tyrosine kinases or genetic abnormalities of key pathway components such as the tumor suppressor phosphatase and tensin homologue on chromosome 10 (PTEN), PIK3CA, and AKT (1, 9). The promising clinical activity observed with agents targeting the mTOR kinase has shown that targeting the PI3K-AKT-mTOR pathway is a rational and effective anticancer approach (10). AKT thus represents an attractive target for the development of anticancer agents (2).

We have used fragment-based in silico screening to identify hits against AKT, which were subsequently validated through structural studies and transformed into potent lead compounds, using structure-based design (11, 12). As a result of these endeavors, CCT128930, a potent small molecule inhibitor of AKT with drug-like properties, was discovered from a series of pyrrolopyrimidines (12). Importantly, CCT128930 shows selectivity for AKT over PKA by targeting a single amino acid difference, which we were the first to show (12, 13). Hence, this was an important compound to profile in model systems and to use as a chemical probe (14).

The development of selective, molecularly targeted therapeutics requires the identification of biologically active concentrations through pharmacokinetic profiling to show appropriate drug exposures and pharmacodynamic measurements that confirm target and pathway modulation in vivo. Such a demonstration of a pharmacologic audit trail necessitates the development and validation of appropriate pharmacodynamic biomarkers (15–17). Although pharmacodynamic biomarker responses in tumors are the most relevant indicator of target modulation, issues of accessibility mean that surrogate normal tissues are frequently used. Because the bulbs of hair follicles contain rapidly proliferating cells, and sampling is both accessible and minimally invasive, plucked hairs are potentially suitable tumor surrogates for serial pharmacodynamic analyses. Previous studies have shown that hair follicles may be used to monitor the pharmacodynamic effects of PI3K inhibitors, and this format may potentially be readily integrated into early clinical trials (18, 19).

In this article, we present the preclinical pharmacology of the novel AKT inhibitor CCT128930 and show its effects on appropriate pharmacodynamic biomarkers, both in vitro and in vivo, and promising antitumor efficacy in human tumor xenograft models with PI3K-AKT-mTOR pathway activation. Using CCT128930 as a preclinical tool, we have also developed a robust and sensitive, noninvasive pharmacodynamic biomarker assay using hair follicles which is currently being used in a clinical trial of an AKT inhibitor (20).

All cell lines were purchased from the American Type Culture Collection and grown in their recommended culture medium, supplemented with 10% FBS at 37°C in 5% CO₂ and passaged for less than 6 months prior to replacement from early passage frozen stocks. CCT128930 (12) and LY294002 were made up as 10 mmol/L stocks in dimethyl sulfoxide (DMSO). All other reagents were purchased from Sigma unless otherwise stated. Cells were regularly screened for Mycoplasma, using a PCR-based assay (VenorGem; Minerva Biolabs).

Profiling against 50 different human kinases was carried out using 10 μmol/L CCT128930 at an ATP concentration equivalent to the K_m for each enzyme (Millipore). All other enzyme assays were carried out at Millipore.

Cells were seeded in 96-well plates and allowed to attach for 36 hours to ensure exponential growth prior to treatment. In vitro antiproliferative activity was determined using a 96-hour SRB assay, and GI₅₀ values were derived as described previously (21).

Drug-treated cells were washed with ice-cold PBS and harvested, lysates were prepared, and protein estimations were carried out as described previously (22). Samples were separated on precast 10% Tris-glycine gels (Novex; Invitrogen) and transferred to polyvinylidene difluoride membrane (Millipore). The membranes were blocked for 1 hour in a blocking buffer [5% dried milk in 50 mmol/L Tris (pH 8.0), 150 mmol/L NaCl, and 0.1% Tween 20], before incubating overnight at 4°C, in the following antibodies diluted in blocking buffer: AKT, phospho-AKT (Ser473), phospho-GSK3β (Ser9), phospho-S6RP (Ser240/244), S6RP, FOXO1, and PRAS40, all at 1:1,000, phospho-S6RP (Ser235/236) at 1:10,000, phospho-FOXO1 (Thr24)/FOX3a (Thr32) at 1:500 (all purchased from Cell Signaling Technology), phospho-PRAS40 (Thr246), p21, and p27 at 1:1,000 (Upstate Biotechnology), GSK3β at 1:1,000 (BD Biosciences), cyclin D1 Ab-1 (DCS-6) at 1:1,000 (NeoMarkers; Lab Vision) and glyceraldhyde-3-phosphate dehydrogenase at 1:1 × 10⁶ (Chemicon International, Millipore). The membranes were washed before incubation in goat anti-rabbit (1:5,000 or 1:10,000) or anti-mouse (1:10,000) horseradish peroxidase–conjugated secondary antibody (Bio-Rad) for 1 hour at room temperature. Protein bands were visualized with enhanced chemiluminescence reagents (Pierce, Thermo Fisher) on Hyperfilm (GE Healthcare), using a Compact X4 developer (Xograft).

PTEN-null U87MG human glioblastoma cells were treated with CCT128930 and labeled with bromodeoxyuridine (BrdUrd) and propidium iodide (23), before being analyzed as previously described (22).

All procedures were carried out in accordance with National Home Office regulations under the Animals (Scientific Procedures) Act 1986 and within guidelines set out by the Institute's Animal Ethics Committee and by the UKCCCR and UK NCRI (24). PTEN-null U87MG human glioblastoma cells (2 × 10⁶) were injected subcutaneously (s.c.) in the right flank of 6 to 8 weeks old female CrTacNCr-Fox1nu mice. For HER2-positive, PIK3CA-mutant BT474 human breast cancer xenografts, cells (5 × 10⁶) were administered s.c. in medium supplemented with Matrigel (1:1) into the mammary fat pads of female mice implanted s.c. with estradiol pellets (0.025 mg, 90-day release #NE-121; Innovative Research of America) 3 days previously. Animals were randomized, and treatment was started with vehicle or CCT128930 when established tumors were approximately 100 mm³ in mean volume. Control mice received vehicle only (10% DMSO, 5% Tween 20, and 85% saline), and treated mice received 50 mg/kg CCT128930 intraperitoneally (i.p.) daily for 5 days (U87MG human glioblastoma xenografts) or 40 mg/kg CCT128930 i.p. twice daily for 5 days (BT474 human breast cancer xenografts). Tumor size and body weight were monitored 3 times a week. Tumor size was evaluated by measurement of 2 orthogonal diameters with calipers and volume was calculated from the following formula: V = 4/3π[(d₁ + d₂)/4]³. At the end of the study, tumors were excised and weighed.

To assess the pharmacokinetic and pharmacodynamic profiles of CCT128930, a single dose of compound (50 mg/kg i.p.) was administered to mice bearing U87MG human glioblastoma xenografts. Plasma and tumor samples were harvested at 2 and 6 hours following dosing. Mice were bled by cardiac puncture, and plasma samples were collected and frozen at −20°C until analysis. Tumors were dissected, divided into 2 approximately equal pieces, and snap frozen in liquid nitrogen until analysis. For pharmacodynamic studies, tumors were homogenized using a buffer containing 50 mmol/L Tris (pH 7.4), 1 mmol/L NaCl, 1 mmol/L EDTA, 1% Triton X-100, 1 mmol/L NaF, 1 mmol/L NaVO₄, 5 μmol/L Fenvalerate, 5 μmol/L Vbphen, 10 mg/mL TLCK, 1× Complete inhibitor tablet per 10 mL of buffer (Roche), protease inhibitor cocktail, and phosphatase inhibitors 1 and 2 (Sigma-Aldrich; ref. 25). Protein content was measured using Bradford reagent (26), and samples were analyzed using immunoblotting as described previously.

Concentrations of CCT128930 in biological samples were determined using liquid chromatography/ mass spectrometry (LC-MS). Drug was extracted using methanol and chromatography carried out on a Synergi-Polar RP column (5.0 cm × 4.6 mm ID, 4-μm particle size; Phenomenex) using a Waters 600MS pump and a 717 autosampler (Waters) with a gradient mobile phase of 0.1% formic acid/methanol at 0.6 mL/min over 12 minutes. Detection was by LC-MS using a TSQ700 triple quadrapole in which the analyte was ionized by electrospray interface in positive mode to monitor the transition [M + H]⁺ 342.8 to 146.6.4. The spray voltage was optimized to 5.5 kV and capillary temperature to 260°C. The assay was linear over the range 10 to 10,000 nmol/L CCT128930.

U87MG human glioblastoma cells were plated at 5 × 10⁴ cells per well on cover slips in 24-well plates for 36 hours, before being treated with increasing concentrations of CCT128930 for 24 hours. Cells were fixed in 3.8% formaldehyde and permeabilized with 0.01% Triton X-100. BALB/c mice were treated with 50 mg/kg CCT128930 i.p. daily for 4 days, before whiskers were plucked and immediately fixed by immersing, root first, in 10% formal saline for 30 minutes before storing in 4% saline at 4°C. Hair follicles from healthy volunteers were collected and processed as described previously. The whisker and hair follicles underwent antigen retrieval with a citrate buffer (DAKO) prior to analysis. The cells, whisker, and hair follicles were stained with anti-phospho-Thr246 PRAS40 (Upstate) and anti-PRAS40 (Cell Signaling Technology) antibodies (1:200 for cells; 1:50 for whisker and hair follicles) and then visualized with 1:1,000 AlexaFluor 488 goat anti-rabbit IgG antibody (Invitrogen). Nuclei were counterstained with 1:10,000 TOPRO-3 (Invitrogen). The cells or follicles were mounted using Vectashield, and images were visualized and captured using a Leica SP1 confocal laser scanning fluorescence microscope (Leica Microsystems). All images were taken with the same detector settings based on the respective vehicle-treated control sample. Optical magnification was 250, with an additional 2-fold software (digital) amplification, giving a total magnification of 500. For follicles, images were taken from the middle of the bulb along the z-axis and included the entire cross-section of the bulb. Fluorescence intensity in individual cells was quantified using the INCell Investigator Developer Toolbox v1.6 software (GE Healthcare). TOPRO-3 was used to identify all nuclei in the image. The areas around the boundary of the nuclei were then expanded to identify cytoplasmic regions. The nuclear and cytoplasmic segmentations were linked together so that only cytoplasmic areas of the image within cells with nuclei were quantified. The mean fluorescence intensity per cell was then reported from both the pThr246 PRAS40 and total PRAS40 images, respectively.

Statistical significance was determined using unpaired t tests with GraphPad Prism 4 software.

CCT128930 was elaborated from an initial hit in a fragment screen by using a structured-based design approach (12, 27). It consists of a pyrrolopyrimidine core, a 4-amino-substituted piperidine, and a 4-chlorobenzyl side chain (Fig. 1A,). CCT128930 is a potent ATP-competitive AKT inhibitor which was initially screened at 10 μmol/L against a panel of kinases representative of the human protein kinome. There was minimal cross-reactivity with 47 other kinases, although inhibition of CHK2 and the AGC kinases p70S6K, PKA, and Rock-II (<25% activity remaining) was observed (Supplementary Fig. 1). In view of the potential of ATP-competitive inhibitors to cross-react with the closely related AGC class of kinases, the IC₅₀ value of CCT128930 against selected AGC kinases was determined. CCT128930 was shown to be a potent and selective inhibitor of AKT2 (IC₅₀ = 6 nmol/L) with a 28-fold selectivity over the closely related PKA kinase (IC₅₀ = 168 nmol/L) through the novel targeting of Met282 of AKT (Met173 of PKA-AKT chimera, Fig. 1B; refs. 12, 27) and a 20-fold selectivity over p70S6K (IC₅₀ = 120 nmol/L). These data indicate that CCT128930 is a potent and selective inhibitor of AKT.

The GI₅₀ values of CCT128930 for growth inhibition (mean ± SD) were 6.3 μmol/L ± 2.2 (n = 3) for U87MG human glioblastoma cells, 0.35 μmol/L ± 0.11 (n = 4) for LNCaP human prostate cancer cells, and 1.9 μmol/L ± 0.80 (n = 5) for PC3 human prostate cancer cells, all of which are PTEN-deficient human tumor cell lines (28).

Following 1 hour treatment of U87MG human glioblastoma cells with increasing concentrations of CCT128930, there was an initial induction of AKT phosphorylation at Ser473 up to 20 μmol/L of CCT128930, followed by a decrease in phosphorylation at higher concentrations (Fig. 2A). Total AKT signal stayed generally constant throughout the treatments. Decreased phosphorylation of direct substrates or downstream targets of AKT was observed with increasing concentrations of CCT128930. Direct substrates of AKT (Ser9 GSK3β, pThr246 PRAS40, and pT24 FOXO1/p32 FOXO3a) were inhibited at 5 μmol/L or more and the downstream target, pSer235/236 S6RP, was inhibited at 10 μmol/L or more of CCT128930, with generally constant levels of the respective total proteins and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; Fig. 2A). There were minimal effects of CCT128930 on pSer240/244 S6RP. Expression of the cell-cycle protein cyclin D1 seemed to decrease at high concentrations (≥40 μmol/L) of compound.

The effects of CCT128930 exposure over time on AKT activity were also investigated (Fig. 2B). Treatment of U87MG human glioblastoma cells with 18.9 μmol/L (3 × GI₅₀) CCT128930 caused an increase in phosphorylation of pSer473 AKT after 30 minutes, which was sustained for 48 hours. Total AKT protein signal decreased gradually from 8 hours to 48 hours of treatment. There was a rapid and profound loss in pSer9 GSK3β phosphorylation starting from 30 minutes of CCT128930 treatment, which was sustained for at least 24 hours, before recovery at 48 hours. The corresponding total GSK3β protein signal remained relatively constant over this interval. CCT128930 induced a marked loss in phosphorylation of pThr246 PRAS40 from 30 minutes, which was sustained throughout 48 hours of treatment, whereas total PRAS40 signal remained unchanged. CCT128930 also caused a marked decrease in FOXO3a phosphorylation at Thr32 (pThr32 FOXO3a) after a 1-hour exposure, whereas phosphorylation of Thr24 of FOXO1 (pThr24 FOXO1) and total FOXO1 remained generally constant. Total FOXO1 signal remained constant throughout treatment. There was a decrease in pSer235/236 S6RP signal at the earliest time point (30 minutes), with complete loss after 2 hours of treatment. In contrast, pSer240/244 S6RP signal decreased from 4 hours and was abolished after 8 hours of CCT128930 treatment. Once again, the total S6RP protein signal remained relatively constant throughout treatment. The cell-cycle marker cyclin D1 decreased after 8 hours of treatment, whereas there was induction of the cyclin-dependent kinase (CDK) inhibitor, p27, from 2 hours of CCT128930 treatment. It is important to note that apoptosis, as measured by PARP cleavage, was also shown to increase after CCT128930 treatment (Supplementary Fig. 2).

The effects of the 1-hour exposure to CCT128930 were also assessed in PTEN-null PC3 and LNCaP prostate cancer cell lines, HCT116 colorectal cancer cells, and MCF7 human breast cancer cells (Fig. 2C). As seen with PTEN-null U87MG human glioblastoma cells, these studies showed an induction in pSer473 AKT, with constant levels of total AKT. Decreased phosphorylation of pSer9 GSK3β was observed in the PTEN-null PC3 and LNCaP human prostate cancer cell lines, PTEN-wild type, PIK3CA-mutated human colorectal cancer cells, and MCF7 human breast cancer cells (28), whereas pSer240/244 S6RP decreased in all 4 cell lines, with total protein levels staying generally constant. Taken together, these molecular biomarker studies clearly show that CCT128930 inhibits AKT activity in cancer cells in vitro. There was no apparent correlation between the SRB assay GI₅₀ and the IC₅₀ for the loss of phosphorylation on Ser9 GSK3β across the panel of 5 cell lines (data not shown), possibly due to the complexities of the survival pathways or small sample size.

Inhibition of the PI3K pathway has been reported to cause a G₁ cell-cycle arrest (29–31) and therefore studies were conducted to investigate the effects of CCT128930 concentration and exposure time on the cell cycle in U87MG human glioblastoma cells. To characterize the effects of CCT128930 on the cell-cycle distribution in U87MG cells over time, a value of 3 × GI₅₀ (18.9 μmol/L) was selected, as PI3K-AKT pathway inhibition had been observed at this concentration (Fig. 2A). CCT128930 treatment of U87MG cells resulted in an increase in G₀/G₁-phase cells from 43.6% to 64.8% (Fig. 3A and B ) after 24 hours of treatment. This was associated with a decrease in S-phase cells from 36.6% in the DMSO control to 2.0% in the drug-treated cells. Cells in the corresponding G₂/M phase increased from 17.8% in the DMSO control to 30.5% after 24 hours of treatment. Nondividing cells in S phase (S') remained relatively constant throughout the treatment.

Cell-cycle analysis of a concentration–response study of CCT128930 showed that 1 × GI₅₀ caused an increase in G₀/G₁-phase cells from 46.4% in the DMSO control to 63.2% after 24 hours of treatment (Fig. 3C ). There was a corresponding decrease in S-phase cells from 32.3% to 13.2% and a slight increase in G₂/M-phase cells from 16.7% to 18.9%. The number of cells in the S' phase was relatively constant. Comparable effects were seen at 3 × GI₅₀. This G₁/S-phase arrest occurred at CCT128930 concentrations of 5μmol/L or more consistent with inhibition of AKT activity (Supplementary Fig. 3 and Fig. 2A). Similar results were obtained after 48 hours of CCT128930 exposure (data not shown). These data show that CCT128930 caused a predominant G₁-phase arrest, with a corresponding decrease in S phase in PTEN-null U87MG human glioblastoma cells in a time- and drug concentration–dependent manner.

The pharmacokinetics of CCT128930 was determined to establish whether therapeutically active drug concentrations could be established in vivo. The pharmacokinetics of CCT128930 after a single dose of 25 mg/kg are illustrated in Figure 4A and summarized in Supplementary Table 1. Following i.v. administration, CCT128930 reached a peak concentration of 6.4 μmol/L in plasma and was eliminated with a relatively short half-life, high volume of distribution, and rapid clearance, giving an area under the curve (AUC)_0–∞ of 4.6 μmol/L h. Following i.p. administration, the peak plasma drug concentration was 4-fold lower and the plasma clearance was similar to that observed with i.v. administration. The corresponding AUC_0–∞ was 1.3 μmol/L h, giving an i.p. bioavailability of 29%.

Oral CCT128930 administration gave a comparable pharmacokinetic profile with other routes, but the peak plasma concentration was only 0.43 μmol/L, with a plasma clearance comparable with that observed with i.v. administration, suggesting no first-pass metabolism (Supplementary Table 1). This resulted in a correspondingly low AUC and an oral bioavailability of only 8.5%. More importantly, following i.p. administration, peak tumor CCT128930 concentrations were 6-fold higher than the corresponding plasma value at 8 μmol/L and there was evidence of drug retention, as shown by the 2-fold longer half-life and 6-fold slower apparent clearance. This resulted in much higher tumor drug exposure relative to plasma with an AUC_0–∞ of 25.8 μmol/L h. Tumor:plasma drug concentrations did not reach a steady state but varied from 4:1 at 30 minutes to 163:1 at 6 hours, confirming tissue drug retention. Assuming linear kinetics, these data supported the use of higher doses and repeat administration to achieve potentially therapeutic tumor drug concentrations in vivo.

Figure 4B shows that following CCT128930 administration at 50 mg/kg i.p. per day for 4 days, drug concentrations in U87MG human glioblastoma tumors were consistently much greater than the corresponding plasma concentrations with tumor:plasma ratios of 27:1 and 42:1 at 2 and 6 hours following the last dose, respectively. Moreover, U87MG tumor drug concentrations greatly exceeded the CCT128930 GI₅₀ for at least 6 hours following the final dose and were consistently 5-fold higher than the concentrations required for in vitro biomarker modulation (Fig. 2). Metabolism studies revealed that only 0.23% of an administered dose (25 mg/kg i.p.) was recovered unaltered in urine after 24 hours (data not shown). These results showed that pharmacologically active concentrations of CCT128930 were achieved in tumor tissue at well-tolerated doses.

Having shown concentration-dependent and time-dependent inhibition of numerous AKT biomarkers by CCT128930 in vitro and promising levels of tumor exposure in vivo, the pharmacodynamic effects of the compound were then evaluated in the same mice bearing U87MG human glioblastoma tumors, as used for the pharmacokinetic studies (Fig. 4B). Figure 4C and D summarizes the effects of CCT128930 treatment (50 mg/kg i.p. × 4 days) on several AKT biomarkers in U87MG xenografts harvested 2 and 6 hours following the last dose (see Fig. 4B). CCT128930 caused a significant increase in Ser473 phosphorylation on AKT at both the 2- and 6-hour time points (P < 0.001 and P < 0.01, respectively), consistent with the biomarker results in vitro (compare Fig. 3 with Fig. 2). In addition, at both 2- and 6-hour time points, there were clear and significant decreases in the phosphorylation of Ser9 GSK3β (P < 0.001 and P < 0.01, respectively), Ser235/236 S6RP (P < 0.05 and P < 0.01, respectively), and Thr246 PRAS40 (P < 0.05 and P < 0.05, respectively), with the total forms of the protein remaining relatively constant (Fig. 4C and D). These observations are consistent with inhibition of AKT activity by CCT128930 occurring in U87MG tumors in vivo.

Next, the antitumor activity of CCT128930 was evaluated in 2 molecularly relevant human tumor xenograft models. Figure 5A shows that there was a marked antitumor effect of CCT128930 at 25 mg/kg i.p. (× 5 of 7 days) in established PTEN-null U87MG human glioblastoma xenografts, giving a treated:control (T/C) ratio on day 12 of 48%. There was no weight loss associated with this regime. Treatment of HER2-positive, PIK3CA-mutant BT474 human breast cancer xenografts (28) with CCT128930 (40 mg/kg twice a day dosing × 5 of 7 days) also had a profound antitumor effect with complete growth arrest and a T/C ratio of 29% on day 22. This regimen was associated with minimal weight loss, with a nadir of only 94.8% of the initial body weight on day 15 of treatment. These results clearly show that CCT128930 has antitumor activity as a single agent in 2 human tumor xenograft models that are molecularly relevant with respect to PI3K pathway activation.

Finally, we developed a quantitative pharmacodynamic hair follicle PRAS40 biomarker assay with CCT128930 for use in clinical trials of AKT inhibitors. The effects of CCT128930 treatment on pThr246 and total PRAS40 were compared in PTEN-null U87MG human glioblastoma cells, whiskers obtained from BALB/c mice treated in vivo, and hair follicles from human healthy volunteers. Initial studies showed that there was a significant decrease in pThr246 PRAS40 (P < 0.001) whereas total PRAS40 stayed generally constant relative to the vehicle control following 24 hour of exposure to increasing concentrations of CCT128930 on U87MG cells (Supplementary Fig. 4). Figure 6A and B shows that there was a significant decrease (P < 0.001) in pThr246 PRAS40 in whiskers plucked from mice treated with CCT128930 (50 mg/kg × 4 days i.p.) relative to controls, with minimal changes in total PRAS40 expression. The influence of diurnal variation was not formally investigated, but no obvious fluctuations in the amount of baseline fluorescence levels (in the absence of AKT inhibitor) were observed. Ex vivo treatment of plucked, human healthy volunteer hair follicles incubated for 24 hours with increasing concentrations of CCT128930 showed that there was a significant decrease (P < 0.001) in pThr246 PRAS40 signal at greater than 3 × GI₅₀ concentrations of CCT128930 (Fig. 6C and D) whereas total PRAS40 stayed generally constant. These results indicate that CCT128930 inhibits AKT activity in a surrogate normal tissue taken from both mouse and human and support this potential use of hair follicle measurements in clinical trials of AKT inhibitors.

The PI3K-AKT signal transduction pathway is a critical driver of tumorigenesis and a number of PI3K/AKT pathway inhibitors have now been disclosed, with several recently entering the clinical arena (2, 9). In this report, we describe the preclinical pharmacology of CCT128930, a recently developed AKT inhibitor, which was discovered using fragment-based lead identification and structure-based design technologies and exhibited improved selectivity over PKA through the targeting of a specific amino acid residue, Met282, of AKT (12). We have confirmed that CCT128930 is a potent and selective AKT inhibitor which causes a predominant G_1/G_o cell-cycle arrest, consistent with other reported specific inhibitors of the PI3K/AKT pathway. CCT128930 inhibited the phosphorylation of several AKT substrates including GSK3β, PRAS40, and FOXO1, as well as downstream targets S6RP and cyclin D1 in multiple cell lines in vitro, consistent with AKT inhibition. These effects were recapitulated in vivo. Antitumor activity was also observed in both PTEN-null U87MG human glioblastoma xenografts and HER2-positive, PIK3CA-mutant BT474 human breast cancer xenografts. Furthermore, a noninvasive and semiquantitative immunofluorescent biomarker assay for AKT inhibition was also developed using whisker follicles from mice treated with CCT128930 and human hair follicles treated ex vivo with this compound.

CCT128930 is a potent, ATP-competitive AKT inhibitor, similar in mechanism to A-443654 (32) and GSK690693 (33) but distinct from the allosteric AKT inhibitor MK2206 (34) or perifosine (35), which interacts with the PIP3-binding pleckstrin homologue domain of AKT. In the present study, CCT128930 caused an initial induction of phosphorylation on Ser473 of AKT at doses up to 20 μmol/L in PTEN-null U87MG human glioblastoma cells, with eventual suppression at higher doses. Similar observations have been made with other ATP-competitive AKT inhibitors (33, 36). This has been hypothesized to be either directly due to inhibitor binding to the ATP binding site of AKT or a result of compensatory but futile feedback loops of AKT pathway regulation (37, 38). Despite this initial induction of AKT phosphorylation, CCT128930 treatment clearly results in a dose- and time-dependent decrease of phosphorylation of both direct and indirect targets of AKT in U87MG human glioblastoma cells (Fig. 2A and B).

We have also shown similar dose-dependent effects on AKT and downstream proteins in PTEN-null PC3 and LNCaP human prostate cancer cells, as well as cells harboring PIK3CA mutations such as HCT116 human colorectal cancer cells and MCF7 human breast cancer cells. Exposures as low as 5 to 10 μmol/L for 1 hour were sufficient to markedly inhibit phosphorylation of GSK3β, S6RP, FOXO1, and PRAS40. Several of these proteins are involved in cell cycle and proliferation (GSK3β, cyclin D1, and p27), cell growth (PRAS40 and S6RP), and apoptosis (FOXO1), and this is consistent with the potent antiproliferative activity of CCT128930 in vitro. The loss of cyclin D1 and induction of the CDK inhibitor p27 are consistent with the cell-cycle arrest at G₁ and inhibition of the PI3K-AKT-mTOR signaling pathway (39–41). In addition, we have observed induction of apoptosis, a known phenotype of AKT inhibition.

One of the key aims of this study was to identify and monitor appropriate biomarkers of AKT inhibition in order to establish pharmacokinetic–pharmacodynamic relationships in vitro and in vivo, as well as to relate these findings to antitumor effects. Pharmacokinetic studies established that following i.p. administration, CCT128930 gave considerably higher and potentially therapeutic drug concentrations in tumor than in plasma.

Pharmacodynamic biomarker and antitumor studies in PTEN-null U87MG human glioblastoma xenografts confirmed that AKT was inhibited by CCT128930 in vivo and that this was associated with a marked antitumor response at doses that were well tolerated. These pharmacodynamic studies were conducted both to enable target and pathway modulation to be monitored throughout the preclinical evaluation process and to enable the construction of a pharmacologic audit trail to ensure that the target inhibition correlated with antitumor activity (17, 42–44). Importantly, we have shown that the single-agent CCT128930 suppressed downstream pharmacodynamic biomarkers in tumor tissues at doses that also inhibited the tumor growth of both PTEN-deficient U87MG human glioblastoma and HER2-positive, PIK3CA-mutant BT474 human breast cancer xenografts. These results also support the notion that the molecular characterization of tumors may be important for single-agent AKT inhibitor activity.

To extrapolate this drug development approach to the clinic, we have developed a novel pharmacodynamic biomarker assay for use in plucked human hair follicles to assess drug effects in clinical trials. In contrast to previous studies, this assay is rapid, sensitive, and quantitative. Importantly, it uses a more practical strategy, using immunofluorescence and confocal microscopy in contrast to procedures involving paraffin embedding and immunohistochemistry used elsewhere (18, 19, 45). This approach may thus be easily incorporated into a clinical trial to serially monitor pharmacodynamic effects in order to potentially guide key decision making in drug development (17, 19, 42–44).

PRAS40 is a binding partner of mTOR which mediates AKT signals to mTOR and is implicated in oncogenesis, insulin resistance, and neuronal cell death (46). PRAS40 was selected as a pharmacodynamic biomarker for CCT128930 inhibition of AKT because it is a direct substrate of AKT and hence its phosphorylation status would accurately reflect AKT modulation. Moreover, phosphorylation of PRAS40 was also acutely sensitive to inhibition by CCT128930 (Fig. 2B). We first developed this strategy in vitro through the treatment of PTEN-null U87MG human glioblastoma cells with CCT128930, followed by the collection of whisker follicles from BALB/c mice treated with CCT128930 in vivo, and finally, ex vivo treatment of healthy volunteer hair follicles. Several limitations exist. Because an intact hair follicle is required, it is not possible to stain for both phosphorylated and total forms of the biomarker on the same follicle simultaneously. Additional assay validation will be useful to define such parameters as lower limit of detection and to evaluate whether a statistically significant decrease from control or a specific percent decrease from control is an appropriate definition of effect.

This assay is also suitable for use with PI3K or allosteric AKT inhibitors, as the pharmacodynamic readout used is the downstream substrate PRAS40. In the future, this technique may potentially be developed further to include the analyses of multiple biomarkers per hair follicle with the use of antibodies directly labeled with distinct fluorophores. This hair follicle assay is currently being used as a pharmacodynamic readout in the Phase I clinical trial of the allosteric AKT inhibitor MK2206 (20).

In conclusion, we have evaluated the pharmacologic and therapeutic properties of the interesting and novel AKT inhibitor CCT128930. This is a leading example of the pyrrolopyrimidines series with which we obtained selectivity for AKT over PKA by targeting a single amino acid difference and which we were the first to show (12, 13). We have shown that CCT128930 suppresses the phosphorylation of downstream markers of AKT in vitro and in vivo and exhibits promising single-agent antitumor activity in molecularly relevant human cancer models with appropriate pharmacokinetic and pharmacodynamic properties. We have also used this AKT inhibitor to develop a novel assay to quantify pharmacodynamic biomarker changes in a normal tissue following PI3K-AKT pathway blockade, and this assay may have broader applications in clinical trials of other related pathway inhibitors.

T.A. Yap, M.I. Walton, L-J.K. Hunter, M. Valenti, A. de Haven Brandon, P. Eve, R. Ruddle, S.P. Heaton, A.Henley, L. Pickard, G. Vijayaraghavan, J.J. Caldwell, W. Aherne, F.I. Raynaud, S.A. Eccles, P. Workman, I. Collins, and M. D. Garrett are current or former employees of The Institute of Cancer Research, which has a commercial interest in the development of AKT inhibitors, including CCT128930, and operates a reward for inventors scheme. N.T. Thompson is an employee of Astex Therapeutics, which also has a commercial interest in the development of AKT inhibitors including CCT128930. Both Astex Therapeutics and The Institute of Cancer Research have been involved in a commercial collaboration with Cancer Research Technology Limited (CRT) to discover and develop inhibitors of AKT and intellectual property arising from this program has been licensed to AstraZeneca.

Grant support was provided to T.A. Yap, M.I. Walton, L-J.K. Hunter, M. Valenti, A. de Haven Brandon, P.D. Eve, R. Ruddle, A. Henley, L. Pickard, G. Vijayaraghavan, J.J. Caldwell, F.I. Raynaud, S.A. Eccles, and M.D. Garrett by Cancer Research UK (CR-UK) grant number C309/A8274, to S.P. Heaton by CR-UK grant number C51/A6883, to P. Workman by CR-UK grant number C309/A8992, and to W. Aherne and I. Collins by CR-UK grant number C309/8365. Additional support was provided to S.A. Eccles and M.D. Garrett by The Institute of Cancer Research and to L-J.K. Hunter by Astex Therapeutics. This work was carried out as part of a funded research collaboration with Astex Therapeutics. We acknowledge NHS funding to the NIHR Biomedical Research Centre.

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.