AKT mutant allele-specific activation dictates pharmacologic sensitivities

Abstract

AKT- a key molecular regulator of PI-3K signaling pathway, is somatically mutated in diverse solid cancer types, and aberrant AKT activation promotes altered cancer cell growth, survival, and metabolism1-8. The most common of AKT mutations (AKT1 E17K) sensitizes affected solid tumors to AKT inhibitor therapy7,8. However, the pathway dependence and inhibitor sensitivity of the long tail of potentially activating mutations in AKT is poorly understood, limiting our ability to act clinically in prospectively characterized cancer patients. Here we show, through population-scale driver mutation discovery combined with functional, biological, and therapeutic studies that some but not all missense mutations activate downstream AKT effector pathways in a growth factor-independent manner and sensitize tumor cells to diverse AKT inhibitors. A distinct class of small in-frame duplications paralogous across AKT isoforms induce structural changes different than those of activating missense mutations, leading to a greater degree of membrane affinity, AKT activation, and cell proliferation as well as pathway dependence and hyper-sensitivity to ATP-competitive, but not allosteric AKT inhibitors. Assessing these mutations clinically, we conducted a phase II clinical trial testing the AKT inhibitor capivasertib (AZD5363) in patients with solid tumors harboring AKT alterations (NCT03310541). Twelve patients were enrolled, out of which six harbored AKT1-3 non-E17K mutations. The median progression free survival (PFS) of capivasertib therapy was 84 days (95% CI 50-not reached) with an objective response rate of 25% (n = 3 of 12) and clinical benefit rate of 42% (n = 5 of 12). Collectively, our data indicate that the degree and mechanism of activation of oncogenic AKT mutants vary, thereby dictating allele-specific pharmacological sensitivities to AKT inhibition.

Conflict of interest statement

The authors declare the following competing interests: L.M.S. reports receiving research funding from AstraZeneca, Puma Biotechnology and Roche Genentech; honoraria from AstraZeneca and Pfizer; travel, accommodations, expenses from Puma Biotechnology, Roche Genentech and Pfizer; and consulting or advisory board activities for Roche Genentech and AstraZeneca. D.B.S. reports advisory board activities for Loxo Oncology, Pfizer, Illumina, Lilly Oncology, Vivideon, and Intezyne. D.M.H. reports receiving research funding from AstraZeneca, Puma Biotechnology, Loxo Oncology and personal fees from Atara Biotherapeutics, Chugai Pharma, Boehringer Ingelheim, AstraZeneca, Pfizer, Bayer, Debiophram Group, and Genentech. B.S.T. reports advisory board activities for Boehringer Ingelheim, Loxo Oncology (a wholly owned subsidiary of Eli Lilly), and honoraria and research funding from Genentech. All stated activities were outside of the work described herein. No other disclosures were noted.

© 2022. The Author(s).

Figures

Fig. 1. Somatic mutations in AKT in human cancers.

a The frequency of known and candidate driver mutations in AKT1, AKT2, and AKT3 in diverse primary and metastatic human cancers as determined from population-scale sequencing of 41,075 patients (frequencies based on only known or candidate driver mutations only, excludes presumptive passenger mutations). b The number of affected samples for each of the individual mutations in AKT1, AKT2, and AKT3 indicates a long tail of increasingly uncommon candidate driver mutations. c The fraction of cases harboring candidate driver mutations in AKT1-3 that are E17K or not. d Schematic representation of the domain structure of AKT1 and the position of different classes of mutations (see inset legend) identified as candidate driver alterations including those in paralogous residues in AKT2 and AKT3 (orange). Arcing lines reflect physically proximity in the cognate folded protein (panel d). Small in-frame indels are shown as horizontal lines and target a paralogous cluster in AKT1 and AKT2. e A cluster of physically adjacent mutations (blue) in the AKT1 protein structure within 5 angstroms of each other and two single-codon hotspots (E17 and D323). The PH and Kinase domains are labeled as in panel (d).

Fig. 2. Diverse AKT alleles hyper-activate PI3K signaling to differing degrees.

a MCF10a cells stably expressing the indicated AKT1 substitutions were incubated in assay medium overnight. Expression and phosphorylation levels were assayed by Western blot, indicating that all mutations identified as hotspots induced activation of AKT and downstream targets, but few other paralogous residues or those identified by protein structure analysis were similarly activating. b As in panel (a) but for a series of small in-frame duplications in AKT1, showing that while E17K and Q79K are activating, AKT1 duplications induce far higher levels of phosphorylated AKT and robust pathway activation. c As in panels (a, b) but showing the effect of AKT2 activating mutations including single-codon AKT2 hotspots (E17K and D324G) and AKT2 indels. d WT, E17K, and various additional indicated missense and indel mutants in AKT1 and AKT2 were stably expressed in murine pro-B Ba/F3 cells and assessed for their ability to induce phosphorylation of AKT and downstream targets in the absence of interleukin-3 (IL-3) by western blot analysis of whole cell lysates. Results shown are representative images from experiments performed at least three times with multiple batches of stable cells. e Mutants as in panel d, assessed for their ability to promote IL3-independent Ba/F3 cell proliferation. Error bars are standard deviations from the mean. Results were derived from at least two independent experiments with triplicates for each experiment. Source data are provided as a Source Data file.

Fig. 3. Structural and signaling impact of AKT in-frame indels.

a Molecular dynamic simulation of AKT1 indicated systematic displacement of the structure by the P68-C77dup mutation as compared to the E17K hotspot, as measured by changes in inter-residue distance between the PH and kinase domains. Duplication mutant-specific displacement targeted the key regulatory phosphorylation site T308, among others. b The structure of AKT1 indicating the region affected by the cluster of paralogous in-frame indels in AKT1 and AKT2 (light blue) in the PH domain (purple). The kinase domain (KD) is shown in light gray and boxed are the two regions of greater detail shown in the indicated panels. c A detailed view of the 13.6 angstrom T308 displacement by the P68-C77dup mutation. d A detailed view of N53 in the inositol-binding region of the PH domain that is displaced by the duplication mutant breaking its WT hydrogen bonding with Q79. e MCF10a cells stably expressing either GFP, WT AKT1, or the indicated missense or duplication mutants and their myristoylated counterparts were incubated in assay medium overnight, and whole-cell lysates were prepared and immunoblotted with the indicated antibodies in order to assess pathway activation. Source data are provided as a Source Data file. f MCF10a cells stably expressing WT or mutant AKT1 and their myristoylated counterparts were incubated with assay media for four hours, after which the cells were fixed, permeabilized, probed with antibodies for p-AKT (T308) and V5 tag (for AKT expression) and examined under fluorescence microscope in order to determine the intracellular localization of the indicated AKT1 constructs. For panels (e–f), results are representative of at least three independent experiments performed. Scale bar: 40 μm.

Fig. 4. Pre-clinical sensitivity of diverse AKT mutations to AKT inhibition.

a The viability of MCF10a cells stably expressing either WT or the indicated mutations in AKT1 was assessed after treatment with 3 μM of the ATP-competitive AKT inhibitor capivasertib. Legend indicates which of the alleles are activating. b As in panel A, cells expressing the indicated mutants were assessed for viability after treatment with either capivasertib or the allosteric inhibitor ARQ092. Points represent mean viability percentage and error bars indicate the 95% confidence intervals. For panels (a-b), mean viability was determined from at least two independent experiments. c MCF10a cells stably expressing wild-type AKT1, the known oncogenic E17K, or indel mutants were treated with indicated concentrations of capivasertib or ARQ092 (top and bottom, respectively), and cell viability was assessed 72 hours post-treatment. Results derived from 18 values from six independent experiments. d Whole-cell lysates from MCF10a cells expressing AKT1 WT, E17K, and P68-C77dup were harvested four hours post-treatment with either capivasertib (top) or ARQ092 (bottom), and pathway inhibition was assessed by western blot. Results are representative images from at least two independent experiments. Error bars in panels (a) and (c) represent standard deviations from the mean. Source data are provided as a Source Data file.

Fig. 5. Clinical sensitivity of diverse AKT mutations in a phase II basket trial of capivasertib.

a The efficacy of ATP-competitive AKT inhibition was explored in a phase II clinical trial of capivasertib in patients harboring AKT1-3 alterations (excluding AKT1 E17K-mutated ER+ breast cancer). The clinical response of patients to capivasertib therapy with tumors harboring non-E17K AKT1-3 mutations is shown. b Computed tomography (CT) indicating a partial response (indicated by arrows) in an endometrial cancer patient with an AKT1 L52R mutation that lasted nearly a year. c Response of a castration-resistant metastatic prostate cancer patient to capivasertib treatment whose tumor harbored a novel AKT2 L78_Q79ins(HANTFVIRCL) indel mutation. Left and right are CT and bone scans at baseline and two months after treatment initiation. The left panel shows the resolution of the soft tissue metastatic compartment (yellow arrow). d Sequencing of pre-treatment and post-progression tumor tissue indicated the acquisition of a focal amplification of IRS2 (blue; x-axis is chromosome 13) present only in the tumor after AKT inhibitor resistance.