Activation of the PI3K pathway occurs commonly in a wide variety of cancers. Experience with other successful targeted agents suggests that clinical resistance is likely to arise and may reduce the durability of clinical benefit. Here, we sought to understand mechanisms underlying resistance to PI3K inhibition in PTEN-deficient cancers. We generated cell lines resistant to the pan-PI3K inhibitor GDC-0941 from parental PTEN-null breast cancer cell lines and identified a novel PIK3CB D1067Y mutation in both cell lines that was recurrent in cancer patients. Stable expression of mutant PIK3CB variants conferred resistance to PI3K inhibition that could be overcome by downstream AKT or mTORC1/2 inhibitors. Furthermore, we show that the p110β D1067Y mutant was highly activated and induced PIP3 levels at the cell membrane, subsequently promoting the localization and activation of AKT and PDK1 at the membrane and driving PI3K signaling to a level that could withstand treatment with proximal inhibitors. Finally, we demonstrate that the PIK3CB D1067Y mutant behaved as an oncogene and transformed normal cells, an activity that was enhanced by PTEN depletion. Collectively, these novel preclinical and clinical findings implicate the acquisition of activating PIK3CB D1067 mutations as an important event underlying the resistance of cancer cells to selective PI3K inhibitors. Cancer Res; 76(5); 1193–203. ©2016 AACR.

The PI3K/AKT/mTOR pathway is pathologically activated in many tumor types. In particular, the phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (PIK3CA) is frequently mutated and activates the pathway (1). The regulatory subunits of PI3K, PIK3R1, and PIK3R2 (2), and the downstream effector AKT1 (3) can also be mutated and hyperactivate the pathway. Activity of the tumor suppressor PTEN is frequently lost in diverse cancer types by mutation or gene deletion, and PTEN-deficient cancers show dramatically elevated PI3K signaling (4). Inhibitors designed to target p110α, mTOR, or AKT, as well as dual PI3K/mTOR inhibitors, are in various stages of clinical development (5), and mTOR is a clinically validated target in renal cell carcinoma (6) and hormone receptor–positive breast cancer (7). However, experience with other successful targeted agents suggests that clinical resistance is likely to arise and may reduce the durability of clinical benefit (8). As with other targeted agents, understanding the diversity of mechanisms that give rise to PI3K inhibitor resistance through preclinical modeling is likely to help guide clinical hypothesis testing. Ultimately, understanding clinical mechanisms of resistance can provide a rationale for therapeutic combinations, sequencing, or alternative therapies that can overcome resistance mechanisms. For example, preclinical modeling of acquired resistance to B-Raf proto-oncogene, serine/threonine kinase (BRAF), and MEK inhibitors has provided the rationale for clinical testing of combination therapy involving BRAF and MEK inhibitors (9).

Previous preclinical studies of acquired resistance to PI3K inhibitors have mostly been conducted in the background of PIK3CA-activating mutations, particularly the hotspot H1047R or E545K alleles. Those studies identified several mechanisms of resistance involving both PI3K pathway reactivation as well as activation of parallel pathways. In the case of downstream PI3K activation, studies in a mouse mammary tumor model engineered to express an activated PIK3CA allele (H1047R) demonstrated that activation of the Myc oncogene rendered these tumors resistant to selective PI3K inhibitors (10, 11). Gene amplification of the downstream effector, eukaryotic translation initiation factor 4E (eIF4E), and overexpression of the ribosomal protein S6 kinase, 90kDa, polypeptides 3 and 4 (RSK3 and RSK4) have also been shown to confer resistance to PI3K inhibitors (11, 12). Mechanisms acting upstream of PI3K have also been described, including dysregulation of the receptor tyrosine kinases AXL and c-MET, and overexpression of the EGFR ligand amphiregulin (AREG) in the background of PTEN loss (10, 13, 14). In terms of mechanisms acting at the level of the target itself, previous work from our laboratory showed that PIK3CA H1047R gene amplification causes resistance to pan-PI3K inhibition by hyperactivating PI3K signaling (15). In short-term experiments, p110β kinase can be adaptively activated and can restore PI3K signaling when p110α is selectively inhibited in cancer cells harboring a PIK3CA mutation (16). In addition, convergent loss of PTEN has been shown to arise in multiple metastatic lesions in a breast cancer patient harboring a PIK3CA mutation treated with a p110α-selective inhibitor (17).

To date, relatively little is known about PI3K inhibitor resistance in PTEN-deficient cancers. In this type of cancer, p110β and mTOR complex 2 (mTORC2) are critical for cancer progression (18, 19). One study has shown that p110β-selective inhibition in PTEN-mutated tumors relieves feedback suppression of several receptor tyrosine kinases and activates p110α, conferring resistance to p110β-selective inhibition (20). However, the diversity of resistance mechanisms in PTEN-deficient cancer is relatively unclear. Here, we sought to understand mechanisms of acquired resistance to PI3K inhibition in PTEN-mutant EVSA-T and ZR75-1 cells by selecting resistant derivatives that were able to grow in the presence of high concentrations (>1 μmol/L) of the pan-PI3K inhibitor, GDC-0941. Whole-exome sequencing revealed a PIK3CB mutation at the D1067 position in EVSA-T cells, and digital PCR identified the same mutation in ZR75-1 cells. Analysis of TCGA datasets revealed recurrent mutations at this amino acid in PIK3CB in diverse cancers. Functional studies showed that knockdown of PIK3CB in resistant clones suppressed cell proliferation, and overexpression of mutant PIK3CB conferred cross-resistance to PI3K inhibitors but not downstream inhibition at the level of AKT or below. Detailed characterization of D1067Y/A/V-mutant proteins suggested that they are all functionally activating and can increase levels of membrane bound PIP3, leading to higher levels of phospho-AKT and overall PI3K signaling, and are transforming in nontumorigenic Rat-2 cells. These results suggest an oncogenic role for PIK3CB-activating mutations, and implicate p110β activation as a novel resistance factor for selective PI3K inhibitors.

Reagents, cell lines, and tumor samples

Inhibitors for GDC-0941, GDC-0032, GDC-0980, GDC-0068, GDC-0349, PD0325901, crizotinib, lapatinib, erlotinib, docetaxel, paclitaxel, and SAHA were produced and supplied by the Genentech Medicinal Chemistry group (21–27). Other compounds are described in Supplementary Table S1. The HEK293, Rat-2, A-498, ZR-75-1, and HCC-1569 cell lines were from the ATCC; the EVSA-T cell line from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH; the SUM-52PE cell line from Asterand. The identity of the isolate used here was verified by genotyping with a Multiplex STR assay (Genetica). EVSA-T parental, resistant clones, and A-498 cells were cultured in RPMI1640 medium, HEK293 cells in Eagle's Minimum Essential Medium, and Rat-2 cells in DMEM. All medium contained 10% FBS and 2 mmol/L l-glutamine. PI3K inhibitor–resistant cells were generated as previously described (15). All lines were obtained between 2000 and 2010. DNA was isolated from archival FFPE tumor tissue from 72 advanced or metastatic renal cell carcinoma patients as previously described (28). Patient tumor samples were from patients enrolled in study PIM4973g (NCT01442090), which was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki (29). Written informed consent was obtained from all patients in agreement with approved protocols from ethics committees at each study site.

Cell viability assay, apoptosis assay, and small RNAi assay

Compound screening was performed and cell viability was assessed using CellTiter Glo ATP Luminescence assay (Promega) as described previously (22) and apoptosis induction was measured using Caspase-Glo 3/7 Assay (Promega). PIK3CA, PIK3CB, and PTEN individual siRNAs and nontargeting control siRNAs were from Dharmacon (Thermo Scientific). Reverse transfections were conducted with Lipofectamine RNAiMAX (Life Technologies). Four days after transfection, cell viability was assessed or cell lysates were collected.

Western blot analysis

Cells were treated with inhibitors or a solvent control (0.1% DMSO) for 4 hours and lysed in T-PER cell extraction reagent (Thermo Scientific) supplemented with protease and phosphatase inhibitor cocktails (Roche). Protein concentrations were determined by BCA Protein Assay (Thermo Scientific). Cell lysates were resolved by electrophoresis and transferred to PVDF or nitrocellulose membranes. Western blot analysis was performed as described previously (30). Antibodies to phospho-Ser473-AKT, phospho-Thr308-AKT, AKT, phospho-Ser240/Ser244-S6, phospho-Ser235/Ser236-S6, S6, p110alpha, p110beta, PDK1, GAPDH, Myc-tag, E-cadherin, PTEN, and horseradish peroxidase–conjugated secondary antibodies against rabbit IgG were from Cell Signaling Technology.

PCR, Sanger sequencing, cloning, and exome sequencing

PCR (42 cycles of 10 seconds at 94°C, 15 seconds at 55°C, and 1 minute at 68°C) was performed with Platinum PCR supermix (Life Technologies) and the primers 5′-GACTCTCTTGCATTAGGG-3′ and 5′-TGCAAAGTCAGCAGGAAATG-3′. PCR products were sequenced with the following primers: 5′-GGGAAGAGTGAAGAAGAAG-3′ and 5′-CATCGGGGATTGTTCAGATT-3′. The pReceiver-M43 plasmid encoding wild-type (WT) PIK3CB (NP_006210.1) was from GeneCopoeia. The D1067Y, D1067A, D1067V, and K805R mutants were generated by the GeneArt Site-Directed Mutagenesis PLUS System (Life Technologies). For generation of Rat-2–stable cell lines, the sequence corresponding to PIK3CB WT or D1067Y was inserted into pEF1/Myc-His vector by GenScript. Exome sequencing of parental and resistant cell lines was done as described previously, and the prevalence of PIK3CB mutations was determined using exome sequencing data from a panel of cancers downloaded from The Cancer Genome Atlas (TCGA) and processed as described previously (31, 32). Sequencing data were filtered for nonsynonymous coding variants not present in dbSNP and present at >20% allele frequency in the resistant pool only.

Gene copy number analysis

Relative gene copy number for a panel of cancer-related genes was determined using quantitative PCR. Threshold cycles (Ct) were measured using cell line genomic DNAs and gene-specific TaqMan Copy Number Assays (Life Technologies) according to the manufacturer's instructions. Two assays for each gene were used. Relative copy numbers were calculated by ΔΔCt method as previously described (28) using RNaseP (Life Technologies) as a reference and pooled human blood genomic DNA (Roche) as a calibrator sample.

Transient transfection and generation of stable transfectants

Cells were seeded at 1–3 × 105 cells per well to a 12-well plate and were transfected with expression constructs on the next day, using Lipofectamine3000 reagent (Life Technologies). At 2 days after transfection, cells were lysed for Western blot analysis. Stable transfectants were generated separately using geneticin or puromycin.

Droplet digital PCR

Droplet digital PCR (ddPCR) probe assays were designed to detect the following single nucleotide variations in the PIK3CB gene: D1067A, D1067V, and D1067Y Assay sequences were: forward, TTAGGAGCGAAGGCTG; reverse, GTGAACTGGATGGCCC; WT probe, ACAGTTCGGAAAGACTACA; D1067A probe, ACAGTTCGGAAAGCCTACAGA; D1067V Probe, ACAGTTCGGAAAGTCTACAGA; D1067Y Probe, ACAGTTCGGAAATACTACA. All assays were performed on the QX200 Droplet Digital PCR System (Bio-Rad) according to the manufacturer's protocol. Samples were assessed for the presence of PIK3CB mutations using 20 ng input DNA to achieve a sensitivity of approximately 0.01%, and, where evaluable, mutant allele frequencies were calculated using QuantaSoft software (Bio-Rad).

PIP3/PI(4,5)P2 quantification and membrane/cytosol fractionation

Cells were seeded at 3 × 106 cells to 10-cm dish and cultured until they reached approximately 80% confluence. For PIP3/PI(4,5)P2 quantification, phospholipids were isolated and PIP3 and PI(4,5)P2 levels were measured using ELISA kits (Echelon). For membrane/cytosol fractionation, cells were lysed and membrane or cytosol proteins were fractionated using a Cell Fractionation Kit (Cell Signaling Technology).

ATP competition assay

The ability of GDC-0941 to compete with ATP for binding to kinases was measured using the Pierce Kinase Enrichment Kit (Thermo Scientific). Cell lysates were incubated with GDC-0941 at the desired concentration and desthiobiotin-ATP probe. Then, biotinylated proteins were immunoprecipitated with avidin agarose beads. Immunoprecipitated proteins were immunoblotted using standard procedures.

Focus formation, colony formation, and spheroid formation assays

Stably transfected Rat-2 cells were seeded to 6-well plates, and after 20 days incubation, focus formation was observed. Colony formation activity of stably transfected Rat-2 cells was measured with a CytoSelect 96-Well Cell Transformation assay kit (Cell Biolabs, Inc.) after 15 days of culture. For the spheroid formation assay, stably transfected Rat-2 cells were seeded to Nunclon Sphera Microplates (Thermo Scientific) and incubated for 7 or 21 days. Cell morphologies and cell viabilities were measured as previously described (33).

PTEN-mutated EVSA-T breast cancer cells show broad cross-resistance to PI3K inhibitors but maintain p110β signal dependency

To understand the resistance mechanisms underlying PI3K inhibition in PTEN-deficient cancer, we generated a model resistance to a pan-PI3K inhibitor, GDC-0941, using the PTEN-mutated luminal breast cancer cell line, EVSA-T. This cell line has previously been shown to be sensitive to GDC-0941 (22). PI3K-resistant EVSA-T cells were selected by culturing in gradually increasing concentrations of GDC-0941 until a pool of cells was able to stably grow at a concentration of 1 μmol/L drug, at which point five single cell clones were isolated. Resistance of the selected EVSA-T clones to GDC-0941 was evaluated in an ATP-based cell viability assay as well as a caspase-based apoptosis induction assay (Fig. 1A). The IC50 for cell viability in each of the clones was 10- to 15-fold higher than that of parental cells, and apoptosis seen in the parental line was not apparent in any of the resistant clones. Consistent with these observations, phosphorylation of AKT and S6 was maintained in the presence of 1 μmol/L GDC-0941 for 4 hours in resistant clones but not parental cells (Supplementary Fig. S1). To further characterize pathway effects of PI3K inhibition in resistant cells, parental and resistant cells were treated for 4 hours with a serial dilution of GDC-0941 and the level of phospho-AKT was analyzed (Fig. 1B). GDC-0941 was less effective at suppressing these phospho-epitopes in resistant clones, resulting in at least a 10- to 30-fold increase in IC50 in both clones analyzed, suggesting that EVSA-T PI3K-resistant clones show resistance to the effects of GDC-0941 on signal transduction. To identify the pathway dependencies in EVSA-T–resistant cells, we determined sensitivity of parental and resistant lines to an array of 20 compounds that inhibit diverse signaling pathways. First, to assess whether EVSA-T–resistant clones maintain any dependency on the PI3K pathway, we evaluated the effects of the AKT inhibitor, GDC-0068 and the TORC1/2 inhibitor, GDC-0349 PI3K signal modulation activity and on cell growth inhibitory activity. Parental and resistant clones also showed similar modulation of phospho-AKT and phospho-S6 in response to GDC-0068 or GDC-0349 (Fig. 1B). The cell growth–inhibitory activity of GDC-0068 or GDC-0349 was also similar in parental cells and resistant clones (Fig. 1C and Supplementary Fig. S2). Resistant clones were less sensitive than parental cells to the dual PI3K/mTOR inhibitor GDC-0980 in terms of cell growth–inhibitory effects and phospho-AKT and phospho-S6 modulation, perhaps due to a difference in potency between GDC-0980 and GDC-0349 on mTOR inhibition (Supplementary Fig. S3). EVSA-T–resistant clones displayed cross-resistance to a wide array of PI3K inhibitors, including another pan-PI3K inhibitor, CH5132799 (34), the PI3Kβ sparing inhibitor, GDC-0032, the PI3K-mTOR inhibitor, GDC-0980, and PI3Kβ-selective inhibitors, TGX-221 (35), AZD6482 (36), and GSK2636771 (37) in cell viability assays (Supplementary Fig. S4). Consistently, phospho-AKT and phospho-S6 were not suppressed in EVSA-T–resistant clone cells by GDC-0032 or TGX-221 (Supplementary Fig. S5). In general, the parental cell and resistant clones showed similar response to other targeted therapies and chemotherapeutic agents (Fig. 1C and Supplementary Figs. S5 and S6). To determine whether resistant clones still depend on PI3K itself or have developed a bypass resistance mechanism, we knocked down PIK3CA and PIK3CB with siRNA treatment. Although PIK3CB siRNA treatment suppressed cell proliferation and phospho-AKT (Fig. 1D) in both EVSA-T parental cells and clones, PIK3CA siRNA treatment showed no effect. Thus, EVSA-T single clones show broad resistance to PI3K inhibition but maintained dependency on p110β expression and signaling.

Figure 1.

PI3K inhibitor–resistant EVSA-T breast cancer cells show broad cross-resistance to PI3K inhibitors and maintained p110β signal dependency. A, cell growth inhibition and apoptosis induction of EVSA-T parental and single clones by GDC-0941. Cells were seeded and treated with GDC-0941 for 4 days and then cell viability and apoptosis induction were measured. The data represent the mean of four replicates. B, signal modulation by GDC-0941 (pan-PI3K inhibitor), GDC-0349 (mTOR kinase inhibitor), or GDC-0068 (AKt inhibitor) in EVSA-T parental and single clones. Cells were treated with a serial concentration of GDC-0941 and incubated for 2 hours before harvesting. C, cell growth inhibitory activity of panel of anticancer agents against EVSA-T parental and single clones. The ratio of IC50 to parental cells is indicated. The IC50 data represent the mean of four replicates. D, cell growth inhibition and PI3K signal modulation by PIK3CA or PIK3CB siRNA in EVSA-T parental and single clones. Cells were seeded and treated with indicated siRNAs for 4 days before viability was measured and cells were lysed and analyzed by Western blot analysis. The cell viability data represent the mean of three replicates.

Figure 1.

PI3K inhibitor–resistant EVSA-T breast cancer cells show broad cross-resistance to PI3K inhibitors and maintained p110β signal dependency. A, cell growth inhibition and apoptosis induction of EVSA-T parental and single clones by GDC-0941. Cells were seeded and treated with GDC-0941 for 4 days and then cell viability and apoptosis induction were measured. The data represent the mean of four replicates. B, signal modulation by GDC-0941 (pan-PI3K inhibitor), GDC-0349 (mTOR kinase inhibitor), or GDC-0068 (AKt inhibitor) in EVSA-T parental and single clones. Cells were treated with a serial concentration of GDC-0941 and incubated for 2 hours before harvesting. C, cell growth inhibitory activity of panel of anticancer agents against EVSA-T parental and single clones. The ratio of IC50 to parental cells is indicated. The IC50 data represent the mean of four replicates. D, cell growth inhibition and PI3K signal modulation by PIK3CA or PIK3CB siRNA in EVSA-T parental and single clones. Cells were seeded and treated with indicated siRNAs for 4 days before viability was measured and cells were lysed and analyzed by Western blot analysis. The cell viability data represent the mean of three replicates.

Close modal

The recurrent PIK3CB mutations D1067Y/A/V are sufficient to confer resistance to PI3K inhibitors

To identify possible mechanisms underlying resistance, we ran a comprehensive panel of assays including whole-exome sequencing and a quantitative copy number variation assay for 35 genes, using DNA derived from both EVSA-T parental cells and resistant derivatives. Whole-exome sequencing identified a mutation in PIK3CB predicted to result in a D1067Y substitution in the C-terminal region of p110β in EVSA-T–resistant pool cells (Supplementary Fig. S7; Supplementary Table S2). Filtering of the sequencing data for variants present at >20% allele frequency in the resistant pool resulted in a total of 43 sequence variants. We focused on the PIK3CB-predicted variant D1067Y, present at 32% allele frequency, due to the known functional role of PIK3CB in PI3K pathway activity. The mutation was confirmed by Sanger sequencing and ddPCR in all five resistant clones but not parental EVSA-T cells (Fig. 2A and Supplementary Fig. S8). Although a mutant allele frequency of 50% might have been expected in these single cell clones if EVSA-T were diploid for the PIK3CB gene, EVSA-T parental cells carry more than two copies of the PIK3CB gene (data not shown). We did not find any other copy number alterations or mutations in PI3K pathway related genes or known genes associated with resistance to PI3K inhibition, such as Myc or c-Met, in EVSA-T–resistant clones (Supplementary Fig. S9; ref. 31). To determine whether this putative mechanism of resistance via PIK3CB mutation was generalizable beyond EVSA-T, we selected for GDC-0941 resistance in three other PTEN-deficient breast cancer cell lines, ZR-75-1, HCC-1569, and SUM-52PE (Supplementary Fig. S10) and screened resistant pools for PIK3CB mutation using ddPCR. We found the same alteration, D1067Y, at an allele frequency of 1.5% in a resistant pool of the luminal PTEN-deficient breast cancer cell line, ZR-75-1 (Fig. 2A). As with EVSA-T, we did not detect the PIK3CB mutation in parental ZR-75-1 cells using ddPCR. If the mutation was present as a pre-existing clone in parental cell lines, it would have occurred at a frequency of less than 1 in 10,000 cells based on the 0.01% sensitivity of the ddPCR assay. To determine whether amino acid substitutions at this position occur more broadly in tumor specimens, we searched TCGA data (38, 39) and found recurrent PIK3CB mutations at position D1067 in cancer patients, with D1067Y, D1067A, and D1067V mutations in PIK3CB all annotated. Also, we found one of 72 patients had the PIK3CB D1067V mutation in a collection of metastatic clear cell renal cell carcinoma samples (Table 1 and Supplementary Fig. S11). To functionally characterize these recurrent mutations and their role in resistance and transformation, we established stable derivatives of EVSA-T cells expressing wild-type (WT), D1067Y, D1067A, or D1067V variants. In cell viability experiments, mutant expressing cells were less sensitive to GDC-0941 than WT-expressing cells (Fig. 2B). In addition, basal phospho-AKT levels were elevated in D1067Y/A/V–stable cell lines compared with WT lines, and 4-hour GDC-0941 treatment did not suppress phospho-AKT in mutant-expressing cells (Fig. 2B). Consistent with the previous data in the selected resistant clones, the EVSA-T expressing PIK3CB mutant showed cross-resistance to GDC-0032 and TGX-221, but maintained sensitivity to GDC-0068 and GDC-0349 (Fig. 2C). Consistent with cross-resistance to GDC-0032 and TGX-221 of both stably and transiently expressing mutants (Supplementary Fig. S12), these inhibitors could only partially suppress PI3K signaling at the level of phospho-AKT and could not suppress phospho-S6 in stably expressing mutants (Supplementary Fig. S13). To additionally test whether an endogenously occurring PIK3CB mutation results in intrinsic resistance to PI3K inhibition, we characterized a renal cancer cell line, A498, harboring a PIK3CB D1067V mutation. Although A498, like EVSA-T, depends on p110β signaling (Supplementary Fig. S14), A498 was substantially less sensitive to GDC-0941 than EVSA-T in a cell viability assay (A498: IC50 1.1 μmol/L, EVSA-T: IC50 0.037 μmol/L) and required higher concentrations to suppress phospho-AKT (Fig. 2D). Taken together, these results suggest that the recurrent PIK3CB mutations, D1067Y, D1067A, and D1067V are sufficient to confer intrinsic and acquired resistance to upstream PI3K inhibition.

Figure 2.

EVSA-T and ZR75-1 harbor C-terminal PIK3CB mutations and ectopic expression of D1067Y/A/V confers resistance to PI3K inhibitors. A, detection of PIK3CB mutation by Sanger sequence and ddPCR. Genomic DNA from EVSA-T parental and resistant clones were sequenced via Sanger sequencing. Genomic DNA from EVSA-T parental and resistant clones, as well as ZR-75-1 parental and resistant pools, were analyzed with ddPCR. Data represent three independent ddPCR reactions. B, cell growth inhibition and signal modulation by GDC-0941 in EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells. Cells were incubated with GDC-0941 for 4 days and cell viability was measured. The cell viability data represent the mean of four replicates. After 2 hours incubation of 1 μmol/L GDC-0941, cells were lysed and analyzed by Western blot analysis. C, cell growth–inhibitory activity of panel of compounds against EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells by indicated compounds. The ratio of IC50 to parental cells is indicated. The IC50 data represent the mean of four replicates. D, cell growth inhibition and signal modulation by GDC-0941 in EVSA-T and A498 cells. Cells were incubated with GDC-0941 for 4 days and cell viability was measured. The cell viability data represent the mean of four replicates. After 2 hours incubation of 1 μmol/L GDC-0941, cells were lysed and analyzed by Western blot analysis.

Figure 2.

EVSA-T and ZR75-1 harbor C-terminal PIK3CB mutations and ectopic expression of D1067Y/A/V confers resistance to PI3K inhibitors. A, detection of PIK3CB mutation by Sanger sequence and ddPCR. Genomic DNA from EVSA-T parental and resistant clones were sequenced via Sanger sequencing. Genomic DNA from EVSA-T parental and resistant clones, as well as ZR-75-1 parental and resistant pools, were analyzed with ddPCR. Data represent three independent ddPCR reactions. B, cell growth inhibition and signal modulation by GDC-0941 in EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells. Cells were incubated with GDC-0941 for 4 days and cell viability was measured. The cell viability data represent the mean of four replicates. After 2 hours incubation of 1 μmol/L GDC-0941, cells were lysed and analyzed by Western blot analysis. C, cell growth–inhibitory activity of panel of compounds against EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells by indicated compounds. The ratio of IC50 to parental cells is indicated. The IC50 data represent the mean of four replicates. D, cell growth inhibition and signal modulation by GDC-0941 in EVSA-T and A498 cells. Cells were incubated with GDC-0941 for 4 days and cell viability was measured. The cell viability data represent the mean of four replicates. After 2 hours incubation of 1 μmol/L GDC-0941, cells were lysed and analyzed by Western blot analysis.

Close modal
Table 1.

PIK3CB mutation in TCGA dataset and Genentech dataset

TCGA IDTissue typePIK3CB mutationPTEN mutation
TCGA-32-2494 Glioblastoma D1067A  
TCGA-06-0190 Glioblastoma D1067V D92E 
TCGA-CV-7421 Head and neck cancer D1067V  
TCGA-A5-A1OF Endometrial cancer D1067V  
TCGA-D1-A16X Endometrial cancer D1067Y R130Q/E201* 
A498 Renal cell carcinoma (Cell line) D1067V  
Study ID Tissue type PIK3CB D1067A mutation  
PIM4973g Renal cell carcinoma 1/72 (1.4%)  
TCGA IDTissue typePIK3CB mutationPTEN mutation
TCGA-32-2494 Glioblastoma D1067A  
TCGA-06-0190 Glioblastoma D1067V D92E 
TCGA-CV-7421 Head and neck cancer D1067V  
TCGA-A5-A1OF Endometrial cancer D1067V  
TCGA-D1-A16X Endometrial cancer D1067Y R130Q/E201* 
A498 Renal cell carcinoma (Cell line) D1067V  
Study ID Tissue type PIK3CB D1067A mutation  
PIM4973g Renal cell carcinoma 1/72 (1.4%)  

Mechanistic characterization of PIK3CB mutation mediated PI3K resistance

To establish that resistance to GDC-0941 was not due to differential binding, we confirmed that the binding affinity of GDC-0941 to p110β kinase in EVSA-T parental cells and resistant clones was comparable by using an ATP competition assay (Supplementary Fig. S15). Then, to clarify how PIK3CB mutants can confer resistance to PI3K inhibitors, we characterized the effect of these mutations on downstream PI3K pathway signaling. Because resistant clones of EVSA-T harboring the PIK3CB D1067Y mutation showed higher phospho-AKT levels than EVSA-T parental cells at baseline (Fig. 3A and Supplementary Fig. S16), we expressed PIK3CB WT or D1067Y mutant transiently in EVSA-T to evaluate whether PIK3CB mutants can further potentiate phospho-AKT levels. p110β WT or D1067Y-mutant protein levels increased in proportion to the amount of input DNA, while phospho-AKT levels, especially at threonine 308, showed a proportional increase in PIK3CB D1067Y-mutant cells but not WT-overexpressing cells (Fig. 3B). Similarly, PIK3CB D1067A or D1067V mutant transiently expressing HEK293 cells showed higher phospho-AKT levels than PIK3CB WT-expressing HEK293 cells (Fig. 3C). As AKT activation may also occur via a kinase activity–independent mechanism, we expressed a kinase-dead (KD)-mutant PIK3CB (K805R) in HEK293 and confirmed that neither PIK3CB WT nor PIK3CB D1067Y/A/V KD mutants could potentiate AKT phosphorylation (Fig. 3D). These results suggest that the PIK3CB D1067Y/A/V mutants have increased catalytic activity and required kinase activity, and potently activate downstream PI3K signaling.

Figure 3.

PI3K pathway activation by p110β D1067Y/A/V mutants is kinase activity dependent. A, baseline AKT phosphorylation in EVSA-T parental and resistant clones. B, increased AKT phosphorylation mediated by p110β D1067Y expression. A serially increasing amount of expression vectors of PIK3CB WT or D1067Y mutant was transfected to EVSA-T. Two days after the transfection, cells were lysed and analyzed by Western blot analysis. C, increased AKT phosphorylation level mediated by p110β D1067Y/A/V expression. The expression vectors of PIK3CB WT, D1067Y, D1067A, or D1067V mutant were transfected with HEK293 cells. Two days after the transfection, cells were lysed and analyzed by Western blot analysis. D, expression of a series of kinase dead variants of p110β D1067Y/A/V. The expression vectors of PIK3CB WT, WT-KD, D1067Y, D1067Y-KD, D1067A, D1067A-KD, D1067V, or D1067V-KD were transfected with HEK293 cells. Two days after the transfection, cells were lysed and analyzed by Western blot analysis.

Figure 3.

PI3K pathway activation by p110β D1067Y/A/V mutants is kinase activity dependent. A, baseline AKT phosphorylation in EVSA-T parental and resistant clones. B, increased AKT phosphorylation mediated by p110β D1067Y expression. A serially increasing amount of expression vectors of PIK3CB WT or D1067Y mutant was transfected to EVSA-T. Two days after the transfection, cells were lysed and analyzed by Western blot analysis. C, increased AKT phosphorylation level mediated by p110β D1067Y/A/V expression. The expression vectors of PIK3CB WT, D1067Y, D1067A, or D1067V mutant were transfected with HEK293 cells. Two days after the transfection, cells were lysed and analyzed by Western blot analysis. D, expression of a series of kinase dead variants of p110β D1067Y/A/V. The expression vectors of PIK3CB WT, WT-KD, D1067Y, D1067Y-KD, D1067A, D1067A-KD, D1067V, or D1067V-KD were transfected with HEK293 cells. Two days after the transfection, cells were lysed and analyzed by Western blot analysis.

Close modal

Because measurement of phosphatidylinositol (3,4,5)-trisphosphate (PIP3) and phosphatidylinositol (4,5)-bisphosphate (PI(4,5)P2) levels provide a direct assay for PI3K catalytic activity (16), we evaluated PIP3 and PI(4,5)P2 levels in resistant clones of EVSA-T harboring the PIK3CB D1067Y mutation. The resistant clones showed approximately 8-fold higher PIP3/PI(4,5)P2 ratio than EVSA-T parental cells at baseline (Fig. 4A). To investigate the biologic consequences of elevated PIP3 levels, we fractionated membrane and cytosolic proteins from EVSA-T parental cells and resistant clone cells, and determined the localization of downstream molecules that are recruited to the membrane by PIP3 binding. Consistent with the higher amount of PIP3 at the membrane, AKT and PDK1 were relatively highly localized to the membrane and highly phosphorylated in EVSA-T–resistant clones (Fig. 4B). We fractionated membrane and cytosolic proteins from EVSA-T PIK3CB mutant stably expressing cells (D1067Y, D1067A, and D1067V), and observed the same phenomenon as we found in EVSA-T–resistant clones, namely elevated membrane-associated AKT that was highly phosphorylated (Fig. 4C). Finally, we elucidated the inhibitory activity of GDC-0941 on PIP3 production in EVSA-T parental cells and resistant clones by treating them with GDC-0941 at 1 μmol/L for 4 hours and measuring the PIP3/PIP(4,5)2 ratio. Baseline levels of PIP3 in resistant clones were significantly higher than that in parental cells, and while GDC-0941 treatment suppressed PIP3 in resistant cells, remaining PIP3 levels were substantially higher than in parental cells treated with similar doses of GDC-0941 (Fig. 4D). These results are consistent with D1067 alterations increasing p110β catalytic activity on its lipid substrate and driving PI3K signaling through enhanced recruitment of AKT to the membrane, causing resistance by increasing the amount of inhibitor required to suppress lipid signaling.

Figure 4.

Activating mutations in p110β result in increased membrane bound PIP3 in elevated pAKT levels. A, baseline PIP3/PI(4,5)P2 ratio in a series of EVSA-T cell lines. The amount of PIP3 and PI(4,5)P2 in EVSA-T parental and resistant clones was measured by ELISA. The data represent the mean of three replicates. Ratios of PIP3/PI(4,5)P2 were calculated. B, fractionation of membrane and cytosol proteins of a series of EVSA-T cell lines. The membrane and cytosol proteins of EVSA-T parental and resistant clones were fractionated and protein levels were analyzed by Western blot analysis. C, fractionation of membrane and cytosol proteins of EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells. D, inhibition of PIP3 production by GDC-0941. The amount of PIP3 and PI(4,5)P2 in EVSA-T parental and resistant clones upon the treatment of GDC-0941 were measured by ELISA kit. The data represent the mean of three replicates.

Figure 4.

Activating mutations in p110β result in increased membrane bound PIP3 in elevated pAKT levels. A, baseline PIP3/PI(4,5)P2 ratio in a series of EVSA-T cell lines. The amount of PIP3 and PI(4,5)P2 in EVSA-T parental and resistant clones was measured by ELISA. The data represent the mean of three replicates. Ratios of PIP3/PI(4,5)P2 were calculated. B, fractionation of membrane and cytosol proteins of a series of EVSA-T cell lines. The membrane and cytosol proteins of EVSA-T parental and resistant clones were fractionated and protein levels were analyzed by Western blot analysis. C, fractionation of membrane and cytosol proteins of EVSA-T mock, PIK3CB WT, D1067Y, D1067A, or D1067V stably expressing cells. D, inhibition of PIP3 production by GDC-0941. The amount of PIP3 and PI(4,5)P2 in EVSA-T parental and resistant clones upon the treatment of GDC-0941 were measured by ELISA kit. The data represent the mean of three replicates.

Close modal

An oncogenic role for mutant p110β

To determine whether expression of the PIK3CB D1067Y mutation is sufficient to transform normal cells, we conducted a series of studies with a normal fibroblast Rat cell, Rat-2, stably expressing PIK3CB WT or the D1067Y mutant (Supplementary Fig. S17). Rat-2 cells expressing PIK3CB D1067Y developed foci but Rat-2 cells expressing PIK3CB WT did not (Fig. 5A). We also measured anchorage-independent growth capability in soft agar and spheroid formation assays, and found Rat-2 cells expressing PIK3CB D1067Y formed more colonies in soft agar and constructed larger spheroids than Rat-2 cells expressing PIK3CB WT (Fig. 5A). Cell viability in the colony formation assay and the spheroid formation assay was measured by an ATP-based cell viability assay and was found to be enhanced in D1067Y cells compared with WT (Fig. 5B). Finally, to determine whether PTEN loss can potentiate transforming activity of the PIK3CB mutation, we performed PTEN siRNA in Rat-2–stable cell lines and showed that spheroid forming activity of Rat-2 cells expressing PIK3CB D1067Y was enhanced by PTEN knockdown (Fig. 5C).

Figure 5.

Oncogenic activity of the p110β D1067Y mutant. A, the PIK3CB D1067Y mutation has enhanced activity in focus formation, colony formation, and spheroid formation assay in normal Rat2 cells. For focus formation assays, cells were seeded in a 6-well plate and incubated for 20 days. For colony formation assays, cells were seeded in a 96-well soft agar plate and incubated for 15 days. For spheroid formation assay, cells were seeded in a 96-well U bottom plate and incubated for 21 days. Rat2 mock, PIK3CB WT, or D1067Y stably expressing cells were used for three assays. Cell morphologies were recorded under a microscope. B, quantification of cell viability in colony and spheroid formation assays as assessed using CellTiter Glo ATP Luminescence assay. C, spheroid-forming activity is enhanced in the presence of PTEN siRNA. The cell viability was assessed using CellTiter Glo ATP Luminescence assay 7 days after the siRNA treatment. PTEN, phospho-AKT, and AKT protein levels were detected by Western blot analysis 4 days after the siRNA treatment.

Figure 5.

Oncogenic activity of the p110β D1067Y mutant. A, the PIK3CB D1067Y mutation has enhanced activity in focus formation, colony formation, and spheroid formation assay in normal Rat2 cells. For focus formation assays, cells were seeded in a 6-well plate and incubated for 20 days. For colony formation assays, cells were seeded in a 96-well soft agar plate and incubated for 15 days. For spheroid formation assay, cells were seeded in a 96-well U bottom plate and incubated for 21 days. Rat2 mock, PIK3CB WT, or D1067Y stably expressing cells were used for three assays. Cell morphologies were recorded under a microscope. B, quantification of cell viability in colony and spheroid formation assays as assessed using CellTiter Glo ATP Luminescence assay. C, spheroid-forming activity is enhanced in the presence of PTEN siRNA. The cell viability was assessed using CellTiter Glo ATP Luminescence assay 7 days after the siRNA treatment. PTEN, phospho-AKT, and AKT protein levels were detected by Western blot analysis 4 days after the siRNA treatment.

Close modal

In this study, we sought to define resistance mechanisms for PI3K inhibition in PTEN-deficient breast cancer. Previous studies have shown that PTEN-deficient cancers tend to signal through p110β rather than p110α (18). Consistent with this, our siRNA experiments indicated that EVSA-T PTEN-deficient breast cancer cells depend on p110β signaling rather than p110α. Comprehensive next-generation sequencing of derived GDC-0941–resistant EVSA-T cells identified a novel PIK3CB D1067Y mutation. This phenomenon was not unique to EVSA-T cells, as we found the identical mutation in a GDC-0941–resistant derivative pool of another PTEN-deficient breast cancer cell line, ZR-75-1. Although present at a low frequency of 1.5% in ZR-75-1 cells, the mutation could be contributing to one of multiple resistance mechanisms in this heterogeneous-resistant pool, and its occurrence in two independent PTEN-deficient lines treated with GDC-0941 supports the notion that PIK3CB is a common target for PI3K inhibitor escape in the PTEN-null setting. These findings suggest that PTEN-deficient cancers are intrinsically dependent on p110β signaling and may acquire the mutation to maintain the p110β signaling in the presence of a PI3K inhibitor. This observation is reminiscent of our observation that PIK3CA-mutant KPL-4 cells develop resistance by amplifying the mutant allele of PIK3CA and hyperactivating signaling to maintain pathway activation in the presence of PI3K inhibitors (15).

Surprisingly, PIK3CB mutations at the identical position were also found recurrently in non-treated cancer patients and the A498 renal cell cancer cell line. As we observed in EVSA-T and ZR-75-1 cells expressing mutant PIK3CB, A498 cells harboring a PIK3CB D1067V mutation showed resistance to GDC-0941. Although the frequency of PIK3CB mutations in non-treated cancer patients is relatively low, it nevertheless comprises a potential innate resistance mechanism for PI3K inhibitors. The situation may be akin to other cases where a mutation that is found infrequently in the pretreated patients is also found after the treatment and can be a prevalent mechanism of resistance. Such is the case with EGFR, where it is well known that the acquired gatekeeper mutation, T790M, in EGFR impairs EGFR inhibitor binding with EGFR and confers resistance (40). Although it is not common, the same mutation T790M in EGFR has been found in EGFR inhibitor–naïve non–small cell lung cancer (41). Likewise, estrogen receptor 1 (ESR1) mutations are rare in endocrine therapy–naïve breast cancer patients, but are found at a higher prevalence in metastatic breast cancers following therapy with aromatase inhibitors. Some of these mutations lead to ligand-independent estrogen receptor (ER) activation and others attenuate binding affinity with ER antagonists (42, 43). Our results suggests a model wherein PTEN-deficient cancers either acquire the mutation in PIK3CB or select for rare pre-existing clones in the presence of PI3K inhibition, suggesting that screening for PIK3CB D1067Y/A/V mutations patients who have progressed on PI3K inhibitor treatment may be warranted.

There are several possible mechanisms by which D1067 alterations cause p110β activation. Class I PI3K consists of heterodimers composed of a catalytic (p110) and a regulatory subunit (p85). p110 kinase activity is associated with the nSH2, iSH2, and cSH2 domains of p85. Therefore, dysregulation of the interaction of p110 with p85 can activate p110 kinase activity. For instance, the nSH2 domain of p85 interacts with the helical domain of p110 and suppresses p110 kinase activity. In certain cancers, PIK3CA is mutated at the E545 position in the helical domain. The p110α E545K mutant no longer interacts with the p85 nSH2 domain and is highly activated (44). The p85 cSH2 domain associates with the C-terminus of p110β but not that of p110α and inhibits basal p110β kinase activity in the absence of upstream signaling (45). The D1067 residue is located in the 12th α-helix of p110β, and a mutationally induced conformational change of this helix could theoretically regulate the kinase activity of p110β by attenuating the binding affinity for the p85 cSH2 domain (46), thereby relieving the inhibitory effect of the p85 cSH2 domain and increasing basal p110β kinase activity. In addition, the C-terminus of p110β plays a critical role in phosphorylation of lipid substrate. Although the C-terminal–truncated p110β has higher ATPase activity than WT p110β due to lack of inhibitory effect by the 12th α-helix, the p110β kinase activity against lipid substrates is decreased and C-terminal–truncated p110β cannot induce AKT phosphorylation in cells (46). In this study, we showed that the increase of AKT phosphorylation by p110β D1067Y/A/V mutants was dependent on p110β kinase activity and the phosphorylation of the direct substrate, PI(4,5)P2, was enhanced in cells expressing the mutant kinase. Our results are consistent with a model where p110β D1067Y/A/V mutants might have higher affinity to lipid substrates than p110β WT, resulting in enhanced signaling compared with WT.

During the preparation of this manuscript, the PIK3CB D1067V mutation was reported as a resistance mechanism for erlotinib in lung cancer and was shown to have oncogenic activity and sensitize cells to PI3K inhibition (47). In this study, Pazarentzos and colleagues showed that specific expression of PIK3CB D1067V in both NIH-3T3 cells and H3255 NSCLC cells conferred increased sensitivity of these cells to the p110β isoform–selective inhibitor TGX-221, suggesting that therapeutic treatment with p110β-selective inhibitors may be effective in cancers harboring PIK3CB-activating mutations. We also demonstrated oncogenic activity of PIK3CB D1067Y mutation in Rat-2 normal cells, though our results are more consistent with PIK3CB mutations conferring resistance to PI3K inhibition, based on the nature of the resistant mutations and the fact that A498 cells harboring a naturally occurring PIK3CB D1067V mutation were relatively resistant to GDC-0941 in terms of growth inhibition and phospho-AKT–inhibitory activity. Interestingly, we observed cross-resistance of EVSA-T clones to p110β-selective inhibitors TGX-221, AZD-6482, and GSK2636771. Furthermore, we found that EVSA-T cells genetically engineered to stably express PIK3CB D1067Y/A/V mutations were resistant to TGX-221 as compared with EVSA-T–expressing PIK3CB WT cells. Although PIK3CB-mutant cells may have been expected to respond to p110β-selective inhibitors, one important difference between the previous study and our study is PTEN status of the PIK3CB-mutant models evaluated. Although PIK3CB D1067V mutation confers sensitivity to a p110β-selective inhibitor in a PTEN WT background in NIH-3T3 and H3255 cells, it appears to confer resistance to this same inhibitor in PTEN deficient EVSA-T and ZR-75-1 cells. These observations support the hypothesis that loss of negative regulation of PI3K activity by PTEN combined with increased kinase activity by PIK3CB mutation results in hyperactivation of PI3K signaling to a level that enables escape from p110β-selective inhibitors. In a PTEN-deficient setting, treatment with inhibitors targeting nodes downstream of p110β may therefore be required. The notion that efficacy of p110β-selective inhibition is context-dependent is consistent with previous work showing an effect of TGX-221 on cell viability in PTEN-negative but not PTEN-positive cells (48). Indeed, Pazarentzos and colleagues also emphasize that the effects of PIK3CB mutation on therapeutic efficacy of PI3K inhibitors are likely to be context-specific. We also propose treatment with AKT or TORC1/2 inhibitors as a therapeutic option for PIK3CB-mutant cancers based on our findings that PIK3CB-mutant cells still maintain sensitivity and pathway modulation in response to these inhibitors, and by analogy to the situation in K-ras–mutant tumors, where inhibition of the downstream node ERK resistance to MEK inhibitors mediated by upstream activating events (49). Inhibiting PI3K pathway activity downstream of p110β by using AKT or TORC1/2 inhibitors may also be desirable in PTEN-deficient settings given the role of p110β in insulin signaling and glucose homeostasis and the need to spare its activity in normal tissues.

In summary, we demonstrate here for the first time that novel activating PIK3CB D1067Y/A/V mutations confer resistance to PI3K inhibitors due to hyperactivation of the PI3K pathway and defining a role for PIK3CB as an activated oncogene. Our results have implications for the management of patients receiving targeted PI3K therapies, and suggest that additional therapeutic options targeted at AKT or mTORC1/2 may be a treatment option in PI3K therapy–resistant patients.

No potential conflicts of interest were disclosed.

Conception and design: Y. Nakanishi, G.M. Hampton, M.R. Lackner

Development of methodology: Y. Nakanishi, K. Walter, J.M. Spoerke, L.Y. Huw, C. O'Brien

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Nakanishi, K. Walter, J.M. Spoerke, L.Y. Huw, C. O'Brien

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Nakanishi, K. Walter, J.M. Spoerke, M.R. Lackner, C. O'Brien

Writing, review, and/or revision of the manuscript: Y. Nakanishi, K. Walter, J.M. Spoerke, G.M. Hampton, M.R. Lackner, C. O'Brien

Study supervision: Y. Nakanishi

The authors thank Zemin Zhang, Jeff Settleman, and Pete Haverty for generating and analyzing whole-genome sequencing data on the resistant lines, Charlie Eigenbrot for p110β protein structural modeling and helpful discussions, Teiko Sumiyoshi and An Do for running quantitative PCR to detect somatic mutations, Junko Aimi and David Tran for managing clinical sample acquisition, and My Vo, Pranamee Sarma, and Cassandra Greene for DNA isolation and aliquoting.

This work was funded solely by Genentech, Inc.

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.

1.
Samuels
Y
,
Wang
ZH
,
Bardelli
A
,
Silliman
N
,
Ptak
J
,
Szabo
S
, et al
High frequency of mutations of the PIK3CA gene in human cancers
.
Science
2004
;
304
:
554
.
2.
Bourgon
R
,
Lu
S
,
Yan
Y
,
Lackner
MR
,
Wang
W
,
Weigman
V
, et al
High-throughput detection of clinically relevant mutations in archived tumor samples by multiplexed PCR and next-generation sequencing
.
Clin Cancer Res
2014
;
20
:
2080
91
.
3.
Carpten
JD
,
Faber
AL
,
Horn
C
,
Donoho
GP
,
Briggs
SL
,
Robbins
CM
, et al
A transforming mutation in the pleckstrin homology domain of AKT1 in cancer
.
Nature
2007
;
448
:
439
44
.
4.
Rodon
J
,
Dienstmann
R
,
Serra
V
,
Tabernero
J
. 
Development of PI3K inhibitors: lessons learned from early clinical trials
.
Nat Rev Clin Oncol
2013
;
10
:
143
53
.
5.
Dienstmann
R
,
Rodon
J
,
Serra
V
,
Tabernero
J
. 
Picking the point of inhibition: a comparative review of PI3K/AKT/mTOR pathway inhibitors
.
Mol Cancer Ther
2014
;
13
:
1021
31
.
6.
Motzer
RJ
,
Escudier
B
,
Oudard
S
,
Hutson
TE
,
Porta
C
,
Bracarda
S
, et al
Efficacy of everolimus in advanced renal cell carcinoma: a double-blind, randomised, placebo-controlled phase III trial
.
Lancet
2008
;
372
:
449
56
.
7.
Baselga
J
,
Campone
M
,
Piccart
M
,
Burris
HA
 III
,
Rugo
HS
,
Sahmoud
T
, et al
Everolimus in postmenopausal hormone-receptor-positive advanced breast cancer
.
N Engl J Med
2012
;
366
:
520
9
.
8.
Lackner
MR
,
Wilson
TR
,
Settleman
J
. 
Mechanisms of acquired resistance to targeted cancer therapies
.
Future Oncol
2012
;
8
:
999
1014
.
9.
Flaherty
KT
,
Infante
JR
,
Daud
A
,
Gonzalez
R
,
Kefford
RF
,
Sosman
J
, et al
Combined BRAF and MEK inhibition in melanoma with BRAF V600 mutations
.
N Engl J Med
2012
;
367
:
1694
703
.
10.
Liu
P
,
Cheng
H
,
Santiago
S
,
Raeder
M
,
Zhang
F
,
Isabella
A
, et al
Oncogenic PIK3CA-driven mammary tumors frequently recur via PI3K pathway-dependent and PI3K pathway-independent mechanisms
.
Nat Med
2011
;
17
:
1116
20
.
11.
Ilic
N
,
Utermark
T
,
Widlund
HR
,
Roberts
TM
. 
PI3K-targeted therapy can be evaded by gene amplification along the MYC-eukaryotic translation initiation factor 4E (eIF4E) axis
.
Proc Natl Acad Sci U S A
2011
;
108
:
E699
U07
.
12.
Serra
V
,
Eichhorn
PJA
,
Garcia-Garcia
C
,
Ibrahim
YH
,
Prudkin
L
,
Sanchez
G
, et al
RSK3/4 mediate resistance to PI3K pathway inhibitors in breast cancer
.
J Clin Invest
2013
;
123
:
2551
63
.
13.
Edgar
KA
,
Crocker
L
,
Cheng
E
,
Wagle
MC
,
Wongchenko
M
,
Yan
Y
, et al
Amphiregulin and PTEN evoke a multimodal mechanism of acquired resistance to PI3K inhibition
.
Genes Cancer
2014
;
5
:
113
26
.
14.
Elkabets
M
,
Pazarentzos
E
,
Juric
D
,
Sheng
Q
,
Pelossof
RA
,
Brook
S
, et al
AXL mediates resistance to PI3Kalpha inhibition by activating the EGFR/PKC/mTOR axis in head and neck and esophageal squamous cell carcinomas
.
Cancer Cell
2015
;
27
:
533
46
.
15.
Huw
LY
,
O'Brien
C
,
Pandita
A
,
Mohan
S
,
Spoerke
JM
,
Lu
S
, et al
Acquired PIK3CA amplification causes resistance to selective phosphoinositide 3-kinase inhibitors in breast cancer
.
Oncogenesis
2013
;
2
:
e83
.
16.
Costa
C
,
Ebi
H
,
Martini
M
,
Beausoleil
SA
,
Faber
AC
,
Jakubik
CT
, et al
Measurement of PIP3 levels reveals an unexpected role for p110beta in early adaptive responses to p110alpha-specific inhibitors in luminal breast cancer
.
Cancer Cell
2015
;
27
:
97
108
.
17.
Juric
D
,
Castel
P
,
Griffith
M
,
Griffith
OL
,
Won
HH
,
Ellis
H
, et al
Convergent loss of PTEN leads to clinical resistance to a PI(3)Kalpha inhibitor
.
Nature
2015
;
518
:
240
4
.
18.
Jia
S
,
Liu
Z
,
Zhang
S
,
Liu
P
,
Zhang
L
,
Lee
SH
, et al
Essential roles of PI(3)K-p110beta in cell growth, metabolism and tumorigenesis
.
Nature
2008
;
454
:
776
9
.
19.
Guertin
DA
,
Stevens
DM
,
Saitoh
M
,
Kinkel
S
,
Crosby
K
,
Sheen
JH
, et al
mTOR complex 2 is required for the development of prostate cancer induced by Pten loss in mice
.
Cancer Cell
2009
;
15
:
148
59
.
20.
Schwartz
S
,
Wongvipat
J
,
Trigwell
CB
,
Hancox
U
,
Carver
BS
,
Rodrik-Outmezguine
V
, et al
Feedback suppression of PI3Kalpha signaling in PTEN-mutated tumors is relieved by selective inhibition of PI3Kbeta
.
Cancer Cell
2015
;
27
:
109
22
.
21.
Folkes
AJ
,
Ahmadi
K
,
Alderton
WK
,
Alix
S
,
Baker
SJ
,
Box
G
, et al
The identification of 2-(1H-indazol-4-yl)-6-(4-methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin-4-yl-t hieno[3,2-d]pyrimidine (GDC-0941) as a potent, selective, orally bioavailable inhibitor of class I PI3 kinase for the treatment of cancer
.
J Med Chem
2008
;
51
:
5522
32
.
22.
O'Brien
C
,
Wallin
JJ
,
Sampath
D
,
GuhaThakurta
D
,
Savage
H
,
Punnoose
EA
, et al
Predictive biomarkers of sensitivity to the phosphatidylinositol 3′ kinase inhibitor GDC-0941 in breast cancer preclinical models
.
Clin Cancer Res
2010
;
16
:
3670
83
.
23.
Ndubaku
CO
,
Heffron
TP
,
Staben
ST
,
Baumgardner
M
,
Blaquiere
N
,
Bradley
E
, et al
Discovery of 2-{3-[2-(1-isopropyl-3-methyl-1H-1,2-4-triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepin-9-yl]-1H-pyrazol-1-yl}-2-methylpropanamide (GDC-0032): a beta-sparing phosphoinositide 3-kinase inhibitor with high unbound exposure and robust in vivo antitumor activity
.
J Med Chem
2013
;
56
:
4597
610
.
24.
Sutherlin
DP
,
Bao
L
,
Berry
M
,
Castanedo
G
,
Chuckowree
I
,
Dotson
J
, et al
Discovery of a potent, selective, and orally available class I phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) kinase inhibitor (GDC-0980) for the treatment of cancer
.
J Med Chem
2011
;
54
:
7579
87
.
25.
Wallin
JJ
,
Edgar
KA
,
Guan
J
,
Berry
M
,
Prior
WW
,
Lee
L
, et al
GDC-0980 is a novel class I PI3K/mTOR kinase inhibitor with robust activity in cancer models driven by the PI3K pathway
.
Mol Cancer Ther
2011
;
10
:
2426
36
.
26.
Blake
JF
,
Xu
R
,
Bencsik
JR
,
Xiao
D
,
Kallan
NC
,
Schlachter
S
, et al
Discovery and preclinical pharmacology of a selective ATP-competitive Akt inhibitor (GDC-0068) for the treatment of human tumors
.
J Med Chem
2012
;
55
:
8110
27
.
27.
Pei
Z
,
Blackwood
E
,
Liu
L
,
Malek
S
,
Belvin
M
,
Koehler
MF
, et al
Discovery and biological profiling of potent and selective mTOR inhibitor GDC-0349
.
ACS Med Chem Lett
2013
;
4
:
103
7
.
28.
Spoerke
JM
,
O'Brien
C
,
Huw
L
,
Koeppen
H
,
Fridlyand
J
,
Brachmann
RK
, et al
Phosphoinositide 3-kinase (PI3K) pathway alterations are associated with histologic subtypes and are predictive of sensitivity to PI3K inhibitors in lung cancer preclinical models
.
Clin Cancer Res
2012
;
18
:
6771
83
.
29.
Powles
TO
,
Escudier
S
,
Brown
BJ
,
Hawkins
JE
,
Castellano
RE
,
Ravaud
DE
, et al
2014 A randomized phase II study of GDC-0980 versus everolimus in metastatic renal cell carcinoma (mRCC) patients (pts) after VEGF-targeted therapy (VEGF-TT)
.
J Clin Oncol
32
:
5s
, 
2014
(
suppl; abstr 4525
).
30.
Nakanishi
Y
,
Akiyama
N
,
Tsukaguchi
T
,
Fujii
T
,
Sakata
K
,
Sase
H
, et al
The fibroblast growth factor receptor genetic status as a potential predictor of the sensitivity to CH5183284/Debio 1347, a novel selective FGFR inhibitor
.
Mol Cancer Ther
2014
;
13
:
2547
58
.
31.
Liu
J
,
McCleland
M
,
Stawiski
EW
,
Gnad
F
,
Mayba
O
,
Haverty
PM
, et al
Integrated exome and transcriptome sequencing reveals ZAK isoform usage in gastric cancer
.
Nat Commun
2014
;
5
:
3830
.
32.
Seshagiri
S
,
Stawiski
EW
,
Durinck
S
,
Modrusan
Z
,
Storm
EE
,
Conboy
CB
, et al
Recurrent R-spondin fusions in colon cancer
.
Nature
2012
;
488
:
660
4
.
33.
Nakanishi
Y
,
Akiyama
N
,
Tsukaguchi
T
,
Fujii
T
,
Satoh
Y
,
Ishii
N
, et al
Mechanism of oncogenic signal activation by the novel fusion kinase FGFR3-BAIAP2L1
.
Mol Cancer Ther
2015
;
14
:
704
12
.
34.
Tanaka
H
,
Yoshida
M
,
Tanimura
H
,
Fujii
T
,
Sakata
K
,
Tachibana
Y
, et al
The selective class I PI3K inhibitor CH5132799 targets human cancers harboring oncogenic PIK3CA mutations
.
Clin Cancer Res
2011
;
17
:
3272
81
.
35.
Frazzetto
M
,
Suphioglu
C
,
Zhu
J
,
Schmidt-Kittler
O
,
Jennings
IG
,
Cranmer
SL
, et al
Dissecting isoform selectivity of PI3K inhibitors: the role of non-conserved residues in the catalytic pocket
.
Biochem J
2008
;
414
:
383
90
.
36.
Nylander
S
,
Kull
B
,
Bjorkman
JA
,
Ulvinge
JC
,
Oakes
N
,
Emanuelsson
BM
, et al
Human target validation of phosphoinositide 3-kinase (PI3K)beta: effects on platelets and insulin sensitivity, using AZD6482 a novel PI3Kbeta inhibitor
.
J Thromb Haemost
2012
;
10
:
2127
36
.
37.
Weigelt
B
,
Warne
PH
,
Lambros
MB
,
Reis-Filho
JS
,
Downward
J
. 
PI3K pathway dependencies in endometrioid endometrial cancer cell lines
.
Clin Cancer Res
2013
;
19
:
3533
44
.
38.
Cerami
E
,
Gao
J
,
Dogrusoz
U
,
Gross
BE
,
Sumer
SO
,
Aksoy
BA
, et al
The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data
.
Cancer Discov
2012
;
2
:
401
4
.
39.
Gao
J
,
Aksoy
BA
,
Dogrusoz
U
,
Dresdner
G
,
Gross
B
,
Sumer
SO
, et al
Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal
.
Sci Signal
2013
;
6
:
pl1
.
40.
Pao
W
,
Miller
VA
,
Politi
KA
,
Riely
GJ
,
Somwar
R
,
Zakowski
MF
, et al
Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain
.
PLoS Med
2005
;
2
:
e73
.
41.
Wu
JY
,
Yu
CJ
,
Chang
YC
,
Yang
CH
,
Shih
JY
,
Yang
PC
. 
Effectiveness of tyrosine kinase inhibitors on “uncommon” epidermal growth factor receptor mutations of unknown clinical significance in non-small cell lung cancer
.
Clin Cancer Res
2011
;
17
:
3812
21
.
42.
Robinson
DR
,
Wu
YM
,
Vats
P
,
Su
F
,
Lonigro
RJ
,
Cao
X
, et al
Activating ESR1 mutations in hormone-resistant metastatic breast cancer
.
Nat Genet
2013
;
45
:
1446
51
.
43.
Toy
W
,
Shen
Y
,
Won
H
,
Green
B
,
Sakr
RA
,
Will
M
, et al
ESR1 ligand-binding domain mutations in hormone-resistant breast cancer
.
Nat Genet
2013
;
45
:
1439
45
.
44.
Burke
JE
,
Perisic
O
,
Masson
GR
,
Vadas
O
,
Williams
RL
. 
Oncogenic mutations mimic and enhance dynamic events in the natural activation of phosphoinositide 3-kinase p110alpha (PIK3CA)
.
Proc Natl Acad Sci U S A
2012
;
109
:
15259
64
.
45.
Burke
JE
,
Williams
RL
. 
Synergy in activating class I PI3Ks
.
Trends Biochem Sci
2015
;
40
:
88
100
.
46.
Zhang
X
,
Vadas
O
,
Perisic
O
,
Anderson
KE
,
Clark
J
,
Hawkins
PT
, et al
Structure of lipid kinase p110beta/p85beta elucidates an unusual SH2-domain-mediated inhibitory mechanism
.
Mol Cell
2011
;
41
:
567
78
.
47.
Pazarentzos
E
,
Giannikopoulos
P
,
Hrustanovic
G
,
St John
J
,
Olivas
VR
,
Gubens
MA
, et al
Oncogenic activation of the PI3-kinase p110beta isoform via the tumor-derived PIK3Cbeta kinase domain mutation
.
Oncogene
2015 May 18. [Epub ahead of print]
.
48.
Edgar
KA
,
Wallin
JJ
,
Berry
M
,
Lee
LB
,
Prior
WW
,
Sampath
D
, et al
Isoform-specific phosphoinositide 3-kinase inhibitors exert distinct effects in solid tumors
.
Cancer Res
2010
;
70
:
1164
72
.
49.
Hatzivassiliou
G
,
Liu
B
,
O'Brien
C
,
Spoerke
JM
,
Hoeflich
KP
,
Haverty
PM
, et al
ERK inhibition overcomes acquired resistance to MEK inhibitors
.
Mol Cancer Ther
2012
;
11
:
1143
54
.

Supplementary data