PIK3CA mutations enable targeting of a breast tumor dependency through mTOR-mediated MCL-1 translation

Abstract

Therapies that efficiently induce apoptosis are likely to be required for durable clinical responses in patients with solid tumors. Using a pharmacological screening approach, we discovered that combined inhibition of B cell lymphoma-extra large (BCL-X_L) and the mammalian target of rapamycin (mTOR)/4E-BP axis results in selective and synergistic induction of apoptosis in cellular and animal models of PIK3CA mutant breast cancers, including triple-negative tumors. Mechanistically, inhibition of mTOR/4E-BP suppresses myeloid cell leukemia-1 (MCL-1) protein translation only in PIK3CA mutant tumors, creating a synthetic dependence on BCL-X_L This dual dependence on BCL-X_L and MCL-1, but not on BCL-2, appears to be a fundamental property of diverse breast cancer cell lines, xenografts, and patient-derived tumors that is independent of the molecular subtype or PIK3CA mutational status. Furthermore, this dependence distinguishes breast cancers from normal breast epithelial cells, which are neither primed for apoptosis nor dependent on BCL-X_L/MCL-1, suggesting a potential therapeutic window. By tilting the balance of pro- to antiapoptotic signals in the mitochondria, dual inhibition of MCL-1 and BCL-X_L also sensitizes breast cancer cells to standard-of-care cytotoxic and targeted chemotherapies. Together, these results suggest that patients with PIK3CA mutant breast cancers may benefit from combined treatment with inhibitors of BCL-X_L and the mTOR/4E-BP axis, whereas alternative methods of inhibiting MCL-1 and BCL-X_L may be effective in tumors lacking PIK3CA mutations.

Figure 1. Inhibition of the MAPK and PI3K pathways can sensitize solid tumors to BH3 mimetics in certain tissues

A,B Solid tumors are insensitive to single agent BH3 mimetics; dashed line indicates a sensitivity threshold of 1 µM that was defined based on evidence that complete BCL-2/BCL-X_L inhibition is achieved at this dose (20). A, Short-term (3 day) growth inhibition (GI50) assay testing 43 cancer cell lines from 10 different tissue types using a dual BCL-2/BCL-X_L inhibitor, ABT737. B, Cancer Cell Line Encyclopedia (CCLE) data of 660 cancer cell lines treated in a GI50 assay with a dual BCL-2/BCL-X_L inhibitor, ABT263. Black bars represent solid tumor lines and red bars represent blood cancers. C, Synergy score between ABT737 and VX-11e, an ERK inhibitor, in the cell line panel from part A. Synergy score is defined as the ratio of the ABT737 GI50 value in the absence of VX-11e to the same value in the presence of VX-11e. Additive effects result in a synergy score of zero, whereas synergistic effects and antagonistic effects yield scores greater than or less than zero, respectively. The GI50 graph to the right features the top scoring cell line (WiDR) treated with ABT737 in the presence or absence of VX-11e. D, Synergy score between ABT737 and BEZ235, a PI3K/mTOR inhibitor, in the cell line panel from part A. The GI50 graph to the right features the top scoring cell line (MDA-MB-453) treated with ABT737 in the presence or absence of BEZ235. For experiments in all panels (except B), data are n= 3.

Figure 2. BCL-X_L is the relevant target of ABT737 in colorectal cancer (CRC) and breast cancer (BC)

A,B The contributions of BCL-2 and BCL-X_L to efficacy in CRC cells. A, GI50 value of a selective BCL-2 inhibitor, ABT199, in four CRC cell lines treated with either vehicle (DMSO) or an ERK inhibitor (VX-11e). B, GI50 value of a selective BCL-X_L inhibitor, WEHI-539, in four CRC cell lines treated with either vehicle (DMSO) or an ERK inhibitor (VX-11e). GI50 graphs below A and B are representative cell lines (CRC119, COLO205) showing the GI50 shifts, or lack thereof, with combination treatment. C,D The contributions of BCL-2 and BCL-X_L to efficacy in BC cells. C, GI50 value of a selective BCL-2 inhibitor, ABT199, in three BC cell lines treated with either vehicle (DMSO) or a PI3K/mTOR inhibitor (BEZ235). D, GI50 value of a selective BCL-X_L inhibitor, WEHI-539, in three BC cell lines treated with either vehicle (DMSO) or a PI3K/mTOR inhibitor (BEZ235). GI50 graphs below are representative cell lines (BT-20, MDA-MB-453, and MCF7) showing the GI50 shifts, or lack thereof, with combination treatment. For all panels, error bars show data ± SEM. For experiments in all panels, data are n = 3.

Figure 3. Breast cancer cell lines with PIK3CA mutations respond to combined inhibition of PI3K/mTOR and BCL-X_L

A, Synergy score between WEHI-539 (BCL-X_L inhibitor) and BEZ235 (PI3K/mTOR inhibitor) in a panel of BC cell lines segregated by PIK3CA mutational status. Synergy score is defined as the ratio of the WEHI-539 GI50 value in the absence of BEZ235 to the same value in the presence of BEZ235. Additive effects result in a synergy score of zero, whereas synergistic effects and antagonistic effects yield scores greater than or less than zero, respectively. The GI50 graphs below are representative cell lines (MDA-MB-361, AU565). The images below the graphs are 2D growth assays in two representative cell lines (MCF7, BT-474) treated with a BCL-X_L inhibitor (WEHI-539) in combination with a PI3K/mTOR inhibitor (BEZ235). B, GI50 value of cell lines treated with a BCL-X_L inhibitor, WEHI-539, in the presence of a PI3K/mTOR inhibitor, BEZ235, stratified by PIK3CA mutation status. C, Apoptosis measurements for BCL-X_L and PI3K/mTOR dual inhibition reported as the percentage of cells that were annexin V+ after 48 h drug treatment. BT-20 and SKBR3 are representative PIK3CA mutant and wild-type cell lines, respectively. D, BT-20 orthotopic xenografts treated with vehicle, ABT737, BEZ235, or the combination for 21 days, shown as tumor size versus time. n = 11 for all arms except BEZ235 where n = 10. Data were analyzed using 2-way ANOVA of matched values followed by Bonferroni multiple comparisons to establish significance (p<0.05) of the difference between groups at each day of treatment. E, Waterfall plot for BT-20 orthotopic xenografts calculated as percentage change in tumor volume as compared to tumor size at the initiation of treatment. For all panels, error bars show data ± SEM. For experiments in all panels (except D and E), data are n = 3.

Figure 4. PIK3CA mutant breast cancers depend on BCL-X_L and MCL-1 for survival

A, BH3 profiling of five PIK3CA mutant breast cancer cell lines. B, Immunoblot of PIK3CA mutant and wild-type cell lines treated with vehicle (DMSO), single agent PI3K/mTOR inhibitor (BEZ235), single agent BCL-X_L inhibitor (WEHI-539), or the combination. CAL51, MDA-MB-453, AU565, and MDA-MB-436 are representative cell lines. P-AKT is at the S473 site and P-S6 is at the S235/236 site. The loading control is vinculin for all cell lines. Doses are indicated in the key. Images are cropped for clarity. C, GI50 value of a BCL-X_L inhibitor, WEHI-539, in cells with or without PI3K/mTOR inhibitor (BEZ235) or cells with MCL-1 knockdown without the PI3K/mTOR inhibitor. T47D and BT-20 are representative PIK3CA mutant cell lines. D, Apoptosis measurements in control cells or MCL-1 knockdown cells treated with vehicle (DMSO), single agent inhibitor PI3K/mTOR inhibitor (BEZ235), single agent BCL-X_L inhibitor (WEHI-539), or the combination. T47D is a representative cell line. E, BH3 profiling in a cell line treated with a PI3K/mTOR inhibitor (BEZ235). F, BH3 profiling in a cell line with MCL-1 knockdown. G, BH3 profiling in a cell line with BCL-X_L knockdown. H, GI50 curve in a PIK3CA mutant cell line treated with a BCL-X_L inhibitor (WEHI-539) and either vehicle (DMSO) or an MCL-1 inhibitor (A-1210477). CAL51 is a representative cell line. I, Apoptosis measurements in PIK3CA mutant cell lines treated with vehicle (DMSO), an MCL-1 inhibitor (A-1210477), a BCL-X_L inhibitor (WEHI-539), or the combination. BT-20 and CAL51 are representative cell lines. For all panels (excluding immunoblots), error bars show data ± SEM. For all experiments, data are n = 3.

Figure 5. MCL-1 translation is controlled by the mTOR/4E-BP axis in PIK3CA mutant tumors

A, GI50 value of a selective BCL-X_L inhibitor, WEHI-539, in three BC cell lines treated with either vehicle (DMSO), a dual PI3K/mTOR inhibitor (BEZ235), a pan-PI3K inhibitor (BKM-120), a p110α isoform specific PI3K inhibitor (BYL-719), a partial allosteric mTORC1 inhibitor (rapamycin), and three mTORC1/2 inhibitors (Torin1, AZD2014, MLN-0128). B, Immunoblot of MCL-1 in a PIK3CA mutant cell line treated with DMSO or an inhibitor of cap-dependent translation, 4EGI-1. Images are cropped for clarity. C, Immunoblot of P-4E-BP1 and total 4E-BP1 in four cell lines treated with vehicle or an mTORC1/2 inhibitor (MLN-0128). Images are cropped for clarity. D, Immunoblot of 4EBP1, eIF4E, and eIF4GI in total cell lysates or m7-GTP pulldown in a PIK3CA mutant cell line treated with vehicle (DMSO) or an mTORC1/2 inhibitor (MLN-0128). Images are cropped for clarity. E, Apoptosis measurements in a PIK3CA mutant cell line treated with vehicle (DMSO), one of two mTORC1/2 inhibitors (MLN-0128 or AZD2014), a BCL-X_L inhibitor (WEHI-539), or the combination of WEHI-539 with either MLN-0128 or AZD2014. F, Immunoblot of MCL-1 in a PIK3CA mutant cell line treated with vehicle, two doses of an mTORC1/2 inhibitor (MLN-0128), a BCL-X_L inhibitor (WEHI-539), or the combination of MLN-0128 and WEHI-539 (2 µM). Images are cropped for clarity. G, BT-20 orthotopic xenografts treated with vehicle (n = 7), ABT737 (n = 7), MLN-0128 (n = 9), or the combination (n = 8) for 21 days, shown as tumor size versus time. Data were analyzed using 2-way ANOVA of matched values followed by Bonferroni multiple comparisons to establish significance (p<0.05) of the difference between groups at each day of treatment. For all panels (except immunoblots), error bars show data ± SEM. For experiments in all panels, except G, data are n = 3.

Figure 6. PIK3CA wild-type breast cancers respond to a BCL-X_L inhibitor in combination with a direct MCL-1 inhibitor

A, BH3 profiling of three PIK3CA wild-type breast cancer cell lines. B, GI50 value of a BCL-X_L inhibitor, WEHI-539, in control cells or cells with MCL-1 knockdown. C, Apoptosis measurements by annexin V binding in control cells or MCL-1 knockdown cells treated with vehicle (DMSO) or single agent BCL-X_L inhibitor (WEHI-539). HCC1143 and HCC1395 are representative cell lines. D, BH3 profiling in a cell line with MCL-1 knockdown. EBH3 profiling in a cell line with BCL-X_L knockdown. F, GI50 curves in a PIK3CA wild-type cell line treated with a BCL-X_L inhibitor (WEHI-539) and either vehicle (DMSO) or an MCL-1 inhibitor (A-1210477). HCC1143 is a representative cell line. G, Apoptosis measurements in PIK3CA wild-type cell lines treated with vehicle (DMSO), an MCL-1 inhibitor (A-1210477), a BCL-X_L inhibitor (WEHI-539), or the combination. HCC1143 and SKBR3 are representative cell lines. For panels A-G, error bars show data ± SEM. For experiments in panels A-G, data are n = 3. H, MDA-MB-436 orthotopic xenografts stably expressing Cas9 along with either sgCtrl or sgMCL-1 and treated with vehicle (sgCtrl n = 6, sgMCL-1 n = 7) or ABT737 (sgCtrl n = 5, sgMCL-1 n = 6) for 21 days, shown as tumor size versus time. Data were analyzed using 2-way ANOVA of matched values followed by Bonferroni multiple comparisons to establish significance (p<0.05) of the difference between groups at each day of treatment. To the right is an immunoblot confirming MCL-1 knockdown in the sgMCL-1 cells before the start of the xenograft experiment. Images are cropped for clarity. I, Immunoblot of P-4E-BP1 (S65) and total 4E-BP1 in two cell lines treated with vehicle or an mTORC1/2 inhibitor (MLN-0128). Images are cropped for clarity. J, Immunoblot of MCL-1 in a PIK3CA wild-type cell line treated with DMSO or an inhibitor of cap-dependent translation, 4EGI-1. Images are cropped for clarity. K, Immunoblot of 4EBP1, eIF4E, or eIF4GI in total cell lysates or m7-GTP pulldown in a PIK3CA wild-type cell line treated with vehicle (DMSO) or an mTORC1/2 inhibitor (MLN-0128). Images are cropped for clarity.

Figure 7. A therapeutic window of combined BCL-X_L and MCL-1 (direct or indirect) inhibition suggests translational potential

A, BH3 profiling of four primary patient samples. B, BH3 profiling of two normal breast epithelial cell lines (HME and HMLE). C, GI50 value for a BCL-X_L inhibitor (WEHI-539) in combination with either vehicle (DMSO) or a PI3K/mTOR inhibitor (BEZ235) in HME cells. D, Apoptosis measurements by annexin V binding in HME cells treated with vehicle (DMSO), a PI3K/mTOR inhibitor (BEZ235), a BCL-X_L inhibitor (WEHI-539), or the combination. E, Log₂ fold change in GI50 to WEHI-539 in platelets isolated from two independent donors and in two PIK3CA mutant breast cancer cell lines treated with vehicle, BEZ235, or MLN-0128. F, Apoptosis measurements in a PIK3CA wild-type cell line (HCC1143) treated with vehicle (DMSO), an MCL-1 inhibitor (A-1210477) and/or a BCL-X_L inhibitor (WEHI-539), doxorubicin, etoposide, or the combination of A-1210477 plus WEHI-539 and doxorubicin or etoposide. G, Apoptosis measurements in a PIK3CA mutant cell line (CAL51) treated with vehicle (DMSO), an MCL-1 inhibitor (A-1210477) and/or a BCL-X_L inhibitor (WEHI-539), doxorubicin, etoposide, or the combination of A-1210477 plus WEHI-539 and doxorubicin or etoposide. H, Apoptosis measurements in a PIK3CA mutant cell line (CAL51) treated with vehicle (DMSO), a PI3K/mTOR inhibitor (BEZ235) and/or a BCL-X_L inhibitor (WEHI-539), doxorubicin, etoposide, or the combination of the PI3K/mTOR plus BCL-X_L inhibitors with either doxorubicin or etoposide. I, Apoptosis measurements in a PIK3CA wild-type cell line (SKBR3) treated with vehicle (DMSO), an MCL-1 inhibitor (A-1210477) and/or a BCL-X_L inhibitor (WEHI-539), lapatinib, or the combination of the MCL-1 and BCL-X_L inhibitors with lapatinib. For all panels, error bars show data ± SEM. For experiments in all panels, data are in n = 3.