Repurposing of mTOR Complex Inhibitors Attenuates MCL-1 and Sensitizes to PARP Inhibition

High levels of prosurvival BCL-2 family proteins cancer causes resistance to treatment (1–4), which makes it hard for the current therapeutic drugs to kill the cancer cells. MCL-1 is a member of the prosurvival BCL-2 proteins, which is overexpressed in both hematologic and solid tumor cancers (5, 6). The presence of MCL-1 in outer mitochondrial membrane inhibits the progression of apoptosis by binding proapoptotic BCL-2 proteins BAK and BAX. These proteins induce pore formation in the mitochondrial membrane and which in turn release executioner caspases 3 and 7, responsible for the majority of the macromolecular degradation observed during apoptosis (3). Abrogation of BAK and BAX polymerization by high expression of MCL-1 can be overcome by either MCL-1 depletion or by inhibiting the MCL-1–binding sites on BAK/BAX with BH3 mimetics (7, 8). In addition to their classic role in regulating the intrinsic pathway of apoptosis, MCL-1 has been shown to be involved in DNA damage repair (9–11). Our recent studies provide an insight into how MCL-1 affects double strand breaks (DSBs) by demonstrating that depleting MCL-1 by siRNA impairs DSB repair by homologous recombination (HR) repair switches DNA repair pathway toward nonhomologous end joining (NHEJ) and inhibits the resolution of stalled replication forks (12). Our study was corroborated by a recent report (13) demonstrating that MCL-1 acts as a functional switch in selection between HR and NHEJ pathways. They show that MCL-1 directly interacts with the dimeric Ku protein complex via its BH1 and BH3 domains and inhibits Ku-mediated NHEJ.

DNA repair pathways help tumor cells to survive DNA damage that is induced by ionizing radiation (IR) and S-phase–specific drugs. Cancer cells with defective DNA repair pathways are sensitive to S-phase–specific drugs (14, 15). PARP inhibitors induce synthetic lethality in tumors with defective HR (16, 17). We postulated that repurposing mTORC inhibitors would decrease phosphorylation of 4EBP1, inhibit cap-dependent translation and hence protein synthesis in cancer cells. Because MCL-1 has a short half-life, inhibition of protein synthesis would deplete MCL-1, suppress HR repair and hence sensitize these HR-deficient cancer cells to a PARP inhibitor in in vitro and in mice. To test the specificity of mTORC inhibitors to kill cancer cells through depletion of MCL-1, we reintroduced MCL-1 in these cells. Further, we show that everolimus depletes MCL-1 in colorectal carcinoma and SCLC but is unable to deplete MCL-1 in glioblastoma multiforme (GBM) with PTEN mutations, which have constitutive phosphorylation of PI3K–mTOR/AKT pathways (18). However, a dual mTORC 1 and 2 inhibitor AZD2014 depletes MCL-1 in these PTEN-mutant GBMs. We also show that combination of everolimus or AZD2014 with olaparib inhibits the growth of colorectal carcinoma and a GBM xenograft in in vivo and decreases clonogenic survival in in vitro compared with individual drugs.

Clone A is a subclone of the DLD-1 or HCT-15 cell line (3, 19, 20). NCI-H196 is an SCLC cell line obtained from ATCC and used in one of our previous studies (7). SF-539 and SNB-75 are GBM cell lines obtained in 2017 from DCTD Tumor repository (NCI Frederick). U87-MG was purchased from ATCC in October 2017. The cells were replenished from frozen stocks after every 10 passages. The cell lines were tested for Mycoplasma (December 2017) and authenticated (May 2018) at IDEXX BioResearch. All the cell lines were maintained and transfected with plasmids as described previously (3, 7, 12). Neuro-Cult NS-A Proliferation medium with supplements was purchased from Stem Cell Technologies. MCL-1–overexpression plasmid pTOPO-MCL-1 was purchased from Addgene. Everolimus, AZD2014, and olaparib were purchased from either Selleckchem or Medchem Express LLC and reconstituted in DMSO at 10 mmol/L stocks. Antibodies were purchased as follows: from Cell Signaling Technology, Chk1 (2360), pChk1 (Ser317; 12302S), ATR (2790S), Phospho-AKT(Ser473)(9271S), AKT antibody (9272S), Phospho-RIP(Ser166)(65746S), Phospho-4E-BP1(Th37/46)(2855S), 4EBP1(9452S), p-ATR (Ser428); 53BP1(4937S), BCL-2 (2872S) from Santa Cruz, Biotechnology MCL-1 (sc819); from Abcam, BRCA1(ab16781), MCL-1 (ab197529), RPA-70(ab79398), and Rad51 (ab63801) antibodies; HP-conjugated GAPDH antibody was purchased from Proteintech (HP-60004). The 4%–20% Mini-Protean gradient TGX Precast Gels were obtained from Bio-Rad. The blots were developed by chemiluminescence using a Syngene G:Box Chemi XT4 imager.

Cell culture in chamber slides, fixation, and immunostaining were done as previously described (12, 21). Cancer cells were pretreated with everolimus (2 μmol/L) or AZD2014 (1 μmol/L) for 16 hours followed by treatment with olaparib (10 μmol/L) for 24 hours. After 24 hours of treatment, olaparib was removed, the cells were washed with 1× PBS, and fresh media were added to the cells. Immunostaining was done either immediately 0 or 5 hours after olaparib release (22). The results shown are from three independent experiments.

The Institutional Animal Care and Use Committee of George Mason University approved the protocols used in the study (GMU IACUC protocol #1708). Following the NIH guidelines of sex as a biological variable, we used a male and a female mouse model to establish the in vivo efficiency of the drug combination. Male and Female Immunodeficient Athymic nude mice Crl:NU(NCr)-Foxn1^nu (homozygous) for the study were purchased from Charles River Laboratories. Clone A cells (3.5 × 10⁶ million) were first injected in the right flank of each male athymic nude mice subcutaneously, to generate a colorectal carcinoma xenograft. The tumors were measured every 2 days, and when the tumor reached a volume of 100 mm³ at day 8, the mice were divided into five groups, with 5 mice in each group: vehicle (everolimus), vehicle (olaparib), everolimus, olaparib, and everolimus + olaparib. The drugs were given intraperitoneally between days (8–17) with everolimus (30 mg/kg) administered first followed by olaparib (50 mg/kg) after 16 hours. Similarly, around 3 × 10⁶ U87-MG cells were injected in the right flank of each female athymic nude, and following the same treatment protocol as described for everolimus, the drugs were administered between days (13–22) intraperitoneally. Mice were sacrificed when the control tumor volume reached 2,000 mm³. Tumor volumes were calculated by the formula: perpendicular length × width². Everolimus was dissolved in 30% propylene glycol (dissolved first) and 5% Tween-80 whereas olaparib and AZD2014 were both dissolved in 4% DMSO and 30% PEG 300. Statistical analysis of the treated tumors relative to control was done using two-way ANOVA, with Sidak multiple-comparisons correction test in GraphPad Prism 7.

The clonogenic survival assay was done as described earlier (12, 21, 22). For the cell viability assays, WST-1 (Sigma-Aldrich) reagent was used according to the directions supplied by the manufacturer. For the 3D cell viability assay, 2,500 cells were seeded in Neurocult media with supplements in 96-well ultralow attachment plates (Corning Inc.) in triplicates as described earlier (19). Cell viability for this assay was measured using alamar blue (Thermo Fisher Scientific) following the directions of the manufacturer.

ANOVA was performed for statistical analysis of multiple comparisons using GraphPad Prism 7. Data in graphs are presented as mean ± SD except where indicated in the text. For the analyses, P < 0.05 was considered to be statistically significant. All experiments were repeated at least twice independently.

Earlier, we have shown that depletion of MCL-1 by siRNA sensitizes cancer cells to IR and the S-phase–specific drug hydroxyurea (HU) by suppressing HR repair (12). In order to deplete MCL-1 in in vivo, targeting MCL-1 with siRNA has its limitations because clinically siRNA has challenges in delivering to cancer cells. The second possibility for targeting MCL-1 would be to use the inhibitors or BH3 mimetics, which can target the MCL-1–binding sites on BAK/BAX and activate apoptosis (8). In order to test whether MCL-1 inhibitors can mimic siRNA MCL-1 and suppress the DNA repair activity in cancer cells, we tested S63845—a highly specific inhibitor of MCL-1 (8)—on clone A cells. Clone A cells were pretreated with DMSO or 5 μmol/L of S63845 for 24 hours and then exposed to 2 mmol/L HU for 24 hours. Lysates were collected 4 hours after release of HU, and Chk1-Ser-317 phosphorylation levels were compared between DMSO pretreated and S63845 pretreated clone A cells in the presence and absence of HU treatment (Fig. 1A). Depletion of MCL-1 with siRNA has been shown to decrease phosphorylation of Chk1, after HU, IR, or etoposide treatment (11, 12), but pretreatment with S63845 fails to change the phosphorylation levels of Chk1 (Fig. 1A), which suggests that S63845 is unable to suppress the DNA repair activity of MCL-1.

The levels of MCL-1 in the cells are regulated by different protein kinases at transcriptional, translational, and posttranslational levels (23–27). A third approach to deplete MCL-1 protein levels would be to repurpose protein kinase inhibitors, which have been either approved or in the pipeline for cancer treatment. Inhibition of mTORC1 results in loss of 4EBP1 phosphorylation, inhibition of cap-dependent phosphorylation, and depletion of MCL-1 in cancer cells (7, 24, 28, 29). Clone A and NCI-H196 cells were treated with everolimus (1 μmol/L and 5 μmol/L) for 16 hours to assess whether everolimus depleted MCL-1. Treatment resulted in the depletion of MCL-1 by 3- to 4-fold relative to the DMSO-treated control cells (Fig. 1B). Depletion of MCL-1 led to the death of clone A and NCI-H196 cells, while ectopic expression of MCL-1 in everolimus (2 μmol/L) pretreated clone A and NCI-H196 cells (Fig. 1C) rescued the survival of these MCL-1–depleted cells. This confirmed that MCL-1 plays a critical role in the cytotoxicity mediated by everolimus in colorectal carcinoma and SCLC cells (Fig. 1D). In order to elucidate the kinetics of MCL-1 depletion by everolimus, clone A cells were treated with everolimus, and the whole-cell lysates were collected at 0, 8, 16, and 32 hours after drug treatment and probed for MCL-1 and BCL-2 protein levels (Fig. 1E). Everolimus depleted MCL-1 protein levels in clone A cells only at 16 hours after treatment, whereas levels of BCL-2 proteins are stable even at 32 hours after everolimus treatment.

The ability of everolimus to deplete MCL-1 was further investigated in the GBM cell lines SNB-75 and SF-539. Everolimus and its paralog rapamycin are unable to deplete MCL-1 in these GBM cell lines even at 10 μmol/L concentration (Fig. 1F). GBM cell lines SF-539, SNB-75, and U87-MG all have PTEN mutations characterized by constitutive PI3K–AKT–mTOR pathway activation (18, 30). Because mTORC1 and mTORC2 have been reported to cross-talk (31), the GBM cells have hyperphosphorylated levels of mTOR and AKT proteins, treatment with mTORC1 inhibitor such as everolimus may not be sufficient to inhibit cap-dependent translation (Fig. 2A). In order to target both mTORC1 and mTORC2 in GBM cells, we treated SNB-75, SF-539, and U87-MG cell lines with the dual mTOR complex inhibitor—1 μmol/L of AZD2014 for 16 hours and the whole-cell lysates were probed for MCL-1 protein levels. AZD2014 efficiently depleted MCL-1 in all three cell lines (Fig. 2B). Ectopic expression of MCL-1 in AZD2014 pretreated U87-MG cells rescued their survival, suggesting that MCL-1 plays a critical role in the cytotoxicity mediated by AZD2014 in GBM cells (Fig. 2C). The kinetics of MCL-1 depletion by AZD2014 was investigated next. MCL-1 protein levels decreased as early as 4 hours, and by 8 hours, the levels were less than a fourth of the MCL-1 level in the controls (Fig. 2D). However, the levels of BCl-2 protein remain unchanged, further supporting that mTORC inhibition affects MCL-1 and fails to change the level of other prosurvival proteins. Thus, dual mTORC inhibition may be faster than single mTORC1 inhibition in depletion of MCL-1 in clone A cells (Fig. 1E).

To elucidate the difference between the two mTORC inhibitors at the cell signaling levels, clone A and SF-539 cells were treated with everolimus (2 μmol/L) and AZD2014 (1 μmol/L) for 16 hours and lysates probed with AKT-Ser-473 (substrate of mTORC2) and 4EBP1-Thr 37/46 (substrate of mTORC1). Everolimus decreases 4EBP1 phosphorylation only in clone A cells (Fig. 2E, left, lane 2), whereas AZD2014 decreases 4EBP-1 phosphorylation in both clone A (Fig. 2E, left, lane 3) and SF-539 cells (Fig. 2E, right, lane 3). Phosphorylation of AKT-Ser-473 decreased in both clone A (Fig. 2E, left, lane 3) and SF-539 cells after AZD2014 (Fig. 2E, right, lane 3) treatment. However, because AKT-Ser-473 is a substrate of mTORC2, mTOR1-specific everolimus has no effect on AKT phosphorylation (Fig. 2E, left lane 2).

Targeting the MCL-1 protein levels in cancer cells results in the defective HR pathway (12, 13). Cancer cells, which are defective in HR repair, have been shown to be sensitive to PARP inhibitors (32). To determine whether depleting MCL-1 levels in cancer cells by treatment with mTORC inhibitor sensitized the cells to FDA-approved PARP inhibitor, clonogenic survival assay was done (Fig. 3A). Pretreatment of cancer cells with everolimus (2 μmol/L) or AZD2014 (1 μmol/L) decreased the clonogenic survival of these cancer cells when treated with 2 to 16 μmol/L olaparib by 2- to 4-fold (Fig. 3A). Further, to confirm the enhanced killing by the drug combination in the cancer cells, we used a 3D cell viability assay that more closely mimics tumor growth in in vivo. Cancer cells were pretreated with everolimus or AZD2014 for 16 hours followed by treatment with a variable dose of olaparib (1–16 μmol/L) and cell viability measured 96 hours after olaparib treatment alamarBlue assay. The combination of mTORC inhibitors (everolimus or AZD2014) with olaparib enhanced the cell killing of all the three cancer cells (clone A, NCI-H196, and U87-MG; Fig. 3B). This suggested that depletion of MCL-1 by mTORC inhibitors sensitizes the cancer cells to olaparib.

MCL-1 is primarily a cytosolic protein that regulates the intrinsic or mitochondrial pathway of apoptosis (33). Because DNA repair is a nuclear event, we sought to determine whether MCL-1 is present in the nucleus. In order to test whether olaparib changes the distribution of MCL-1 in clone A cells, the cells were treated with 10 μmol/L of olaparib for 16 hours and after olaparib treatment the cells were stained to determine the intracellular location of MCL-1 (Supplementary Fig. S1). The nuclear MCL-1 levels increased 3- to 4-fold after olaparib treatment in these cells with an increase in MCL-1 foci formation after olaparib treatment (Supplementary Fig. S1). However, treatment of clone A cells with everolimus completely abolished the levels of cytosolic and nuclear MCL-1 (Supplementary Fig. S1) and as a result due to low levels functional MCL-1, the cancer cells show increased sensitivity to olaparib.

Further to elucidate that mTORC inhibitors sensitize the cancer cells to olaparib primarily through depletion of MCL-1, the protein was ectopically expressed in clone A and SF-539 cells, which were then treated with the combination of olaparib and an mTORC inhibitor. mTORC inhibitors depleted MCL-1 in control cells of both clone A and SF-539 cell lines (lane 3 of Fig. 3Cb and 3Db) but not in the cancer cells overexpressing MCL-1, where MCL-1 protein levels were similar to those in the DMSO control (compare lane 1 with lane 6 of Fig. 3Cb and 3Db). Thus, ectopic expression of MCL-1 rescued the clonogenic survival of clone A and SF-539 cells when the cells are treated with a combination of mTORC inhibitor and olaparib. This supports the concept that MCL-1 plays a critical role in the cytotoxicity mediated by combination of everolimus or AZD2014 with olaparib in cancer cells.

To elucidate the cell death pathways involved, we examined both caspase-dependent and -independent cell death pathways. We first looked at PARP cleavage to determine whether the caspase 3–mediated intrinsic pathway is activated in the combination drug treatment of clone A and SF-539 cell lines. The combination treatment of everolimus or AZD2014 with olaparib failed to induce PARP cleavage in either clone A or SF-539 cell line (Supplementary Fig. S3), suggesting that caspase 3–induced cell death is not involved in the cell death by the drug combination. Next, we looked at the cleavage of one of the major autophagic markers—light chain 3 (LC3)—to evaluate the possible role of autophagy. The cleavage of LC3BI to LC3BII was not increased in cells treated with the combination of everolimus and olaparib or AZD2014 and olaparib, respectively, suggesting that autophagy has no role to play in enhanced cell death (Supplementary Fig. S2). We then investigated the role of necroptosis by examining the phosphorylation levels of RIPK1 (RIPK1 Ser 166), a major kinase involved in regulation of necroptosis (34), in control- and drug-treated cells. RIPK1 Ser 166 phosphorylation levels were increased in the clone A and SF-539 cells treated with combination drug treatment (lanes 4 and 8 in Fig. 3E). This indicated that pretreatment with an mTORC inhibitor sensitizes cancer cells to olaparib treatment by inducing necroptosis.

PARP inhibition causes PARP-1 trapping onto DNA repair intermediates, inducing replication stress, collapse of replication forks, and formation of DSBs (35). To determine whether the increase in sensitivity of mTORC inhibitor–pretreated cancer cells to olaparib was due to an increase in formation of DNA DSBs and hence increased DNA damage, we examined whether pretreatment with mTORC inhibitor affected the phosphorylation of H2AX at serine 139 (γ-H2AX), a surrogate marker for DNA DSBs by immunostaining for γ-H2AX foci as described earlier (36). The DMSO-treated (control) and everolimus- or AZD2014-treated cells were exposed to 10 μmol/L of olaparib. Twenty-four hours after olaparib treatment, the cells were washed with PBS, and fresh medium was added to release the cells from the PARP inhibition. Immunostaining for γ-H2AX foci was done at 0, 2, 4, and 24 hours after olaparib release. The initial level of γ-H2AX foci in mTORC-treated and MCL-1–depleted cells was similar to that in control cells at the time of the release of the drug (0 hours). Two to 4 hours after olaparib release, the control (DMSO-pretreated) cells continued to increase γ-H2AX foci significantly more than in the mTORC-pretreated cells (Fig. 4Ab and Bb). However, at 24 hours after olaparib release, the fraction of control cells with more than 10 γ-H2AX foci declined while the fraction of mTORC inhibitor pretreated exhibited higher frequency of cells with more than 10 γ-H2AX foci (Fig. 4). The sustained levels of residual γ-H2AX foci in mTORC pretreated cells indicate more DSBs, which suggests that combination of everolimus or AZD2014 with olaparib increases DNA damage in cancer cells.

PARP inhibitors stimulate replication stress and induce lethal DNA DSBs in HR-defective cells (35, 37). To determine whether increased formation of DSBs in cells treated with combination of mTORC inhibitor and olaparib was due to defective HR, we used immunostaining to compare the foci formation at critical steps of the HR repair pathway between control- and mTORC inhibitor–treated cells. The recruitment of 53BP1 at the DSBs enhances DNA repair by NHEJ and inhibits the engagement of resection proteins important for the HR repair pathway (20). To test whether depletion of MCL-1 alters the formation of 53BP1 foci at DSBs, immunostaining was done as described in Materials and Methods.

In the mTORC inhibitor–pretreated cells, there was a greater formation of 53BP1 foci than control cells at 5 hours after olaparib release, which suggests that depletion of MCL-1 by mTORC inhibitors switches the DNA repair pathway toward NHEJ (Fig. 5A and B). Because the frequency of 53BP1 foci formation is higher in mTORC inhibitor–treated cancer cells and 53BP1 decides the repair pathway selection between HR and NHEJ, we next examined whether MCL-1 depletion by everolimus or AZD2014 changes the recruitment of proteins involved in DSB repair by HR. HR involves the repair of DSBs by BRCA1 recruitment to replace 53BP1, which inhibits the MRN (Mre11-Rad50-Nbs1) and CtIP complex mediated end resection step of HR repair (38). Because we observed an increased number of 53BP1 foci in everolimus- or AZD2014-treated cells (Fig. 5A and B), we decided to test whether MCL-1 depletion affected BRCA1 foci formation. We compared the BRCA1 foci formation between control- and everolimus- or AZD2014-pretreated cells, 0 and 5 hours after olaparib release by determining the percentage of cells with >5 BRCA1 foci. Pretreatment with mTORC inhibitors decreased BRCA1 foci formation at 5 hours after olaparib release, with 40% of control cells in clone A and 50% of cells in SF-539 had greater than 5 foci per cell compared with 17% in everolimus (Fig. 5C) and 20% in AZD2014 (Fig. 5D)- pretreated cells. BRCA1 assists in the recruitment RPA and Rad51 to the sites of DNA damage. RPA is an important step in the HR pathway and is recruited to the ssDNA formed during DNA end resection and is also detected by foci in the nucleus formed 5 hours after olaparib release. Fewer cells were detected with greater than 10 RPA foci in everolimus- or AZD2014-pretreated cells (30%–40%) than was the case with control cells (60%–70%; Fig. 5E and F; Supplementary Fig. S4). Similarly, the frequency of cells with >10 Rad51 foci was also lower in everolimus- or AZD2014-pretreated cancer cells than in the control cells at both 0 and 5 hours after olaparib release (Fig. 5G and H; Supplementary Fig. S5).

The ssDNA formed during the end resection step of repair by HR also activates ATR/Chk1 kinases (39). To determine whether depletion of MCL-1 by everolimus and AZD2014 decreases the phosphorylation of the ATR/Chk1 complex after olaparib treatment, the mTORC inhibitor pretreated cells were exposed to olaparib for 24 hours, the drug was released, and whole-cell lysates were collected 5 hours later. The lysates were probed for phosphorylation of ATR-Ser 428 and Chk1-Ser 317 in DMSO-pretreated (control) and everolimus (2 μmol/L)- or AZD2014 (1 μmol/L)-pretreated cells. The levels of phosphorylation of ATR-Ser 428 and Chk1-Ser 317 are low when mTORC inhibitor–pretreated cells are exposed to olaparib (10 μmol/L; lane 4 for Fig. 5Ia and 5Ib) compared with treatment of control cells (lane 2 for Fig. 5Ia and 5Ib) with the same drug. Taken together, our results suggested that HR is defective in everolimus- or AZD2014-pretreated cells and that these MCL-1–depleted cancer cells are sensitive to olaparib treatment.

To determine whether depletion of MCL-1 by mTORC inhibitors sensitizes the cancer cells to olaparib in in vivo, xenografts were created as described in Materials and Methods. Everolimus or vehicle was administered intraperitoneally (i.p.) to deplete MCL-1 followed 16 hours later by olaparib or vehicle (Fig. 6Aa). The mice were sacrificed at day 20 when the tumors of the control groups of mice treated with vehicle alone reached around 2,000 mm³ and tumors were harvested to determine the levels of MCL-1 in control and treated mice (Fig. 6Ab). The levels of MCL-1 were low in everolimus- and combination-treated mice. The volume of tumors in mice treated with the single-agent everolimus or olaparib was not significantly reduced compared with the size of the vehicle controls (Fig. 6A). However, the growth of tumors in mice treated with combination of (everolimus and olaparib) was significantly inhibited compared with vehicles (Fig. 6Aa), and at day 20, the volume of the tumors treated with the combination drug treatment was one-tenth the volume of vehicle-treated controls (P < 0.00001; Fig. 6Aa). Supplementary Fig. S6 represents the striking difference in the tumor sizes between control- and drug-treated groups.

Further, to determine the efficacy of AZD2014 to sensitize a GBM cell line to olaparib in in vivo, a U87-MG xenograft was generated as described in Materials and Methods and followed the same treatment protocol as described for everolimus. The mice were sacrificed between days 36 and 40 when the tumors of the mice treated with vehicle, AZD2014, and olaparib reached around 2,000 mm³ (Fig. 6B). Similar to the clone A xenograft, the growth of tumors in mice treated with the drug combination (AZD2014 and olaparib) was significantly inhibited throughout the growth curve, and at day 40, the volume of the tumors treated with the combination drug treatment was one-seventh of the volume of vehicle-treated controls (P < 0.00001). The drug combination treatment in both xenograft models (colorectal carcinoma and GBM) resulted in around 10% to 15% reduction in body weight during the course of the treatment, but the mice regained body weight 3 to 4 days after the treatment period ended. These results support the postulate that depletion of MCL-1 by everolimus or AZD2014 enhances the efficacy of olaparib in a colorectal carcinoma and a GBM cancer model.

The overexpression of MCL-1 protein is one of the major causes of drug resistance in cancers, including colorectal carcinoma, SCLC, and GBMs (1, 3, 4, 7, 40). However, MCL-1 has a short half-life, and its levels have been shown to decrease after inhibition of protein synthesis (7). Here, we show that depletion of MCL-1 by everolimus or AZD2014 suppresses HR repair in cancer cells, which in turn increases DNA damage and sensitizes the cells to olaparib (Fig. 6C). Our results suggest that repurposing of mTORC inhibitors depleted MCL-1 in colorectal carcinoma, SCLC, and GBM cell lines at clinically achievable concentrations (1–2 μmol/L). Further, our xenograft studies showed that administering everolimus or AZD2014 in combination with olaparib inhibited the growth of tumors in a colorectal carcinoma and a GBM mouse model. Because the depletion of MCL-1 is important to suppress HR repair, the time between administering everolimus or AZD2014 and olaparib to patients could be critical for the success of these drugs in the clinic. For the mice xenograft studies, we administered the drugs at 16-hour time interval, which was supported by in vitro results that showed that at 16 hours, both everolimus and AZD2014 deplete MCL-1 protein levels by 3- to 5-fold. However, everolimus depleted MCL-1 in cancer cells more slowly than AZD2014, and because AZD2014 starts depleting MCL-1 in GBM cells as early as 4 hours, the combination of AZD2014 with olaparib could be administered to patients at a shorter interval than 16 hours. We show that mTORC inhibitors failed to change the protein levels of BCL-2, ATR, Chk1, AKT, and 4EBP1, which further supports the fact that inhibition of protein synthesis specifically depletes MCL-1 in cancer cells and fails to change the levels of other prosurvival proteins such as BCL-2.

A recent study has shown that MI-223, a novel BH1/BH3 domain specific inhibitor of MCL-1, blocked the binding of MCL-1 to Ku-70. This inhibition in binding lead to sensitization of a non-small cell lung cancer cell (NSCLC) line to Olaparib (13). The major drawback with the clinical future of this inhibitor is that, although specific, it is still under early stages of development, and it has to go through different stages preclinical and clinical testing before it can be used in patients. Further, the ability of MI-223 to inhibit tumor growth in combination with olaparib or other S-phase–specific drug in different cancer mouse models besides NSCLC has not been tested so far. However, repurposing of drugs such as everolimus, AZD2014, and olaparib has the clinical edge because the drugs are either FDA approved (everolimus and olaparib) or in advanced stages of clinical trials for treating different cancers (AZD2014), with the toxicity of all three drugs well established in patients with cancer. Moreover, mTORC and PARP inhibitors have been shown to pass the blood–brain barrier (41–44), so it would be interesting to test the efficacy of the drug combination in intracranial GBM models in the future.

Our results show that MCL-1 depletion is the primary reason for increased sensitivity to drug combination. Earlier, Mcl-1 has been identified as a translational regulated genetic determinant of mTORC1-dependent survival in cancer cells and the extent by which rapamycin modulates expression of Mcl-1 as an important feature of the rapamycin response (26). Further, the cap-dependent translation has been shown to be important for proficient translation of MCL-1 mRNA in various cancer cells (24, 28). Consistent with the literature, here we show that status of 4EBP1 phosphorylation, which is a marker for cap-dependent phosphorylation, determines whether everolimus or AZD2014 can deplete MCL-1 in cancer cells. So phosphorylation levels of 4EBP1-Thr 37/46 can be used as a marker to determine the drug activity in cancer cells of patients. Additionally, the mutation background of the cancer cell line could be critical in deciding whether to use an everolimus or AZD2014 for depleting MCL-1 in cancer cells. Our results show that only dual mTORC inhibitor AZD2014 can deplete MCL-1 in GBM cells which have a PTEN mutation. This means that in brain, breast cancer, and prostate tumors, where PTEN is one of the most commonly lost or downregulated genes (45), AZD2014 or a dual mTORC inhibitor would be the only agent to deplete MCL-1.

Our results also demonstrate that MCL-1 depletion by everolimus or AZD2014 suppresses HR repair in cancer cells. It has been reported that ssDNA tails formed during DNA end resection bind RPA, and this RPA-ssDNA not only recruits proteins important for repair of DNA but also activates ATR/Chk1 checkpoint kinases, which slow down cell-cycle progression and give the cell time to repair the DNA (46, 47). In the mTORC inhibitor–pretreated cells, both ATR and Chk1 phosphorylation is impaired, and the RPA and Rad51 foci recruitment is also reduced in response to treatment with olaparib, which suggests that DNA end resection is defective in these cancer cells. Earlier, Chk1 phosphorylation has been shown to be important for interaction with Rad51 and that Chk1 inhibition has been shown to result in reduced Rad51 foci formation (48). Our results showing that combination of everolimus or AZD2014 and olaparib decreased Chk1 phosphorylation and reduced Rad51 foci formation provide further support to the fact that pretreatment with mTORC inhibitors suppresses HR repair in cancer cells. The defective HR repair in turn increases the formation of DSBs as indicated by higher residual γ-H2AX foci, resulting in decreased clonogenic survival of MCL-1–depleted cells in response to olaparib. Our results clearly show that necroptosis as the major form of cell death in colorectal carcinoma and GBM cells treated with a combination of everolimus/AZD2014 and olaparib as indicated by the activation of RIPK1 Ser 166 phosphorylation. Necroptosis of cancer cells releases damage-associated molecular patterns, which can induce strong immune response and lead to activation of CD8⁺ T cells, NKT cells, and other immune cells (49). Therefore, the activation of necroptosis using the drug combination of everolimus or AZD2014 with olaparib may have the additional benefit of eliciting the antitumor immunity and improve the therapeutic intervention of cancers such as colorectal carcinoma and GBM. In summary, these results elucidate that the mTORC inhibitors everolimus and AZD2014 sensitize cancer cells to a PARP inhibitor by depletion of MCL-1. Thus, a therapeutic approach of repurposing mTORC inhibitors everolimus or AZD2014 in combination with olaparib may benefit the patients across a wide spectrum of cancers.

No potential conflicts of interest were disclosed.

Conception and design: A.R. Mattoo

Development of methodology: A.R. Mattoo, A. Joun

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.R. Mattoo, A. Joun

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.R. Mattoo, J. M. Jessup

Writing, review, and/or revision of the manuscript: A.R. Mattoo, J. M. Jessup

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.R. Mattoo

Study supervision: A.R. Mattoo

Other (oversaw Dr. A.R. Mattoo's general research): J. M. Jessup

This project was funded by generous philanthropic support from Inova Schar Cancer Institute.

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.