Mutant BRAF Upregulates MCL-1 to Confer Apoptosis Resistance that Is Reversed by MCL-1 Antagonism and Cobimetinib in Colorectal Cancer

Mutational activation of the BRAF oncogene drives tumorigenesis through constitutive activation of the MAPK signaling pathway (1). The most frequent somatic alteration in BRAF is the point mutation T1799A encoding BRAFV600E which results in a several hundred-fold increase in its kinase activity (2). BRAF^V600E is detected in approximately 10% of human colorectal cancers (CRC) where it is associated with poor responsiveness to chemotherapy (3, 4), and poor prognosis (5–9) with frequent peritoneal metastasis (10–12). The subset of CRCs with microsatellite instability (MSI) is enriched with BRAF mutations that are associated with poor outcome in the metastatic setting as is also observed in microsatellite-stable CRCs (13). In contrast to BRAF^V600E melanoma, metastatic CRCs with the same BRAF^V600E mutation have shown a lack of sensitivity to BRAF or MEK inhibitor monotherapy in early clinical trials (3). Whereas approximately 50% of patients with metastatic melanoma responded to BRAF inhibition by vemurafenib, the response rate was only 5% among patients with BRAF^V600E metastatic CRC (3, 14, 15). The inability of vemurafenib to kill colon cancer cells was shown to be related to rebound EGFR activation and the addition of inhibitors of EGFR signaling suppressed MEK/ERK activation and induced a synergistic apoptosis (16, 17). These data suggested that apoptosis resistance can be overcome by blockade of reactivated MAPK signaling and in this regard, combined BRAF and MEK inhibition led to further suppression of MAPK signaling and increased therapeutic efficacy (17, 18). Potent inhibition of active, phosphorylated MEK is required for strong inhibition of the MAPK pathway in BRAF^V600E tumors, and was achieved using cobimetinib which resulted in superior efficacy in BRAF^V600E versus KRAS mutant cancer cells (1). Among patients with BRAF^V600E-mutant metastatic melanoma, the addition of cobimetinib to vemurafenib was associated with a significant improvement in progression-free survival in a phase III clinical trial that led to its approval by the FDA (19). In contrast to results in patients with melanoma, however, treatment of patients with metastatic CRC with the combination of the BRAF inhibitor dabrafenib and the MEK inhibitor trametinib showed only a 12% response rate including one complete response (4). These data underscore the need to further elucidate the mechanism of BRAF^V600E-mediated drug resistance and, thereby, improve the responsiveness of these cancers using novel therapeutic approaches.

The MAPK pathway links extracellular signals to the machinery that controls cellular processes including apoptosis (20). Existing data indicate a continued dependency on BRAF signaling for cancer cell survival, although the mechanisms by which BRAF^V600E or its downstream effectors confer apoptosis resistance are poorly understood. In BRAF^V600E melanoma cells, the MAPK pathway is an important regulator of cell survival through the suppression of proapoptotic BH3 family proteins and potentially by upregulation of antiapoptotic proteins (21–23). The importance of antiapoptotic proteins in CRC has been previously reviewed (24, 25). BIM is a BH3-only protein that induces cell death through antagonizing the antiapoptotic effects of BCL-2, BCL-XL, and MCL-1 (26). Treatment with inhibitors of MEK and BRAF were shown to prevent the proteasomal degradation of BIM that led to apoptosis (27, 28). Further support for apoptosis as a contributor to mutant BRAF-mediated resistance is suggested by the observation that inhibition of BRAF^V600E combined with a BCL-2/BCL-xL antagonist can significantly enhance the cytotoxic effect of vemurafenib in melanoma cell lines (29) which may be relevant to CRC.

We tested the hypothesis that BRAF^V600E can confer apoptosis resistance by upregulating the expression of an antiapoptotic BCL-2 family protein(s). In this report, we demonstrate that BRAF^V600E can upregulate antiapoptotic MCL-1 (30) via its phosphorylation/stabilization to confer apoptosis resistance in human CRC cells. This resistance mechanism was reversed by concurrent MCL-1 antagonism and MEK/ERK inhibition that increased antitumor efficacy in vitro and in vivo.

The BRAF-mutant RKO, HT29, and WiDr cell lines were obtained from the ATCC. Isogenic RKO [A19 (BRAF^{V600E/−/−}), T29 (BRAF^WT/−/−)], VACO432 [parental (BRAF^V600E/WT), VT1 (BRAF^WT/−)], and HCT116 #152 (KRAS^WT/−) human colorectal cancer cell lines were obtained (2015) from Dr. B. Vogelstein (GRCF Biorepository & Cell Center, Johns Hopkins University, Baltimore, MD). Parental RKO (BRAF^{V600E/V600E/WT}) cells are triploid, harboring two mutant BRAF^V600E alleles and one wild-type (WT) BRAF allele. RKO A19 (BRAF^{V600E/−/−}) cells carry a mutant BRAF^V600E allele and have knockout of the WT allele. RKO T29 (BRAF^WT/−/−) cells contain knockout of two mutant BRAF^V600E alleles and carry one WT allele. In contrast, VACO432 cells and HCT116#152 cells are diploid: VACO432 VT1 (BRAF^WT/−) cells and HCT116#152 (KRAS^WT/−) cells contain knockout of one BRAF^V600E allele or one KRAS-mutant allele, respectively. Authentication of RKO, HT29, and WiDr cell lines was not performed within the previous 6 months. However, cell lines were routinely tested for Mycoplasma contamination every 3 months with a MycoAlert mycoplasma detection set (Lonza). For isogenic BRAF cells, GRCF utilizes s a short tandem repeat profiling for authentication. All cell lines, except HEK293T cells, were grown as monolayers in RPMI medium (Invitrogen, cat # 11875) supplemented with 10% (v/v) FBS and 1% antibiotic–antimycotic (Invitrogen, cat # 15240). HEK293T cells were utilized for pseudovirus production and were grown in DMEM (Sigma, cat # D5796) with supplementation as above. Cells were treated with cobimetinib (GDC-0973/XL-518; ActiveBiochem, cat # A-1180), A-1210477 (Selleckchem, cat # S7790) or their combination. Both cobimetinib and A-1210477 were prepared as 50 mmol/L and 10 mmol/L stock solutions in DMSO (Sigma, cat # D2650), respectively, and stored at -20°C.

Virus production in HEK293T cells and transduction of target cells with lentivirus were performed utilizing a standard procedure described previously (31). The nontargeting shRNA expression vector was obtained from Addgene (cat #1864). Lentiviral BIM and MCL-1 shRNA constructs were described previously (31). Lentiviral BRAF^V600E was generated by subcloning its cDNA (Addgene, cat # 15269) into the lentiviral vector pCDH1-puro-2HA. For transduction of lentiviral constructs (packaged as pseudotyped viral particles) into target cells, the growth medium of recipient cells was replaced with Opti-MEM (Invitrogen, cat # 31985) containing 8 μg/mL Polybrene (Sigma, cat # 107689) and lentivirus. The cells were incubated overnight at 37°C, and the medium was replaced the following day. Puromycin (2–4 μg/ml, Sigma; cat # P8833) was added 48 hours after transduction, and the puromycin-resistant cells were used for subsequent experiments.

Wild-type MCL-1 cDNA was purchased from Origene (cat # RC200521). MCL-1 AA or DD mutants at T92/T163 were generated using a mutagenesis kit (Agilent, cat # 200523) and synthesized primers containing the desired mutations, which were then cloned into retroviral pBape-puro-2HA vector.

Cells were seeded 1 day before transfection at 30%–50% confluence in growth medium without antibiotics. ERK siRNA (Cell Signaling Technology, cat # 6560) and Lipofectamine RNAiMax (Invitrogen, cat # 13778150) were diluted in Opti-MEM medium, mixed gently, and incubated to allow complex formation. The cells were then transfected by adding the RNAi–Lipofectamine complex dropwise to medium to achieve an siRNA concentration of 100 nmol/L. Cells were then incubated at 37°C, and knockdown efficiency was determined 48 hours after transfection.

Total RNA was extracted from cells using the RNEasy Mini Kit (Qiagen, cat # 74104), and RNA integrity was evaluated using an Agilent Bioanalyzer 2000. Competitive RT-PCR was performed with a one-step RT-PCR kit (Qiagen, cat # 210210) using the following primer sets containing an equimolar ratio of MCL-1 (forward, 5′-GGGCAGGATTGTGACTCTCATT-3′; reverse, 5′-GATGCAGCTTTCTTGGTTTATGG-3′) against β-actin (forward, 5′-TCACCCACACTGTGCCCATCTACGA-3′; reverse, 5′-CAGCGGAACCGCTCATTGCCAATGG-3′). Reverse transcription was coupled with PCR (25 cycles) on a thermocycler (Eppendorf, Mastercycler GX2). PCR products were quantified on the Agilent Bioanalyzer 2000 using the DNA 12,000 kit (cat # 5067-1508). In brief, samples were loaded onto DNA microchips, and the DNA fragments were then separated by capillary electrophoresis. The target DNA sizes and relative quantities were calculated on the basis of DNA ladders and an internal marker, respectively. The associated software then generates agarose gel-like images.

Apoptosis was analyzed by Annexin V⁺ staining and quantified by flow cytometry as described previously (32). Briefly, cells were incubated with the study drugs at prespecified time points. Trypsin was added to detach adherent cells that were then combined with floating cells. Cells were pelleted by centrifugation, and the pellet was washed three times in cold PBS. Cells were incubated with Annexin V conjugated with FITC (BD Biosciences, cat # 556419). The labeled cell populations were then quantitated by flow cytometry. For analysis of drug synergy, we calculated a combination index (CI) in cells treated with combination of cobimetinib and A-1210477 at a fixed dose ratio. The means of triplicate experiments were used to compute the CI using CompuSyn software (ComboSyn, Inc.). CI <1 indicated drug synergy.

Protein samples were prepared in lysis buffer [5 mmol/L MgCl₂, 137 mmol/L KCL, 1 mmol/L EDTA, 1 mmol/L EGTA, 1% CHAPS, and 10 mmol/L HEPES (pH 7.5)] supplemented with a protease inhibitor cocktail (Sigma, cat # P8340). The protein concentration of the samples was measured using the nanodrop method (Thermo Scientific). Cell lysates were incubated with primary antibodies for 3 hours at 4°C. Immunocomplexes were captured using magnetic beads conjugated with protein A/G (Pierce) and then washed three times in lysis buffer. Immunoprecipitated proteins were eluted with 2× LDS sample buffer (Invitrogen, cat # NP0008) and loaded onto a 14% SDS-PAGE gel for separation, followed by an electrical transfer onto polyvinylidene difluoride membranes (Bio-Rad, cat # 1620177). Immunoblotting was then performed as described above. For immunoprecipitation, MCL-1 antibody (BD Biosciences, cat # 554103) and conformation-specific antibodies for BAK (Millipore, cat # 06-536) or BAX (Sigma, cat # B8429) was used as primary antibody. For immunoblotting, primary antibodies included those against MCL-1 (Santa Cruz Biotechnology, cat # sc-819) and tubulin (Sigma, T4026). All other antibodies were purchased from Cell Signaling Technology.

HT29 cells containing stable expression of control (#293) or MCL-1 (#50) shRNA were injected (2 × 10⁶ cells per injection) into the right flank of male immunodeficient nu/nu, SCID mice (Charles River Laboratories) at 6–8 weeks of age. When flank tumors reached approximately 100 mm³, cobimetinib (Genentech) at a dose of 15 mg/kg or vehicle (5% DMSO/30% PEG 300/5% Tween 80/ddH₂O) was administered once every 3 days by oral gavage for 14 consecutive days. A replicate of 8 mice for vehicle and 6 mice for cobimetinib treatment per experimental condition were utilized. Two mice from vehicle groups were sacrificed before treatment to examine the baseline efficiency of MCL-1 knockdown. Tumor volume was calculated using the following formula: length × width² × 0.5. All animals were sacrificed at 72 hours after the last treatment dose and tumor tissue was snap frozen at −80°C for immunoblotting experiments. All the animal experiments were performed under an animal protocol approved by the Mayo Clinic Institutional Animal Care and Use Committee.

Statistical analysis was performed on data generated for the extent of apoptosis using the Annexin V assay. Experimental data represent the mean ± SD for triplicate experiments. Statistical significance was determined using Student t test in R programming language. P values <0.05 were considered statistically significant.

Evidence suggests that BRAF^V600E confers apoptosis resistance by an as yet undefined mechanism. To determine whether BRAF^V600E is associated with an increase in the expression of antiapoptotic BCL-2 family proteins, we utilized the isogenic CRC cell lines, RKO parental (BRAF^{V600E/V600E/WT}), RKO A19 (BRAF^{V600E/−/−}), and RKO T29 (BRAF^wt/−/−). In cells with BRAF^V600E, MEK and ERK were activated and an increase in the number of mutant BRAF alleles was associated with upregulation of MCL-1 in a gene dose-dependent manner (Fig. 1A). In contrast, the gain of BRAF^V600E alleles was not associated with an increase in either BCL-XL or BCL-2 expression (Fig. 1A). To confirm the ability of BRAF^V600E to upregulate MCL-1, we ectopically expressed BRAF^V600E in RKO T29 (BRAF^wt/−/−) and VACO432 VT1 (BRAF^wt/−) cell lines that was shown to activate MEK/ERK and to upregulate MCL-1, but not BCL-XL or BCL-2, and to increase MCL-1 phosphorylation at Thr¹⁶³ (pMCL-1^Thr163) (Fig. 1B). As there was no significant change in MCL-1 transcription by ectopic BRAF^V600E, MCL-1 upregulation is likely to occur by a posttranscriptional mechanism (Fig. 1B). We also ectopically expressed BRAF^V600E in HCT116#152 (KRAS^wt/−) cells derived from a KRAS-mutant background which showed similar upregulation of MCL-1 and pMCL-1^Thr163 (Fig. 1B). In contrast to BRAF^V600E, ectopic expression of constitutively active AKT did not increase pMCL-1^Thr163 (Fig. 1C). In a human colon cancer with mutant BRAF^V600E, MCL-1 protein expression was shown to be upregulated in the cytoplasm of tumor cells compared with normal colonic epithelial cells indicating the relevance of MCL-1 as a therapeutic target (Fig. 1D).

Mutant BRAF^V600E is known to cause sustained activation of the MEK/ERK MAPK signaling cascade that controls tumor cell proliferation, apoptosis, and invasion (2). To determine whether mutant BRAF^V600E–induced MEK/ERK signaling mediates MCL-1 expression, we silenced ERK by RNA interference in the mutant BRAF^V600E VACO432 CRC cell line. In addition to suppression of ERK signaling, knockdown of ERK was shown to inhibit MCL-1 expression and phosphorylation (Fig. 2A). In VACO432 VT1 cells with ectopic BRAF^V600E expression, knockdown of ERK was also shown to attenuate MCL-1 expression and to inhibit pMCL-1^Thr163 (Fig. 2A).

To demonstrate the ability of MCL-1 phosphorylation to confer protein stability, we ectopically expressed an HA-tagged, phosphorylation-mimicking double MCL-1 mutant [T92D/T163D (DD)], a nonphosphorylated mutant [T92A/T163A], or wild-type MCL-1 (Fig. 2B). These two residues (T92/T163) have been reported to mediate MCL-1 protein stability by ERK (33). Analysis of protein turnover in cycloheximide chase experiments revealed that Mcl-1-92/163DD, a phosphorylation-mimicking mutant, was more stabilized than was WT Mcl-1 or Mcl-1-92/163AA, a nonphosphorylation mutant (Fig. 2C). While the nonphosphorylation mutant Mcl-1-92/163AA was rapidly degraded, the phosphorylation-mimicking mutant Mcl-1-92/163DD exhibited greater protein stability compared with WT Mcl-1 (Fig. 2C). Taken together, these findings indicate that MEK-ERK signaling phosphorylates and stabilizes MCL-1 to prevent its degradation and thereby, increase its expression.

Given the ability of MEK/ERK to regulate MCL-1 expression downstream of BRAF^V600E, we determined whether the MEK/ERK inhibitor cobimetinib can suppress MCL-1 and thereby, sensitize mutant BRAF CRC cell lines to apoptosis. Cobimetinib was shown to potently inhibit the phosphorylation of ERK and MCL-1, and to attenuate MCL-1 expression in a dose- and time-dependent manner (Fig. 3A and B) that occurred in association with BIM induction and PARP cleavage (Fig. 3A). In contrast to the sustained effects of cobimetinib, the BRAF inhibitor vemurafenib produced only an early suppression of ERK and MCL-1 expression and phosphorylation (Fig. 3B) whose rebound may be related to EGFR-mediated rebound activation of MAPK signaling (16). To determine whether the observed effects are generalizable, cobimetinib was shown to suppress ERK activation and to downregulate MCL-1 in multiple BRAF^V600E CRC cell lines. Cobimetinib was shown to consistently induce proapoptotic BIM and to downregulate MCL-1 expression in BRAF-mutant colorectal cancer cell lines that was associated with cleavage of PARP and to a lesser extent, caspase-3 (Fig. 3C). Compared with isogenic WT BRAF cell lines, the presence of one or two mutant BRAF^V600E alleles was associated attenuated cobimetinib-induced apoptosis that may be related to their MCL-1 upregulation (Fig. 3D).

To demonstrate the role of MCL-1 in conferring resistance to cobimetinib, we suppressed MCL-1 using shRNA in RKO and HT29 cell lines and examined drug-induced apoptosis. Knockdown of MCL-1 was shown to increase cobimetinib-induced apoptosis compared with control shRNA cells, as shown by Annexin V labeling and cleaved PARP and caspase-3 (Fig. 4A and B); this effect was confirmed with a second shRNA with a different MCL-1 targeting sequence (Fig. 4A and B). These data indicate that antagonism of MCL-1 in mutant BRAF^V600E cells treated with cobimetinib can elicit a robust apoptotic response.

We next determined whether suppression of MCL-1 can enhance cobimetinib-induced tumor regression in vivo. HT29 cells stably expressing control (#293) or MCL-1 (#50) shRNA were grown and then transplanted subcutaneously into SCID mice to generate tumor xenografts that were subsequently treated with vehicle or cobimetinib (15 mg/kg every three days). Prior to treatment, potent suppression of MCL-1 was demonstrated in HT29 tumor xenograft-bearing mice by immunoblotting (Fig. 4D, top). Compared with control, MCL-1 knockdown was shown to suppress tumor growth in untreated mice (Fig. 4C). Treatment with cobimetinib significantly suppressed tumor growth in xenograft-bearing mice and was shown to significantly enhance tumor growth inhibition in MCL-1 shRNA tumor xenografts (Fig. 4C).

In post-treatment tissue samples from xenograft-bearing mice, we analyzed and compared MEK/ERK activation, MCL-1 expression, and apoptotic markers (cleaved PARP and cleaved caspase-3) among the 4 treatment groups. Suppression of MCL-1 was maintained in knockdown cells (Fig. 4D, bottom). Consistent with in vitro results, inhibition of ERK phosphorylation by cobimetinib led to downregulation of MCL-1 expression in control shRNA tumor xenografts (Figs. 3 and 4D, bottom). Furthermore, cobimetinib treatment increased cleavage of both PARP and caspase-3 which was further enhanced in MCL-1 knockdown tumor xenografts (Fig. 4D, bottom).

To gain further mechanistic insight, we evaluated the small-molecule MCL-1 inhibitor A-1210477 (34) that has been shown to demonstrate selectivity for MCL-1 in cancer cells (34). As previously reported in other tumor cell types (34–36), we found that treatment with A-1210477 upregulates MCL-1 expression in BRAF-mutant CRC cells and in the melanoma cell line A375 in a dose-dependent manner (Fig. 5A). Treatment with A-1210477 was shown to upregulate MCL-1 together with phosphorylation of ERK and MCL-1, all of which were abrogated by the addition of cobimetinib (Fig. 5A). The combination of cobimetinib and A-1210477 resulted in MCL-1 downregulation, and was associated with increased cleavage of PARP and caspase-3 in CRC and melanoma cell lines (Fig. 5A). Furthermore, the drug combination triggered a 2-fold increase in apoptosis, shown by Annexin-V labeling, compared with either drug alone in BRAF-mutant CRC cell lines (Fig. 5B). A synergistic interaction was seen between cobimetinib and A-1210477 as determined by calculation of the combination index (CI; CI < 1; data not shown). Compared with cobimetinib, the combination of vemurafenib with A-1210477 triggered less apoptosis and the triple drug combination failed to further enhance apoptosis. These data indicate that cobimetinib shows greater efficacy combined with an MCL-1 antagonist than it does with the BRAF inhibitor, vemurafenib (Fig. 5C).

We further examined the mechanism underlying the observed synergistic interaction between cobimetinib and A-1210477. We found that the drug combination activated BAX, as shown by its conformational change, and BAK to a greater extent than did either drug alone (Fig. 6A). Analysis of protein–protein interactions using immunoprecipitation revealed that A-1210477 disrupted the association between MCL-1 and BAK as well as MCL-1 and BIM in RKO and HT29 cell lines (Fig. 6B), confirming the ability of A-1210477 to antagonize MCL-1. The ability of BIM to promote apoptosis induced by the drug combination was demonstrated using BIM shRNA that attenuated BAX, but not BAK, activation.

To date, the treatment of human CRC with mutant BRAF has been largely unsuccessful due to intrinsic/acquired resistance that includes reactivation of signaling through the MAPK pathway (37). Studies indicate that the MAPK pathway is reactivated in 70%–79% of melanomas with acquired vemurafenib or dabrafenib resistance (37, 38). The recovery of MAPK signaling can occur by rebound EGFR activation and/or the acquisition of mutations in NRAS, MEK1/2, and BRAF-splice form mutants (39–41). These data indicate the need for potent inhibition of activated MEK and support a strategy of combined BRAF and MEK inhibition in CRC which has been shown to be effective in BRAF-mutant melanoma (19). However, this strategy was recently shown to be largely ineffective against human CRC (4). We, therefore, tested the hypothesis that apoptosis resistance is a major contributor to therapeutic failure in mutant BRAF CRCs. We found that activation of MEK/ERK by mutant BRAF can upregulate antiapoptotic MCL-1 by its phosphorylation and stabilization to prevent protein turnover. Mcl-1 is known to play a critical role in tumor cell survival and is overexpressed in many human cancers. We demonstrated that phosphorylation can stabilize MCL-1 using a MCL-1 phosphorylation-mimicking mutant and an unphosphorylated MCL-1 mutant. Phosphorylation is known to influence the ubiquitination and thus degradation of the modified protein, with examples including c-Myc (42, 43), the androgen receptor (44), or the yeast transcriptional factor, Rpn4 (45). In a BRAF-mutant human colon cancer, we observed that MCL-1 proteins were overexpressed in tumor cells compared with normal colonic epithelia. Upregulation of MCL-1 protein was increased in a BRAF^V600E gene dose-dependent manner in CRC cell lines isogenic for BRAF (that differ in the number of mutant BRAF alleles), and MCL-1 upregulation/phosphorylation was confirmed in cells with ectopic BRAF^V600E expression. In human melanoma cells, mutant BRAF was reported to increase MCL-1 expression by the STAT3 transcription factor (46). However, MCL-1 was not transcriptionally regulated in our CRC cell lines indicating that its posttranscriptional regulation results in enhanced protein stability.

As mutant BRAF activates downstream MAPK signaling, we determined whether ERK activation is an important contributor to MCL-1 upregulation. Knockdown of ERK was shown to attenuate MCL-1 expression and phosphorylation. Furthermore, we made the novel observation that the selective MEK/ERK inhibitor cobimetinib can suppress MCL-1 expression and phosphorylation in CRC cells. In melanoma cell lines, conflicting data exist for the effect of MEK inhibitors on MCL-1 expression in that MEK inhibition by U0126 (47) was shown to downregulate MCL-1, whereas trametinib (48) increased MCL-1 expression. Cobimetinib-induced apoptosis was attenuated by mutant BRAF in isogenic CRC cells where MCL-1 was shown to be upregulated. Suppression of MCL-1 by shRNA or using the selective small-molecule MCL-1 inhibitor A-1210477 was shown to potently increase cobimetinib-induced apoptosis, indicating that MCL-1 can mediate apoptosis resistance by mutant BRAF. We confirmed this observation in colon cancer xenograft-bearing mice whereby either MCL-1 antagonism or cobimetinib treatment suppressed tumor growth compared with their respective controls, and cobimetinib significantly enhanced tumor growth inhibition in MCL-1 knockdown versus control xenograft-bearing mice. Treatment with cobimetinib enhanced cleavage of both PARP and caspase-3 in tumor xenografts derived from MCL-1 knockdown compared with control shRNA cells.

To gain mechanistic insight into the ability of MCL-1 antagonism to enhance cobimetinib-induced apoptosis, we utilized the small-molecule MCL-1 inhibitor, A-1210477, which has shown “on-target” cancer cell cytotoxicity as monotherapy (34). By inhibiting MCL-1, A-1210477 was shown to interact synergistically with the BCL-2/BCL-xL inhibitor ABT-263 (navitoclax) to kill a variety of cancer cell types (34–36). We found that the combination of A-1210477 and cobimetinib was synergistic as shown by a mechanism that involved disruption of the interaction of MCL-1 with the proapoptotic BH3-only proteins BIM and BAK by A-1210477, and by the ability of cobimetinib to suppress MCL-1 and induce BIM expression. MCL-1 is known to bind to and neutralize BIM and BAK (49), and release of BIM from MCL-1 contributes to apoptosis induction. Studies have shown that BIM induction is an important mechanism by which MEK/ERK inhibitors, including cobimetinib, can cooperatively kill tumor cells (50–52). Whereas ERK phosphorylation of BIM targets the protein for polyubiquitination and proteasomal degradation (53), we previously reported that the MEK/ERK inhibitor, GDC-0623, inhibits BIM phosphorylation at Ser⁶⁹ to increase protein stability (50). Depending on the specific motif that is modified, phosphorylation can either inhibit ubiquitination-triggered protein degradation (e.g., BIM) or promote proteins stability (e.g., MCL-1; ref. 54). The importance of BIM induction in triggering apoptosis was shown in BIM knockdown cells where BAX activation was attenuated by drug combination, consistent with evidence that BIM preferentially activates BAX, whereas BID preferentially activates BAK (55). These data support dual targeting of MCL-1 by reducing its expression, antagonizing its function, and promoting BIM which cooperatively enhance apoptosis.

We found that cobimetinib inhibited ERK activation and suppressed MCL-1 to a greater extent than did vemurafenib in vitro. Compared to its combination with cobimetinib, the addition of A-1210477 to vemurafenib triggered less apoptosis, and the triple drug combination failed to further enhance apoptosis that was likely due to the observed weaker ability of vemurafenib to induce BIM and inhibit MCL-1 expression. These data, albeit limited, suggest that the combination of a MEK/ERK inhibitor with an MCL-1 antagonist may be a more potent inducer of apoptosis compared with its combination with a BRAF inhibitor in CRC cell lines. Furthermore, the ability of MCL-1 to confer resistance to MEK/ERK and BRAF inhibition may explain, in part, the disappointing response rate for the combination of dabrafenib and trametinib in patients with metastatic CRC (4). Cobimetinib is being studied in clinical trials in patients with metastatic CRC and was found to be well tolerated at the same dose (60 mg daily; ref. 56) which is approved by the FDA for treatment of metastatic melanoma in combination with vemurafenib (19). To date, neither A-1210477 nor other inhibitors of MCL-1 have undergone testing in humans. However, our demonstration that cobimetinib in combination with MCL-1 antagonism can enhance tumor growth inhibition in vivo suggests the clinical relevance of this strategy.

In conclusion, we identify MEK/ERK-mediated MCL-1 upregulation by mutant BRAF to be an important mechanism of apoptosis resistance that contributes to poor therapeutic efficacy in BRAF-mutant CRC cells. In addition to potent MEK/ERK inhibition, cobimetinib was shown to attenuate MCL-1 phosphorylation/stabilization to increase apoptotic susceptibility that was enhanced by concurrent MCL-1 antagonism. The combination of cobimetinib with MCL-1 gene knockdown significantly inhibited BRAF^V600E-mutant CRC tumor growth in xenograft-bearing mice that was associated with enhanced tumor cell apoptosis, suggesting a promising and novel therapeutic strategy for the treatment of mutant BRAF CRC.

No potential conflicts of interest were disclosed.

Conception and design: H. Kawakami, S. Huang, D. Mukhopadhyay, F.A. Sinicrope

Development of methodology: H. Kawakami, F.A. Sinicrope

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Kawakami, S. Huang, K. Pal, S.K. Dutta, F.A. Sinicrope

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Kawakami, S. Huang, K. Pal, D. Mukhopadhyay, F.A. Sinicrope

Writing, review, and/or revision of the manuscript: H. Kawakami, S. Huang, F.A. Sinicrope

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Huang, F.A. Sinicrope

Study supervision: S. Huang, D. Mukhopadhyay, F.A. Sinicrope

The authors wish to thank Deborah Frank for her very capable secretarial assistance in the preparation of this manuscript.

Hisato Kawakami was a recipient of a fellowship grant from the Uehara Memorial Foundation, Japan. This work was supported, in part, by grants from the National Cancer Institute/NIH (grant number R01 CA113681; to F.A. Sinicrope), the Mayo Clinic Cancer Center (grant number P30 CA15083), and the Mayo Clinic Center for Cell Signaling in Gastroenterology (grant number P30 DK084567).

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.