A Novel Combination Treatment Targeting BCL-X_L and MCL1 for KRAS/BRAF-mutated and BCL2L1-amplified Colorectal Cancers

Colorectal cancer is the third most common cancer worldwide and ranks as the fourth most prevalent cause of cancer-related deaths (1). Recent advances in high-throughput sequencing have enabled the molecular classification of colorectal cancers into hypermutated [microsatellite instability (MSI)/CpG island methylation phenotype (CIMP)-high] and nonhypermutated [microsatellite stable (MSS)/chromosomal instable (CIN)] subgroups (2). In addition, gene expression profiling studies have suggested further classification of the nonhypermutated group into epithelial proliferative and mesenchymal/invasive stromal subgroups (3). Although these classifications suggest potential avenues for targeted therapeutics, there has actually been limited success in targeted therapies for colorectal cancers.

One subset of colorectal cancer cases for which targeted therapeutics will be particularly important are those patients with mutations in either the KRAS (30%–45% frequency) or BRAF (5%–15% frequency) genes. The prognosis for these patients is significantly worse than patients without these mutations (4, 5). These cancers are resistant to currently employed anti-EGFR treatments, such as panitumumab and cetuximab (6, 7), apparently due to activation of signaling pathways (including MEK-ERK and PI3K pathways) that are important for cell survival and drug resistance (8, 9). Despite extensive attempts to regulate these signaling pathways, single drug treatments have shown minimal therapeutic effect (10, 11). More recently, several combination treatment strategies have been pursued for colorectal cancers with KRAS or BRAF (KRAS/BRAF) mutations. These combination treatments are mainly based on MEK inhibitors (12, 13). Although some of these combination treatments are now under evaluation in clinical trials (13), it is still unclear which combinations will be clinically applicable. Furthermore, having alternate treatment strategies for KRAS/BRAF–mutated colorectal cancer will be important to mitigate colorectal cancer cases acquiring resistance to selected targeted therapies.

In this study, we have found that as much as 21.7% of colorectal cancers can be categorized into a novel colorectal cancer subgroup defined by KRAS/BRAF mutations and BCL2L1 (encoding BCL-X_L) amplification. Concomitant treatment of a BCL-X_L inhibitor, ABT-263, and an MCL1 inhibitor, YM-155, showed synergistic antitumor effects in both in vitro cell line models and in vivo patient-derived xenograft (PDX) models. These results propose a novel combination targeted treatment strategy for a subgroup of recalcitrant colorectal cancers.

All of the human colorectal cancer cell lines, authenticated using DNA fingerprint analysis, were provided by the Korean Cell Line Bank in 2015. Cells were passaged for fewer than 6 months after resuscitation. Cells were cultured in RPMI1640 medium (Life Technologies) with 10% FBS (Life Technologies), penicillin (100 U/mL; Life Technologies), and streptomycin (100 U/mL; Life Technologies). All cells were maintained in a humidified incubator with 5% CO₂ at 37°C.

Cell lysates were prepared using RIPA buffer (Thermo Scientific) with protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Roche), incubated for 15 minutes in ice and centrifuged at 16,800 × g for 10 minutes at 4°C. BCA method (Thermo Scientific) was used to determine the protein concentration. Proteins were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membrane. After blocking using 5% skim milk, membranes were probed with anti-BCL-X_L (Cell Signaling Technology), anti-MCL1 (Santa Cruz Biotechnology), anti-cleaved caspase-3 (Cell Signaling Technology), anti-PARP (Cell Signaling Technology), anti-survivin (Cell Signaling Technology), anti-caspase-8 (Cell Signaling Technology), anti-cleaved caspase-8 (Cell Signaling Technology), anti-caspase-9 (Cell Signaling Technology), anti-BIM (Cell Signaling Technology), anti-AKT (Cell Signaling Technology), anti-phospho-AKT (Cell Signaling Technology), anti-S6K (Cell Signaling Technology), anti-phospho-S6K (Cell Signaling Technology), anti-4EBP (Cell Signaling Technology), anti-phospho-4EBP (Cell Signaling Technology), and anti-Actin (Sigma-Aldrich Corporation) antibody. The membranes were washed and incubated with horseradish peroxidase–conjugated secondary antibody, followed by enhanced chemiluminescence development according to the manufacturer's instructions (Pierce).

The extracted genomic DNA was restricted with EcoRI (New England Biolabs) enzyme for 1 hour at 37°C. The PCR mixture was assembled in 20-μL solution containing 1× droplet digital PCR (ddPCR) supermix (Bio-Rad), 1× probe and primer premix for determining BCL2L1 gene and internal control gene, RNase P (final concentration of 250 nmol/L for probe and 900 nmol/L for each primer; Applied Biosystems), and 10 ng of the restricted DNA. The reaction mixture and droplet generation oil (Bio-Rad) were loaded into the droplet generator (QX-200; Bio-Rad). The droplets were transferred to a 96-well PCR plate and PCR reaction was performed as follows: enzyme activation for 10 minutes at 95°C, 40 cycles of 94°C for 30 seconds, 60°C for 1 minute, and 98°C for 10 minutes, followed by enzyme deactivation for 10 minutes at 98°C and 4°C hold (performed with a ramp rate of 2°C/sec in all steps). The PCR plate was placed in a droplet reader (Bio-Rad). After the reading, the copy number variation of target genes was analyzed by Quanta software (Bio-Rad) accompanied by the droplet reader. The amplification threshold value was set at 3.0 for tissues and cell lines.

Frozen tissue and formalin-fixed paraffin-embedded (FFPE) tissue samples of colorectal cancer were obtained from individuals who underwent colorectal cancer surgery at Gachon University Gil Medical Center (Incheon, South Korea). Total DNA was extracted from frozen tissues using the QIAamp DNA Mini kit (Qiagen) for ddPCR, and FFPE tissue samples were used for histologic analysis and IHC. All samples were obtained with informed consent at the Gachon University Gil Medical Center, and the study was approved by the institutional review board in accordance with the Declaration of Helsinki.

After removal, tumor tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. Four-micrometer-thick consecutive sections that covered the entire tumor were cut and mounted on silanized slides. Hematoxylin and eosin staining (H&E) was performed according to standard protocols. For IHC, slides were deparaffinized in xylene and rehydrated in a series of graded alcohols, and the antigen was retrieved in Antigen unmasking solutions (Vector Laboratories). Sections were then treated with 3% of hydrogen peroxide to inhibit endogenous peroxidase. The sections were stained with primary antibodies [anti-BCL-X_L (Cell Signaling Technology), anti-MCL1 (Santa Cruz Biotechnology), anti-cleaved caspase-3 (Cell Signaling Technology), anti-Ki67 (Thermo Fisher Scientific)], followed by appropriate secondary antibody, and then visualized by use of the appropriate substrate/chromogen (Diaminobenzidine, DAB) reagent according to the manufacturer's recommendations (Vector Laboratories). TUNEL staining was performed using DeadEnd Colorimetric TUNEL System (Promega) according to the manufacturer's instructions. Counterstaining was performed using Hematoxylin QS (Vector Laboratories). The intensities of stain were assessed using “ImageJ” software (https://imagej.nih.gov/ij), which is a Java image processing and analysis program. We utilized “ImageJ” software as previously reported by applying same threshold for each antibody staining and selecting “Analyze Particles” option (14).

For the cell viability assays, about 5,000 cells in 96-well plates were treated with indicated drugs for 72 hours and cell viabilities were estimated using the EzCytox WST assay kit according to the manufacturer's instructions (Daeil Lab). Cell viabilities were estimated as relative values compared with untreated controls. Drug synergism was quantified by calculating combination index (CI) based on the multiple drug effect equation (15). The CI was based on the mass action law in biophysics and biochemistry, and calculated by the Chou–Talalay equation, which considers both drug potency and the shape of the dose–effect curve (15). The CI for each concentration of drugs was calculated by CompuSyn Software (ComboSyn Inc.) and a CI lower than 0.9 indicates synergism.

The SW620 cells were seeded into 60-mm culture dishes (1 × 10⁶ cells/well) and were treated with indicated drugs. The treated cells were collected and washed in PBS. The apoptotic cells were analyzed by FITC Annexin V Apoptosis Detection Kit (BD Pharmingen) according to the manufacturer's protocol. The suspended cells were treated with 10 μL of Annexin V-FITC and 10 mL of propidium iodide for 5–15 minutes in the dark. Stained cells were analyzed with a FACScan flow cytometer (Becton Dickinson).

For the transfection of small interfering RNA (siRNA) to downregulate target genes, specific siRNAs targeting for MCL1 and survivin were purchased from Genolution. Cells were transfected using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol. After 24 hours, cells were collected and plated in 96-well plates for drug test. The effects of siRNA knockdown were measured by Western blot analysis with the indicated antibodies. The sequence of siRNA for MCL1 was 5′- GUGCCUUUGUGGCUAAACATT-3′, for survivin was 5′- AAGGCUGGGAGCCAGAUGACGUU-3′, and for GFP was 5′- GACGUAAACGGCCACAAGUTT-3′.

After treatment of indicated drugs, cell lysates were prepared using immunoprecipitation (IP) buffer (Thermo Scientific) with protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail (Roche), incubated for 15 minutes on ice, and centrifuged at 16,800 × g for 10 minutes at 4°C. Protein G Dynabeads (Thermo Scientific) were incubated with anti-BIM for 2 hours at 4°C with rotation. The antibody-bound magnetic beads were incubated with the cell lysate for 2 hours at 4°C with rotation. The beads were washed for three times with IP buffer and eluted in 4× SDS sample buffer (Bio-Rad) with boiling. Immunoprecipitated proteins were evaluated by Western blot analysis.

Total RNA from cells was extracted using RNeasy Plus Mini Kit (Qiagen). Reverse transcription with 1 μg of total RNA was performed using Maxime RT PreMix (Intron Biotechnology). We performed real-time PCR using SYBR Green PCR Master Mix (Applied Biosystems) and checked MCL1, AKT, S6K, DR4, DR5, and TRAIL mRNA level normalized by GAPDH used as internal control. The sequences of primers were as follows: for MCL1, 5′-TAAGGACAAAACGGGACTGG-3′ and 5′- ACCAGCTCCTACTCCAGCAA-3′; for AKT, 5′- TCTATGGCGCTGAGATTGTG-3′ and 5′- CTTAATGTGCCCGTCCTTGT-3′; for S6K, 5′- GTGGAGGAGAACTATTTATGCAGTTAGAAAGAG-3′ and 5′- TCCAGCGTCCCCAGGACCAGCTC-3′; for DR4, 5′-GGCTGAGGACAATGCTCACA-3′ and 5′-TTGCTGCTCAGAGACGAAAGTG-3′; for DR5, 5′-GACTCTGAGACAGTGCTTCGATGA-3′ and 5′-CCATGAGGCCCAACTTCCT-3′; for TRAIL, 5′-GCTCTGGGCCGCAAAAT-3′ and 5′-TGCAAGTTGCTCAGGAATGAA-3′; and for GAPDH, 5′- CGCTCTCTGCTCCTCCTGTT-3′ and 5′- CCATGGTGTCTGAGCGATGT-3′.

The target sequence of DR5 with BsmI (New England Biolabs) restriction site was cloned into the lentiCRISPR v2 plasmid (Addgene #52961). HEK293T were transfected using Lipofectamine 2000 reagent (Invitrogen) with lentiCRISPR v2 (DR5 sgRNA), psPAX2, and VSV-G plasmids. After 48 hours, the supernatants were collected. SW620 cells were incubated with lentivirus-containing media for 48 hours and selected by puromycin (1 μg/mL) for 72 hours. The effects of gene knockdown were measured by Western blot analysis with the indicated antibodies.

Total RNA from cells after drug treatment was extracted using RNeasy Plus Mini Kit (Qiagen). Whole-transcriptome expression profiles were generated by RNA sequencing using Mapsplice and RSEM with TCGA RNASeq v2 pipeline (https://wiki.nci.nih.gov/display/TCGA/RNASeq+Version+2; refs. 16, 17), and then calculated reads counts for each gene were used for analyses of differentially expressed genes utilizing the DESeq2 packages (18). ClueGO plug-in (v2.2.5, http://www.ici.upmc.fr/cluego/) in Cytoscape software (v3.3.0, http://cytoscape.org/) was used to analyze GO and functional groups in networks for upregulated genes in combination treatment. ClueGO is a tool combining GO terms to create functionally grouped annotations in a network (19). GO Biological Process database (http://www.geneontology.org/) was used for functional enrichment analysis. Significantly enriched GO terms were calculated by a two-sided hypergeometric test with a Bonferroni correction (P < 0.05) and the degree of connectivity between terms in the network is calculated using kappa statistics (kappa score of 0.4). The RNA sequencing data have been deposited in the European Nucleotide Archive (accession no. PRJEB20153).

Mice were cared for according to institutional guidelines of the Institutional Animal Care and Use Committee (IACUC) of the Seoul National University (No. 14-0016-C0A0). For patient-derived xenograft (PDX) models, the surgically resected tissues were minced into pieces approximately 2 mm in size and injected into the flanks of 4-week-old NOD/SCID/IL2γ-receptor null (NSG) female mice. For cell line xenograft models, the flanks of 4-week-old NSG female mice were injected subcutaneously with 1 × 10⁶ cells in 100-μL PBS. Drug treatments began after tumors reached approximately 200 mm³. Mice were randomly divided into four treatment groups consisting of 4–5 mice in each group: (i) vehicle only; (ii) ABT-263 only (Selleckchem, 100 mg/kg, daily), (iii) YM-155 only (Selleckchem, 5 mg/kg, daily), and (iv) ABT-737 plus YM-155. The vehicle for ABT-263 was 10% (v/v) ethanol (Merck), 30% (v/v) polyethylene glycol (Sigma), and 60% (v/v) Phosal 50 PG (Lipoid), and the vehicle for YM-155 was 0.9% normal saline. ABT-263 was administered via oral injection and YM-155 was subcutaneously administrated using micro-osmotic pump (Alzet model 1007D, Durect) for 21 days.

Statistical calculations were performed using Prism 5.0 (GraphPad). Differences between multiple variables were assessed by one-way ANOVA with Tukey multiple comparison test. The difference of tumor growth rate between differently treated groups was analyzed by growth curve analysis, which is a multilevel regression technique designed for analysis of time course or longitudinal data (20). Growth curve analysis was performed using the R software package lme4. The difference was considered significant if the P value was less than 0.05.

Mutations in KRAS have been shown to increase protein expression of BCL-X_L (21, 22). Because KRAS and BRAF are in the same signaling pathway (23), it is possible that BRAF mutations also similarly increase protein expression of BCL-X_L. Therefore, we investigated BCL-X_L protein expression in colorectal cancer cell lines that carry either KRAS or BRAF mutations versus having wild-type KRAS/BRAF genes. As expected, the protein levels of BCL-X_L were seen to be elevated in most KRAS/BRAF-mutated colorectal cancer cells (Fig. 1A and B; Supplementary Fig. S1A), which was confirmed using KRAS/BRAF-mutated isogenic SW48 cells (Supplementary Fig. S1B). We further determined the copy number of BCL2L1 in the colorectal cancer cell lines examined using droplet digital PCR (ddPCR) and found five cell lines (5/13, 38.5%) that had increased gene copy number of BCL2L1 (Fig. 1A). Among them, four cell lines with both KRAS/BRAF mutation and BCL2L1 amplification showed significantly increased BCL-X_L expression levels by Western blotting compared with wild-type cells (Fig. 1A and B; Supplementary Fig. S1A). However, another antiapoptotic protein, BCL2, was not nearly expressed in these four cell lines (Supplementary Fig. S1C). Moreover, KRAS/BRAF-mutated, BCL2L1-amplified colorectal cancer tissues from three unrelated patients also showed elevated BCL-X_L expression via immunostaining compared with colorectal cancer tissues from patients with wild-type KRAS/BRAF (Fig. 1C). In our colorectal cancer cohort, we found simultaneous KRAS/BRAF mutation and BCL2L1 amplification in 21.7% (5/23) of patients (Supplementary Fig. S2). Because BCL-X_L proteins were highly expressed in colorectal cancers with KRAS/BRAF mutation and BCL2L1 amplification, this subset of colorectal cancers could be considered for BCL-X_L–targeted treatments (24).

Inhibition of BCL-X_L alone does not appear to be effectively cytotoxic as BCL-X_L inhibitors lead to upregulation of MCL1, which in turn leads to cellular resistance to the original BCL-X_L–targeted treatments (12, 25). Therefore, to effectively inhibit the action of BCL-X_L, it may be necessary to inhibit both BCL-X_L as well as MCL1. We identified nine chemicals that were previously shown to downregulate MCL1 proteins (Supplementary Table S1) and combined each of these with the BCL-X_L–inhibiting drug, ABT-263 (26), on a representative KRAS/BRAF–mutated, BCL2L1-amplified cell line (SW620; Supplementary Fig. S3) as well as a representative KRAS/BRAF wild-type, BCL2L1-diploid cell line (SNU1684; Supplementary Fig. S4). The drug combination of ABT-263 + YM-155 appeared to have a synergistic effect in the KRAS/BRAF–mutated, BCL2L1-amplified cell lines with little cytotoxic or synergistic effect in the KRAS/BRAF wild-type and/or BCL2L1-diploid cell lines (Fig. 2; Supplementary Fig. S5). However, KRAS/BRAF–mutated and BCL2L1-amplified SW620 cells treated with siRNA for BCL-X_L did not show the synergistic effect of ABT-263/YM-155 combination (Supplementary Fig. S6). In addition, concomitant treatment of YM-155 abrogated the ABT-263–induced increase of MCL1 levels only in the KRAS/BRAF–mutated, BCL2L1-amplified cell lines (Fig. 2; Supplementary Figs. S5, S7, and S8). The synergistic effect of combination treatment is partly attributed to the increase of cellular apoptosis, which was validated by the observation of increased cleavage of PARP and caspase-3 (Fig. 2; Supplementary Figs. S5, S7, and S8), and the increase of Annexin-V–positive and Annexin-V/propidium iodide (PI) double-positive cells in FACS analyses (Supplementary Fig. S9A–S9C).

YM-155 has two possible functions, to inhibit survivin and to inhibit MCL1 (27, 28). When combined with ABT-263, siRNA experiments against MCL1 dramatically decreased the relative cell viability as compared with the siRNA control or siRNA to survivin (Fig. 3A; Supplementary Fig. S10A). This result suggests that targeted cell death may have been occurred due to YM-155′s action against MCL1 rather than survivin.

When BCL-X_L is inhibited, BIM (Bcl-2 interacting mediator of cell death) is released, leading to mitochondrial-induced apoptosis in cells having increased BCL-X_L expression (29, 30). However, BCL-X_L inhibition also upregulates MCL1, which is then capable of binding the free BIM molecules, preventing cellular apoptosis (12). To address the question of whether inhibition of both BCL-X_L and MCL1 leads to significant release of BIM molecules, we conducted immunoprecipitation experiments and found that the combination of YM-155 with ABT-263 led to the reduction of BCL-X_L–BIM and MCL1–BIM complexes and increased amounts of free BIM within the targeted cells (Fig. 3B; Supplementary Fig. S10B).

Downregulation of MCL1 by YM-155 could be due to (i) decreased MCL1 mRNA expression, (ii) decreased MCL1 protein synthesis, or (iii) increased MCL1 protein degradation. We examined MCL1 mRNA levels in two different KRAS/BRAF–mutated, BCL2L1-amplified cell lines (SW620 and HCT116) and found no significant decrease of MCL1 mRNA levels after treatment with YM-155 and after treatment with the combination of YM-155 and ABT-263 (Fig. 3C). Next, we examined the rate of MCL1 protein degradation in the SW620 cell line after treatment with YM-155 or combination of YM-155 and ABT-263, and found no evidence for accelerated MCL1 protein degradation in the presence of YM-155 alone or combined with ABT-263 (Fig. 3D; Supplementary Fig. S10C). In addition, inhibition of proteosomal degradation did not block the degradation of MCL1 by combination treatment of ABT-263 and YM-155 (Fig. 3E; Supplementary Fig. S10D).

To investigate whether downregulation of MCL1 is due to decreased MCL1 protein synthesis, we evaluated three proteins, S6K, eIF4E-binding protein 1 (4EBP1), and AKT, that are in the mTOR pathway, which regulates protein translation of MCL1 (31). We found that the combination treatment led to a deficiency of phosphorylated and total S6K and AKT in KRAS/BRAF–mutated, BCL2L1-amplified (SW620) cells (Fig. 4A; Supplementary Fig. S11A). In addition, the combination treatment led to the absence of phosphorylated 4EBP1 in KRAS/BRAF–mutated, BCL2L1-amplified (SW620) cells (Fig. 4A; Supplementary Fig. S11A). However, these effects could not be observed in a KRAS/BRAF wild-type and BCL2L1-diploid cell line (SNU C1; Fig. 4A; Supplementary Fig. S11A). Taken together, this suggests that the downregulation of MCL1 in these combination-targeted cells is in part due to the reduction of S6K, 4EBP1, and AKT, which may inhibit the protein synthesis of MCL1.

Decreased levels of S6K protein does not appear to be due to inhibition of mRNA expression (Fig. 4B) or proteasomal degradation (Fig. 4C and D, Supplementary Fig. S11B and S11C), but rather via caspase-3–mediated protein degradation (Fig. 4E; Supplementary Fig. S11D). Downregulation of AKT was due to both decrease of mRNA expression (Fig. 4B) and decrease of protein stability (Fig. 4C; Supplementary Fig. S11B) via caspase-3–mediated protein degradation (Fig. 4E; Supplementary Fig. S11D). Indeed, activation of caspase-3, which detected by cleavage of PARP and caspase-3, preceded the degradation of AKT and S6K (Fig. 4F; Supplementary Fig. S11E). In addition, inhibition of caspase-3 restored MCL1 levels even in the presence of the combination treatment of ABT-263 and YM-155 (Fig. 4E; Supplementary Fig. S11D).

YM-155 was reported to activate death receptor pathway in pancreatic cancer cells (32), and we checked this possibility in our combination treatment for colorectal cancer cells. Cotreatment of YM-155 and ABT-263 leads to increased expression of death receptor 4 (DR4)/DR5 and TRAIL (Fig. 5A and B), and activated caspase-8 (Fig. 5C; Supplementary Fig. S12A) only in the KRAS-mutated and BCL2L1-amplified cell line (SW620). In contrast, when we used CRISPR/Cas9 to specifically downregulate DR5, sgRNA transfection deactivated caspase-3 and led to a subsequent increase of MCL1 protein levels in the presence of the combination treatment of ABT-263 and YM-155 (Fig. 5D; Supplementary Fig. S12B). Taken together, cotreatment of ABT-263 and YM-155 appears to lead to death receptor–mediated activation of caspase-3/8, inducing the degradation of AKT and S6K, and resulting in translational inhibition of MCL1. When both MCL1 and BCL-X_L are effectively inhibited in KRAS/BRAF–mutated and BCL2L1-amplified cells, free BIM molecules can trigger mitochondria-dependent cell apoptosis.

To overview the global effect of each drug and combination on KRAS/BRAF–mutated and BCL2L1-amplified cells, we investigated the transcriptomic changes induced by drug treatment using RNA-seq. Total of 2,223 genes were significantly altered by at least one treatment condition (P < 0.05; Fig. 6A; Supplementary Table S2). Compared with ABT-263 treatment, which resulted in minimal gene expression changes (128 genes), YM-155 and combination treatment significantly altered expressions of 1,781 and 1,199 genes (P < 0.05), respectively, and exhibited a number of common altered genes (Fig. 6B). Gene Ontology (GO) functional enrichment analysis using ClueGo (19) indicated that 737 genes, which were significantly deregulated in combination treatment sample compared with other samples (Supplementary Table S3), are involved in several biological processes, including cell development, secretion, cell migration, regulation of signal transduction, cell surface receptor signaling pathway, and response to hypoxia (Fig. 6C; Supplementary Table S4). These results suggest that YM-155 has more profound effects on gene expression compared with ABT-263, and combination treatment of ABT-263 and YM-155 produces stressful condition of cells by activating apoptosis pathway and induced reactive hyperactivation of several cellular processes especially in secretion, cell migration, and signaling transduction.

We established PDX mouse models from patient C033 (Fig. 7A) and C178 (Supplementary Fig. S13A), known to be colorectal cancers with KRAS mutations (G13D for C033, G12V for C178) and BCL2L1 amplification (3 copies for C033, 4 copies for C178), for in vivo drug efficacy testing. These mice exhibited a synergistic effect by the combined use of ABT-263 and YM-155 (Fig. 7A; Supplementary Fig. S13A). However, the synergistic effect of ABT-263 and YM-155 was not observed in PDX models from a KRAS wild-type and BCL2L1-diploid patient (Supplementary Fig. S13B). We found similar synergistic effect by the combination of ABT-263 and YM-155 in SW620 cell line xenograft models [KRAS-mutated (G12V) and BCL2L1-amplified (4 copies); Supplementary Fig. S14A], but not in Colo320HRS cell line xenograft models (KRAS wild-type and BCL2L1-diploid; Supplementary Fig. S14B). Histologic analyses showed increased expression of MCL1 by the ABT-263 treatment, which was abrogated by concomitant treatment with YM-155 (Fig. 7B). Furthermore, the concentration of apoptotic cells was increased in tumor tissue after the combination treatment based on TUNEL and cleaved caspase-3 staining (Fig. 7B). However, there was little difference in cell proliferation in single-treated or combination treated cells, as exhibited by Ki67 immunostaining (Fig. 7B). These results are consistent with the notion that the combination treatment of ABT-263 and YM-155 is effective in this colorectal cancer subgroup, genomically defined by KRAS/BRAF mutations and BCL2L1 amplification.

Anti-EGFR agents, including cetuximab and panitumumab, are effective for the treatment of some patients with colorectal cancer (33–35). However, these anti-EGFR treatments are less effective in patients with KRAS or BRAF mutations (11, 36, 37). Therefore, there has been an urgent need to develop new therapeutic strategies for targeting KRAS/BRAF–mutated colorectal cancer patients. In this study, we have identified a genomically defined subgroup having KRAS/BRAF mutations and BCL2L1 amplification. This subgroup, found in as much as 21.7% of colorectal cancers examined, characteristically exhibits high expression of BCL-X_L. Here, we show that these colorectal cancers can be effectively treated using a combination therapy targeting both BCL-X_L (with ABT-263) and MCL1 proteins (with YM-155). We have validated the efficacy of this cotreatment using in vitro cell lines and in vivo PDX models.

To overcome drug resistance in colorectal cancer patients with KRAS/BRAF mutations, multiple combination treatment approaches have already been undertaken. Because KRAS/BRAF mutations result in the activation of MEK/MAPK pathway, MEK inhibitor–based combination treatments have been most often tested, with combination that include a PI3K inhibitor (38), an mTOR inhibitor (39), an AKT inhibitor (40), an IGF1R inhibitor (41), an STK33 inhibition (42), and a BCL-X_L inhibitor (43). In addition, other combination strategies have been attempted, including a combination of BCL-X_L and mTOR inhibitors (12), a combination of EGFR and PI3K inhibitors (44), and a combination of proteasome and topoisomerase inhibitors (45). Among these combination approaches, no single effective therapy in clinic has been established. Furthermore, it is probable that even KRAS/BRAF–mutated colorectal cancers exhibit heterogeneous responses to specific combination treatments.

The novelty of our study is that the dual inhibition of BCL-X_L and MCL1 is consistently effective in colorectal cancer cells with both KRAS/BRAF mutation and BCL2L1 amplification, not in colorectal cancer cells with KRAS/BRAF mutation and BCL2L1 diploid. We showed that colorectal cancer cells with KRAS/BRAF mutation and BCL2L1 diploid exhibited diverse BCL-X_L expression levels (Fig. 1B), and combination of ABT-263 and YM-155 was not highly effective in these cell lines (Fig. 2D; Supplementary Fig. S5D). On the contrary, colorectal cancer cells with both KRAS/BRAF mutation and BCL2L1 amplification showed significantly increased BCL-X_L expression levels (Fig. 1B), and combination of ABT-263 and YM-155 was consistently effective in these cell lines (Fig. 2A and B; Supplementary Fig. S5A and S5B). These results suggest that the colorectal cancer subtype with KRAS/BRAF mutation can be divided into two subgroup according to BCL2L1 copy number alteration (BCL2L1 diploid vs. BCL2L1-amplified), and colorectal cancer cells with both KRAS/BRAF mutation and BCL2L1 amplification can be the target of our combination treatment.

Previous studies found that mutant KRAS increased the expression levels of BCL-X_L via ERK and STAT3-mediated pathways by both transcriptional and posttranscriptional mechanisms (21, 22). BCL-X_L was also reported to be associated with the resistance of MEK and PI3K inhibitors in KRAS-mutant cancers (43, 46). Therefore, overexpressed BCL-X_L in colorectal cancer patients with KRAS mutation is a feasible target of BCL-X_L inhibitor–based combination treatment. In addition, MCL1 levels were considered to be a predictive marker for BCL-X_L inhibitor, ABT-263 response (47). This study demonstrated that, when combined with ABT-263, the YM-155 activates a cell death receptor, leading to caspase-8 and caspase-3–mediated inhibition of the mTOR pathway by degradation of AKT and S6K and subsequent downregulation of MCL1 protein translation. Simultaneous inhibition of BCL-X_L and MCL1, using ABT-263 and YM-155, treatment showed cytotoxic effects in glioblastoma and lung cancer cell lines (48). Recently, a specific MCL1 inhibitor, A-1210477, showed synergistic activity with ABT-263 in pancreas, bladder, lung cancer, and myeloma cell lines (49), and this combination needs to be evaluated for colorectal cancers with KRAS/BRAF mutation and BCL2L1 amplification. Therefore, simultaneous targeting of BCL-X_L and MCL1 is a judicious strategy for KRAS/BRAF–mutated and BCL2L1-amplified colorectal cancers.

Although we suggested that ABT-263 and YM-155 inhibited the activity of BCL-X_L and the expression of MCL1, respectively, these two drugs might have additional targets in inducing cell death. ABT-263 was known to inhibit other BCL2 family proteins including BCL2. We found that colorectal cancer cell lines with KRAS/BRAF mutation and BCL2L1 amplification had lower protein expression levels of BCL2 (Supplementary Fig. S1D), suggesting that the effect of BCL2 inhibition by ABT-263 was minimal in combination treatment of ABT-263 and YM-155. YM-155 was originally developed as a survivin inhibitor, but we exhibited that inhibition of survivin is dispensable for combination effect of ABT-263 and YM-155 via siRNA experiment (Fig. 3A). Therefore, the main targets of ABT-263 and YM-155 are BCL-X_L and MCL1. However, these two drugs can have new targets that are not known until now, and the effect on these new targets could improve the combination effect of the drugs in the treatment of colorectal cancers.

No potential conflicts of interest were disclosed.

Conception and design: S.-Y. Cho, J.Y. Han, D. Na, J.-I. Kim, H. Park, W.-S. Lee, C. Lee

Development of methodology: J.Y. Han, J. Lee, H. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.-Y. Cho, J.Y. Han, W. Kang, A. Lee, J. Kim, J. Lee, S. Min, J. Kang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.-Y. Cho, J.Y. Han, J. Kim, J. Chae

Writing, review, and/or revision of the manuscript: S.-Y. Cho, J.Y. Han, H. Park, C. Lee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.Y. Han, J. Kim, J. Lee

Study supervision: J.Y. Han, J.-I. Kim, H. Park, C. Lee

The authors would like to thank Drs. Asis Das and Dave Mellert for critical reading of this manuscript.

This work was supported by Gil Hospital Research Grant (GCU-2015-5092 and FRD20142802; to W.-S. Lee); the Korean Healthcare Technology R&D project through the Korean Health Industry Development Institute (KHIDI) funded by the Ministry of Health & Welfare, Republic of Korea (grant no. HI13C2148; to C. Lee); the International Research & Development Program of the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (grant no. 2015K1A4A3047851; to C. Lee); and the National Cancer Institute of the NIH (grant P30CA034196). C. Lee is a distinguished Ewha Womans University Professor supported in part by the Ewha Womans University Research grant of 2016.

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.