Vincristine Enhances the Efficacy of MEK Inhibitors in Preclinical Models of KRAS-mutant Colorectal Cancer

Colorectal cancer is a heterogeneous disease with various driver genetic mutations. Metastatic colorectal cancer (mCRC) is the second leading cause of cancer-related deaths in the United States (1). More than 50% of patients with mCRC harbor mutations in the oncogenic driver genes KRAS or NRAS (2). Cancer-associated mutations in KRAS cluster in three hotspots (G12, G13, and Q61); 77% of KRAS mutations are associated with G12. Across different cancers, the predominant G12 variant is KRAS G12D (35%), followed in frequency by KRAS G12V (29%) and KRAS G12C (21%; ref. 3). In patients with mCRC, RAS mutations are associated with poor prognosis (4). Direct targeting of most RAS mutants has been technically challenging. However, encouraging progress has been made with the development of small molecules that selectively and irreversibly bind to KRAS G12C (5). Recent development of a high-affinity, noncovalent small molecule against KRAS G12D, illustrated the therapeutic feasibility of targeting this mutant variant (6). Based upon the results of phase II studies, the KRAS G12C inhibitor sotorasib was approved by the FDA for the treatment of lung cancer; however, it failed to significantly improve outcomes for patients with mCRC (7, 8). Currently, combination therapies sotorasib with panitumumab (9), and adagrasib with cetuximab (10) are showing promising results for patients with mCRC harboring KRAS G12C mutations. Several other combination regimens, such as sotorasib with an SHP2 inhibitor, are presently in clinical trials (NCT05480865). However, KRAS G12C is only present in 3% of patients with mCRC and effective therapies for the remaining patients with mCRC harboring RAS mutations are urgently needed.

To target KRAS-driven tumors, indirect inhibition of downstream mediators of RAS, such as the mitogen-activated protein kinase kinase (MEK), has been studied (11). However, single-agent targeting of MEK has not been clinically effective, especially in patients with mCRC. Clinical trials of combinations of PI3K/AKT/mTOR (12–16), BCL-XL (17), or CDK4/6 (18–20) inhibitors with MEK inhibitors have also had only limited clinical success. Recently, MEK inhibitors in combination with autophagy inhibitors in colorectal cancer and pancreatic cancer have shown promise in vitro and are being further studied in anticipation of advancing to clinical trials (21–23). However, to identify successful combination therapies that improve the efficacy of MEK inhibitors in patients with RAS-mutant tumors, especially those with mCRC, further efforts are warranted.

In this study, we hypothesized that an unbiased screening approach could identify the most effective chemotherapy or targeted therapy agent for use in combination with a MEK inhibitor. Furthermore, there has been a shift in the field of high-throughput screening (HTS) to integrate more physiologically relevant model systems that better recapitulate the tumor microenvironment, such as establishing cell–cell and cell–matrix interactions (24, 25). Hence, we chose to establish a three-dimensional (3D) HTS assay using KRAS-mutant colorectal cancer spheroids to test drug combinations. The MEK inhibitor trametinib was used as the base drug and tested for synergistic effects with two “clinically ready” compound libraries of approved and phase III drugs. We determined that vincristine was synergistic with trametinib in KRAS-mutant colorectal cancer spheroids. The efficacy of the combination therapy was further evaluated using both in vitro and in vivo models. Our unbiased HTS along with in vitro and in vivo validation studies demonstrated that combining trametinib with vincristine can enhance the efficacy of MEK-targeted therapy in KRAS-mutant colorectal cancer. Thus, our preclinical studies serve as the basis for future clinical studies to determine the efficacy of this drug combination in patients with KRAS-mutant mCRC.

KRAS-mutant human colorectal cancer cell lines HCT116 (RRID: CVCL_0291), SW620 (RRID: CVCL_0547), LS174T (RRID: CVCL_1384), LoVo (RRID: CVCL_0399), and SW480 (RRID: CVCL_0546) were purchased from ATCC. The pair of parental SW620 and SW620/AD300 cell lines were generous gifts from Ze-Sheng Chen (St John's University, New York, NY) and Susan Bates (Columbia University, NY; refs. 26, 27). The colorectal cancer cells were cultured in minimum essential medium with 5% FBS (Atlanta Biologicals). Media were also supplemented with recommended concentrations of vitamins, nonessential amino acids, penicillin/streptomycin, sodium pyruvate, and l-glutamine (Thermo Fisher Scientific). In vitro experiments were performed using cells that were within 15 passages. Cell lines were validated at the MD Anderson Cancer Center Characterized Cell Line Core Facility using short tandem repeat DNA profiling and the MycoAlert Mycoplasma detection kit (Lonza. catalog no.: LT07-218) was used to confirm that all cell lines were free of Mycoplasma during the experiments. The cell number and viability were determined using a Cellometer (Nexelom Biosciences) according to the manufacturer's instructions before plating for each in vitro experiment.

The optimal cell seeding density and ability to form a uniform, sustainably growing spheroid in 384-well ultra-low attachment (ULA) round (U)-bottom plates (Corning, catalog no. 3830) were determined in a preliminary characterization assay. Five cell seeding densities were tested by performing a 1:1 serial dilution starting with the maximal target dilution of 3,000 cells/well. Cells were plated into ULA U-bottom plates with a multichannel pipette for the characterization assay and by Multidrop dispenser (Thermo Fisher Scientific) for the HTS assays. The cells were then incubated in a robotically integrated Cytomat 6,000 cell culture incubator (Thermo Fisher Scientific) for 10 days in a humidity-controlled (>95%) cell incubator at 37°C in an atmosphere of 5% CO₂. From this analysis, three KRAS-mutant colorectal cancer cell lines, HCT116 (seeding density 190 cells/well), SW620 (3,000 cells/well), and LS174T (1,500 cells/well), were identified as robust 3D models amenable to spheroid-based screens.

For screening assays, drug treatment was performed 48 hours after plating, which was adequate time for all models to form spheroids. For the primary combination screen, we tested a library consisting of 112 FDA-approved and phase III investigational drugs acquired from the NCI (NCI_AOD_vr5), both as single agents and in multipoint pairwise combinations with trametinib. All drugs were diluted in DMSO on Echo-certified low dead volume plates (Labcyte) and were transferred from the source plates into assay plates using an Echo 550 liquid handling platform (Labcyte). For primary screening assays, a fixed range of probe concentrations were tested (1.0 μmol/L, 300 nmol/L, 100 nmol/L, 30 nmol/L). For validations, a tuned range was determined from the single-agent response where the highest concentration would reflect approximately a 50% response. Wells containing spheroids were treated with a fixed amount of DMSO 0.5% [volume for volume (v/v)]; for combination assays DMSO was backfilled to a 0.5% (v/v) final concentration. Wells containing DMSO and media only served as on-plate negative controls.

For primary screens, an area-based readout was used. Here assay plates were serially imaged on an imageXpress microconfocal using a 10x NA = 0.45 objective. A single centrally located z-plane was captured using brightfield, FITC (Cytosolic GFP, reflective of translationally active live cells), and Cy5 (Draq7, Dead Cells). Image analysis was performed using a custom script develop in using the advanced imaging collection of Pipeline Pilot (RRID: SCR_014917), which segmented the live-cell Cytosolic GFP area. The raw area was then normalized using the Hafner growth rate equation, such that growth/shrinkage could be evaluated. For validation assays, an orthogonal viability readout was used to test that generalizability of the results. After a 7-day drug exposure, plates were leveled to 35 μL using a HydroSpeed plate aspirator (Tecan), and CellTiter-Glo 3D (Promega, catalog no. G9681) was added in a 1:1 (v/v) ratio. Plates were then incubated for at least 30 minutes at room temperature and luminescence were measured on a Synergy Neo2 plate reader (Agilent BioTek).

High-throughput combinatorial screens were performed by testing varying ratios of the anchor and probes. The data were then fit to a 3D surface using a support vector machine–based method to provide additional rigor and automated outlier detection as described previously (28). The Bliss independence model was then used to calculate the theoretical additivity surface (29). Interactions between the two drugs were then characterized as antagonistic, additive, or synergistic by comparing the empirically determined drug effect with the calculated Bliss independence. As a subjective cutoff, we used a volumetric difference (i.e., the sum of all pairwise interactions) of −1 or 1 to define antagonism or synergy, respectively.

For cell viability assays, colorectal cancer cells (1,000–3,000 cells/well) were plated in 96-well flat-bottom plates. After 24 hours, cells were treated with trametinib, vincristine, or a combination (Selleck Chemicals). For the combination, various concentrations of vincristine were combined with a fixed concentration of trametinib (5 nmol/L). DMSO was used as a control. The cells were allowed to grow further for 48 hours with or without drugs, and cell survival was measured using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reagent (30).

For colony formation assays, colorectal cancer cells were plated in 12-well dishes (1,000–3,000 cells/well). After 24 hours, trametinib, vincristine, or a combination was added, and cells were allowed to grow for an additional 7 days. DMSO was used as the control. The surviving colonies were stained with 0.05% methylene blue solution and imaged. Then, 1% sodium dodecyl sulfate was used to extract the cell-bound dye, and the intensity of the colored solutions was measured at 600 nm.

For two-dimensional (2D) assays, HCT116, SW620, SW480, LS174T, and LoVo cells (0.1 × 10⁶ to 0.3 × 10⁶ cells/well) were seeded in 6-well plates. After 24 hours, cells were treated with trametinib (5 nmol/L), vincristine (2 nmol/L), or the combination and allowed to grow for 24 hours. For 3D assays, HCT116 and SW620 cells (1 × 10³ cells/well) were seeded in ULA U-bottom 96-well plates and incubated for 72 hours to form spheroids. Spheroids were treated with trametinib, vincristine, or their combination and allowed to grow for 48 hours. Proteins were extracted from cells and spheroids and separated by SDS-PAGE following a standard protocol and transferred to Immobilon polyvinylidene membranes (EMD Millipore). The membranes were blocked with 5% milk in TBS with 0.1% Tween 20 (TBST) for 1 hour, followed by incubation with a primary antibody (diluted in 3% BSA in TBST) overnight at 4°C. Membranes were then washed three times in TBST and reincubated with horseradish peroxidase–labeled secondary antibodies for 1 hour. Finally, the membranes were washed three times in TBST and exposed to autoradiography films. Signals were detected by chemiluminescence (Thermo Fisher Scientific). Antibodies for cleaved PARP (catalog no. 9541; RRID:AB_331426), cleaved caspase 8 (catalog no. 9496, RRID:AB_561381), p-SAPK/JNK (catalog no. 4668, RRID:AB_823588), p27 Kip1 (catalog no. 2552, RRID:AB_10693314), mTOR-pS2448 (catalog no. 2971, RRID:AB_330970), p-P70-S6K (catalog no. 9205, RRID:AB_330944), and p-eIF4B (catalog no. 3591, RRID:AB_2097522) were from Cell Signaling Technology), and vinculin (catalog no. sc-25336, RRID:AB_628438) was from Santa Cruz Biotechnology. All antibodies were used according to manufacturers’ instructions. All cell lysates for 2D cultures were prepared in RIPA with protease and phosphatase inhibitors as described previously (31). For the lysate preparation from 3D cultures, the spheroids were suspended in RIPA buffer and lysed by sonication.

HCT116 and SW620 cells (0.1 × 10⁶ to 0.3 × 10⁶ cells/well) were seeded in 6-well plates. After 24 hours, cells were treated with trametinib (5 nmol/L), vincristine (2 nmol/L), or the combination and allowed to grow for another 24 hours. Cell lysates were prepared as described above. Reverse phase protein array (RPPA) analyses were performed at MD Anderson Cancer Center's Functional Proteomics RPPA Core facility as described previously (32).

HCT116 and SW620 cells (0.1 × 10⁶ to 0.3 × 10⁶ cells/well) were seeded in 6-well plates. After 24 hours, cells were treated with trametinib (10 nmol/L) for 48 hours or positive control verapamil (10 μmol/L) for 24 hours. Then daunorubicin (10 μmol/L) was added singly or in combination with trametinib or verapamil for 2 hours. Cells were then washed to remove excess daunorubicin, collected by trypsinization and analyzed by flow cytometry. Daunorubicin levels were detected using a Beckman Coulter Gallios instrument using the 575 nm band pass filter.

For immunofluorescence staining, HCT116 and SW620 cells (0.1 × 10⁶ to 0.3 × 10⁶ cells/well) were seeded in 6-well glass-bottom plates (Chemglass Life Sciences, catalog no. CLS-1812-006). After 24 hours, cells were treated with trametinib (10 nmol/L) for 48 hours. For positive control, cells were incubated with verapamil (10 μmol/L) for 24 hours. Next, daunorubicin (10 μmol/L) was added singly or in combination with trametinib or verapamil for 2 hours, followed by replacement of the drug-treated media with fresh media, and images were obtained using a fluorescence microscope (Olympus IX71) using a 20× objective lens. Because of its autofluorescence, daunorubicin was visualized using the RFP filters. Quantification of the images were done using ImageJ (RRID:SCR_003070). From each image, cells (n = 20, randomly selected) were outlined with the Freehand ROI tool. Integrated density for each cell was measured. A number of small areas of the image that had no fluorescence were selected and fluorescence was measured. This served as the background. Mean fluorescence of background readings was calculated. Corrected total cell fluorescence (CTCF) for each cell was calculated as described previously (33) using the following formula: CTCF = Integrated Density – (Area of Selected Cell × Mean Fluorescence of Background readings).

For immunostaining of mitotic cells, HCT116 cells (0.1 × 10⁶ cells/well) were seeded in 6-well glass-bottom plates (Chemglass Life Sciences, catalog no. CLS-1812-006). After 24 hours, cells were either untreated or treated with trametinib (5 nmol/L), vincristine (2 nmol/L), or their combination for 48 hours. Cells on coverslips were fixed with chilled methanol and rehydrated in TBST. Cells were then incubated with a monoclonal anti-α-tubulin antibody (Santa Cruz Biotechnology, catalog no. sc-32293, RRID: AB_628412) diluted at 1:100 in TBST + 3% BSA for 1 hour at room temperature. Following washing, the coverslips were reincubated with an anti-mouse secondary antibody labeled with Alexa-488 (Thermo Fisher Scientific, catalog no. A-11001, RRID: AB_2534069) diluted at 1:500 in TBST + 3% BSA. DNA was labeled using Hoechst 33342 (1 μg/mL; Thermo Fisher Scientific, catalog no. H1399). Cover slips were washed in TBST for 15 minutes and mounted on glass slides. Tubulin and DNA were visualized and imaged using a fluorescence microscope (Olympus IX71) using a 60× oil objective lens.

To determine the relative abundance of intracellular vincristine, extracts were prepared and analyzed by ultra-high-resolution mass spectrometry. Approximately 1 × 10⁶ HCT116 cells were plated on 100 cm petridishes in triplicate and treated with or without trametinib (10 nmol/L) for 48 hours. Cells were then incubated with vincristine (10 nmol/L) for 4 hours. Then the cells were quickly washed with ice-cold PBS to remove extra medium components. Metabolites were extracted using cold methanol and extracts were centrifuged at 17,000 × g for 5 minutes at 4°C, and supernatants were transferred to clean tubes, followed by evaporation to dryness under nitrogen. Dried extracts were reconstituted in 100 μL of 90/10 Acetonitrile/Water, and 10 μL was injected for Vincristine analysis by LC/MS. LC mobile phase A was 95/5 (v/v) water/acetonitrile containing 20 mmol/L ammonium acetate and 20 mmol/L ammonium hydroxide (pH∼9), and mobile phase B (MPB) was acetonitrile. The mobile phase flow rate was 300 μL/minute under isocratic elution program with 85% MPB. The total run time was 10 minutes. Data were acquired using a Thermo Orbitrap Exploris 240 Mass Spectrometer with electrospray ionization (ESI)-positive ionization mode at a resolution of 240,000. Raw data files were imported to Thermo Trace Finder software for final analysis. Known amounts of pure vincristine were used to generate standard curves and the relative abundance of each metabolite was normalized by cell numbers.

Three KRAS-mutant (G12C, G12D, G13D) colorectal cancer patient-derived xenografts (PDX) were used in this study. They were obtained from a repository at MD Anderson Cancer Center through a collaboration with Dr. E. Scott Kopetz.

All mice used in the in vivo studies were obtained from Experimental Radiation Oncology at MD Anderson Cancer Center. PDXs that had been verified to have KRAS mutations (PDX C1117 having KRAS G12D mutation, PDX C1138 having G13D, and PDX B8239 having G12C) were initially grown subcutaneously in male NSG mice as described previously (18, 34). When tumors reached approximately 1 cm in diameter, mice were euthanized and the tumors were harvested. The tumors were then dissected into small (∼2 mm) pieces. The tumor pieces were then implanted subcutaneously into the flanks of anesthetized 4–6 weeks old nude mice (female mice for PDX C1138; male mice for PDX C1117 and PDX B8239) and allowed to grow to approximately 100–200 mm³. Animals were then randomly assigned to treatment with vehicle, trametinib (0.2 mg/kg, 5 days/week, orally), vincristine (0.5 mg/kg, once/week, intravenously), or a combination of trametinib and vincristine (10 mice/ treatment group) for nearly 3 weeks. Trametinib (Selleck Chemicals) was prepared in a suspension of 0.5% H-methyl cellulose and 0.5% Tween 80 and was administered by oral gavage 5 days a week. Vincristine (MD Anderson Cancer Center Pharmacy) was prepared in 0.9% saline solution and was administered intravenously once a week. Tumor size was measured using digital calipers and mice were weighed twice a week by observers who were blinded to the treatment groups. At the end of the experiments, mice were euthanized by CO₂ inhalation and tumors were harvested and weighed. All animal experiments were performed in compliance with institutional guidelines and regulations after approval by the Institutional Animal Care and Use Committee (IACUC) at MD Anderson Cancer Center.

Note that only for studies with PDX C1138, owing to differences in take rate and growth kinetics, mice were initially randomized into four treatment groups with five animals in each group to ensure similar tumor sizes at the start of the experiment. After a few days, when tumor sizes were adequate to initiate the experiments, the remaining mice were again randomly assigned to four groups of 5 mice each. The treatment schedule was the same as that of the prior group to reduce experimental divergence.

To evaluate the toxicity profile, nude mice were randomly assigned to treatment with vehicle, trametinib, vincristine, or a combination of trametinib and vincristine. At the end of treatment, mice were euthanized, and the heart, liver, lung, spleen, and a piece of small intestine were collected for histologic staining and examination by the MD Anderson Cancer Center Department of Veterinary Medicine and Surgery.

All in vivo experiments utilizing PDXs were performed according to NIH NCI recommendations summarized in SOP50102: PDX Implantation, Expansion and Cryopreservation (Subcutaneous). Tumor specimens were obtained from patients with mCRC under a research laboratory protocol (LAB10-0982) approved by UT MD Anderson Cancer Center Institutional Review Board, and all patients provided written informed consent for specimens to be used for research purposes including implantation in xenografts. Xenografts were established in 6–8 weeks old female NSG mice. All in vivo studies to determine drug efficacy were performed using 6–8 weeks old nude mice. Either male or female mice were used for experiments with a specific PDX. All studies were performed in accordance with accepted guidelines for housing, euthanasia and treatment, under a protocol (# 00001368-RN00) approved by UT MD Anderson Cancer Center IACUC.

Tumors collected at the experimental endpoint were fixed in formalin and embedded in paraffin by Research Histology Core Laboratory at MD Anderson Cancer Center. For IHC staining, tissue sections were first deparaffinized in xylene and 100% ethanol twice for 10 minutes each, then in 70% and 50% ethanol for 5 minutes each. Next, the tissues were washed in distilled water and subjected to an antigen retrieval process using antigen unmasking solution, citrate buffer (pH 6.0; Invitrogen) at 95°C for 30 minutes. Sections were then washed and allowed to cool in distilled water. ImmPRESS universal reagent (Vector Laboratories, catalog no. MP-7500) was used according to the manufacturer's protocol for blocking. After blocking, sections were incubated with primary antibodies against Ki-67 (1:250, Cell Signaling Technology, catalog no. 9027, RRID: AB_2636984) and cleaved caspase 3 (1:250, Cell Signaling Technology, catalog no. 9661, RRID: AB_2341188). ImmPRESS universal reagent was also used according to the manufacturer's protocol for the secondary antibody step. Immunoreactivity was visualized by using a DAB substrate kit (Cell Marque, catalog no. 957D-60) according to the manufacturer's protocol. Stained slides were imaged using an Aperio CS2 microscope slide scanner (Leica Biosystems, RRID:SCR_020993). For quantification of both Ki-67 and cleaved caspase 3, Aperio ImageScope software was used to calculate the percentage of positive nuclei.

Graphical representations and statistical analysis were done using GraphPad Prism 9 (RRID:SCR_002798) and Microsoft Excel (RRID:SCR_016137). Two-tailed Student t tests were performed to compare groups. Results are expressed as mean ± SEM. P < 0.05 was considered as significant. For in vitro assays, all quantitative values represent at least three replicates. For in vivo assays, 8–10 tumors were measured for each treatment group.

The data generated in this study are available within the article and its Supplementary Data.

We performed unbiased HTS using the NCI-approved Oncology Library (version 5) and a small, focused library of clinical oncology drugs. We tested the drugs in the libraries as single agents or in multiple pairwise concentrations with trametinib, using HCT116 spheroids as a prototypical KRAS-mutant colorectal cancer model. From the single-agent screens, we found that 124 drugs with a broad range of mechanisms of action (Supplementary Data S1) resulted in robust inhibition of spheroid growth or reduction of spheroid size. We then used a combinatorial drug screening approach to identify drugs that either additively or synergistically improved the efficacy of trametinib (Fig. 1A). From this analysis, we found that proteotoxic agents, microtubule inhibitors, anthracyclines, and topoisomerase inhibitors had either highly additive or synergistic cytotoxic effects (Supplementary Data S2). From this screen, one of the most efficacious combinations was trametinib with the mitotic inhibitor vincristine (Supplementary Data S2). This prompted us to perform orthogonal validations in HCT116 in addition to SW620 and LS174T to further establish generalizability, which similarly identified vincristine as a top-performing synergistic combination (Fig. 1B and C). Collectively, these initial data provided a strong rationale to advance vincristine in combination with trametinib into additional model systems.

The effect of combining trametinib with vincristine on long-term cell viability was determined by colony formation assays in a panel of KRAS-mutant colorectal cancer cells. The extent of cell growth retardation was also measured by extracting the accumulated dye in each well. The combination of trametinib and vincristine more effectively suppressed cell growth and colony formation than did the monotherapies in HCT116, SW620, and LS174T cell lines (Fig. 2A). Similar effects were also seen in KRAS-mutant LoVo and SW480 cell lines (Supplementary Fig. S1A).

We further validated the effect of combining trametinib and vincristine on cell viability by MTT assays in KRAS-mutant colorectal cancer cell lines. The combination showed a greater effect on the inhibition of cell survival than did trametinib or vincristine alone or DMSO control in HCT116, SW620, and LS174T cell lines (Fig. 2B).

We next examined the effect of combining trametinib and vincristine on induction of apoptosis and inhibition of cell-cycle progression by Western blot analysis in HCT116 and SW620 cell lines and cell line–derived 3D spheroids. [Note: We used fixed concentrations of trametinib (5 nmol/L) and vincristine (2 nmol/L) for these studies based on their ability to partially inhibit colorectal cancer cell growth as single agents but reliably demonstrate enhanced drug efficacy in short term in vitro studies.] Markedly higher levels of three markers of apoptosis cleaved PARP, cleaved caspase 8, and p-SAPK/JNK, and a marker of cell-cycle progression inhibition, p27/Kip-1, were observed in cell lines and spheroids treated with the drug combination compared with untreated or single agent–treated cell lines and spheroids (Fig. 3A). Similar effects were also seen in KRAS-mutant LS174T, LoVo, and SW480 cell lines (Supplementary Fig. S1B).

These effects of the combination were validated by RPPA using protein lysates from HCT116 and SW620 cell lines, which showed higher levels of cleaved caspase 7, pH2AX, p-SAPK/JNK, PAR, and p27/Kip-1 in lysates of cells treated with the trametinib and vincristine combination than in those treated with trametinib or vincristine alone (Fig. 3B). Together, these findings indicate that the combination of trametinib and vincristine synergistically induces apoptosis and inhibits cell-cycle progression.

To confirm the antitumor effects observed in our in vitro data, we treated mice bearing PDXs with three different KRAS mutations with trametinib, vincristine, or their combination. In all three PDX models (C1117, C1138, and B8239), the combination of trametinib and vincristine resulted in significantly lower tumor volumes than did control or single-agent treatments (Fig. 4A, C, and E; Supplementary Fig. S3A). At the end of the experiments, the average tumor weight was also lower in the combination group than in either monotherapy group or the control group (Supplementary Figs. S2 and S3B). Representative images of the tumors are shown in Supplementary Fig. S2.

Finally, after the end of the treatment regimen, IHC staining for Ki-67 and cleaved caspase 3 was done from the harvested tumors. In all the PDX models, the tumors that were treated with the combination of trametinib and vincristine had significantly lower expression of the cell proliferation marker Ki-67 compared with the single agents. The combination-treated tumors also exhibited significantly higher expression of the apoptosis marker cleaved caspase 3 compared with the monotherapies. Representative images of the stained tumor tissues and graphs of staining quantification are shown in Fig. 4B, D, and F. Together, these results validate our in vitro data and support our hypothesis that the combination of vincristine and trametinib induces tumor growth inhibition via cooperative induction of apoptosis.

In terms of the toxicity profile, trametinib and vincristine combination treatment did not cause a significant decrease in the body weight of mice during the experimental period (Supplementary Fig. S4A). The histopathologic reports of the organs collected from combination-treated mice showed no indications of drug-related toxicity, suggesting that the drug combination was well tolerated (Supplementary Fig. S4B).

To determine the mechanism of action of the drug combination, RPPA was performed using protein lysates from HCT116 and SW620 cell lines. These analyses demonstrated significantly lower levels of p-mTOR, eIF4G, and P70-S6K, signaling molecules of possible bypass mechanisms of the RAS-RAF-MEK pathway, in the combination treatment group compared with the monotherapy-treated groups in both cell lines (Fig. 5A). Validating these effects of the drug combination, Western blot analyses confirmed significantly lower levels of p-mTOR, p-P70-S6K, and p-eIF4B in the combination-treated cells than in the single agent–treated cells (Fig. 5B). These findings suggest that the combination of vincristine and trametinib, but not the single agents, significantly inhibits both activation of mTOR and the MAP kinase pathway (Fig. 5C).

Overexpression of ATP-binding cassette (ABC) transporters, predominantly ABCB1, is one of the foremost drivers of multidrug resistance in cancer cells (35, 36). ABCB1 can transport chemotherapeutic drugs such as vincristine and daunorubicin out of cancer cells (35). Inhibitors of these transporters can restore chemotherapy sensitivity in resistant cancer cells (37). Trametinib is known to interact with and inhibit the function of ABCB1 (38). Figure 6A shows a proposed model, based on the literature (38), showing the activity of ABCB1 on chemotherapeutic drug efflux in the absence and presence of trametinib. To determine whether trametinib inhibits the drug efflux activity of ABCB1, we measured changes in the intracellular accumulation of vincristine and daunorubicin. Mass spectrometric analyses of HCT116 cells demonstrated that intracellular levels of vincristine were substantially (∼2.5-fold) increased in the presence of trametinib (Fig. 6B). We utilized the autofluorescence of daunorubicin to measure its levels via flow cytometry and immunofluorescence staining in colorectal cancer cells. Verapamil, a positive inhibitor of ABCB1, was used to compare the activity of trametinib. Flow cytometric analysis showed that intracellular levels of daunorubicin were significantly increased following trametinib treatment in both the cell lines and that this effect was greater than that of verapamil (Supplementary Fig. S5). Immunofluorescence staining further confirmed the higher intracellular levels of daunorubicin in both cell lines treated with trametinib as compared with controls (Supplementary Fig. S5).

We also examined the effects of combining trametinib with vincristine on SW620/AD300 cells that express elevated levels of ABCB1 (26). We reasoned that while trametinib can enhance activity of vincristine in wild-type SW620 cells leading to synergistic effects, overexpression of ABCB1 in SW620/AD300 cells would reduce the ability of trametinib to enhance activity of vincristine. We found that while trametinib can significantly enhance the efficacy of vincristine in wild-type SW620 cells, there were no measurable increases in efficacy in the ABCB1 overexpressing SW620/AD300 cells under similar experimental conditions (Fig. 6C). These results suggest that trametinib can likely inhibit ABCB1 activity, thereby causing accumulation of chemotherapeutic drugs such as vincristine and daunorubicin in colorectal cancer cells and leading, in turn, to enhanced cytotoxicity in the presence of the drug combination.

It is expected that if trametinib enhances the intracellular concentration of vincristine, higher concentration of vincristine would likely result in enhanced defects in colorectal cancer cell division. We measured the number of mitotic cells with normal or defective spindles in presence of single or combination of the drugs. Colorectal cancer cells treated with the combination of trametinib and vincristine demonstrated significantly higher number of cells with defective mitotic spindles as compared with vincristine or trametinib alone (Fig. 6D).

KRAS-mutant colorectal cancer tumors have shown short-term sensitivity to MEK and RAF inhibitors, and vertical inhibition of the RAS-RAF-MEK pathway is not enough to inhibit RAS oncogenic functions. Adaptive resistance through activation of the PI3K/AKT pathway has been a significant obstacle in achieving durable remission with MEK inhibitors (39).

In this study, we identified vincristine as synergistic with trametinib in KRAS-mutant colorectal cancer models. Our data suggested that the combination: (i) led to increased cytotoxicity in both colorectal cancer cell lines and spheroids, (ii) reduced mTOR activation in colorectal cancer cell lines, (iii) suppressed tumor growth and promoted caspase-dependent apoptosis in colorectal cancer PDX models, and (iv) has no indication of drug-related toxicity in mice.

We used unbiased HTS to identify a combinatorial strategy using a MEK inhibitor and a chemotherapeutic or targeted-therapy agent to treat KRAS-mutant colorectal cancer tumors. Usually, 2D screens are used in HTS to identify combination therapies. However, multicellular 3D spheroids are of intermediate complexity between in vivo tumors and monolayer cultures and their growth emulates the heterogeneity of solid tumors with varying degrees of necrosis and hypoxia (24). The ability of multicellular spheroids to simulate a tumor-like environment (25), together with the ease at which these models can be applied in high-throughput screen makes this a better model for the identification of novel therapeutics using HTS. We used 3D colorectal cancer spheroids in these HTS studies because they can best identify clinically relevant drug combinations (40). So that our findings can be quickly translated to the clinic, we used the MEK inhibitor trametinib (approved by the FDA for melanoma) as the anchor drug with two different “clinically ready” compound library sets. Vincristine was identified as synergistic with trametinib in KRAS-mutant colorectal cancer spheroids.

Note: Vinorelbine, a drug in the same class as vincristine was also identified in our initial HTS to be synergistic with trametinib (Supplementary Fig. S6). Identification of a second drug, vinorelbine, in the same functional class validated the initial observations with vincristine. Also, this provides with a second drug combination that can be clinically examined should the initial drug combination result in significant drug-related toxicities in patients.

We validated our initial screening results showing synergistic effects of combining trametinib and vincristine by using different in vitro approaches in multiple KRAS-mutant colorectal cancer cell lines. Clonogenic assays are used to measure the ability of cells to retain tumor-initiating capabilities over a prolonged period of time. Long-term exposure to the combination resulted in significantly reduced colony formation and cell proliferation in colorectal cancer cell lines with different KRAS mutations, suggesting that cotargeting by MEK and microtubule inhibitors (trametinib and vincristine) led to increased cytotoxicity. We also evaluated the changes in apoptosis-related proteins following the combination treatment. We observed higher levels of the apoptosis markers cleaved PARP, cleaved caspase 8, and p-SAPK/JNK in the combination group than in single agent–treated or control-treated cells. These observations were further validated by RPPA studies showing higher levels of cleaved caspase 7, p-H2AX, p-SAPK/JNK, and PAR, indicating that the drug combination led to an increase in apoptosis. Altogether, data from our in vitro studies confirmed that the efficacy of trametinib was significantly enhanced by the addition of vincristine and led to substantial increases in cell death in KRAS-mutant colorectal cancer cells.

To demonstrate the preclinical relevance of combining trametinib with vincristine, we examined three different KRAS-mutant PDX models in mice. The drug doses used in the mice were equivalent to those used in humans and were determined as described previously (41). The drug doses were also verified in light of previous literature demonstrating that similar doses led to target inhibition (42, 43). Consistent with our in vitro findings, the drug combination in vivo significantly suppressed tumor growth and promoted caspase-dependent apoptosis, marked by increased levels of cleaved caspase 3, compared with tumors treated with single agents.

We further elucidated the molecular mechanisms underlying the enhanced cytotoxic effects of combining trametinib and vincristine in colorectal cancer cells. We found that the combination of trametinib and vincristine decreased the levels of p-mTOR and downstream components of the mTOR pathway, such as P70-S6K, eIF4B, and eIF4G; however, neither single agent by itself had a significant effect on mTOR or its downstream molecules. Two major targetable cell signaling pathways in cancer are the PI3K-AKT-mTOR and the RAS-RAF-MEK pathways (44). The components of these two pathways engage in cross-talk (45, 46). The PI3K pathway downstream component mTOR is critical for cellular growth (44), and the PI3K-AKT-mTOR pathway can act as a compensatory signaling pathway when MEK is inhibited (47). Thus, the combination of trametinib and vincristine, by reducing mTOR activity, can slow cell growth and increase apoptosis. A previous study demonstrated that AMP-activated protein kinase–dependent inhibition of mTOR complex 1 is involved in vincristine-induced apoptosis of melanoma cells (48). However, we found that mTOR activity was not significantly reduced by vincristine alone at clinically relevant concentrations. Only when combining trametinib with vincristine did we observe reduced mTOR activity. However, why vincristine in combination with trametinib reduces mTOR activation requires further study.

Vincristine is a vinca alkaloid that exerts its cytotoxic effects mainly by binding to tubulin dimers, thereby inhibiting the assembly of microtubule structures, arresting mitosis in metaphase and causing apoptosis in cells undergoing mitosis (49). Overexpression of ABC transporters, particularly ABCB1, is one of the most common causes of vincristine resistance (35), and ABCB1 is highly expressed in colorectal cancer tumors (50). ABCB1 acts as a drug efflux pump, lowering the intracellular concentration of cytotoxic drugs such as vincristine, daunorubicin, paclitaxel, and etoposide (35, 51). Trametinib has been shown to interact with ABCB1 on its large drug-binding cavity, thereby blocking ABCB-1–induced drug efflux and increasing the intracellular concentration of other cytotoxic drugs (38). We performed mass spectrometry to determine intracellular levels of vincristine and utilized the inherent fluorescence to measure intracellular levels of daunorubicin. These studies demonstrated that trametinib can lead to significant increases in the intracellular accumulation of these cytotoxic drugs. Increases in intracellular vincristine likely led to higher microtubule damage and thereby resulted in increased mitotic defects and colorectal cancer cell death.

Systemic and targeted therapies are required for patients with metastatic colorectal cancer; preclinical in vivo models that recapitulate metastatic disease would be ideal for studies that examine efficacy of investigational therapeutics. Thus, demonstrating that the combination of trametinib and vincristine is more effective using in vivo models to mimic liver metastasis, the major site of colorectal cancer metastasis, is ideal. However, developing liver metastases using intrasplenic or intravenous injections in a large cohort of animals is technically difficult, especially while transplanting colorectal cancer PDXs. Thus, we have utilized the widely accepted methodology of developing subcutaneous tumors that not only allows implantation of PDXs but also allows easy measurement of tumor growth. Although this is one of the shortcomings of our studies, we posit that these carefully conducted proof-of-principle studies using the simpler subcutaneous model demonstrates enhanced drug efficacy. Future studies using liver metastatic models will likely further validate these findings.

The combination of trametinib and vincristine resulted in strong growth inhibition in vitro. In vivo, while the drug combination demonstrated significant decreases in tumor volume, differences in average tumor weight were modest as compared with trametinib alone. However, we reason that changes in tumor volume, similar to how tumor responses to therapeutics are measured in patients, may better demonstrate the effects of the drug combination. While challenges exist in translating preclinical studies to clinical settings, we believe our studies using clinically relevant doses of the drugs primarily targeting the tumor cells themselves and colorectal cancer PDXs present a strong case for further evaluation of this drug combination in clinical studies.

In summary, we performed unbiased 3D HTS studies that identified vincristine as a synergistic agent with the MEK inhibitor trametinib in KRAS-mutant colorectal cancer. We found combinatorial effects that led to significant cell growth inhibition, decreased cell survival and proliferation, and increased apoptosis in vitro. The combination of trametinib and vincristine, used at clinically achievable doses, also significantly suppressed tumor growth in vivo in different KRAS-mutant PDXs. Toxicity profiling of the combination demonstrated that it was well tolerated by mice. Hence, our studies indicate that this drug combination is effective preclinically in the treatment of KRAS-mutant colorectal cancer and strongly supports further clinical studies to determine its efficacy in patients with KRAS-mutant mCRC.

S. Kopetz reports other support from Lutris, Iylon, Frontier Medicines, Xilis, Navire, Genentech, EMD Serono, Merck, Holy Stone Healthcare, Novartis, Lilly, Boehringer Ingelheim, AstraZeneca/MedImmune, Bayer Health, Redx Pharma, Ipsen, HalioDx, Lutris, Jacobio, Pfizer, Repare Therapeutics, Inivata, GlaxoSmithKline, Jazz Pharmaceuticals, Iylon, Xilis, Abbvie, Amal Therapeutics, Gilead Sciences, Mirati Therapeutics, Flame Biosciences, Servier, Carina Biotech, Bicara Therapeutics, Endeavor BioMedicines, Numab, Johnson & Johnson/Janssen, Genomic Health, Frontier Medicines, Replimune, Taiho Pharmaceutical, Cardiff Oncology, Ono Pharmaceutical, Bristol-Myers Squibb-Medarex, Amgen, Tempus, Foundation Medicine, Harbinger Oncology, Takeda, CureTeq, Zentalis, Black Stone Therapeutics, NeoGenomics Laboratories, Accademia Nazionale Di Medicina, Sanofi, Biocartis, Guardant Health, Array BioPharma, Genentech/Roche, EMD Serono, MedImmune, Novartis, Amgen, Lilly, and Daiichi Sankyo outside the submitted work. L.M. Ellis reports other support from Fibrogen, New Beta Innovations, Oncohost, and Actuate Therapeutics. No disclosures were reported by the other authors.

S. Ghosh: Data curation, investigation, writing–original draft, writing–review and editing. F. Fan: Investigation, methodology. R.T. Powell: Formal analysis, methodology, writing–review and editing. J. Roszik: Formal analysis, visualization. Y.S. Park: Investigation. C. Stephan: Formal analysis, supervision. M. Sebastian: Formal analysis. L. Tan: Investigation. A.V. Sorokin: Resources, methodology. P.L. Lorenzi: Investigation, methodology. S. Kopetz: Resources. L.M. Ellis: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. R. Bhattacharya: Conceptualization, formal analysis, supervision, funding acquisition, investigation, writing–original draft, writing–review and editing.

This work was supported in part by Department of Defense grants CA181043 (to R. Bhattacharya) and CA140515 (to L.M. Ellis), the Ruben Distinguished Chair in Gastroenterology Cancer Research (to L.M. Ellis), the NIH/NCI under award number P30CA016672 (to support the Flow Cytometry and Cellular Imaging Core, Functional Proteomics RPPA Core, Cytogenetics and Cell Authentication Core and Research Histology Core Facilities at MD Anderson Cancer Center), the Cancer Prevention & Research Institute of Texas grant numbers RP130397 (to support the Metabolomics Core Facility at MD Anderson Cancer Center) and RP200668 and Multi-Investigator Research Award RP110532 (to support the Center for Translational Cancer Research (RRID: SCR_022214) at Institute of Biosciences and Technology). The authors acknowledge Amy Ninetto from Research Medical Library, MD Anderson Cancer Center for editorial assistance. The authors also would like to acknowledge Ze-Sheng Chen (St John's University, New York, NY) and Susan Bates (Columbia University, NY) for the kind gift of the pair of parental SW620 and SW620/AD300 cell lines.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.