Inhibition of MCL-1 and MEK overcomes MEK inhibitor resistance in triple-negative and inflammatory breast cancers

Abstract

The MAPK pathway can drive resistance in highly aggressive breast cancers. Our previous work showed that MEK inhibitor (MEKi) AZD6244 (selumetinib) prevented lung metastasis in a breast cancer xenograft model. In clinical studies, MEKis as single agents have had only modest activity against solid tumors due to the onset of resistance. Using synthetic-lethality siRNA screening, we identified myeloid cell leukemia-1 (MCL-1) as a potential contributor to AZD6244 resistance. We hypothesized that MCL-1 promotes MEKi resistance in highly aggressive breast cancers and that MCL-1 inhibition overcomes AZD6244 resistance. We established two AZD6244-resistant cell lines: MDA-MB-231-R (triple-negative breast cancer) and SUM149-R (triple-negative inflammatory breast cancer). These resistant cells were characterized with respect to different parameters, and a combination of an MCL-1 inhibitor (MCL-1i) together with a MEKi was evaluated in vitro and in vivo to overcome the acquired resistance. Compared with their respective parental cells, MDA-MB-231-R and SUM149-R cells showed increased proliferation, colony formation, stemness, anchorage-independent growth, and MCL-1 expression levels. MCL-1 knockdown in resistant cells decreased cell proliferation and colony formation, increased apoptosis, and was associated with high expression of the pro-apoptotic proteins PUMA, NOXA, BAK, and BAX. MEKi resistance was overcome when resistant cells were treated with MCL-1i and MEKi combined. In an in vivo mouse model, inhibition of MCL-1 restored sensitivity to AZD6244. Our results suggest that MCL-1 is a driver of MEKi resistance and that combining an MCL-1i with a MEKi warrants further investigation in triple-negative and triple-negative inflammatory breast cancer.

Introduction

Triple-negative breast cancer (TNBC) and inflammatory breast cancer (IBC) are highly aggressive breast cancers (1). TNBC accounts for 15–20% of all breast cancer diagnoses but up to 30% of breast cancer-related deaths (1). IBC accounts for 2–4% of breast cancer diagnoses in the United States but 8%−10% of all breast cancer-related deaths, largely because of its aggressive biology and propensity to metastasize (2). TNBC and triple-negative (TN)-IBC, a subtype of IBC, lack expression of the estrogen, progesterone, and HER2 receptors, which are common targets in breast cancers. A lack of specific, well-validated targets and a high frequency of chemoresistant metastases contributes to the ongoing challenges in TNBC/IBC management.

Contributing to the onset of drug resistance is dysregulation of signaling pathways (3,4). Dysregulation of the MAPK pathway, also known as the ERK pathway, is crucial in metastasis and drug resistance (5). Activation of the MAPK/ERK pathway results in sequential activation of multiple kinases—namely, Ras, Raf, MEK, and ERK. The MAPK/ERK pathway plays an essential role in regulating numerous processes of cancer progression, including cell proliferation, differentiation, apoptosis, stress responses, inflammation, tumorigenesis, and epithelial-to-mesenchymal transition (6). Despite the low number of activating mutations in the MAPK/ERK pathway, this pathway is hyperactive in breast cancer (7). Elevated activation of the MAPK/ERK pathway has been observed in patients with TNBC despite a lack of specific RAS/RAF mutations, which are present in other cancers (8). Increased activation of the MAPK/ERK pathway has been associated with a high recurrence rate in TNBC (6). Targeting MEK is gaining attention as a possible treatment option for tumors driven by aberrations in MAPK/ERK signaling, which has led to the development of numerous MEK inhibitors (MEKis). We found that the MAPK/ERK pathway plays an important role in metastasis in TNBC (9). Further, we showed that patients with high-ERK2-expressing TNBC tumors had a lower survival rate than patients with low-ERK2-expressing tumors (9). Thus, inhibition of ERK activation via MEK inhibition may be a promising therapeutic strategy for TNBC.

Preclinical studies of the MEKis AZD6244 (selumetinib, ARRY-142886), pimasertib (AS703026), cobimetinib, binimetinib (ARRY-438162), trametinib, TAK-733, GDC-0623, and G-573 have shown promising effects in various xenograft models of human cancer, including models of BRAF-mutant melanoma and KRAS-mutant colorectal, non-small cell lung, pancreatic, and breast cancers (10,11). We previously showed that AZD6244 inhibited cell proliferation, epithelial-to-mesenchymal transition, and lung metastasis in a TNBC xenograft model (12). AZD6244 is an active, highly selective, non-ATP-competitive MEKi (developed by Array BioPharma; licensed to AstraZeneca) (12,13). At the time of this writing, AZD6244 was being tested in 15 clinical trials (NCT00463814, NCT01362803, NCT00600496, NCT02407405, NCT03433183, NCT01364051, NCT02151084, NCT02839720, NCT02188264, NCT01089101, NCT01306045, NCT03213691, NCT02546661, NCT01933932, and NCT03162627) for different types of cancers, including a trial for metastatic breast cancer. AZD6244 was approved by the US Food and Drug Administration in April 2020 as a single agent to treat pediatric inoperable plexiform neurofibromas, which exhibit loss of NF-1, a GTPase-activating protein (and was the first drug approved in the US to treat this disease) (14). However, although MEKis have shown promise, their use as single agents has not been successful in cancer treatment due to the gradual development of resistance (15, 23). Combination therapies have been shown to prolong response, delaying acquired resistance (24). We therefore sought to identify mediators of MEKi resistance and develop a therapy combining MEKi with another targeted therapy with a different mechanism of action.

In the present study, to identify mediators contributing to AZD6244 resistance, we conducted a synthetic-lethality whole-genome siRNA-library screen on TNBC cells (MDA-MB-231). MCL-1 (myeloid cell leukemia-1 or induced myeloid leukemia cell differentiation protein) was identified as a top candidate whose silencing sensitized TNBC cells to AZD6244. MCL-1 is a member of the Bcl-2 family of anti-apoptotic proteins and is overexpressed in many human cancers, including TNBC (15). MCL-1 is involved in cell survival, cell immortalization, malignant transformation, chemoresistance (16), and hematopoietic stem cell survival (17). A previous study has shown a reduction of MCL-1 levels after single-agent AZD6244 treatment in siRNA-sensitized EGFR-mutant non–small cell lung cancers (18). MCL-1 is amplified in 55% of TNBCs after preoperative chemotherapy, implying both functional and chemoresistance significance (19). In addition, elevated levels of MCL-1 have been implicated in mediating resistance to lapatinib and trastuzumab in HER2-overexpressing breast cancer cells (20), and increased expression of MCL-1 is associated with poor overall survival in patients with breast cancer (21).

Our current understanding of mediators of resistance to MEKis in TNBC/IBC is limited. In this study, we attempt to understand the mechanisms underlying MEKi resistance in TNBC/IBC, by investigating the biological role of MCL-1 as a mediator of MEKi resistance. We also show how MCL-1 could be utilized as a potential target to overcome MEKi resistance. We demonstrate that targeting MCL-1 sensitizes MEKi-resistant cells to a MEKi.

Materials and Methods

Cell lines and drugs

TNBC cell lines used in the study were MDA-MB-231 (RRID: CVCL_0062) and MDA-MB-468 (RRID: CVCL_0419), purchased from American Type Culture Collection; E0771 (RRID: CVCL_GR23), purchased from CH3 BioSystems; and 4T1.2 (RRID: CVCL_GR32), obtained from Dr. Robin Anderson (Peter MacCallum Cancer Centre, Victoria, Australia). IBC cell lines used were SUM149 (RRID: CVCL_3422) and SUM190 (RRID: CVCL_3423), purchased from Asterand; MDA-IBC3 (RRID: CVCL_HC47), obtained from Dr. Wendy Woodward (MD Anderson Cancer Center); BCX010 (RRID:), obtained from Dr. Funda Meric-Bernstam (MD Anderson); KPL4 (RRID:, obtained from Kawasaki Medical School (Okayama, Japan); and FC-IBC02 (RRID: CVCL_B7S0), obtained from Dr. Fredika Robertson (MD Anderson) and Dr. Massimo Cristofanilli (Fox Chase Cancer Center). MCF10A (RRID: CVCL_0598) normal breast cells were purchased from American Type Culture Collection.

MDA-MB-231, MDA-MB-468, SUM149, SUM190, MDA-IBC3, BCX010, and MCF10A cell lines were validated by DNA fingerprinting at the MD Anderson Cancer Center Cytogenetics and Cell Authentication Core (4/30/2025) using a short tandem repeat method based on a primer extension to detect single-base deviations and confirmed to be free of mycoplasma using the MycoAlert Mycoplasma Detection Kit (Lonza). Additional details are provided in Supplementary Materials and Methods. AZD6244 (MEKi) and AZD5991 (MCL-1 inhibitor [MCL-1i]) were procured from Selleck Chemicals LLC and AstraZeneca, respectively(22).

Establishment of MEKi-resistant cell lines

To develop AZD6244-resistant cell lines, a TNBC cell line, MDA-MB-231, and a TN-IBC cell line, SUM149 were continuously exposed to AZD6244 in increasing concentrations until a final concentration of 15 μM AZD6244 was reached over 6–8 months.

Whole-genome synthetic-lethality siRNA screen in TNBC cells

To identify molecules contributing to resistance to AZD6244 in TNBC cells, we performed a synthetic-lethality whole-genome siRNA-library screen. We selected MDA-MB-231, a TNBC cell line that is relatively resistant to the growth-inhibitory effects of AZD6244 in a two-dimensional culture but sensitive to AZD6244 in a three-dimensional culture model system. Additional details are provided in Supplementary Materials and Methods.

MCL-1 siRNA transfection

MCL-1 siRNA transient transfection was performed in both parental and resistant cell lines using four MCL-1 siRNAs: #1, oligo SASI_Hs01_00162658; #2, oligo SASI_Hs01_00162659; #3, oligo SASI_Hs01_00223435; and #4, oligo SASI_Hs01_00223436 (Sigma-Aldrich). Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) was used for the transfection. Cells (3 × 10⁵ cells/well) were seeded in six-well plates, and transfection was performed for 48 hours according to the manufacturer’s protocol.

Development of shMCL-1 clones

MCL-1-knockdown (shMCL-1) clones were generated in resistant SUM149 cells (SUM149-R) by using two shRNAs (#1, V2LHS_72724; #2, V3LHS_393782) specific to MCL-1 from the GIPZ lentiviral RNA system (Dharmacon/Horizon) per the manufacturer’s protocol. The transfected clones were selected using puromycin (3 μg/mL) as a selection marker and green fluorescent protein (GFP)-based fluorescence-activated cell sorting.

Cell viability and clonogenic assays

The following cell lines were subjected to a Cell Titer-Blue cell viability assay (Promega) to assess cell proliferation and clonogenic assay to assess colony formation: a panel of TNBC/IBC cells (MDA-MB-231, MDA-MB-468, E0771, 4T1.2, FC-IBC02, SUM190, MDA-IBC3, KPL4, BCX010, and SUM149) and immortalized normal breast cells (MCF10A); MDA-MB-231 and SUM149 parental and resistant cells; MCL-1-knockdown MDA-MB-231 and SUM149 parental and resistant cells; and SUM149R-shMCL1 clones. Additional details are provided in Supplementary Materials and Methods.

Comparison of parental and resistant cells

The MDA-MB-231 and SUM149 parental and resistant cells were subjected to the following assays. Migration of the cells was analyzed by transwell migration assay. The stem cell population was assessed using surrogate markers, including CD44^+/high CD24^−/low markers, and mammosphere formation assays. Anchorage-independent growth was analyzed by soft agar assay. Expression of MCL-1 and multi-drug resistance-associated (MDR) gene transcripts were analyzed by quantitative PCR (qPCR) (Table S1). Additional details are provided in Supplementary Materials and Methods.

Apoptosis assay

The Annexin V–FITC/PI kit (Bio-Rad) was used to check the different phases of apoptosis in the MCL-1-knockdown MDA-MB-231 and SUM149 parental and resistant cells and the cells treated with the combination of the MCL-1i AZD5991 and the MEKi AZD6244. Additional details are provided in Supplementary Materials and Methods.

Western blot analysis

Western blot analysis was used to detect the expression of MCL-1 in MDA-MB-231 and SUM149 parental and resistant cells and the expression of binding partners of MCL-1 (BAK, BAX, PUMA, and NOXA) in the AZD6244-treated MCL-1-knockdown MDA-MB-231 and SUM149 parental and resistant cells. Additional details are provided in Supplementary Materials and Methods.

In vivo studies

To examine the effect of MCL-1 inhibition alone on the tumor formation capability of highly aggressive breast cancer, the anti-tumor effect of MCL-1i AZD5991 was evaluated using a SUM149 orthotopic xenograft mouse model.

To compare tumorigenicity between TNBC parental and resistant cell-induced xenograft models, parental and resistant SUM149 cells were transplanted into the mammary fat pads of female NU/J mice, and the breast tumor tissues from the parental and resistant groups were subjected to immunohistochemistry (IHC).

To assess the therapeutic effect of MEKi AZD6244 in mice harboring MCL-1-deficient tumors, SUM149-R-shRNA CTRL, SUM149-R-shMCL-1-clone #1, and SUM149-R-shMCL-1-clone #2 were transplanted into the mammary fat pads of female NU/J mice. Animal care and use followed institutional and National Institutes of Health guidelines, and the experiments were approved by the MD Anderson Institutional Animal Care and Use Committee. The breast tumor tissues from the control and treatment groups were subjected to IHC. Additional details are provided in Supplementary Materials and Methods.

Statistical analysis

All experiments were performed in triplicate. Kaplan-Meier survival analysis and analyses of statistical significance (p < 0.05) were done using GraphPad Prism software, version 9.0 (RRID: SCR_002798). Comparisons between two groups were done using an unpaired t-test, and comparisons between three groups were done using the one-way ANOVA Tukey post hoc test. Data are presented as average ± SEM (n = 3–5).

Data availability

The siRNA lethal screening data generated in this study are available upon request from the corresponding author.

Results

MEKi-resistant cells show increased proliferation, colony formation, and migration with high expression of MDR proteins.

To model MEKi resistance, SUM149 and MDA-MB-231 AZD6244-resistant cell lines (SUM149-R and MDA-MB-231-R, respectively) were generated. The proliferation of parental and resistant cells in the presence and absence of MEKi (AZD6244) was tested over 96 hours. IC₅₀ values were calculated from the percentages of viable cells after exposure to MEKi at various concentrations. We found that AZD6244 significantly reduced the proliferation of the parental cells (MDA-MB-231-P and SUM149-P) (IC₅₀ values, 3.17 μM, and 7.84 μM, respectively). MEKi-resistant MDA-MB-231-R and SUM149-R cells showed significantly (P ≤ 0.0001) higher proliferation rate (IC₅₀ values, 22 and 24 μM, respectively) than their respective parental controls, confirming resistance to AZD6244 (Fig. 1A).

Fig. 1. MEKi-resistant cells show increased proliferation, colony formation, and migration.

(A) CellTiter-Blue proliferation assay results showed that AZD6244 significantly reduced the proliferation of both parental cell lines (MDA-MB-231-P and SUM149-P) in a dose-dependent manner but did not have a significant effect on the proliferation of the resistant cell lines (MDA-MB-231-R and SUM149-R). (B) Clonogenic assay results showed that AZD6244 resulted in a significant dose-dependent decrease in colony formation capability of both parental cell lines, but the resistant cell lines showed higher colony formation capability than the parental cell lines. (C) Transwell migration assay results showed that both resistant cell lines had increased migration compared to their respective parental controls. 231P, MDA-MB-231-P; 231R, MDA-MB-231-R; 149P, SUM149-P; 149R, SUM149-R. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

In a clonogenic assay, AZD6244 significantly decreased the colony formation capability of the MDA-MB-231-P and SUM149-P parental cells in a dose-dependent manner compared with untreated controls. However, when the resistant cells were treated with AZD6244, they retained their colony formation capability even at a high drug concentration (12 μM); MDA-MB-231-R and SUM149-R cells formed colonies at the rates of 92.84% (p = 0.09) and 51.63% (p = 0.42) compared to the respective untreated resistant cells (Fig. 1B).

Next, we evaluated the migration capability of the parental and resistant cells. Both MDA-MB-231-R and SUM149-R cells showed increased migration compared to their respective parental controls. The untreated MDA-MB-231-R cells showed a 2.23-fold ± 0.08 increased relative migration percentage compared to the untreated MDA-MB-231-P cells, and the AZD6244-treated MDA-MB-231-R cells showed a 5.23-fold ± 0.06 increased relative migration percentage compared to the AZD6244-treated MDA-MB-231-P cells. Likewise, the untreated SUM149-R cells showed a 0.410-fold ± 0.1 increased relative migration percentage compared to the untreated SUM149-P cells, and the AZD6244-treated SUM149-R cells showed a 1.13-fold ± 0.1 increased relative migration percentage compared to the AZD6244-treated SUM149-P cells (Fig. 1C). These results showed that resistant cells were less sensitive to the MEKi.

High expression levels of MDR proteins and cancer stem cell markers are among the most common and important factors associated with cancer recurrence. We performed quantitative real-time PCR analysis to compare gene expression levels between the parental and resistant cells. Both MDA-MB-231-R and SUM149-R cells showed significantly higher expression of the MDR protein ABCB1 than their respective parental controls. In addition, MDA-MB-231-R cells showed significantly higher expression of MYC, and SUM149-R cells showed significantly higher expression of OCT4, NANOG, and SOX2 than their respective parental controls (Fig. S1).

MEKi-resistant cells show higher anchorage-independent growth and an increased stemness phenotype.

The soft agar colony formation assay, an indirect measure of tumorigenicity, showed that the resistant cells had a greater capacity than the parental cells to form colonies in soft agar. The average number of colonies was 894 ± 83 for the MDA-MB-231-R cells and 171 ± 30 for the MDA-MB-231-P cells. Similarly, the average number of colonies was 902 ± 152 for the SUM149-R cells and 242 ± 4 for the SUM149-P cells (Fig. 2A).

Fig. 2. MEKi-resistant cells show an increased stemness phenotype, high anchorage-independent growth, and high expression of MCL-1 compared with parental cells.

(A) Both resistant cell lines showed more colonies than the respective parental control cell lines in an anchorage-independent soft agar assay. (B) Both resistant cell lines (MDA-MB-231-R [231R] and SUM149-R [149R]) formed more mammospheres than the respective parental control cell lines (MDA-MB-231-P [231P] and SUM149-P [149P]) with or without AZD6244 treatment. (C) Both resistant cell lines showed increased stemness properties, via an increased CD44⁺/CD24^low subpopulation, compared to their respective controls, with or without AZD6244 treatment. (D, E) Both resistant cell lines showed higher expression of MCL-1 than the respective controls in western blot (D) and qPCR (E) analysis. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

To compare the stemness phenotype of the resistant cells and their parental controls, we first performed a secondary mammosphere assay to evaluate the stem cell-like capacity of these cells. The average mammosphere counts were higher in the untreated and treated MDA-MB-231-R cells (168 ± 1 and 145 ± 1, respectively) than in the untreated and treated MDA-MB-231-P cells (126 ± 1 and 29 ± 2, respectively). Similarly, the average mammosphere counts were higher in the untreated and treated SUM149-R cells (275 ± 24 and 297 ± 31, respectively) than in the untreated and treated SUM149-P cells (193 ± 18 and 91 ± 6, respectively) (Fig. 2B).

Next, we assessed stem cell markers CD44⁺ CD24^−/low in the resistant and parental cells using flow cytometry analysis. The results of this assay confirmed the increased stemness of the resistant cells compared to their parental controls. In the absence of AZD6244, there was no significant difference in the average percentage of the CD44⁺ CD24^−/low subpopulation between MDA-MB-231-R cells (99.19% ± 0.1) and MDA-MB-231-P cells (98.24% ± 0.03). After AZD6244 treatment, the average percentage of the CD44⁺ CD24^−/low subpopulation was higher in MDA-MB-231-R cells (95.27% ± 0.1) than in MDA-MB-231-P cells (73.47% ± 1). In the absence of AZD6244, the average percentage of the CD44⁺ CD24^−/low subpopulation was 59.13% ± 1 in SUM149-R cells and 42.86% ± 1 in SUM149-P cells. After AZD6244 treatment, SUM149-R cells still showed a higher average percentage of the CD44⁺ CD24^−/low subpopulation: 37.45% ± 5 compared to 14.98% ± 7 (Fig. 2C).

siRNA screening uncovers targets whose knockdown further sensitizes TNBC cells to AZD6244.

To explore the potential for combination therapy with a MEKi, we performed a synthetic-lethality screen using a whole-genome siRNA library to identify genes whose silencing sensitizes MDA-MB-231 TNBC cells to AZD6244. The analysis showed 26 top targets that were consistently identified as differing between AZD6244-treated and untreated cells using three different statistical methods (Fig. S2; Table S2).

One of these top targets was MCL-1, which is an anti-apoptotic protein and a member of the BCL-2 family of proteins. We thus evaluated the expression of MCL-1 in our parental and resistant cells through western blot and qPCR analysis. Both western blot analysis and qPCR analysis confirmed increased expression of the MCL-1 protein and MCL1 gene in the MDA-MB-231-R and SUM149-R cells compared to their respective parental controls (Fig. 2D, E).

MCL-1 inhibition affects the proliferation and colony formation of TNBC and IBC cells and reduces tumor volume in an orthotopic xenograft model.

To evaluate the effect of MCL-1 inhibition in highly aggressive breast cancers, we examined the effect of the MCL-1i AZD5991 on cell proliferation and tumorigenicity of TNBC and IBC (TN and HER2+) cells. The cell proliferation assay showed IC₅₀ values ranging from 0.23 to 36.61 μM (Fig. 3A, table). MDA-MB-231, MDA-MB-468, E0771, FC-IBC02, MDA-IBC3, SUM149, SUM190, and BCX010, cells showed high sensitivity and a significant dose-dependent decrease in the proliferation rate, while 4T1.2 and KPL4, cells,, together with MCF10A normal breast cells, showed resistance with no significant decrease in the proliferation rate (Fig. 3A). The clonogenic assay confirmed these findings except for MDA-MB-231 cells and KPL4 cells (Fig. 3B). In the colony formation assay, AZD5991 decreased colony formation on soft agar in a dose-dependent manner (in MDA-MB-468 and SUM149 cells decreased colony formation but not in a dose-dependent manner in KPL4and had no significant effect on colony formation in MDA-MB-231 cells (Fig. 3C). In a SUM149 orthotopic xenograft mouse model, AZD5991 as a single agent at a dose of 30 mg/kg resulted in significant tumor regression compared to what was observed in the vehicle-treated mice (P = 0.0006), but AZD5991 at a dose of 15 mg/kg did not result in significant regression of tumor (Fig. 3D). The average tumor volume was 971.99 ± 127 mm³ in the control group, 900.73 ± 85 after treatment with AZD5991 at the 15-mg/kg dose, and 686.73 ± 48 after treatment with AZD5991 at the 30-mg/kg dose, with no significant change in the average weight of the mice.

Fig. 3. MCL-1 inhibition affects the proliferation and colony formation of TNBC/IBC cells and reduces tumor volume in an orthotopic xenograft model.

(A) A Cell Titer-Blue (CTB) proliferation assay showed that MCL-1i AZD5991 significantly reduced the proliferation of most of the TNBC/IBC cell lines studied. The table shows the IC₅₀ values from the CTB proliferation assay. (B, C) Clonogenic (B) and soft agar (C) assays showed that AZD5991 significantly decreased the colony formation capability of most of the TNBC/IBC cell lines. (D) The IBC cell line SUM149 was injected into the mammary fat pads of female SCID mice. When the tumor size reached 75–150 mm³, mice were treated with AZD5991 (15 or 30 mg/kg) or vehicle once weekly via tail-vein injection for 4 weeks. A dose of 30 mg/kg resulted in significantly reduced tumor volume compared to vehicle treatment with no significant change in the average weight of mice. Table shows the average tumor volume of the control and treatment groups. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, nonsignificant.

Knockdown of MCL-1 sensitizes MEKi-resistant cells and decreases proliferation and colony formation.

The biological effect of MCL-1 in MEKi resistance was evaluated using MCL-1 siRNA transient transfection in MDA-MB-231-R and SUM149-R cells and shRNA-based stable MCL-1-knockdown in SUM-149-R cells. Western blot analysis showed a significant reduction in expression of the MCL-1 gene in transient transfection in both MDA-MB-231-R and SUM-149-R cells (Fig. 4A(i)) and in stable clones in SUM-149-R cells (Fig. 4B(i)).

Fig. 4. MCL-1 knockdown restores MEKi sensitivity in MEKi-resistant TNBC cells.

(A) Transient knockdown of the MCL-1 gene using siRNA transfection in MDA-MB-231 and SUM149 cells. (i) Western blot analysis confirmed the siRNA-mediated silencing of the MCL-1 gene in both parental and resistant cells. (ii) A Cell Titer-Blue (CTB) proliferation assay with the parental and resistant cells with transient MCL-1 knockdown showed a decrease in the proliferation of these cells in the presence of AZD6244. (iii) A colony formation assay after knockdown of MCL-1 and treatment with different doses of AZD6244 showed a significant decrease in the colony formation capability of both parental and resistant cells. (B) Stable knockdown of the MCL-1 gene via a shRNA system in SUM149-R cells. (i) Western blot analysis confirmed knockdown of the MCL-1 gene. (ii) A CTB proliferation assay with stable MCL-1-knockdown clones showed a significant dose-dependent decrease in proliferation in the presence of AZD6244. (iii) A colony formation assay showed a significant dose-dependent decrease in colony formation capability in the presence of AZD6244.

Next, we assessed cell proliferation in both parental and resistant cells transfected with MCL-1 siRNA and treated with AZD6244. We observed inhibition of cell proliferation in both parental and resistant cells after AZD6244 treatment (Fig. 4A(ii)). In MDA-MB-231-P cells, the IC₅₀ values were 1.39 μM and 1.92 μM, respectively, for siRNAs #1 and #2, compared to 6.35 μM and 6.18 μM, respectively, for the scrambled and untransfected controls. The MDA-MB-231-R cells showed some sensitization towards AZD6244. In SUM149-P cells, the IC₅₀ values were 1.97 μM and 0.26 μM, respectively, for siRNAs #1 and #2, compared to 2.23 μM and 3.88 μM, respectively, for the scrambled and untransfected controls. The SUM149-R cells also showed sensitization towards AZD6244; IC₅₀ values were 2.87 μM and 0.47 μM, respectively, for siRNAs #1 and #2, compared to 12.93 μM and 23.54 μM, respectively, for the scrambled and untransfected controls.

Similarly, we assessed the proliferation of SUM149-R shMCL-1 clones in the presence of AZD6244. The stable knockdown clones (SUM149-R shMCL-1 #1 and #2) showed significant dose-dependent inhibition of proliferation in the presence of AZD6244, with IC₅₀ values of 3.13 μM and 7.46 μM, respectively, compared to 51.73 μM and 46.20 μM for the shRNA and untransfected controls, respectively (Fig. 4A(ii) and 4B(ii)).

These findings were further validated in the clonogenic assay; AZD6244 significantly decreased the colony formation capability in the transiently MCL-1-knockdown parental (MDA-MB-231-P and SUM149-P) and resistant (MDA-MB-231-R and SUM149-R) cells in a dose-dependent manner (Fig. 4A(iii)). Likewise, SUM149-R-shMCL-1 clones showed a significant dose-dependent decrease in colony formation capability in the presence of AZD6244 (Fig. 4B(iii)). These results showed that the knockdown of MCL-1 sensitized the parental and resistant cells to the MEKi AZD6244.

Knockdown of MCL-1-induces apoptosis in MEKi-resistant cells.

To assess the mechanism underlying the sensitization of MEKi-resistant cells by MCL-1 knockdown, we performed an Annexin V-FITC/PI apoptosis assay. We found that knockdown of MCL-1 in MDA-MB-231-R and SUM149-R cells increased the proportions of cells in the early and late apoptotic phases (Annexin V-FITC⁺/PI⁻ and Annexin V-FITC⁺/PI⁺, respectively) compared to the scrambled vector control in the presence of AZD6244 (Fig. 5A). In MDA-MB-231-R cells with MCL-1 knockdown using siRNAs #1, #2, #3and #4, showed average percentages of cells in the early apoptotic phase were 8.3% ± 0.3, 8.8% ± 0.3, 7.1 ± 0.1 and 6.4 ± 0.7, respectively, compared with 0.56 % ± 0.03 for the scrambled vector control, and average percentages of cells in the late apoptotic phase were 11.36% ± 0.6, 26.26% ± 1, 10.2 ± 0.4, 10.16 ± 0.7, respectively, compared with 6.53 % ± 0.3 for the scrambled vector control. Similarly, in SUM149-R cells, with MCL-1 knockdown using siRNAs #1, #2, #3, and #4, showed average percentages of cells in the early apoptotic phase were 35.66% ± 1, 16.20% ± 1, 27.50% ± 1, and 3.80% ± 0.1, respectively, compared with 1.23% ± 0.08 for the scrambled vector control, and average percentages of cells in the late apoptotic phase were 26.83% ± 1, 16.86% ± 1, 24.43% ± 1, and 15.26% ± 0.8, respectively, compared with 2.6% ± 0.2 for the scrambled vector control.

Fig. 5. Mechanism of MCL-1-knockdown-induced apoptosis in MEKi-resistant cells and combination therapy for resistance.

(A) An apoptosis assay showed high percentages of cells in the early and late apoptotic phases in MCL-1-knockdown MDA-MB-231-R and SUM149-R cells after treatment with AZD6244. (B) Western blot analysis showed the expression of the binding partners of MCL-1 in the MCL-1-knockdown MDA-MB-231 and SUM149 parental and resistant cells. (C) A Cell Titer-Blue proliferation assay showed that a combination of MEKi AZD6244 and MCL-1i AZD5991 sensitized both resistant cell lines and showed a significant dose-dependent decrease in the proliferation of these cells. The combination index (CI) analysis through Calcusyn software (table) showed the synergism of the combination. (D) A clonogenic assay showed a significant dose-dependent decrease in the colony formation capability of SUM149-R cells after treatment with the combination of AZD6244 and AZD5991. (E) An apoptosis assay showed the percentages of SUM149-R cells in the early and late apoptotic phases after treatment with the combination of AZD6244 (1μM) and AZD5991 (1μM). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, not significant.

MEKi-resistant cells with MCL-1 knockdown show high expression of MCL-1 binding partners.

To further understand the mechanism underlying apoptosis in the MEKi-resistant cells, the expression of binding partners of MCL-1 was evaluated in MEKi-resistant cells with MCL-1 knockdown. Western blot analysis showed that MCL-1 knockdown in MDA-MB-231-R and SUM149-R cells in the presence of AZD6244 resulted in high expression of the BH3-activating protein PUMA and the BH3-sensitizing protein NOXA (in both MDA-MB-231-R and SUM149-R cells) as well as the proapoptotic molecules BAK (in MDA-MB-231-R cells) and BAX (in SUM149-R cells) (Fig. 5B). These results suggested that AZD6244-induced apoptosis in the MCL-1-knockdown MDA-MB-231-R and SUM149-R cells may be mediated through both BH3 sensitizers and direct apoptosis activators.

MEKi-resistant cells show decreased proliferation and colony formation and increased apoptosis after treatment with MCL-1i AZD5991 plus MEKi AZD6244.

To evaluate the therapeutic efficacy of the combination of an MCL-1i and a MEKi, we performed a proliferation assay with the MDA-MB-231 and SUM149 parental and resistant cell lines. All four cell lines showed sensitization and decreased proliferation in a dose-dependent manner after treatment with equal concentrations of AZD5991 for 24 hours followed by AZD6244 for 72 hours for a total 96-hour incubation period (Fig. 5C; Fig. S3). In the MDA-MB-231-P and MDA-MB-231-R cells, the IC₅₀ values for AZD6244 were 10.94 μM and 40.79 μM, respectively, and the IC₅₀ values for AZD5991 and AZD6244 in combination were significantly lower: 3.22 μM and 7.22 μM, respectively. Similarly, in the SUM149-P and SUM149-R cells, the IC₅₀ values for AZD6244 were 3.80 μM and 22.76 μM, respectively, and the IC₅₀ values for AZD5991 and AZD6244 in combination were significantly lower: 0.20 μM and 1.68 μM, respectively. These findings were validated by the clonogenic assay, which showed that the combination of AZD5991 and AZD6244 significantly reduced the colony formation capability of the SUM149-R cells in a dose-dependent manner (Fig. 5D). The combination indexes for both parental and resistant cells were less than 1, which showed synergism of AZD5991 and AZD6244 (Fig. 5E). Similar results were obtained with the MCL-1i S63845 (Fig. S4). These data indicated that a combination of MCL-1i and MEKi has potential as a therapeutic approach to overcome MEKi resistance in TNBC.

Next, to determine whether AZD5991 in combination with AZD6244 induces apoptosis, an Annexin-V-FITC/PI apoptosis assay was performed. The treated SUM149-R cells showed average percentages of 3.43% ± 0.2 of cells in the early apoptotic phase (Annexin V-FITC⁺/PI⁻) and 11.54% ± 0.8 of cells in the late apoptotic phase (Annexin V-FITC⁺/PI⁺), compared to the untreated control’s average percentages of 0.50% ± 0.03 and 1.63% ± 0.09 in the early and late apoptotic phases, respectively (Fig. 5F). This result showed that a combination of AZD5991 and AZD6244 can induce apoptosis in SUM149-R cells.

MEKi-resistant cells induce high expression of MCL-1, with high proliferation and stemness, in an orthotopic xenograft mouse model.

We next compared tumor formation capability between SUM149 parental and resistant cells in a xenograft mouse model. Mice injected with SUM149-R cells exhibited a significant time-dependent increase in tumor volume as compared to mice injected with SUM149-P cells (P < 0.0001) (Fig. 6A(i)), with no significant change in the average weight of the mice (Fig. 6A(ii)). Moreover, IHC analysis showed higher expression of MCL-1, proliferative marker Ki67, and stemness markers CD44 and ALDHA1 in breast tumor tissues of the SUM149-R-injected mice than in the tissues of SUM149-P-injected mice (Fig. 6A(iii)). These results showed that AZD6244-resistant cells induced high proliferation and stemness, which resulted in higher tumor growth than was observed with the parental cells in the in vivo model.

Fig. 6. MEKi AZD6244 repressed tumor growth and reduced stemness in an MCL-1-knockdown MEKi-resistant orthotopic xenograft mouse model.

(A) MEKi-resistant cells showed high expression of MCL-1 with high proliferation and stemness in an orthotopic xenograft mouse model. (i) Comparison of the tumor formation capability of SUM149-P and SUM149-R cells injected into NU/J mice. (ii) Average weights of the mice of different groups. (iii) IHC analysis showed higher expression of ALDHA1, CD44, Ki67, and MCL-1 in breast tumor tissues from the SUM149-R-injected mice than in tissues from the SUM149-P-injected mice. (B) MEKi reduced tumor growth and increased survival in the MCL-1-knockdown orthotopic xenograft mouse model. (i) Schematic representation of the animal study. (ii) Comparison of the tumor growth of SUM149-R shRNA CTRL, SUM149-R shMCL-1 #1 and SUM149-R shMCL-1 #2 cells injected into NU/J mice. (iii) Kaplan-Meier overall survival curve of the mice of different treatment groups for 82 days. The mice injected with shMCL-1 clone #1 and clone #2 showed significantly higher survival rates than the mice injected with the shRNA CTRL cells after treatment with AZD6244 (p = 4.15e-05 and p = 2.37e-05) and vehicle (p = 1.20e-06 and p = 7.20e-06) (iv) Average weight of the mice of different groups. (v) IHC analysis showed AZD6244 reduced the expression of ALDHA1, CD44, Ki67, and MCL-1 in breast tumor tissues compared to the vehicle-treated breast tumor tissues from the mice injected with SUM149-R shMCL-1 #1 and SUM149-R shMCL-1 #2 cells.

MEKi suppressed tumor growth and reduced stemness in an MCL-1-knockdown orthotopic xenograft mouse model.

To translate our in vitro findings, we evaluated the effect of MEKi in an orthotopic xenograft model involving two different MCL-1-knockdown clones (shMCL-1 clone #1 and clone #2) of SUM149-R cells (Fig. 6B). Treatment with AZD6244 significantly reduced tumor growth in mice injected with SUM149-R-shMCL-1 clone #1 and #2 compared to mice injected with a SUM149-R-shRNA control (P < 0.0001) (Fig. 6B(i)).

Moreover, the mice injected with SUM149-R shMCL-1 clones #1 and #2 showed a higher survival rate compared to the mice injected with SUM149-R shRNA control after treatment with AZD6244 (p = 4.15e-05 and p = 2.37e-05) (Fig. 6B(ii)) with no significant change in the average weight of the mice (Fig. 6B(iii)). In addition, IHC analysis showed lower expression of MCL-1, proliferative marker Ki67, and stemness markers CD44 and ALDHA1 in breast tumor tissues of the mice in the SUM149-R shMCL-1 clone #1 and #2 groups than in the tissues of the mice in the SUM149-R shRNA control group (Fig. 6B(iv). Together, these results showed that the knockdown of MCL-1 in the presence of MEKi AZD6244 resulted in reduced tumor growth, reduced proliferation, reduced stemness, and a high overall survival rate in the in vivo model.

Discussion

Our present study shows the biological role of MCL-1 in TNBC and explores MCL-1 as a potential target in the presence of MEKi resistance in TNBC. We found that a combination of an MCL-1i together with a MEKi overcame this developed resistance in TNBC cells and an orthotopic xenograft model. This strategy thus has the potential to improve the therapeutic efficacy of MEKis in TNBC and warrants further investigation.

MCL-1 emerged as a potential target in our siRNA synthetic-lethality whole-genome screen of MDA-MB-231 TNBC cells after treatment with AZD6244. High expression of MCL-1 protein supports cancer cells in evading apoptosis and inhibits sensitivity to main-line chemotherapeutic agents, which ultimately leads to the development of resistance against commonly used therapies like radiotherapy and chemotherapy. Therefore, targeting MCL-1 has the potential to overcome the drug-resistance property of cancer cells and to efficiently enhance chemosensitivity (23,24). Campbell et al also showed the dependency of TNBC on MCL-1 by demonstrating the relationship between high MCL-1 and poor prognosis in TNBC (15). Many cancers with dysregulation of MCL-1 show dependency on MCL-1 for survival and development of resistance against chemotherapy (25). MCL-1 overexpression has also been reported in ABT737 (antagonist of BCL-2) resistance in lymphoma cells (26). In addition, patients with lymphoid malignancies whose disease was resistant to rituximab (an anti-CD20 antibody used to treat B-cell malignancies) also showed an increased level of MCL-1, and its knockdown sensitized cells to rituximab treatment (27). MCL-1 knockdown also sensitized pancreatic carcinoma cells to gemcitabine (28).

Our findings are in line with previous reports that MCL-1 is associated with increased tumorigenicity and stemness. MCL-1 overexpression was previously reported to be associated with high-grade morphology and high proliferation in mantle cell lymphoma (29), and its interaction with the voltage-dependent anion channel at the mitochondrial outer membrane has been shown to promote cell migration in non-small cell lung carcinoma cells (30). MCL-1 has also been reported to have a critical role in the stemness of breast cancer cells, and its high expression correlated with stem cell markers in tumors (31). In our study reported here, cells resistant to a MEKi displayed higher stemness —i.e., a higher population of CD44^+/high CD24^−/low cells and a higher mammosphere count—than their parental cells. CD44^+/high CD24^−/low ratio and mammosphere count are extensively used as indicators of cancer stem cell population in breast cancer research (32,33). These resistant cells showed higher anchorage-independent growth, an indirect measurement of tumorigenicity, than their parental cells (34). In our study, we found that AZD6244-resistant MDA-MB-231-R and SUM149-R cells had greater proliferation, colony formation, and migration capability than the respective parental controls and had higher expression of MCL-1 and MDR proteins. All these findings were corroborated by our in vivo study, in which mice injected with resistant cells showed a high level of MCL-1 together with high tumor volume, proliferation, and, most interestingly, a high degree of stemness. Wei et al demonstrated that knockdown of MCL-1 resulted in reduced tumorigenicity in a mouse xenograft model (28).

Our findings shed light on the biological role of MCL-1 in MEKi resistance. We found that parental and resistant MDA-MB-231 and SUM149 cells with MCL1 knockdown via transient transfection and SUM149-R cells with MCL1 knockdown via stable transfection showed decreased proliferation and colony formation capabilities compared to the respective controls in the presence of AZD6244. A previous study also showed a reduction in cell viability and colony formation capability after the knockdown of MCL-1 in pancreatic cell lines (28). Interestingly, the MCL-1-knockdown MDA-MB-231-R and SUM149-R cells showed increased apoptosis in the presence of AZD6244. Many previous studies also have shed light on the essentiality of MCL-1 for cancer cell survival and have shown that knockdown of MCL-1 via different approaches resulted in the induction of apoptosis in various cancer cells (35–38).

There are two main apoptosis pathways: the extrinsic pathway, which is triggered by different death signals, and the intrinsic pathway, which is regulated by the BH3 (Bcl-2 homology 3) family of proteins. These BH3 proteins are of two types: direct activators and sensitizers/de-repressors. The direct activators, including Bim and Bid, bind and trigger effectors such as BAK and BAX directly (39–41). The sensitizers/de-repressors (40,42–44), including Bad, Bmf, Hrk (Harakiri), NOXA, and PUMA, can inhibit the anti-apoptotic Bcl-2 network by means other than direct activation of BAK or BAX. To reveal how the combination of MCL-1i AZD5991 and MEKi AZD6244 sensitizes and induces apoptosis in MEKi-resistant cells, we assessed the expression pattern of these known activators and sensitizers. Our results showed that inhibition of MCL-1 in the presence of AZD6244 resulted in increased expression of the BH3 sensitizers PUMA and NOXA in both MDA-MB-231-R and SUM149-R cell lines, together with high expression of the direct activators BAK and BAX. Our results highlight a probable mechanism behind the induction of apoptosis via targeting MCL-1 in the presence of AZD6244 in MEKi-resistant cells. However, a slight deviation in the mechanism was observed between the two resistant cell lines (Fig. 5B). MDA-MB-231-R cells showed increased expression of BAK, while SUM149-R cells showed increased expression of BAX. Both BAK and BAX form pores in the mitochondrial outer membrane, which leads to the intrusion of cytochrome c with other apoptotic proteins in the cytoplasm and promotes apoptotic microsome development, caspase activation, cell lysis, and cell death (45). The differences between MDA-MB-231-R and SUM149-R cells in the expression of direct activators may be due to differences in the nature, origin, and behavior of the two cell lines.

Having verified the role of MCL-1 in MEKi resistance in TNBC, we sought to further validate and translate these findings by testing a combinatorial approach that might be applied to overcome MEKi resistance in patients. Many MCL-1i are currently being used, including S63845, AMG 176, AMG 397, and AZD5991 (22,46–48). The first MCL-1i published was S63845, but AZD5991 is the first MCL-1i to enter the clinical trial phase (NCT03218683). AZD5991, a rationally developed macrocyclic molecule, has shown high selectivity and affinity for MCL-1. In myeloma and acute myeloid leukemia cells, AZD5991 induced apoptosis and showed anti-tumor effects in an in vivo model (22). On the basis of these clinical data, we explored the therapeutic efficacy of combining the MCL-1i AZD5991 with the MEKi AZD6244 as a combination therapy in our model. The combination of these two drugs showed a significant decrease in proliferation, colony formation, and induced apoptosis in the MEKi-resistant cells and validated the findings of the MCL-1-knockdown studies. The combination of a MEKi with an MCL-1i has also shown effectiveness in melanoma and rhabdomyosarcoma (49,50). In melanoma cells, the combination of the MEKi trametinib with the MCL-1i AZD5991 re-sensitized the cells, causing a reduction in in vivo tumor growth (49). Similarly, the MEKi trametinib together with the MCL-1i S63845 decreased tumorigenicity in rhabdomyosarcoma cells as well as in a patient-derived xenograft mouse model (50).

By elucidating the biological role of MCL-1 in the setting of MEKi resistance in TNBC, our study provides a foundation for future research. Despite encouraging results with the MCL-1is AZD5991 and S63845 in combination with AZD6244, emerging cardiotoxicity halted the clinical trial of AZD5991 (NCT03218683) (51), which limits its use as an effective combination partner to target MCL-1. However, there are other approaches to decrease MCL-1 levels, such as translation inhibitors and transcription inhibitors, including CDK7/9 inhibitor approaches. The development of a highly potent MCL-1i with minimal or no side effects remains a challenge. Our study generates hope for future research that by developing a potent, selective, and safer MCL-1i in combination with a MEKi, we can combat highly aggressive breast cancers.