Identification of Lethal Inhibitors and Inhibitor Combinations for Mono-Driver versus Multi-Driver Triple-Negative Breast Cancer Cells

Simple Summary Triple-negative breast cancers lack estrogen and progesterone receptors and HER2. They lack effective targeted therapies and tend to be more aggressive, with the worst five-year overall survival and a higher rate of recurrence. This study investigates the oncogenic signaling mechanisms of two model cell lines, DU-4475 and MDA-MB-231. The former is a mono-driver cancer cell line relying on a BRAF V600E mutation and the latter is a multi-driver cancer dependent on a KRAS-mutation-activated MAP kinase pathway and Src kinase. The results of this study reveal that while the mono-driver cancer cells can be effectively killed by one drug blocking their predominant driver, multi-driver cancer cells can only be killed by drug combinations blocking all drivers. The drug combination in MDA-MB-231 achieves strong synergy and potent synthetic lethality. This study suggests pharmacological synthetic lethality as a foundation for combination targeted therapy for multi-driver cancers. Abstract There are no signaling-based targeted therapies for triple-negative breast cancer. The development of targeted cancer therapy relies on identifying oncogenic signaling drivers, understanding their contributions to oncogenesis and developing inhibitors to block such drivers. In this study, we determine that DU-4475 is a mono-driver cancer cell line relying on BRAF and the mitogen-activated protein kinase pathway for viability and proliferation. It is fully and lethally inhibited by BRAF or Mek inhibitors at low nM concentrations, but it is resistant to inhibitors targeting other signaling pathways. The inhibitory lethality caused by blocking Mek or BRAF is through apoptosis. In contrast, MDA-MB-231 is a multi-driver triple-negative breast cancer cell line dependent on both Src and the KRAS-activated mitogen-activated kinase pathway for proliferation and viability. Blocking each pathway alone only partially inhibits cell proliferation without killing them, but the combination of dasatinib, an Src inhibitor, and trametinib, a Mek inhibitor, achieves synthetic lethality. The combination is highly potent, with an IC50 of 8.2 nM each, and strikingly synergistic, with a combination index of less than 0.003 for 70% inhibition. The synthetic lethality of the drug combination is achieved by apoptosis. These results reveal a crucial difference between mono-driver and multi-driver cancer cells and suggest that pharmacological synthetic lethality may provide a basis for effectively inhibiting multi-driver cancers.


Introduction
Breast cancer is the most common cancer in women and metastatic breast cancer is the second-leading cause of cancer-related deaths in American women [1,2]. Triple-negative breast cancer (TNBC) lacks an estrogen receptor (ER), progesterone receptor (PR), and HER2 receptor. It accounts for~15% of all breast cancers [3][4][5] and tends to be more aggressive [6][7][8], with higher mean tumor size, tumor grade, the worst five-year overall Dye MA or MTT dye assay (Thermo Fisher Scientific) per the manufacturer's instructions. DU-4475 cells were cultured in suspension and plated at 10,000 cells per well in a 100 µL medium containing the indicated drug and 1% DMSO. The cells were treated with indicated drugs for 48 h at 37 • C with 5% CO 2-before the cell viability was determined using the Biolog Redox Dye MA by calculating the differences in absorbance at 590 and 750 nm. MDA-MB-231 cells are adherent cells and were seeded at 12,000 cells per well in 120 µL and allowed to attach overnight. A culture medium containing the drug and 5% DMSO (30 µL) was added to each well and the cells were cultured for 48 h before the cell viability was determined by the MTT assay. The formazon product was determined by absorbance at 490 and 750 nm using a Biotek Microplate Reader. In the MTT assay, the A 490 -A 750 values were taken as indicators of cell viability. All cell growth and drug inhibition experiments were performed at least twice in triplicate.

Curve Fitting by the Hill Equation and the Biphasic Equation
Dose-response data were fitted to the Hill equation and the biphasic equation in Microsoft Excel using the Solver add-in program. The Hill equation used was I = I max × D n /(IC 50 * n + D n ), where I max is the maximal inhibition by a drug, IC 50 * is the IC 50 for inhibiting the portion of cell viability that is sensitive to the drug, and is the Hill coefficient or slope. The biphasic equation used was I = F 1 where the inhibition of cell viability (I) as a function of variable drug concentration is determined by three constants: F 1 , K d1 , and K d2 . The F 2 was not an independent variable but instead calculated as 100%-F 1 . In Hill equation fitting, the root mean square error (RMSE) was minimized using the I max , IC 50 * and n as variables. In biphasic curve fitting, the RMSE was minimized using the F 1 , K d1 , and K d2 as variables.

Drug Synergy Analysis and Combination Index Calculation
Drug synergy was evaluated by the combination index (CI) and the dose reduction index (DRI), as described previously [53,54]. Cell viability was determined after incubation in the presence of each drug alone or both drugs (1:1 ratio) at 16 concentrations ranging from 0.6 nM to 20 µM. The CI was calculated according to Chou [55], using the Chou and Talalay Method [55] with the following equation: CI = IC x-AB /IC x-A + IC x-AB /IC x-B , where IC x-A , IC x-B , and IC x-AB are the concentrations of drug A, drug B, and drug AB combination, causing X% inhibition of cell viability, respectively. DRI was calculated as 1/CI.

Time Course Experiments
To determine the time-dependent effects of drug treatments on cell viability of DU-4475 and MDA-MB-231, cells were treated with drugs at indicated concentrations as described in Section 2.2 and the cell viability was determined 1 h, 24 h, 48 h, and 72 h after the treatments were initiated.

Cell Treatments and Western Blots
To determine the effects of a protein kinase inhibitor or a combination of inhibitors on the signaling proteins, cells were seeded at 70% confluency and treated with drugs at indicated concentrations for 1 h under normal cell culture conditions. After the treatment, the cells were placed on ice, washed 1× with chilled PBS, and lysed in RIPA buffer containing a protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA) and protein phosphatase inhibitors (PhosSTOP, Sigma-Aldrich) for 30 min at 4 • C. Lysates were cleared at 21,000× g for 10 min at 4 • C. Protein concentrations of the supernatants were then quantified using a BioRad Bradford Protein Assay. Then, 12-40 µg of protein was separated using a gradient (4-20%) SDS-PAGE gel and transferred onto a nitrocellulose membrane. The membranes were then blocked in 5% non-fat milk in TBST (Tris-buffered Saline with Tween 20) for 1 h before probing with the indicated primary antibody overnight. Membranes were imaged using LI-COR Odyssey CxL Imager and analyzed using Image Studio software (LI-COR Biosciences, Lincoln, NE, USA). All antibodies were purchased from Cell Signaling Tech-Cancers 2022, 14, 4027 4 of 20 nology (Danvers, MA, USA). The density of each protein band in the Western blots was measured using ImageJ 1.53K software [56].

DU-4475 Apoptosis and Necrosis Assay
DU-4475 cell apoptosis and necrosis in response to trametinib treatment are monitored using the RealTime-Glo Anexin V Apoptosis and Necrosis assay (Promega, Madison, WI, USA) [57]. DU-4475 cells were seeded at 10,000 cells per well in 100 µL medium with or without trametinib in 96-well white plates with clear bottoms. Apoptosis and necrosis assay reagents (100 µL) were added and the cells were incubated at 37 • C and 5% CO 2 . Each well contained 1% DMSO, 20 nM trametinib (treatment) or no drug (control). The plate was shaken for 30 s at 500-700 rpm before the luminescence and fluorescence (485 nm Ex /528 nm Em ) in each well were recorded every 6 h, using a Biotek Synergy HTX multimode reader (Agilent Technologies, Santa Clara, CA, USA). The assay was performed in quadruplicates and the average ∆Lum (luminescence of treatment-control) and ∆Fluo (fluorescence of treatment-control) are reported.

DU-4475 Cell Line Is Exceptionally Sensitive to BRAF and Mek Kinase Inhibitors
We previously reported that TNBC cell lines, MDA-MB-231 and MDA-MB-468, are each dependent on two signaling pathways and can be effectively inhibited by inhibitor combinations blocking both pathways [53]. To evaluate if most or all TNBC cell lines are multi-driver cancer cells that cannot be blocked by any individual signaling kinase inhibitors, we examined the drug response data of all TNBC cell lines in the Genomics of Drug Sensitivity in Cancer database [58]. Table 1 lists 17 TNBC cell lines and the IC 50 for each cell line's most potent protein kinase inhibitors for targeted therapy. Most of the TNBC cell lines do not respond potently to any protein kinase inhibitors, displaying high nM to low µM IC 50 s. The only exception is DU-4475, which is potently inhibited by the BRAF inhibitor dabrafenib (IC 50 = 6.3 nM) and the Mek inhibitor trametinib (IC 50 = 0.5 nM). These data suggest that most TNBC cell lines do not depend on any single kinase for viability, while DU-4475 is dependent on BRAF and Mek for viability. This suggestion is consistent with DU-4475 being a mono-driver cancer cell line and the others reported are multi-driver cancer cell lines. In this study, we aimed to determine whether DU-4475 was truly a mono-driver cancer cell line. DU4475 was derived from a cutaneous metastatic nodule from a patient with advanced TNBC [59,60]. Even though DU-4475 has been widely used as a model cell line to study the oncogenesis of TNBC, the oncogenic mechanism of this cell line is not established. According to the COSMIC database [61], DU-4475 contains two notable mutations: a truncation mutation at E1577 in the APC gene and a point mutation (V600E) in BRAF. APC is a tumor-suppressor gene and its deletion and point mutations correlate to the development of numerous types of cancers, especially colorectal cancers. BRAF with a V600E mutation is a well-established oncogene in melanoma and colorectal cancer, but it is rarely found in breast cancer. We investigated what signaling drivers were responsible for the proliferation and viability of this cell line.

Probing Oncogenic Protein Kinase Drivers in DU-4475 Cell Line Viability
To determine what signaling pathways are essential for the viability of DU-4475 cells, we screened it against a panel of 20 protein kinase inhibitors (PKI) ( Table 2). These inhibitors included 10 PKIs against various receptor protein tyrosine kinases that have been associated with some cancer types. Other intended targets included protein kinases in the MAPK pathway, PI3K pathway, and Src kinases. Many of these kinases are suggested to be involved in TNBC development. Most of the inhibitors are approved as treatments for appropriate cancers. Figure 1 shows the responses of DU-4475 to these drugs at four concentrations from 10 nM to 10 µM. DU-4475 cells did not strongly respond to Akt inhibitors (Figure 1a), Src or Abl inhibitors (Figure 1b), nor ten inhibitors against various receptor PTKs (Figure 1c). However, they were potently inhibited by a BRAF inhibitor, dabrafenib, and two Mek inhibitors, trametinib and binimetinib (Figure 1d). The most potent inhibitor was trametinib, which appeared to fully inhibit DU-4475 viability at 10 nM. These results are consistent with the GDSC1 data.
To better characterize the response of DU-4475 to these drugs, cell responses to the BRAF and Mek inhibitors at 16 concentrations were determined (Figure 2a). The doseresponse data to these four drugs were fitted to the Hill equation and the inhibitory parameters are shown in Table 3. The data confirmed that DU-4475 cells are most sensitive to trametinib, followed by dabrafenib and binimetinib. The IC 50 of 0.28 nM for trametinib is more potent in DU-4475 than previously reported inhibition of any other cancer cell line [62]. It is also notable that vemurafenib showed only mild inhibition, with an IC 50 of 507 nM, while dabrafenib was much more potent, with an IC 50 of 2.4 nM. All four drugs inhibited the cell viability to 95% or more (I max ) and the inhibition displayed mild positive cooperativity in inhibiting cell viability, with the n values above 1. These characteristics indicate that each drug can shut down DU-4475 viability by binding to a single high-affinity target, BRAF or Mek. This result suggests that DU-4475 cells are solely dependent on BRAF signaling for viability. Considering that DU-4475 contains an oncogenic BRAF V600E mutation, these data suggest that DU-4475 is a mono-driver cancer cell line. To better characterize the response of DU-4475 to these drugs, cell responses to the BRAF and Mek inhibitors at 16 concentrations were determined (Figure 2a). The doseresponse data to these four drugs were fitted to the Hill equation and the inhibitory parameters are shown in Table 3. The data confirmed that DU-4475 cells are most sensitive to trametinib, followed by dabrafenib and binimetinib. The IC50 of 0.28 nM for trametinib is more potent in DU-4475 than previously reported inhibition of any other cancer cell line [62]. It is also notable that vemurafenib showed only mild inhibition, with an IC50 of 507 nM, while dabrafenib was much more potent, with an IC50 of 2.4 nM. All four drugs inhibited the cell viability to 95% or more (Imax) and the inhibition displayed mild positive cooperativity in inhibiting cell viability, with the n values above 1. These characteristics indicate that each drug can shut down DU-4475 viability by binding to a single high-affinity target, BRAF or Mek. This result suggests that DU-4475 cells are solely dependent on BRAF signaling for viability. Considering that DU-4475 contains an oncogenic BRAF V600E mutation, these data suggest that DU-4475 is a mono-driver cancer cell line.

Trametinib and Dabrafenib Fully Block Mek and Erk Activation in DU-4475
To confirm that these drugs are blocking the intended targets, we determined the effects of drug treatments on the phosphorylation states of key protein kinases in the MAP kinase pathway (Figure 2b). BRAF was phosphorylated on Ser445 in untreated cells and the phosphorylation level was not affected by any of the inhibitors. Trametinib and dabrafenib nearly fully blocked the phosphorylation of Mek (Ser217 and Ser221) and Erk (Thr202/Tyr204) at 10 nM and 100 nM, respectively, while vemurafenib only inhibited

Trametinib and Dabrafenib Fully Block Mek and Erk Activation in DU-4475
To confirm that these drugs are blocking the intended targets, we determined the effects of drug treatments on the phosphorylation states of key protein kinases in the MAP kinase pathway (Figure 2b). BRAF was phosphorylated on Ser445 in untreated cells and the phosphorylation level was not affected by any of the inhibitors. Trametinib and dabrafenib nearly fully blocked the phosphorylation of Mek (Ser217 and Ser221) and Erk (Thr202/Tyr204) at 10 nM and 100 nM, respectively, while vemurafenib only inhibited their phosphorylation at 1000 nM. These observations are consistent with the potency of each drug, inhibiting cell viability.
It is notable that trametinib not only inhibited Erk phosphorylation by Mek, but also inhibited Mek phosphorylation by BRAF. This pattern of inhibition by trametinib is consistent with its reported mechanism of action [63][64][65]. Even though trametinib is generally referred to as an Mek inhibitor, it more potently inhibits Mek phosphorylation by BRAF than it inhibits Mek phosphorylation of Erk. This is achieved by interfering with BRAF's ability to recognize Mek, even though it is not a general BRAF inhibitor in phosphorylating other BRAF substrates [63]. Thus, trametinib should be more accurately referred to as an inhibitor of Mek activation by BRAF rather than an inhibitor of Mek activity.

Blocking BRAF or Mek Inhibits Proliferation and Causes Cell Death in DU-4475
A cancer drug may inhibit cell proliferation and/or cause cell death. To determine the mechanism of inhibition of trametinib and dabrafenib on DU-4475, cell viability over time of treatments was monitored. Figure  It is notable that trametinib not only inhibited Erk phosphorylation by Mek, but also inhibited Mek phosphorylation by BRAF. This pattern of inhibition by trametinib is consistent with its reported mechanism of action [63][64][65]. Even though trametinib is generally referred to as an Mek inhibitor, it more potently inhibits Mek phosphorylation by BRAF than it inhibits Mek phosphorylation of Erk. This is achieved by interfering with BRAF's ability to recognize Mek, even though it is not a general BRAF inhibitor in phosphorylating other BRAF substrates [63]. Thus, trametinib should be more accurately referred to as an inhibitor of Mek activation by BRAF rather than an inhibitor of Mek activity.

Blocking BRAF or Mek Inhibits Proliferation and Causes Cell Death in DU-4475
A cancer drug may inhibit cell proliferation and/or cause cell death. To determine the mechanism of inhibition of trametinib and dabrafenib on DU-4475, cell viability over time of treatments was monitored. Figure    hibition of DU-4475 proliferation, while complete inhibition of BRAF or Mek signaling fully blocked DU-4475 proliferation and caused cell killing. These results strongly suggest that DU-4475 is a mono-driver cancer cell line predominantly supported by a single oncogenic signaling driver, BRAF V600E.
We determined if the DU-4475 killing by trametinib was via apoptosis. DU-4475 cells treated with trametinib for 48 h were collected and the cleavage of apoptotic caspases and poly (ADP-ribose) polymerase (PARP) was analyzed. The cleavage of caspases activates them and initiates apoptosis. The cleavage of PARP inactivates it and suppresses DNA repair, which is also associated with apoptosis. Caspases 3, 7, and 9 were cleaved to expected sizes in response to trametinib treatments (Figure 3d). PARP was cleaved into an 89 kD fragment, which is also an indicator of apoptosis. These results suggest that blockage of the predominant driver pathway by trametinib is sufficient to inhibit cell proliferation and induce apoptosis in DU-4475 cells.
We used the Annexin V apoptosis and necrosis assay to confirm that trametinib caused cell death in DU-4475 through apoptosis. The RealTime-Glo™ Annexin V Apoptosis and Necrosis Assay (Promega, Madison, WI, USA) measures the real-time translocation of phosphatidylserine on the outer leaflet of cell membranes during early apoptosis [57]. The assay also measures the loss of membrane integrity using a DNA-binding dye, which enters the cell and generates a fluorescent signal. Figure 4a displays the increase in the luminescence and fluorescence in DU-4475 cells due to trametinib treatment. The differential in luminescence signal between treated and control cells peaked at 11 h of treatment, an indication of early apoptosis. The subsequent loss of the luminescence signal is an indication of the loss of membrane integrity. This indicated that early apoptosis started at~10 h treatment. The fluorescence signal increased gradually following early apoptosis, corresponding to the loss of membrane integrity due to necrosis. These results are consistent with the Western blotting results, indicating that trametinib caused apoptosis and secondary necrosis.    [66,67]. Second, we previously determined that MDA-MB-231 is driven by both Src and the MAPK pathway [53], making MDA-MB-231 a multi-driver cancer cell line. A comparison between DU-4475 and MDA-MB-231 might offer insights into the multi-driver versus mono-driver oncogenic mechanisms.
We compared the inhibition of MDA-MB-231 by different Mek inhibitors and dasatinib, an Src inhibitor (Table 4). Consistent with our previous report on MDA-MB-231, individual Mek inhibitors, trametinib, selumetinib, or binimetinib, are not potent inhibitors of MDA-MB-231, reaching only a maximal inhibition of 43% for selumetinib and somewhat less for the other two inhibitors. However, even with partial inhibition, trametinib displayed a better potency (K d = 13 nM) than the other Mek inhibitors (K d of 563 nM and 199 nM for selumetinib and binimetinib, respectively) at achieving the maximal inhibition. The lack of full inhibition of MDA-MB-231 cell viability by these inhibitors is not due to restricted cell permeability nor failure to bind to their intended targets. Each drug effectively inhibited the phosphorylation and activation of Erk, trametinib at 10 nM and both binimetinib and selumetinib at 100 nM ( Figure 5). Consistent with trametinib being an inhibitor of Mek activation by BRAF, trametinib also inhibited Mek phosphorylation, while binimetinib and selumetinib did not. These results suggest that blocking Mek and Erk in the MAP kinase pathway is not sufficient to block the viability of MDA-MB-231 cells. This suggestion is consistent with MDA-MB-231 being a multi-driver cancer cell line.
Dasatinib inhibited MDA-MB-231 in a biphasic manner, with the first phase accounting for 55% of cell viability and a K d1 of 27 nM. This pattern is similar to what has been previously reported [53]. These analyses confirmed that the MAP kinase pathway and Src are independent drivers in MDA-MB-231. Thus, we tested the effects of dasatinib and trametinib combination on MDA-MB-231 viability. The combination of equal concentrations was an extremely potent inhibitor cocktail for this cell line, with an IC 50 of 8.2 nM (Figure 6a and Table 4), about 10-fold more potent than dasatinib + selumetinib or dasatinib + binimetinib combinations (Table 4). At the end of 48 h treatments, dasatinib (100 nM) and trametinib (100 nM)-treated cells displayed similar morphology to control cells. In contrast, the combination-treated cells were killed, with some cell debris remaining (Figure 6b). effectively inhibited the phosphorylation and activation of Erk, trametinib at 10 nM and both binimetinib and selumetinib at 100 nM ( Figure 5). Consistent with trametinib being an inhibitor of Mek activation by BRAF, trametinib also inhibited Mek phosphorylation, while binimetinib and selumetinib did not. These results suggest that blocking Mek and Erk in the MAP kinase pathway is not sufficient to block the viability of MDA-MB-231 cells. This suggestion is consistent with MDA-MB-231 being a multi-driver cancer cell line.  Dasatinib inhibited MDA-MB-231 in a biphasic manner, with the first phase accounting for 55% of cell viability and a Kd1 of 27 nM. This pattern is similar to what has been previously reported [53]. These analyses confirmed that the MAP kinase pathway and Src are independent drivers in MDA- Especially remarkable is the incredible synergy between dasatinib and trametinib at inhibiting MDA-MB-231 cell viability. The combination reached near-complete inhibition of MDA-MB-231 viability at 100 nM, while either drug alone did not reach that level of inhibition at 20 µM. The synergy is strong at all inhibition levels and it becomes especially striking at 60% or higher levels of inhibition. For example, the IC 70 (drug concentration for 70% inhibition) is 25 nM for the combination, 12.6 µM for dasatinib and above 20 µM for trametinib. These numbers translate into a combination index (CI) below 0.0039 and a dose-reduction index (DRI) above 257 [55,68] (Figure 6c). Thus, the drug combination is >250-fold more potent than dasatinib and trametinib combined if there was no synergy. This combination offers the most potent targeted drug combination for this cell line. The potency is even more favorable than some targeted drugs for mono-driver cancers approved for clinical applications.
To determine if the dasatinib/trametinib combination is specific for MDA-MB-231 cells, we compared the inhibition of MDA-MB-231 to that of MDA-MB-468 (Figure 6d). MDA-MB-468 has been shown to overexpress EGFR and lack PTEN and it is sensitive to inhibition by the combination of lapatinib and GSK-690693. MDA-MB-468 cells are >1000-times more resistant to the combination of dasatinib and trametinib, with an IC 50 above 10 µM. This result demonstrates that the sensitivity of MDA-MB-231 to the dasatinib/trametinib combination is determined by the unique oncogenic driving mechanism in this cell.
MB-231. Thus, we tested the effects of dasatinib and trametinib combination on MDA-MB-231 viability. The combination of equal concentrations was an extremely potent inhibitor cocktail for this cell line, with an IC50 of 8.2 nM (Figure 6a and Table 4), about 10-fold more potent than dasatinib + selumetinib or dasatinib + binimetinib combinations (Table 4). At the end of 48 h treatments, dasatinib (100 nM) and trametinib (100 nM)-treated cells displayed similar morphology to control cells. In contrast, the combination-treated cells were killed, with some cell debris remaining (Figure 6b). Especially remarkable is the incredible synergy between dasatinib and trametinib at inhibiting MDA-MB-231 cell viability. The combination reached near-complete inhibition of MDA-MB-231 viability at 100 nM, while either drug alone did not reach that level of inhibition at 20 μM. The synergy is strong at all inhibition levels and it becomes especially striking at 60% or higher levels of inhibition. For example, the IC70 (drug concentration for 70% inhibition) is 25 nM for the combination, 12.6 μM for dasatinib and above 20 μM for trametinib. These numbers translate into a combination index (CI) below 0.0039 and a dose-reduction index (DRI) above 257 [55,68] (Figure 6c). Thus, the drug combination is >250-fold more potent than dasatinib and trametinib combined if there was no synergy. This combination offers the most potent targeted drug combination for this cell line. The potency is even more favorable than some targeted drugs for mono-driver cancers approved for clinical applications.
To determine if the dasatinib/trametinib combination is specific for MDA-MB-231 cells, we compared the inhibition of MDA-MB-231 to that of MDA-MB-468 (Figure 6d). MDA-MB-468 has been shown to overexpress EGFR and lack PTEN and it is sensitive to inhibition by the combination of lapatinib and GSK-690693. MDA-MB-468 cells are >1000-

Blocking Each Driver Partially Inhibits Cell Proliferation, While Blocking Both Drivers Induces Apoptosis in MDA-MB-231
The strong synergy between dasatinib and trametinib prompted us to further investigate the mechanism of the drug combination. Figure  We next determined the concentration-dependent effects of trametinib, dasatinib, and the combination on cell viability in 48 h treatments (Figure 7b-d). Treatment with trametinib for 48 h resulted in a dose-response curve above the 1 h treatment curve at all concentrations (Figure 7b), indicating that trametinib only mildly inhibited cell growth but did not kill the cells up to 1 µM. Similarly, the 48 h dose-response curves for dasatinib were above the 1 h treatment at all concentrations (Figure 7c), indicating that dasatinib inhibited cell growth but did not kill the cells up to 1 µM. In contrast, the dose-response curve for the 48 h combination treatment was above the 1 h curve below 10 nM and crossed over to below the 1 h curve at 100 and 1000 nM (Figure 7d). This cross-over pattern indicated that the number of viable cells after 48 h treatment with the drug combination dropped below that of the amount observed at 1 h, indicating that cells were killed in the combination but not with trametinib or dasatinib alone. Western blotting (Figure 7e) demonstrated that the drug combination at 100 nM caused significant cleavage of caspases 3, 7, and PARP, indicating cell death is through apoptosis.

Induces Apoptosis in MDA-MB-231
The strong synergy between dasatinib and trametinib prompted us to further investigate the mechanism of the drug combination. Figure 7a displays the growth curves of MDA-MB-231 in the presence of 1 μM trametinib, dasatinib, and their combination. Dasatinib or trametinib alone at 1 μM resulted in a slower increase in cell viability up to 72 h, but neither decreased cell viability. The combination of dasatinib and trametinib at 1 μM resulted in a slight increase in cell viability up to 24 h and a dramatic decrease in cell viability in 48 h and 72 h, indicating that the combination resulted in near-complete cell killing (Figure 7a).  The increase in apoptosis when both pathways are blocked also provides a mechanistic explanation for the dramatic synergy observed earlier. When only one driver is blocked by a drug, the cells are still able to survive or even grow with the support of the other pathway. When both drivers are blocked, the cells lose all growth stimulation and initiate apoptosis. This is in direct contrast with that of DU-4475, for which apoptosis is induced by one drug. These results suggest that a drug blocking one driver can kill mono-driver cancer cells, but a multi-driver cancer cell can be killed only by a drug combination blocking both signaling pathways.

Discussion
Despite dramatic progress in targeted cancer therapies blocking oncogenic signaling, only a relatively small number of cancers have benefitted from effective therapy and most, if not all, effectively treated cancers are mono-driver cancers. Most cancers are dependent on multiple drivers, making them naturally resistant to treatments targeting any single driver. Triple-negative breast cancers appear to be mostly multi-driver cancers. Understanding the molecular basis of multi-driver versus mono-driver oncogenesis is a major challenge for developing targeted therapies for TNBC.

DU-4475 as a Mono-Driver Cancer Cell Model
This study provides evidence that DU-4475 is a mono-driver cancer cell line. DU-4475 cells harbor a BRAF V600E mutation, but the functional significance of this mutation in this cell line is not fully clear. While BRAF V600E is a predominant driver in some cells, especially in melanoma, other cells with this mutation are not entirely dependent on it for oncogenic proliferation. In DU-4475, blocking BRAF or Mek kinase alone is sufficient to completely block cell viability and cause apoptosis (Figures 2-4). These results suggest that BRAF V600E-activated MAP kinase signaling is essential for the survival of DU-4475 cells and blocking this pathway at either step is sufficient to kill DU-4475 cells. To our knowledge, this is the only TNBC cell line that is dependent on a single oncogenic driver, i.e., a mono-driver cancer cell line.
The inhibition of a mono-driver cancer cell line by the inhibitor for its driver displays three characteristics when the dose-response data are fitted into the Hill equation: an IC 50 consistent with the target kinase inhibition, an I max close to 100%, and a Hill slope, n, around 1. When a cell response to a given PKI meets these criteria, the target kinase of the drug can be considered the predominant driver and the cancer cell can be considered a mono-driver cancer cell. DU-4475 cells meet these criteria as they are potently and fully inhibited by BRAF or Mek inhibitors, demonstrating the cells' dependence on BRAF V600E for cell viability. The most noteworthy drug is trametinib, which inhibited DU-4475 cells with an IC 50 of 0.28 nM and reached 95% inhibition at about 1 nM. It can be expected that cancer with a similar mechanism would be susceptible to targeted therapy with trametinib.

BRAF V600E as a Therapeutic Target in TNBC
A BRAF mutation is rare in breast cancer and TNBC, but several-BRAF V600E-driven TNBC cases have been reported. Pircher et al. reported [69] that a 38-year-old female TNBC patient with a BRAF V600E mutation developed multiple lung metastases. The patient was treated with vemurafenib, which resulted in clear radiological improvements within three months. At the time of the report (19 months after the therapy started), the lung metastases remain radiologically stable and the patient remains in good clinical condition. The second case of a TNBC patient with a BRAF V600E mutation was reported by Wang et al. [70]. This 60-year-old patient was first treated with chemotherapy to achieve a 7-month progression-free survival (PFS). The patient received vemurafenib and albumin-bound paclitaxel as a second-line therapy, exhibiting regression of some metastatic pulmonary lesions with concomitant progression of other lesions and achieved a 4.4-month PFS. Genetic testing of the progressed pulmonary lesion revealed the presence of the BRAF V600E mutation together with other newly acquired mutations in PDGFRB, NF2, GRM3, MLH1, FOXA1, LRP1B, and amplification of androgen receptor. The patient ultimately died of multiple organ failure and achieved 12 months of overall survival. The third TNBC patient with a BRAF V600E mutation was a 57-year-old woman with metastatic TNBC and chemotherapy-refractory massive pleural effusion [71]. She also had oncogenic PIK3CA H1047R mutation, among others. After failure with anthracycline-and taxanebased chemotherapy and palliative radiotherapy, the patient was treated with a dabrafenib and trametinib combination. The patient exhibited improved conditions with decreases in swelling and pain, a decrease in pleural fusion, and a reduction in the size of the axillary lymph nodes for several weeks. However, the patient developed another subcutaneous tumor and died 12 weeks after initiating the dabrafenib/trametinib treatment. All three case reports support BRAF V600E as a valid treatment target in TNBC. These case studies also make it clear that the acquisition of additional oncogenic driver mutations would make it ineffective to treat the patient with a BRAF inhibitor alone and finding effective drug combinations that block all oncogenic drivers would be necessary. The presence of the H1047R mutation in PIK3CA, a well-established oncogenic driver [72], in the third patient, suggests that the PI 3-kinase pathway is also activated in this patient, likely making it a multi-driver cancer. A drug combination blocking both BARF and the PI 3-kinase pathway, such as PI 3-kinase inhibitors or Akt inhibitors, may have been more effective than the dabrafenib and trametinib combination.

Blocking Both Oncogenic Drivers in MDA-MB-231 Causes Synthetic Lethality
Most cancers are likely dependent on multiple oncogenic drivers. We previously demonstrated that several colorectal cancers and TNBC cell lines are multi-driver cancer cells, including MDA-MB-231. The identification of a mono-driver cancer cell model in DU-4475 enabled us to compare the drug response patterns of a mono-driver versus a multi-driver cancer cell model. While blocking the predominant driver in a mono-driver cancer cell line is sufficient to fully inhibit cell proliferation and cause cell death, blocking a driver in a multi-driver cancer model only causes partial inhibition to cancer viability and does not kill the cancer cells. The MDA-MB-231 cells are dependent on the MAP kinase pathway and Src signaling for survival and blocking either the MAP kinase pathway or Src only partially inhibits cell viability and without causing cell death. Only the combination of Src and Mek inhibitors completely inhibits cell proliferation and causes programmed cell death. This crucial difference between mono-driver and multi-driver cancer cells provides an explanation why single-agent targeted therapy only works well for mono-driver cancer cells but not for multi-driver cancer cells and supports combination targeted therapy as a viable strategy against multi-driver cancers.
The drug combination is also strikingly synergistic. Dasatinib displays a biphasic inhibition with a potent phase 1 (F 1 = 55% and K d1 = 27 nM) and an F 2 of 45% with K d2 of 18 µM. With both phases combined, dasatinib is still not an effective inhibitor, reaching 70% inhibition at 12.5 µM. Trametinib alone is even less effective, only causing a maximal inhibition of~20%. However, the combination is much more potent, with an IC 50 of 8.2 nM, and achieved near-complete inhibition at about 100 nM. The strong synergy is also reflected in the CI of <0.003 and DRI > 300. This striking synergy can be explained by the multi-driver oncogenic mechanism. Because the cells are supported by two pathways, blocking either Src or the MAP kinase pathway alone still leaves the cells viable and only blocking both pathways causes cell death. The striking synergy supports the potential effectiveness of combination targeted therapy for multi-driver cancers.

Single-Drug Lethality versus Synthetic Lethality
An ideal cancer drug should inhibit cancer cell growth and cause cancer cell lethality with specificity. Our data demonstrate that for mono-driver cancer cells, such as DU-4475, the specific lethality is achieved with a single drug at low nM concentrations. This is presumably the underlying basis for the success of targeted therapies against mono-driver cancers. Because multi-driver cancer cells can survive with the support of multiple drivers, pharmacological lethality is not achievable with any one targeted drug blocking any driver and is only achievable when a drug combination blocking all drivers is used. Combinational inhibition of MDA-MD-231 can be considered a form of pharmacological synthetic lethality. This pharmacological synthetic lethality is rooted in the multi-driver nature of MDA-MB-231 cells, as they have acquired an overlapping dependence on the MAP kinase pathway and Src kinases. The partial functional redundancy between these two pathways likely makes MDA-MB-231 cells much more resilient and gives them a significant growth advantage and adaptability. The advantage of this multi-driver overlapping mechanism to the cancer cells is evident and likely applicable to other multi-driver cancer cells. Further research is needed to establish this as a broadly utilized mechanism.
Traditionally, synthetic lethality is defined at the genetic level, where the loss-offunction mutations in two genes cause lethality [73,74], while the loss of either gene alone is not lethal. This concept was further expanded to include the lethality caused by genetic mutation in one gene and chemical inhibition of another. For example, BRCA1 and BRCA2 mutations and PARP inhibitors are synthetically lethal, which has become the conceptual basis for the new therapeutic paradigm of PARP inhibition for cancer patients carrying BRCA1 and BRCA2 mutations [75]. The pharmacological synthetic lethality between dasatinib and trametinib observed in MDA-MB-231 cells carries this concept a step further. This concept could provide a molecular basis for combination targeted therapies in multidriver cancers.

Conclusions
Triple-negative breast cancer is notoriously heterogeneous, with different patients using different genetic and biochemical mechanisms of oncogenesis. This heterogeneity is clearly illustrated in the two cancer cell lines used in this study. Even though both DU-4475 and MDA-MB-231 have mutations in the MAP kinase pathway, BRAF V600E in DU-4475 and KRAS G13D in MDA-MB-231, these two cell lines have one critical difference. While DU-4475 is a mono-driver cancer cell line, solely dependent on BRAF for survival and proliferation, MDA-MB-231 relies on both the MAP kinase pathway and Src signaling. Comparing these cell lines offers critical insights into mono-driver versus multi-driver oncogenesis and strategies for blocking them.
The results demonstrated that DU-4475 cells, as mono-driver cancer cells, can be blocked and killed by BRAF or Mek inhibitors. The MDA-MB-231 cells, in contrast, can still survive when either the MAP kinase pathway or Src kinases are fully inhibited. The cells can only be killed when both pathways are blocked. This pharmacological synthetic lethality offers an underlying basis for killing multi-driver cancer cells by targeted drug combinations. In both cases, cell killing was achieved through apoptosis. These results explain why successful targeted therapies are against mono-driver cancers and offer a proof of concept for a mechanism-based combination targeted therapy strategy for multi-driver cancers. It is well observed that broadly targeting protein kinase inhibitors are generally more toxic toward more cells than those with narrow inhibitory specificity. However, such broadly targeted drugs tend to cause general toxicity. Identifying the oncogenic drivers and drug combinations to specifically block the oncogenic drivers through combination therapy is a plausible precision approach to achieve multi-target inhibition without causing general toxicity. Further validation is needed to determine if this strategy is broadly applicable. While most TNBC cancers are expected to be multi-driver cancers, the functional signaling drivers have yet to be identified in most TNBC models. The elucidation of signaling drivers in other TNBC cell models will further validate this strategy.