Chemical Systems Biology Studies of Triple Negative Breast Cancer Cell Lines

Triple negative breast cancer (TNBC) is a highly aggressive type of breast cancer that accounts for 15-20% of breast cancer cases. Targeted therapy remains to be established for TNBC that lacks estrogen receptors, progesterone receptors and human epidermal growth factor receptor HER2. This limits the therapy to traditional chemotherapy, radiation and surgery, which is only beneficial to a fraction of TNBC patients. Transcriptomics-based subtyping of TNBC into six classes, but it is unclear how the transcriptomics-based subtypes link to effective therapeutic strategies, resulting in a poor clinical prognosis in comparison to other breast cancer subtypes. Hence, there is an imminent need for identifying molecular markers and druggable targets against TNBC. This study is focused on the establishment of functional profiling of TNBC cell lines based on their drug vulnerabilities, and to identify novel druggable signaling nodes. We studied a panel of 16 TNBC cell lines using a functional profiling approach in which we measured the responses of TNBC cells to 304 oncology compounds and 355 GSK published kinase inhibitors. The clustering analysis based on overall drug-responses did not match the transcriptomics-based subtypes, suggesting the presence of extensive heterogeneity in TNBC and that the genomic or transcriptomic profiles do not always reflect the functional behavior of these cells. First, to go beyond standard anti-proliferative drug effects, we established a multiplexed readout for both cell viability and cytotoxicity. We identified many drug classes (such as anti-mitotics, anti-metabolites), which generally are assumed to have cytotoxic effects, mostly exhibited strong effects on cell viability but failed to kill the cells. However, in a subset of the cell lines, they induced a selective cell death. In those cases, we identified differential levels of protein markers linked to the cytotoxic responses (e.g. high level PAI-1 linked to anti-mitotics), suggesting their potential use in clinics for therapeutic decision. These results highlighted that simple multiplexed cell viability and cytotoxicity measurements provide more insight in cellular responses towards the treatment and thereby may help in providing better translationally predictive readouts. Second, we devised a novel drug response metric, called normalized drug response (NDR), which accounts for many kinds of screening artifacts such as signal growth rate differences in positive and negative control, as well as in drug-treated conditions. We found that the NDR metric is a time-independent method and it significantly improved the drug response curve fitting. Our NDR will be of great value in cell-based high throughput drug screening approaches as it cuts down the cost and time for the replicate experiments and further validation with cytotoxicity assay. Lastly, we used computational approach to decipher the kinase signal addiction of breast cancer cell lines by integrating vulnerabilities to kinase inhibitors and their polypharmacology data. We developed the kinase inhibition sensitivity score (KISS) to predict single and combinatorial signal addictions. For this study, we used 40 kinase inhibitors with well-defined target selectivities. With this approach, we predicted and validated novel synergistic inhibitor combinations against TNBC cells, such as dasatinib with axitinib, bosutinib with foretinib combinations for HCC1937 cells. This study suggests that drug sensitivity profiling is a powerful strategy for de-convolving cancer cell specific target addictions.N