(222f) A Systems Approach to Understanding Drug Response in a Heterogeneous Tumor-Cell Population

Single-cell heterogeneity is pervasive across multiple molecular levels within a tumor and supports multiple mechanisms through which cellular subpopulations that are inherently drug resistant can arise or can acquire resistance during treatment. These issues hinder our ability to develop effective treatment strategies. A quintessential example of tumor cell heterogeneity is Glioblastoma (GBM), a highly aggressive and lethal form of primary brain cancer. One approach to address GBM cellular heterogeneity has been to apply high-throughput drug screens in order to identify novel single or combination drug therapies that inhibit growth of glioma stem-like cells (GSCs), a clinically relevant subpopulation of tumor cells that drive tumor formation/recurrence and are resistant to the current standard of care (surgical resection and treatment with the chemotherapeutic temozolomide). Results from our collaborators have demonstrated remarkable differences in the response of patient-derived GSCs to the drug pitavastatin, which has shown potential to inhibit GSC growth. These differences are due in part to the heterogeneous population structure of GSCs, which contain stem-like cells that differ in their tumor-initiation ability, molecular signatures, and drug responses. Understanding how a tumor-cell population is structured (i.e. proportions of subpopulations) and the regulatory mechanisms (e.g. transcription factor and miRNA regulators) that relate or distinguish these subpopulations would provide deeper insight into how tumor-cell heterogeneity contributes to overall tumor-cell population drug response. In this work, we apply a systems approach to determine the regulatory mechanisms that relate or distinguish cellular subpopulations by analyzing the transcriptomic responses of patient-derived GSCs and their time-course response to pitavastatin. Using network inference methodologies, we have identified several transcriptional programs (i.e., transcription factors and corresponding co-regulated target genes) that underlie these distinct subpopulations. This analysis provides insight into potential transcription factors that can be targeted to modulate the drug-response of subpopulations that are resistant to pitavastatin. More generally, this approach will inform the rational selection of molecular targets to attack specific drug-resistant subpopulations within a tumor.