Systems biology of intratumoral heterogeneity in glioblastoma

PROPOSAL SUMMARY (revisions in orange font ) Patients with glioblastoma (GBM) have a 12-14 month median survival rate, ~10% chance of 5-year survival, and ~90% likelihood of recurrence, even after receiving standard of care (SOC), which involves tumor resection, fractionated radiation therapy (XRT), and chemotherapy with temozolomide (TMZ). There is growing evidence that this poor prognosis and dismal therapy responsiveness emerges from interplay of tumor cell heterogeneity and non-genetic, treatment-induced shifts of cellular phenotypic states. Notably, the SOC has been shown to drive a shift of tumor cells from a drug-susceptible proneural (PN) subtype to a drug-resistant mesenchymal (MES) subtype. This partly explains why primary GBM tumors of the classical or PN subtype often recur as the more aggressive and drug-resistant MES subtype. To complicate matters further, extrinsic signals and stressors can drive dedifferentiation of a heterogeneous tumor cell population into immature, glioma stem-like cells (GSCs), which have been implicated in tumor recurrence. GSCs are resistant to multiple cytotoxic drugs like TMZ, which motivates the need for discovering novel cytotoxic drugs, including drugs repurposed from other indications, to treat GBM. Notably, we have discovered that off-label FDA-approved drugs are effective against patient-derived GSCs (PD-GSCs) increasing median survival of patients by >3X, but can also induce transition of a surviving subpopulation from a susceptible PN subtype to a MES subtype – called PN-to-MES transition (PMT). Here, we propose to elucidate at single-cell resolution the mechanisms by which diverse drugs induce PMT within a heterogeneous population of GSCs. We hypothesize that early response to drug treatments will vary by mechanisms of action of drugs and patient-specific characteristics of PD-GSCs, but cytotoxic events will drive these responses onto a common pathway that can be targeted with genetic and chemical interventions to block drug-induced PMT. We will test this hypothesis by single-cell profiling of longitudinal changes in transcription (scRNA-seq), chromatin accessibility (scATAC-seq), and phenotypes of up to 34 patient-derived GSCs (PD-GSCs) across 76 FDA approved anti-proliferative compounds. We will integrate the longitudinal multi- omic profiles to discover the transcriptional regulatory network (TRN) that mechanistically drives drug-induced PMT in each PD-GSC. By comparing TRNs across PD-GSCs and drug treatments, we will identify, perturb, and characterize mechanisms of drug-induced PMT in each PD-GSC. Using FDA-approved drugs mapped to validated mechanisms, we will perform high throughput screens to evaluate the sensitivity and specificity of our model-driven approach to identify drug combinations that synergistically kill PD-GSCs, without inducing PMT. Outcomes of this project include (i) methodology to elucidate single-cell resolution TRNs within subpopulations of a heterogeneous tumor, (ii) insight into mechanisms of drug-induced PMT in PD-GSCs, and (iii) a rational strategy to repurpose, and tailor FDA-approved combination drug regimens for off-label use in treating GBM. Lines: 30 (0 under/over)