(208d) Identifying Regulatory Mechanisms Driving Adaptive Transitions and Treatment Escape in Glioblastoma Stem-like Cells to Improve Drug Efficacy

Intratumoral heterogeneity, a defining feature of glioblastoma (GBM), hinders our ability to treat this disease effectively. Heterogeneous GBM tumor cell populations harbor a rare subpopulation of cells that exhibit stem cell-like properties and have tumorigenic properties, making them a clinically relevant target. Moreover, these GBM stem-like cells (GSCs) can undergo adaptive transitions that enable them to escape drug treatment. Multiple studies have shown that GSCs undergo drug-induced transitions from a drug-susceptible proneural molecular subtype to a more drug-resistant, aggressive mesenchymal subtype, i.e., proneural-to-mesenchymal transition (PMT). Here, we characterized the longitudinal response of two patient-derived GSC (PD-GSC) populations, one sensitive and one resistant to the drug pitavastatin, a statin having anti-proliferative effects on glioma cells. Our results showed that distinctive adaptive transitions occurred between the two phenotypes – the resistant PD-GSCs transition into multiple distinct transcriptional states while the surviving sensitive PD-GSCs underwent PMT, transitioning across continuum of states to become resistant to pitavastatin. To elucidate the regulatory mechanisms underlying these distinctive responses, we used single-cell RNA-seq profiling and network inference and developed two distinct models of TF-TF interaction networks that putatively drove the transcriptional responses of each drug-response phenotype. Interestingly, distinct TF-TF network topologies characterized each PD-GSC population – a highly interconnected network defined the responsive PD-GSC drug response while a sparse network defined the resistant PD-GSC response. To understand how these distinct topologies contributed to the dynamic responses of these PD-GSCs, we created ODE models to simulate the dynamics of the TF-TF networks and showed that network topologies are a driving factor in the types of phenotypic transitions exhibited by each PD-GSC population. Finally, through in silico perturbation simulations, we identified and experimentally verified multiple TF targets that when knocked down in vitro via siRNAs led to increased cell death during pitavastatin treatment. Our results highlight the need to understand the non-genetic mechanisms associated with cell state transitions and may open an avenue toward the development of new therapeutic strategies aimed at preventing adaptive transitions that enable GSCs to escape treatment.