Glioblastoma is the most common and deadly cancer of the brain in adults, with abysmal patient survival times. Current therapies are ineffective due in part to the phenotypic plasticity of glioblastoma cells, which enables them to become more chemoresistant in response to inflammatory cytokines, the hypoxic tumor microenvironment, and DNA damaging chemotherapeutics. It is unknown if these varied drivers of chemoresistance operate through common signaling pathways, but the proteins that play critical roles in promoting and maintaining the chemoresistant phenotype represent potential drug targets for glioblastoma combination therapies. To identify signaling pathways that promote chemoresistance, we developed an <i>in vitro </i>model of chemoresistance acquisition in glioma-initiating cells (GICs) grown in two-dimensional adherent culture. Pretreating GICs with transforming growth factor beta (TGFβ) or tumor necrosis factor alpha (TNFα), or culturing them in a hypoxic environment (1% oxygen), altered sensitivity to the chemotherapeutics temozolomide and carboplatin in different ways. Specifically, some conditions surprisingly enhanced response to carboplatin while suppressing the response to temozolomide, and others suppressed the efficacy of both drugs. Analysis of RNA sequencing data using gene set variation analysis identified pathways of likely relevance to chemoresistance to be measured quantitatively with a greater temporal resolution using Luminex bead-based phosphoprotein assays. Luminex assays measured a variety of signaling pathways, including those related to MAP kinase signaling, DNA damage, and other stress-related processes. The high-dimensional signaling data will be used to construct a partial least squares regression (PLSR) machine learning model, a dimensionality reduction approach that will generate testable predictions of the druggable signaling pathways that most strongly control glioblastoma chemoresistance. Testing of those model predictions may identify novel combination therapy approaches for glioblastoma and more broadly demonstrate the utility of data-driven modeling to select useful combinations of drugs.

Data-driven modeling approaches to identify cell signaling mechanisms regulating glioblastoma chemoresistance

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.