Quantitative Analysis of Signaling Networks

Glioblastoma (GBM) is the most common malignant form of brain cancer. Even with treatment including surgery, radiation, and temozolomide chemotherapy, the 1 year survival rate is only 35%. To identify specific mediators of GBM progression in a genetically engineered murine model of proneural GBM, we quantified signaling networks using mass spectrometry. We identified oncogenic signaling associated with the GBM model, such as increased phosphorylation of ERK1/2, P13K, and PDGFRA, relative to murine brain. Phosphorylation of CDK1 Y15, which causes G2 /M cell cycle arrest, was measured to be the most differentially phosphorylated site, with a 14-fold increase in the tumors. We used syngeneic cell lines to investigate this checkpoint further and treated these cells with MK-1775, an inhibitor of Weei, the kinase responsible for phosphorylation of CDK1 Y15. MK-1775 treatment resulted in mitotic catastrophe of these cells, as measured by increased DNA damage, abnormal percentages of cells in cell cycle phases, and death by apoptosis. This response was abrogated by inhibiting CDK1 with roscovitine, a CDK inhibitor, demonstrating the necessity of active CDK1 for MK-1775 induced mitotic catastrophe. To assess the extensibility of targeting Weei and the G2 /M checkpoint in GBM, we treated patientderived xenograft (PDX) cell lines with MK-1775. The response was more heterogeneous, but we measured decreased CDK1 phosphorylation, increased DNA damage, and death by apoptosis. These results were validated in a flank GBM PDX model where treatment with MK-1775 increased mouse survival by 1.74-fold. We also quantified the signaling differences in our murine GBM model after treatment with sunitinib, an inhibitor of its driver receptor tyrosine kinase, PDGFRA. Treatment increased survival but lead to a morphological change causing a more invasive phenotype. Pro-migratory signaling was characterized by mass spectrometry, such as increased phosphorylation of Enoi, ELMO2, and tubulins. Invasion was further characterized in a lung cancer model where we identified signaling specific to different ligands that result in similar levels of invasion. We have demonstrated that unbiased, quantitative phosphotyrosine proteomics has the ability to reveal therapeutic targets in tumor models and signaling differences between treatments. Thesis Supervisor: Forest M. White Title: Professor of Biological Engineering 3