Abstract In glioblastoma (GBM), cancer cells break away from the tumor mass and infiltrate into adjacent brain tissue. Like other poor-prognosis cancers, GBM has been extensively analyzed by genome-wide transcriptomic analyses. This has led to the identification of 3-4 subtypes that span a spectrum of states from “Proneural” (PN) to “Mesenchymal” (MES). While the identification of subtypes is intriguing, it has yet to produce clinically- actionable mechanistic insight. In our unpublished work, we discovered key mechanical signatures of these two subtypes. Using our Sleeping Beauty (SB) immunocompetent genetically-induced mouse glioma model, we found that the oncogenic driver NRasG12V promotes a MES-like phenotype and the oncogenic driver PDGFβ promotes a PN-like phenotype. In addition, we found that NRas-driven tumors migrate fast and generate large traction forces, while PDGFβ-driven tumors migrate slowly and generate weaker traction forces, features we also observe with human cells in brain tissue. Thus, the two subtypes may each have their own distinct mechanical weaknesses that can be effectively targeted. Since brute force trial-and-error of possible targets is not feasible, we will manage complexity using the modeling approach that is widely used in engineering. As pointed out in the Overall section of this proposal, the mobility of the cancer cells and the antitumoral T cells are both critical determinants of tumor progression/regression, so we will apply our recently published “Cell Migration Simulator” (CMS1.0) to cancer and immune cell migration and use experimental microscopy measurements made in brain tissue to identify the model parameters for the two GBM subtypes. This will then allow us to identify key mechanical vulnerabilities that will be tested using digital multiplex T cell genome engineering (as described in Project 3) and will provide a computational platform for application to pancreatic cancer and immune cells (in Project 2). To simulate the multicellular migration, proliferation, and immune-mediated killing dynamics, we will apply our “Brownian Dynamics Tumor Simulator” (BDTS1.0) to predict the overall tumor dynamics of the NRas (MES) and PDGFβ (PN) tumors. Interestingly, like the human disease, the NRas (MES) tumors are immunologically ‘hot’, while the PDGFβ (PN) tumors are immunologically ‘cold’. Thus, the BDTS1.0, once developed for these two subtypes of brain tumors, will allow us to predict the effects of emergent immunotherapy concepts developed by our team, including CD200 peptide therapy and Peptide Alarm Therapy. By constraining the simulators with data obtained by live cell fluorescence microscopy, we will develop a multiscale computational model that provides mechanistic de-risking and optimization to maximize the physical proximity and encounter frequency between antitumoral T cells and cancer cells. Together the modeling and experiments will allow us to test our central hypothesis that T cell proximity to cancer cells is a major determinant of successful immunotherapy of solid tumors.

NARRATIVE: PROJECT 1 Brain tumors remain among the most malignant and deadly cancers, with adult glioblastoma (GBM) patients having a median survival of 15 months and five-year survival of less than 10 percent. We aim to use new and innovative simulation tools to better screen GBM immunotherapies before they reach clinical trials. We will also use these simulators to develop and test immunotherapies in mouse GBM models.