PROJECT SUMMARY This grant is in response to PAR-18-206, Bioengineering Research Grants (BRG). Our goal is to adapt a cutting-edge proteomic network analysis platform, Quantitative Multiplex co-Immunoprecipitation or QMI, to chimeric antigen receptor (CAR) T cell signaling. We will then use CAR-QMI to characterize signal transduction network activation downstream of the CAR, to both understand how the CAR instructs a T cell to attack and destroy cancerous targets, and to make batch-specific predictions about efficacy and side-effect profiles of CAR T cell products. CAR T cells are a breakthrough anti-cancer therapy that recently won FDA approval for relapsed B cell lymphomas. A true “personalized medicine”, CAR T cells are manufactured for each patient from that patient's own T cells by transducing T cells collected by leukopheresis with a viral vector encoding a CAR. However, since each batch is unique, some batches perform better than others in terms of producing remissions and/or deleterious and sometimes fatal side effects including cytokine storms and neurotoxicity. The goal of this project is to develop a “personalized signal transduction network analysis platform” that can screen each batch of CAR T cells and predict the efficacy and side-effect potential of that specific batch. Because signal transduction networks integrate information from multiple input sources- for example costimulatory and immunosuppressive cell surface receptors, patient genetic background, and T-cell specific history of activation- we hypothesize that this readout will be a powerful predictor of function. Our preliminary data show that small changes in CAR design parameters such as scFV binding domain affinity produce measurable changes in signal transduction network state that correlate with functional variables such as target killing ability and cytokine release. Further, we show that there exists considerable individual-to- individual variation in batches of CAR T cells produced from different donors. Therefore, the two prerequisites for an individualized predictive assay are present- variation in our measurement across the population, and the functional relevance of our measurement to outcome parameters. Our interdisciplinary team consists of experts in CAR development, signal transduction, proteomics, and bioinformatics. Our ambitious but achievable goals are to expand the QMI panel to include CAR-specific components; to understand how CAR design parameters influence both signal transduction network states and functional performance measures; and to develop a predictive machine learning algorithm that translates QMI-derived signal transduction network states into a functional biomarker of in vivo clinical efficacy. Successful completion these aims will (1) identify specific proteins or protein interactions that determine clinically-relevant outcomes such as cytokine production or cell killing ability, allowing CAR designers to rationally modify the design of CA...