Abstract
Spacer or co-stimulatory components in chimeric antigen receptor (CAR) design influence CAR T cell effector function. Few preclinical mouse models optimally support CAR candidate pre-selection for clinical development. Here we use a model in which murine CAR T cells can be exploited with human tumor xenografts. This mouse-in-mouse approach avoids limitations caused by species-specific factors crucial for CAR T cell survival, trafficking and function. We compared trafficking, expansion and tumor control for T cells expressing different CAR construct designs targeting two antigens (L1CAM or HER2), structurally identical except for spacer (long or short) or co-stimulatory (4-1BB or CD28) domains to be evaluated. Using monoclonal, murine-derived L1CAM-specific CAR T cells in Rag-/- mice harboring established xenografted tumors from a human neuroblastoma cell line revealed a clear superiority in CAR T cell trafficking using CD28 co-stimulation. L1CAM-targeting short spacer-CD28/ζ CAR T cells expanded the most at the tumor site and induced initial tumor regression. Treating patient-derived neuroblastoma xenografts with human L1CAM-targeting CAR T cells confirmed the superiority of CD28 co-stimulus. CD28 superiority was also demonstrated with HER2-specific CAR T cells (targeting ovarian carcinoma xenografts). Our findings encourage incorporating CD28 signaling into CAR design for adoptive T cell treatment of solid tumors.