A robust model of human T cell exhaustion links surface marker phenotype to dysfunction
Evan W. Weber1*, Rachel C. Lynn1, Meena Malipatlolla1, Robert C. Jones1,2, Crystal L. Mackall1
1Stanford Cancer Institute, Stanford University, Stanford, CA, 94305; 2Fluidigm Corporation, South San Francisco, CA, 94080
During chronic viral infection and cancer, continuous antigen exposure results in T cell exhaustion, which is characterized by a hierarchical loss of effector function and sustained co-expression of multiple inhibitory receptors (ex., PD-1, TIM-3, LAG-3). Human T cells expressing a GD2-targeting chimeric antigen receptor (CAR, GD2.28z) develop phenotypic and functional hallmarks of exhaustion after only 10 days in culture due to promiscuous clustering and tonic signaling (Long et. al, Nat Med, 2015). In the present study, we utilized the GD2.28z CAR as a controlled and reproducible in vitro model to interrogate the relationship between canonical exhaustion marker expression and T cell dysfunction. CAR T cells were removed from culture on days 10 and 14 after activation and sorted for phenotypic and functional analysis. GD2.28z CAR T cells co-expressing PD-1, TIM-3, and LAG-3 displayed a marked reduction in IL-2 secretion in response to antigen compared to cells that lacked co-expression of all three markers. Interestingly, PD-1 expression alone did not confer a defect in cytokine response, as PD-1 high expressing cells secreted more cytokine in response to antigen than those expressing lower levels, suggesting that PD-1 may be more strongly associated with activation state rather than exhaustion. Subsequent CyTOF and viSNE analyses of GD2.28z CAR T cells revealed a distinct exhaustion-associated cell population in which canonical exhaustion markers, T cell differentiation markers, and inhibitory receptor ligands were differentially expressed compared to other GD2.28z CAR T cells in the same sample. In contrast to PD-1, PD-L1 emerged from these studies as a marker that more closely correlates with dysfunction and not activation state. In summary, the GD2.28z CAR is a robust model to identify and characterize functionally exhausted T cell subsets.