Engineering TCR-controlled fuzzy logic into CAR T cells enhances therapeutic specificity

Abstract

Chimeric antigen receptor (CAR) T cell immunotherapy represents a breakthrough in the treatment of hematological malignancies, but poor specificity has limited its applicability to solid tumors. By contrast, natural T cells harboring T cell receptors (TCRs) can discriminate between neoantigen-expressing cancer cells and self-antigen-expressing healthy tissues but have limited potency against tumors. We used a high-throughput platform to systematically evaluate the impact of co-expressing a TCR and CAR on the same CAR T cell. While strong TCR-antigen interactions enhanced CAR activation, weak TCR-antigen interactions actively antagonized their activation. Mathematical modeling captured this TCR-CAR crosstalk in CAR T cells, allowing us to engineer dual TCR/CAR T cells targeting neoantigens (HHAT^L8F/p53^R175H) and human epithelial growth factor receptor 2 (HER2) ligands, respectively. These T cells exhibited superior anti-cancer activity and minimal toxicity against healthy tissue compared with conventional CAR T cells in a humanized solid tumor mouse model. Harnessing pre-existing inhibitory crosstalk between receptors, therefore, paves the way for the design of more precise cancer immunotherapies.

Keywords: CAR T cells; TCR; cancer immunotherapy; fuzzy logic; humanized in vivo toxicity model; on-target/off-tumor toxicity; robotics; systems immunology; theoretical modeling.