Non-toxic cross-reactivity with healthy tissues supports in vivo expansion and therapeutic efficacy of T cells engineered with synthetic antigen receptors

Background & Aim Clinical success of engineered T cell therapeutics against blood cancers has fueled rapid evolution of the field of T cell-based therapies and created demand for new antigen-binding domains. Among different tumor-targeting domains, single-chain antibodies (scFvs) remain the most prevalent. Extensive screening of scFvs for unwanted cross-reactivities with healthy tissues can reduce the likelihood of selecting ligands reactive to host tissues but not completely eliminate the possibility. Here we describe our experience with two BCMA-specific scFvs and the effect of non-toxic tissue cross-reactivity on the performance of engineered T cells. Methods, Results & Conclusion BCMA-specific scFvs derived from the monoclonal antibodies C11D5.3 and J22.9-xi were used as antigen-binding domains on two synthetic antigen receptor frameworks, T cell antigen coupler (TAC) receptor and chimeric antigen receptor (CAR), and tested against multiple myeloma xenografts. When T cells were engineered with the TAC framework, both scFvs yielded TAC T cells with equivalent in vitro performance; however, the J22.9-xi TAC T cells were more therapeutically robust in vivo. Further analysis revealed superior in vivo expansion of J22.9-xi TAC T cells, compared to C11D5.3 TAC T cells. Surprisingly, J22.9-xi TAC T cells expanded equally well in tumor-free and tumor-bearing mice in the absence of toxicity, suggesting that cross-reactivity to the host supported in vivo expansion and efficacy of J22.9-xi TAC T cells. In the context of a CD28-based CAR, T cells engineered with the J22.9-xi CAR displayed markedly greater levels of exhaustion, compared with T cells engineered with the C11D5.3 CAR, as evidenced by upregulation of checkpoint receptors, higher basal cytokine production, and reduced ability to upregulate cytokine production upon stimulation. Despite functional in vitro impairments, the J22.9-xi-28ζ CAR T cells proved to be the most efficacious and persistent in vivo when compared to the other engineered T cell populations; again, therapeutic efficacy was not associated with overt toxicities. Taken together, these observations suggest that non-lethal cross-reactivity of engineered T cells can augment in vivo potency and enable therapeutic efficacy of T cell products with evidence of functional exhaustion.