Biology-based AI Predicts T-cell Receptor Antigen Binding Specificity

Abstract

Adoptive cell transfer (ACT) using T cells modified by the T cell receptor (TCR) gene is an exciting and rapidly developing field. Numerous preclinical and clinical studies have shown varying feasibility, safety, and efficacy of using TCR-engineered T cells to treat cancer and viral infections. Although there is evidence that their use is effective, to what extent and how these therapies can be improved is still a question of research. Since TCR affinity has been generally accepted as the primary role in defining T cell specificity and sensitivity, selecting and generating high-affinity TCRS remains a fundamental approach to designing more effective T cells. However, the traditional approach of increasing affinity by random mutagenesis can cause adverse cross-reactions that result in on-target and off-target adverse events, produce depleted effectors through overstimulation, and ignore other kinetic and cellular parameters that have been shown to affect antigen specificity. In this paper, we review the preclinical and clinical potential of TCR-modified T cells, summarize contributions that challenge the role of TCR affinity in antigen recognition, and explore how structure-guided design can be used to manipulate antigen specificity and TCR cross-reactivity to improve the safety and efficacy of TCR-modified T cells for ACT.