Mechano-dynamic-principle-driven rational design of next-generation TCRs with high specificity and sensitivity

Background: Naturally evolved T-cell receptors (TCRs) achieve extraordinary antigen specificity through a flexible binding interface that forms force-stabilized catch bonds with cognate peptide-MHC (pMHC) complexes. This critical mechano-regulatory feature is often lost in conventional high-affinity engineered TCRs, whose rigid interfaces can paradoxically cause cross-reactivity with self-antigens and off-target toxicity, a major barrier to safe TCR-based immunotherapies. Purpose: Building on our prior elucidation of this mechano-regulation mechanism, we aimed to establish a novel engineering platform for designing next-generation TCRs-including TCR mimics, TCR-engineered cells (TCR-T), and T cell engagers (TCEs)-with enhanced specificity and sensitivity, using the KRAS-G12D neoantigen (presented by HLA-A*11:01) as a proof of concept. Methods: Following an initial screen, top-performing TCR clones were selected for comprehensive characterization. We evaluated their biophysical binding and in vitro function (single-molecule force spectroscopy, kinetic-functional mapping, serial tumor killing, memory phenotype differentiation), in vivo anti-tumor efficacy (KRAS-G12D cell-derived xenograft models), and safety profile (whole-proteome cross-reactivity screening). Results: Mechano-optimized TCRs exhibited ~40-fold higher antigen sensitivity in vitro compared to conventional high-affinity TCRs, with no detectable cross-reactivity to wild-type (WT) KRAS; TCR-T cells engineered with these optimized TCRs showed superior serial tumor killing, reduced exhaustion (37% lower PD-1 expression vs. controls), and enhanced differentiation into long-lived memory phenotypes (2.2-fold higher CD62L⁺CD44⁺ cells); In KRAS-G12D CDX models, mechano-optimized TCR-T cells achieved robust tumor clearance (mean tumor volume reduction of 82% vs. 41% in controls at day 21) and displayed a more infiltrative, less exhausted T-cell phenotype; Whole-proteome cross-reactivity screening confirmed exceptional specificity, with no off-target binding to 10,000+ human proteins. Conclusions: Our study establishes a mechano-principle-driven paradigm for rational TCR engineering. By translating fundamental insights into TCR-pMHC mechanics into a practical design framework, we generated next-generation TCRs -including TCEs, TCR mimics, and TCR-T cells -with unprecedented specificity, sensitivity, and potency. This platform offers a promising route toward safer and more effective immunotherapies for solid tumors and leukemia. Our lead TCR-T candidate is currently advancing into IND-enabling studies, alongside a complementary in vivo TCR engineering program aimed at expanding the therapeutic reach of this mechano-principled approach.