Peptide-HLA (pHLA)- targeting therapeutics, such as T cell receptor-engineered T cells (TCR-T), have had clinical success in treating solid tumors. However, challenges related to safety exist: major concerns remain surrounding the cross-reactivity of T cell receptors (TCRs) as well as the ability of therapeutics to discriminate between on-target vs off-target pHLAs while maintaining high potency. Therefore, approaches that survey the diversity of the T cell repertoire to discover optimal TCRs, as well as platforms to comprehensively identify potential off-target liabilities, are critical to de-risking and accelerating the development of this promising class of pHLA-targeting therapeutics. We have developed a strategy that (1) queries the TCR repertoire to enrich and identify multiple active, sequence-distinct endogenous TCRs; (2) uses 3T-TRACE, a high-diversity pHLA library, to screen for cross-reactivity; and (3) exploits functional selections to simultaneously optimize for TCR potency and specificity. We applied this approach to identify TCRs of optimal specificity and potency targeting a peptide derived from the cancer-testis antigen NY-ESO-1 (SLLMWITQC) displayed by HLA-A2. We profiled 6 sequence-distinct TCRs using 3T-TRACE and validated potential endogenous off-target cross-reactivities. Many of the identified off-targets exhibited little to no sequence homology to the NY-ESO-1 epitope, highlighting the importance of diverse combinatorial libraries in identifying unexpected cross-reactivities. Leveraging the off-target liabilities identified by 3T-TRACE, we designed a functional library and selection scheme that enabled the identification of TCRs with increased potency and specificity. Optimized TCRs exhibited enhanced killing activity and improved safety against an NY-ESO-1-expressing melanoma cell line compared to benchmark TCRs, indicating that this approach has potential to improve clinical safety and efficacy. Using this multi-faceted and comprehensive approach we rapidly identified highly potent and specific TCRs against NY-ESO-1. Identifying cross-reactivities using 3T-TRACE proved to be critical in selecting TCRs suitable for engineering and functional selections. This approach can be extended to any pHLA target to create safe and effective TCR-Ts for clinical development. Citation Format: John D. Leonard, Alejandro Ramirez, Jason Romero, Jake Kleiner, Joanna Dreux, Bindu Hegde, Bryan Xie, Akshay Sharathchandra, Nathan Katz, Venita I. DeAlmeida, Hans-Peter Gerber, Marvin H. Gee, Leah V. Sibener. A functional approach to identifying and engineering TCRs results in highly-potent and specific TCRs for TCR-T cell therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 578.

Background T cell engagers (TCEs) are bi-functional biologics that bind a target on tumor cells and the CD3 of the TCR-complex on T cells to induce immune-mediated killing of tumors expressing the epitope of interest. While most TCEs on the market and in development target cell surface antigens with tumor- or lineage-specific expression, this approach cannot exploit the tumor-specific expression of intracellular proteins, which make up the majority of the human proteome. However, peptides derived from virtually all intracellular proteins are presented on the cell surface by Human Leukocyte Antigen (HLA) and can be targeted therapeutically to exploit the tumor-specific expression of intracellular proteins. A key safety concern for therapeutics targeting peptide-HLA complexes is cross-reactivity, as unintended binding of peptides expressed on healthy tissues can lead to lethal off-target toxicity. Methods Here, we utilize Athebody® DARPins (Designed Ankyrin Repeat Proteins), engineered proteins based on an ankyrin repeat scaffold, to generate highly specific binders against an epitope from the cancer-germline antigen MAGE-A4 bound to HLA-A2. Using our proprietary 3T-TRACE™ platform, we perform a global cross-reactivity assessment to identify potential off-target liabilities, many of which are sequence-dissimilar to the intended MAGE-A4 epitope and would therefore be undetectable by traditional methods. The interplay of this cross-reactivity risk evaluation, informing tailored binder selection, allowed us to identify leads of unparalleled efficiency and safety profiles. TCEs armed with these Athebody® DARPins induce high potency killing of MAGE-A4-expressing cancer cell lines with no detectable killing of antigen-negative cells. We further tune the format of these molecules to retain high killing activity while minimizing cytokine release, a key determinant of dose-limiting toxicity in the clinic. Conclusions These optimized TCEs express at high titers and display a favorable developability profile. The work presented here demonstrates a novel approach to develop TCEs with excellent potency and specificity against tumor-restricted intracellular targets.