Engineering Endogenous T Cell Receptors to Recognize Cancer Neoantigens Using a Hybrid Physics-AI Approach

Area of Science:

• Immunology
• Bioengineering
• Artificial Intelligence

Background:

• T cell receptors (TCRs) are crucial for immune surveillance, recognizing peptide epitopes on MHC complexes.
• Current TCR generation through genetic recombination is stochastic, limiting *in silico* design of functional TCRs.
• Designing TCRs *in silico* for specific epitopes is highly desirable but has been unsuccessful until now.

Purpose of the Study:

• To develop a novel artificial intelligence (AI)-powered approach for the *in silico* design of functional TCRs.
• To engineer TCRs capable of targeting specific cancer neoantigens presented by MHC-I.
• To create TCRs with enhanced specificity and anti-cancer activity.

Main Methods:

• Utilized a hybrid approach combining physics-based simulation and generative AI.
• Designed TCRs targeting a HERC1 neoantigen and an EGFR T790M neoepitope.
• Validated engineered TCRs for T cell activation, cancer cell killing, and specificity.

Main Results:

• Successfully engineered multiple functional TCRs against the HERC1 neoantigen.
• Engineered TCRs demonstrated enhanced T cell activation and superior cancer cell killing compared to patient-derived TCRs.
• Developed TCRs targeting the EGFR T790M neoantigen with improved specificity for the mutant sequence.

Conclusions:

• An AI-based approach enables the directed engineering of functional TCRs with enhanced features.
• This method holds broad utility for TCR engineering and the development of novel cell therapies.
• The AI approach represents a significant advancement in designing targeted immunotherapies.