Computational Reprogramming of T Cell Antigen Receptor Binding Properties.

Abstract

T-cell receptor (TCR) binding to peptide/MHC is key to antigen-specific cellular immunity, and there has been considerable interest in modulating TCR affinity and specificity for the development of therapeutics and imaging reagents. While in vitro engineering efforts using molecular evolution have yielded remarkable improvements in TCR affinity, such approaches do not offer structural control and can adversely affect receptor specificity, particularly if the attraction towards the MHC is enhanced independently of the peptide. Here we describe an approach to computational design that begins with structural information and offers the potential for more controlled manipulation of binding properties. Our design process models point mutations in selected regions of the TCR and ranks the resulting change in binding energy. Consideration is given to designing optimized scoring functions tuned to particular TCR-peptide/MHC interfaces. Validation of highly ranked predictions can be used to refine the modeling methodology and scoring functions, improving the design process. Our approach results in a strong correlation between predicted and measured changes in binding energy, as well as good agreement between modeled and experimental structures.