T-cell receptors that are k-binding have defined sequence features.

Area of Science:

• Immunology
• Computational Biology
• Structural Biology

Background:

• T cell receptors (TCRs) recognize diverse peptide-MHC targets for immune surveillance.
• The sequence basis of TCR cross-reactivity remains poorly understood.

Purpose of the Study:

• To characterize TCR k-binding across millions of TCR sequences and seven related peptides.
• To identify the role of TCR sequences, particularly CDR3 regions, in encoding cross-reactivity.

Main Methods:

• Utilized an in vitro assay to test k-binding of ~47 million TCRs against seven peptide-MHC targets.
• Developed a machine learning model using BLOSUM-50 embeddings to predict TCR-peptide-MHC binding.
• Assessed CDR3 residue importance by masking individual residues in the machine learning model.

Main Results:

• Identified a hierarchy of TCR CDR3 residue importance influencing k-binding.
• Achieved high predictive performance (F1=0.698, AUPRC=0.745) for TCR-pMHC binding using the BLOSUM-50 embedded model.
• Found that residue importance rankings correlated with computational alanine scan results.

Conclusions:

• TCR sequence, especially CDR3, dictates cross-reactivity.
• Machine learning models can effectively predict TCR-pMHC binding and CDR3 residue importance.
• Therapeutic TCRs require careful evaluation for specificity due to inherent cross-reactivity.