NEURAL-NETWORK-BASED PREDICTION OF CANDIDATE T-CELL EPITOPES

Activation of T cells requires recognition by T-cell receptors of spec ific peptides bound to major histocompatibility complex (MHC) molecule s on the surface of either antigen-presenting or target cells, These p eptides, T-cell epitopes, have potential therapeutic applications, suc h as for use as vaccines. Their identification, however, usually requi res that multiple overlapping synthetic peptides encompassing a protei n antigen be assayed, which in humans, is limited by volume of donor b lood. T-cell epitopes are a subset of peptides that bind to MHC molecu les. We use an artificial neural network (ANN) model trained to predic t peptides that bind to the MHC class II molecule HLA-DR4(0401). Bind ing prediction facilitates identification of T-cell epitopes in tyrosi ne phosphatase IA-2, an autoantigen in DR4-associated type1 diabetes. Synthetic peptides encompassing IA-2 were tested experimentally for DR 4 binding and T-cell proliferation in humans at risk for diabetes. ANN -based binding prediction was sensitive and specific, and reduced the number of peptides required for T-cell assay by more than half, with o nly a minor loss of epitopes. This strategy could expedite identificat ion of candidate T-cell epitopes in diverse diseases.