Antigenicity prediction of peptides generated by chimeric transcripts provides proof-of-concept of tumor-specific cryptic neoepitopes

Abstract

Considerable recent interest lies in the immunogenicity of non-canonical neoantigens harboured in alternative open reading frames generated by non-synonymous mutations, nucleotide insertions / deletions, exonization of untranslated regions of the genome, non-coding transcripts, transposable elements, etc. Amongst these ‘cryptic’ epitopes, those generated by chimeric transcripts [CTs], which represent sequences from more than one gene remain undefined, and their translational potential and immunogenicity is relatively unexplored. In the present study, we predicted and assessed neoepitopes generated by CTs in ovarian cancer. Development and application of a customized targeted peptide database, further led to the identification of unique peptides and reading-frame isoforms [sense/anti-sense] of such transcripts. Prediction of antigenicity of these novel peptides and proteins was achieved considering their stability, proteasomal processing, and likelihood of being presented to MHC-I in an allele-specific manner. Further docking of peptide MHC and peptide-MHC-TCR complexes provided validation and proof-of-concept of generation of a small number of putative neoepitopes of CTs that are recognized by specific TCRs. Importantly, our findings identify that relaxed/perturbed rigidity of canonical transcription and translation in cancer may present new opportunities in immunotherapy.