The peptides presented by Human Leukocyte Antigens (p-HLA) play a critical role in adaptive immunity, allowing the surveillance of intra-and extracellular states by T cells. Significant progress has been made in the treatment of different cancer types, including melanoma, using immunotherapeutic approaches. However, this strategy is highly dependent on the identification of tumor-specific antigens(1). Peptides that originate from somatically mutated parts of the genome and carry the mutated sequence and are outstanding candidates for cancer immunotherapy. In addition to mutated p-HLA, recent studies have shown a prominent subset of p-HLAs are modified by post-translation modifications (PTMs) such as phosphorylation(2) and more complex modifications such as ligation of different protein segments by proteasomal splicing(3). Advancements in mass spectrometry have enabled the routine identification of HLA-bound peptides, however the identification of mutated, PTM and spliced p-HLAs remains challenging(4). To address this limitation, we have recently developed workflows which enable us to access to this “dark side” of the Melanoma immunopeptidome.
We have isolated p-HLAs from various Melanoma cell lines by immunoprecipitation followed by RP-HPLC fractionation. Peptide-containing fractions were analysed by LC-MS/MS to generate rich peptidomic datasets. In addition to data generated in our lab, we have also re-analysed a number of publicly available data from Melanoma immunopeptidomic studies(1).
Here we develop a sophisticated data-driven workflow to identify peptides generated by cis-splicing (derived from the same antigen) as well as trans-splicing of melanoma antigens (where distinct proteins contribute peptide segments to the hybrid peptide). By using our workflow, we found around 5-10%, and 10-20% of p-HLAs in Melanoma tumours are cis and trans-spliced, respectively. We have identified over 500 spliced peptides derived from known melanoma antigens, including the MAGE superfamily and tyrosinase. In addition, to account for neopeptides in the peptidome, we have generated a comprehensive Melanoma-specific proteome. In this database, we have incorporated more than 300,000 melanoma-related mutations from the Cancer Genome Atlas into the UniProt human proteome database. Interrogation of our data with this database has led to the identification of more than 50 novel neopeptides from patient tumor samples derived from known/common mutations in Melanoma. We also found more than 20 novel phosphorylated and 12 deamidated p-HLA from known melanoma antigens. Preliminary immunogenicity evaluation of a subset of these identified neopeptides, has identified 15 neoepitopes of relevance for Melanoma immunotherapy.
Multiple novel bioinformatics workflows have been proposed for unrestricted identification of melanoma immunopeptidome. Our results highlight the complexity and diversity of Melanoma neoepitopes and broaden our understanding of potential targets of T cell immunity and have significant implications for further melanoma immunotherapy approaches.