Background:
Although immune checkpoint inhibition has transformed melanoma treatment, 40%–50% of patients exhibit primary or acquired resistance. As immunotherapy responsiveness is underpinned by immune cell recognition of the tumour that in turn depends on tumour antigen processing and presentation, mapping the cell-surface melanoma immunopeptidome is vital for discovering efficacious targets. Beyond immunopeptides derived from canonical proteins coded by conventional open reading frames, a proportion originate from non-canonical translational products, such as human endogenous retroviruses (HERVs) and other genomic regions traditionally thought to be “non-coding”. Termed ‘dark’ peptides, these molecules cannot be readily identified using reference proteomes, precluding the identification of all targetable melanoma antigens for effective immunotherapy development.
Methods:
To systematically identify dark peptides, we leveraged genomics, transcriptomics, mass spectrometry-based proteomics and immunopeptidomics, and computational approaches using cell lines and blood samples from Stage 3/4 melanoma patients undergoing immunotherapy.
Results:
In total, >166K and >128K unique peptides with 8-14 amino acids were identified from six melanoma cell lines with and without interferon-gamma (IFNγ) treatment, respectively. Among these peptides, 199 (123 from IFNγ-treated and 76 from untreated melanoma cells) and 2408 (1445 from IFNγ-treated and 963 from untreated melanoma cells) peptides that originated from HERVs and non-traditional open reading frames (ORFs) were identified via searches against public databases. With the help of our large-scale mutant peptide database (CAN-IMMUNE), we further identified >70 potential neoantigens using public tumour-associated mutation data. More importantly, 48 peptides from HERVs and 472 peptides from non-traditional ORFs (out of >80K unique 8-14-mer peptides) were identified using 34 blood samples from 19 Stage 3/4 melanoma patients who underwent at least one cycle of checkpoint inhibitor immunotherapy. We are currently leveraging cell line-specific whole exome sequencing (WES) and RNA-seq data and our ‘immunopeptidogenomics’ (IPG) workflow to further search and profile sample-specific dark immunopeptides.
Conclusions:
Systematic identification of non-canonical antigen sources opens up a novel direction for the exploration of the immunopeptidome. Using multiomic experimental and computational approaches, we identified a number of dark peptides from various non-traditional translation products, thereby providing candidates for immunogenicity analysis and informing antigen-based immunotherapy development for Stage 3/4 melanoma.