Abstract
Tumor epitopes—peptides presented on surface-bound MHC I proteins—provide critical targets for cancer immunotherapy. Traditionally, these have been identified within the annotated protein-coding regions of the genome. Motivated by the discovery of translated novel unannotated open reading frames (nuORFs) using ribosome profiling (Ribo-seq), we hypothesized that cancer-associated processes generate nuORFs that serve as a new source of tumor antigens. By generating Ribo-seq profiles for 29 malignant and healthy samples and developing a sensitive hierarchical ORF prediction approach, we constructed a high-confidence database of translated nuORFs. Analysis of mass spectrometry data revealed that peptides from 3,555 unique nuORFs were presented on MHC I, with a >20-fold enrichment in the immunopeptidome compared to the whole proteome. We further identified somatic mutations and tumor-specific translation in nuORFs across melanoma, chronic lymphocytic leukemia, and glioblastoma, significantly expanding the pool of targetable tumor-specific peptides.
Introduction: Beyond the Annotated Genome
The success of modern cancer immunotherapy, including checkpoint inhibitors and personalized vaccines, hinges on the immune system's ability to recognize neoantigens. These are small protein fragments (peptides) displayed on the cell surface by Major Histocompatibility Complex class I (MHC I) molecules. To date, the search for these "ID badges" has focused almost exclusively on the 2% of the human genome containing known, annotated protein-coding genes.
However, many patients possess "cold" tumors with low mutational burdens in these annotated regions, leaving them with few actionable targets. This research investigates the "dark matter" of the genome—specifically novel unannotated open reading frames (nuORFs)—to determine if they serve as a hidden reservoir of tumor-specific antigens that could expand the reach of immunotherapy.
Methods: A Multi-Omic Pipeline for nuORF Discovery
To identify and validate these cryptic proteins, the researchers developed a comprehensive multi-omic pipeline:
- Ribosome Profiling (Ribo-seq): Unlike standard RNA sequencing, which measures transcript levels, Ribo-seq identifies the specific sequences actively being translated into proteins by ribosomes. The team generated profiles from 29 samples, including Chronic Lymphocytic Leukemia (CLL), Glioblastoma (GBM), and melanoma.
- Hierarchical ORF Prediction: A sensitive analytic approach was developed to pool Ribo-seq data across individual samples and tissue types. This hierarchical method allowed for the detection of lowly expressed or tissue-specific nuORFs that are typically missed by standard algorithms.
- nuORFdb v1.0: These efforts culminated in a high-confidence database containing over 237,000 nuORFs.
- Mass Spectrometry (MS/MS) Validation: To prove these nuORFs are actually processed and presented to the immune system, the team searched mass spectrometry data from 92 different HLA (MHC) alleles against their new database.
- Genomic Comparison: The researchers compared Whole Genome Sequencing (WGS) with standard Whole Exome Sequencing (WES) to evaluate how many potential mutations in these regions are currently overlooked in clinical settings.
Key Insights: The Efficiency of nuORF Presentation
The study revealed several striking characteristics of nuORFs that make them ideal candidates for immune recognition:
- "Ready-Made" Antigens: Some nuORFs produce proteins that are naturally 8-12 amino acids long. This is the optimal size for MHC I presentation, allowing these peptides to bypass the complex cellular machinery usually required to chop larger proteins into fragments.
- High Immunogenic Potential: nuORF peptides were found to be 20-fold more likely to be presented on MHC I molecules than proteins from the standard "whole proteome." This suggests that non-canonical translation products are highly efficient sources of immune signals.
- Clarifying "Spliced Peptides": The findings provide a new perspective on "spliced peptides" (fragments of different proteins fused together). The authors demonstrate that many signals previously attributed to splicing are actually standard linear peptides derived from previously unknown nuORFs.
Results: Expanding the Target Landscape
The integration of Ribo-seq and mass spectrometry confirmed that 3,555 unique nuORFs are presented on MHC I. In a patient-derived melanoma sample, the inclusion of nuORFs increased the number of potential mutated neoantigens by 25%.
Furthermore, the study identified cancer-restricted nuORFs that could serve as "off-the-shelf" targets. Examples include nuORFs within the genes ARHGAP44 (specific to CLL) and SOX2-OT (specific to GBM). Overall, nuORFs contribute approximately 2% of the total antigen repertoire presented on MHC I in cancer cells, representing a significant expansion of the known immunopeptidome.
The WES Blind Spot and Future Directions
A critical finding for clinical oncology is the "WES blind spot." Standard clinical sequencing (WES) typically ignores the untranslated regions (UTRs) of genes. The authors found that only about 20% of 5’ UTRs are covered by WES, meaning current diagnostic tools are blind to a vast majority of potential cancer mutations occurring in nuORFs.
While nuORFdb v1.0 represents a massive leap forward, the authors note it is likely not yet saturated. Profiling additional tissue types will likely reveal even more hidden proteins. While these nuORFs are excellent immune targets, their biological functions—whether they are functional proteins or simply "translational noise" exploited by the immune system—remain a subject for future investigation. Regardless of their function, their presence on the cell surface offers a powerful new toolkit for personalized cancer therapy.
Summary updated