Unannotated open reading frames expand the antigen repertoire in T cell acute lymphoblastic leukemia

Abstract Background: T cell acute lymphoblastic leukemia (T-ALL) is an aggressive hematological malignancy in which immature T cells clonally expand and displace normal immune cells in the bone marrow and peripheral blood. Standard chemotherapy induces remission in many patients, yet relapse rates remain high and are difficult to manage. A major challenge in treating T-ALL is its low mutational burden relative to most adult tumors which limits the availability of neoantigens for immunotherapy. No immunotherapies are currently approved for T-ALL. In this study we aimed to identify novel unannotated open reading frames as a source of immunogenic antigens in T-ALL. Objective: To identify immunogenic antigens encoded by novel unannotated open reading frames that are uniquely presented on MHC class I by T-ALL cells and absent from healthy tissues thereby expanding the actionable antigen repertoire for precision immunotherapy. Methods: We applied an integrated approach combining MHC class I immunoprecipitation coupled to liquid chromatography tandem mass spectrometry and RNA sequencing in 15 T-ALL patient derived xenograft samples (PDX). Mass spectrometry spectra were searched against a curated database of proteins from novel unannotated open reading frames. RNA expression of genomic regions encoding the candidate peptides was quantified with BamQuery and normalized to reads per hundred million (rphm). To prioritize peptides unlikely to be presented in healthy cells and tissues we set a minimum threshold of 5.92 rphm, corresponding to a greater than 5 % probability of generating an MHC-I-associated peptide (MAP), based on RNA sequencing data from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) T-ALL cohort (n=265). Peptides with significantly higher expression in both our PDXs and the TARGET T-ALL cohort compared to normal thymocytes (n=11) and Genotype-Tissue Expression (GTEx) collection (30 tissues, 10 donors each) were selected. We performed silico immunogenicity predictions using PRIME 2.0, a computational framework that integrates MHC-I binding predictions with T cell receptor recognition propensities. We classified a peptide as immunogenic when two conditions were met: its %rank was < 0.5, indicating that fewer than 0.5% of random peptides of 8-14 residues achieve an equal or higher predicted immunogenicity score, and the HLA allele predicted to present that peptide was confirmed to be expressed in the same sample. Results: We identified 688 peptides across 15 T-ALL PDX samples (median=71) that mapped to the novel open reading frame database. These MAPs originated from introns 13% (n=89), noncoding RNAs 29% (n=200), intergenic regions 2% (n=14), endogenous retroelements 3% (n=21), and protein coding regions including 5‘ and 3‘ untranslated regions 54% (n=372). Of the 688 peptides discovered, we retained 637 peptides (92.5%) that had RNA expression greater than 0rphm in matched PDX samples. Filtering out those with expression above 5.92rphm in normal thymocytes yielded 97 peptides (14%). Applying a two-fold expression increase filter (p < 0.05) in our PDXs and the TARGET cohort compared to normal thymocytes (n=11) identified 40 peptides (5.8%). Further, we excluded peptides with expressions above 5.92rphm in other healthy cells and tissues including medullary thymic epithelial cells (n=8), dendritic cells (n=12), CD4 T cells (n=4), CD8 T cells (n=4), NK cells (n=4), B cells (n=4), granulocytes (n=5), and GTEx (n=30) retained 5 peptides (0.7%). Each candidate antigen was significantly overexpressed (p < 0.05) in our PDXs (n=33) and TARGET (n=265) relative to GTEx (n=30). These candidate T-ALL-specific peptides were also highly expressed in two other T-ALL patient cohorts: the Dana-Farber Cancer Institute (Tran et al., Blood Advances, 2022; n=37) and Verboom et al. (Haematologica, 2018; n=25). Importantly, all five peptides scored as strongly immunogenic with PRIME %rank below 0.5. Conclusion: We demonstrated that T-ALL cells present novel open reading frame derived MHC class I peptides with high predicted immunogenicity. These results suggest that antigens from non-canonical open reading frames could form the basis of tumor specific immunotherapies in T-ALL.