Work

126 Discovery of endogenous retroviral antigens as candidate therapeutic targets in STAG2 -mutant myeloid malignancies

Year: 2026

Type: conference-abstract

Abstract: Background Loss-of-function mutations in the cohesin complex subunit STAG2 are recurrent genetic drivers in myelodysplastic neoplasms (MDS) and acute myeloid leukemia (AML) and are associated with poor clinical outcomes. Despite their prevalence, therapeutic strategies that selectively target STAG2-mutant malignancies remain limited. Notably, patients with STAG2-mutant MDS and AML have demonstrated positive clinical responses to a combination of hypomethylating agents and immune checkpoint blockade, suggesting T cells may play an important role in mediating anti-cancer immunity in this molecular subtype. Previous studies have shown that STAG2-mutant cells show increased transcription of open chromatin in intergenic regions. Given the abundance of retroviral sequences in these regions, we hypothesize that altered chromatin structure and subsequent expression of endogenous retrovirus (ERV) loci could present potential therapeutic targets for STAG2-mutant myeloid malignancies.Methods We employed a multi-omics approach to determine whether ERVs in STAG2-mutant MDS and AML are transcribed, translated, and presented by human leukocyte antigen (HLA) complexes. Total RNA sequencing was performed on isogenic STAG2 wild-type and knockout cell lines to characterize differential gene and transposable element expression and generate custom mass spectrometry (MS) search databases. HLA-I immunoprecipitation and subsequent peptide elution coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS-based immunopeptidomics) detected HLA-bound ERV-derived peptides. Candidate ERV-derived peptides were validated using machine learning-based prediction tools, including HLAthena for HLA binding, Prosit for spectral prediction, and DeepLC for retention time prediction, to increase confidence in peptide identifications.Results Transcriptomic and immunopeptidomic analyses identified differentially expressed transposable elements in STAG2-KO cell lines. In the absence of STAG2, we observed differential transcription of specific loci from the ERV subfamily HERV3-int and its flanking LTR4. Immunopeptidomics identified dozens of ERV-derived HLA-I peptides, including two unique HERV3-int/LTR4-derived peptides upregulated in STAG2-KO cells relative to wild-type controls. These peptides were predicted strong binders of HLA-A*03:01 and HLA-B*18:01 by HLAthena and were further supported by Prosit-predicted tandem mass spectra (MS/MS) and DeepLC-predicted chromatographic retention times.Conclusions Our findings demonstrate that STAG2 loss generates a distinct HLA-I immunopeptidomic landscape characterized by the increased presentation of ERV-derived peptides. The integration of transcriptomics, immunopeptidomics, and machine learning-based validation provides a framework for the systematic discovery of tumor-specific antigens in cohesin-mutant malignancies. These results identify candidate therapeutic targets and support the development of T cell-based immunotherapies for patients with STAG2-mutant MDS and AML.

Source: Journal for ImmunoTherapy of Cancer

DOI: 10.1136/jitc-2026-sitc2026.0126