Activation of endogenous retroviral elements drives inflammation and transformation in STAG2-mutant myeloid malignancies

Abstract

Abstract Cohesin is a multimeric protein complex that regulates three-dimensional chromatin architecture and gene regulation. In myeloid malignancies, mutations in the cohesin subunit STAG2 are common and associated with poor outcomes. STAG2 mutations frequently emerge as secondary drivers in myelodysplastic syndrome (MDS), following mutations in epigenetic modifiers such as TET2 and ASXL1 in clonal hematopoiesis of indeterminate potential (CHIP). While a chronic inflammatory milieu can select for mutant hematopoietic stem cells (HSCs), the specific cause of STAG2-mutant clonal expansion in MDS remains unknown. Using models of cohesin-mutant MDS, here we describe endogenous retroviral element (ERV) reactivation as a driver of inflammation and stem cell transcriptional programs during MDS development. Interferon (IFN) signaling is crucial for antiviral responses and has been linked to the transcriptional reactivation of genomic repeats, which can in turn create double-stranded RNA (dsRNA). We found that patients with STAG2-mutant MDS (n=39) had increased expression of Type 1 IFN signaling compared to wild type (WT) patients (n=126). To understand the mechanism behind STAG2-mutant IFN signaling and its role in cohesin-mutant myeloid malignancies, we first examined isogenic cell line models with and without STAG2 mutations. STAG2-mutant cells showed increased Type 1 IFN signatures, an accumulation of dsRNA, and increased transcription of open chromatin in intergenic regions, as identified by ATAC-Seq. Given the abundance of retroviral sequences in these regions, we hypothesized that altered chromatin structure and subsequent expression of ERV loci could be key drivers of inflammation and subsequent progressionof STAG2-mutant MDS. To address this hypothesis, we characterized differentially expressed transposable elements and their chromatin state in multiple isogenic STAG2-KO leukemia cell lines. We observed transcription of a specific ERV subfamily, HERV3-int and its flanking LTR4 in the absence of STAG2. While these loci were accessible in WT cells, the STAG2 mutation resulted in a loss of the repressive H3K9me3 histone mark and gain of the activating H3K27ac mark, driving transcription, and in specific cases also translation of ERV-derived peptides, which were presented on MHC Class I. Further analysis revealed the presence of the AP-1 transcription factor motifs within these overexpressed ERV loci and binding of canonical AP-1 complex members Fos and Fosl1. We also observed a significant increase in ERV expression in STAG2-mutant MDS patients, further highlighting the association between STAG2 mutations, Type 1 IFN signaling, and ERV reactivation in MDS. Experimentally, repression of the HERV3-int/LTR4 subfamily loci using CRISPRi reversed Type 1 IFN signaling in STAG2-mutant cell lines. Finally, we examined the role or ERV expression in disease progression from CHIP to STAG2-mutant MDS. We generated transgenic mouse models of Stag2-mutant MDS in the context of most frequently co-occurring CHIP lesions. These models showed features of MDS, including macrocytic anemia, thrombocytopenia, and monocytosis with an increase in immature bone marrow populations and significant B-cell aplasia. Analogously to human cells, we identified increased expression of the murine IAPEY4 ERV subfamily and its associated LTR in hematopoietic stem and progenitor cells across all of our Stag2-mutant MDS mouse models. Using scRNA-Seq from sorted lineage-Sca1+kit+ cells, we mapped the highest abundance of IAPEY4 expression to short- and long-term HSCs. To probe the role of ERV reactivation in Stag2-mutant MDS disease progression, we simultaneously repressed multiple members of the IAPEY4 subfamily using CRISPRi in Tet2/Stag2-mutant HSCs in vivo. ERV repression led to decreased inflammatory gene expression signatures and reversed multiple features of the MDS phenotype, including the expansion of short-term HSCs and B-cell aplasia. In summary, our data suggest that STAG2 loss of function mutations lead to chromatin remodeling that promotes ERV expression. These transcripts can then induce an IFN response, which drives HSC dysfunction in MDS, highlighting a cell-autonomous mechanism for inflammation, clonal selection, and MDS progression. We also find that STAG2-mutant MDS cells present ERV-encoded peptides on MHC Class I molecules. Thus, our data not only inform the pathogenesis of MDS development but also have significant therapeutic implications.