STAG2 Mutations Drive Development of Myelodysplastic Syndromes by Promoting R-Loop-Mediated Genomic Instability and Activation of cGAS-STING

Genes encoding components of the cohesin complex, including STAG2, are frequently mutated in myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML), and serve as predictors of high-risk disease and poor outcomes. While it is known that cohesin mutations selectively impair DNA damage repair (DDR) and gene-regulatory functions, their role in transcription-coupled DNA damage is unknown. We have previously characterized transcriptional and splicing consequences of STAG2 mutations and identified genetic and pharmacologic dependencies on DDR and splicing. Furthermore, we identified upregulation of type 1 interferon and cGAS/STING pathways in STAG2 knockout (KO) AML cell lines, both known to be triggered by DNA-RNA hybrids (R-loops) in the cytoplasm. We therefore set out to investigate the role of transcription-coupled R-loops in the development of cohesin-mutant MDS and AML, and their potential as a novel therapeutic liability.

We first analyzed the effects of STAG2 loss-of-function mutations on transcriptional regulation of a curated set of R-loop regulators and associated proteins, and identified significant differential expression of 109/461 (24%) genes, including downregulation of TOP2A and AQR involved in R-loop resolution. This was associated with increased levels of R-loops, as assessed by immunofluorescence staining and dot blot, G1/S cell cycle arrest and DNA damage accumulation in STAG2 KO cells. R-loop resolution using RNase H1 overexpression restored near-normal levels of dsDNA breaks, as well as cell cycle arrest and apoptosis.

To investigate the role of R-loops during MDS development in vivo, we established a new transgenic model of MDS by crossing conditional knockout mice of Tet2 and Stag2, which are frequently co-mutated in MDS. Bone marrow analysis of Tet2/Stag2-mutant mice demonstrated multiple features of MDS, including ineffective hematopoiesis, erythroid and megakaryocyte dysplasia, and poor overall survival due to leukemic transformation. We observed an elevation of R-loops in hematopoietic stem and progenitor cells (HSPC, Lin-cKit+) during the transition from Tet2-mutant clonal hematopoiesis of indeterminate potential (CHIP) to Tet2/Stag2-mutant MDS. This was associated with accumulation of gH2Ax, increased myeloid:erythroid (M:E) colony ratio, and increased clonal potential and serial replating. Notably, overexpression of RNase H1 rescued erythroid differentiation and DNA damage defects, and significantly diminished the serial replating ability of Tet2/Stag2-mutant HSPC, suggesting a causal role of increased R-loops in the MDS phenotype.

We next mapped genome-wide distribution of R-loops during MDS development using the RNase H-based sequencing method, MapR. We detected a significant increase in R-loop peaks in Tet2/Stag2-mutant MDS but not CHIP or WT HSPC. Furthermore, we noted an enrichment of DNA damage signal (gH2Ax ChIP-Seq) at sites of R-loops, and observed increased mutation burden in Tet2/Stag2-mutant bone marrow. Our in vivo Tet2/Stag2-mutant MDS model therefore demonstrated accumulation of R-loops during MDS progression, which was associated with DNA damage accumulation and increased mutation burden. Importantly, R-loop resolution led to significant rescue of the MDS phenotype, including erythroid defects, DNA damage, and normalization of M:E ratio and serial replating.

Finally, we explored the role of R-loops in augmenting inflammatory responses via the activation of the cGAS-STING pathway. STAG2 loss leads to upregulation of the cGAS-STING activated genes, which are fully rescued with RNase H1 overexpression, similar to cGAS or STING knockout. This led us to hypothesize that the cGAS/STING pathway could serve as a therapeutic vulnerability in STAG2-mutant cells. We treated STAG2-wild type and KO cells with the STING agonist diABZI, and observed a significant on-target sensitivity of STAG2 KO cells.

In conclusion, our studies uncover a role for the cohesin complex in R-loop regulation and demonstrate that R-loop perturbation is a key mechanism underlying the development of STAG2-mutant MDS and AML through DNA damage and cGAS/STING activation. Given previous reports of R-loop accumulation in spliceosome and DDX41-mutant MDS and AML, we suggest that R-loop accumulation may be a key unifying downstream mechanism of myeloid malignancy development across multiple genetic backgrounds.