Genomic-DNA Exposed By Somatic Gene Mutations Engages the cGAS/STING Axis to License the NLRP3 Inflammasome in Myelodysplastic Syndromes

Background: The pathogenesis of Myelodysplastic Syndromes (MDS) is linked to constitutive innate immune stimulation that converges upon the NLRP3 inflammasome to induce pyroptosis, a caspase-1 dependent cell death. We have shown that inflammasome assembly is initiated by both cell-extrinsic stimuli such as S100A9 elaborated by Myeloid-Derived Suppressor Cells (MDSC), as well as cell-intrinsic somatic gene mutations (SGM) (Basiorka A, et. al. Blood 2016). SGM of varied classes evoke replication stress caused by transcriptional pauses that can expose genomic DNA to cytosolic sensors through unresolved R-loops or micronuclei formation. The cGMP-AMP Synthase-Stimulator of Interferon Genes (cGAS-STING) is a cell-intrinsic DNA surveillance pathway recognizing both cytosolic pathogenic and autologous DNA, leading to interferon stimulated gene (ISG) transcription and NLRP3 inflammasome activation, key biological features of MDS (Pellagatti A, et. al. Blood 2006; 108:337.). Here, we investigate the contribution of genomic cytosolic DNA engagement by cGAS-STING to NLRP3 inflammasome activation in MDS.

Methods: MDS patient and healthy donor bone marrow mononuclear cells (BMMC) were isolated by Ficoll®-Hipaque method from consented participants at the Moffitt Cancer Center or the National Taiwan University Hospital (NTUH). Immortalized murine C57BL/6 Tet2-/- and MX1Cre/SRSF2P95H as well as respective wild type (WT) control BMMCs were used as MDS SGM models.

Results: We first assessed cGAS-STING activation in MDS BMMC by measuring ISG response by microarray, demonstrating profoundly increased expression of ISG15, CXCL10, Samd9l, and Ifi27l2 in MDS BMMC (n=213) compared to healthy control BMMC (n=20) (p<0.0001, p=0.013, p=0.0007 and p<0.0001, respectively). Likewise, both the Tet2 and SRSF2 SGM models have increased ISG expression compared to WT. Importantly, treatment with the cGAS inhibitor, RU.521 (0.1-1µM, 24 hours), significantly suppressed ISG expression in the SGM models. Further, we confirmed DNA sensor pathway activation in the SRSF2 SGM model by demonstrating upregulated phosphorylation of the interferon-regulatory factor (IRF)3 and IRF7 transcription factors compared to WT controls, and treatment with RU.521 decreased IRF3 phosphorylation. Caspase-1 cleavage in Tet2 and SRSF2 SGM models confirmed NLRP3 inflammasome activation in mutant cells compared to WT controls, and RU.521 treatment decreased active caspase-1 generation in both SGM models. RU.521 treatment of MDS BMMC harboring DNMT3 and Tet2 mutations induced terminal differentiation as evidenced by increased CD11b expression detected by flow cytometry and morphological assessment of Wright-Giemsa stains. Similarly, treatment with RU.521 promoted cytological differentiation in the Tet2 SGM model. We next investigated cytosolic genomic-DNA sources including micronuclei and unresolved DNA:RNA hybrids (R-loops) in the SGM models and primary MDS BMMC. Using immunofluorescence (IF), we found a significantly increased number of micronuclei in MDS BMMC (n=8) compared to controls (n=4), p=0.0248. Additionally, R-loops were increased in MDS primary BMMC (n=5) harboring varied SGM classes vs. controls (n=5) as well as in the SGM models. Finally, to confirm genomic-DNA engagement by cGAS, we used IF to assess cGAS co-localization. cGAS co-localized with micronuclei at sites of envelope collapse as well as with R-loops in MDS BMMC, thereby demonstrating cGAS/STING engagement of cytosolic self-DNA (n=4, each).

Conclusion: These data indicate that cGAS-STING engages redundant sources of cytosolic genomic-DNA in MDS to initiate a Type I interferon response and NLRP3 inflammasome activation. Inhibition of the cGAS-STING axis may represent a novel therapeutic strategy for investigation in MDS.