Epigenetic STING silencing is developmentally conserved in gliomas and can be rescued by methyltransferase inhibition

The stimulator of interferon genes (STING) is a critical signaling transducer of cytosolic DNA sensing, eliciting IRF3- and NFκB-dependent transcription of type I IFNs and pro-inflammatory cytokines. The importance of tumor-localized STING activation in priming endogenous antitumor immunity is well established. A pan-cancer analysis showed that STING signaling is frequently disrupted through loss-of-function mutations or hypermethylation of the STING or cGAS promoters (Konno et al., 2018Konno H. Yamauchi S. Berglund A. Putney R.M. Mulé J.J. Barber G.N. Suppression of STING signaling through epigenetic silencing and missense mutation impedes DNA damage mediated cytokine production.Oncogene. 2018; 37: 2037-2051https://doi.org/10.1038/s41388-017-0120-0Crossref PubMed Scopus (111) Google Scholar). Here we investigate STING signaling in glioblastoma (GBM) and normal brain and find that STING expression is suppressed in both normal brain and glioma cells, but not in tumor-associated immune cells or stroma. We identify a CpG site that is methylated in normal brain, gliomas, and other neuroectoderm-derived cancers, but not in most extracranial cancers. We demonstrate that STING expression is rescued by decitabine, a DNA methyltransferase inhibitor (DNMTi) that is used to treat myelodysplastic syndrome and acute myeloid leukemia. Our work raises the potential of DNMTis to reconstitute STING signaling in GBM tumor cells. We examined whether the cGAS/STING pathway is disrupted in GBM, as has been demonstrated in other cancers (Xia et al., 2016aXia T. Konno H. Ahn J. Barber G.N. Deregulation of STING signaling in Colorectal Carcinoma Constrains DNA damage Responses and Correlates with tumorigenesis.Cell Rep. 2016; 14: 282-297https://doi.org/10.1016/j.celrep.2015.12.029Abstract Full Text Full Text PDF PubMed Scopus (323) Google Scholar, Xia et al., 2016bXia T. Konno H. Barber G.N. Recurrent loss of STING signaling in melanoma Correlates with Susceptibility to Viral Oncolysis.Cancer Res. 2016; 76: 6747-6759https://doi.org/10.1158/0008-5472.can-16-1404Crossref PubMed Scopus (0) Google Scholar). We recently reported that the STING agonist 2′3′-cGAMP (cGAMP) induced type I IFNs and CXCL10 in ex vivo GBM tissue slice cultures that retained the native tumor micro-environment (Brown et al., 2021Brown M.C. Mosaheb M.M. Mohme M. McKay Z.P. Holl E.K. Kastan J.P. Yang Y. Beasley G.M. Hwang E.S. Ashley D.M. et al.Viral infection of cells within the tumor microenvironment mediates antitumor immunotherapy via selective TBK1-IRF3 signaling.Nat. Commun. 2021; 12: 1858https://doi.org/10.1038/s41467-021-22088-1Crossref PubMed Scopus (23) Google Scholar). Here we separated GBM cell suspensions by surface CD14 expression, a specific marker of tumor-associated myeloid cells (TAMCs). cGAMP-mediated CXCL10 and TNF induction occurred only in CD14+-containing glioma cell suspensions, and the interferon-stimulated gene IFIT1 was induced selectively in TAMC after ex vivo GBM slice culture treatment (Figure S1A), indicating that STING signaling is absent in the neoplastic, but not myeloid, compartment of human GBM samples. We investigated baseline STING expression using single-cell RNA-sequencing profiles in neoplastic, immune, and stromal cells from newly diagnosed GBM patient samples (Figure S1B). Single-cell clustering analysis identified three major clusters: neoplastic (SOX2), immune (PTPRC), and vascular smooth muscle cells (COL3A1). Detectable STING mRNA was observed only in immune and stromal cells, but not in neoplastic cells. Multiplex immunofluorescence analysis of GBM tumor sections was performed to colocalize STING expression with CD3 (T-cells), CD64 (macrophage/microglia), CD31 (endothelial cells), and vimentin (mesenchyme) (Figure S1C). STING protein expression was restricted to myeloid and endothelial/stromal cells and was absent in GBM tumor cells and T cells (Figure S1C). Despite absent STING expression and signaling in GBM cells, STING gene mutations are rare, with only one out of 378 samples in TCGA GBM PanCancer Atlas exhibiting a missense mutation. We hypothesized that epigenetic silencing of STING could account for its reduced expression. We performed Illumina methylation arrays on 64 GBM patient samples, analyzed the methylation values across the STING gene, and compared them with those from normal fetal and adult brain datasets (Gasparoni et al., 2018Gasparoni G. Bultmann S. Lutsik P. Kraus T.F.J. Sordon S. Vlcek J. Dietinger V. Steinmaurer M. Haider M. Mulholland C.B. et al.DNA methylation analysis on purified neurons and glia dissects age and Alzheimer’s disease-specific changes in the human cortex.Epigenetics Chromatin. 2018; 11: 41https://doi.org/10.1186/s13072-018-0211-3Crossref PubMed Scopus (115) Google Scholar). The CpG site cg16983159, located in the STING promoter (Wang et al., 2016Wang Y.-Y. Jin R. Zhou G.-P. Xu H.-G. Mechanisms of transcriptional activation of the stimulator of interferon genes by transcription factors CREB and c-Myc.Oncotarget. 2016; 7: 85049-85057https://doi.org/10.18632/oncotarget.13183Crossref PubMed Scopus (11) Google Scholar), is consistently hypermethylated (Figure S1D). TCGA GBM Illumina methylation array data are limited to seven probes across the STING gene. Each of these probes demonstrates negative correlations with mRNA expression, with the strongest correlation being with probe cg16983159 (r = −0.76, Figure S1D). These data suggest that cg16983159 methylation suppresses STING mRNA expression. The presence of cg16983159 hypermethylation in normal adult and fetal brains suggests that STING promoter hypermethylation is not a consequence of gliomagenesis, but rather that GBM forms in the context of a normally STING-silent environment that is conserved through brain development and subsequent tumorigenesis. Since the Illumina methylation array probes only 11 CpG sites distributed across the STING gene, we sequenced two bisulfite-treated glioma cell lines to evaluate the methylation state of 13 CpG sites in an ∼400 nt window centered on cg16983159. We chose cell lines that, like patient-derived samples, exhibit baseline high STING cg16981359 methylation and low STING expression. Moderate (LN229, 50%–75%) to high (U138, 75%–100%) methylation was seen at most CpG sites in this region. We hypothesized that STING expression and signaling are suppressed by methylation in the vicinity of cg16983159, which could be reversed by DNMTis. We treated cells with the DNMTi decitabine and found that cg16983159 and upstream CpG sites are demethylated, while downstream sites are not (Figure S1D). We additionally specifically measured STING cg16983159 methylation, RNA and protein expression. Decitabine treatment suppressed DNMT1, decreased cg16983159 methylation, and increased STING mRNA and protein expression (Figure S1E). Moreover, cell lines previously unresponsive to the STING agonist cGAMP were able to activate innate immune and interferon-stimulated genes (e.g., p-IRF3, IFIT1, pSTAT1, and ISG15) after decitabine treatment. Finally, we queried whether STING hypermethylation occurs in other tumor types. We plotted average cg16983159 methylation values for TCGA solid cancers versus non-cancer tissue (Figure S1F). Most cancers arising outside the nervous system displayed cg16983159 hypomethylation (beta <0.5), as did their normal tissues of origin. Conversely, the few extracranial cancers that exhibited cg16983159 hypermethylation also demonstrated cg16983159 hypermethylation in their tissues of origin. Analysis of methylation datasets for other primary brain tumors from the Gene Expression Omnibus revealed cg16983159 hypermethylation, with two exceptions: primary CNS lymphoma and meningioma. Similar to primary brain cancers, neuroectoderm-derived tumors are also methylated at the STING promoter cg16983159 site (Figure S1F). An exception to this finding was melanoma (TCGA codes UVM and SKCM), which is derived from neuroectoderm cells but is unmethylated at this site. GBM carries frequent extrachromosomal and cytoplasmic DNA (Kim et al., 2020Kim H. Nguyen N.-P. Turner K. Wu S. Gujar A.D. Luebeck J. Liu J. Deshpande V. Rajkumar U. Namburi S. et al.Extrachromosomal DNA is associated with oncogene amplification and poor outcome across multiple cancers.Nat. Genet. 2020; 52: 891-897https://doi.org/10.1038/s41588-020-0678-2Crossref PubMed Scopus (123) Google Scholar) that should invoke a cGAS-STING signal yet has a “cold” micro-environment and is notoriously resistant to immunotherapy. Our work suggests that hypermethylation in the STING promoter mediates STING silencing in GBM and may contribute to its intrinsic immunosuppression. The utility of DNMTi to rescue STING signaling and induce sensitivity to STING agonists has been demonstrated in KRAS-LKB1-mutant lung cancer, where LKB1 loss results in hyperactivation of DNMT1 and STING promoter methylation (Kitajima et al., 2019Kitajima S. Ivanova E. Guo S. Yoshida R. Campisi M. Sundararaman S.K. Tange S. Mitsuishi Y. Thai T.C. Masuda S. et al.Suppression of STING associated with LKB1 loss in KRAS-Driven lung cancer.Cancer Discov. 2019; 9: 34-45https://doi.org/10.1158/2159-8290.cd-18-0689Crossref PubMed Scopus (0) Google Scholar). Here, we highlight STING epigenetic silencing as characteristic of both the normal brain and primary brain tumors. We propose that reconstituting endogenous STING signaling using DNMTi may be a promising approach for inducing immunotherapy sensitivity in GBM. Support was received from the Zachem and Lu families in support of Brain Tumor Research. Conceptualization, D.M.A. and J.T.L.; Methodology, V.C., J.T.L., M.C.B., M.L.B.; Investigation, J.T.L., V.C., M.L.B., M.C.B., A.B., K.S., R.F., A.M.M., J.H., S.D., A.B., N.C.W.; Writing – Original Draft, J.T.L..; Writing – Review & Editing, J.T.L., D.M.A., V.C., M.C.B, M.S.W, M.K., A.H., M.G., S.H., and S.Y.W.; Funding Acquisition, D.M.A.; Resources, D.M.A, D.D.B, S.G., S.K.N. and M.G.; Supervision, D.M.A. The authors declare no competing interests. Download .pdf (3.55 MB) Help with pdf files Document S1. Figure S1, supplemental methods, and supplemental references