A DNA Methylation Fingerprint of Cellular Senescence

Recent years have witnessed an influx of therapies aimed at targeting the biological aging process. While many hold promise for potentially delaying the simultaneous onset of multiple chronic conditions, a major hurdle in developing interventions to slow human aging is the lack of reliable and valid endpoints from which to evaluate potential candidates. The aim of this study was to develop a biomarker that could serve as a potential endophenotype in evaluation of one of the most promising and exciting aging therapeutics being developed--senolytics. Senolytics are compounds which selectively clear senescent cells by targeting anti-apoptotic pathways. Using DNAm data from fibroblasts and mesenchymal stromal cells (MSC) in culture, we developed a predictor of cellular senescence that related to three distinct senescence inducers (replicative, oncogene induced (OIS), and ionizing radiation (IR)). Our measure, termed DNAmSen, showed expected classification of validation data from embryonic stem cells (ESC), induced pluripotent stem cells (iPSC), and near senescent cells in culture. Further, using bulk data from whole blood, peripheral tissue samples, and postmortem tissue, we observed robust correlations between age and DNAmSen. We also observed age adjusted associations between DNAmSen and idiopathic pulmonary fibrosis (IPS), COPD, lung cancer, and Werner syndrome compared to controls. When characterizing the 88 CpGs in the DNAmSen measure, we observed a enrichment for those located in enhancer regions and regulatory regions marked by DNase I hypersensitive sites (DHSs). Interestingly most of the CpGs were observed to be moderately hypomethyled in early passage cells, yet exhibited further hypomethylation upon induction of cellular senescence, suggesting senescence is accompanied by programmatic alterations to the epigenome, rather than entropic drift.