# DNA Methylation Signatures of Cellular Senescence Are Not Reversed by Senolytic Treatment

## TL;DR

This study shows that DNA methylation changes linked to cellular senescence are not reversed by senolytic treatments, suggesting limitations in using methylation clocks to track senescence or aging interventions.

## Contribution

The study identifies a small subset of DNA methylation sites linked to senescence, age, and mortality, and reveals that senolytic treatments do not reverse these methylation signatures.

## Key findings

- Senescence-related DNA methylation signatures overlap minimally with age and mortality-related CpGs.
- Senolytic treatment does not reverse or slow down methylation clocks trained to detect senescence or aging.
- The results challenge the assumption that geroscience interventions reduce aging biomarkers.

## Abstract

Epigenetic clocks are commonly used aging biomarkers based on DNA methylation that predict long‐term morbidity and mortality risk. Increased cellular senescence with age is also posited to contribute to age‐related disease and mortality. However, prior studies have found that existing epigenetic clocks show inconsistent associations with cellular senescence and no reductions after senolytic treatment. We hypothesize this reflects that senescence‐related CpGs are a small proportion of age‐related CpGs, and that an epigenetic clock focused on a core senescence signal conserved across different cell types and different senescence inducers would be a better tool for monitoring senescence and senolytic treatment compared to traditional epigenetic clocks. In our study, we find that senescence, age and mortality risk intersect at a small subset of the DNA methylome (9363 CpGs out of 396,333 analyzed; 2.4%). Utilizing these CpGs, we generated three different epigenetic clocks trained to predict in vitro senescence, age, and mortality, respectively. Surprisingly, all three of these predictors stayed the same or even accelerated after senolytic treatment in both in vivo and in vitro data. Our findings not only call into question whether cellular senescence can be captured by DNA methylation but also challenge the assumption that aging biomarkers decrease after geroscience interventions.

We found very little overlap between CpGs that were correlated with in vitro senescence, chronological age, and mortality. While we were able to train epigenetic clocks with CpGs that accelerated with cellular senescence, these clocks did not decelerate with Senolytic treatment.