DNA methylation-based epigenetic clock construction has generally relied on bulk datasets representing findings from thousands or millions of cells, and now, a new single-cell epigenetic clock age predictor known as scEpiAge may help to resolve the impact of heterogeneity. Can improved resolution offer more profound insight into the aging process?
A team led by Marc Jan Bonder, Stephen J. Clark, Wolf Reik, and Ferdinand von Meyenn understood that single-cell genome-wide DNA methylation datasets differed from bulk datasets in that they generate binary results (methylated or not), sparse results, and random genomic coverage, which represent challenges to single-cell epigenetic clock construction. With this in mind, they generated a single-cell DNA methylation and transcriptome dataset from around 800 nucleated peripheral blood cells of mice spanning a broad range of ages (∼10-100 weeks) using single-cell methylation and transcriptome sequencing (scM&T-seq) and constructed an accurate single-cell DNA methylation age predictor– scEpiAge.
Let’s hear more from Bonder and Colleagues on scEpiAge brings single-cell resolution to aging research:
- DNA methylation levels increase with age at CpG islands but decrease at repeat regions, with results indicating the potential masking of accumulated cell type-specific changes by cell heterogeneity
- As opposed to traditional epigenetic clocks, scEpiAge reverses the standard elastic net regression setup, predicts DNA methylation from age, and does not require the coverage of the same set of CpG sites in each cell
- scEpiAge also leverages redundancy and models a range of possible relationships between age and DNA methylation
- scEpiAge supports highly accurate epigenetic age prediction in single cells while also applying well to (sparse) bulk bisulfite data, low-depth long-read sequencing methods, and large-scale screening approaches
- The best scEpiAge model displays a cross-validation median absolute error of ∼8 weeks
- Comparing predicted ages with known chronological ages reveals that bulk datasets and the average age of all cells of a given age group agreed with the expected (chronological) age (although variance increases with age)
- The analysis also revealed cell type-dependent variation in the epigenetic age of mouse blood cells of a given chronological age, suggesting that T- and B-cell aging differs
Overall, scEpiAge can cut through heterogeneity and aid the exploration of individual cell aging within multicellular organisms – do cells age uniformly and lose functional capacity homogeneously, or does the relative proportion of young and old cells determine tissue age? Furthermore, we can now start to address questions related to cell type-specific age-related changes and better understand rejuvenatory strategies at a single-cell resolution.
For more on how scEpiAge brings single-cell resolution to epigenetic clock heterogeneity, see Nature Communications, August 2024.