A new study considers epigenetic aging from a single-cell perspective and highlights its having both co-regulated and stochastic components.
The process of aging is associated with several alterations at the cellular, subcellular and nuclear levels, including telomere attrition, protein misfolding, epigenetic alterations, mitochondrial dysfunction, cellular senescence, loss of proteostasis and others. Among them, epigenetic alterations are significant and involve changes in DNA methylation, ubiquitination, phosphorylation and histone acetylation. Epigenetic alterations can occur due to both intrinsic and extrinsic factors and lead to age-related pathologies [1], and this led to the development of epigenetic clocks which can predict the chronological age of tissues, cells and organisms.
Longevity.Technology: All epigenetic clocks are constructed on DNA methylation (DNAm) levels but it is challenging when there is mixed DNAm signal from single cells. Several factors including errors in DNAm maintenance during cell division, changes in cell type composition and clonal expansion are considered to contribute to tissue DNAm age-related changes. However, pure clock-based approaches that can accurately explain epigenetic aging are still lacking.
A new study published in Nature Aging used single-cell DNA methylation (scDNAm) aging and embryonic development data in a mouse model to determine the mechanisms behind epigenetic aging. The researchers categorized aging into stochastic (lacking consistent DNAm pattern across cells and organisms)and co-regulated (coherent DNAm patterns across different cells and animals) for the study.
The researchers divided the DNAm behavior over the life course of mice as per the typical aging dynamics into development, functional aging and multimorbidity – categories that also apply to humans. The developmental stage was observed to be tightly regulated and programmed. Following after, the functional aging stage was characterized by a gradual decrease in function without significant comorbidities. The final stage was multimorbidity, where significant functional decline affected the overall survival of the organism.
The study analyzed blood bulk DNAm changes of male mice and identified 268,044 CpG sites (dinucleotides that are often methylated) that were associated with age (later reduced to 16,889 CpG sites after Bonferroni correction) [2]. However, the overall change in DNAm for each age-elated CpG site was reported to be small during functional aging, where most CpG sites showed less than 10% changes throughout the lifespan.
The researchers then classified the aging trajectories based on their initial methylation levels at the end of development. Dynamics 1, 4 and 7 represented CpG sites whose methylation remained constant with age. Dynamics 2 and 3 represented those with a gain of methylation, and dynamics 5 and 6 represented loss of methylation with age [2]. Results from genomic enrichment analyses indicated that clustering of the dynamics were based on the direction of DNAm levels change during functional aging rather than the initial methylation level. The results also reported that the dynamics associated with higher or loss of methylation levels were found in the non-functional genomic regions. Such results suggest that the features of DNAm aging changes indicate a stochastic process.
Results from the stochastic single-cell model showed that the model predictions fit the observed behavior of clock CpG aging trajectories. Researchers then used two single-cell DNA methylation (scDNAm) datasets (mouse embryos before and during gastrulation and another from aging muscle stem cell) and reported out of 502 CpGs, 40 showed co-regulated and 482 showed stochastic aging. Analysis of embryonic DNAm data for testing co-regulation scenario reported CpGs to change in a program-like manner during development which is different from what is seen during functional aging and reported 191 co-regulated CpGs and 5809 stochastic CpGs [2].
The results also found co-regulated CpG clusters to be associated with functional genomic regions and active transcriptional processes while this was not the case for stochastic CpG clusters. Co-regulated clusters were also found to be more evolutionary conserved than stochastic regions. Fewer splicing age-associated splicing events and association with specific transcription factors were observed in co-regulated clusters as compared to stochastic clusters.
A Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis reported co-regulated genes to be enriched for muscle-specific pathways and stochastic for cell morphogenesis and neuron differentiation pathways. The results also reported that genes associated with co-regulated CpG clusters had a higher level of transcriptomic coordination as compared to those associated with stochastic clusters. Finally, the co-regulated genes were observed to maintain a stable expression profile through aging, consistent with epigenetic co-regulation [2].
The study highlights that aging scDNAm changes involve co-regulated changes and a major stochastic component. The stochastic component may be an indicator of the cumulative damage arising due to the environment of organisms, while the co-regulated component helps to detect the effects of target-specific antiaging interventions. As the authors put it, these “analyses increase our understanding of the basis of epigenetic clocks and highlight potential opportunities for targeting aging and evaluating longevity interventions [2].”
[1] https://www.sciencedirect.com/science/article/abs/pii/S1568163722001854
[2] https://www.nature.com/articles/s43587-024-00616-0
