DNA methylation is a form of epigenetic alteration to DNA, influencing the rate at which specific proteins are produced from their blueprint genes. The pattern of methylation shifts constantly in response to circumstances, but since the damage that causes aging is much the same in everyone, it is possible to identify patterns that correlate well with age. Here researchers investigate DNA methylation changes in aging skin:
Epigenetic changes represent an attractive mechanism for understanding the phenotypic changes associated with human aging. Age-related changes in DNA methylation at the genome scale have been termed 'epigenetic drift', but the defining features of this phenomenon remain to be established. Human epidermis represents an excellent model for understanding age-related epigenetic changes because of its substantial cell-type homogeneity and its well-known age-related phenotype. We have now generated and analyzed the currently largest set of human epidermis methylomes (N = 108) using array-based profiling of 450,000 methylation marks in various age groups. Data analysis confirmed that age-related methylation differences are locally restricted and characterized by relatively small effect sizes. Nevertheless, methylation data could be used to predict the chronological age of sample donors with high accuracy.
In agreement with our previous studies that were carried out either at lower resolution or with smaller sample sizes, we find that age-related methylation changes appear rather moderate and do not compromise the overall integrity of the epidermis methylome. Nevertheless, we identified a variety of specific age-related methylation changes. In contrast to prior work by others, where whole-blood samples and different tissues were used to develop a predictive signature of biological age, we achieved significantly improved prediction accuracy by training the prediction algorithm on epidermis samples. In agreement with previous analyses, we observed a significant age-related hypermethylation of CpG island-associated probes. Interestingly, this effect was strongly enriched during two specific age windows, at 40-45 and 50-55 years. Considering that our samples were exclusively derived from female volunteers, it seems reasonable to link the latter window to menopause, which is also known to distinctly accelerate skin aging. The high temporal and spatial specificity of these methylation changes suggests that defined signaling pathways, such as estrogen signaling, may be involved in their establishment.
Our results also describe an age-related erosion of DNA methylation patterns that is characterized by two distinct features: (i) While the topology of young methylomes is characterized by sharply demarcated regions of (almost) complete and (almost) absent methylation, old methylomes appeared to be less clearly defined, which is reflected by the significantly reduced variance and spatial correlation within methylomes. (ii) While young methylomes are highly similar among each other, old methylomes appeared to be substantially more heterogeneous. Hence, while methylation patterning within an individual becomes more homogeneous with age, the differences between individuals increase. The effects of age-related methylation changes on gene expression patterns have been analyzed in several previous studies. Somewhat surprisingly, however, no global correlations could be established and methylation-related expression changes generally appeared very limited. These findings support the notion that age-related methylation changes function to stabilize pre-existing gene expression patterns. Alternatively, age-related gene expression changes might also be too subtle to achieve statistical significance in classical differential expression analyses. The analysis of gene co-expression networks provides an opportunity to analyze transcriptional deregulation at a higher level of complexity, and our findings demonstrate a reduced connectivity of gene expression in old samples. These results are in agreement with earlier findings in aging mice and suggest that the age-related erosion of methylation patterns is accompanied by a reduced fine-tuning in the transcriptional circuitry, possibly through methylation-dependent changes in transcription factor binding.