Epigenetic assimilation in the aging human brain

Abstract

Background: Epigenetic drift progressively increases variation in DNA modification profiles of aging cells, but the finale of such divergence remains elusive. In this study, we explored the dynamics of DNA modification and transcription in the later stages of human life.

Results: We find that brain tissues of older individuals (>75 years) become more similar to each other, both epigenetically and transcriptionally, compared with younger individuals. Inter-individual epigenetic assimilation is concurrent with increasing similarity between the cerebral cortex and the cerebellum, which points to potential brain cell dedifferentiation. DNA modification analysis of twins affected with Alzheimer's disease reveals a potential for accelerated epigenetic assimilation in neurodegenerative disease. We also observe loss of boundaries and merging of neighboring DNA modification and transcriptomic domains over time.

Conclusions: Age-dependent epigenetic divergence, paradoxically, changes to convergence in the later stages of life. The newly described phenomena of epigenetic assimilation and tissue dedifferentiation may help us better understand the molecular mechanisms of aging and the origins of diseases for which age is a risk factor.

Keywords: Aging; Alzheimer’s disease; DNA methylation; Dedifferentiation; Epigenetic drift; Epigenetics; Genomic organization; Transcriptome.

Fig. 1

ICC analysis of DNA modification and gene transcription within the cerebral cortex and the cerebellum. The histograms represent the densities of the permuted mean ICC coefficients from samples of all ages and the red dashed lines show the mean ICC in the older individuals (>75 years). a Mean ICC of DNA modification in the cerebral cortex of older individuals (permuted p = 7 × 10⁻⁶). b Mean ICC of DNA modification in the cerebellum of older individuals (permuted p = 9 × 10⁻⁴). c Mean ICC of the transcriptome in the cerebral cortex of older individuals (permuted p = 5 × 10⁻⁶). d Mean ICC of the transcriptome in the cerebellum of older individuals (permuted p < 10⁻⁶)

Fig. 2

Loss of tissue-specific DNA modification and gene transcription patterns in the aging brain. The histograms represent the densities of the permuted mean ICC coefficients between two different brain regions (cerebral cortex and cerebellum) from samples of all ages: a DNA modification (permuted p < 10⁻⁶); b transcriptome (permuted p < 10⁻⁶). The red dashed lines show the mean cortex–cerebellum ICCs in the older individuals (>75 years)

Fig. 3

Unsupervised hierarchical clustering of DNA modification in the brains of EAO and LAO AD twins. a The red boxes indicate clades with higher than 80 % bootstrapping probability. Clustering, using the top 5 % of the most differentially modified loci, showed that cerebellum (CB) and EAO cerebral cortex form a single clade 95 % of the time while the cortex from the LAO co-twins are in a separate clade 95 % of the time. b In the top 82 AD onset-associated loci, the cerebral cortex of EAO twins and the cerebellum clustered into a single clade 98 % of the time, while LAO co-twins separated into a different clade 98 % of the time

Fig. 4

Examples of expanding DNA modification and transcriptomic domains. The contour plot represents the correlation coefficients between the nearest neighboring probes, where high correlation (r = 1) is represented in white, no correlation (r = 0) in light green, and anti-correlation (r = −1) in dark green. a The plot shows correlations between 11 DNA modification probes representing a ~4 Mb region on chr14: 34,414,883–38,642,244. Three distinct domains with high inter-probe correlation were detected in the young individuals (1–20 years), while the boundaries of these domains merged in the old individuals (76–96 years). The middle-aged individuals (40–63 years) showed an intermediate pattern. b The plot shows correlations between 11 transcripts representing a ~1 Mb region on chr2:190,753,842–191,603,958. Like the DNA modification data, the young individuals show distinct transcriptomic boundaries that diminished with age