Extensive epigenetic reprogramming in human somatic tissues between fetus and adult

Abstract

Background: Development of human tissue is influenced by a combination of intrinsic biological signals and extrinsic environmental stimuli, both of which are mediated by epigenetic regulation, including DNA methylation. However, little is currently known of the normal acquisition or loss of epigenetic markers during fetal and postnatal development.

Results: The DNA methylation status of over 1000 CpGs located in the regulatory regions of nearly 800 genes was evaluated in five somatic tissues (brain, kidney, lung, muscle and skin) from eight normal second-trimester fetuses. Tissue-specific differentially methylated regions (tDMRs) were identified in 195 such loci. However, comparison with corresponding data from trisomic fetuses (five trisomy 21 and four trisomy 18) revealed relatively few DNA methylation differences associated with trisomy, despite such conditions having a profound effect on development. Of interest, only 17% of the identified fetal tDMRs were found to maintain this same tissue-specific DNA methylation in adult tissues. Furthermore, 10% of the sites analyzed, including sites associated with imprinted genes, had a DNA methylation difference of >40% between fetus and adult. This plasticity of DNA methylation over development was further confirmed by comparison with similar data from embryonic stem cells, with the most altered methylation levels being linked to domains with bivalent histone modifications.

Conclusions: Most fetal tDMRs seem to reflect transient DNA methylation changes during development rather than permanent epigenetic signatures. The extensive tissue-specific and developmental-stage specific nature of DNA methylation will need to be elucidated to identify abnormal patterns of DNA methylation associated with abnormal development or disease.

Figure 1

Unsupervised clustering of fetal tissues demonstrates that each tissue has a distinct DNA methylation profile. Sample names are shown with labelling of the corresponding tissue types. Tissue samples were clustered by hierarchical clustering of β values based on the formula 1-r (Beadstudio software; Illumina), where r is the correlation coefficient between samples. Specific tissue types clustered together, with strong correlation between samples derived from the same tissue. All tissues had distinct clustering from the other groups, except for one muscle sample from T21 (FT1_t21_muscle), which clustered with the skin sample from the same fetus.

Figure 2

Heat-map of 98 loci showing hypermethylated or hypomethylated tissue-specific differentially methylated regions (tDMRs) in particular tissues. Probes and sample names are shown and with hierarchical clustering of β values based on the formula 1-r (Beadstudio software; Illumina), where r is the correlation coefficient between samples. A β value of 0 (bright green) represents an unmethylated locus, and a value of 1 (bright red) represents a methylated locus. Hypermethylated and hypomethylated loci were defined as those having an average β value in that tissue of >0.2 above or below the overall mean for all tissues. Fetal brain had the highest number of tDMRs, with 30 hypermethylated and 23 hypomethylated loci.

Figure 3

Venn diagram having the number of age-dependent methylated loci/genes between brain, kidney and lung. (A) Of 89 age-dependent methylated loci (CpG sites), only four loci were common to different tissues. (B) Similarly, of the 75 associated genes in the 89 age-dependent methylated loci, only four genes were common to different tissues. Most age-dependent differentially methylated regions (aDMRs) were specific for one tissue, with 24 such loci identified in brain, 11 in kidney and 25 in lung. DMGs = differentially methylated genes. ↑ = Hypermethylated; ↓ = Hypomethylated.

Figure 4

Lack of conservation in tissue-specific differentially methylated loci between fetus and adult. Methylation level (β value) of (A) MEST_P4 of the MEST gene, (B) CDH17_E31 of the CDH17 gene and (C) CRK_P721 of the CRK gene in fetal and adult tissues. Each bar represents a different sample. Fetal tissue-specific indicator loci were not indicative of tissue origin in adult tissues.

Figure 5

Characteristics of differentially methylated regions. Characteristics of (A) tissue-specific differentially methylated regions (tDMRs) and (B) age-dependent differentially methylated regions (aDMRs). The characteristics of Polycomb complex binding targets and histone markers were based on the previous report on embryonic stem cells whereas the information on CG islands (CGI) was available from Illumina. *P < 0.05, **P < 0.005, ***P < 0.0005. 'Percentage of loci' refers to the percentage of loci in the microarray that contains the specified features.

Figure 6

Dynamic changes of DNA methylation. (A) RAB32_P493 had hypermethylation in fetal brain, but hypomethylation in embryonic stem (ES) cells and adult brain. (B) HPN_P823 had hypermethylation in ES cells and adult kidney, but hypomethylation in fetal kidney.

Figure 7

Summary of data and findings in this study. Number of probes and tissues is shown for each comparison between sample groups. A brief description of main findings is included.