Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: A comprehensive analysis of 195 DNA methylomes reveals shared and cell-specific features of partially methylated domains

Fig. 3

Heterochromatization accompanied by DNA methylation erosion at PMDs in cancers. a A snapshot of 14 Mb of chr3 showing the relevant epigenetic marks. Top: distinct DNA methylation tracks and the MethylSeekR segmentation of liver tissue, isolated hepatocytes (PHH), liver cancer tissue, HepaRG, and HepG2 cell lines, respectively. PMDs of primary cells and normal and cancerous tissues are extensively and selectively less methylated in cancer cell lines (largely converted into unmethylated regions). Middle: histone marks H3K27me3, H3K9me3, and H3K36me3 in the same samples. Bottom: ChromHMM segmentation based on these three histone modifications in addition to H3K4me3, H3K4me1, H3K27ac, and Input (see details in the “Methods” section). bK-means clustering (k = 6) based on the averaged methylation in 10 Kb bins. Cluster 1 represents the most (almost fully) methylated bins across all samples, while the other clusters are ordered according to the progressive erosion of methylation in PMDs. Bar plots (left) beside the heatmap show the percentage of the annotated bins as HMD, PMDs, and UMR for each sample in each cluster. c Progressive change of DNA methylation in PMDs across cancer cell lines. The top of the figure shows classified and grouped PMDs (three classes) based on the average PMD methylation levels in PHH and their corresponding overall levels in HepaRG and HepG2, respectively. Note the intermediate status of HepaRG, e.g., with a higher similarity to PHH in class_I (most highly methylated), an intermediate status in class_II and a higher similarity to HepG2 in class_III (lowest methylation level). The bottom shows the PMD wide changes in heterochromatic marks across the clusters defined by DNA methylation. The inverse correlation to DNA methylation is most obvious for HepaRG (class_I and class_III)

Back to article page