GENOME-WIDE METHYLATION PATTERN OF REPETITIVE ELEMENTS IN THE HUMAN GENOME
We investigated how individual occurrences of repetitive elements were methylated in the human genome, as summarized in Fig. 7.2A. Fractions of hypomethylated repeat occurrences vary considerably among different classes of repetitive elements, from ~1% for L1 and Alu to ~50% for MIR and >70% for simple repeats and low-complexity regions. To validate our prediction regarding the repeat occurrences, we selected 21 regions for bisulfite Sanger sequencing, designed primers for nested PCR, and could
Figure 7.2 Epigenetic landscape of repetitive elements in the human genome. (A) Summary of methylation status on repetitive elements. Because some occurrences of repetitive elements contain no or very few CpG sites, we only consider those occurrences with at least 10 CpGs to exclude other less informative cases (a). First, we checked whether single-molecule real-time (SMRT) reads could address the repetitive regions in a useful manner for methylation analysis. Specifically, we considered a repeat occurrence to be covered by uniquely mapped SMRT reads if the interpulse duration ratio was available on >50% of CpGs (b), and found that >96% were covered for every repeat type. To draw robust conclusions, we further applied a stringent quality control process to each repeat occurrence such that the read coverage was >5 (c). Although this step reduced the number of repeat occurrences under consideration by 3-18%, this reduction could be mitigated simply by producing more data. Finally, we treated an occurrence as hypomethylated if >50% of CpGs were predicted as hypomethylated (d). (B and C) Distribution of CpG density (B) and sequence divergence from the representative in each repeat class (C) for methylated (cyan) and hypomethylated (pink) repeat occurrences. The asterisks indicate statistical significance (p < 1%) determined by the U test. (D-F) Genome-wide distribution of hypomethylated repetitive elements. The ratio of hypomethylated repeat occurrences to all occurrences in each 5-Mb bin is indicated by color shadings.
amplify six regions, indicating the difficulty in observing DNA methylation of repetitive elements by using traditional bisulfite Sanger sequencing. In the six amplified regions, we confirmed the consistency between our prediction and the methylation state observed by bisulfite Sanger sequencing.
We then examined the features for characterizing the differences between hypermethylated and hypomethylated repetitive elements. First, CpG density was significantly higher in the hypomethylated occurrences in almost all classes of repetitive elements (p < 1%, Fig. 7.2B). This observation was consistent with the known association between CpG-rich regions and hypomethylation because hypermethylation leads to depletion of CpG sites through deamination (Cooper & Krawczak, 1989). Second, sequence divergence from the representative in each repeat class also showed a correlation with methylation status (Fig. 7.2C). For most classes, with the apparent exception of simple repeats, low-complexity regions, and MIR elements, hypomethylated occurrences were significantly more divergent than were hypermethylated occurrences (p< 1%, Fig. 7.2C), presumably because younger copies of a repeat element are less divergent and are likely to be targets of DNA methylation.
Next, we examined whether the hypomethylated repeat occurrences were distributed uniformly or nonuniformly throughout the entire genome. We selected three major classes of repetitive elements for this analysis: long interspersed nuclear element (LINE), Alu, and long terminal repeat (LTR). We calculated the ratios of hypomethylated copies to all repetitive elements in individual nonoverlapping bins 5 Mb in size (Fig. 7.2D-F). The nonrandom distribution patterns were more evident for LINE and LTR than for Alu. For example, we found hypomethylated LINEs to be enriched in the p-arm of chromosome 1 and in chromosomes 17 and 19. There were hypomethylation “hot spots” of LTR elements, for example, in chromosomes 6 and 9. It is intriguing that some of these hypomethylation hot spots, such as those in the p-arms of chromosomes 6 and Y, seem to be shared among different classes of repetitive elements.