Temporal uncoupling of the DNA methylome and transcriptional repression during embryogenesis

Abstract

DNA methylation is a tightly regulated epigenetic mark associated with transcriptional repression. Next-generation sequencing of purified methylated DNA obtained from early Xenopus tropicalis embryos demonstrates that this genome is heavily methylated during blastula and gastrula stages. Although DNA methylation is largely absent from transcriptional start sites marked with histone H3 lysine 4 trimethylation (H3K4me3), we find both promoters and gene bodies of active genes robustly methylated. In contrast, DNA methylation is absent in large H3K27me3 domains, indicating that these two repression pathways have different roles. Comparison with chromatin state maps of human ES cells reveals strong conservation of epigenetic makeup and gene regulation between the two systems. Strikingly, genes that are highly expressed in pluripotent cells and in Xenopus embryos but not in differentiated cells exhibit relatively high DNA methylation. Therefore, we tested the repressive potential of DNA methylation using transient and transgenic approaches and show that methylated promoters are robustly transcribed in blastula- and gastrula-stage embryos, but not in oocytes or late embryos. These findings have implications for reprogramming and the epigenetic regulation of pluripotency and differentiation and suggest a relatively open, pliable chromatin state in early embryos followed by reestablished methylation-dependent transcriptional repression during organogenesis and differentiation.

Figure 1.

(A) Genome Browser view of sequenced DNA methylation tracks: late blastula (stage 9) and late gastrula (stage 12.5) tracks of both salt elutions (500 mM and 600 mM/700 mM). The X. tropicalis genome is robustly methylated in both intergenic and intragenic regions during blastula and gastrula stages. (B) Boxplots showing the distribution of the CpG observed/expected ratio (CpG O/E) and the GC content (GC%) of DNA methylation peaks. Peaks of DNA methylation are enriched for CpG dinucleotides, whereas their GC content is similar to the genome average. JGI 4.1 refers to the genome assembly used (Joint Genome Institute genome assembly version 4.1). (C) Bisulfite sequencing of randomly selected MethylCap peaks (see also Supplemental Fig. S2B). (Right) Embryonic stages and MethylCap salt elutions of the DNA methylation tracks; (black boxes) PCR amplicons; (below) bisulfite sequencing profiles of the amplicons. (Black circles) Methylated CpGs; (white circles) unmethylated ones.

Figure 2.

DNA methylation of repetitive elements. (A) Distribution profiles of DNA methylation over three distinct repeat tracks demonstrate an enrichment of repetitive elements for DNA methylation. Microsatellite DNA lacks CpG dinucleotides and therefore is not methylated. However, the RepeatMasker track is depleted for CpG dinucleotides but enriched for DNA methylation, while in the case of the Simple Repeats track DNA, methylation positively correlates with CpG density. (B) Extent of overlap between repeat types and DNA methylation (merged MethylCap peaks). (Top panel) Percentage of methylated genomic copies. (Bottom panel) CpG density of different repeat types. There is an overall correlation of DNA methylation and CpG density. However, some repeats such as Sat1 are highly methylated in spite of their low CpG density.

Figure 3.

DNA methylation and transcriptional regulation. (A) Venn diagrams showing a direct overlap of merged DNA methylation peaks (141,246 regions, all four peak sets combined), the 5′ ends of genes (left: 8625 XTEV_VV gene models; right: 27,916 Joint Genome Institute, JGI, FilteredModels), and genomic CpGis (24,283 Takai-Jones CpGis). DNA methylation is largely absent from 5′ ends of genes. A large portion of the genomic CpGis are methylated (overlap between CpGis and MethylCap peaks). (B) K-means clustering of DNA methylation, histone methylation, and active transcription within CpGis. Clusters 1 and 3 correspond to methylated CpGis with different methylation profiles. Cluster 2 represents exons of genes expressed during gastrulation. Clusters 4 and 5 consist of active CpGi promoters. Cluster 6 corresponds to CpGis that are neither associated with active transcription units nor enriched for either histone modification or RNAPII. (C) Distribution of DNA methylation over CpG and non-CpGi promoters. Both CpG and non-CpG promoters show a dip in DNA methylation around the TSS. To define CpGi and non-CpGi promoter subsets, the TSS regions of XTEV_VV gene models were intersected with Takai-Jones CpGis. (D) Boxplots showing the abundance of DNA methylation within promoter regions of genes with distinct expression intensities. (E) Percentage of promoters (region 1 kb upstream of TSS) overlapping (≥500 bp) with peaks of DNA methylation.

Figure 4.

DNA methylation and H3K4me3/H3K27me3 are mutually exclusive on the majority of genomic locations. (A) Distribution of DNA methylation over boundaries of H3K4me3 peaks. DNA methylation is excluded from the bulk of genomic H3K4me3 sites even though above average DNA methylation can be observed on sequences flanking the H3K4me3 peaks. (B) Genomic example of the exclusion of DNA methylation and H3K4me3. (C) Broad H3K27me3 domains are mostly DNA methylation-free, even though smaller H3K27me3 peaks can be methylated. (D) hoxb cluster as an example of the mutual exclusion of H3K27me3 and DNA methylation. (E) H3K4me3 and H3K27me3 peaks are enriched for CpG dinucleotides but depleted of DNA methylation.

Figure 5.

Conserved patterns of epigenetic regulation in human ES cells and Xenopus embryos. (A) Distribution of RNAPII, H3K4me3, H3K27me3, and DNA methylation (average values of all four tracks) around TSSs corresponding to Xenopus orthologs of two gene groups differing in their expression status among fetal fibroblasts (Fib) and ES cells (H1). The genes with higher expression in ES cells (H1-exp) are also more highly expressed in Xenopus embryos and show increased DNA methylation both upstream of and downstream from the TSS. These orthologs also have a significantly higher DNA methylation in the region spanning 10 kb upstream of the TSS (P-value = 0.031, Wilcoxon rank test), the region 10 kb downstream from the TSS (P-value = 0.0026), and the total region from −10 kb to +10 kb relative to the TSS (P-value = 0.0023). (B) Hierarchical clustering performed on Xenopus orthologs of H1-exp and Fib-exp genes shows an inverse relationship between DNA methylation and H3K27me3. It also shows a high level of DNA methylation on promoters of robustly expressed H1-exp orthologs.

Figure 6.

Methylated promoters in early Xenopus embryos are able to initiate transcription. (A) Unlike Xenopus oocytes, gastrula embryos fail to repress transcription from the injected hsp70 methylated (SssI) promoter. Levels of injected plasmid DNA recovered from embryos were similar for all samples. (B) Effects of DNA methylation on stably integrated transgenes. A fully methylated CMV-luciferase transgene is robustly expressed compared to its unmethylated counterpart during blastula and gastrula stages. Strong DNA methylation-dependent repression observed in the oocyte is restored gradually at neurula and tailbud stages. The results represent three independent experiments and were normalized for the number of transgene insertions. (C) When targeted to the injected hsp70 promoter by the heterologous Gal4 DNA-binding domain in oocytes, the transcription repression domain of MECP2 shuts down reporter construct transcription. Such an effect is not observed in gastrula embryos where the Gal4-TRD fusion is expressed at equivalent levels (see Supplemental Fig. S15). Error bars represent the SEM of three experiments.