Dynamic Changes of Genome-Wide DNA Methylation during Soybean Seed Development

Abstract

Seed development is programmed by expression of many genes in plants. Seed maturation is an important developmental process to soybean seed quality and yield. DNA methylation is a major epigenetic modification regulating gene expression. However, little is known about the dynamic nature of DNA methylation and its effects on gene expression during plant development. Through whole-genome bisulfite sequencing, we showed that DNA methylation went through dynamic changes during seed maturation. An average of 66% CG, 45% CHG and 9% CHH contexts was methylated in cotyledons. CHH methylation levels in cotyledons changed greatly from 6% at the early stage to 11% at the late stage. Transcribed genes were approximately two-fold more likely to be differentially methylated than non-transcribed genes. We identified 40, 66 and 2136 genes containing differentially methylated regions (DMRs) with negative correlation between their expression and methylation in the CG, CHG and CHH contexts, respectively. The majority of the DMR genes in the CHH context were transcriptionally down-regulated as seeds mature: 99% of them during early maturation were down-regulated, and preferentially associated with DNA replication and cell division. The results provide novel insights into the dynamic nature of DNA methylation and its relationship with gene regulation in seed development.

Figure 1

Genome-wide features of DNA methylation and transcriptome of soybean cotyledons. (a) The genome-wide percentage of methylated CG, CHG and CHH as a proportion of the total CG, CHG, and CHH, respectively, in leaves and cotyledons at different stages. (b) The relative levels of methylated CG, CHG and CHH among the total methylated cytosine in leaves and cotyledons at different stages. (c) A circle plot of DNA methylation, and transcriptome in soybean cotyledon at the S6 stage. The outermost circle represents the 20 soybean chromosomes, the numbers 0, 20, 40 outside the circle represent 0 Mb, 20 Mb, and 40 Mb positions on the chromosome, respectively, and solid gray boxes and black bars indicate relative locations of pericentromeric regions and centromeric repeats, respectively. The middle circle shows the percentage of DNA methylation in the CG (black), CHG (red), and CHH (blue) contexts in 1 million base pair (bp) windows that scanned the entire genome with 100,000 bp steps. The innermost circle is a heatmap of gene expression for all expressed genes in the log₂ FPKM values. A gene with darker bar was expressed at a higher level.

Figure 2

DNA methylation patterns in protein-coding genes and transposons in soybean cotyledons. (a) End analysis of ^mCG, ^mCHG, and ^mCHH levels for each bin in gene bodies and in 4 kb upstream of the transcription start site (TSS) and 4 kb downstream of the transcription termination site (TTS) in leaf and cotyledon at S2, S6, and S8. (b) End analysis of ^mCG, ^mCHG, and ^mCHH levels in leaf and cotyledon in the following transposons and their 4 kb upstream and downstream flanking regions: Helitron DNA transposons, Long Interspersed Nuclear Element (LINE) retrotransposons, Long Terminal Repeat (LTR) retrotransposons, and Terminal Inverted Repeat (TIR) DNA transposons. The upstream, gene body and downstream regions were divided into 100 bins respectively. Percentage of methylation in each bin is shown on Y-axis.

Figure 3

Distinct expression patterns of genes that were potentially regulated by DNA methylation during seed maturation. (a) Expression patterns of gene clusters based on gene transcription patterns in cotyledon at three stages and DNA methylation in DMRs in ^mCHH, ^mCHG and ^mCG contexts. The green to red color gradient represents low to high gene expression, respectively. Genes with 1) more than 30% DNA methylation changes among three different seed stages S2, S6, and S8; 2) statistically significant changes in gene expression; and 3) negative correlation (PCC < −0.85) between gene expression and methylation levels, were used for a cluster analysis. (b) Relationship between gene transcription and DNA methylation in clusters. For each specific cluster, the Z-score of log₂ expression (Red) and the Z-score of DNA methylation levels (Green) were shown at each stage. The mean PCC between Z-score of log₂ expression and Z-score of DNA methylation levels was also calculated and shown. The complete data set is shown in Figure S3.

Figure 4

Cluster analysis of 77 genes with seed-specific CHH DMRs based on gene expression at stages S2, S6, and S8. Gene clusters based on gene transcription patterns in cotyledon at three stages and DNA methylation in DMRs in ^mCHH, ^mCHG and ^mCG contexts. The green to red color gradient represents low to high gene expression, respectively. Genes with 1) more than 30% DNA methylation changes among three different seed stages S2, S6, and S8, 2). statistically significant changes in gene expression and 3). a negative correlation (PCC < −0.85) between gene expression and methylation level were used for cluster analysis.