On the origin and evolutionary consequences of gene body DNA methylation

Abstract

In plants, CG DNA methylation is prevalent in the transcribed regions of many constitutively expressed genes (gene body methylation; gbM), but the origin and function of gbM remain unknown. Here we report the discovery that Eutrema salsugineum has lost gbM from its genome, to our knowledge the first instance for an angiosperm. Of all known DNA methyltransferases, only CHROMOMETHYLASE 3 (CMT3) is missing from E. salsugineum Identification of an additional angiosperm, Conringia planisiliqua, which independently lost CMT3 and gbM, supports that CMT3 is required for the establishment of gbM. Detailed analyses of gene expression, the histone variant H2A.Z, and various histone modifications in E. salsugineum and in Arabidopsis thaliana epigenetic recombinant inbred lines found no evidence in support of any role for gbM in regulating transcription or affecting the composition and modification of chromatin over evolutionary timescales.

Keywords: CHROMOMETHYLASE 3; DNA methylation; epigenetics; gene body methylation; histone modifications.

Fig. 1.

CMT3 and gbM are absent in E. salsugineum. (A and B) Metagene plots of DNA methylation across (A) gene bodies and (B) repeats including 4 kb up- and downstream. TTS, transcriptional termination site. (C) A syntenic block of sequence is shared between A. thaliana and E. salsugineum. The black box in the A. thaliana block indicates the location of CMT3, which is absent in E. salsugineum. The red shaded areas indicate regions of shared synteny. (D) Boxplot representation of methylation levels of individual methylated cytosines within each sequence context. (E) A bar plot of methylation levels at symmetric CHG methylated sites in A. thaliana, cmt3-11t, and E. salsugineum. (F) Density plot representation of CMT3-dependent versus CMT3-independent CHG methylation.

Fig. 2.

CMT3 is required for CHG DNA methylation and establishment of gbM in angiosperms. (A) Similar to E. salsugineum, loss of CMT3 in C. planisiliqua leads to genome-wide reductions of CHG DNA methylation at a per-site level. Closely related species that possess CMT3 maintain higher per-site CHG DNA methylation compared with C. planisiliqua. (B) The loss of CMT3 also causes the loss of gene body methylation in C. planisiliqua. (C) The loss of CMT3 (–) in E. salsugineum and C. planisiliqua represents two independent events separated by at least 13.5 My; loss of CMT3 in C. planisiliqua and E. salsugineum occurred within the last 27.41 and 40.92 My, respectively (20). However, loss in C. planisiliqua could be more recent, ≤12.27 My (20).

Fig. 3.

De novo gbM accumulates incrementally over generational time. (A) A genetic map of met1 epiRIL line 1 using only the methylation status of gbM loci from wild-type Col-0 as markers. The orange line indicates the expected heterozygous methylation levels from Col-0 and met1 epiRIL-1. Thus, the blue line indicates inheritance of methylation from Col-0 (>1) or met1 epiRIL-1 (<1). (B and C) Metaplots of CG methylation in (B) genes and (C) transposons including 4 kb up- and downstream of the TSS and TTS from the Col-0– or the met1-derived regions of the epiRIL. (D) Examples of loci in met1 epiRIL-1 where gbM has partially returned in loci that are located in met1-derived regions of the genome. Black arrows indicate where mCG returned.

Fig. 4.

Gene expression and histone modifications are not affected by the loss of gbM in E. salsugineum. (A) Comparison of gene expression levels between gbM and non-gbM within each species listed. Orthologs of gbM loci from A. thaliana were used to identify loci from E. salsugineum, and FPKM (fragments per kb of transcript per million mapped reads) values were plotted. (B) Metagene plots of H2A.Z enrichment in gbM versus unmethylated genes in A. thaliana (black line) and E. salsugineum (orange line). The y axis on the left is associated with A. thaliana, and the one on the right is associated with E. salsugineum. (C) Metagene plots of H2A.Z enrichment in two met1 epiRILs of gbM loci derived from either the Col-0 (dotted lines) or the met1 (solid lines) parent. (D and E) Heatmap representation of histone modification distributions and patterns in gene bodies of (D) A. thaliana and (E) E. salsugineum. The genes in each heatmap are ranked from highest to lowest expression levels. The vertical lines indicate the position of the transcriptional start and stop sites; 1 kb upstream of the TSS and downstream of the TTS are included in the heatmaps.