The evolution of CHROMOMETHYLASES and gene body DNA methylation in plants

Abstract

Background: The evolution of gene body methylation (gbM), its origins, and its functional consequences are poorly understood. By pairing the largest collection of transcriptomes (>1000) and methylomes (77) across Viridiplantae, we provide novel insights into the evolution of gbM and its relationship to CHROMOMETHYLASE (CMT) proteins.

Results: CMTs are evolutionary conserved DNA methyltransferases in Viridiplantae. Duplication events gave rise to what are now referred to as CMT1, 2 and 3. Independent losses of CMT1, 2, and 3 in eudicots, CMT2 and ZMET in monocots and monocots/commelinids, variation in copy number, and non-neutral evolution suggests overlapping or fluid functional evolution of this gene family. DNA methylation within genes is widespread and is found in all major taxonomic groups of Viridiplantae investigated. Genes enriched with methylated CGs (mCG) were also identified in species sister to angiosperms. The proportion of genes and DNA methylation patterns associated with gbM are restricted to angiosperms with a functional CMT3 or ortholog. However, mCG-enriched genes in the gymnosperm Pinus taeda shared some similarities with gbM genes in Amborella trichopoda. Additionally, gymnosperms and ferns share a CMT homolog closely related to CMT2 and 3. Hence, the dependency of gbM on a CMT most likely extends to all angiosperms and possibly gymnosperms and ferns.

Conclusions: The resulting gene family phylogeny of CMT transcripts from the most diverse sampling of plants to date redefines our understanding of CMT evolution and its evolutionary consequences on DNA methylation. Future, functional tests of homologous and paralogous CMTs will uncover novel roles and consequences to the epigenome.

Keywords: CHROMOMETHYLASE; DNA methylation; Phylogenetics; WGBS; Whole-genome bisulfite sequencing.

Fig. 1

Phylogenetic relationships of CMTs across Embryophyta. a CMTs are separated into four monophyletic clades based on bootstrap support and the relationship of A. thaliana CMTs: (1) the gbM-dependent CMT superclade with subclades CMT1, CMT3, ZMET, and A. trichopoda; (2) CMT2; and (3) homologous (hCMT) α and β. CMT1 and CMT3 clades only contain eudicot species of plants suggesting a eudicot-specific duplication event that occurred after the divergence of eudicots from monocots and monocots/commelinids. Sister to CMT1 and CMT3 is the monophyletic group ZMET, which contains monocots, monocots/commelinids, and magnoliids. CMT2 is sister to CMT1 and CMT3. Lastly, the polyphyletic hCMT clades are sister to all previously mentioned clades. HCMTα is sister to CMT2 and the CMT superclade and contains gymnosperm and ferns. HCMTβ contains gymnosperms, ferns, and other early diverging land plants. b A collapsed CMT gene family tree showing the seven clades described in (a). Pie charts represent species diversity within each clade and are scaled to the number of species. Two duplication events shared by all angiosperms (ε) and eudicots (ϒ) gave rise to what is now referred to as CMT1, CMT2, and CMT3. These duplication events correspond to what was reported by Jiao et al. (2011). Values at nodes in (a) and (b) represent bootstrap support from 1000 replicates and (a) was rooted to the clade containing all liverwort species

Fig. 2

Non-neutral evolution of CMT3 in the Brassicaceae is correlated with reduced levels of genic mCG and numbers of gbM loci. a Distribution of mCG upstream, downstream, and within gene bodies of Brassicaceae species and outgroup species Prunus persica. MCG levels within gene bodies of Brassicaceae species are within the bottom 38% of 34 angiosperms. Data used represent a subset of that previously published [19, 25]. TSS transcriptional start site, TTS transcriptional termination site. b Similarly, the number of gbM genes within the genome of Brassicaceae species are within the bottom 15% of 34 angiosperms. The size of the circle corresponds to the number of gbM genes within each genome. Data used represent a subset of that previously published [19, 25]. c Changes at the amino acid level of CMT3 is correlated to reduced genic levels of DNA methylation and number of gbM genes in the Brassicaceae. An overall higher rate of molecular evolution, measured as the number of non-synonymous substitutions per non-synonymous site (dN) divided by the number of synonymous substitutions per synonymous site (dS) (ω), was detected in the Brassicaceae. Also, a higher rate ratio of ω was detected in the Brassicaceae clade containing B. rapa and closely related species compared to the clade containing A. thaliana and closely related species. The higher rate ratio in the Brassicaceae, compared with the background branches, was not attributed to positive selection

Fig. 3

Variation in levels of DNA methylation within gene bodies across Viridiplantae. a DNA methylation at CG, CHG, and CHH sites within gene bodies can be found in the majority of species investigated. Variation of DNA methylation levels within gene bodies at all sequence contexts is high across all land plants and within major taxonomic groups. mCG levels are typically higher than mCHG, followed by mCHH. However, levels of mCG and mCHG within genes are similar in gymnosperms and ferns. Error bars represent 95% confidence intervals for species with low sequencing coverage. Cladogram was generated from Open Tree of Life [53]. b The distribution of DNA methylation within genes (all [dashed lines] and mCG-enriched/gbM [solid lines]) has diverged among taxonomic groups of Viridiplantae represented by specific species. Based on the distribution of DNA methylation, and number of mCG-enriched genes, gbM is specific to angiosperms. However, mCG-enriched genes in P. taeda share some DNA methylation characteristics to A. trichopoda. However, other characteristics associated with gbM genes remains unknown at this time for mCG-enriched genes in gymnosperms and other early diverging Viridiplantae. The yellow highlighted line represents the average from 100 random sampling of 100 gbM genes in angiosperms and was used to assess biases in numbers of mCG-enriched genes identified. NCR non-conversion rate, TSS transcriptional start site, TTS transcriptional termination site

Fig. 4

Presence/absence (+/–) of genes likely involved in the evolution of gbM and heterochromatin formation for various taxonomic groups of Viridiplantae. Families (orthogroups) of gbM- and heterochromatin-related genes are taxonomically diverse. However, after phylogenetic resolution, clades containing proteins of known function in A. thaliana are less diverse. Specifically, the CMT3 and orthologous genes (ZMET2 and ZMET5, and A. trichopoda CMT3) and IBM1 are angiosperm-specific. Other clades – SUVH4 and homologous SUVH5/6 (hSUVH5/6) – are more taxonomically diverse, which might relate to universal functions in heterochromatin formation