The Epigenomic Landscape of Prokaryotes

Abstract

DNA methylation acts in concert with restriction enzymes to protect the integrity of prokaryotic genomes. Studies in a limited number of organisms suggest that methylation also contributes to prokaryotic genome regulation, but the prevalence and properties of such non-restriction-associated methylation systems remain poorly understood. Here, we used single molecule, real-time sequencing to map DNA modifications including m6A, m4C, and m5C across the genomes of 230 diverse bacterial and archaeal species. We observed DNA methylation in nearly all (93%) organisms examined, and identified a total of 834 distinct reproducibly methylated motifs. This data enabled annotation of the DNA binding specificities of 620 DNA Methyltransferases (MTases), doubling known specificities for previously hard to study Type I, IIG and III MTases, and revealing their extraordinary diversity. Strikingly, 48% of organisms harbor active Type II MTases with no apparent cognate restriction enzyme. These active 'orphan' MTases are present in diverse bacterial and archaeal phyla and show motif specificities and methylation patterns consistent with functions in gene regulation and DNA replication. Our results reveal the pervasive presence of DNA methylation throughout the prokaryotic kingdoms, as well as the diversity of sequence specificities and potential functions of DNA methylation systems.

Fig 1. Methylomes of 230 prokaryotes.

A) Phylogenetic tree of 230 sequenced organisms. Outer bars indicate the number and types of active MTases detected per genome B) Number of methylated motifs and MTase genes identified across all 230 organisms. C) Breakdown of 583 annotated DNA MTases by type. D) Proportion of annotated DNA MTases with and without cognate restriction enzymes. For Type IIG systems, MTase and restriction activities are encoded in the same peptide.

Fig 2. Novelty and diversity of methylome data.

A) Novelty of MTases annotated in this study. The 583 newly annotated MTases were compared against all existing annotated MTases in the REBASE database [5], and classed as novel or homologous to existing MTases on the basis of their sequence specificities. B) Impact of this study on the diversity of known MTase specificities. ‘Previously existing’ refers to all specificities in the REBASE database prior to this study. New specificity refers to methylated motifs identified in this study. C) Diversity of DNA recognition by Type I RM systems. Type I specificity subunits recognize bipartite motifs. In this data, all aspects of this architecture are found to vary. Red bases indicate the positions of 6mA methylation (A), or their reverse complement (T). Sub-motif sequence variation shows the IUPAC codes for ambiguous nucleotides found within the bipartite DNA sequences.

Fig 3. Identification and classification of ‘Orphan’ Type II MTases.

A) Proportion of Type II MTases encoded with (RM-system) or without (‘orphan’) a cognate restriction system. B) Proportion of sequenced organisms with orphan Type II MTases. C) Comparison of evolutionary conservation properties of Type II MTases. An MTase is considered conserved if orthologs are present in >50% species in the respective taxonomic class. D) Table of orphan MTase families identified based on protein sequence clustering (Methods and S2 Fig). Tree is based on agglomerative clustering of protein sequences. Bar charts represents fraction of related species with (green) or without (grey) a copy of the orphan MTase gene (diagonal lines, < 5 sequenced genomes).

Fig 4. Orphan MTases are associated with unmethylated sites in the gene regulatory regions.

A) Scatter plot of DNA modification scores on forward and reverse strands of each Dam MTase target motif (GATC) in the E. coli genome. The ipdR (inter-pulse duration ratio) is the primary metric in DNA modification detection, and corresponds to the time delay in incorporation of successive bases in a sample versus an unmodified control. This plot reveals a distinct set of sites that is unmethylated on both strands of the genome (highlighted in red). B) Scatter plot of DNA methylation scores at Dam target motif sites in Salmonella bongorii reveals a similar set of unmethylated sites. C) Scatter plot of DNA methylation scores GATC-specific RM-system MTase in Clostridium thermocellum. In this case all sites in the genome are methylated. D) Systematic analysis of the number of unmethylated motifs (on both strands of the genome) associated with orphan MTases (blue panel), and RM-system MTase (red panel). Orphan MTase names and gene orders correspond to Fig 3. E) Fold enrichment of all motifs (grey bars) and unmethylated motifs (black bars) in gene regulatory regions. * = significantly enriched (p <0.01, Fishers exact). Letters indicate enrichment at specific functional categories of genes based on COG category analysis. K = transcription, T = signal transduction, H = coenzyme metabolism.

Fig 5. Examples of candidate regulatory unmethylated sites.

In all panels bar charts show the extent of DNA methylation (inter-pulse duration ratio) at the candidate regulatory unmethylated site (blue) and, for comparison, at the ten immediately flanking upstream and downstream motif instances (red). The sequence interval covered in each chart varies due to the density of motifs across the respective genomes. A) Unmethylated sites are present upstream of the Hpa operon in four Enterobacteria species. In 3 cases, the unmethylated site is at the orthologous GATC, in S. bongori, the unmethylated site is located ~100bp upstream of the conserved sites. B) Conserved unmethylated site upstream of a PadR transcriptional regulator in Arthrobacter species. C) Cluster of unmethylated sites upstream of transcriptional regulator and sugar degradation operon in Spirochaeta smaragdinae. D) Cluster of non- or weakly-methylated sites throughout a non-ribosomal peptide synthase operon.

Fig 6. Identification of putative novel orphan MTase regulators of DNA replication.

Four orphan MTases were found to be associated with enriched clusters of motifs in non-coding regions of the genome (Methods). Plots show density of motifs across a 50kb region of the genome flanking the motif cluster. Data is shown for the organism in which the pattern was originally identified, along with related organisms from the same taxonomic group. For comparison, plots were also generated from closely related organism lacking the orphan MTase. In each panel, a reference MTase sequence is selected (from Fig 1). The similarity score of the best scoring orthologs in each genome is represented in the MTase column using a white-green scale. A) Density of GATC motifs flanking the gidA gene (origin of replication) in Enterobacteria and other Gammaproteobacteria. B) Density of CTCGAG motifs flanking the dnaA gene terminus in Nocardiaceae and other Actinobacteria species. C) Density of TTAA motifs flanking the dnaA gene terminus in Arthrobacter and other Actinobacteria species. D) Density of CTAG motifs flanking the orc1/cdc6 gene start in Haloarchaea and other Euryarchaeota species (Bold and underlined characters indicate methylated bases. Underlined characters indicate reverse complement of methylated bases). In each example, the presence of motifs clusters correlates with the presence of the respective MTase in the same genome.