Reappraisal of the DNA phosphorothioate modification machinery: uncovering neglected functional modalities and identification of new counter-invader defense systems

Abstract

The DndABCDE systems catalysing the unusual phosphorothioate (PT) DNA backbone modification, and the DndFGH systems, which restrict invasive DNA, have enigmatic and paradoxical features. Using comparative genomics and sequence-structure analyses, we show that the DndABCDE module is commonly functionally decoupled from the DndFGH module. However, the modification gene-neighborhoods encode other nucleases, potentially acting as the actual restriction components or suicide effectors limiting propagation of the selfish elements. The modification module's core consists of a coevolving gene-pair encoding the DNA-scanning apparatus - a DndD/CxC-clade ABC ATPase and DndE with two ribbon-helix-helix (MetJ/Arc) DNA-binding domains. Diversification of DndE's DNA-binding interface suggests a multiplicity of target specificities. Additionally, many systems feature DNA cytosine methylase genes instead of PT modification, indicating the DndDE core can recruit other nucleobase modifications. We show that DndFGH is a distinct counter-invader system with several previously uncharacterized domains, including a nucleotide kinase. These likely trigger its restriction endonuclease domain in response to multiple stimuli, like nucleotides, while blocking protective modifications by invader methylases. Remarkably, different DndH variants contain a HerA/FtsK ATPase domain acquired from multiple sources, including cellular genome-segregation systems and mobile elements. Thus, we uncovered novel HerA/FtsK-dependent defense systems that might intercept invasive DNA during replication, conjugation, or packaging.

Graphical Abstract

Figure 1.

Updated phyletic distribution of Dnd systems. The heatmap shows the overall phyletic distribution of Dnd genes where the boxes are coloured based on the number of unique NCBI tax IDs at the species level against each phylum or taxonomic group shown on the Y-axis. The panel on the right shows the range and the respective colour code. The individual Dnd genes and their occurrence percentage are shown at the top. DndD and CxC ABC-ATPase distributions are presented separately.

Figure 2.

Sequence-Structure synapomorphies of DndE clades. (A–D) Representative structure topology diagrams showing key features: (A) a simple tri-helical HTH; (B) a canonical RHH MetJ/Arc-like dimer containing two RHH domains; (C) clade-1 DndE; (D) clade-2 DndE. Key residues on the core stabilizing helix and the antiparallel strands are depicted. (E–H) Representative 3D structures showing key structural synapomorphies: (E) clade-1 DndE; (F) clade-2 DndE; (G) ball and stick representation of key residues forming the DNA binding interface and interactions between the strands and the helices in DndE; (H) superimposition of DndE and F-plasmid transfer factor TraY showing identical 3D structural topology; (I) multiple sequence alignment showing consensus representative sequence of each DndE subclade where the background color code represents the mean column-wise Shannon entropy values of the corresponding position obtained from separate multiple alignments for each subclade. The entropy of the overall alignment reflects the high sequence heterogeneity and fast-evolving aspects of DndEs. Key residues that are moderately conserved within each subclade are color-coded based on their properties: positively charged (green); negatively charged (red); polar (light blue); hydrophobic (yellow); and small (grey). (J, K) Subclade-wise entropy plots of DndEs. Shannon entropy data, computed using the regular 20 amino acid alphabets, are depicted above the zero line in orange shades. Shannon entropy data, computed using a reduced 8-residue alphabet (based on chemical properties), are displayed below the zero line in blue shades. High entropy in both alphabets indicates potential positive selection for amino acids with diverse chemical characteristics at those positions.

Figure 3.

Classification and congruent evolutionary trajectories of the DndE and DndD/CxC pairs. (A) Maximum-likelihood tree topologies of DndE and DndD-CxC are shown as mirror trees. The tree on the left shows the grouping of DndE sequences into two major clades (Clade-1 and Clade-2) and multiple subclades. Likewise, the tree topology on the right clearly distinguishes the operonically linked DndD and CxC into two major clades and multiple subclades, corresponding to Clade-1 and Clade-2 DndEs, respectively. Percentage bootstrap values obtained using 1000 replicates are shown for each major node. (B, C) CLANS-based clustering using the pairwise ‘one-to-one’ alignment-based approach precisely reproduces the grouping of DndE and DndD/CxC, respectively, into major clades and subclades as shown in the phylogenetic tree topologies.

Figure 4.

Gene-neighborhoods and phyletic distribution of operonically linked DndDE and CxC-DndE anchored systems and the rectified annotation of PbeABCD system. (A: i–iii) Representative gene-neighborhoods showing the tandem or immediately adjacent occurrences of (i) Dnd systems centered on DndDE and CxC-DndE core; (ii) DndDE pairs and other Dnd genes without the CxC-DndE pair; (iii) DndDE pairs and associated Dnd genes occurring along with the CxC-DndE anchored system. Gene neighborhoods are shown as box arrows, with the direction of each arrow pointing to the orientation of individual genes. The domain names and the domain architectures of each gene product are shown within box arrows and are colored accordingly. The dotted lines at the bottom mark the boundaries of the DndDE (red) and CxC-DndE (blue) anchored systems. Each domain is assigned a unique color, with nuclease domains highlighted in the same color for clear distinction and prominent visibility within the neighborhoods. SP in the gene-neighborhoods denotes the small disordered protein (A: iv) Representative gene-neighborhoods showing the occurrence of cytosine methylases with the CxC-DndE anchored systems. (B) Rectified annotation of the so-called pbeABCD system. Gene-neighborhoods were reconstructed using the same assembly as utilized in the earlier study and are mapped to the CxC-centered system. (C) Representative gene neighborhoods showing the occurrence of diverse nucleases within the gene neighborhoods of the Dnd modification system. D) The phyletic distribution of various contextual connections and occurrences of Dnd modification systems across different phyla or taxonomic groups. The Y-axis shows different categories: (i) the complete DndABCDE operon; the DndBDE triad while lacking DndA and DndC; only the DndDE dyad being present without other PT modification genes. The ‘Duplication’ category includes instances where DndD and CxC anchored systems are found sequentially within the same genome, as indicated in panel A. The ‘Nucleases’ category encompasses all occurrences of multiple distinct nucleases associated with the Dnd modification systems. As shown in Figure 1, the boxes are color-coded to represent the number of unique NCBI tax IDs at the species level within each phylum or taxonomic group.

Figure 5.

Structural features and domain architectures of DndG and DndF. (A, B) 3D structure and domain architectures of DndG and DndF, respectively. (C–E) structural topology diagrams showing key features of individual domains present in the DndF. (F) Phyletic distribution of multiple domain fusions occurring N-terminal to the P-loop kinase domain of DndF. As shown in Figure 1, the boxes are color-coded based on the number of unique NCBI tax IDs at the species level against each phylum or taxonomic group.

Figure 6.

Structural features and domain architectures of DndH. (A) Overall 3D structure and representative domain architectures of DndH showing all individual domains. (B–D) 3D structure and structure topology diagrams showing key features of inactive C-terminal domain of SWI2/SNF2-type SF2 helicase module fused to an HTH domain (B), β-sandwich domains (C), and target recognition domain (D). (E) 3D structure of TRD + iMtase + REase triad and its comparison with equivalent domains from BpuSI (Type II-G). (F–H) 3D structure and structure topology diagrams showing key features of inactive Mtase (F), REase (G) and HerA/FtsK ATPase (H).

Figure 7.

Diversity and functional radiation of HerA/FtsK capture systems. Maximum-likelihood tree topology anchored on HerA/FtsK domain of DndH informs six distinct and well-supported clades. HerA/FtsK Clades 1–6 are colored separately, and corresponding bootstrap values obtained from 1000 replicates are shown as percentages on major nodes. The ancestral counterparts of six DndH HerA/FtsK clades formed well-defined groups (highlighted as oval) and are clustered basal to the corresponding DndH HerA/FtsK clades with high confidence support. The contextual connections of the selected members of these ancestral counterparts are shown adjacent to each clade. The selected members are shown with filled circles with respective color codes of the cluster. As shown in Figure 2, gene neighborhoods are shown as box arrows, and the domain architectures of each gene product are shown within box arrows and are colored accordingly.

Figure 8.

Phyletic decoupling, new functional models, and the overall summary of DndABCDE with associated nucleases and DndFGH systems. (A) Phyletic distribution showing different contextual occurrences and decoupling of the Dnd systems. As shown in Figure 1, the boxes are color-coded based on the number of unique NCBI tax IDs at the species level against each phylum or taxonomic group. (B) Schematic diagram showing the functional mechanisms of the newly proposed and hypothesized model of DndABCDE and associated nucleases incorporating both the modification and restriction components. (C) Schematic diagram showing the hypothesized functional mechanisms of DndFGH operating as a standalone counter-invader defense system.