A novel immunity system for bacterial nucleic acid degrading toxins and its recruitment in various eukaryotic and DNA viral systems

Abstract

The use of nucleases as toxins for defense, offense or addiction of selfish elements is widely encountered across all life forms. Using sensitive sequence profile analysis methods, we characterize a novel superfamily (the SUKH superfamily) that unites a diverse group of proteins including Smi1/Knr4, PGs2, FBXO3, SKIP16, Syd, herpesviral US22, IRS1 and TRS1, and their bacterial homologs. Using contextual analysis we present evidence that the bacterial members of this superfamily are potential immunity proteins for a variety of toxin systems that also include the recently characterized contact-dependent inhibition (CDI) systems of proteobacteria. By analyzing the toxin proteins encoded in the neighborhood of the SUKH superfamily we predict that they possess domains belonging to diverse nuclease and nucleic acid deaminase families. These include at least eight distinct types of DNases belonging to HNH/EndoVII- and restriction endonuclease-fold, and RNases of the EndoU-like and colicin E3-like cytotoxic RNases-folds. The N-terminal domains of these toxins indicate that they are extruded by several distinct secretory mechanisms such as the two-partner system (shared with the CDI systems) in proteobacteria, ESAT-6/WXG-like ATP-dependent secretory systems in Gram-positive bacteria and the conventional Sec-dependent system in several bacterial lineages. The hedgehog-intein domain might also release a subset of toxic nuclease domains through auto-proteolytic action. Unlike classical colicin-like nuclease toxins, the overwhelming majority of toxin systems with the SUKH superfamily is chromosomally encoded and appears to have diversified through a recombination process combining different C-terminal nuclease domains to N-terminal secretion-related domains. Across the bacterial superkingdom these systems might participate in discriminating `self' or kin from `non-self' or non-kin strains. Using structural analysis we demonstrate that the SUKH domain possesses a versatile scaffold that can be used to bind a wide range of protein partners. In eukaryotes it appears to have been recruited as an adaptor to regulate modification of proteins by ubiquitination or polyglutamylation. Similarly, another widespread immunity protein from these toxin systems, namely the suppressor of fused (SuFu) superfamily has been recruited for comparable roles in eukaryotes. In animal DNA viruses, such as herpesviruses, poxviruses, iridoviruses and adenoviruses, the ability of the SUKH domain to bind diverse targets has been deployed to counter diverse anti-viral responses by interacting with specific host proteins.

Figure 1.

(A) Multiple sequence alignment of the different groups of the SUKH superfamily. The consensus was derived using the following amino acid classes: a, aromatic (FHWY, black on orange); b, big (EFHIKLMQRWY, black on light blue); h, hydrophobic (ACFGHILMTVWY, black on yellow); l, aliphatic (ILV, black on dark yellow); p, polar (CDEHKNQRST, blue on gray); s, small (ACDGNPSTV, black on blue); t, tiny (AGS, white on dark blue). Secondary structures derived from PDB structures or predicted using Jpred program are indicated above the alignment (‘e’ in blue, β-sheet; ‘h’ in red, α-helix). The numbers in bracket are indicative of the excluded residues from sequences. (B) Cartoons of known structures and a homology model of the US22 IRS-N domain made using Modeller are shown in approximately similar orientation. The α-helices are shown in purple, β-sheets in yellow, and loops in gray. Surface diagrams are colored based on their positions relative to the center of the structure (outside to inside: blue to red) to illustrate the cleft. (C) Domain architectures of representatives of the SUKH superfamily. Other than the domain abbreviations already provided in the text and the rest of the domains are the Ig-fold domain overlaps with PFAM DUF525; MoeA is a domain found in the MoeA protein of the molybdopterin biosynthesis pathway; U5, herpesvirus U5-like family (PF05999).

Figure 2.

Gene neighborhoods of representative SUKH superfamily genes in bacteria are shown. Individual genes are represented as box arrows pointing from the 5′ to the 3′-end of the coding frame. Genes were named by their domain architectures. For each operon, the gi of the SUKH gene (marked with a star) and species name are indicated. Most ORFs shown as gray boxes are small ones (<80 amino acids) that appear to be false gene predictions, or in a few cases are uncharacterized genes. For species abbreviations see ‘Materials and Methods’ section.

Figure 3.

Multiple sequence alignments and structural scaffolds of the distinct families of the HNH/EndoVII fold recovered in SUKH neighborhoods: HNH, NucA, WHH, LHH, AHH, DH-NNK and GH-E. Their secondary structures are indicated above the alignments (‘e’ in blue, β-sheet; ‘h’ in red, α-helix). The numbers in bracket are indicative of the excluded residues from sequences. ‘hash’ indicates the residues involved in metal ion-binding, ‘percent’ symbol indicates the conserved histidine which is required for activation of the water molecule for hydrolysis and ‘asterisk’ indicates the conserved asparagines. On the right, structures of HNH and EndoG families are shown as cartoon representations with the central structural core colored by structural element type (α-helices in purple, β-sheets in yellow), and key catalytic residues highlighted. For those newly identified families, inferred topology diagrams of their core nuclease domains are shown with conserved catalytic residues.

Figure 4.

(A) Multiple sequence alignment of the EndoU family emphasizing the new bacterial versions found in this study. The eukaryotic EndoU domain (PDB: 2c1w) is shown to the right to indicate the spatial position of the conserved elements and the two units with three-strands each. (B) Multiple sequence alignment of the newly identified REase family. The structure of the archaeal Holliday junction resolvase (PDB: 1OB8) is shown to the right to indicate the spatial location of the conserved residues in this fold. Secondary structure elements are indicated above the alignments (‘e’ in blue, β-sheet; ‘h’ in red, α-helix). The numbers in brackets represent excluded residues from sequences and ‘hash’ indicates the catalytic residues.

Figure 5.

Domain architectures of selected examples of nuclease toxins encoded in the neighborhood of the SUKH superfamily genes. A domain architecture network of these toxins is shown to illustrate the directionality and syntactical features of their organization. Arrows indicate the polarity of domain arrangement in a polypeptide with the arrowhead pointing to the C-terminus. Newly identified domains include DUF637-N, A-link (α-helical PT domain), WXG-like, LDXD, NUC_N, PT-TGE which are non-catalytic domains, and AHH, LHH, WHH, DHNNK, GH-E, EndoU, REase, [NS]HH, DEAM, which are toxin domains. The CdiAC domain is a predicted nucleic acid modifying domain that is present in the C-termini of CdiA proteins of Photorhabdus and E. coli.