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

Zhang, Dapeng; Iyer, Lakshminarayan M.; Aravind, L.

We also recovered three other novel families of domains, which are respectively typified by nearly absolutely conserved tripeptide sequence motifs LHH, WHH and AHH (Figure 3 ). Most CDI operons, which encode a SUKH domain immunity protein, have proximal toxin genes with a LHH domain as the polymorphic C-terminal unit of their products ( Figure 2) . Additionally, the LHH domain is found in products of genes adjacent to the SUKH superfamily gene outside of proteobacteria in several other bacterial lineages such as firmicutes, actinobacteria, bacteroidetes and planctomycetes ( Figure 2 ). Although we also found the WHH domain as the polymorphic toxin unit of a subset of proteobacterial CDI systems, none of these have a SUKH superfamily immunity protein. However, we found several non-CDI gene neighborhoods, which are likely to define distinct but analogous toxin systems, in proteobacteria, firmicutes, actinobacteria, synergistetes and bacteroidetes that combine genes for WHH and SUKH domain proteins (Figure 2 ). The AHH domain is also found in similarly organized gene-neighborhoods from the same bacterial lineages as those in which the WHH and LHH domains are found. Profile-profile comparisons with multiple alignments of all these three novel domains indicated that the best matches are families of the HNH fold. Indeed, a visual examination of the conservation patterns of these three domains showed that the HH dyad shared by them corresponds to the HH or DH dipeptide found in the first strand of treble-clef fold of the classical HNH domains (52). The first H forms one of the catalytic metal-chelating ligands and the second H contributes to the active site that directs the water for phosphoester hydrolysis (58) . Further, the sequence alignments of the LHH, WHH and AHH motifs revealed two further conserved histidines, which were associated with the helix of the treble-clef fold and aligned with the two C-terminal metal-chelating residues in the profile of the classical HNH domains (58, 59) . These observations indicated that the LHH, WHH and AHH domains are highly derived versions of the HNH fold. The eighth family of HNH fold enzymes emerging from this analysis comprises of proteins typified by the protein Dd586_1447 (gi: 271499995) from D. dadanti found in predicted toxins in SUKH neighborhoods and also in CDI operon products which do not contain a SUKH-type immunity protein (Figure 2) . A subset of these domains constitutes the PFAM model for a 'domain of unknown function', DUF1994, that does not define the boundaries of this domain precisely. We were able to define the proper boundaries of this domain by using the diversity of distinct architectural contexts in which we detected it and used the refined alignment for profile-profile comparisons. This comparison revealed the representatives of HNH domains as the best hits and indicated a perfect match between the polar residues conserved in this domain and catalytic and active-site metal chelating residues of the classical HNH domains. We named this family of HNH domains as DH-NNK after the conserved DH dyad in the strand-1 and the two asparagines and lysine which are conserved in the helix of the core treble-clef fold (Figure 3 ). While all these above versions have lost the cysteines of the ancestral treble-clef, they nevertheless, retain the catalytic configuration typical of those nucleases. Hence, we predict that these domains are likely to be nucleases with a similar catalytic mechani