A model for the evolution of prokaryotic DNA restriction-modification systems based upon the structural malleability of Type I restriction-modification enzymes.

Restriction Modification (RM) systems prevent the invasion of foreign genetic material into bacterial cells by restriction and protect the host's genetic material by methylation. They are therefore important in maintaining the integrity of the host genome. RM systems are currently classified into four types (I to IV) on the basis of differences in composition, target recognition, cofactors and the manner in which they cleave DNA. Comparing the structures of the different types, similarities can be observed suggesting an evolutionary link between these different types. This work describes the 'deconstruction' of a large Type I RM enzyme into forms structurally similar to smaller Type II RM enzymes in an effort to elucidate the pathway taken by Nature to form these different RM enzymes. Based upon the ability to engineer new enzymes from the Type I 'scaffold', an evolutionary pathway and the evolutionary pressures required to move along the pathway from Type I RM systems to Type II RM systems are proposed. Experiments to test the evolutionary model are discussed.

DNA target recognition domains in the Type I restriction and modification systems of Staphylococcus aureus.

Staphylococcus aureus displays a clonal population structure in which horizontal gene transfer between different lineages is extremely rare. This is due, in part, to the presence of a Type I DNA restriction-modification (RM) system given the generic name of Sau1, which maintains different patterns of methylation on specific target sequences on the genomes of different lineages. We have determined the target sequences recognized by the Sau1 Type I RM systems present in a wide range of the most prevalent S. aureus lineages and assigned the sequences recognized to particular target recognition domains within the RM enzymes. We used a range of biochemical assays on purified enzymes and single molecule real-time sequencing on genomic DNA to determine these target sequences and their patterns of methylation. Knowledge of the main target sequences for Sau1 will facilitate the synthesis of new vectors for transformation of the most prevalent lineages of this 'untransformable' bacterium.

Mutations of the domain forming the dimeric interface of the ArdA protein affect dimerization and antimodification activity but not antirestriction activity.

ArdA antirestriction proteins are encoded by genes present in many conjugative plasmids and transposons within bacterial genomes. Antirestriction is the ability to prevent cleavage of foreign incoming DNA by restriction-modification (RM) systems. Antimodification, the ability to inhibit modification by the RM system, can also be observed with some antirestriction proteins. As these mobile genetic elements can transfer antibiotic resistance genes, the ArdA proteins assist their spread. The consequence of antirestriction is therefore the enhanced dissemination of mobile genetic elements. ArdA proteins cause antirestriction by mimicking the DNA structure bound by Type I RM enzymes. The crystal structure of ArdA showed it to be a dimeric protein with a highly elongated curved cylindrical shape [McMahon SA et al. (2009) Nucleic Acids Res 37, 4887-4897]. Each monomer has three domains covered with negatively charged side chains and a very small interface with the other monomer. We investigated the role of the domain forming the dimer interface for ArdA activity via site-directed mutagenesis. The antirestriction activity of ArdA was maintained when up to seven mutations per monomer were made or the interface was disrupted such that the protein could only exist as a monomer. The antimodification activity of ArdA was lost upon mutation of this domain. The ability of the monomeric form of ArdA to function in antirestriction suggests, first, that it can bind independently to the restriction subunit or the modification subunits of the RM enzyme, and second, that the many ArdA homologues with long amino acid extensions, present in sequence databases, may be active in antirestriction.