Crystal structure of the Holliday junction- resolving enzyme T7 endonuclease I at 2.1Å resolution

Jon Hadden, Maire Convery and Simon Phillips

Introduction

Homologous genetic recombination is important in DNA rearrangements and in the repair of double-strand breaks in DNA. The central DNA intermediate in this process is the four-way (Holliday) junction, and recognition and manipulation of this structure by proteins are important elements of the mechanism. The penultimate stage of recombination requires the resolution of the junction into component duplex species by junction-specific nucleases. In general these relatively small, basic proteins bind to DNA junctions in a highly structure-selective manner. Structure-selective interactions are fundamental to the mechanism of genetic recombination.

Bacteriophage T7 DNA undergoes genetic recombination during infection. The phage-encoded junction-resolving enzyme is endonuclease I. Mutants in the gene encoding this enzyme are deficient in recombination and accumulate branched DNA intermediates. The crystal structures of two junction-resolving enzymes have been previously reported. The structure of RuvC has been determined at 2.5 Å resolution, while that of T4 endonuclease VII has recently been described at 2.4 Å resolution, together with an inactive mutant at 2.1 Å resolution. No structural similarity is discernible between these two enzymes. We have determined the crystal structure of a third junction-resolving enzyme T7 endonuclease I at 2.1 Å resolution. This is the first structure of a junction-resolving enzyme that falls into the superfamily of proteins that includes the restriction enzymes.

Figure 1. a) Overall structure of an endonuclease 1 homodomer. Individual monomers are shown in red and green
b) Close up view of the active site of endonuclease 1.

Endonuclease I forms an intimately associated symmetrical homodimer comprising two domains (Fig. 1a). Each domain is composed of residues 17-44 from one subunit and residues 50-145 from the other. The two domains are connected by a bridge that forms part of an extended ß-sheet (ß2). Endonuclease I has an extensive dimer interface with many interactions along the ß-sheet bridge and between the N-terminal region (17-44) and the body of the other subunit. This is consistent with our observation that in free solution, subunit exchange in endonuclease I is undetectable in the absence of denaturing agents.

Each domain comprises a central five-stranded mixed ß-sheet, flanked by five ß-helices with one strand and one helix contributed by the other subunit in the dimer. The compact nature of each domain suggests a stable structure that may function independently.

Mutagenesis experiments have previously identified five acidic residues which could potentially be involved in catalysis in endonuclease I, Glu 20, Glu 35, Asp 55, Glu 65 and Asp 74. Glu 20, Asp 55 and Glu 65 (mutated to Lys 65 in the crystallised protein) are clustered and form an acidic cluster on the surface of the protein and we suggest that these residues constitute the active site of the nuclease. Closer examination of this region (Fig. 1b) reveals that the arrangement of residues is closely similar to that found in the active sites of a number of well-characterised restriction endonucleases, including, BglI, EcoRV, EcoRI and FokI. The restriction endonucleases normally possess three or four active site residues, typically two (BamHI, EcoRI, FokI) or three (BglI, EcoRV) acidic amino acids and a lysine residue are involved in catalysis. The acidic amino acids are thought to be involved in chelating one or two metal ions, whilst the lysine residue probably plays a role in stabilising the transition state or the product of the cleavage reaction. A structural alignment of a number of restriction enzymes suggests that the active site of endonuclease I belongs to the same family as BglI and EcoRV, and that lysine 67 could also be involved in catalysis.

T7 endonuclease I shows little structural similarity with T4 endonuclease VII. The latter is almost totally a-helical except for a section of b-sheet organised as part of a zinc domain that carries a number of acidic and histidine side chains required for catalysis. RuvC has an a/b fold, but with different topology from endonuclease I. The protein domain-recognition program DALI failed to identify any significant structural similarity between the three proteins.

By contrast, there is clearly a significant structural similarity between the active site of endonuclease I and those of a number of restriction enzymes, discussed above. Moreover, the archaeal junction-resolving enzyme Hjc has an active site that is predicted, by sequence motif and site-directed mutagenesis, to be very similar to that of endonuclease I and the restriction enzymes. These enzymes, together with MutH and l-exonuclease, are therefore likely to be related to an ancestral nuclease, forming a superfamily of nucleases. The essential nuclease function has acquired different types of specificity in the various enzymes, so that the restriction enzymes exhibit high sequence specificity, while the resolving enzymes such as endonuclease I are highly selective for the branched structure of DNA junctions.

Collaborators: A.-C. Déclais, D.M.J. Lilley CRC Nucleic Acid Structure Research Group, Department of Biochemistry, University of Dundee, DD1 4HN, UK.

References

Hadden, J.M., Convery, M.A., Déclais, A-C., Lilley, D.M.J. & Phillips, S.E.V. (2001).

Crystal Structure of Bacteriophage T7 Endonuclease I: A Holliday Junction Resolving Enzyme. Nature Structural Biology, 8, 62-67.

Funding: We acknowledge the support of The Wellcome Trust