Commentary
Resolving the relationships of resolving enzymes David M. J. Lilley* and Malcolm F. White*† CRC Nucleic Acid Structure Research Group, Department of Biochemistry, The University of Dundee, Dundee DD1 4HN, United Kingdom
olliday junction-resolving enzymes are metal ion-dependent endonucleases that recognize and cleave four-way DNA junctions arising during recombination (reviewed in ref. 1). Examples include RuvC in eubacteria (2, 3), Cce1 in fungal mitochondria (4), Hjc in archaea (5, 6), and three enzymes in bacteriophage [RusA (7), T4 endonuclease VII (8), and T7 endonuclease I (9, 10)]. These six nucleases display no detectable sequence similarity with standard techniques, and the crystal structures of RuvC (11), T4 endonuclease VII (12), and T7 endonuclease I (S. E. V. Phillips and D.M.J.L., unpublished observations) are topologically dissimilar. This plethora of seemingly unrelated proteins carrying out essentially similar roles in different organisms has been a source of some puzzlement, as it has suggested that the evolution of Holliday junction resolution had occurred relatively recently, postdating the divergence of the three domains of life. This was at odds with the generally accepted view that homologous recombination is a fundamental process present in virtually all life forms and that strand-exchange proteins RecA兾Rad51 are universal. This paradox has been resolved by the identification of the pox viral junctionresolving enzyme (13), based on weak sequence similarity to both Cce1 and RuvC, and allows us to apply Occam’s Razor to simplify the resolving enzyme family tree. Garcia et al. made the connection between the eubacterial, pox, and mitochondrial resolving enzymes by using RuvC as a probe for the program PSI BLAST (14). This allowed identification of weak homologs in the pox viruses, including protein A22 in vaccinia. Subsequent searches with these proteins picked up the similarity with the fungal mitochondrial resolving enzymes. The vaccinia A22 protein was subsequently expressed in Escherichia coli and was shown to be a functional resolving enzyme in vitro (13). By approaching from another direction, we came to the same conclusion by defining four conserved motifs in the fungal resolving enzymes (Fig. 1A), containing residues identified by site-directed mutagenesis as essential for substrate binding or catalysis. These motifs were used to search for homolo-
COMMENTARY
H
Fig. 1. Conserved motifs in cellular junction-resolving enzymes. (A) Sequence alignment showing the four sequence motifs common to the mitochondrial, eubacterial, and pox viral Holliday junction resolving enzymes. The residue numbering for E. coli RuvC is indicated below the sequences. The four conserved acid residues highlighted in red (D7, E66, D138, and D142) cluster in the active site of the RuvC crystal structure (11), where they provide ligands for the catalytic metal ions. Mutagenesis studies have shown that all four of these acidic residues are essential for catalysis, as are the equivalent residues E145 (E66) and D294 (D138) in the S. cerevisiae enzyme. Basic residues conserved between RuvC and some representatives of both the mitochondrial and pox enzymes are highlighted in blue. Only one, K107, is absolutely conserved in all sequences. Conserved neutral amino acids (yellow) probably play important structural roles. Sequences are: S. cere, Saccharomyces Cce1 (GenBank no. 416770); C. albi, Candida Cce1 (17); S. pombe, Schizosaccharomyces Cce1 (GenBank no. 1256525); Vaccin, Vaccinia virus A22 protein (GenBank no. 93258); Myxoma, myxoma virus A22 homolog (GenBank no. 6523967); Mollusc, Molluscum virus A22 homolog (GenBank no. 1492070); Yaba, monkey pox virus A22 homolog (GenBank no. 6682807); RuvC, E. coli RuvC (GenBank no. 95767). (B) The four acidic residues (red) and four basic residues (blue) conserved in the eubacterial, mitochondrial, and pox viral junction resolving enzymes are highlighted by using a space-filling representation of a RuvC monomer (11). The residues all group around the active site, suggesting that motifs I to IV of A are conserved because of common roles in substrate recognition and catalysis.
gous proteins (using the program MOTIFGCG Wisconsin package), leading to the identification of the pox viral A22 homologs, which in turn allowed the connection with RuvC. The pox virus enzymes are slightly more similar to the fungal than to the eubacterial enzymes, with 31:51 (% identity兾% similarity) for the entomopox兾Ydc2 pair, and 28:39 for the entomopox兾RuvC pair. SEARCH ,
The mitochondrial enzymes, ranging in size from 258 to 353 residues, are signifi-
See companion article on page 8926 in issue 16 of volume 97. *To whom reprint requests should be addressed. E-mail:
[email protected] or
[email protected]. †Present address: Centre for Biomolecular Sciences, Univer-
sity of St. Andrews, North Haugh, St. Andrews, Fife, KY16 9ST, United Kingdom.
PNAS 兩 August 15, 2000 兩 vol. 97 兩 no. 17 兩 9351–9353
cantly larger than the pox viral (156–262 aa) or eubacterial (157–191 aa). The former have presumably accumulated a variety of insertions, some of which may serve other functions such as interactions with other recombination proteins. Despite the low levels of absolute sequence identity, which have obscured the relationships between these proteins to date, there is absolute conservation of the four acidic residues in RuvC (D7, E66, D138, and D142) known to constitute the binding site catalytic metal ion(s). Three of these residues are also conserved in other members of the integrase superfamily, where they probably perform similar roles (15, 16). In the fungal enzymes, two of these residues have been mutated and shown to play a direct role in metal ion binding by using isothermal titration calorimetry (17), and two have been mutated in vaccinia A22 by Garcia et al. and shown to be catalytically incompetent. Four basic residues are also highly conserved between representatives of the eubacterial, mitochondrial, and pox viral resolving enzymes, although only one of these (K107) is completely conserved. When the conserved acidic and basic residues are highlighted in the RuvC crystal structure, they form a clear grouping on one face of the protein (Fig. 1B), yielding further weight to the hypothesis that these enzymes share a common fold. Interestingly, the conserved residue K107 occupies a position reminiscent of the catalytic lysine found in the nuclease superfamily members [K92 in EcoRV (18) and K57 in Hjc (19)]. This residue has not been the target of any mutagenesis studies to date in either the RuvC or Cce1 model systems and would clearly be a good candidate for further analysis. Function of the Pox Viral Resolving Enzyme.
Pox viruses replicate in the cytoplasm of host cells and therefore encode all the pro-
teins necessary for DNA replication and recombination, including DNA polymerase, DNA ligase, topoisomerase, single-stranded DNA-binding protein, helicase, etc. (reviewed in ref. 20). The poxviridae replicate their DNA to produce a tandem array of linear genomes, linked by telomeric sequences that are capable of forming cruciform structures (21, 22). The enzyme responsible for the resolution of these structures has long been sought, and although the viral topoisomerase is capable of performing this role in vitro, it is probably too inefficient to carry out the reaction in vivo (23, 24). Holliday junction-resolution activity in pox virus-infected cells has been detected by biochemical screens (25), and we can now say with some degree of confidence that A22 was responsible for that activity and is involved in homologous recombination, a highly active process in the poxviridae (reviewed in ref. 26). Further studies of the effects of conditional lethal mutations of the viral resolving enzyme will no doubt shed light on these possibilities in due course. The resolving enzyme may act in concert with other proteins to catalyze telomere resolution. For example, it is likely that a different enzyme acts to form the hairpin termini of mature genomes after telomere resolution. The recent genome-wide two-hybrid screen for vaccinia protein interactions turned up the finding that A22 interacted only with itself, which is consistent with the dimeric composition of all of the resolving enzymes (27). However, it is possible that A22 interacts with other proteins via the viral nucleic acid, and this possibility requires both genetic and biochemical approaches. Wider Evolutionary Relationships. The breathtaking pace of both genomic sequencing and crystallography has revolutionized our understanding of the phylogenetic relationships of proteins. It is clear that, having hit on a stable protein fold, Nature tends to use that fold to solve a
Fig. 2. Superfamilies containing Holliday junction-resolving enzymes. Structural studies of diverse metal ion-dependent nucleases have revealed the existence of two superfamilies. The integrase superfamily, whose founding member is RNaseH, includes viral transposases and integrases as well as the Holliday junction resolving enzymes RuvC, Cce1, and poxviral A22. The nuclease superfamily, with founder member EcoRV, includes all of the type II and probably all of the type I restriction enzymes, as well as diverse DNA repair endo- and exonucleases, the archaeal resolving enzyme Hjc, and T7 endonuclease I. The bacteriophage resolving enzymes RusA and T4 endonuclease VII cannot be assigned to either of these superfamilies at present, although the lineage of the former may be discernible on solution of the crystal structure. The line lengths above bear no relationship to the degree of divergence of family members. 9352 兩 www.pnas.org
variety of seemingly unrelated problems. Solution of the crystal structures of RuvC, RnaseH1, MuA transposase, and HIV integrase revealed an unsuspected superfamily sharing the same core topology and critical metal binding residues (16). Similarly, a conserved domain consisting of a five-stranded -sheet flanked by ␣-helices and ligands for two catalytic metal ions has cropped up in a diverse collection of nucleases that includes the type I, II, and III restriction enzymes (28, 29) and the repair proteins MutH (30) and -exonuclease (31). The archaeal junction resolving enzyme Hjc is also predicted from sequence and biochemical data to share this domain (19), and T7 endonuclease I possesses the same domain and catalytic residues in a novel configuration (S. E. V. Phillips and D.M.J.L., unpublished observations). These relationships are shown in Fig. 2. Hence it appears that a domain capable of metal-activated hydrolysis of phosphodiester bonds has been grafted into structures capable of targeting specific nucleic acid structural features, nucleotide sequence, or base modification. The integrase and nuclease superfamilies share a similar ␣- structure and grouping of metal ion ligands but differ in topological organization and are thus probably related by convergent rather than divergent evolution. Future Prospects. Where now for the resolv-
ing enzymes? The unification of the cellular enzymes throws up some interesting new leads. The first concerns the identity of the elusive nuclear junction-resolving enzyme, which has resisted all attempts at identification by genetic or biochemical means. It seems highly probable that this enzyme too will share the signature of either the integrase or nuclease superfamily. One approach would be systematically to identify and screen candidates in the simplest complete eukaryotic genome, Saccharomyces cerevisiae, where nuclear recombination is a highly active process. It will be interesting to compare the mechanisms of substrate recognition and catalysis of resolving enzymes from the two superfamilies, which have arrived at the similar end points from different origins. For example, we can speculate from current biochemical evidence that the resolving enzymes from the integrase family are likely all to have sequence specificity, whereas those in the nuclease superfamily have a relaxed sequence preference, at least for the catalytic step. As bioinformatic techniques targeted to the identification of highly divergent homologous proteins continue to improve, we are likely to witness many new connections, and surprises, in this field.
We thank Mamuka Kvaratskhelia for helpful discussion. Lilley and White
(1987) J. Mol. Biol. 193, 359–376. 11. Ariyoshi, M., Vassylyev, D. G., Iwasaki, H., Nakamura, H., Shinagawa, H. & Morikawa, K. (1994) Cell 78, 1063–1072. 12. Raaijmakers, H., Vix, O., Toro, I., Golz, S., Kemper, B. & Suck, D. (1999) EMBO J. 18, 1447–1458. 13. Garcia, A. D., Aravind, L., Koonin, E. & Moss, B. (2000) Proc. Natl. Acad. Sci. USA 97, 8926–8931. 14. Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997) Nucleic Acids Res. 25, 3389–3402. 15. Rice, P. & Mizuuchi, K. (1995) Cell 82, 209–220. 16. Yang, W. & Steitz, T. A. (1995) Structure (London) 3, 131–134. 17. Wardleworth, B. N., Kvaratskhelia, M. & White, M. F. (2000) J. Biol. Chem., in press. 18. Winkler, F. K., Banner, D. W., Oefner, C., Tsernoglou, D., Brown, R. S., Heathman, S. P., Bryan, R. K., Martin, P. D., Petratos, K. & Wilson, K. S. (1993) EMBO J. 12, 1781–1795. 19. Kvaratskhelia, M., Wardleworth, B. N., Norman, D. & White, M. F. (2000) J. Biol. Chem., in press. 20. Beaud, G. (1995) Biochimie 77, 774–779.
21. Baroudy, B. M., Venkatesan, S. & Moss, B. (1982) Cell 28, 315–324. 22. Dickie, P., Morgan, A. R. & McFadden, G. (1987) J. Mol. Biol. 196, 541–558. 23. Sekiguchi, J. A., Seeman, N. C. & Shuman, S. (1996) Proc. Natl. Acad. Sci. USA 93, 785–789. 24. Palaniyar, N., Gerasimopoulos, E. & Evans, D. H. (1999) J. Mol. Biol. 287, 9–20. 25. Stuart, D., Ellison, K., Graham, K. & McFadden, G. (1992) J. Virol. 66, 1551–1563. 26. Condit, R. C. & Niles, E. G. (1990) Curr. Top. Microbiol. Immunol. 163, 1–39. 27. Uetz, P. & Hughes, R. E. (2000) Curr. Opin. Microbiol. 3, 303–308. 28. Kovall, R. A. & Matthews, B. W. (1999) Curr. Opin. Chem. Biol. 3, 578–583. 29. Davies, G. P., Martin, I., Sturrock, S. S., Cronshaw, A., Murray, N. E. & Dryden, D. T. (1999) J. Mol. Biol. 290, 565–579. 30. Ban, C. & Yang, W. (1998) EMBO J. 17, 1526– 1534. 31. Kovall, R. A. & Matthews, B. W. (1998) Proc. Natl. Acad. Sci. USA 95, 7893–7897.
COMMENTARY
1. White, M. F., Giraud-Panis, M.-J. E., Po ¨hler, J. R. G. & Lilley, D. M. J. (1997) J. Mol. Biol. 269, 647–664. 2. Iwasaki, H., Takahagi, M., Shiba, T., Nakata, A. & Shinagawa, H. (1991) EMBO J. 10, 4381–4389. 3. Connolly, B., Parsons, C. A., Benson, F. E., Dunderdale, H. J., Sharples, G. J., Lloyd, R. G. & West, S. C. (1991) Proc. Natl. Acad. Sci. USA 88, 6063– 6067. 4. Kleff, S., Kemper, B. & Sternglanz, R. (1992) EMBO J. 11, 699–704. 5. Komori, K., Sakae, S., Shinagawa, H., Morikawa, K. & Ishino, Y. (1999) Proc. Natl. Acad. Sci. USA 96, 8873–8878. 6. Kvaratskhelia, M. & White, M. F. (2000) J. Mol. Biol. 297, 923–932. 7. Mahdi, A. A., Sharples, G. J., Mandal, T. N. & Lloyd, R. G. (1996) J. Mol. Biol. 257, 561–573. 8. Mizuuchi, K., Kemper, B., Hays, J. & Weisberg, R. A. (1982) Cell 29, 357–365. 9. Dickie, P., McFadden, G. & Morgan, A. R. (1987) J. Biol. Chem. 262, 14826–14836. 10. de Massey, B., Weisberg, R. A. & Studier, F. W.
Lilley and White
PNAS 兩 August 15, 2000 兩 vol. 97 兩 no. 17 兩 9353
