Site-directed Mutagenesis of the Yeast Resolving Enzyme Cce1 Reveals Catalytic Residues and Relationship with the Intron-splicing Factor Mrs1*

The Holliday junction-resolving enzyme Cce1 is a magnesium-dependent endonuclease, responsible for the resolution of recombining mitochondrial DNA molecules inSaccharomyces cerevisiae. We have identified a homologue of Cce1 from Candida albicans and used a multiple sequence alignment to predict residues important for junction binding and catalysis. Twelve site-directed mutants have been constructed, expressed, purified, and characterized. Using this approach, we have identified basic residues with putative roles in both DNA recognition and catalysis of strand scission and acidic residues that have a purely catalytic role. We have shown directly by isothermal titration calorimetry that a group of acidic residues vital for catalytic activity in Cce1 act as ligands for the catalytic magnesium ions. Sequence similarities between the Cce1 proteins and the group I intron splicing factor Mrs1 suggest the latter may also possess a binding site for magnesium, with a putative role in stabilization of RNA tertiary structure or catalysis of the splicing reaction.

The Holliday junction-resolving enzyme Cce1 is a magnesium-dependent endonuclease, responsible for the resolution of recombining mitochondrial DNA molecules in Saccharomyces cerevisiae. We have identified a homologue of Cce1 from Candida albicans and used a multiple sequence alignment to predict residues important for junction binding and catalysis. Twelve site-directed mutants have been constructed, expressed, purified, and characterized. Using this approach, we have identified basic residues with putative roles in both DNA recognition and catalysis of strand scission and acidic residues that have a purely catalytic role. We have shown directly by isothermal titration calorimetry that a group of acidic residues vital for catalytic activity in Cce1 act as ligands for the catalytic magnesium ions. Sequence similarities between the Cce1 proteins and the group I intron splicing factor Mrs1 suggest the latter may also possess a binding site for magnesium, with a putative role in stabilization of RNA tertiary structure or catalysis of the splicing reaction.
Holliday junctions arising during homologous recombination are resolved by a class of structure specific endonucleases, yielding recombinant DNA duplex products. Holliday junctionresolving enzymes are ubiquitous, although the nuclear enzyme(s) have not yet been identified. They are all dimeric, metal-dependent endonucleases with strong specificity for the structure of the Holliday junction but differ in their requirements for specific nucleotide cleavage recognition sequences (reviewed in Refs. 1 and 2). Cce1 resolves recombining mtDNA genomes in yeast mitochondria, and cce1 mutants display highly branched mtDNA, linked by unresolved Holliday junctions (3), and have an increased frequency of petite mutants due to impaired mitochondrial function (4). Cce1 binds Holliday junctions as a dimer (5), holding the junction in a square, 4-fold symmetric conformation (6), with disruption of base pairing at the point of strand exchange (7). Junctions are cleaved only when the recognition sequence 5Ј-CT/ is positioned at or adjacent to the junction center (8). A homologue of Cce1, Ydc2, has been identified in the distantly related fission yeast Schizosaccharomyces pombe (9 -11), suggesting that this enzyme has a conserved role in mtDNA recombination in the fungi and possibly in higher eukaryotes.
S. cerevisiae Cce1 has previously been observed to share 29% sequence identity with the yeast protein Mrs1 (5). Mrs1 (Pet157) is a nuclear encoded protein from S. cerevisiae that is required for splicing of two group I introns in the mitochondrion (12,13). It may function in stabilizing the folded structure of the group I intron to promote self-splicing, as has been shown for the protein Cbp2 (14,15).
In this paper, we report the identification of a further homologue of Cce1, from the fungus Candida albicans. Sequence alignments of the three Cce1 proteins along with the group I intron accessory protein Mrs1 highlighted subsets of amino acids either conserved in all four proteins or only in the Holliday junction-resolving enzymes. An extensive program of sitedirected mutagenesis coupled with kinetic analysis and isothermal titration calorimetry has been carried out to address the roles of these conserved residues and the relationship of Cce1 with Mrs1. The results suggest roles in junction binding and catalysis for basic and acidic residues in Cce1 and a metal ion binding pocket potentially conserved in Cce1 and Mrs1.

EXPERIMENTAL PROCEDURES
Identification of the Cce1 Homologue of C. albicans-Sequence data for C. albicans was obtained from the Stanford DNA Sequencing and Technology Center site on the World Wide Web. Sequencing of C. albicans was accomplished with the support of the NIDR and the Burroughs Wellcome Fund.
Oligonucleotides-Junction 1 is a fixed four-way junction with 20base pair arms, described previously (5). Junction Z1 is a fixed four-way junction with 15-base pair arms, the h strand containing a consensus site for cleavage by Cce1, described previously (8).
Oligonucleotide Synthesis and Assembly of DNA Junctions-Oligonucleotides were synthesized and DNA junctions were assembled as described previously (5).
Protein Purification-Native recombinant Cce1 was expressed and purified to near homogeneity as described previously (6).
Single Turnover Kinetic Analysis-Determination of the first order rate constants of junction cleavage were carried out by first incubating 8 ϫ 10 Ϫ8 M 5Ј-32 P-labeled junction J1 (or junction Z1 for some mutants) with 8 ϫ 10 Ϫ7 M Cce1 dimer in binding buffer (20 mM Tris, pH 8.0, 200 mM NaCl, 0.2 mM dithiothreitol, 1 mM EDTA, 0.1 mg/ml bovine serum albumin). Samples were pre-equilibrated at 37°C, and reactions were initiated by the addition of MgCl 2 to a final concentration of 15 mM. At set time points, an aliquot of the reaction mix was removed, and the reaction was stopped by the addition of an equal volume of formamide loading buffer (95% (v/v) formamide, 50 mM EDTA, pH 8.0, 0.1% bromphenol blue, 0.1% xylene cyanol FF) and heating at 80°C for 2 min, followed by storage on ice. Reaction products were analyzed by denaturing gel electrophoresis and phosphor imaging as described previously (8). All experiments were carried out in triplicate, and, where not specifically stated, S.E. values were typically less than 5%. For mutants E145Q, D293N, and D294N, activity was also assayed using the fourway junction Z1, which is cleaved approximately 20-fold more quickly than J1 by wild-type Cce1. Use of junction Z1 provided a more sensitive assay for residual enzyme activity, allowing detection of up to a 60,000fold reduction in activity compared with the wild-type enzyme.
Measurement of Equilibrium Dissociation Constants-Equilibrium dissociation constants were determined by gel electrophoretic retardation analysis and quantified by phosphor imaging, and the data were fitted to an equation for the binding model, as described in Ref. 8.
Site-directed Mutagenesis-Site-directed mutagenesis of CCE1 was carried out in plasmid pUC119-CCE1 (5) using the QuikChange protocol (Stratagene). After mutagenesis, DNA sequencing was used to confirm that no spurious mutations had been introduced. The CCE1 mutants were subcloned into pET19b, expressed, and purified as for wildtype Cce1.
Isothermal Titration Calorimetry-ITC experiments were carried out at 25°C using VP-ITC titration calorimeter (MicroCal, Northampton, MA). All solutions were degassed before the titrations. Cce1 samples were extensively dialyzed against 20 mM Tris-HCl buffer, pH 8.0, containing 200 mM NaCl, and MgCl 2 solutions were prepared in the same buffer. Titration was carried out using a 370-l syringe with stirring at 400 rpm. Each titration consisted of a preliminary 1-l injection followed by 20 -30 subsequent 10-l injections into a cell containing approximately 1.4-ml enzyme solutions. Calorimetric data were analyzed using MicroCal ORIGIN software. All measurements of binding parameters presented are the means of duplicate experiments.

RESULTS
Identification of a Third Sequence for Cce1 from C. albicans-Searches of the unfinished C. albicans genome sequence data (on the World Wide Web) using Saccharomyces cerevisiae Cce1 as a probe revealed a significant match to a predicted open reading frame in contig 1 4-2987. The Cce1 sequences from S. cerevisiae, S. pombe, and C. albicans share approximately 33% sequence identity. Multiple sequence alignment of the three Cce1 sequences, together with that of Mrs1 from S. cerevisiae, was carried out using the program ClustalX, followed by manual adjustment (Fig. 1). A limited number of residues are conserved in all the Cce1 sequences, and a subset of those are also conserved in Mrs1. These amino acids are likely to play important roles in the structure and function of the proteins.
Site-directed Mutagenesis of S. cerevisiae Cce1-Ten residues in conserved regions of Cce1 were selected for site-directed mutagenesis in order to investigate their importance in fourway DNA junction binding and cleavage. Mutations were introduced using the Stratagene QuikChange system and checked by sequencing of the entire CCE1 coding sequence. The mutated proteins were expressed in E. coli strain BL21 (DE3) using the vector pET19b and purified as described previously for the wild-type enzyme. All the mutants eluted from size exclusion chromatography with a retention time similar to the wild-type enzyme, suggesting that the dimeric quaternary structure of the wild-type enzyme is unaltered in the mutants. For each mutant, catalytic activity was measured using a single turnover kinetic assay, with the four-way junction J1, radioactively labeled on the r strand, as a substrate. For some mutants, the four-way junction Z1, which is cleaved about 20-fold more quickly that J1 (8), was used to provide a more sensitive assay of reduced enzyme activity. Equilibrium dissociation constants were measured by gel electrophoretic retardation analysis in the presence of EDTA. The data for the wild-type and all mutant enzymes is summarized in Table I.
Basic Residues in Cce1-Arg 146 and Arg 150 are both conserved in three of the four proteins. The R150A mutant gave the smallest decrease in activity (4-fold) of any of the mutants tested and a moderate decrease in binding affinity, suggesting that it does not play an important role in catalysis or substrate recognition, whereas the equivalent mutation in Arg 146 resulted in a modest decrease in junction binding affinity but a relatively large effect on catalytic activity, which is 100-fold reduced compared with the wild-type enzyme. Lysine 291 is specific to the junction-resolving enzymes. It appears to have an important role in catalysis, since mutation to an alanine results in a complete loss of detectable activity. The more conservative replacement with an arginine is also catalytically inactive, suggesting that the residue plays a more specific role in catalysis, rather than merely contributing a positive charge. Indeed, while the K291A mutant still binds four-way DNA junctions relatively well, the K291R mutant displays significantly weaker binding. Possibly, the presence of the more bulky arginine residue is sterically unfavorable in the protein-DNA complex. Arginine 231 is conserved in all four proteins, and the R231A mutant displays the weakest binding of all of the mutants tested, with a 330-fold decrease in binding affinity. The more conservative replacement with a lysine rescues this phenotype to some extent, yielding 2.5-fold tighter binding and a detectable activity. On the basis of these data, we predict that Arg 231 plays an important role in binding the nucleic acid substrate, although probably not in catalysis.
Acidic Residues-Four acidic residues in Cce1 have been mutated to their corresponding amide forms. None of these 1 The abbreviation used is: contig, group of overlapping clones. mutations results in significant changes in the equilibrium binding affinity of the enzyme for the four-way junction. Asp 292 is not conserved in any of the other sequences, and not surprisingly the D292N mutant retains appreciable catalytic activity. The significant decrease in catalytic activity observed (80-fold) may reflect the fact that this residue is clearly adjacent to the catalytic site, and any mutation at this position might well influence the conformation or overall charge of the active site. Asp 293 is conserved in the resolving enzymes but not in Mrs1. The D293N mutant displays a 600-fold decrease in catalytic activity (using junction Z1 as a substrate), suggesting an important role in catalysis, possibly by providing a ligand for one of the catalytic magnesium ions. Two acidic residues (Glu 145 and Asp 294 ) are conserved in all four proteins. Mutation of either results in complete loss of activity with the junction J1 substrate. Using the faster cutting junction Z1 enabled the detection of very weak cleavage activity, reduced 60,000-fold compared with the wild-type enzyme for D294N, and no detectable activity for E145Q.
Neutral Amino Acids-Replacement of the absolutely conserved residue Phe 79 with alanine results in a significant drop in junction binding affinity and a complete loss of detectable activity. This residue could conceivably play a role in catalysis in Cce1, for example by aromatic interactions with disrupted bases near the cleavage site, but we cannot rule out the possibility that it has an important structural role in the proteins, which could be disrupted by the nonconservative change to an alanine. The Q147A mutation results in a moderate effect on binding, but a 70-fold drop in the cleavage rate. Gln 147 is conserved in the junction-resolving enzymes and is replaced by the conservative substitution of a glutamate in Mrs1. Gln 147 does not appear to play a role in metal ion binding (see below) but may conceivably play another role in catalysis.
Isothermal Titration Calorimetry-We have shown previously that wild-type Cce1 binds two magnesium ions per monomer with a micromolar dissociation constant (16). We selected three mutants, E145Q, Q147A, and D294N, for direct analysis of metal ion binding by isothermal titration calorimetry (Fig.  2). Q147A was found to bind two metal ions, as observed for wild-type Cce1, but binding was abolished in the E145Q and D294N mutants. These residues are thus likely to act as ligands for the catalytic metal ions at the active site.

DISCUSSION
Site-directed Mutagenesis of Cce1-In common with other nucleases, Holliday junction-resolving enzymes are thought to hydrolyze phosphodiester bonds by means of a metal-activated water molecule. Acidic residues typically function as ligands for catalytic metal ions in nucleases. Mutation of each of three acidic residues (Glu 145 , Asp 293 , and Asp 294 ), which are conserved in all three Cce1 homologues, resulted in very large decreases in the catalytic activity without affecting junction binding affinity. Previously, the mutation D226N in S. pombe a These mutants were also assayed using junction Z1, which provided a more sensitive assay of residual activity, as discussed under "Experimental Procedures."  (16). In contrast, no heat effect was observed from titration of a solution of magnesium into either the D294N (Ⅺ) or E145Q (q) enzymes under identical conditions, suggesting disruption of the metal binding site in these mutants.
Ydc2 (equivalent to D294N in S. cerevisiae Cce1) was shown to reduce drastically catalytic activity without affecting binding specificity (17). Similar phenotypes have been demonstrated for mutations of acidic residues in RuvC (18), T4 endonuclease VII (19 -21), T7 endonuclease I (22), RusA (23,24), and Hjc (25), and in the case of RuvC the acidic residues in question are known to cluster in a metal binding pocket at the active site (18). Isothermal titration calorimetry and EPR have been used to demonstrate that Cce1 binds two magnesium or manganese ions (16) and may thus have a two-metal ion mechanism for hydrolysis of phosphodiester bonds, similar to many nucleases (26). Use of ITC to examine metal ion binding in the Cce1 mutants E145Q, D294N, and Q147A clearly shows that binding is abolished in the former two, but not in the latter, thus providing strong evidence that Glu 145 and Asp 294 (and probably by implication Asp 293 ) form part of the metal binding pocket in Cce1.
Basic residues in nucleases are required for binding the nucleic acid substrate, either through contacts with the phosphodiester backbone or by contacting specific bases in the major or minor grooves. A second role for basic residues is in catalysis; e.g. the active site residue Lys 92 in EcoRV is essential for activity and is thought to stabilize the negative charge developing on the pentavalent phosphorus in the transition state (27). Our mutagenesis studies of conserved basic residues in Cce1 appear to have uncovered examples of residues important in each of these roles. Notably, Arg 231 has been demonstrated to play an important role in junction binding, and the R231A mutant can be partially rescued by replacement with a lysine residue. In contrast, Lys 291 , which is close to the active site residues Asp 293 and Asp 294 , appears to have an important role in catalysis, and substitution with an arginine at this position fails to restore activity and is detrimental to binding affinity. The archaeal junction-resolving enzyme Hjc has recently been shown to possess a conserved nuclease domain found in type II restriction enzymes such as EcoRV as well as many other diverse nucleases. This domain includes three acidic residues that function as ligands for the catalytic metal ions and a conserved lysine residue (Lys 57 ) that is essential for catalysis. Mutation of any of these four residues in Hjc results in a severe decrease in catalytic competence without affecting junction binding (25). Given the proximity of Lys 291 in Cce1 to two acidic residues (Asp 293 and Asp 294 ) involved in binding catalytic metal ions, It is tempting to suggest that this lysine plays a similar role to Lys 57 in Hjc and Lys 92 in EcoRV (27), in the stabilization of the transition state.
Comparisons with Mrs1-The 29% sequence identity between Cce1 with Mrs1 has always been an intriguing observation, since at first sight the two proteins perform radically different functions. Cce1 is a structure-specific DNA endonuclease, while Mrs1 is thought to have a nonenzymatic function in stabilizing the tertiary conformation of a self-splicing group I intron (13). Presumably, one was recruited to perform the role of the other. The observation that Mrs1 appears specific to the genus Saccharomyces whereas Cce1 is widely distributed in the fungi suggests that Cce1 is the ancestral enzyme and was recruited to a role in RNA splicing. Structural studies of both the group I and group II introns suggest parallels with the stacked X structure of the four-way DNA junction (28 -31), and it is plausible to suggest that these similarities allowed recruitment of Cce1 as an RNA splicing cofactor or chaperone.
Given the limited sequence identity apparent from Fig. 1 for all four Cce1 and Mrs1 sequences, conserved residues are likely to have an important role in the structure or function of the two classes of protein. Significantly, these include two of the acidic residues (Glu 145 and Asp 294 ) shown by mutagenesis and ITC to be ligands for the catalytic magnesium ions. The question thus arises as to their role in Mrs1. If Mrs1 does bind one or more magnesium ions, what are their functional roles? Possibly, these metal ion(s) have assumed a structural role in the Mrs1 protein, but a more exciting possibility is that they play an active role either in the stabilization of the tertiary structure of the catalytic RNA or indeed participate directly in catalysis of RNA splicing. These questions are currently being addressed.