Substrate specificity of the SpCCE1 holliday junction resolvase of Schizosaccharomyces pombe.

SpCCE1 from Schizosaccharomyces pombe is an endonuclease that resolves Holliday junctions in vitro. SpCCE1 also binds and cleaves a range of other DNAs (Y-junction; flap; and flayed, nicked, and partial duplexes) with varying efficiency. Cleavage sites are always 3' of thymine nucleotides positioned at or close to the branch point or strand interruption. SpCCE1's favored substrate is the X-junction. Up to two dimers of SpCCE1 can bind concurrently to the same X-junction at its crossover point. From mixing experiments of SpCCE1 and the Escherichia coli RuvA protein, we show that each dimer of SpCCE1 binds to a different face of the X-junction and that both are seemingly competent for strand cleavage. We propose that this provides a mechanism whereby SpCCE1 can scrutinize all four junction strands simultaneously for cleavable thymine nucleotides. SpCCE1 appears to resolve X-junctions by a nick and counter-nick mechanism. Therefore, to ensure a high probability of bilateral strand cleavage, SpCCE1 has a relatively long lifetime on X-junctions. This mechanism has the drawback of limiting dissociation from noncleavable junctions. We discuss why this might not be a problem in vivo.

SpCCE1 from Schizosaccharomyces pombe is an endonuclease that resolves Holliday junctions in vitro. SpCCE1 also binds and cleaves a range of other DNAs (Y-junction; flap; and flayed, nicked, and partial duplexes) with varying efficiency. Cleavage sites are always 3 of thymine nucleotides positioned at or close to the branch point or strand interruption. SpCCE1's favored substrate is the X-junction. Up to two dimers of SpCCE1 can bind concurrently to the same X-junction at its crossover point. From mixing experiments of SpCCE1 and the Escherichia coli RuvA protein, we show that each dimer of SpCCE1 binds to a different face of the X-junction and that both are seemingly competent for strand cleavage. We propose that this provides a mechanism whereby SpCCE1 can scrutinize all four junction strands simultaneously for cleavable thymine nucleotides. SpCCE1 appears to resolve X-junctions by a nick and counter-nick mechanism. Therefore, to ensure a high probability of bilateral strand cleavage, SpCCE1 has a relatively long lifetime on X-junctions. This mechanism has the drawback of limiting dissociation from noncleavable junctions. We discuss why this might not be a problem in vivo.
Homologous recombination plays a central role in DNA metabolism; e.g. it provides a mechanism for the faithful repair of double strand breaks, facilitates the removal of lesions in single strand gaps, ensures the completion of DNA replication by helping to reform collapsed replication forks, and functions during meiosis to promote the correct segregation of homologous chromosomes. In addition, the genome rearrangements that are a consequence of recombination are an important component in the generation of genetic diversity.
A key intermediate of the recombination reaction is the Holliday junction, which consists of two double-stranded DNA molecules linked by a reciprocal crossover of single strands. Cleavage of a pair of symmetrical strands at the Holliday junction by a junction-specific endonuclease and sealing of the nicks by a DNA ligase generates classical "patch" or "splice" recombinant products. To date, six junction-resolving enzymes have been cloned representing eubacteria (RuvC), bacteriophage (T4 endonuclease VII, T7 endonuclease I, and lambdoid RusA), and eucarya (CCE1 from Saccharomyces cerevisiae and SpCCE1 from Schizosaccharomyces pombe) (1)(2)(3)(4). The detection of junction-specific endonuclease activities from eucaryal viruses and mammalian cells provides further evidence of the ubiquitous nature of resolvases (1).
The known junction resolvases share little or no amino acid sequence similarity. Even CCE1 from S. cerevisiae exhibits only weak sequence identity with its homologue from S. pombe, SpCCE1 (2)(3)(4). Despite this, the resolvases share certain physical and mechanistic attributes; e.g. they are mostly small basic proteins (1), bind to DNA in a structure-specific manner (2)(3)(4)(5)(6)(7)(8)(9)(10)(11), bind to junctions as dimers (3, 10 -13), distort the global conformation of the junction upon binding (9, 11, 14 -16), and resolve Holliday junctions by introducing a pair of symmetrically related nicks in strands of the same polarity at or close to the junction crossover point (2-4, 7, 8, 10, 17-20). Aside from these common properties, there are further features that divide resolvases into two main groups. Group I enzymes, which include RuvC, RusA, CCE1, and SpCCE1, are distinguished by their high level of selectivity for Holliday junctions and by a degree of sequence specificity at the resolution step (2, 3, 5-8, 10, 13, 21, 22), whereas group II enzymes, which include T4 endonuclease VII and T7 endonuclease I, have more relaxed structural and sequence requirements for DNA cleavage; e.g. T4 endonuclease VII cleaves various branched DNAs (17,23), mismatches (24), heteroduplex loops (25), and lesions in DNA (26) as well as Holliday junctions. In accord with these in vitro properties, group I enzymes appear to play specialized roles in recombination, whereas group II enzymes act both in recombination and in the removal of branch structures from phage DNA prior to packaging.
SpCCE1 (or YDC2) is the most recently cloned and characterized Holliday junction resolvase (2)(3)(4). Its activity was originally identified in fractionated extracts from S. pombe, but it was cloned based on its sequence identity to CCE1 from S. cerevisiae. The 30.2-kDa SpCCE1 protein shares 28% sequence identity and approximately 50% sequence similarity with the 41-kDa CCE1 protein. Both enzymes cleave X-junctions 3Ј of thymine nucleotides positioned at or just 3Ј of the crossover point. However, an observed difference in preferred cut sites within a synthetic Holliday junction indicates that they do have different sequence specificities for cleavage (3).
The likeness between SpCCE1 and CCE1 indicates that they may perform similar functions in vivo. In accord with this, both enzymes are encoded within the nucleus but localize to mitochondria (27). 1 Furthermore, mutations in both CCE1 and SpCCE1 have no observable effect on nuclear recombination (28). 1 In cce1 mutants, mitochondrial DNA is aggregated seemingly due to an accumulation of recombination junctions linking the DNAs together (29). As a consequence of this, mitochondrial DNA segregation is effected and results in an increase in the frequency of petite cells (28,29). We are currently investigating whether our Spcce1 mutant has similar phenotypes.
In this work, we concentrate on the in vitro properties of SpCCE1, extending our and others' previous characterization of the protein. In particular, we focus on how SpCCE1 recognizes and targets Holliday junctions, a question that remains largely unanswered for any resolvase. We also examine how SpCCE1 ensures that junctions are resolved properly by the introduction of two symmetrical strand incisions.

MATERIALS AND METHODS
Enzymes and Reagents-RuvA was a gift from Robert Lloyd (University of Nottingham) and was purified as described (30). SpCCE1 was purified as described below. Amounts of both proteins were estimated by a modified Bradford assay using a Bio-Rad protein assay kit and bovine serum albumin (Amersham Pharmacia Biotech) as the standard. Concentrations of these proteins are expressed in moles of monomers. T4 polynucleotide kinase was from Boehringer Mannheim, and [␥-32 P]ATP was from Amersham Pharmacia Biotech. All other reagents were from Sigma, BDH, and NBL Gene Sciences Ltd. and were of analytical grade.
Overexpression of SpCCE1-1-Liter batches of BL21 (DE3) plysS (32) transformed with pMW206 (2) were grown with aeration at 37°C in Luria-Bertani broth containing 100 g/ml ampicillin and 20 g/ml chloramphenicol. At a cell density corresponding to an A 600 of 0.5, SpCCE1 was induced by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM and incubated for a further 3 h. The cells were then harvested by centrifugation, resuspended in 30 ml of lysis buffer (100 mM Tris-HCl, pH 8.0, 2 mM EDTA, 5% glycerol), and stored at Ϫ80°C until required. Under these conditions, SpCCE1 was induced to approximately 5% of the total cell protein.
Purification of SpCCE1-The purification of SpCCE1 was monitored throughout by SDS-polyacrylamide gel electrophoresis analysis and by assaying for junction cleavage activity using X-12. Frozen cells (120 ml) from 4 liters of induced culture were thawed at room temperature and then mixed on ice with 30 ml of 5 M NaCl, 0.75 ml of 200 mM DTT, 1.5 ml of 10% Triton X-100, and 0.75 ml of 200 mM phenylmethylsulfonyl fluoride. Lysis was completed by a single passage of the mixture through a French pressure cell at 30,000 p.s.i., and the cell debris was removed by centrifugation at 19,000 rpm for 60 min at 4°C using a Beckman JA20 rotor. Approximately 60% of the SpCCE1 was removed with the cell debris. Following lysis, all steps were performed at 4°C. The soluble lysate fraction (ϳ150 ml) was dialyzed against 4 liters of buffer A (Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM DTT, 10% glycerol) containing 0.5 M KCl and 1 mM phenylmethylsulfonyl fluoride. Following centrifugation at 15,000 rpm for 10 min to remove traces of insoluble material, the dialyzed sample was applied to a phosphocellulose P11 (Whatman) column (30-ml bed volume) equilibrated with buffer A containing 0.5 M KCl. The column was washed with 205 ml of the same buffer before eluting bound proteins (including SpCCE1) with 150 ml of buffer A containing 0.7 M KCl. Eluted SpCCE1 was dialyzed against 4 liters of buffer A containing 0.3 M KCl for 3 h and applied to a doublestranded DNA cellulose (Sigma) column (8-ml bed volume) equilibrated with the same buffer. The column was then washed with 64 ml of the same buffer, and bound protein was eluted with 40 ml of buffer A containing 0.45 M KCl. The eluted protein was dialyzed against 2 liters of buffer A containing 0.2 M KCl for 3 h and applied to a prepacked 1-ml Mono S column (Amersham Pharmacia Biotech), which was then washed with 10 ml of the same buffer. Bound protein was eluted with a linear gradient from 0.2 to 0.6 M KCl in 8 ml and then by a further 6 ml of buffer A containing 0.6 M KCl. SpCCE1 eluted in the 0.6 M KCl column fractions. The peak of these fractions (3 ml) was pooled and applied in 0.5-ml batches to a prepacked Superose 12 HR gel filtration column (Amersham Pharmacia Biotech). The six column runs were each developed with 25 ml of buffer A containing 0.1 M KCl. The majority of SpCCE1 eluted as a single peak with a molecular mass determined to be 33 kDa by comparison with protein standards (Bio-Rad) analyzed using the same buffer. However, a significant proportion of the protein eluted in later fractions, suggesting a tendency for SpCCE1 to stick to the Superose 12 column. The peaks of SpCCE1 from the six Superose 12 column runs were pooled, dialyzed against protein storage buffer (Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM DTT, 0.1 M KCl, 50% glycerol), and stored in aliquots at Ϫ80°C. The yield was approximately 2 mg of SpCCE1 (in 10 ml). SpCCE1121RWGTP, which will be described in greater detail elsewhere, was purified from 1 liter of induced cells using a scaled down version of the above strategy.
Oligonucleotides-Oligonucleotides were synthesized on an Applied Biosystems 380B DNA synthesizer using cyanoethyl chemistry. Each oligonucleotide was deprotected, precipitated in ethanol, and purified on a 12% (w/v) polyacrylamide gel containing 7 M urea. The bands containing full-length oligonucleotides were cut out and extracted from the gel by soaking in water overnight.
DNA Substrates-DNA substrates were made by annealing combinations of oligonucleotides as indicated in Table I following the procedures of Parsons et al. (33). Prior to annealing, DNA substrates were labeled at the 5Ј-end of one of their component oligonucleotides as indicated using [␥-32 P]ATP and polynucleotide kinase. Annealed substrates were purified by nondenaturing electrophoresis on 6% polyacrylamide gels followed by electroelution. The concentration of DNA substrates was estimated by relating the specific activity of the labeled oligonucleotide to the activity of the purified substrate and is expressed in molar concentrations of DNA substrate.
Binding Assays-Reaction mixtures (20 l) contained labeled substrate DNA in binding buffer (25 mM Tris-HCl, pH 8.0, 1 mM DTT, 100 g/ml bovine serum albumin, 6% glycerol) containing NaCl (200 mM), EDTA (5 mM), and MgCl 2 (10 mM) as indicated. Reactions were started typically by the addition of protein, were incubated on ice or at 37°C as indicated, and then were loaded onto a 4% native polyacrylamide gel in low ionic strength buffer (6.7 mM Tris-HCl, pH 8.0, 3.3 mM sodium acetate, 2 mM EDTA). Competitor DNA (poly(dI⅐dC)-poly(dI⅐dC)) (Amersham Pharmacia Biotech) added during the course of some reactions was mixed in by pipetting. Electrophoresis of gels was typically for 2 h at 160 V. For time course experiments assessing binding stabilities, samples were loaded onto gels already running at 200 V. Following loading of all of the samples, the voltage was reduced to 160 V, and electrophoresis continued for a further 1 h and 20 min. For all experiments, both buffer and gel were precooled at 4°C, and the electrophoresis was at room temperature with continuous buffer recirculation. Gels were dried on 3 MM Whatman paper, quantified using a model SF PhosphorImager and ImageQuant software (Molecular Dynamics), and autoradiographed.
Junction Cleavage Assays-Reaction mixtures (20 l) contained labeled substrate DNA in binding buffer containing NaCl (200 mM), MgCl 2 (10 mM), and protein as indicated. After incubation at 37°C typically for 20 or 30 min, reactions were terminated by adding one-fifth volume of stop mix (2.5% SDS, 200 mM EDTA, 10 mg/ml proteinase K) and incubating for a further 10 min at 37°C to deproteinize the mixture. Products were analyzed by electrophoresis through 10% native polyacrylamide gels at 190 V using a standard Tris borate buffer system, or 11% denaturing gels containing 7 M urea. For native gels, deproteinized reactions were loaded directly onto the gel. For denaturing gels, reactions were extracted with phenol/chloroform/isoamyl alcohol (25:24:1), and the DNA was precipitated with ethanol, resuspended in gel loading buffer (0.05% (w/v) bromphenol blue, 0.05% (w/v) xylene cyanol, 10 mM EDTA, pH 7.5, 97.5% (v/v) formamide), and denatured by boiling for 2 min before loading onto the gel. To map the sites of cleavage, detected by denaturing gel analysis, Maxam-Gilbert sequencing ladders of each labeled oligonucleotide were used (34). A 1.5-base allowance was made to compensate for the nucleoside eliminated in the sequencing reaction. Gels were dried on 3 MM Whatman paper, quantified using a model SF PhosphorImager and ImageQuant software (Molecular Dynamics), and autoradiographed.
In Situ 1,10-Phenanthroline-Copper Footprinting-Binding reactions (40 l) containing 86 ng of SpCCE1 and 32 ng of X-0, labeled in one of its four strands, were set up, in binding buffer containing 2 mM EDTA and 200 mM NaCl, in parallel with control reactions containing no protein. Following incubation on ice, binding reactions were run on a 6% nondenaturing polyacrylamide gel in low ionic strength buffer. Electrophoresis was at 4°C for 2 h at 160 V with continuous circulation of buffer. X-0 and SpCCE1-X-0 complexes were detected by autoradiography at 4°C, and the appropriate bands were excised from the gel, chopped into evenly sized small pieces, and immersed in 100 l of 50 mM Tris-HCl (pH 8.0). 10 l of OP-Cu mix (2 mM 1,10-phenanthroline and 0.45 mM CuSO 4 ) and 10 l of 58 mM mercaptoproprionic acid were then added, and the mixture was incubated at room temperature for 10 min. The reaction was stopped by the addition of 20 l of 28 mM 2 The abbreviation used is: DTT, dithiothreitol.
2,9-dimethyl-1,10-phenanthroline (35). 270 l of 0.5 M ammonium acetate, 1 mM EDTA was then added, and the DNA was eluted from the gel by incubation overnight at 37°C with agitation in a thermomixer (Eppendorf). The DNA was then precipitated with ethanol, washed twice with 70% ethanol, resuspended in loading dye (98% formamide, 10 mM EDTA, 0.05% xylene cyanol, 0.05% bromphenol blue) plus 10 mM NaOH, and electrophoresed on a 15% polyacrylamide sequencing gel containing 7 M urea. Maxam-Gilbert sequencing ladders of each labeled oligonucleotide were used to map the protected sites. Gels were dried and analyzed using a model SF PhosphorImager and ImageQuant software (Molecular Dynamics) and by autoradiography.

RESULTS
Purification of SpCCE1-Previously, we have described the purification of the SpCCE1 protein of S. pombe as a histidinetagged recombinant from Escherichia coli cells (2). In the present study, we have developed a new strategy for purifying SpCCE1 without a histidine tag from E. coli cells (see "Materials and Methods"). This yielded approximately 2 mg of SpCCE1 from 4 liters of induced cells at Ͼ95% homogeneity as judged by Coomassie Blue, silver, and Sypro Orange staining of SDS gels ( Fig. 1 and data not shown). This relatively poor yield of protein was due to the insolubility of the majority of the induced protein. Subsequent studies have shown that inducing cells at 25°C greatly improves the solubility of SpCCE1, resulting in much higher yields of protein using the same purification strategy (data not shown). For the studies presented here, we also purified a mutant form of SpCCE1 containing a five-amino acid (RWGTP) insertion after the tyrosine at position 121 using the same strategy as for wild-type SpCCE1 (Fig.  1, lane c). SpCCE1121RWGTP binds to Holliday junctions like wild-type protein but has little cleavage activity. 1 Holliday Junction Binding by SpCCE1-A number of proteins, including SpCCE1, have been described that bind to Holliday junctions in a structure-specific rather than sequencespecific manner. To provide some insight into how this is achieved by SpCCE1, we began to characterize its interaction with model Holliday junctions in vitro. For these studies, we used two different but sequence-related Holliday junctions. X-12 is composed of four 50 -52-base-long oligonucleotides that anneal to form an X-junction with a central 12-base pair core of homology in which the junction point is free to branch migrate. X-0 shares one common oligonucleotide with X-12 but has no central core of homology, so its junction point is fixed. Binding of SpCCE1 to both X-12 and X-0 is readily detected by a band shift assay ( Fig. 2A). When increasing concentrations of SpCCE1 are incubated with either X-12 or X-0 in the presence of 200 mM NaCl and a buffer containing 6% glycerol, two protein-DNA complexes are formed. Complex 1 is formed by the binding of a dimer of SpCCE1 (lanes b and c and lanes g and h), and complex 2 results from the binding of two dimers of SpCCE1 (lanes d-f and i-k) (2,3). The omission of glycerol from the binding buffer dramatically reduces the formation of complex 2 (Ref. 3 and data not shown). Furthermore, the half-life of complex 2 is considerably shorter than that of complex 1 ( Fig.  9 and data not shown). These data show that complex 2 is less stable than complex 1 and indicate that the binding of a second dimer of SpCCE1 to an X-junction is hindered by the dimer showing SpCCE1 binding to X-12 and X-0 junctions. Reaction mixtures (20 l) contained the indicated amounts of SpCCE1 with 0.9 nM of either 32 P-labeled X-12 or X-0 in binding buffer plus 200 mM NaCl. Reactions were incubated on ice for 10 min before loading onto a 4% polyacrylamide binding gel as described under "Materials and Methods." B, in situ 1,10-phenanthroline-copper footprinting of complex 1 of SpCCE1 with X-0 isolated from a 6% polyacrylamide binding gel. Reactions were run on a 15% sequencing gel beside G ϩ A sequencing ladders (see "Materials and Methods"). Bars beside ϩ SpCCE1 lanes indicate regions of protection as determined by PhosphorImager analysis. C, schematic of X-0 showing regions protected by SpCCE1 (shaded). that is already bound and/or that binding occurs at a site that is not equivalent to that of the first dimer.
Footprinting the SpCCE1-X-junction Complex-In situ 1,10phenanthroline copper footprinting was used to map where SpCCE1 was binding on X-junction DNA. X-0 was used for these studies so that the position of the junction crossover point would be fixed and therefore known precisely. SpCCE1-X-0 junction complexes were excised from band shift gels containing 2 mM EDTA like those shown in Fig. 2A and reacted with the 1,10-phenanthroline copper reagent. Four reactions were performed in parallel, each containing X-0 labeled in a different strand. The products of these reactions were analyzed on a denaturing gel (Fig. 2B). Regions of protection from attack by the 1,10-phenanthroline copper reagent were detected in all four strands of X-0, which was bound by SpCCE1 to form complex 1. These were located at the point of strand crossover in X-0 and extended 10 -13 base pairs along each of the four flanking DNA arms (C). The same pattern of protection was obtained when complex 2 was footprinted (data not shown). However, as noted above, the SpCCE1-X-0 complex 2 is less stable than complex 1, and therefore it is possible that dissociation of SpCCE1 from X-0 to form complex 1 could have occurred during the footprinting reaction. In the absence of metal ions, electrostatic repulsion between the four arms of an X-junction forces them into a 4-fold symmetrical arrangement in which each arm subtends an angle of 90°with its neighbor (36). The pattern of protection obtained when SpCCE1 is bound to X-0 is also approximately 4-fold symmetrical, suggesting that SpCCE1 sits squarely on the junction center. However, this is not what would be expected for the interaction of a single dimer of SpCCE1 with X-0, and we suspect that the footprint reflects the average of two different SpCCE1-X-0 complexes. In agreement with this, the amount of protection in each of the four strands is approximately 50%, suggesting that in only half of the junction molecules was SpCCE1 binding at that site.
A Dimer of SpCCE1 Can Bind to an X-junction That Is Already Bound by a Tetramer of RuvA-To provide further clues as to how SpCCE1 interacts with X-junction DNA, we investigated what effect the E. coli RuvA protein might have on SpCCE1's ability to bind to X-12 and X-0. RuvA forms a square planar tetramer onto which an X-junction can lie with its arms held in a square planar configuration (Fig. 3A) (37)(38)(39). A second tetramer of RuvA can bind to the opposite face of the X-junction such that the DNA becomes fully encased in an octameric shell of RuvA protein (Fig. 3C). An octamer of RuvA would be expected to exclude binding to the center of the X-junction by other junction-specific binders. However, a single tetramer of RuvA leaves one face of the X-junction exposed to binding by other proteins. Binding here would be possible as long as the protein could recognize the open square planar configuration of the X-junction and interact principally with only one face of the structure. Previously it has been shown that a dimer of RuvC can bind to an X-junction that is already bound by a tetramer of RuvA (40). However, the same does not appear to be true for RusA. 3 Both RuvC and RusA manipulate the structure of the X-junction into an extended, unstacked configuration upon binding to it (14,16). However, whereas the RuvC-X-junction structure is approximately square planar with 2-fold symmetry (14) (i.e. not too dissimilar to the structure imposed by RuvA), RusA imposes a very different structure that has been interpreted as a tetrahedral arrangement of junction arms (16).
To see if SpCCE1 is capable of binding to an X-junction that is bound by RuvA, we incubated X-12 with three different concentrations of RuvA, added SpCCE1 to each of these, and analyzed the reactions on a band shift gel (Fig. 4A). In the absence of SpCCE1, RuvA formed two complexes with X-12. Complex 1 (lanes b and f) is formed by the binding of a single tetramer of RuvA to X-12, and complex 2 (lanes b, f, and j) is formed by the binding of an octamer of RuvA. When SpCCE1 was titrated into the binding reaction, the remaining free X-12 was bound to form SpCCE1-X-junction complexes 1 and 2 (lanes c-e and k-p). In addition, we observed the formation of a novel protein-DNA complex whose mobility was intermediate between RuvA complexes 1 and 2 (lanes c-e, g-i, and k-m). This novel complex was not detected when either RuvA or SpCCE1 was absent (lanes b, f, j, and n-p) and appears to form at the expense of both RuvA complexes. Its mobility is consistent with the binding of a tetramer of RuvA and a dimer of SpCCE1 to the same junction. A further experiment was performed where X-12 was incubated with SpCCE1 before the addition of increasing concentrations of RuvA (Fig. 4B). We chose a concentration of SpCCE1, which shifted all of the X-junction into SpCCE1 complexes 1 and 2 in the absence of RuvA (lane b). When RuvA was titrated into the binding reaction, the novel complex was again detected and increased with increasing concentrations of RuvA at the expense of both SpCCE1 complex 1 and 2 such that the only band detectable at the highest concentrations of RuvA was that of the novel complex (lane h). In none of our experiments could we detect any other complexes that were dependent on the presence of both SpCCE1 and RuvA; therefore, we assume that SpCCE1 cannot bind to the RuvA octamer X-junction complex and that only one dimer of SpCCE1 can bind to an X-junction that is already bound by a tetramer of RuvA. Our data are therefore most consistent with the models in Fig. 3, which predict that (i) dimers of SpCCE1  bind to opposite faces of an X-junction (E) and (ii) a tetramer of RuvA readily replaces one dimer of SpCCE1 to form a complex in which the junction is sandwiched between a tetramer of RuvA and a dimer of SpCCE1 (D).
Specificity of Substrate Binding by SpCCE1-SpCCE1 has a strong binding specificity for X-junction DNA (2)(3)(4). This presumably reflects its ability to recognize some structural feature of this type of DNA. To gain a better understanding of what this might be, we constructed a range of DNA substrates of decreasing complexity (Table I) and tested which ones were bound by SpCCE1 using the band shift assay. Binding was tested both in the presence and absence of 200 mM NaCl (Fig. 5). Discrete retarded species were detected with X-0 (lanes 2-13), Y-junction (lanes [15][16][17][18][19][20][21][22][23][24][25][26], flayed duplex (lanes 28 -37), and linear duplex DNAs (lanes 42-49) but not with single-stranded DNA (lanes 39 -40). The binding of the flap substrate was essentially the same as the flayed duplex, whereas the binding of nicked and partial duplex substrates was very similar to that of the linear duplex (data not shown). For each of the bound substrates, bands with similar mobilities to complexes 1 and 2 formed with X-junction DNA were evident. However, in the case of linear duplex, nicked duplex, and partial duplex DNA, very little of complex 1 was observed relative to the amount DNA migrating at the position of complex 2 (lanes 42-49 and data not shown). In the absence of NaCl, additional complex bands were observed with each of the bound substrates compared with the equivalent reactions containing NaCl. Bands with intermediate mobility to complexes 1 and 2 probably represent monomer SpCCE1 binding to naked or dimer-bound DNA respectively, and the slower migrating species observed with X-0 and Y-junction at higher concentrations of SpCCE1 represent additional monomers and dimers binding to junctions already bound by two dimers of SpCCE1 (lanes 6 and 7  and lanes 19 and 20). The low abundance of these complexes and their susceptibility to NaCl indicates that monomers of SpCCE1 bind less stably to DNA than dimers. Quantitation of gels as in Fig. 5 enabled us to estimate apparent dissociation constants (K D values) for the interactions (Table II). These data show that SpCCE1's binding affinity for X-and Y-junctions is very similar and, in the absence of NaCl, approximately 2-3fold better than for flap and flayed duplex and 6 -11-fold better than for linear, nicked, and partial duplex DNAs. In the presence of NaCl, binding to X-and Y-junctions is improved slightly, whereas binding to flap, flayed duplex, and linear duplex substrates is reduced. Under these conditions, binding to flap and flayed duplex is approximately 10 -12-fold less and binding to linear substrates is 23-35-fold less than for X-and Y-junctions. From these data, we conclude that the requirements for efficient DNA binding by SpCCE1 are satisfied almost as well by a three-way (Y-) junction as by a four-way (X-) junction. The minimal substrates bound by SpCCE1 were the linear duplex, nicked duplex, and partial duplex DNAs, but significantly better levels of binding were observed with flap and flayed duplex DNA, indicating a preference for branch points.
Specificity of Substrate Cleavage by SpCCE1-Previously, we and others have shown that SpCCE1, in the presence of Mg 2ϩ , cleaves X-junctions by cutting 3Ј of thymine nucleotides positioned predominantly at or one nucleotide 3Ј from the point of strand crossover (2)(3)(4). Other DNAs such as a three-strand junction and a Y-junction were not substrates for cleavage. With the availability of highly active non-histidine-tagged SpCCE1 and the observation that SpCCE1 binds to substrates other than Holliday junctions, we have investigated further SpCCE1's ability to cleave different DNAs. The substrates detailed in Table I were incubated with increasing quantities of SpCCE1 in buffer containing 10 mM MgCl 2 . Following deproteinization of reaction mixtures, the products of each reaction were analyzed on nondenaturing polyacrylamide gels (Fig. 6). In accord with our previous results, SpCCE1 cleaved both X-0 and X-12 to generate nicked linear duplex products (A, cleavage product 2, lanes b-d and g-i) and failed to cleave YB, doublestranded, and single-stranded DNAs (data not shown). However, contrary to expectation, we also detected cleavage products with YA (lanes l-n), flayed duplex A and B (lanes q-s and data not shown), flap DNA (B, lanes b-d), nicked duplex DNA (B, lanes g-i), and partial duplex DNA (B, lanes l-n). These included products that migrated approximately at the position of nicked duplex DNA (cleavage product 2) for YA, flap, and flayed duplex substrates; cleavage product 1 for YA; and cleavage product 3 for YA, flap, nicked duplex, and partial duplex substrates. Surprisingly, we also observed cleavage products 1 and 3 with both X-12 and X-0 substrates. With all of these substrates, relatively little or no cleavage was detected using the SpCCE1121RWGTP mutant protein that binds to DNA but is catalytically disabled (A, lanes e, j, o, and t; B, lanes e, j, and  o).
To determine the way in which SpCCE1 was cleaving these DNA substrates, we mapped all of the strand cleavages using denaturing gels and related these to the cleavage products detected on native gels. A summary of this data is presented in FIG. 4. Binding to X-12 by SpCCE1 and RuvA. A, RuvA was incubated with 1.6 nM 32 P-labeled X-12 in binding buffer plus 200 mM NaCl (20-l reaction volume) for 5 min on ice. SpCCE1 (2 l) in binding buffer was then added, and incubation continued for a further 5 min on ice. Following incubation, samples (12 l) from each reaction were loaded onto a 4% polyacrylamide binding gel and electrophoresed for 2.5 h at 160 V. See "Materials and Methods" for further details. B, as A except SpCCE1 was incubated with 1.1 nM 32 P-labeled X-12 in binding buffer plus 200 mM NaCl (20-l reaction volume) for 5 min at 37°C. RuvA (2 l) in binding buffer was then added, and incubation continued for a further 5 min at 37°C. Fig. 7 and Table III. Both X-12 and X-0 were cleaved at the sites that had been determined previously (2). All cleavage sites were 3Ј of thymine nucleotides positioned within the mobile core of X-12 or up to two nucleotides from the crossover point in X-0. The majority of strand cleavages in X-12 occurred in pairs to liberate two linear nicked duplex products (cleavage product 2). However, a small percentage (3.2%) of X-12 was cleaved across a single junction arm to liberate a small linear duplex product (cleavage product 3) and a nicked Y product (cleavage product 1). Such aberrant resolution appears to come predominantly from cleavage at a T-A and an A-T step within the homologous core of X-12 (Fig. 7, sites 1-4 and 5-8). Sites of cleavage on each strand of X-0 were detected. However, the majority of strand cleavage was in oligonucleotide 2, which was cleaved at a level equivalent to the amount of resolution of X-12 (Fig. 7, site 3, and Table III). The less frequent cleavage of oligonucleotides 5-7 in X-0 probably reflects both a sequence and a positional preference for cleavage by SpCCE1. It is apparent from these data that strand cleavages in X-0 occur independently of each other. However, combinations of two Each substrate is made from the oligonucleotides indicated by number on each respective schematic. The number is positioned at the 5Ј-end of its respective oligonucleotide. YA, flap, flayed duplex A and B, nicked duplex, and partial duplex DNA substrates, which are each composed from elements of the X-0 structure, are cleaved at sites equivalent to those cleaved in X-0 (Fig. 7). One site that is cleaved in all of these substrates, except flayed duplex B, is site 3 in oligonucleotide 2. A comparison of the amount of cleavage at this site in each of the substrates after a 30-min reaction shows that it is cleaved most efficiently within the X-0 structure with approximately 2-fold more cleavage than in the flap substrate, 7-10-fold more cleavage than in YA or flayed duplex A, and 40 -80-fold more cleavage than in either the nicked or partial duplex DNA substrates (Table III). However, much larger differences in cleavage efficiencies were apparent when the rates of cleavage were measured (data not shown). Furthermore, the addition of 200 mM NaCl stimulated cleavage of X-12 and X-0 but generally reduced cleavage of all of the other substrates (data not shown). From this we conclude that, although SpCCE1 is capable of cleaving a range of different DNAs, under physiological levels of salt it displays a high degree of specificity for cleaving X-junctions.
Flayed duplex B is unlike the other substrates that are composed from parts of X-0 because it is not cleaved at site 3 in oligonucleotide 2; instead, it is cleaved at site 2 in oligonucleotide 8 (Fig. 7). Why site 2 is not cleaved in flayed duplex B when it is a favored cleavage site in each of the other substrates is not certain. One possibility is that SpCCE1 binds principally to duplex DNA and only makes incisions in strands that are adjacent to its binding site; i.e. binding and cleavage sites have to be separated by a branch point or strand interruption.
Aberrant Resolution of X-junctions by SpCCE1-Our working model for SpCCE1's interaction with X-junction DNA predicts that a dimer of SpCCE1 locates on the junction such that each monomer subunit's active site can cleave diametrically opposed junction strands (Fig. 3B). If this is correct, then cleavage of adjacent strands necessary to give cleavage products 1 and 3 from X-12 and X-0 should be impossible for a single dimer to achieve in one reaction. Cleavage products 1 and 3 could have arisen from two rounds of SpCCE1 junction binding and unilateral strand cleavage. To test this idea, we preincubated SpCCE1 with X-12 or X-0 at 37°C so that all of the junction was bound into complex 1 and 2 as determined on a band shift gel (data not shown). After 5 min, MgCl 2 was added to initiate strand cleavage. Samples were then taken at timed intervals, deproteinized, and analyzed on native gels (Fig. 8, A and B). As expected, cleavage of both X-12 and X-0 yielded the three products seen earlier in Fig. 6, which each accumulated over time (A and B, lanes b-f). We reasoned that if cleavage products 1 and 3 were derived from two rounds of binding and cleavage, then the effective removal of SpCCE1 from the reaction, following its dissociation from the junction, by adding an excess of competitor DNA at the same time as the MgCl 2 , should inhibit product 1 and 3 formation by preventing a second round of binding and cleavage. However, the addition of sufficient competitor DNA to abolish detectable binding (data not shown) and greatly reduce strand cleavage if added before preincubation of SpCCE1 with either X-12 or X-0 (A and B, lane m), did not significantly reduce the amount of cleavage products 1 and 3 when added at the same time as the MgCl 2 (A and B, lanes g-l). These data indicate that cleavage products 1 and 3 do not arise from two separate rounds of dimer binding and unilateral strand cleavage.
One way in which cleavage products 1 and 3 could appear to be derived from a single binding reaction is if the strand cleavages are made by two separate monomers or dimers of SpCCE1 bound concurrently to the same junction. We can rule out FIG. 6. Structure specificity of DNA cleavage by SpCCE1. A and B, reactions (20 l) contained SpCCE1 or SpCCE1121RWGTP (120 nM) and 0.5 nM substrate DNA (each 32 P-labeled in oligonucleotide 2) in binding buffer plus 10 mM MgCl 2 as indicated. Following incubation at 37°C for 30 min, reactions were stopped, deproteinized, and run on a 10% polyacrylamide gel as described under "Materials and Methods." Most of the substrates contain a background of unannealed and partially annealed oligonucleotides liberated during its preparation or reaction. In most cases, these are not observed to alter significantly upon the addition of protein and are therefore regarded to have little if any effect on the overall reaction. However, in some cases (e.g. the nicked and partial duplexes), we observe a reduction in the amount of free oligonucleotide in the presence of protein. This is due to the protein binding and stabilizing the substrate during the course of the reaction.  Fig. 5 were quantitated by PhosphorImager analysis, and the percentage of DNA bound for each concentration of SpCCE1 was plotted against the logarithm of the protein molarity (data not shown). From these graphs, the apparent dissociation constants (K D ) were estimated. Average K D values from two or three separate experiments are shown. b NB, no binding was detected using either oligo 2 or 5 up to 150 nM SpCCE1.
monomers, since products 1 and 3 are not reduced by NaCl concentrations that markedly diminish monomer junction binding (data not shown). If dimers are responsible, then RuvA bound to one face of the X-junction, excluding one dimer of SpCCE1 from binding to that junction (see Fig. 4), should inhibit their formation. To test this idea, we first incubated X-12 and X-0 with enough SpCCE1 to bind all of the DNA into complexes 1 and 2. Increasing amounts of RuvA were then added to form the RuvA-SpCCE1-X-junction complex. Finally, MgCl 2 was added to initiate strand cleavage, and after 5 min the reactions were stopped and analyzed on native gels (Fig.  8C). In the absence of RuvA, both X-12 and X-0 were cleaved to give products 1-3 (lanes b and h). The addition of just 20 nM RuvA to the X-12 reaction abolished cleavage product 1 and 3 formation but only reduced cleavage product 2 formation by approximately half (lane c). The addition of more RuvA up to a maximum of 200 nM made little difference to this result (lanes d-f), indicating that the level of cleavage product 2 formation is that obtainable from the RuvA-SpCCE1-X-12 complex. In the case of X-0, the addition of RuvA abolished the formation of all cleavage products (lanes i-l). Denaturing gel analysis showed that RuvA reduced cleavage in oligonucleotide 2 by about 90% (data not shown). These data indicate that, at least in the case of X-12, SpCCE1 is capable of resolving X-junctions that are bound by a single tetramer of RuvA. They are also consistent with the idea that two dimers of SpCCE1 bound simultaneously to the same junction are required for the formation of cleavage products 1 and 3. However, the reduction in cleavage product 2 by RuvA was not expected. In the case of X-12, this could be due to RuvA fixing the junction crossover point at noncleavable or poorly cleavable sites. However, this does not explain why X-0 cleavage product 2 is abolished. An alternative explanation is that SpCCE1 cleaves some sites by interaction with one face of the junction, while other sites are cleaved via the opposite face of the junction. One face of the open square planar X-junction displays major grooves adjacent to the crossover point, whereas its other face displays minor grooves. At least for one X-junction, RuvA favors binding to the major groove face (38). If the same is true for X-0 and X-12, RuvA  Table III.  Following incubation at 37°C for 30 min, reactions were stopped and deproteinized, and half the mixture was run on a 10% polyacrylamide gel and the other half on a 11% denaturing gel. Strand cleavage was determined from quantification of the denaturing gel.
c Determined from analysis of cleavage products on a nondenaturing gel.
should selectively abolish cleavages made via the major groove face of each junction. This may be why cleavage of oligonucleotide 2 in X-0 is greatly reduced by RuvA. Denaturing gel analysis of reactions like in Fig. 8C shows that cleavages made at sites 5 and 6 in X-12 are reduced twice as much as those at sites 1-4 by RuvA (data not shown). This may mean that SpCCE1 cleaves sites 5 and 6 via a particular junction face. However, sites 1-4 were inhibited the same as each other even when RuvA was preincubated with X-12 before the addition of SpCCE1 (data not shown). For these data to be consistent with our hypothesis, either RuvA does not favor binding to a particular face of X-12 and therefore does not selectively inhibit cleavages, or SpCCE1 binds to either face without preference and in so doing forces RuvA to dock on whichever face is free. Further analysis will be required to test these ideas.
Stability of the SpCCE1-X-junction Complex-SpCCE1's ability to make unilateral strand incisions in certain DNA substrates (see above) provides clear evidence that its two active sites function independently. Therefore, to ensure a high probability of Holliday junction resolution as opposed to unilateral nicking, the lifetime of the SpCCE1-X-junction complex must be long enough to provide ample time for two strand cleavages to be made before dissociation. We tested this prediction by monitoring the dissociation of SpCCE1 from X-12 and X-0 under a range of different conditions. SpCCE1 was incubated with X-junction at 37°C in the presence of 200 mM NaCl. After 5 min, an excess of unlabeled double-stranded DNA was added to act as a sink for protein that was not bound to X-junction. Binding was then monitored by loading samples directly onto a band shift gel running at 200 V (Fig. 9A). Starting from a position where junction DNA was bound by two dimers of SpCCE1 (complex 2), the first dimer dissociated in a few minutes, in the presence of EDTA, following the addition of the unlabeled DNA (lanes a-d and h-j). However, under these conditions, the binding of a single dimer of SpCCE1 to either X-12 or X-0 remained remarkably stable, with less than 15% of bound DNA dissociating over a 25-min period in each case (A,  lanes a-g and h-n, and B). To determine the effect that MgCl 2 has on the stability of SpCCE1-X-junction complex, we used the X-0 junction. X-0 is cleaved efficiently in only one of its strands and therefore remains largely intact (albeit with a nicked strand) in the presence of SpCCE1 and MgCl 2 . SpCCE1-X-0 junction complexes were formed in the absence of MgCl 2 . MgCl 2 (10 mM) was then added along with the competitor DNA, and binding was monitored as above. Under these conditions, the dissociation of SpCCE1 from X-0 was much faster than without MgCl 2 , and more than 70% of bound X-0 was converted to free junction within 15 min (B). When CaCl 2 , which does not support strand cleavage, was used instead of MgCl 2 , the rate of dissociation was much slower, albeit faster than in the absence of any divalent cation (B). These data indicate that divalent cations accelerate the dissociation of SpCCE1 from X-junction Methods." The control reaction was essentially the same as that in lanes g-l except the poly(dI⅐dC)-poly(dI⅐dC) was added before SpCCE1. Only the 10-min time sample from the control reaction is shown (lane m). B, effect of competitor DNA on the aberrant resolution of X-0 (essentially the same as in A except X-0 was used instead of X-12). C, effect of RuvA on the aberrant resolution of X-12 and X-0. Reactions (20 l) contained a 2.2 nM concentration of either X-12 or X-0 (each 32 P-labeled in oligonucleotide 2) in binding buffer plus 200 mM NaCl and contained protein as indicated. Reactions were preincubated with SpCCE1 for 6 min at 37°C and then for a further 6 min after the addition of RuvA. Strand cleavage was then initiated by the addition of 10 mM MgCl 2 and continued for 6 min before termination. Reactions were analyzed on native 10% polyacrylamide gels as described under "Materials and Methods." See the legend to Fig. 6 for comments on the background of unannealed and partially annealed oligonucleotides.
FIG. 8. Effect of competitor DNA and RuvA on the aberrant resolution of X-junctions. A, effect of competitor DNA on the aberrant resolution of X-12. 9 nM SpCCE1 was preincubated with 2.3 nM X-12 ( 32 P-labeled in oligonucleotide 2) in binding buffer plus 200 mM NaCl (total volume of 80 l) for 5 min at 37°C. After preincubation, a 10-l sample was withdrawn and terminated for analysis (lanes a and g). The reaction was then initiated by the addition of 10 l of either MgCl 2 in binding buffer plus NaCl (lanes b-f) or MgCl 2 plus 16 g of poly(dI⅐dC)-poly(dI⅐dC) in binding buffer plus NaCl (lanes h-l). In both cases, the final concentration of MgCl 2 in the reaction was 10 mM. 10-l samples were withdrawn at stated intervals, terminated, and analyzed on native 10% polyacrylamide gels as described under "Materials and DNA; however, the higher rates of dissociation observed with MgCl 2 suggest that the act of strand nicking also promotes dissociation. To provide a further measure of the importance of strand cleavage for dissociation, we used the SpCCE1121RW-GTP mutant protein that binds X-junctions but has little or no cleavage activity (Fig. 9C). The rate of dissociation of SpCCE1-121RWGTP from X-12 in the absence of MgCl 2 was slightly less than wild-type protein (compare B and C). The inclusion of 10 mM MgCl 2 increased the rate of dissociation but not to the same extent as seen with wild-type protein. These data indicate that MgCl 2 promotes the dissociation of SpCCE1 from junction DNA principally by enabling strand cleavage. To determine whether unilateral strand cleavage leads to the immediate dissociation of SpCCE1 from junction DNA, we analyzed the state of strand cleavage in complex 1 and free X-0 at all time points following MgCl 2 and competitor DNA addition. Strand cleavages were detected in both bound and free X-0 (data not shown). Therefore, we conclude that strand cleavage does not lead to the immediate dissociation of SpCCE1 from X-junction DNA but instead accelerates its rate.

DISCUSSION
Holliday Junction Binding-Using in situ 1,10-phenanthroline-copper footprinting, we have shown that SpCCE1 targets Holliday junctions by binding directly to the junction crossover point. Up to two dimers of SpCCE1 can bind concurrently to the same junction, although the second dimer binds with much lower stability than the first. Mixing experiments with RuvA provided further insight into the nature of this interaction. A dimer of SpCCE1 can bind and cleave a junction that is already bound by a tetramer of RuvA, but the second SpCCE1 dimer is excluded. Since a tetramer of RuvA binds principally to one face of the Holliday junction, holding it in an open and approximately square planar configuration, we surmise that (i) a dimer of SpCCE1 must be able to bind and resolve an X-junction with a 4-fold symmetrical configuration of duplex arms via interaction on only one face of the junction structure, and (ii) dimers of SpCCE1 bind to opposing junction faces. Consistent with the former idea, CCE1 imposes an extended, unstacked structure on the X-junction much the same as RuvA (15), we suspect that SpCCE1 does likewise. With respect to the second idea, the differences between the two junction faces could explain why two dimers of SpCCE1 bind to the same junction with different stabilities.
Substrate Specificity-We have shown that SpCCE1 is able to bind and cleave a range of different DNA substrates in addition to X-junctions. The minimal substrate that we have detected binding to is a linear double-stranded DNA. In comparison, little or no binding to single-stranded DNA was detected. Although SpCCE1 will bind to linear double-stranded DNA, no cleavage of this substrate was detected. Strand cleavage was only observed with substrates containing strand interruptions or branch points (junctions) with thymine nucleotides positioned close by. This substrate specificity is reminiscent of T4 endonuclease VII, which cleaves a range of substrates that have in common two mutually inclined DNA helices (23,41). Lilley and co-workers (41) have proposed that the two subunits within a dimer of T4 endonuclease VII are optimally arranged to contact the helices that are mutually inclined. However, it is evident that most, if not all, Holliday junction binding proteins alter the conformation of the junction upon binding to it (9, 11, 14 -16, 39). Therefore, binding and catalytic activation for enzymes like T4 endonuclease VII and SpCCE1 probably depend more on the intrinsic ability of DNA to be molded into the binding and catalytic centers of the protein rather than recognition of the initial conformation of the DNA.
T4 endonuclease VII cleaves X-junctions, Y-junctions, flayed duplex, nicked duplex, and partial duplex DNAs with the same efficiency (23). This is very different from SpCCE1, which has a marked preference for cleaving X-junctions especially at physiological ionic strengths. What underlies SpCCE1's substrate specificity? DNA binding is clearly one way in which SpCCE1 discriminates between potential substrates. SpCCE1 binds far better to DNAs containing junctions with either three or four fully duplex arms (X-and Y-junctions) than to linear substrates. As with cleavage, this binding specificity is enhanced at physiological ionic strengths. Presumably, efficient DNA binding requires that the substrate can accommodate a dimer of SpCCE1 such that both protein subunits can interact with the DNA and contribute to the stability of the complex. This would be necessary particularly under stringent conditions. However, SpCCE1's ability to bind to a substrate does not in all cases correlate with its ability to cleave that substrate; e.g. YA is bound almost as well as X-0 and yet, despite it containing the same cleavable site as X-0, is cleaved with considerably less efficiency. This indicates that DNA binding is not the only determinant in substrate cleavage specificity. Possibly, a fixed disposition of active sites within a dimer of SpCCE1 precludes either from accurately positioning for efficient strand cleavage on substrates other than X-junctions. Furthermore, efficient strand cleavage may require both active sites within a dimer of SpCCE1 to be positioned appropriately at the phosphodiester backbone. This would be reminiscent of FIG. 9. Dissociation of SpCCE1 from X-junctions. A, band shift analysis of dissociation of SpCCE1 from X-12 and X-0. Reactions (80 l) contained 18 nM SpCCE1 and 2.3 nM 32 P-labeled X-12 or X-0 in binding buffer plus 200 mM NaCl and 5 mM EDTA. Reactions were incubated at 37°C for 5 min, following which a 10-l sample was withdrawn and loaded directly onto a 4% polyacrylamide gel running at 200 V as described under "Materials and Methods" (lanes a and h). 16 g of poly(dI⅐dC)-poly(dI⅐dC) was then added to each reaction. 10-l samples were then withdrawn and loaded onto the gel at intervals as described above. B, quantitated data from A together with that from reactions where MgCl 2 or CaCl 2 was added along with the competitor DNA. C, quantitated data from reactions similar to those above except SpCCE1 was substituted by SpCCE1121RWGTP.
RuvC that requires both of its subunits to be positioned at a consensus cleavage sequence before either will act (12).
Aberrant Resolution of Holliday Junctions-During our studies, we observed that SpCCE1 occasionally cleaved off a single arm of X-12 and X-0 rather than resolving them into nicked linear products. This aberrant resolution appears to come from two dimers of SpCCE1 each making a single strand incision, since it was abolished by the binding of a tetramer of RuvA to the junction. We can imagine that such aberrant resolutions, if they occur in vivo, could provide a mechanism for forming a replication fork directly from a Holliday junction. SpCCE1 functions in the mitochondria of S. pombe, where it is expected to resolve recombination intermediates in mitochondrial DNA to ensure their efficient segregation. The mitochondrial DNA of S. pombe, as with many other fungi and higher plants, is proposed to replicate by a rolling circle mechanism (42). Holliday junctions formed from strand exchange between circular and linear molecules of mitochondrial DNA could be aberrantly resolved by SpCCE1 as a way of initiating rolling circle replication.
Resolution by a Nick and Counter-nick Mechanism-SpCCE1's unilateral strand cleavage in X-0 indicates that each active site in a monomer subunit functions independently. This suggests that Holliday junction resolution occurs by a nick and counter-nick mechanism. If this is true, then how does SpCCE1 achieve a high success rate of correct Holliday junction resolution? From measurements of the stability of the SpCCE1-X-junction complex, the answer appears to be that the lifetime of the protein-junction complex is long enough, even after nicking one strand, to ensure a high probability that both sites will be cleaved before dissociation. The same mechanism appears to be used by T4 endonuclease VII (41), whereas RuvC employs a more sophisticated mechanism that requires both active sites to be positioned at consensus cleavage sequences before either will act (12).
The stability of the SpCCE1-X-junction complex, although ensuring that there is a high probability of bilateral strand cleavage before dissociation, presents an interesting problem as to how SpCCE1 is able to dissociate from junctions that it cannot cleave due to the absence of appropriately positioned thymine nucleotides. Not all junction resolvases are presented with this problem; e.g. T4 endonuclease VII and T7 endonuclease I have more relaxed sequence requirements for cleavage, so they can cleave most if not all junctions that they bind to (1), and RuvC appears to have a much shorter lifetime on junctions (16) and associates with the RuvAB branch migration enzyme, which can presumably move junctions to sites that it can cleave (40,43). In contrast, RusA appears to be more like SpCCE1, because it binds with high stability to junctions and only cleaves them if they contain the right sequence (16). RusA only functions well in E. coli in the presence of the RecG branch migration protein (44). Possibly, RecG is required to locate junctions at cleavable sequences. Interestingly, SpCCE1, which can function in E. coli, also requires RecG activity in order to complement the UV sensitivity of a ruvAC mutant strain. 1 Perhaps SpCCE1, when functioning in E. coli, also makes use of RecG to locate junctions at sequences that it can cleave. Of course, SpCCE1's natural environment is the mitochondrion of fission yeast; how does it function effectively there? One possibility is that there are branch migration enzymes analogous to RuvAB and RecG. Alternatively, other Holliday junction binders actively displace SpCCE1 from junctions, e.g. a homologue of the Abf2p high mobility group protein that is involved in the recombination of mitochondrial DNA in S. cerevisiae (45). However, SpCCE1 may be able to function effectively in vivo without the need for additional enzymatic intervention, because approximately 70% of the mitochondrial genome of S. pombe consists of adenine and thymine nucleotides. Therefore, there is a high probability that any junction that is bound by SpCCE1 will contain appropriately positioned thymine nucleotides for it to cleave. Furthermore, the ability to load two dimers of SpCCE1 on a junction may provide a mechanism whereby SpCCE1 can scrutinize all four junction strands for cleavable thymine nucleotides at the same time, thereby maximizing the chance that every junction that is bound is resolved.
Conclusion-In this paper, we have suggested how SpCCE1 might target and interact with Holliday junctions in ways that both ensure efficient resolution and possibly, in some instances, initiate replication. Furthermore, we have shown that SpCCE1 is capable of cleaving a wider range of substrates than was previously believed. How much of this will be applicable to other resolvases, especially CCE1, remains to be determined.