Cleavage Properties of an Archaeal Site-specific Recombinase, the SSV1 Integrase*

SSV1 is a virus infecting the extremely thermophilic archaeon Sulfolobus shibatae. The viral-encoded integrase is responsible for site-specific integration of SSV1 into its host genome. The recombinant enzyme was expressed inEscherichia coli, purified to homogeneity, and its biochemical properties investigated in vitro. We show that the SSV1 integrase belongs to the tyrosine recombinases family and that Tyr314 is involved in the formation of a 3′-phosphotyrosine intermediate. The integrase cleaves both strands of a synthetic substrate in a temperature-dependent reaction, the cleavage efficiency increasing with temperature. A discontinuity was observed in the Arrhenius plot above 50 °C, suggesting that a conformational transition may occur in the integrase at this temperature. Analysis of cleavage time course suggested that noncovalent binding of the integrase to its substrate is rate-limiting in the cleavage reaction. The cleavage positions were localized on each side of the anticodon loop of the tRNA gene where SSV1 integration takes place. Finally, the SSV1 integrase is able to cut substrates harboring mismatches in the binding site. For the cleavage step, the chemical nature of the base in position −1 of cleavage seems to be more important than its pairing to the opposite strand.

Site-specific recombination catalyzed by tyrosine recombinases plays a number of critical roles in prokaryote and eukaryote kingdoms. Well documented examples in lower eukaryotes and bacteria include generation of genetic variability, plasmid copy control and/or stable inheritance, resolution of bacterial chromosome dimers or viral DNA integration in host chromosomes (for reviews, see Refs. [1][2][3]. Members of the tyrosine recombinases family catalyze site-specific recombination between two DNA sites by using a topoisomerase IB-like mechanism to cut and religate DNA strands (4,5). Unlike topoisomerases, tyrosine recombinases perform the ligation step after strand exchange between the two DNA partners. Sitespecific recombination requires the assembly of a synaptic complex containing, at least, four enzyme protomers and the two DNA sites. The recombination reaction occurs by cutting and exchanging the two pairs of DNA strands in two temporally distinct steps. In the first step, cleavage occurs on the top strands of each DNA site. For each site, a 3Ј-phosphotyrosine DNA-protein covalent complex is formed and a free 5Ј-OH DNA end generated. The leaving strands then attack the phosphotyrosine link of the recombination partner, thus releasing the recombinase subunits. After this first round of strand cleavagestrand exchange, a Holliday junction is formed. The bottom strands are then cut and religated, thus resolving the Holliday junction and completing the recombination reaction.
Site-specific recombination in archaea is not as well known. So far, the only studied system is SSV1, a virus of the extremely thermophilic archaeon Sulfolobus shibatae. In the cell, the 15.5-kb genome of SSV1 is present both as a circular DNA and as a provirus stably integrated into an arginine tRNA gene (6,7). Site-specific integration of SSV1 into its host chromosome is catalyzed by the virus-encoded integrase (Int SSV ). This enzyme catalyzes recombination between viral and chromosomal attachment sites, attP and attB (the latter previously denoted attA), to generate a left (attL) and right (attR) prophage att sites (7,8). In all four att sites, a 44-bp invariant sequence (previously referred as to the core sequence (7)) is found. This 44-bp segment comprises the 3Ј-half of the arginine tRNA gene and flanks the provirus as direct repeats (7). In vitro, Int SSV was shown to recombine DNA substrates both in the integrative and excisive pathways (8,9).
Based on sequence alignments, Int SSV was proposed to be a member of the tyrosine recombinases family (10). One striking aspect of this family is the lack of global homology among its more than 130 members. Nevertheless, a conserved signature is found in the C-terminal part of all the proteins. All members of the family harbor two short regions of similarity, box I and box II, sharing an invariant RHRY amino acids tetrad (11)(12)(13)(14). In different systems, mutations introduced at any of these positions produced proteins deficient in recombination, as expected for active site residues. The conserved tyrosine of the tetrad is the nucleophilic group which attacks the scissile phosphate of each DNA site, while the conserved RHR are involved in the correct positioning of the phosphate group and behave as a charge relay system within the catalytic pocket. For Int SSV , Tyr 314 was proposed to be the catalytic residue (10). However, there was no biochemical evidence that Int SSV would catalyze recombination in the same way as other members of the family. Furthermore, the observations that Int SSV is the most distantly related member of the tyrosine recombinases family (13) and that it harbors substitutions at several conserved positions (14), raised the possibility that the recombination mechanism might be different in archaea.
To gain more insights about the Int SSV recombination mechanism, we cloned the wild type Int SSV gene, as well as a Tyr 314 to Phe mutant (Int SSV -Y314F), in an E. coli expression vector. We developed a purification procedure that yields homoge-neously purified untagged recombinant enzymes. The present work focuses onto the first stage of catalysis which leads to the cleavage of the DNA substrate. We show that, in vitro, the SSV1 integrase is able to cut a double-stranded synthetic substrate harboring a specific target sequence. The efficiency of the cleavage reaction is strictly dependent on temperature. Analysis of the reaction kinetics suggests a rapid equilibrium between cleavage and religation within the integrase-substrate complex, a feature that can be expected for a topoisomerase IB-like mechanism. Moreover, the absence of substrate cleavage when using the Int SSV -Y314F mutant strongly suggests that like other members of the tyrosine recombinases family, Int SSV mediates cleavage through a 3Ј-phosphotyrosine intermediate. Finally, we have identified the cleavage points on the double-stranded substrate and found that they correspond to the borders of the anticodon loop of a tRNA arginine gene. This finding is reminiscent of the integration of some bacterial phages and plasmids into tRNA genes (15). This is the first case described in archaea, suggesting that targeting of tRNA genes is an ancient process that was conserved during evolution of bacteriophages.

MATERIALS AND METHODS
Bacterial Strains and Media-The Escherichia coli K12 strains used were as follows. C600 recA was used for cloning experiments and DNA amplification, except for the T-tailing vector and derivatives which were transformed in strain Mos Blue (Amersham Pharmacia Biotech). MC1061 (16) was transformed by plasmid pSG3 (described below) to give the integrase producing strain, MSG91. When necessary, Luria-Bertani medium was supplemented with ampicillin (100 g/ml) or tetracyclin (12.5 g/ml).
DNA Techniques-Large scale or minipreparations of plasmid DNA were performed as described previously (17). Restriction and modification enzymes were used as recommended by the manufacturers. Purification of DNA fragments (GeneClean, Bio 101), cloning of PCR fragments in a T-tailing vector (pMOSBlue T-vector kit, Amersham Bioscience), and DNA sequencing (Thermo Sequenase cycle sequencing kit, Amersham Bioscience) were done according to the supplier's instructions. Oligonucleotides were purchased from Genset and MWG.
Cloning of the Native Int SSV Gene under the Arabinose Promoter-Plasmid pBAD18, a pBR322 derivative carrying the para BAD promoter of E. coli and its regulator gene araC (18), was modified to be used as the expression vector. The unique NdeI site of pBAD18 was removed by fill in (pSG1). A XbaI-HindIII fragment containing the Shine-Dalgarno sequence and initiating ATG (overlapping a NdeI site) from gene 10 of bacteriophage T 7 was then subcloned into the corresponding sites of pSG1 (pSG2). The Int SSV coding sequence was amplified by PCR using plasmid pMT4 (8) as template and the following primers 5Ј-CGCGTG-GACATATGACGAAAG-3Ј (ND5) and 5Ј-CAAGCTTCAGACCCCTTT-TAGCCATT-3Ј (MUT3). The underlined positions correspond to the NdeI and HindIII sites created. In bold are shown the initiating codon (ND5) and stop codon (MUT3). Italics indicate the mutations introduced. The resulting 1,025-bp fragment was cloned into the pMosBlue vector, and its sequence verified. The NdeI-HindIII 1,010-bp fragment was then subcloned into the corresponding sites of pSG2, to give the expression plasmid pSG3.
Site-directed Mutagenesis of the Gene Encoding Int SSV -The 1,049-bp XbaI-HindIII fragment containing the Int SSV coding sequence was subcloned into the corresponding restriction sites of M13mp18 (19). Sitedirected mutagenesis was performed on the uracil-containing template of the resulting phage as described previously (20) but with T 7 DNA polymerase for the elongation step. The mutation (underlined position) Y314F was introduced by using the following oligonucleotide: 5Ј-GAGC-GATACGAAATGTTGCG-3Ј. The nucleotide sequence of the entire mutated fragment was then verified by sequencing prior to recloning in pSG3.
Expression and Purification of Int SSV -An overnight culture of strain MSG91 in Luria-Bertani, ampicillin, 0.2% glucose, was diluted to A 550 ϭ 0.07 in Luria-Bertani, ampicillin. The culture was grown at 37°C to A 550 ϭ 0.6 and expression of Int SSV was induced by the addition of arabinose (1% final concentration). 2 h after induction (A 550 ϭ 1.1), cells were harvested by centrifugation. The cell pellet was washed with TN buffer (10 mM Tris, 100 mM NaCl) and centrifuged again. The cell pellet was resuspended in 5 ml of A 0.1 buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5 mM ␤-mercaptoethanol) per g of wet cells. Lysozyme was added (1.7 mg/g of cells), and incubation performed for 1 h at 4°C. 3 cycles of freezing/thawing were then performed, and the resulting crude extract was ultracentrifuged 45 min at 4°C and 200,000 ϫ g. The supernatant was collected (fraction FI) and submitted to streptomycin sulfate fractionation (2% final concentration). After 30 min stirring at 4°C, the extract was centrifuged for 15 min at 4°C and 8,500 ϫ g. The pellet was resuspended in 2.5 ml of A 1.0 buffer (same as A 0.1 , but 1 M NaCl), and solubilization obtained by stirring at 4°C for 1 h (fraction FII). The solution was then centrifuged 10 min at 4°C and 12,500 ϫ g. The supernatant was collected and incubated 30 min at 37°C after addition of RNase A (2 g/ml final concentration) then dialyzed overnight at room temperature against 500 ml of A 1.0 buffer. The dialyzed sample was collected and centrifuged at room temperature 10 min at 10,000 ϫ g. The supernatant (fraction FIII) was collected and heated for 10 min at 65°C. After centrifugation at 10,000 ϫ g, the sample (fraction FIV) was further purified by size exclusion chromatography performed at room temperature, onto a Sephacryl S300 column (Amersham Bioscience) equilibrated in A 1.0 buffer. The peak fractions containing the integrase were pooled and concentrated in a Vivaspin concentrator (MWCO 30,000), then dialyzed overnight against AЈ 0.5 buffer (same as A 0.1 but 500 mM NaCl and 20% glycerol). The protein (fraction FV) was either stored at 4°C for up to 1 month or at Ϫ70°C for up to 6 months. The yield of purification is of 450 g of pure Int SSV per liter of bacterial culture.
Fluorescence-Protein fluorescence was followed with a Bio-Tek SFM 25 fluorimeter. Fluorescence measurements were carried out in a buffer composed of 50 mM Tris, pH 7.5, 125 mM NaCl, 1 mM EDTA, containing 1 mM dithiothreitol. Protein concentration was 0.5 or 3.5 M when recording excitation or emission spectra, respectively. Emission spectra of the protein excited at 295 nm were recorded at different temperatures.
Substrates Preparation-Oligonucleotides were either 3Ј-end labeled by using [␣ 32 P]ddATP and terminal nucleotidyl transferase or 5Ј-end labeled by [␥-32 P]ATP and polynucleotide kinase. Unincorporated nucleotides were removed by spin dialysis, and the labeled oligonucleotide was then hybridized with a 2-fold excess of unlabeled complementary strand in TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA).
Cleavage Assays-The cleavage reactions were carried out in 20 l of reaction mixtures consisting of 30 mM Tris, pH 7.5, 50 g/ml bovine serum albumin, 125 mM NaCl, and the indicated amounts of 5Ј-end labeled substrate and protein. The reaction mixture was incubated at 65°C (unless noted otherwise) and at the times indicated quenched in Laemmli loading buffer (final concentrations: 40 mM Tris, pH 6.8, 3% SDS, 8% glycerol, 250 mM ␤-mercaptoethanol, 0.005% bromophenol blue) and heated for 5 min at 98°C. The reaction products were then analyzed by electrophoresis through 12% SDS-polyacrylamide gel. The gels were autoradiographed, and bands corresponding to protein-DNA complex and substrate DNA were quantitated using the Scan Analysis 2.21 software. The results presented are the mean of at least three independent experiments.
Determination of the cleavage position was performed by incubating 2 M Int SSV with 50 nM 3Ј-end labeled double-stranded substrate for 4 h at 65°C in the same buffer conditions. The reaction was stopped by addition of formamide dye mixture (97.5% deionized formamide, 10 mM EDTA, 0.3% bromophenol blue, 0.3% xylene cyanol blue), and the sample heated for 5 min at 98°C. The reactions were analyzed on a 18% polyacrylamide gel (19:1) containing 8 M urea, in TBE buffer (90 mM Tris, 90 mM boric acid, 1 mM EDTA). Electrophoresis was performed at 52 V/cm. The cleavage positions were determined by comparing electrophoretic mobilities of the samples to that of specific and nonspecific ladders (see "Results").

Expression and Purification of the Recombinant Int SSV -
Cloning of the Int SSV gene (SSV1 open reading frame D-335) was first described by Muskhelishvili et al. (8). However, the strategy of cloning into pGEX-2T expression vector led to the addition of 2 residues at the N terminus and 7 residues at the C terminus of the protein. To investigate the biochemical properties as well as to start up crystallographic studies, we wanted to express the untagged integrase protein. The cloning strategy of the Int SSV gene into a pBAD vector is described under "Materials and Methods." When induced with arabinose, strain MSG91 produced a recombinant Int SSV with a molecular mass on SDS-PAGE close to the predicted value of 38.9 kDa. Under the expression conditions described under "Materials and Methods," Int SSV accounts for 10 -15% of total protein mass.
The purification scheme of the recombinant Int SSV takes advantage of two intrinsic properties of the enzyme, its affinity for nucleic acids and its thermostability, so that the enzyme could be easily purified to near homogeneity. At low ionic strength in the presence of nucleic acids, Int SSV remains in the soluble fraction (fraction I) after centrifugation at 200,000 ϫ g (Fig. 1, lane 1). When adding streptomycin sulfate (2% final) to fraction I, Int SSV co-precipitates with nucleic acids. About 80% of the proteins are eliminated during this fractionation step, while most of Int SSV remains in the nucleic acids pellet. The pellet is resolubilized in a high ionic strength buffer (A 1.0 ) to give fraction II (Fig. 1, lane 2). Remaining RNA contaminants are eliminated by RNase A treatment of fraction II. After dialysis, the sample (Fig. 1, lane 3) is submitted to thermal denaturation for 10 min at 65°C, and the insoluble material removed by centrifugation. This step eliminates 65% of the remaining contaminant proteins. The collected soluble fraction (Fig. 1, lane 4) is injected onto a Sephacryl S300 size exclusion column equilibrated in A 1.0 buffer. Int SSV elutes in a single peak, with an apparent molecular mass of 42.7 kDa indicating that, under these conditions, the protein behaves as a monomeric species. The peak fractions are then pooled and concentrated by ultrafiltration (Fig. 1, lane 5) to a final concentration of 0.5 to 1 mg/ml. The purified integrase can be stored at 4°C for up to 1 month or at Ϫ70°C for up to 6 months without significant loss of cleavage activity. The recombinant Int SSV purified from E. coli retains its thermostability and affinity for nucleic acids, suggesting that, even though expression is performed at a low temperature, the recombinant enzyme is correctly folded.
Int SSV Is Able to Cleave a Minimal Substrate in Vitro, Forming a Covalent Link via Tyr 314 -The reaction catalyzed by tyrosine recombinases involves a transient protein-DNA covalent intermediate between a DNA 3Ј-phosphoryl group and the conserved tyrosine residue. This mechanism is supposed to be conserved within the family. Therefore, the prediction on Int SSV activity was that the enzyme would be able to cleave a specific DNA substrate and to form a 3Ј-phosphoprotein covalent intermediate.
To monitor the cleavage reaction, we designed a 19-bp double-stranded synthetic substrate (XTB, Table I). XTB contains the 18 bp of the attP site protected by Int SSV (9). Int SSV (1 or 2 M) was incubated 4 h at 65°C with the XTB substrate (12.5 nM), in conditions described under "Materials and Methods." After heat denaturation, the reaction products were analyzed by SDS-PAGE ( Fig. 2A). The appearance of a low mobility product (lanes 2 and 4) indicates that a covalent complex is formed between the integrase and the substrate. The same result is obtained when the bottom strand of the substrate is 5Ј-end labeled ( Table I). Transfer of the labeling to the enzyme reveals that it is covalently bound to the 3Ј-DNA end, confirming that the polarity of strand cleavage by Int SSV is the same as that of other members of the tyrosine recombinases family.
The specificity of the cleavage reaction was further assessed by adding increasing amounts of nonspecific DNA (Table I).
Cleavage on the top and bottom strands was not significantly reduced in the presence of 0.5 g of poly(dI-dC) 2 used as nonspecific competitor DNA. These results indicate that the reaction is specific to the XTB-encoded sequence since a large excess of nonspecific DNA (ratio site:competitor in bp of 1 to 150) has very little effect on the cleavage efficiency. The XTB substrate thus contains enough information to allow specific cleavage by Int SSV on the top and bottom strands.
Alignment of Int SSV amino acid sequence with the other members of the family predicted that Tyr 314 is the active site tyrosine (10,13,14). To demonstrate that Tyr 314 is indeed involved in the covalent complex formation, the activity of the mutant Int SSV -Y314F has been investigated. The purification procedure applied to the mutant protein gave identical results to that obtained with the wild type enzyme (data not shown). As presented in Fig. 2A (lanes 3 and 5), the Int SSV -Y314F mutant is totally inactive in the cleavage reaction, indicating that the hydroxyl group of Tyr 314 is essential for the formation of the covalent complex. Thus, like bacterial and eukaryotic tyrosine recombinases, the archaeal integrase mediates cleavage via a 3Ј-phosphotyrosine intermediate.
It was suggested that tRNA genes may have been the recognition sequences for an ancestral site-specific recombination enzyme (7). It was also proposed that "the frequent use of tRNA genes as insertion sites relates to the suitability of anticodon stem sequences as elements for core sites" (15). Based on these assumptions, we predicted that the site specificity of Int SSV would lead to strand cleavages on each side of the anticodon loop (Fig. 2B).
To identify the Int SSV cleavage sites, the XTB substrate was 3Ј-end labeled on its top or bottom strands. After incubation with the enzyme, the reactions were analyzed on a denaturing gel (Fig. 2C). Preliminary experiments revealed that the oligonucleotides used as minimal substrates had abnormal migration behavior either on native or denaturing gels. Therefore, to unambiguously localize cleavage positions, two kinds of radiolabeled size standards were used: a nonspecific size ladder ranging from 8 to 32 nucleotides (Fig. 2C, lanes M) and two specific oligonucleotides corresponding to the expected prod-ucts of cleavage on the top (14 nucleotides) and bottom (12 nucleotides) strands (Fig. 2C, lanes 3 and 6). Analysis of the reactions (Fig. 2C, lanes 2 and 5) reveals that Int SSV cuts XTB at specific positions which are located on each side of the anticodon loop sequence (Fig. 2B). 3Ј-End labeling produces a blunt end substrate which has a cleavage efficiency lower than that of a 5Ј-end labeled substrate. This behavior correlates the a All substrates were 5Ј-end labeled. The amount of covalent complex formed was quantified by scanning of the autoradiograms (covalent ϩ unreacted ϭ 100%) then normalized with cleavage of the XTB top strand taken as 100% (original covalent product formed after 4 h at 65°C, 38%). b 0.5 g of poly(dI-dC) 2 added in the reaction as nonspecific competitor. c ND, not done. yields obtained when using 5Ј-end labeled blunt end substrates. 1 The same cleavage positions were observed when a longer substrate (FTB , Table I) was used, although with a lower efficiency (data not shown). These results indicate that XTB has all the minimal sequences required by Int SSV to get cleavage specificity.
As described for the pSAM2 integrase (28) and for the HP1 and L5 integrases (25,27), the specificity of insertion is directed toward the anticodon loop. This result strengthens the prediction made by Campbell (15) that when attB sites overlap a tRNA anticodon stem-loop, the anticodon loop coincide with the exchange (O) region, while the sequences of dyad symmetry composing the tRNA anticodon stem are likely to be the integrase core-binding sites from the left and right parts of attB and attP (named B, BЈ, P, and PЈ by analogy to the core sites, see Fig. 3A). For Int SSV , alignment of these core type sequences reveals that as predicted by Campbell (15) the BЈ and PЈ sites are identical while the B and P sites differ from each other (Fig.  3B). The core sequence (i.e. the minimal sequence required for recombination) should therefore be 27 bp long (Fig. 3A) rather than 44 bp long as previously proposed (7). This is the first case described in archaea, suggesting that targeting of tRNA genes is an ancient process that was conserved during evolution of bacteriophages.
The Int SSV -XTB Complex Is in Equilibrium between Covalent and Noncovalent Species-The extent of XTB cleavage was studied, revealing that the maximal activity is reached for a molar ratio of 40:1 (enzyme:substrate), whatever strand is cleaved (not shown). We then followed the time course formation of the covalent complex between Int SSV and the XTB substrate at 65°C for 24 h for a fixed XTB and different enzyme concentrations (Fig. 4A) molar excess of enzyme (as high as 160-fold). It was striking that the final value of 0.4 of f cov was (almost) the same, independently of the total enzyme concentration E 0 in the range from 0.1 to 2 M (Fig. 4A). This suggested to us that the covalently bound substrate was not in direct equilibrium with free enzyme. In the different assays the total integrase E 0 (between 0.05 and 2 M) was in molar excess over the substrate (12.5 nM), so that, above 0.1 M integrase, the concentrations of free and total integrase were very close.
The simplest scheme describing this situation is, where ES and ES cov are, respectively, a noncovalent and a covalent complex between the integrase and its substrate. Scheme 1 assumes that the cleavage step is at equilibrium relative to the binding step, and thus that binding of the substrate is concomitant with its cleavage. The alternative model, with the reverse assumption of rapid binding and slow cleavage, is not consistent with the data. According to Scheme 1, the fraction of substrate covalently bound to the enzyme, f cov , is equal to f cov ϭ ES cov /S 0 , where S 0 is the total substrate concentration, S 0 ϭ S ϩ ES ϩ ES cov ϭ ES cov (1 ϩ 1/K ϩ k Ϫ1 /(k 1 E 0 K)), since E 0 Ͼ Ͼ S 0 , E (representing the free enzyme) can be approximated to E 0 , the total amount of enzyme. The reciprocal of the fraction f cov of covalently bound substrate at final equilibrium is then given by, 1/f cov ϭ 1 ϩ (1 ϩ k Ϫ1 /k 1 ⅐E 0 )⅐(1/K). Since f cov was found independent of E 0 , then k Ϫ1 /k 1 ⅐E 0 was smaller than 1, and the ratio k Ϫ1 /k 1 was smaller than the lowest E 0 (0.1 M), that is k Ϫ1 /k 1 Ͻ 10 Ϫ7 M. In this case, the partitioning between the covalent ES cov and the noncovalent ES complexes would remain independent of the total protein concentration because (almost) all the substrate would be bound (covalently or noncovalently) to the integrase, with (almost) no free substrate remaining. The technique used here could not determine the relative amount of free and noncovalently bound substrate, but we could verify on nondenaturing gels that most of the substrate was bound to Int SSV as predicted by our model (Fig. 4B).
Determination of the Kinetic Parameters of the Reaction-In contrast to the final plateau, the rate at which the substrate distributed itself between covalent and noncovalent complexes was markedly dependent upon total Int SSV concentration E 0 (Fig. 4A). This was also in agreement with Scheme 1, which predicts that formation of the covalent complex ES cov should follow first-order kinetics, with an apparent rate constant given by the following equation.
First-order rate constants k app were obtained from Fig. 4A by fitting the kinetics to exponential curves, and Fig. 4C shows that the values of k app increased linearly with E 0 . The rate constant for E 0 ϭ 0.05 M was obtained by assuming that the final value of f cov was 0.4 as for the other values of E 0 . According to Equation 1, the slope of the straight line gave k 1 ϭ (2.5 Ϯ 1) ⅐ 10 2 M Ϫ1 s Ϫ1 , and the vertical intercept gave k Ϫ1 ϭ (0.8 Ϯ 0.6)⅐10 Ϫ5 s Ϫ1 . The ratio k Ϫ1 /k 1 ϭ 3 ϫ 10 Ϫ8 M obtained from these kinetics was indeed lower than the limit of 10 Ϫ7 M deduced from the lack of variation of f cov with E 0 . The rate-limiting step for the cleavage reaction would therefore be the formation of a noncovalent complex between the integrase and its substrate: high concentrations of enzyme and/or substrate would be needed to form this complex at an appreciable rate. It is a high (of the order of micromolar) concentration of (at least) one of the reagents that would be required for cleavage, rather than a high molar ratio of enzyme to substrate. Another conclusion is that after binding noncovalently to the enzyme, only about 40% of the substrate undergoes cleavage with concomitant covalent binding to the protein. The equilibrium between ES and ES cov (Scheme 1), that is between intact and cleaved substrate on the surface of the integrase, would have an equilibrium constant K of about 0.4/0.6 ϭ 0.7. This constant has a value near unity, consistent with an equilibrium involving an exchange between 2 phosphodiester bonds, the phosphodiester bond of the intact substrate and the phosphodiester bond with the hydroxyl group of Tyr 314 on the cleaved substrate. An equilibrium constant near unity also indicates that the cleavage reaction could be easily reversed, thus leading to possible exchange with another DNA partner.
Affinity of the Enzyme for the Synthetic Substrates-Analysis of the relative affinity of Int SSV for the XTB substrate was performed by adding the pKS-attB plasmid harboring the chromosomal attB site (8) as specific competitor into the cleavage reaction.
The enzyme (1 M) was incubated with 25 nM 5Ј-end labeled XTB and increasing amounts of competitor DNA. The results are shown in Fig. 5. A molar ratio of 1 attB for 5 XTB decreases formation of the covalent complex by 50%. Higher amounts of pKS-attB totally abolish formation of the Int SSV ⅐XTB covalent complex, indicating that Int SSV shows a better affinity for a full recombination site than for the synthetic minimal site. Although less efficiently cleaved, a longer substrate of 29 bp (FTB, see Table I), forms a covalent complex more resistant to competition by pKS-attB since only 55% of the complex can be competed (Fig. 5). The FTB site contains the complete 27 bp proposed attP core sequence and may thus form a stable synaptic complex with pKS-attB. The competition experiments indicate that the affinity of the enzyme for its synthetic substrates does not correlate its cleavage efficiency (see Table I) a finding consistent with the kinetic data.
The Recombinant Int SSV Is Thermostable and the Cleavage Activity Increases with Temperature-The strategy employed to purify Int SSV revealed that the recombinant enzyme is thermostable since it remains soluble after 15 min at 65°C, a behavior expected for a protein from a hyperthermophilic organism. To further access the thermodynamic parameters of Int SSV , we studied its thermal stability and the thermodependence of the cleavage reaction.
Since the optimal growth temperature of S. shibatae is around 80°C, Int SSV is expected to be stable at such temperature. The enzyme was preincubated for different periods of time at 82°C without DNA, and then incubated with XTB for 2 h at 65°C (Fig. 6A). The results obtained show that the purified recombinant Int SSV is relatively stable at 82°C, with a half-life of inactivation of 32 min.
The ability of Int SSV to mediate cleavage was investigated at different temperatures. The enzyme (1 M) was incubated for 4 h with the XTB site (12.5 nM), in a temperature range of 40 to 65°C and the data normalized to the amount of covalent complex formed at 65°C (Fig. 6B). Higher temperatures were not tested to prevent interference by substrate melting. The results obtained show that the efficiency of the cleavage reaction increases with temperature. While cleavage is almost negligible at 40°C, the activity strongly increases between 45 and 60°C reflecting the thermophilicy of the enzyme. This increase in  6. Effect of the temperature on Int SSV activity. A, thermal stability of the enzyme. Int SSV was incubated at 82°C without DNA. At different times, 1 M protein samples were transfered into a test reaction containing 12.5 nM XTB and incubated for 2 h at 65°C. The reaction products were analyzed by SDS-PAGE and the amount of covalent complex formed were determined by scanning the autoradiogram of the gel. The data were normalized by taking the amount of complexes formed without preincubation as 100%. The resulting relative activity values were plotted against the time of preincubation. B, temperature dependence of the cleavage activity. Int SSV (1 M) was incubated for 4 h with 12.5 nM XTB substrate at six different temperatures ranging from 40 to 65°C. The reaction products were analyzed as described in A except for normalization. The inset shows an Arrhenius plot of the data. activity probably reflects that the overall velocity of the reaction is a function of temperature. Representation of the data in an Arrhenius plot is given in the inset of Fig. 6B. Linear approximation of the profile yielded a "break" point located around 50°C. Below 50°C, the calculated activation energy E a is of 169.8 kJ/mol while above 50°C, it decreases to 66.5 kJ/ mol. Such a profile for an Arrhenius plot can be the result of a lower than expected rate at high temperatures. A trivial explanation for such behavior is that the substrate starts to melt around 50°C, thus becoming rate-limiting above this temperature. Another possible explanation for the bimodal profile of thermodependence of the cleavage activity (inset, Fig. 6B), is that conformational transitions may occur within the enzyme in the studied temperature range, affecting either the binding of the enzyme to its substrate, or simply the equilibrium between cleavage and religation.
Temperature-induced Changes of Int SSV Conformation-Fluorescence measurements allow to monitor the local environment of the aromatic side chains, mostly the Trp residues (29). The 3 Trp residues present in Int SSV may therefore be probes to follow the potential conformational changes induced by temperature. Like its absorption spectrum, the fluorescence excitation spectrum of Int SSV at 20°C has a maximum at 278 nm (data not shown). However, when exciting at this wavelength the fluorescence signal is essentially due to Tyr residues (27 present in Int SSV ). We therefore followed the protein fluorescence upon excitation at 295 nm at different temperatures. The fluorescence spectra emitted by Int SSV in a temperature range of 20 to 58°C are characterized by a maximal emission wavelength of 343 nm (Fig. 7A). When increasing the temperature, the fluorescence at 343 nm is reduced, to reach at 58°C about 40% of the intensity observed at 20°C. The decrease in emitted fluorescence is reversible and the full intensity recovered when going back to 20°C. The decrease in fluorescence intensity indicates a restricted flexibility of the aromatic side chains as a probable consequence of strengthened hydrophobic interactions. Tightening of the hydrophobic interactions with increasing temperature would compensate for the increased thermal agitation, thus preventing thermal unfolding. When plotting the maximal emitted fluorescence as a function of temperature ( Fig. 7B) a discontinuity of fluorescence decrease is observed around 45°C. Interestingly, the temperature of the discontinuity correlates to the transition temperature in the Arrhenius plot (inset, Fig. 6B), suggesting that the break in this plot is due to conformational changes in the protein rather than to alteration of the substrate. Such a behavior has been previously reported for other thermophilic enzymes (30,31). strate melting could also contribute to slow down the cleavage activity between 60 and 65°C (Fig. 6B), we wanted to study the ability of Int SSV to cleave single-stranded DNA. Int SSV cleavage activity was measured on each unannealed strand of the XTB substrate after 4 h at 65°C, under the conditions described for the double-stranded substrate. The results are presented in Table I. Unexpectedly, efficient formation of covalent complex was observed for both XT and XB single strands, ruling out that substrate melting may reduce the covalent complex formation above 60°C and suggesting that Int SSV may recognize and cleave single-stranded substrates. However, we suspected that both the XT and XB single-stranded oligonucleotides could anneal on themselves to form imperfect double-stranded substrates that may be cleaved by Int SSV (Fig. 3C). This homopairing would generate a BspEI site, and we used this property to determine the structures of the substrates that were incubated with the integrase. Indeed, digestion of the XT substrate by BspEI prior to incubating with Int SSV totally abolished the covalent complex formation (data not shown) suggesting that the XT sequence could adopt a double-stranded structure recognized as a substrate by Int SSV . In the light of a doublestranded structure for XT, the cleavage efficiency would reflect the sum of "top" (CT in Fig. 3C) and "bottom" (CB in Fig. 3C) strand cleavages, since both strands of the substrate are 5Ј-end labeled in this case. If so, the XT cleavage efficiency is likely to be lower than the XTB cleavage where only one strand is labeled.
Comparison of the XT and XB homopairing shows that both produce a mismatch at position Ϫ1 of the "CT" cleavage site and a mismatch at position Ϫ2 of the "CB" cleavage site (Fig.  3C). However, incubation of Int SSV with either the XT or XB substrates leads to formation of a covalent complex suggesting that mismatches in the binding sites do not alter substrate binding and cleavage by the enzyme. As shown in Table I, Int SSV seems to cleave more efficiently the XT substrate than the XB substrate. Furthermore, formation of a covalent complex between the enzyme and the XT substrate is not affected by large amounts of nonspecific competitor (Table I: compare XT cleavage to XTB top strand cleavage with or without poly(dI-dC) 2 ) which is not the case when using XB as a substrate (Table I: compare XB cleavage to XTB bottom strand cleavage). It should be noted that the Ϫ1 position of the CT cleavage site is identical to the XTB site in the XT:XT appariement, but different in the case of the XB:XB appariement (Fig.  3C). For Int SSV the chemical nature of the base in position Ϫ1 of the top strand cleavage seems to be more important than its pairing to the opposite strand, at least for the cleavage step.
Cleavage of mismatched DNA substrates was studied for the vaccinia virus type IB topoisomerase (32) and revealed a hierarchy of mutational effects based on the nature of the mutation and the position mutated relative to the cleavage site. At positions equivalent to the Ϫ1 and Ϫ2 of the Int SSV site, the most drastic cleavage inhibition (Ͼ100-fold) was observed for any mutation at position Ϫ2. In the case of Int SSV the cleavage inhibition is not so drastic, but the Ϫ1 and Ϫ2 mismatches should be uncoupled to quantify the relative influence of each position on cleavage. Indeed we cannot rule out that only one of the two mismatches is responsible for the inhibition observed. The XT:XT pairing and XB:XB pairing are chemically and structurally different, the former harboring two pyrimidine: pyrimidine mispairs while the latter has two purine:purine mispairs. The differences observed between the XT and XB cleavage could therefore reflect that Int SSV is more sensitive to pu:pu mispairing than to py:py misparing in either position Ϫ1 or Ϫ2 of the cleavage site.
Conclusion-Analysis of some of the catalytic properties of Int SSV revealed that the enzyme exhibits a cleavage mechanism similar to that observed for tyrosine recombinases and type IB topoisomerases (4). The cleavage step is dependent on Tyr 314 , the proposed active site tyrosine (10,13,14), and leads to the formation of a 3Ј-phosphoprotein intermediate. Enzymatic analysis of the cleavage reaction revealed that the rate-limiting step of the reaction is the formation of a noncovalent complex between Int SSV and its substrate. All the sequence information for binding and cleavage are located within a minimal 19-bp sequence, and correspond to the anticodon stem-loop sequence of the tRNA gene where SSV1 integration takes place. Campbell (15) proposed that elements integrating in tRNA genes may use the symmetry of the anticodon stem-loop sequence as the inverted core sequences in attB. Indeed, Int SSV cleavage specifity targets the borders of the anticodon loop as it is described for mesophilic bacteriophages (25,27). This observation strongly supports the hypothesis that tRNA genes may have been the recognition sequences for an ancestral site-specific recombinase (7).
Int SSV is also able to bind and cut substrates harboring mismatches in the binding sites. The chemical nature of the base in position Ϫ1 of the top strand cleavage seems to be more important than its pairing to the opposite strand, at least for the cleavage step. The influence of positions Ϫ1 and Ϫ2 on cleavage efficiency as well as substrate mismatching is now under investigation.