Mutational analysis of the archaeal tyrosine recombinase SSV1 integrase suggests a mechanism of DNA cleavage in trans.

The only tyrosine recombinase so far studied in archaea, the SSV1 integrase, harbors several changes in the canonical residues forming the catalytic pocket of this family of recombinases. This raised the possibility of a different mechanism for archaeal tyrosine recombinase. The residues of Int(SSV) tentatively involved in catalysis were modified by site-directed mutagenesis, and the properties of the corresponding mutants were studied. The results show that all of the targeted residues are important for activity, suggesting that the archaeal integrase uses a mechanism similar to that of bacterial or eukaryotic tyrosine recombinases. In addition, we show that Int(SSV) exhibits a type IB topoisomerase activity because it is able to relax both positive and negative supercoils. Interestingly, in vitro complementation experiments between the inactive integrase mutant Y314F and all other inactive mutants restore in all cases enzymatic activity. This suggests that, as for the yeast Flp recombinase, the active site is assembled by the interaction of the tyrosine from one monomer with the other residues from another monomer. The shared active site paradigm of the eukaryotic Flp protein may therefore be extended to the archaeal tyrosine recombinase Int(SSV).

The only tyrosine recombinase so far studied in archaea, the SSV1 integrase, harbors several changes in the canonical residues forming the catalytic pocket of this family of recombinases. This raised the possibility of a different mechanism for archaeal tyrosine recombinase. The residues of Int SSV tentatively involved in catalysis were modified by site-directed mutagenesis, and the properties of the corresponding mutants were studied. The results show that all of the targeted residues are important for activity, suggesting that the archaeal integrase uses a mechanism similar to that of bacterial or eukaryotic tyrosine recombinases. In addition, we show that Int SSV exhibits a type IB topoisomerase activity because it is able to relax both positive and negative supercoils. Interestingly, in vitro complementation experiments between the inactive integrase mutant Y314F and all other inactive mutants restore in all cases enzymatic activity. This suggests that, as for the yeast Flp recombinase, the active site is assembled by the interaction of the tyrosine from one monomer with the other residues from another monomer. The shared active site paradigm of the eukaryotic Flp protein may therefore be extended to the archaeal tyrosine recombinase Int SSV .
Tyrosine recombinases form a large family of site-specific recombinases comprising more than 150 members, most of which were identified on the basis of sequence similarities (1,2). Within this family, several subfamilies can be defined such as the -phage integrase family, the Xer recombinases family, or the yeast plasmid recombinases family (1). The hallmark of tyrosine recombinases is the conservation of six noncontiguous residues: Arg I , Lys ␤ , His II , Arg II , His/Trp, Tyr (Table I). This motif is directly involved in catalysis of DNA strand cleavage and strand exchange (for review, see Ref. 3). Five of the six residues are located within the highly conserved boxes I and II found in tyrosine recombinases (1,4,5), whereas the sixth residue, Lys ␤ , was identified by alignments with the eukaryotic topoisomerases IB (6). Two different structural organizations of this motif have been described from crystallographic data. In prokaryotic tyrosine recombinases XerD (7), Cre (8), HP1 integrase (9), and -Int (10,11), the six active site residues come from a single monomer, whereas the eukaryotic Flp recombinase presents a shared active site, where the catalytic tyrosine is provided by one monomer, and the five other residues are from another monomer (12). In this latter case, the active site is created by dimer association. As a consequence of this organization, Flp realizes trans cleavage (13,14), whereas prokaryotic recombinases act in cis (8,9,(15)(16)(17). Cis cleavage is the result of cis activation/cis cleavage where the tyrosine of the bound monomer attacks the nearby activated phosphate. In trans cleavage, binding of a monomer to its site leads to activation of the adjacent phosphodiester that will be attacked by a nucleophile (here a tyrosine) provided in trans by a partner monomer, a mechanism that can be described as cis activation/ trans cleavage. In both cases the chemistry of the reaction is conserved and is similar to that used by topoisomerases IB (18). The Arg I , His II , and Arg II side chains coordinate the scissile phosphate, activating it for nucleophilic attack by the tyrosine and stabilizing the resulting transient penta-coordinated phosphate. In some proteins, the His/Trp side chain forms a hydrogen bond to the nonbridging oxygen of the scissile phosphate (19), whereas in Flp this residue is more likely involved in protein-protein interactions (20). The Lys ␤ residue is critical for activity of topoisomerases IB and tyrosine recombinases (6,21,22). Enzymatic analysis of vaccinia topoisomerase IB mutants revealed that this residue is the general acid catalyst that protonates the 5Ј-oxygen of the leaving strand (21). Crystal structures of Cre (8), Flp (12), -Int (10), and human topoisomerase IB (19) reveal that this residue contacts the base adjacent to the cleavage site in the minor groove. Other structural data show that Lys ␤ is located on a loop displaying a high conformational flexibility (9,23). Therefore the Lys ␤ residue was proposed to serve similarly as a general acid in the reaction mechanism catalyzed by tyrosine recombinases (21).
So far, the only studied archaeal member of the tyrosine recombinases family is the SSV1 integrase (Int SSV ) 1 encoded by SSV1, a virus of the extremely thermophilic archaeon Sulfolobus shibatae. Int SSV catalyzes the site-specific integration of the viral DNA into the host chromosome using viral and chromosomal attachment sites attP and attB (24,25). In a previous work we have shown that Int SSV exhibits a cleavage mechanism dependent on Tyr 314 , leading to the formation of a 3Јphosphoprotein intermediate like other tyrosine recombinases (26). However, Int SSV harbors substitutions at several conserved positions ( Fig. 1 and Ref. 2) and cannot be classified phylogenetically in any subgroup of the tyrosine recombinases family. Whether the recombination reaction catalyzed by Int SSV would follow the general mechanism described for tyrosine recombinases or would be different in archaea remained an open question (27). To characterize further the site-specific recombination mechanism in archaea we have generated Int SSV mutants and analyzed their enzymatic properties. Eight positions were targeted, including the putative six conserved catalytic residues identified by sequence alignments (Fig. 1). In the absence of a full recombination assay available in vitro, we have tested the different mutants for their binding to target sequence, covalent complex formation, and cleavage/religation properties. We found that Int SSV has a type IB topoisomerase activity, a property shared by some other tyrosine recombinases (28 -31). All of the mutants generated alter the enzymatic properties of Int SSV , indicating that the residues targeted are likely involved in the catalytic process. The conservative changes of some catalytic residues in Int SSV , with regard to the consensus, may reflect an adaptation of the active site to thermophily and/or may be the signature of the primitive active site that evolved to give the tyrosine recombinase and topoisomerase IB catalytic pockets (18,32). Importantly, in vitro complementation assays between Y314F and all other Int SSV inactive mutants restored both cleavage and DNA relaxation activities, suggesting that as for the yeast Flp recombinase, the Int SSV active site could be assembled by dimer association. The shared active site model may therefore be extended to an archaeal tyrosine recombinase.

MATERIALS AND METHODS
Bacterial Strains-Escherichia coli strain DH5␣ was used for cloning and plasmid amplification, except for pGEM-SN180 and pGEM-P207, which were constructed in strain One shot Top10 (Invitrogen). Strains TG1 and RZ1032 were used for M13-based site-directed mutagenesis. Overexpression of recombinant wild-type or Int SSV mutants was carried out in strain MC1061. Strains TG1 and RZ1032 were grown in 2YT medium and all others in Luria-Bertani medium supplemented with 100 g/ml ampicillin when necessary.
Site-directed Mutagenesis and Protein Purification-The R304H mutant was obtained by using the QuikChange site-directed mutagenesis kit (Stratagene) according to the supplier's instructions, with plasmid pSG3 (26) as a template. The others mutations were introduced by M13 site-directed mutagenesis as described previously (33) but with T 7 DNA polymerase for the elongation step. The nucleotide sequence of the different mutated fragments was verified by sequencing (Applied Biosystem, ABI 370) prior to recloning in the expression vector pSG3. The integrase mutants were expressed and purified as described previously (26), with addition of a protease inhibitor mixture (Sigma) in the different buffers up to the gel filtration step.
Negatively supercoiled plasmids used in the relaxation assay were purified on a CsCl gradient. Positively supercoiled DNA was obtained by incubating 2.5 g of negatively supercoiled pKSattB with purified Thermotoga maritima reverse gyrase (kindly provided by C. Bouthier de la Tour). The reaction was done in 50 mM Tris, pH 8.0, 1 mM ATP, 10 mM MgCl 2 , 120 mM NaCl, 0.5 mM dithiothreitol, and 30 g/ml BSA for 30 min at 85°C. The sample was then phenol extracted and DNA ethanol precipitated. The positively supercoiled plasmid was resuspended in water at a final concentration of 100 ng/l. For band shift experiments the P207 and SN180 PCR products were purified by using the Nucleospin Extract kit (Macherey-Nagel). The P207 fragment was then 5Ј-end labeled on both strands using T 4 polynucleotide kinase and [␥-32 P]ATP, 5000 Ci/mmol (Amersham Biosciences). Unincorporated nucleotides were removed by spin dialysis.
DNA Binding Reactions and Electrophoretic Mobility Shift Assay-The DNA binding reactions were carried out in 20 l of a mixture composed of 10 nM 5Ј-end-labeled P207 substrate, 0.2 M wild-type or mutant Int SSV , in a binding buffer composed of 50 mM Tris, pH 7.5, 125 mM NaCl, and 1 mM EDTA. Incubation was performed for 1 h at room temperature. 5 l of 5ϫ loading buffer (10 mM Tris, pH 7.5, 20% glycerol, 1 mM EDTA, 0.1 mg/ml BSA, and 0.1% xylene cyanol) was then added to the reaction mix. The DNA binding reactions were loaded onto an 8% nondenaturating acrylamide gel containing 5% glycerol. Electrophoresis was performed in 1ϫ TGE buffer (50 mM Tris, pH 7.5, 8 mM glycine, 0.1 mM EDTA) at 4°C for 4 h at 7.5 V/cm. The DNA-protein complexes were visualized by autoradiography and phosphorimaging.
Filter Binding Assay-12.5 nM 5Ј-end-labeled XTB substrate was incubated at room temperature for 1 h with 2 M wild-type or mutant Int SSV in the binding buffer described previously. Each reaction mixture was spotted under vacuum on a nitrocellulose membrane (Protran 0.2 m, Schleicher & Schuell) and the spot washed three times by 20 l of binding buffer. The membrane was heated 2 h at 80°C, and the protein-DNA complexes were visualized by phosphorimaging. The binding efficiency of each protein was quantified using the ImageQuant version 1.2 software. The results presented are the mean of three independent experiments.
Cleavage Assays-For covalent complex formation, 12.5 nM 5Ј-endlabeled XTB substrate was incubated with 1 M wild-type or mutant integrase in 30 mM Hepes, pH 7.5, 50 g/ml BSA, and 125 mM NaCl for 3 h at 65°C. Reactions were stopped by the addition of Laemmli loading buffer (final concentrations: 40 mM Tris, pH 6.8, 3% SDS, 8% glycerol, 250 mM ␤-mercaptoethanol, 0.005% bromphenol blue) and heating for 5 min at 98°C. The reaction products were analyzed by electrophoresis through 12% SDS-polyacrylamide gel, and covalent complex formation was visualized by autoradiography and phosphorimaging of the gel.
Determination of the cleavage position was performed by incubating 1 or 2 M wild-type or mutant Int SSV with 12.5 nM 3Ј-end-labeled XTB substrate in the conditions described above. Reactions were stopped by the addition of formamide dye (97.5% deionized formamide, 10 mM EDTA, 0.3% bromphenol blue, 0.3% xylene cyanol blue) and heating for 5 min at 98°C. One-fifth of each reaction was loaded on an 18% polyacrylamide gel (19:1) containing 8 M urea, in TBE buffer (90 mM Tris, 90 mM boric acid, 1 mM EDTA), and electrophoresis was performed FIG. 1. Assignment of Int SSV catalytic residues by alignment with members of the tyrosine recombinase family. The original alignment was performed by ClustalX with 29 protein sequences including other thermophilic tyrosine recombinases. The consensus sequence results from this alignment. Only the catalytic signatures (Box I, K ␤ , Box II) of crystallized tyrosine recombinases are represented together with Int SSV . Int SSV Arg 240 rather than Lys 243 was considered to be the Lys ␤ equivalent based on alignment of neighboring residues. Substitutions were considered conservative when following the Dayhoff exchange groups (Ser, Pro, Ala, Gly, and Thr; Ile, Leu, Met, and Val; Asp, Glu, Asn, and Gln; Phe, Trp, and Tyr; Arg,, Lys, and His; Cys). Black boxed residues, 100% identity; dark gray boxed residues, Ն80% conservation; light gray boxed residues, Ն60% conservation. White letters, residue identity Ն80%. Within the consensus, boxed residues correspond to catalytic positions with less than 100% identity. Asterisk (*), any residue within an exchange group (Ն60%). at 52 V/cm. Cleavage products were visualized by autoradiography and phosphorimaging of the gel.
DNA Relaxation Assay-12.5 nM plasmid pGEM-P207, pGEM-SN180, or pKSattB was incubated with the indicated amount of Int SSV (wild-type or mutant) in 30 mM Hepes, pH 7.5, 50 g/ml BSA, and 125 mM NaCl for 3 h at 65°C. poly(dI-dC)⅐poly(dI-dC) was used as nonspecific competitor when indicated. After incubation, samples were treated with SDS (0.5% final) and 10ϫ loading buffer (100 mM EDTA, 5% SDS, 40% glycerol, and 0.35% bromphenol blue) was added. Reaction mixes were loaded on a 2% agarose gel, and electrophoresis was performed in 1ϫ TEP buffer (90 mM Tris phosphate, 1 mM EDTA) at room temperature for 4 h at 3.5 V/cm. DNA was visualized by staining with ethidium bromide.
Relaxation assays with calf thymus topoisomerase IB (Invitrogen) were done as follow. 12.5 nM plasmid was incubated for 30 min at 37°C with 1 l of topoisomerase (corresponding to 5-15 units) in 50 mM Tris, pH 7.5, 50 mM KCl, 10 mM MgCl 2 , 0.5 mM dithiothreitol, 0.1 mM EDTA, and 30 g/ml BSA in a final volume of 50 l. Reactions were stopped and analyzed as described above.
To monitor the activity of Int SSV on positively or negatively supercoiled substrates, the reaction products were analyzed by two-dimensional agarose gel electrophoresis (34). After the first dimension was performed as described above, the second dimension was run in a perpendicular direction in TEP buffer containing 3 g/ml chloroquine for 15 h at 0.8 V/cm. Chloroquine was removed by washing the gel in water for at least 3 h. The distribution of topoisomers was then visualized by staining with ethidium bromide.

Mutagenesis Strategy and Production of Int SSV Mutants within the Conserved Motifs of Tyrosine Recombinases-
The signature of tyrosine recombinases is the conservation of two short regions of similarity in their C-terminal part (1,2,4,5). Within these regions, six residues are strictly conserved and were shown to be directly involved in catalysis by biochemical studies and crystal structures of tyrosine recombinases and topoisomerases IB (for a review, see Ref. 3). Alignment of Int SSV with members of the tyrosine recombinase family suggested that they correspond to residues Arg 211 , Arg 240 , Lys 278 , Arg 281 , Arg 304 , and Tyr 314 in Int SSV ( Fig. 1 and Table I). All of these positions were subjected to site-directed mutagenesis, as well as two other less conserved positions Glu 214 and Lys 282 , which correspond to positions conserved at more than 80% within the tyrosine recombinases conserved boxes I and II (Fig.  1). The two latter residues were shown to play a role in the catalytic process mediated by the Flp recombinase (35)(36)(37). Within Int SSV three of the six conserved residues are divergent from the tyrosine recombinases consensus, namely Arg 240 , Lys 278 , and Arg 304 . We have replaced all divergent residues with the consensus residues, and in addition we have introduced either aliphatic or residues with an opposed charge at these positions (Table I). Residues Glu 214 and Lys 282 were targeted with the same rationale.
None of the mutations introduced altered the level of expression nor the purification behavior of proteins (Fig. 2), although two mutants showed an increased sensitivity to proteolysis (R281L and R304L). All mutant proteins were therefore purified as described previously for wild-type Int SSV (26) with the addition of a mixture of inhibitors in all buffers up to the chromatographic step.
All Int SSV Mutants Are Still Able to Bind the att Sites-Int SSV activity depends on its ability to recognize and bind specific DNA sites. We analyzed the binding efficiency of Int SSV on a substrate that carries the POPЈ core sequence of the attP site (Fig. 3A) by electrophoretic mobility shift assay (Fig. 3B). In the absence of competitor DNA, three protein-DNA complexes of retarded electrophoretic mobility can be observed, referred as to CI, CII, and LMC (low mobility complexes), the major species obtained corresponding to the CII complex (Fig.  3B, lane 2). Adding increasing amounts of nonspecific DNA (SN180) to the reaction leads to the disappearance of the LMC to the benefit of CII (Fig. 3B, lanes 3 and 4). At a molar ratio of up to 1:20 (specific:nonspecific), the CI and CII complexes are also competed (lane 5). The apparent affinity of Int SSV for its core site is relatively low but in the same order of magnitude as that observed for binding of the -Int to its POPЈ site in the absence of IHF (38). A simple interpretation of this profile is that CI could reflect the binding of one monomer of Int SSV to one "core-type" site (P or PЈ) and CII the binding of a dimer to both core-type sites (P and PЈ). The heterogeneous third complex, LMC, could either represent higher order structures of CI and/or CII or more likely be generated by nonspecific binding of more than two monomers of Int SSV along the substrate.
To verify whether the Int SSV mutants were still able to bind DNA, we analyzed their ability to generate protein-DNA complexes with attP. Given the apparent low affinity of Int SSV for this substrate, the experiment was realized in the absence of competitor DNA. The DNA binding profile of most of the mutants is similar to that of the wild-type protein (Fig. 3C). However, the ratio of complexes over the free substrate depends on the mutation introduced. Mutants E214N, R240K, R281L, K282L, R304H, R304L, and Y314E behave as the wildtype protein with regard to the amount of CII formed. Mutants K278H, K278L, and K282H generate a somewhat reduced amount of CI and CII. At the same time the LMC disappear as the free substrate amount increases. The last class of mutants (R211L, R211Q, E214K, R240L, and Y314F) generate less CI

SSV1 Integrase Mutants
and CII but about the same amount of LMC as the wild-type protein. Among these, some mutants retain a large amount of substrate in the wells, possibly by nonspecific aggregation (see E214K, R240L, and R281L in Fig. 3C). Nevertheless, all of the mutants generally conserved the property to generate CI and CII, and CII is systematically the most abundant, suggesting that it is the more stable Int SSV -DNA association. Int SSV Has a Topoisomerase IB Activity-Previous work on Int SSV described an in vitro test for recombination (24,39). Unfortunately, like others, we were unable to reproduce the published data. While attempting to set up new conditions for an in vitro recombination assay, we found that Int SSV has a DNA relaxation activity (Fig. 4) as some other tyrosine recombinases (28 -31). As for recombination, DNA relaxation requires strand cleavage and strand religation steps. The optimum for relaxation activity is observed when incubating 2 M Int SSV with 12.5 nM supercoiled DNA substrate for 2 h at 65°C in the presence of 125 mM NaCl, without ATP or MgCl 2 . Unlike XerC and XerD (29), the relaxation activity is not stimulated by the presence of glycerol in the assay. Incubation of purified Int SSV with the plasmid pGEM-P207, containing the attP site, led to the appearance of DNA topoisomers (Fig. 4A, lane 3). Neither phenol extraction nor proteinase K treatment of the samples altered the migration pattern, indicating that there were no protein-DNA covalent complexes present at the end of the reaction (data not shown). The appearance of topoisomers reflects that Int SSV is able to promote relaxation of a supercoiled DNA. On time course experiments, the amount of each topoisomer increased, but their distribution did not vary, and gels containing chloroquine revealed that the totality of substrate DNA was at least partly relaxed (data not shown). This strongly suggests that Int SSV is more distributive than processive for DNA relaxation.
The influence of the substrate topology on Int SSV relaxation activity was also tested. Interestingly, Int SSV relaxes both negatively and positively supercoiled plasmids (Fig. 4C) even though the relaxation of both substrates is incomplete (compare II and IV in Fig. 4C). Relaxation of negative and positive supercoils is specific for type IB topoisomerases, ruling out any possible contamination by an E. coli topoisomerase. So far, only Int SSV and -Int (30) have been shown to behave specifically as type IB topoisomerases, although a type I activity was demonstrated for both Xer proteins (29) and Cre (28). This is consistent with the idea that tyrosine recombinases and type IB topoisomerases evolved from an ancestral common catalytic module.
The Int SSV DNA relaxation activity is not site-specific because it can also be observed with plasmid pGEM-SN180, which does not contain any att site (Fig. 4B, lane 3). Once again, this behavior is similar to that of -Int (30). Addition of poly(dI-dC)⅐poly(dI-dC) as nonspecific competitor DNA (Fig. 4,  A and B, lanes 4 -7) was slightly more inhibitory when no att   FIG. 2. Purification of Int SSV mutants. The purified wild-type (WT) and Int SSV mutants were analyzed by SDS-PAGE (12% acrylamide). 1.5 g of each sample was loaded on the gel. Proteins were visualized by Coomassie staining. Lane MW, molecular mass markers (values in kDa are indicated on the left). The faint band migrating around 30 kDa corresponds to an Int SSV proteolytic product that appears as the purified protein decays. site was present on the substrate (compare lanes 4 -7 in the A and B panels). This most likely reflects a better affinity of Int SSV for pGEM-P207 than for pGEM-SN180 rather than a site-specificity for the topoisomerase activity.
Although no strand exchange occurs, the topoisomerization reaction requires strand cleavage and strand religation, as in a complete recombination process. In the absence of a reliable recombination assay we decided to use the relaxation assay, in addition to the binding and cleavage assays, as a reporter of activity for Int SSV and its mutants.
Most Int SSV Mutants Are Deficient in DNA Relaxation-The ability to promote strand cleavage and religation was evaluated for all mutants by using the relaxation assay. As expected, the Y314F mutant is unable to relax supercoiled DNA, once again ruling out a topoisomerase contaminant in the protein samples. Although most Int SSV mutants are inactive in the relaxation assay, three mutants (R240K, K278H, and R304H) retain a low but significant level of activity (Fig. 5). Interestingly, these mutants correspond to the restoration of the tyrosine recombinases consensus within the catalytic motif. In addition, mutant K278L shows a faint level of activity (Fig. 5), suggesting that position Lys 278 would not be strictly essential for DNA relaxation. A reason for the loss of activity induced by most of the mutations could be that the mutants are less thermostable than the wild-type enzyme. All proteins were purified according to the procedure set up for the wild-type, which contains a thermal denaturation step (26), without affecting the purification yield. It thus seemed unlikely that for all of these point mutations the loss of activity could be the result of a decrease in thermostability. Nevertheless, we tested the activity of wild-type and some mutant proteins at lower temperatures (37, 45, and 55°C). In any case we did not observe any change in activity compared with incubation at 65°C (data not shown). This indicates that mutations impair more catalysis than thermostability. Moreover, none of the three mutations restoring the consensus at a catalytic residue (R240K, K278H, R304H) shows a better activity at lower temperature. One possible interpretation is that thermal adaptation of the catalytic pocket requires more than one of these adjustments in the charge relay system.
Most Int SSV Mutants Are Impaired in Covalent Complex Formation but Not in Their Ability to Bind a Minimal Substrate-We have shown previously that incubation of Int SSV with a synthetic minimal 19-bp substrate (referred as to XTB) leads to the formation of a covalent complex, resulting from cleavage of a phosphodiester bond and creation of a phosphotyrosine intermediate (26). This property was used to evaluate the effect of the mutations on the cleavage step of the reaction. Only one mutant, R304H, is still able to generate protein-DNA covalent complex, although with a low efficiency corresponding to about 9% of the wild-type activity (see Fig. 7B). All other Int SSV mutants are unable to catalyze covalent complex formation (see Fig. 7B), although in some cases a faint signal with a lower mobility than the protein-DNA covalent complex can be detected (see for example mutant E214K in Fig. 7B). Strikingly, mutants that retained a low DNA relaxation activity were deficient in covalent complex formation, suggesting that the cleavage of a small synthetic substrate may be a more stringent assay than the relaxation assay. Indeed, it was shown previously for the Flp recombinase that some mutants showing activity on large substrates were deficient in cleavage activity when tested on small synthetic substrates (40).
Because the lack of covalent complex formation could arise from defective binding to the substrate, we checked whether Int SSV mutants were able to bind XTB by using a filter binding assay. The results presented in Fig. 6 show that most mutants bind the synthetic substrate quite efficiently. Two mutants showed an important defect in substrate binding (R240K and K278L), the latter being totally unable to retain XTB on a filter assay. Quite surprisingly, K278L was able to bind a larger DNA substrate although less efficiently than the wild-type enzyme (see Fig. 3C). This suggests that the K278L substitution alters the DNA binding properties of Int SSV and that this effect is more drastic on the short 19-bp substrate. The other substitution made on the same position (K278H) does not significantly alter XTB binding but still affects the cleavage activity, as it is the case for most of the mutants. Remarkably, all other mutants are unable to promote covalent complex formation (Fig. 7B) even though they can bind significantly to the synthetic substrate (Fig. 6). This indicates that all of the residues targeted are likely involved in the cleavage reaction, suggesting that Int SSV follows the mechanism proposed for tyrosine recombinases and topoisomerases IB (32).
Complementation between Mutants Restores Both Strand Cleavage and DNA Relaxation Activity-Several lines of evidence suggest that at moderate ionic strength Int SSV behaves as a dimer in solution. 2 This led us to test whether the active species of Int SSV could be dimeric rather than monomeric, in other words, whether Int SSV could follow the shared active site model (13,14). We therefore set up in vitro complementation assays with the collection of defective mutants (Fig. 7).
One fully inactive mutant of each targeted residue was coincubated with the Y314F mutant in the relaxation assay (Fig.  7A). The activity was restored in all cases, indicating that Y314F can complement all of the mutants affected on any other position. On the other hand, coincubation of Y314F with Y314E (Fig. 7A) and all others mutant combinations did not restore the relaxation activity (data not shown). Furthermore, the addition of 1 mM tyrosine in the relaxation assay with Y314F did not produce topoisomers (data not shown) making it unlikely that a loose tyrosine residue from a monomer may interfere in the reaction. This strongly suggests that, for one considered catalytic site, Tyr 314 is brought by a monomer, and all the other residues are provided by the second one. In the complementation experiment three different types of dimers can be formed, two homodimers and a heterodimer. Only the heterodimer is expected to be a catalytically active species with one active site restored. It is thus not surprising to observe that the combination of mutants involving 2 M total proteins in the reaction is less efficient in relaxation than 1 M wild-type integrase (Fig. 7A).
Covalent complex formation is also restored when any mutant is mixed with the Y314F mutant (Fig. 7B). Interestingly, complementation is efficient between Y314F and K278L, which is the only mutant unable to bind the XTB substrate (Fig. 6). Heterodimer formation may therefore compensate for the binding defect of K278L. This suggests that protein-protein interactions established between two monomers contribute to stabilize the dimer at least on the synthetic substrate. Four mutants, E214N, R240K, R304H, and R304L, restore the wildtype level of cleavage when complemented by Y314F (Fig. 7B). These mutants bind P207 as efficiently as the wild-type (Fig. 3) and except for R240K also bind efficiently the XTB substrate (Fig. 6), suggesting again that protein-protein contacts are important for proper protein-DNA interactions and therefore catalysis. The behavior of Lys 282 mutants is more puzzling. Both mutants of this position bind the XTB substrate with a good efficiency (Fig. 6) and efficiently produce the CII complex with the P207 substrate (Fig. 3). However, the two mutations introduced at position 282 yield a lower level of complementation than the other mutants for covalent complex formation (Fig. 7B). The corresponding residue in Flp is His 309 . The Flp 2 M. C. Serre, unpublished data.  (26). The 5Ј-endlabeled strand is indicated by an asterisk. The ability of wild-type and Int SSV mutants to bind XTB was measured by a filter binding assay. Protein-DNA complexes were visualized by phosphorimaging and quantified by using the ImageQuant software. The data were normalized by taking the amount of complexes formed by wild-type as 100%. structure reveals that this residue is not involved in DNA contacts. His 309 from one Flp monomer rather interacts with His 345 from the second monomer, which brings the catalytic Tyr 343 . The His 309 -His 345 interaction was proposed to allow a correct positioning of Tyr 343 toward the scissile phosphate in the Flp catalytic pocket (12). If the mode of dimer assembly is conserved between Flp and Int SSV , the Lys 282 residue would then be critical for stabilizing the dimer in an active conformation. Heterodimers K282L/His-Y314F could then have a loose interface that may alter their stability and thus their catalytic efficiency.
The complementation reaction was analyzed further by verifying that the cleavage specificity was not altered when mixing two mutants (Fig. 7C). The small synthetic substrate XTB was 3Ј-end labeled on the top strand and incubated with wild-type, mutants, and mutant mixes. As was pointed out previously (26), the amount of products observed with the 3Ј-end-labeled substrate is lower than that obtained for the 5Ј-end-labeled substrate in all cases. Analysis of the reactions on a denaturing gel shows that the cleavage product obtained in the mixing experiments is the same as that obtained with the wild-type Int SSV (Fig. 7C), indicating that complementation between active sites mutants leads to the correct cleavage site specificity. The simplest interpretation for these results is that Int SSV active site assembly proceeds through monomer association.
The results obtained in the different complementation assays are consistent with the hypothesis that Int SSV requires protein-protein interactions to assemble active sites, either by assembling two shared active sites within a dimer or, like Flp, four active sites within a tetramer (12). DISCUSSION We report here the characterization of 15 Int SSV mutants. The targeted residues were comprised in the canonical active site motif common to tyrosine recombinases and topoisomerases IB (Table I). Crystallographic data and mechanistic analysis support the hypothesis that these two enzyme families would have evolved from an ancestor catalytic domain with a strand transferase activity (41). In a previous work we have shown that like other virus-encoded integrases, Int SSV cleavage FIG. 7. Complementation between Int SSV mutants in relaxation and cleavage assays. A, functional complementation between Int SSV mutants in the relaxation assay. The supercoiled pGEM-P207 plasmid was incubated with 0.5 or 1 M wild-type Int SSV (WT 0.5 and WT 1) or with a 1 M concentration of each indicated mutant integrase alone or in combination with 1 M Y314F mutant. C indicates that no protein was added. Supercoiled and open circular forms are indicated by arrows (FI and FII). B, restoration of covalent complex formation between the XTB substrate and Int SSV mutants. The sequence of the XTB substrate is indicated at the bottom. The overlap region (O) is boxed and corresponds to the anticodon loop sequence of the tRNA. Arrows indicate the cleavage positions (26). The 5Ј-end-labeled strand in the assay is indicated by an asterisk. 12.5 nM XTB was incubated for 3 h at 65°C with 1 M wildtype or mutant integrases. When indicated, 1 M Y314F mutant was added in the assay. Covalent complex formation was revealed by SDS-PAGE analysis and phosphorimaging. Note that a small but reproducible amount of covalent complex was obtained with the R304H mutant alone, representing about 9% of the wildtype activity. sites specifically target the border of the anticodon loop of a tRNA gene and that cleavage is mediated via a 3Ј-phosphotyrosine intermediate (26). However, several discrepancies in the conserved catalytic residues raised the possibility that the catalytic process would be different in the archaeal kingdom (27). The results presented here provide evidence to unravel this point.
First, we have shown that the wild-type Int SSV is able to promote DNA relaxation. The relaxation activity does not require ATP or MgCl 2 and is monitored both on positively or negatively supercoiled DNA, a property specific for topoisomerases IB. As for the -Int (30) and unlike Cre or XerC/XerD (28,29) the relaxation activity is not dependent on a specific DNA site. Detection of relaxation activity for Cre, Flp, -Int, and Xers proteins is possible only under conditions where (almost) no recombination occurs (28 -31). For Int SSV no recombination products would be expected from the substrates used in the relaxation assay. These data are consistent with the idea that the DNA relaxation activity of tyrosine recombinases is more likely a bypass of the recombination reaction.
The second point of evidence for mechanism conservation in Int SSV is provided by the properties of the mutants studied here. None of the mutations abolished the DNA binding properties of Int SSV , although one mutant (K278L) was totally defective for binding to a small synthetic substrate, and some mutants may be impaired in establishing protein-protein interactions (K282H, K282L). However, mutations at any of the conserved positions of the catalytic pocket strongly alter Int SSV activities. All mutants but one (R304H) are unable to generate a protein-DNA covalent complex on a small synthetic substrate. This is consistent with the proposed role in catalysis for these conserved residues (4,22,35,37,(42)(43)(44)(45)(46)(47). Hence, Arg 211 , Lys 278 , and Arg 281 in Int SSV could coordinate the scissile phosphate, thus activating it for nucleophilic attack by Tyr 314 . Residue Arg 240 , corresponding to the conserved Lys ␤ , is also critical for Int SSV activity. Replacement of Arg 240 by a leucine abolishes the relaxation activity, whereas the R240K mutation only reduces it. Furthermore, complementation of R240K by Y314F on the small synthetic substrate allows more covalent complex to be formed than that observed for R240L. This indicates that a positive side chain at this position is important for catalysis. Mutants made on equivalent residues in vaccinia topoisomerase IB (48), XerD (6), and more recently Flp (20) have the same behavior, i.e. binding to substrate but deficiency in cleavage and relaxation or recombination, suggesting that Arg 240 could fulfill the role of Lys ␤ in the Int SSV active site. As for tyrosine recombinases (20) and topoisomerases IB (45,47), residue Arg 304 corresponding to the conserved His/Trp is essential for Int SSV activity. Although the R304L mutant is unable to cleave a synthetic substrate and relax a supercoiled substrate, the R304H mutant retains a slight level of cleavage (Fig. 7B) and has a low relaxation activity (Fig. 5). The increased sensitivity of R304L toward proteolysis as well as altered fluorescence spectra for R304H and R304L 2 suggest that this position is important for proper folding of the protein. However, whether Arg 304 is involved in docking the helix (or structural motif) providing the catalytic tyrosine as for Flp (20) or in forming a hydrogen bond to the nonbridging oxygens of the scissile phosphate as for the human topoisomerase IB (19) cannot be inferred from our experiments. Mutants at position Glu 214 bind the different substrates with almost the same efficiency as wild-type, indicating that a negatively charged side chain at this position is not required for substrate recognition. Furthermore, the complementation experiment indicates that E214K is less efficient in cleavage than E214N, suggesting that the length of the side chain at this position may be important for correct dimer assembly. Even though the strong conservation of this acidic residue in tyrosine recombinases and different mutational analysis (35,43) suggested that it is involved in catalysis, crystal structures indicate that its role would rather be architectural by maintaining the active site geometry via interactions with the nearby conserved Arg I . Finally, mutants at residue Lys 282 behave as mutants at the equivalent positions in the R and Flp recombinases (36,37). Furthermore, the behavior of the Lys 282 mutants in the complementation assays is consistent with the proposed role of this residue in Flp (12), i.e. correct positioning of the catalytic tyrosine mediated by interactions with a residue located on the helix that delivers the tyrosine.
Changes of conserved residues within the putative Int SSV active site may be the signature of a chemical adaptation to thermophily. Indeed the conserved histidine residues of tyrosine recombinases are replaced by arginine or lysine residues in Int SSV . The pK a decrease of bases at high temperatures, correlated with the pK a variability of buried histidines (49) could account for these replacements in Int SSV catalytic pocket. Around 80°C lysines and arginines could therefore have a pK a close to the pK a of histidines at lower temperature, thus being chemically equivalent for catalysis. It should be noted however that thermophilic Xer-like proteins do not have this signature, although there is no biochemical evidence for an enzymatic activity of these proteins (50). Another possible interpretation for the His 3 Arg/Lys replacements in Int SSV is that they could be the remnant of the primitive catalytic module that evolved to give the topoisomerase IB and tyrosine recombinase catalytic pockets.
Finally, the results obtained in the complementation assays strongly suggest that Int SSV would follow the shared active site paradigm. The shared active site model has been controversial for a long time (51)(52)(53)(54). For Flp, the recent crystal structures (12,55) are consistent with trans delivery of the catalytic tyrosine, which was inferred from complementation assays (13,56). The trans activity of Cre (57) was ruled out by crystallographic data (8,23), and further bulk experiments provided evidence for cis cleavage (58). In the case of -Int, two different enzymatic analyses showed divergent results with regard to the cis versus trans origin of the catalytic tyrosine (17,59). The first structural data did not allow discrimination because the crystal obtained was disordered in the region of the catalytic tyrosine (11). More recently, the structure of -Int bound on a suicide substrate shows that on a substrate allowing binding of only one monomer the catalytic tyrosine is delivered in cis (10).
It was suggested that trans cleavage may be artifactual (54), either because the substrates used were suicide substrates, and may thus alter the protein-DNA interactions, or because mixing two mutant proteins would allow another residue to play the role of a surrogate nucleophile in the cleavage reaction. As was pointed out recently, the use of an assay scoring not only cleavage but also the strand-joining event can avoid artifacts in the determination of cis or trans cleavage (60). We show here that cleavage specificity, covalent complex formation, and relaxation activities of Int SSV can be restored when mixing mutants of each conserved residues with the Y314F mutant. This is the only combination allowing complementation, making it unlikely that another residue may be a surrogate nucleophile in the assays. Furthermore, the relaxation activity requires not only cleavage of the DNA substrate but also strand religation. The level of complementations observed are also consistent with a ''half of the sites'' activity. Indeed, the wild-type level of plasmid relaxation is never reached when mixing two mutants. By contrast, full complementation can be observed in the assay measuring covalent complex formation. In this latter case, the rationale of the experiment consisted in measuring a stoichiometric activity with a large excess of protein with respect to the 19-bp oligonucleotide. Some mixing reactions give lower amounts of products than wild-type, suggesting that on small synthetic substrates the stability of the Int SSV dimer alone or in complex with DNA is important for activity. The simplest interpretation of these results is that the Int SSV active site is assembled at a dimer interface with Tyr 314 provided by one protomer and all other residues targeted here by another protomer. However the exact architecture of the intasome, i.e. Int SSV assembled on DNA for efficient recombination, cannot be assessed by the assays used in this study. Therefore, a cis-acting mechanism for the full recombination process cannot be totally excluded. A final answer to this question will be brought by structural analysis of Int SSV . If indeed an archaeal tyrosine recombinase is following the shared active site assembly, the positioning of the catalytic tyrosine with respect to the scissile phosphate may have evolved further in the eukaryal and eubacterial kingdoms. Stabilization of the site assembly in the eukaryal kingdom may have occurred by the addition of a C-terminal module improving contacts between protomers, whereas in the eubacterial kingdom the helix swapping may have been lost, leading to the cis cleavage mechanism.