Synapsis and strand exchange in the resolution and DNA inversion reactions catalysed by the β recombinase

Abstract

In the presence of a sequence-independent chromatin-associated protein, such as Hbsu or HMGB, the β recombinase catalyses resolution between two directly oriented recombination sites (six sites) and both resolution and DNA inversion between two inversely oriented six sites. Assembly of the synaptic complex requires binding of the β recombinase to the six sites and the presence of Hbsu. Whether resolution or inversion will take place depends on the relative orientation of the two six sites, the level of DNA supercoiling and the amounts of Hbsu. In this work, the topologies of the products of the resolution and inversion reactions were analysed. The resolution reaction generated mainly singly catenated DNA circles, while DNA inversion gave rise to unknotted circles and small amounts of DNA molecules containing 3- or 5-noded knots. In spite of the distinctive features of the β system, the topology of synapsis and strand exchange during the resolution reaction is similar to that of Tn3 and γδ resolvases. The ability of the β recombinase to catalyse both inversion and resolution reactions probably reflects different possible architectures of the synaptic complex, which to a large extent depends on Hbsu.

INTRODUCTION

Site-specific DNA recombination reactions are catalysed by polynucleotidyl transferases that are specific to each reaction system (1). The two major families of site-specific recombinases (tyrosine and serine recombinases) use different mechanisms for cutting and rejoining the DNA strands at the recombination crossover sites. The tyrosine recombinases complete the cleavage, exchange and rejoining of one pair of DNA strands with the generation of a Holliday junction as a recombination intermediate, before initiating the same set of reactions on the other pair of DNA strands. If the recombination sites are on different DNA molecules, the net result of the process is the integration of the two molecules into a single one. However, if the recombination sites are on the same DNA molecule, the recombination reaction can lead to either the deletion or the inversion of the DNA sequences located between the crossover sites, depending on their relative orientation and the number of interdomainal supercoils trapped on synapsis (reviewed in 2–4).

Two subfamilies of serine recombinases are known. One includes small recombinases (<250 amino acids long) that primarily catalyse intramolecular recombination reactions (4). The second subfamily comprises large recombinases (>400 amino acids long) that catalyse both inter- and intramolecular recombination events (5). The small serine recombinases can be classified into three distinct groups depending on whether they catalyse resolution (DNA resolvases), inversion (DNA invertases) or both (resolvase-invertases). Recombination proceeds via a concerted four-strand cleavage and rejoining mechanism (3,4,6). Resolvases (γδ and Tn3 as prototypes) selectively catalyse excision between two directly oriented res sites. The res site, 114 bp in length, contains three adjacent subsites with dyad axis symmetry, named I, II and III. Each subsite binds a resolvase dimer. Assembly of a synaptic complex requires binding of three resolvase dimers to each res site and DNA supercoiling. No additional factors are needed. Three highly condensed supercoils are trapped in the synaptic complex through interactions of subsites II and III. Recombination takes place at the centre of subsite I (4,6,7). DNA invertases (Hin and Gin as prototypes) specifically catalyse inversion between two inversely oriented recombination sites. Invertase dimers bound to these sites can interact to form an inactive synaptic complex. The assembly of the productive synaptic complex, trapping two interdomainal supercoils, requires a 65 bp enhancer located on the same DNA molecule, to which FIS binds (8–10). Mutant derivatives of DNA resolvases and DNA invertases that are permissive for inter- and intramolecular events have been isolated. These activated mutant recombinases bypass the requirement for a complex synaptic structure to initiate recombination and can thus catalyse both resolution and inversion (11–13).

Resolvase-invertases (β and Sin as prototypes) are peculiar because they do not show the same high selectivity as resolvases and invertases for resolution or inversion (14,15). The β recombinase, for example, can catalyse resolution between two directly oriented recombination sites and both resolution and inversion between two inversely oriented sites (16). The resolution reaction requires a supercoiled DNA substrate, but inversion does not (16). Both of these reactions may be important for the biological role of β recombinase in plasmid maintenance (17,18). In contrast to resolvases and DNA invertases, binding of the β recombinase to two complete recombination sites is not sufficient to generate a synaptic complex. Productive synapsis requires the presence of a chromatin-associated protein, namely the bacterial Hbsu/HU or eukaryotic HMGB protein, which is believed to provide the appropriate topological arrangement of the recombinase dimers and the DNA (15,19–21). The target of the β recombinase, the 90 bp six site, contains two subsites (Fig. 1). Subsite I has dyad symmetry, with the crossover point localised at the centre. Subsite II shows no evident symmetry (22,23). A dimer of the β recombinase binds to each subsite, although the protein–DNA interactions at subsites I and II are markedly different (22). Binding of the β recombinase bends the DNA at both subsites and enhances loading of the Hbsu or HMGB protein (24).

Figure 1.

Recombination substrates used. (A) Plasmids pCB8 and pCB12, containing two recombination sites (six sites) in either direct or inverse orientation, respectively, are represented. The six sites are depicted as open and filled arrows. Restriction sites relevant for the analysis of the recombination products are indicated. (B). Orientation and sequence of the six sites present in the plasmids used in this work. Each six site is composed of two subsites, named I and II, which are binding sites for the β recombinase. Subsite I has a dyad symmetry and is represented by arrows. Subsite II has no obvious symmetry and is represented by an open box. The length of the DNA segments (in bp) is indicated in Arabic numbers. The sequence of the crossover site, which is located at the centre of subsite I, is shown for each six site of the plasmids used. Plasmid pCB8 and pCB12 have two identical wild-type six sites. Plasmids pCB81 and pCB141 have one copy of a wild-type six site and another copy of a six site with an A→C substitution at the crossover site. The relative orientation of each site is indicated with open or filled arrows.

It was not clear whether the distinctive properties of the β recombinase are due to a greater flexibility in the structure of the synapse (or in its assembly pathway) or to flexibility in the strand exchange mechanism (e.g. allowing a specific synaptic structure to generate both inversion and resolution products). This question has now been addressed by analysing the topology of recombination between wild-type and mutant six sites, present in direct or inverse orientation. The results suggest that a degree of flexibility in the structure and assembly of the synaptic complex, rather than any difference in strand exchange, distinguishes β recombinase from prototypical resolvases and DNA invertases.

MATERIALS AND METHODS

DNA substrates, proteins and enzymes

Plasmids pCB8 and pCB12 (14) (Fig. 1A) contain two six sites in direct and inverse orientation, respectively. Plasmids pCB81 and pCB141 are equivalent to pCB8 and pCB12, respectively, but contain an A→C mutation at the central dinucleotide of the crossover site in one of the six sites (23,25) (Fig. 1B).

The β and Hbsu proteins were purified and their concentrations determined as previously described (19,26). Since both proteins are dimers in solution, their concentrations are expressed as mol protein dimers. Topoisomerase II and DNase I were obtained from commercial sources.

In vitro assays for site-specific recombination

Reaction mixtures contained 10 nM plasmid DNA, 50 nM purified β recombinase, 50, 300 or 400 nM purified Hbsu protein (as specified), 10 mM bis,tris-propane-HCl, pH 7.0, 10 mM MgCl₂, 10 mM NaCl, 1 mM DTT, in a total volume of 25 µl. Reactions were initiated by the addition of β protein and incubated at 30°C for 30 min for plasmids with directly oriented six sites (pCB8 and pCB81) or for 90 min for plasmids with inversely oriented six sites (pCB12 and pCB141). To analyse the recombination products, the DNA was digested with two restriction enzymes, one cutting the vector and the other cutting the region lying between the two six sites (PstI andSalI for pCB8 and pCB81, EcoRV and KpnI for pCB12 and pCB141).

Topological analysis of the recombination products

The products of the recombination reactions were either digested with appropriate restriction enzymes to quantify recombination efficiency or nicked with DNase I to remove supercoils and analyse the knot or catenane structure of the products. DNase I reactions were stopped by addition of EDTA (20 mM final concentration) and proteinase K (300 µg/ml final concentration) and incubated for 30 min at 37°C. Samples were then loaded onto 0.7% (w/v) agarose gels and separated by electrophoresis in the absence of ethidium bromide (EtBr) for 24 h at 50 V in TAE running buffer (14). The number of knot or catenane nodes in the reaction products was deduced by comparing their migration to that of a marker ladder of topoisomers (shown in Figs 2 and 3) or knots and catenanes generated by treating the same substrate DNA with topoisomerase II or by comparison to authentic 2-noded and 4-noded species generated by Tn3 resolvase (data not shown). DNA was visualised using UV illumination (254 nm) and photography after staining the gels with EtBr (0.5 µg/ml).

Figure 2.

Topology of the resolution products of β recombinase. The β recombinase (50 nM) and Hbsu (50 nM) proteins were incubated with either pCB8 or pCB81 (10 nM) for 30 min at 37°C. (A) The substrates (lanes 1 and 3) and the reaction products (lanes 2 and 4) were digested with SalI and PstI and separated in a 0.8% agarose gel. Lin, linear product; NR, non- recombinant molecules; RP, resolution products. The small 0.5 kb SalI DNA segment is not shown. (B) The different topoisomers present in the undigested substrates (lanes 1 and 3) and recombination products were resolved in 0.8% agarose gels (lanes 1–4 and 7–10), in parallel with a ladder of topoisomers obtained by partial treatment of plasmid pCB8 with topoisomerase II (lanes 5 and 6). In lanes 7–10, the DNA was nicked by a partial treatment with DNase I. Numbers on the right indicate the migration of knots or catenanes containing 0, 2, 4 or 8 nodes; sc denotes the position of supercoiled DNA. (C) Predicted topology of the reaction products if three supercoils are trapped. One round of recombination (180°) would generate a 2-noded catenane. A second round of strand exchange (360°) would generate a non-recombinant 4-noded knot. Based on Stark et al. (28).

Figure 3.

Topology of the reaction products obtained with a supercoiled or relaxed plasmid containing two inversely oriented six sites. The β recombinase (50 nM) and Hbsu (300 nM) proteins were incubated with either pCB12 or pCB141 (10 nM) for 90 min at 37°C, the substrates and reaction products digested with KpnI and EcoRV and separated in a 0.8% agarose gel. (A) The supercoiled substrates (lanes 2 and 4) and the reaction products (lanes 3 and 5) were digested and separated. (B and C) The different species present in the undigested substrates and recombination products were resolved in 0.8% agarose gels in parallel with a ladder of topoisomers obtained by partial treatment of plasmid pCB12 with topoisomerase II (lanes 7 and 8 in panel B and lane 5 in panel C). Where indicated, the DNA was nicked by a partial treatment with increasing concentrations of DNase I. (D) The relaxed substrate (lane 2) and the reaction product (lanes 3) were digested with KpnI and EcoRV and separated in a 0.8% agarose gel. (E) The topoisomers present in the undigested substrates and recombination products were nicked by a partial treatment with increasing concentrations of DNase I (lanes 1–4) and resolved in 0.8% agarose gels in parallel with a ladder of topoisomers obtained by partial treatment of plasmid pCB12 with topoisomerase II (lane 5). Lin, linear product; IP, inversion products; NR, non-recombinant molecules; RP, resolution products. Numbers on the right of the figure indicate the migration of topoisomers containing 0, 2, 3, 4 or 5 supercoils; sc and lin denote the position of supercoiled and linear DNA, respectively.

RESULTS AND DISCUSSION

Topology of recombination between directly repeated six sites

Both in vivo and in vitro, β recombinase catalyses resolution between two directly oriented 90 bp six sites in the presence of a chromatin-associated protein, such as Hbsu or HMGB (19–21,23). To analyse the topology of a resolution reaction catalysed by the β recombinase, supercoiled pCB8 DNA was used. This plasmid contains two identical six sites in direct orientation, separated by ∼2.7 kb (see Fig. 1). The extent of the in vitro recombination reaction was followed by digesting the reaction products with endonucleases SalI and PstI (14) (Fig. 2A). When the result of the recombination reaction was analysed by electrophoresis without restriction enzyme digestion, a complex mixture of species was observed; the major product migrated slightly faster than the supercoiled plasmid substrate (Fig. 2B, lanes 1 and 2). The topological complexity of the products was analysed by nicking them with a limiting amount of DNase I and resolving the products by gel electrophoresis. This eliminates all DNA supercoils, except those trapped as knot or catenane nodes by the recombination reaction (27,28). As shown in Figure 2B (lanes 7 and 8), the major reaction product migrated as a 2-noded molecule. Taking into account the restriction products observed in Figure 2A, which clearly indicate a resolution reaction, this molecule must be a 2-noded catenane. Small amounts of 4-noded and more complex products were also observed, but no significant amount of unlinked resolution circles. This suggests that strand exchange takes place in a synaptic complex topologically equivalent to those assembled by Tn3 and γδ resolvases, with three interwound negative (–) supercoils trapped between the crossover sites (15,27–29). Additional minor products are also observed with these recombinases, and are thought to be the consequence of successive rounds of strand exchange. A second round generates a non-recombinant 4-noded knot (see Fig. 2C) and additional rounds of strand exchange generate more complex products (15,27–30).

Elucidation of the strand exchange mechanism for the Tn3 and γδ resolvases was facilitated by the use of recombination substrates in which the two recombination sites differed at one position in the central dinucleotide of the crossover site (28). Synaptic complexes can be formed with these substrates, but after the DNA strands are cleaved and rotated by 180°, the mismatch between the crossover sites is revealed, preventing correct base pairing in the recombinant. The mismatch induces a second 180° rotation that gives a correctly base paired, non-recombinant 4-noded knot (28) (see Fig. 2C). To test whether the β recombinase followed a similar reaction scheme, plasmid pCB81 was used as recombination substrate. This plasmid contains two directly oriented six sites, but one of them contains an A→C substitution at the central dinucleotide of the crossover site (see Fig. 1B). As a consequence, the β protein, with the help of Hbsu, is able to form a synaptic complex but unable to make any recombinant product (23) (see Fig. 2A, lanes 3 and 4). Nevertheless, incubation of supercoiled pCB81 with β recombinase and Hbsu did generate a complex mixture of novel products (Fig. 2B, lanes 3 and 4). Nicking of these products with DNase I showed that the 2-noded catenane was not present, while molecules migrating as 4-noded or 8-noded knots were abundant (Fig. 2B, lanes 9 and 10). This result strongly supports the idea that resolution by β recombinase follows the same topological pathway as is used by Tn3 and γδ enzymes, as representatives of the resolvase subfamily, and by Sin of the resolvase-invertase subfamily (15,27–30).

Unlike the Tn3/γδ systems, which have three recombinase binding sites within res, the resolvase-invertases have only two recombinase binding subsites (I and II), within a 90 bp recombination site. Therefore, the protein–DNA architecture of the synapse must be different for DNA resolvases and for resolvase-invertases (see Introduction). Our data indicate that resolution by β recombinase proceeds through a synaptic complex with a precise and unique topology, trapping three supercoils, as in the Tn3 resolvase system. Strand cleavage, followed by a right-handed 180° rotation and religation, would give the observed resolution product, a 2-noded catenane. The topological changes seen with β recombinase, notably the ‘double-round’ knotting reaction seen with a ‘mismatch’ resolution substrate, are consistent with rotational models for strand exchange, in which the rotation of the DNA half-sites is coupled to rotation of two subunits (or parts of two subunits) of a recombinase tetramer bound at the two crossover sites (13,27).

Topology of recombination between inversely oriented six sites

As stated above, when acting on a supercoiled plasmid containing two inversely oriented six sites, the β recombinase can catalyse either resolution or inversion of the DNA segment between the two six sites. When the amounts of Hbsu are low (3–10 Hbsu dimers/DNA molecule), resolution is favoured, while at higher Hbsu concentrations (>30 Hbsu dimers/DNA molecule) inversion predominates (14,16). This suggests that two different synaptic complexes can form, depending on the availability of Hbsu, each one having a distinct geometry. To investigate the mechanism of the inversion reaction, the topology of the products was analysed using as substrate plasmid pCB12, which contains two inversely oriented six sites (see Fig. 1). Negatively supercoiled pCB12 DNA was incubated with β recombinase and ∼30 Hbsu dimers/DNA molecule, which allows the visualisation of both excision and DNA inversion. Digestion of the reaction products with endonucleases KpnI and EcoRV indicated that both resolution and inversion had occurred (Fig. 3A, lanes 2 and 3). The recombination products were incubated with DNase I to nick the DNA and resolved by agarose gel electrophoresis. As seen in Figure 3B (lanes 4–6), 2-noded catenane was the most abundant product, and this presumably accounts for most of the resolution products seen by restriction (Fig. 3A, lane 3). Unlinked circular resolution products were not detected, but the minor 4-noded product may also be associated with resolution. The inversion reaction was significantly more efficient than resolution, as judged by restriction (Fig. 3A, lane 3). However, the observed knotted products (Fig. 3B, lane 4) do not account for this high level of inversion. Thus, the major product of the inversion reaction must be an unknotted circle, which co-migrates with the pCB12 substrate. Minor products with the mobility of 3- and 5-noded knots (Fig. 3B, lane 4; highlighted with a star, although the 5-noded knot is barely visible) could represent alternative topologies for the inversion reaction or products generated by multiple rounds of strand exchange, as previously described for DNA invertases (31).

The behaviour of a plasmid (pCB141) containing two inversely oriented six sites, one of them containing an A→C substitution at the central dinucleotide of the crossover site, was also investigated. It was predicted that neither the resolution nor the inversion synapses would generate recombinant molecules from such a substrate. As shown in Figure 3, incubation of supercoiled pCB141 with β recombinase and Hbsu generated exclusively non-recombinant products (Fig. 3A, lanes 4 and 5, and Fig. 3C, lanes 1–4). The 2-noded catenane was not detected, but there was an increased yield of 4-noded knot, compared to reactions with pCB12. This suggests that in the resolution reaction between inversely oriented sites, the β recombinase uses the same topology as with a conventional resolution substrate and that the mismatch induces a double round of strand exchange, giving the 4-noded knot. The experiment provided no further insight into the topology of the inversion reaction.

Under the conditions used above it is difficult to study separately the resolution and inversion reactions. An interesting property of the β recombinase allowed us to get around this problem. This recombinase requires a supercoiled DNA substrate for efficient resolution, but supercoiling is not required for DNA inversion (16,23). Therefore, by using a relaxed plasmid containing two inversely oriented wild-type six sites, the resolution reaction is markedly reduced, while the inversion reaction is not. Relaxed plasmid pCB12 was incubated with β recombinase and ∼40 Hbsu dimers/DNA molecule. Restriction analysis of the reaction products showed that 35–40% of the input substrate was inverted, while resolution was not detectable (Fig. 3D, lane 3). When the reaction products were analysed after nicking with DNase I, no major novel species could be detected. The reaction generated very small amounts of species migrating as 3-noded knots and minute amounts of 5-noded knots (Fig. 3E). This implies that the major product of the inversion reaction is an unknotted circle that co-migrates with the nicked substrate. Consistent with the absence of resolution, 2-noded products were not detected (Fig. 3E, lanes 2–4). Thus, the inversion reaction catalysed by the β recombinase resembles that of the DNA invertases Gin and Hin in that an unknotted DNA circle is the primary product. The reaction differs in showing no requirement for a (–) supercoiled substrate or an enhancer element and in its requirement for multiple recombinase binding sites (23). The current data do not make clear whether the β recombinase traps two supercoils in the synapse, as in the DNA invertase systems, or whether zero interdomainal supercoils are trapped. It is clear, however, that the inversion synapse cannot have the same local structure as the resolution synapse, with three interwound supercoils, as this would necessarily give knotted inversion products, as seen with Tn3/γδ resolvase (7).

A model for the β recombinase synaptic complex

Resolvases, DNA invertases and resolvase-invertases are all related serine recombinases. By means of different protein– protein and protein–DNA interactions they form distinctive synaptic complexes (reviewed in 4,6,18). Resolvase mutants have been described that can catalyse both resolution and inversion reactions on DNA substrates containing only subsite I (the crossover site), but not subsites II and III, which are normally required for synaptosome assembly (12). Similarly, mutants of the Gin invertase are known that are independent of the FIS protein and have also lost their selectivity for inversely oriented recombination sites (11). This suggests that resolvases and DNA invertases could use the same strand exchange mechanism, the selectivity being imposed by the constraints on formation of the synaptic complex (see 8,32).

Current evidence suggests that the core of the synapse is formed by a tetrameric arrangement of recombinase dimers at the crossover sites (13) and there is also experimental evidence for functionally relevant interactions between resolvase dimers through the crystallographic 2,3′ interface (4,33). The role of subsites II and III for resolvases and of the FIS enhancer for the DNA invertases would be in the selection of appropriately oriented recombination sites. Two general models have been proposed for the structure of the synaptic complexes formed by resolvases and DNA invertases. In one, the DNA would lie on the inside of the complex and the synaptic interactions would most likely involve the DNA-binding domains of the recombinases, rather than the catalytic domains (10,34). Alternatively, the DNA could be on the outside of the synaptic complex (13,16,27,33).

In the case of the β recombinase, available information about the protein–DNA interactions at the six sites (22), the effects of mutations in the recombinase and at the crossover site (16,22,23) and the topological properties of the reaction reported here are fully consistent with a mechanism of strand exchange similar to that reported for DNA resolvases and DNA invertases. Therefore, the distinctive properties of the β recombinase may reside in the structure and assembly of synaptic complexes, rather than in the mechanism of strand exchange.

Accordingly, we propose models of the synapse that are consistent with the available information. Selectivity of the recombinase for directly or inversely oriented six sites and the generation of resolution or inversion products would be determined by the conformation imposed on the DNA by subsite II and Hbsu (see Fig. 4). The proposals are consistent with a number of experimental findings and inferences: (i) in the presence of a sequence-independent DNA-bending protein, such as Hbsu or HMBG, four β dimers at the two directly oriented 90 bp six sites, separated by an intervening sequence as short as 99 bp on a (–) supercoiled substrate, recombine with a wild-type frequency; (ii) Hbsu or HMGB bends the spacer DNA between subsites I and II; (iii) the two β dimers at the symmetric subsite I synapse (16,19–23).

Figure 4.

Proposed models for β-mediated recombination in the presence of Hbsu. (A and B) The two six sites for the β recombinase are represented. The binding sites for β recombinase are indicated with filled arrows (subsite I) or filled boxes (subsite II). The length of the DNA segments (in bp) is indicated in Arabic numbers. The β recombinase dimers, which interact differently with subsites I and II (see 22), are represented with filled spheres. The horizontal arrows above the six site denote the relative orientation of the recombination sites. In (A)–(D) the lower case letters are used to position the respective subsite in the synaptic complex. Speculative models for synaptic complexes containing β recombinase and Hbsu are presented for DNA substrates containing two six sites in direct orientation (C) or in inverse orientation (D). The large spheres correspond to β recombinase, while small spheres denote the Hbsu protein. The model for the resolution synapse shows the three trapped (–) supercoils, while that for the inversion synapse assumes that no supercoils are trapped.

In the resolution reaction, we propose that the β recombinase is activated by an interwound synapse; resolution is thus stimulated by (–) supercoiling (Fig. 4C). The deduced synapse topology, with three interwound supercoils, is characteristic of enzymes that function as resolvases (reviewed in 7, 15). The proposed arrangement of the recombinase subunits resembles a recent model for the Sin system (15) and a subset of the arrangement proposed for the Tn3/γδ system (13). In the inversion reaction, we suggest that the β recombinase is activated by a synapse with a different (but related) structure (Fig. 4D), perhaps explaining why inversion can use relaxed or (–) supercoiled substrates at high Hbsu concentrations. The resolution and inversion reactions are relevant for different aspects of the proposed biological function of the β recombinase, in plasmid replication and stable segregation (17,18). Using different synaptic structures to regulate these two reactions could perhaps be an important adaptive feature of the system. The third type of reaction, resolution between inverted sites [stimulated by (–) supercoiling and limiting concentrations of Hbsu] has no obvious biological relevance. Our data suggest that it proceeds through a degenerate structure that resembles the standard resolution synapse.

The topology of the β recombinase inversion reaction is most likely different from the DNA invertase systems and the nucleoprotein architecture will differ in many respects. In the DNA invertase systems, the arrangement of the recombinase dimers at the two crossover sites is likely to be similar to that for resolvases and resolvase-invertases bound at subsite I. However, the FIS protein bound to the enhancer is thought to contact the DNA invertase dimers in the invertosome, defining the topology of the active synapse. FIS could contact amino acid residues near the dimer interface to induce a conformational change that coordinately brings the catalytic sites close to the scissile phosphodiester bonds (10,34). In contrast to the DNA invertase systems (see 35), the topology of the β synaptic complex would be fully defined by the structure of the six sites, through synapsis of the two β dimers at subsite I and the two β dimers at subsite II, assisted by Hbsu or HMGB (Fig. 4D). The models are consistent with the observation that the HMGB protein from different sources (e.g. mammalian or plant origin) stimulates DNA inversion (and resolution) by β recombinase (20,21,36). This suggests that specific contacts between the recombinase and the chromatin-associated protein are not essential for recombination to occur (15,20).