The active-site interactions involved in the catalysis of DNA site-specific recombination by the serine recombinases are still incompletely understood. Recent crystal structures of synaptic γδ resolvase–DNA intermediates and biochemical analysis of Tn3 resolvase mutants have provided new insights into the structure of the resolvase active site, and how interactions of the catalytic residues with the DNA substrate might promote the phosphoryl transfer reactions.
Introduction to site-specific recombination
Conservative site-specific recombination is a process whereby DNA is cut at two specific sites and the ends are rejoined to new partners. Each site-specific recombination system encodes a recombinase enzyme that recognizes the sequence of the appropriate sites and catalyses DNA strand exchange within the sites. Site-specific recombination differs from homologous recombination in that extensive DNA sequence homology within the region of strand exchange is not required. No synthesis or degradation of DNA is involved (hence the term ‘conservative’), and no high-energy co-factors are required .
Two families of conservative site-specific recombination systems can be identified, on the basis of similarity of the recombinase sequences and reaction mechanisms [1,2]. Catalysis takes place in a synaptic complex involving a protein tetramer, and recombination proceeds via a protein–DNA covalent intermediate (Figure 1A). The tyrosine recombinases use an invariant active-site tyrosine residue to attack the DNA phosphodiester in the DNA-cleavage step of the recombination reaction, whereas the serine recombinases use an invariant serine residue. Cleavage is believed to occur through the in-line nucleophilic displacement of a DNA hydroxy leaving group by the hydroxy group of the tyrosine or serine side chain, via a pentaco-ordinate phosphorus transition state [1,3]. Re-ligation of the DNA strands is proposed to be by a similar mechanism, with the nucleophile being a DNA hydroxy group and the leaving group the tyrosine or serine hydroxy group (Figure 1A illustrates these steps for the serine recombinases).
Strand exchange by Tn3 resolvase
The serine recombinases include transposon resolvases (for example, Tn3, γδ and Tn21), DNA invertases (for example, Gin, Hin and Cin) and phage integrases (for example, φc31, Bxb1 and TP901) . Members of the serine recombinase family can be identified by alignment of the amino acid sequences of their catalytic domains. They are typically more selective than the tyrosine recombinases with respect to the reactions that they catalyse, usually giving characteristic changes in DNA topology and specific types of recombination product . DNA cleavage occurs at the centre of the ‘crossover site’, producing 2 bp staggered double-strand breaks with recessed 5′ ends and a 5′ DNA–phosphoseryl intermediate. Cleavage is followed by a rearrangement of the DNA ends, then re-ligation by attack of a DNA 3′-hydroxy group on each phosphoseryl group  (Figure 1A). The topological changes in the DNA observed in a number of systems after resolution and inversion can be accounted for if rearrangement of the DNA ends in the post-cleavage intermediate is by a right-handed 180° rotation of one pair of half-sites relative to the other pair. The strands are then re-ligated, and the product complex dissociates to release the recombinant sites.
Recombination by Tn3 resolvase
Tn3 resolvase functions in replicative transposition to split co-integrate intermediates into two circles, by catalysing a site-specific recombination reaction between the two copies of the transposon in the cointegrate. The res recombination site (114 bp) acted upon by Tn3 resolvase binds three resolvase dimers at the three sites I, II and III shown in Figure 1(B). Resolvase (185 amino acids) has two functional domains (Figure 2): an N-terminal domain (residues 1–140), which contains the residues that form the active site for catalysis of strand exchange and is involved in subunit interactions, and a C-terminal domain (residues 141–185), which makes a sequence-specific interaction with a motif at each end of the dimer-binding sites in res. Recombination occurs in a synaptic complex of two resolvase-bound res sites, which therefore contains 12 resolvase subunits in total. The recombining res sites must be in the same molecule in direct repeat, and the DNA must be negatively supercoiled . The synapse can be regarded as comprising two basic functional units. The catalytic unit (or ‘site I synapse’), where the DNA strands are cleaved, rearranged and rejoined consists of a resolvase tetramer bridging the two copies of binding site I. The regulatory unit formed by the synapsis and intertwining of the two copies of the resolvase-bound sites II/III controls the assembly and activation of the catalytic unit (Figure 1C). Structural information is available for the catalytic unit [6–8] (see below), but not as yet for the regulatory unit.
Structure of a synaptic tetramer of a γδ resolvase activated mutant with two cleaved site Is
The proposed mechanism of strand exchange via subunit rotation (Figure 1A) is supported by topological evidence  and more recently by structures of cleaved resolvase–DNA covalent complexes [7,8] (Figure 2). The two halves of the site I synaptic complex in these structures are separated by a flat interface that is essentially hydrophobic. Li et al.  proposed that this interface allows the two halves of the cleaved intermediate to rotate relative to each other without major steric hindrance.
Chemical strategies in phosphoryl transfer reactions
The simplest strategy to effect phosphoryl transfer would be via acid–base catalysis, as used in reactions of the tyrosine recombinases and other enzymes involved in nucleic acid transactions . Figure 3(A) shows a scheme that depicts the minimum interactions that would be required for this mechanism. One key element is the need for a general base to activate Ser10 for nucleophilic attack on the phosphorus atom of the scissile phosphodiester; the Ser10 hydroxy group is expected to be a poor nucleophile in the absence of a base to remove a proton from it. The rejoining step will likewise require the generation of an effective nucleophile from the 3′ hydroxy group of the DNA half-site. A general acid may also be required in order to generate good leaving groups by protonation of the deoxyribose 3′ oxygen or the phosphoseryl oxygen [3,9]. The residues that fulfil these proposed acid–base functions have not been identified for the serine recombinases.
The Tn3 resolvase active site
As proposed for most phosphoryl transfer reactions involving phosphodiester groups, the serine recombinases are expected to catalyse recombination via an associative mechanism  involving a pentaco-ordinated phosphorus transition state in which the incoming nucleophile and the departing leaving group are both covalently attached to the phosphorus atom. (In contrast, in a dissociative mechanism, the leaving group would partially or completely break its attachment to phosphorus before the incoming nucleophile attacks .) The negatively charged transition states of enzyme-catalysed phosphoryl transfer reactions are usually stabilized by bivalent metal ions and/or positively charged residues in the active site . Interestingly, however, resolvase-catalysed recombination does not require the participation of metal ions [10,11].
Studying catalysis in an activated resolvase background
Our knowledge of how the catalytic residues of the resolvase active site promote the chemical steps of recombination is still very incomplete. One reason for this is the difficulty of studying catalysis within the highly regulated context of the wild-type system. The kinetics are very complex because of the many pre-catalytic steps involved in building the 2-res synapse (Figure 1C), mutations of candidate catalytic residues may have ancillary effects on accessory site synapsis and regulation, and the use of chemically modified substrates is difficult because of the requirement for long recombination sites and supercoiled DNA. Nevertheless, a number of highly conserved residues have been characterized as being involved in catalysis by studying the properties of mutants . However, even in the recent structures of reaction intermediates (see above), some of these putative active-site residues are not in close contact with the DNA, and how they might come together to form a functional active site is unclear.
To circumvent some of these problems, we used a variant version of Tn3 resolvase to investigate the putative catalytic residues that participate in recombination. Certain ‘activated mutant’ resolvases do not require sites II and III, directly repeated sites or negative supercoiling . These enzymes can recombine linear oligonucleotide substrates containing just the site I sequence, and form stable resolvase–site I synapses that can be observed by gel electrophoresis. By studying mutants of one of these activated Tn3 resolvase variants, we identified the core catalytic residues that are essential to DNA cleavage and rejoining . In this study, we introduced a new assay to simplify further the analysis of catalysis. The mechanism for recombination outlined in Figure 1(A) involves eight chemical reaction steps (cleavage of four DNA strands followed by their re-ligation), so the kinetics of recombination are likely to be complex even for the minimal site I×site I reaction. In our new assay , which uses a modified site I DNA substrate, we can observe a single chemical reaction (cleavage of a single strand by attack of the resolvase catalytic serine residue; Figure 1D).
The core catalytic residues involved in DNA cleavage
The three highly conserved arginine residues Arg8, Arg68 and Arg71, located close to Ser10 (Figure 3B) have for some time been regarded as core catalytic residues [1,5,14]. This cluster of arginine residues could help to mediate polarization of P–O bonds, an essential element of phosphoryl transfer via an associative mechanism (Figure 3A) . The crystal structures  show the guanidinium moiety of the Arg8 side chain in contact with the two non-bridging oxygen atoms of the phosphate group which is covalently linked to Ser10, and the Arg68 side chain contacting one of the oxygen atoms. Arg71 does not interact with the scissile phosphate group in the crystal structures, but is close enough to do so at some stage of the reaction. Consistent with polarization roles for these residues, loss of the Arg8 side chain leads to complete abolition of cleavage activity, whereas only traces of activity were seen with the equivalent Arg68 and Arg71 mutants . The structural and biochemical data thus suggest a more essential catalytic role for Arg8 than for Arg68 and Arg71. We hypothesize that this role might include deprotonation of Ser10 (see below), as well as stabilization of transition states by polarization of non-bridging P–O bonds.
On the basis of the results of our biochemical analysis and on the structural data, we proposed a relay mechanism to deprotonate the hydroxy group of Ser10, involving the three conserved residues Arg8, Tyr6 and Asp36 . The guanidinium group of Arg8 is proposed to act as the primary base. More detailed biochemical analysis will be required to test this model. Loss of either the Tyr6 or the Asp36 side chain results in a more than 103-fold reduction in reaction rate, consistent with a catalytic role for these residues, which interact via hydrogen bonds. Unexpectedly, however, Tyr6 and Asp36 also have a role in synapsis; mutants at these positions retain DNA-binding proficiency, but the stability of the site I synapse is markedly reduced. We propose that this is evidence for coupling of synapsis with assembly of the active-site residues on the substrate. The Tn3 resolvase tetramer in the site I synapse is only formed in the presence of DNA , and its stabilization might thus depend on specific interactions between active-site residues and the site I DNA. Substrate-induced assembly of a functional active site could be a pervasive strategy for imposing specificity on otherwise promiscuous phosphoryl transfer catalysts.
Resolvase has two well-conserved glutamine residues near the active-site serine residue, Gln14 and Gln19, whose functions remain rather mysterious. Gln19 might have a non-catalytic structural role; it interacts with Arg8 and is close to Arg68, and could help to position these residues optimally for catalysis in the active site [3,7,15]. Replacement of Gln19 with the bulky positively charged residues lysine or arginine abolished catalytic activity, as did replacement with the isosteric but negatively charged glutamate, which also caused severe destabilization of the site I synapse . The other well-conserved glutamine residue, Gln14, contacts neither DNA nor neighbouring residues in the crystal structures. Replacement of Gln14 with the positively charged lysine or arginine resulted in loss of DNA cleavage and recombination activity, but increased yields of the site I synapse, suggesting that Gln14 and its mutant side chains might be close enough to interact with the negatively charged DNA phosphodiester backbone . Further biochemical analysis will be needed to define more precisely the roles of these conserved glutamine residues.
A number of other well-conserved charged or polar resolvase residues (including Ser39, Arg45, Glu118, Arg119, Glu124 and Arg125) are apparently less important than the aforementioned ones for catalysis in an activated resolvase background , but preliminary evidence suggests that some of these residues may be essential for regulated recombination by the wild-type enzyme (F.J. Olorunniji and W.M. Stark, unpublished work). It will be interesting to compare and contrast the behaviour of mutants of these residues in activated and wild-type resolvase to gain insight into their roles in the wild-type recombination system.
Having identified the principal participant amino acid residues in the chemical mechanism of Tn3 resolvase (and, by extension, other serine recombinases), the challenge remains to assign definite functions to these residues and construct a detailed catalytic scheme for the system. Understanding the mechanism of this important class of enzymes will undoubtedly enhance our more general appreciation of how enzymes can mediate phosphoryl transfer reactions. It is also an essential prelude to elucidation of the currently mysterious processes that regulate activity in natural serine recombinase systems to ensure that only biologically appropriate recombination reactions take place.
Machines on Genes: Enzymes that Make, Break and Move DNA and RNA: A Biochemical Society Focused Meeting held at Robinson College, Cambridge, U.K., 12–14 August 2009. Organized and Edited by Richard Bowater (University of East Anglia, U.K.), Ben Luisi (University of Cambridge, U.K.) and Marshall Stark (University of Glasgow, U.K.).
We are very grateful to Marko Prorocic and Jan-Gero Schloetel for a critical reading of the manuscript draft.
This work is supported by the Biotechnology and Biosciences Research Council [grant number BB/E022200] and a UK Commonwealth Academic Staff Scholarship Award to F.J.O.