A remarkable feature of the serine resolvases is their regulation: the wild-type enzymes will catalyse intra- but not inter-molecular recombination, can sense the relative orientation of their sites and can exchange strands directionally, despite the fact that there is no net release of chemical bond energy. The key to this regulation is that they are only active within a large intertwined complex called the ‘synaptosome’. Because substrate topology greatly facilitates (or, in other cases, inhibits) formation of the synaptosome, it acts as a ‘topological filter’. Within the defined topology of the synaptosome, strand exchange releases supercoiling tension, providing an energy source to bias the reaction direction. The regulatory portion of this complex contains additional copies of the recombinase and sometimes other DNA-bending proteins. We are using a combination of X-ray crystallography, biochemistry and genetics to model the full synaptic complex and to understand how the regulatory portion activates the crossover-site-bound recombinases.
Site-specific recombinases rearrange DNA by cutting and pasting at defined locations. There are two large groups of such enzymes, which are generally referred to as the serine and the tyrosine recombinase families, based on their active-site nucleophiles . Although both store the energy of the scissile phosphodiester bonds as covalent protein–DNA intermediates, their structures and strand-exchange mechanisms are entirely unrelated. Nevertheless, their biological roles overlap greatly, with members of both families acting as bacteriophage integrases, DNA invertases and DNA resolvases.
The present paper focuses on the regulation of a pair of serine resolvases: Sin and Tn3. Both convert a large circular DNA molecule into two smaller ones, but for different reasons. Sin is found on certain large multi-resistance plasmids from Staphylococcus aureus [2,3]. It presumably aids in stable plasmid maintenance by converting plasmid dimers into monomers (such dimers can be formed by RecA-mediated restart of stalled replication forks). Tn3 resolvase is encoded by the Tn3 transposon of Escherichia coli. This mobile element uses a replicative transposition mechanism that results in fusion of the original donor and recipient molecules with a full copy of the transposon at each junction. Tn3 resolvase acts at a site within each copy of the transposon, separating the donor and recipient molecules, but now with each product circle containing a copy of the transposon.
Extensive topological, biochemical and structural studies have elucidated the mechanism of strand exchange for serine resolvases (Figure 1 and reviewed in ). The two crossover sites are bound by dimers of the recombinase that, when activated, come together in a synaptic tetramer. The active-site serine residues attack the scissile phosphate groups, displacing the 3′ hydroxy groups. This creates double-strand breaks with 2 nt 3′ overhangs, and with each DNA half-site covalently linked to a protein catalytic domain via a 5′ phosphoserine bond. Two subunits then rotate 180° relative to the other two, realigning the broken ends, which are religated by the attack of the free 3′ hydroxy groups on the phosphoserine bonds. The structure of an activated post-DNA cleavage form of γδ resolvase (which is closely related to Tn3 resolvase) showed that the central interface of the tetramer, about which this rotation should occur, is large, hydrophobic and unusually flat, so that it would pose relatively small kinetic barriers to rotation while resisting complete dissociation .
Recombination sites and synaptosome cartoons
The activity of serine resolvases is tightly regulated, and the wild-type protein normally exists as an inactive dimer. The structure of wild-type γδ resolvase bound to crossover site DNA shows the complex in a pre-cleavage state, where substantial conformational changes would be needed to dock the scissile phosphate groups near the active-site serine residues . Large conformational changes are also needed to morph from this inactive dimer to the post-cleavage activated tetramer structure. Although the system may include many sub-states, our working hypothesis is that the enzyme can see-saw between two primary states: an inactive dimer and an active tetramer. The equilibrium normally lies on the dimer side, but the balance can be tipped by the natural regulatory complex described below. It can also be tipped by mutation: numerous activating mutations have been selected that allow recombination to occur in the absence of the regulatory complex [6–9]. Such mutations were used to obtain the activated tetramer structure.
Regulation is important for several reasons. First, double-strand breaks are dangerous to genomic integrity, so their introduction and lifetime should be minimized. Secondly, the crossover sites themselves can be fully symmetrical, yet to create the proper products they must be synapsed in one relative orientation rather than the other. Thirdly, resolution requires intramolecular recombination of two sites on the same DNA circle. Recombination between sites on different replicons would create fusion products rather than resolve them. Finally, despite the substrate and product being isoenergetic in terms of bond energy, these resolvases can drive the product/substrate ratio beyond 50%.
The key to regulation is that, under normal circumstances, serine resolvases are only active in the context of a larger multisubunit protein–DNA complex termed a synaptosome (Figure 1). For Sin, the full recombination site includes not only the crossover site, but also binding sites for a DNA-bending protein and for a second dimer of Sin . In vivo, the DNA-bending protein is presumably HU, but it can be replaced in vitro by IHF (integration host factor), which is closely related, but displays much higher sequence specificity . For Tn3 resolvase, the full recombination site binds two additional dimers of the recombinase . In both cases, the half sites of these ‘accessory’ recombinase dimer-binding sites are arranged differently from how they are at the crossover site, and the proteins bound to them remain chemically inactive.
Also in both cases, the synaptosome is formed by two full recombination sites wrapping around one another with a defined topology: three right-handed nodes are trapped (Figure 1B) [12–14]. Formation of such a complex between two recombination sites within the same DNA circle is favoured by negative supercoiling, which naturally results in such right-handed crossings, whereas formation of such a complex between sites on two different molecules would be topologically hindered. This tightly defined synaptic complex also brings together the two crossover sites in the correct alignment for productive recombination. Finally, it harnesses supercoiling as an energy source for directionality: within the context of the synaptosome, a 180° right-handed rotation of subunits at the crossover site releases four negative supercoils.
Sin–site II structure
We wanted a structural model of the synaptosome to better understand how it functions. Sin was chosen as the best system for this because it is the most compact: the full recombination site comprises only ~90, rather than 114 or more bp as in the case of Tn3/gd resolvase. Furthermore, only one additional structure was required to have models in hand for all of the subcomplexes within the synaptosome. The structure of an IHF–DNA complex is known , and because Sin and γδ resolvase are 30% identical, the Steitz group's crossover-site-bound γδ structures provide reasonable models for the Sin–crossover site complex. Therefore we co-crystallized Sin bound to its regulatory site, ‘site II’, to acquire the remaining piece of the jigsaw puzzle.
The Sin–site II structure is shown in Figure 2(A) . The catalytic domain forms a dimer very similar to that seen in inactive wild-type γδ resolvase structures. The organization of the DNA and DNA-binding domains, however, is necessarily quite different because the individual half-sites that make up site II are in direct, rather than inverted, repeat. The protein can accommodate this varying geometry by rotating about a very flexible point in the last helix of the N-terminal catalytic domain. As with γδ resolvase, the section of this helix past the hinge interacts with the minor groove of the DNA, bending it away from the catalytic domains. However, the Sin–site II DNA complex is more asymmetric than the γδ resolvase–crossover site structures, and only one such interaction occurs within the dimer. If the Sin and γδ catalytic domain dimers are superimposed, the DNAs are nearly perpendicular to one another.
Building a structural model of the Sin synaptosome
Two copies of site II can be synapsed by two dimers of Sin. Multiple protein–protein interactions help hold the crystal lattice together (Figure 2B). However, the dimer–dimer contact that buries the largest surface area is also the only one for which the DNA crossing has the appropriate sign to fit in a synaptosome model. This tetramer is mediated by unexpected interactions between DNA-binding domains, and differs dramatically from the activated γδ resolvase tetramer, which is mediated by the catalytic domains. The relevance of this interaction was supported by an elegant screen for mutations that interrupt synapsis of two site IIs: all of the ‘hits’ map to the new protein–protein interface (Figure 2C) [9,16].
Since there is no reason to doubt that an activated Sin–crossover site complex will include a catalytic-domain-mediated tetramer similar to that formed by γδ resolvase, it appears that Sin is a versatile protein that forms different protein–protein contacts when bound to different DNA sites. The DNA-binding-domain-mediated tetramer probably forms only at site II because of that site's special geometry: the direct repeats position the DNA-binding domains so that identical contacts are formed between both pairs of binding domains. This lends a synergy to these interactions that would not be possible for crossover-site-bound dimers, where, because of the different relative orientations of the DNA-binding domains, only one such DNA-binding-domain-mediated protein–protein interaction could occur.
A model for the synaptic complex was built using individual structures of the activated γδ resolvase–crossover site complex, the IHF–DNA complex and the Sin–site II complex (Figure 3) . Docking these structures together as rigid bodies produced a model that shows how the Sin synaptosome can trap three highly condensed supercoiling nodes between partners while bringing the two recombining sites together in close proximity and in the correct alignment. The site II-bound tetramer acts as a binder clamp to hold the central crossing together, while IHF introduces the necessary sharp DNA bends.
Model of the Sin synaptosome
How does the regulatory part of the synaptosome activate the crossover-site-bound proteins? The simplest hypothesis is Mass Action: if the tetramerization is required for catalytic activity, then the regulatory complex may tip the conformational equilibrium from two dimers to one tetramer simply by greatly increasing the local concentration of dimers. The synaptosome confines the two crossover-site-bound dimers to a box of ~130 Å3 (1 Å=0.1 nm), which corresponds to a local concentration of 30 mM. In vitro, we find that low concentrations of wild-type Sin are catalytically inactive on isolated crossover site duplexes, but we see low levels of activity even at concentrations three orders of magnitude less than 30 mM (P.A. Rice and K.W. Mouw, unpublished work).
The regulatory portion of the synaptosome probably also acts more directly. Two residues from the catalytic domain, Phe52 and Arg54, are needed for full stimulatory activity of the regulatory module . These make dimer–dimer contacts in the crystal and, with some minor readjustments to the synaptosome model, could make protein–protein contacts between the regulatory and catalytic tetramers. Whereas these contacts may simply help to localize and orient the crossover-site-bound dimers, they may also induce a conformational strain that would destabilize the dimer, tipping the equilibrium towards the tetramer.
How universal is this synaptosome model? Tn3 and γδ resolvases are both ~30% identical in sequence with Sin, and all three are regulated by a three-noded synaptosome. However, the components of the synaptosome are slightly different for Tn3/γδ (Figure 1): two additional recombinase dimers rather than one, and no other DNA-bending proteins. However, as with Sin, the geometry of the additional resolvase-binding sites differs significantly from that of site I. Also, Tn3/γδ residues equivalent to 52 and 54 of Sin have long been known to be important for contacts between certain regulatory subunits and the catalytic ones [17,18]. We therefore wondered whether site II of Tn3 might play a simple DNA-bending role similar to that of IHF for Sin, and site III would form a DNA-binding-domain-mediated tetramer. However, recent preliminary structural and domain-swap data from our laboratories lend no support to this hypothesis (P.A. Rice, S.P. Montano, S.-J. Rowland and S. Holt, unpublished work). The new structures also show that, as in the γδ resolvase–site I and Sin–site II complexes, the DNA in the Tn3 resolvase dimer–site III structure bends away from, not towards the catalytic domains. This is the opposite bend direction to that seen in all published models of Tn3/γδ synaptosomes [17,19,20]. We are thus left with the conundrum that all of the existing models for the Tn3/γδ complex will need at least some modification.
Sin and the Tn3/γδ transposon resolvases are so closely related that they must have diverged from a common ancestor, yet they may have independently converged on the trick of using a three-noded synaptic complex for regulation. Interestingly, the unrelated tyrosine recombinase XerC/D, when acting as a plasmid dimer resolvase, is also regulated by a three-noded synaptic complex . This may be due to the intrinsic usefulness of this topology in ensuring that these site-specific recombinases catalyse only resolution and not other physiologically deleterious reactions.
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.).
This work was supported by the National Institute of Health [grant number R01 G086826 (to P.A.R.)], a Medical Scientist National Research Service Award [T32 G07281 (to K.W.M.)] and The Wellcome Trust (to W.M.S.).