The process of information exchange between two homologous DNA duplexes is known as homologous recombination (HR) or double-strand break repair (DSBR), depending on the context. HR is the fundamental process underlying the genome shuffling that expands genetic diversity (for example during meiosis in eukaryotes). DSBR is an essential repair pathway in all three domains of life, and plays a major role in the rescue of stalled or collapsed replication forks, a phenomenon known as recombination-dependent replication (RDR). The process of HR in the archaea is gradually being elucidated, initially from structural and biochemical studies, but increasingly using new genetic systems. The present review focuses on our current understanding of the structures, functions and interactions of archaeal HR proteins, with an emphasis on recent advances. There are still many unknown aspects of archaeal HR, most notably the mechanism of branch migration of Holliday junctions, which is also an open question in eukarya.
The classical HR (homologous recombination) pathway (Figure 1) starts with a DSB (double-strand break) that can arise either due to the action of a cellular enzyme or as a consequence of DNA damage. DSBs are ‘resectioned’ to produce 3′ ss (single-stranded) DNA tails that are suitable for strand invasion into a homologous duplex DNA and as a primer for DNA synthesis, producing a joint molecule linked by an HJ (Holliday junction). HJs are four-stranded DNA structures that are mobile and can move by the process of branch migration, catalysing the exchange of genetic information and formation of heteroduplexes as a consequence. HJs are removed by an HJ-resolving enzyme, releasing nicked daughter duplexes that are ligated, completing the HR process. Several alternative pathways, such as SDSA (synthesis-dependent strand annealing) play an important role in certain contexts, but all involve strand exchange reactions between homologous sequences.
The pathway and proteins of HR in the archaea
The proteins that catalyse HR in archaea are highly conserved and nearly ubiquitous, reflecting the importance of HR and its role in the rescue of stalled replication forks. Some of these proteins, such as Mre11, Rad50 and RadA, are conserved between archaea and eukarya and have proven to be useful particularly in structural biology to help us to understand the equivalent eukaryal system. Others, such as Hjc, NurA and HerA, represent unique archaeal solutions to the problem of DSBR (DSB repair). The present review summarizes the current state of knowledge of archaeal HR and highlights some future prospects for the field.
The first step in HR following initiation by a DSB is resectioning to remove the 5′-strands, leaving 3′-DNA overhangs that are suitable both for strand invasion into a homologous DNA duplex and to act as primers for DNA synthesis. In bacteria such as Escherichia coli, resectioning is catalysed by the RecBCD complex that includes helicase and nuclease subunits . In eukarya, the Mre11 and Rad50 proteins form a complex that is implicated in this step of HR (reviewed in ). These proteins are also conserved in archaea, frequently in an operon with the hexameric HerA helicase  and NurA nuclease . HerA was shown to form a stable complex with Mre11 and Rad50 in vitro [5,6]. In an elegant series of experiments, Hopkins and Paull  have demonstrated that Pyrococcus furiosus Rad50, Mre11, HerA and NurA co-operate to catalyse 3′-end resectioning for HR. The Mre11–Rad50 complex was shown to generate a short 3′-overhang by limited degradation of the 5′-strand. This allowed the HerA–NurA complex to initiate processive 5′-strand resection, generating a substrate suitable for RadA-catalysed strand exchange. By including all four proteins along with the RadA recombinase, a blunt DNA substrate could be processed and used to catalyse a strand invasion (D-loop formation) reaction in vitro . A more nuanced role for Mre11–Rad50 has been suggested on the basis of genetic studies of recombination in Haloferax volcanii. Here, the Mre11–Rad50 complex is proposed to act as a gatekeeper, slowing entry to the HR pathway following DNA damage, perhaps to limit unrestrained chromosome rearrangements .
Strand invasion and exchange
Once end resectioning is completed, a number of proteins are thought to participate in the homology search, strand invasion and exchange steps of HR, which can ultimately result in the formation of an HJ. These proteins are considered in turn below.
SSBs (ssDNA-binding proteins)
SSBs [also known as RPA (replication protein A) in eukaryotes] act as DNA-damage sensors and ssDNA chaperones in many DNA repair, replication and recombination pathways in all forms of life. SSBs come in many different varieties: homodimers and homotetramers in bacteria, homodimers and heterotrimers in eukarya, and monomers, dimers and trimers in archaea . The core functional domain of the canonical SSB is the OB-fold (oligonucleotide-binding fold), which is a twisted β-barrel with a binding site that accommodates four nucleotides of ssDNA . In euryarchaea such as P. furiosus, the SSB is heterotrimeric and resembles eukaryal RPA closely . It is often named as RPA for this reason. In crenarchaea, the SSB is monomeric in solution  and can bind to and unwind damaged DNA by trapping transient ss forms . The crenarchaeal SSB more closely resembles the hSSB1 and hSSB2 proteins recently described in humans .
In HR, SSB is expected to bind to the 3′-ssDNA ends generated by the action of Mre11–Rad50–HerA–NurA. The SSB must therefore be displaced to allow formation of a RadA nucleoprotein filament that is essential for strand exchange. In vitro, Sulfolobus solfataricus SSB has been shown to inhibit a key step in strand exchange, the invasion step known as D-loop formation . This inhibition may be relieved by RadA paralogues catalysing SSB displacement [15,16], as has been observed in eukaryal organisms, or may result from a physical interaction between SSB and RadA [16,17]. Although a role for archaeal SSBs in HR is beyond doubt, there are still some conflicting observations, particularly regarding physical and functional protein interactions involving SSBs, that remain to be resolved. For example, the C-terminal tail of S. solfataricus SSB plays no part in DNA binding and resembles the protein interaction domain of E. coli SSB . S. solfataricus SSB has been reported to interact physically with RNA polymerase , the nuclease NurA  and the recombinase RadA . Some caution is required in the interpretation of these observations, however, as interactions via bridging DNA have not always been ruled out.
The RecA protein family, comprising Rad51 and its paralogues in eukarya, RadA in archaea and RecA in bacteria, is one of the few universally conserved DNA-repair proteins. RecA family members are DNA recombinases catalysing strand-exchange reactions that are central to HR and DSBR. The role of RecA family recombinases in vivo is primarily to catalyse the incorporation of a protein-coated ssDNA filament into a homologous duplex DNA species, leading to the formation of recombination intermediates such as heteroduplexes, D-loops and HJs. Disruption of RecA in bacteria, or Rad51 in yeast, is highly deleterious to the cell, whereas metazoan Rad51 is an essential protein . The archaeal Rad51 protein is much more similar to Rad51 than RecA, sharing the ds (double-stranded) DNA-binding NTD (N-terminal domain), and lacking the RecA-specific CTD (C-terminal domain) . RadA has been deleted successfully in H. volcanii, leading to severe DNA recombination, repair and growth defects . The structural and biochemical properties of RadA in general match those of Rad51 [22–24] and will not be discussed in detail in the present paper owing to space limitations.
In eukaryotes, Rad51 paralogues (Rad51C, Rad51D, XRCC2, XRCC3) co-operate with Rad51 in strand-exchange reactions in vitro, and are required for damage-specific Rad51 repair foci in vivo . Most archaeal genomes, even that of Nanoarchaeum equitans, encode one or more RadA paralogues (reviewed in ). The first to be characterized was RadB, which lacks the NTD of Rad51 and does not support strand exchange in vitro [26,27]. Genetic studies have revealed an important role for RadB from H. volcanii in recombination and repair ; however, its function in vivo remains unclear. S. solfataricus encodes three RadA paralogues, which are all expressed at lower levels than RadA . The crystal structure of one, Sso2452, has been solved, revealing a structure similar to RadB with the core ATP-binding domain of RadA, but lacking the NTD . Sso2452 binds ssDNA more tightly than RadA and inhibits D-loop formation by RadA in vitro. This inhibition is observed even when a ssDNA–RadA nucleoprotein filament is pre-formed . This antirecombinase activity may be physiologically relevant as a mechanism to limit HR in vivo. However, the equivalent protein from Sulfolobus tokadaii has been suggested to function as an activator of RadA-mediated strand exchange, albeit based on quite preliminary data . Genetic studies are likely to be required to help tease out the roles of the archaeal RadA paralogues, and it is worth considering that the functions of Rad51 paralogues in eukarya are still not understood fully.
Branch migration: the missing motor
In bacteria, the molecular motor RuvAB binds to HJs and catalyses strand exchange during HR (reviewed in ). Such an activity is required during HR to provide directionality to strand exchange; however, the equivalent proteins have not been identified in eukarya or archaea despite many years of study. Currently, the best candidate for this role is the helicase Hel308 (also known as Hjm). Hel308 is widely conserved in archaea and also in eukarya, where it functions in recombination . Archaeal Hel308 can unwind branched DNA structures and has limited HJ branch migration activity in vitro [30–32]. Although these properties argue more strongly for a role in replication fork restart than branch migration, the enzyme has recently been shown to have a physical and functional interaction with Hjc . At present, the jury is out on whether Hel308 catalyses branch migration in archaea or whether there is another motor protein that has yet to be discovered.
HJ resolution: Hjc and Hje
The archaeal HJ-resolving enzyme, Hjc (for HJ cleavage), was identified by the Ishino laboratory in 1999 by screening for a junction resolution activity in crude cell extracts followed by heterologous expression in E. coli . Hjc was shown to have properties analogous to the E. coli resolving enzyme RuvC, although the two proteins did not share any sequence similarity. Genetic studies confirmed that expression of an archaeal Hjc gene in E. coli could rescue a ruvC mutant . Subsequently, it was demonstrated that the crenarchaeon S. solfataricus had two distinct resolving enzyme activities, one encoded by the hjc gene and the other named Hje (for HJ endonuclease) which is homologous with Hjc [36,37]. Hjc orthologues shared a very limited number of invariant amino acid residues, including the E…PD…ExK signature conserved in a large number of nucleases and restriction enzymes, confirming that Hjc belonged in the nuclease superfamily of Mg2+/Mn2+-dependent endonucleases that includes the Type I and Type II restriction enzymes as well as DNA-repair enzymes such as MutH, λ-exonuclease and the phage resolving enzyme T7 endonuclease I [38,39]. Subsequently, an orthologue of Hjc was identified in the genome of SIRV (Sulfolobus islandicus rod-shaped virus), a rudivirus with a linear covalently closed genome . This led to suggestions that the archaeal SIRV replicated through concatameric intermediates that required an HJ-resolving enzyme to resolve them, as is seen the pox family of eukaryotic viruses which also encode their own resolving enzyme .
Careful analysis of Hjc and Hje suggest that they cleave HJs at different positions with respect to the junction centre, suggesting a differential positioning of the pair of active sites. They also cut HJs on different pairs of arms (exchanging arms of a stacked-X junction for Hjc and continuous strands for Hje), suggesting differences in the way they manipulate the junction on binding . Comparison of the crystal structures of Hjc and Hje suggests that there has been a reorientation of the dimer interface in the latter, resulting from quite subtle changes in the interface. This may help to explain some of the differences in substrate specificity.
Unlike other cellular resolving enzymes such as RuvC and CCE1, Hjc and Hje lack any detectable sequence preference for cleavage of HJs . Sequence specificity has been suggested to act as a filter to ensure that only four-way junctions are substrates for this class of enzyme , making the exquisite specificity of Hjc and Hje for HJs all the more remarkable. The crystal structures of Pyrococcus and Sulfolobus Hjc were reported in 2001 [43,44], confirming the restriction-enzyme-like active site predicted by bioinformatic analyses, and revealing a highly basic binding interface for the HJ, suggesting that the HJ would be distorted into a planar X-shape on binding. This suggestion was confirmed by subsequent FRET (fluorescence resonance energy transfer) studies . A crystal structure for Hje followed in 2004 , revealing the structural basis for the differences in substrate specificity and cleavage that had been observed previously and highlighting the role of an absolutely conserved serine residue in the catalytic mechanism. Subsequent kinetic studies of Hje revealed a rapid multiple turnover reaction cycle limited by the catalytic step, suggesting that this enzyme is a suitable candidate for applications that require the detection and resolution of HJs . Meanwhile, the canonical Hjc enzyme was shown to interact physically with PCNA (proliferating-cell nuclear antigen), the ring-shaped sliding clamp protein, via a C-terminal PIP (PCNA-interacting peptide) motif . Subsequently this interaction was shown to be functionally relevant, as PCNA stimulates the activity of Hjc in vitro . This interaction may be relevant in vivo for targeting of Hjc to stalled replication forks that are rescued by HR .
Molecular Biology of Archaea II: A Biochemical Society Focused Meeting held at Robinson College, Cambridge, U.K., 16–18 August 2010. Organized and Edited by Stephen Bell (Oxford, U.K.) and Finn Werner (University College London, U.K.).
Thanks to all current and past members of the White laboratory for their contributions to research in this area.
This work has been supported over the last 12 years by the Biotechnology and Biological Sciences Research Council.