A novel family of regulated helicases/nucleases from Gram‐positive bacteria: insights into the initiation of DNA recombination

Homologous DNA recombination can be defined at a molecular level as the exchange of genetic information between two homologous DNA molecules. It is critically important in fundamental processes such as the repair of genotoxic lesions, the creation of genetic diversity, the proper segregation of chromosomes and the restart of stalled DNA replication forks. Detailed genetic and biochemical analyses in the bacterium Escherichia coli and in the yeast Saccharomyces cerevisiae led to the realization that DNA recombination can be divided into four successive steps, with each step requiring the co-ordinated action of a large number of proteins (for a review, see Kowalczykowski et al., 1994; Pâques and Haber, 1999). At the heart of homologous recombination lies the DNA strand-exchange step. This step involves the alignment of two homologous DNA molecules and the exchange of their complementary strands to produce a joint DNA mole-cule. In E. coli, the main protein involved at this step is the RecA protein. Structural and functional homologues of RecA protein have been discovered throughout Eukarya and Archaea, thus emphasizing their critical role in all forms of cellular life (Bianco et al., 1998; Seitz et al., 1998). In all cases, DNA strand-exchange proteins are able to bind to single-stranded DNA (ssDNA) in a co-operative way, forming a highly ordered nucleoprotein fila-ment in which the DNA is extended. This filament is the active species of the protein: it carries out the search for a homologous duplex DNA partner, pairs the two molecules and catalyses the exchange of their strands. Despite some biochemical differences between the various known DNA strand-exchange proteins, ssDNA is universally used by all of them to initiate the homologous pairing and DNA strand-exchange events, although other modes of initiation are possible (Mazin and Kowalczykowski, 1999; Zaitsev and Kowalczykowski, 2000). Since DNA in cells is normally found in the double-stranded B form (dsDNA), homologous recombination calls for a first step in which the dsDNA is processed to produce a suitable substrate – ssDNA – for the binding of DNA strand-exchange proteins. This first phase is referred to as the initiation phase and will be the main focus of this review. Initiation sites include both dsDNA breaks, either ‘normal’ physiological breaks (e.g. programmed during the course of meiosis or formed at a stalled replication fork) or pathological breaks (e.g. caused by exposure to γ-radiation or certain chemicals), and single-strand DNA gaps (e.g. such as those which exist after the collapse and restart of a DNA replication fork) (Kowalczykowski, 2000). This review will primarily discuss dsDNA breaks as initiation sites. Once a dsDNA break has been introduced into DNA, the creation of ssDNA can be achieved by several means (Fig. 1). DNA helicases can catalyse the separation of the duplex DNA into its individual strands. Initiation by helicases has been documented in vivo as well as in vitro in bacteria (Thaler et al., 1989; Harmon and Kowalczykowski, 1998; Churchill et al., 1999; Cohen and Sinclair, 2001). Double-stranded exonucleases degrading one strand of the duplex can also be used to produce a single-stranded region and such processing has also been documented both in vivo and in vitro in different organisms (Little, 1967; Kolodner et al., 1994). Another possibility is to associate the activities of a DNA helicase and a single-stranded DNA nuclease to produce the desired ssDNA region. This can be achieved by the coupling of the action of two distinct enzymes, each contributing one of the two activities (Lovett et al., 1988), or from the action of one enzyme containing both helicase and nuclease activities. The RecBCD enzyme of E. coli has, for a number of years, been the prototype for this latter category and has been the focus of considerable attention. Recently, however, a novel class of enzymes that are structurally different from the RecBCD enzyme, but that carry both helicase and nuclease activities, has emerged from studies of DNA recombination in the Gram-positive bacterium Bacillus subtilis. This discovery allowed the E. coli paradigm to be made universal among Bacteria and to highlight interesting differences in the use of such enzymes to initiate DNA recombination. This review will briefly summarize our current knowledge of the RecBCD class of enzymes (for reviews, see also Kowalczykowski et al., 1994; Myers and Stahl, 1994) and will then emphasize the recent developments and their implications for our understanding of DNA recombination. Of central importance for this review will also be a class of short DNA sequences, collectively referred to as Chi sequences, that have the unique ability to regulate the activities of their own cognate helicase/nuclease enzyme and thereby effect DNA recombination. Creation of single-stranded DNA from a double-stranded DNA break (top) can be achieved by the action of either DNA helicases (e.g. E. coli RecBC or RecQ proteins – left), double-stranded DNA-specific exonucleases (e.g. E. coli RecE protein or phage lambda Redα protein – middle), or a combination of helicases and nucleases (e.g. E. coli RecQ–RecJ proteins, RecBC–RecJ proteins, or RecBCD enzyme and χ– right). The newly created region of ssDNA will then be used to initiate homology-dependent pairing and DNA strand exchange, a reaction catalysed by the RecA protein. The RecBCD enzyme is a large (330 kDa) heterotrimeric complex composed of the RecB, RecC and RecD proteins that initiates the main pathway of recombination in E. coli. Mutations in the recB or recC genes cause about a 100-fold decrease in the frequency of conjugal recombination (Howard-Flanders and Theriot, 1966; Emmerson, 1968) and strongly sensitize cells to DNA-damaging treatments such as UV light or X-ray radiation (Sargentini and Smith, 1986). The RecBCD enzyme has long been known to be a potent ATP-dependent dsDNA exonuclease, often referred to as ExoV (Telander-Muskavitch and Linn, 1981). Subsequent studies showed that RecBCD is also a processive DNA helicase, for which unwinding is coupled to exonucleolytic degradation of the DNA. Notably, RecBCD enzyme has the unique ability to load onto a blunt or near-blunt dsDNA end and to unwind more than 30 kilobases per binding event (Taylor and Smith, 1985; Roman et al., 1992; Bianco et al., 2001). Its degradative function, however, seems to be at odds with the view that RecBCD is involved in initiating DNA recombination, a process which requires preservation of DNA. This paradox was solved by the discovery that the nuclease properties of the enzyme are regulated upon recognition of a specific DNA sequence by the translocating RecBCD enzyme. This sequence, called Chi (χ; Lam et al., 1974; Stahl et al., 1975), corresponds to the octamer 5′-GCTGGTGG-3′ (Smith et al., 1981; Bianco and Kowalczykowski, 1997). Recognition of χ by RecBCD is polar: χ will only be recognized by a RecBCD molecule travelling through dsDNA from right to left as written above. This recognition event elicits a number of changes in the behaviour of the RecBCD enzyme (Fig. 2, left). Upon recognition, the vigorous 3′ to 5′ nuclease that was present before χ is strongly downregulated, while a weaker 5′ to 3′ nuclease activity is activated on the opposite strand (Dixon and Kowalczykowski, 1991, 1993; Anderson and Kowalczykowski, 1997a). After recognition of χ by the translocating RecBCD enzyme and its eventual dissociation from the DNA, the modifications described above enable the RecBCD enzyme to produce a dsDNA molecule with a 3′-ssDNA tail terminating at χ that can be bound by the RecA protein to initiate the homologous pairing phase of genetic recombination. Recent studies also showed that the χ-activated RecBCD enzyme directly facilitates the loading of the RecA protein onto the χ-terminated ssDNA. This ensures maximal participation of this χ-protected ssDNA in the process of genetic recombination (Anderson and Kowalczykowski, 1997b) and effectively provides for co-ordination of the two first steps of this process. The mechanism by which χ recognition is translated into a change of enzymatic activity has been the subject of much debate and is still not clearly understood (for reviews, see Kowalczykowski et al., 1994; Myers and Stahl, 1994). It appears, however, that the RecD subunit of the complex plays a major role in this regulation. Indeed, although dissociation of the RecD subunit does not occur upon χ recognition (Dohoney and Gelles, 2001), the RecBC enzyme displays most of the recombinationally important biochemical activities of the χ-modified RecBCD enzyme. Most notably, it simply functions as a DNA helicase lacking the potent nuclease activity of the RecBCD enzyme, it is unable to respond to χ, but it retains the ability to load the RecA protein onto the 3′-terminated DNA strand (Dixon et al., 1994; Anderson et al., 1997; Churchill et al., 1999). Processing of χ-containing DNA by E. coli RecBCD enzyme (left) and B. subtilis AddAB enzyme (right). See text for details. Although the RecBCD enzyme and the regulation of its activities by χ have long been considered a model for understanding the initiation of DNA recombination in Bacteria, evidence supporting this assumption was, until very recently, quite limited. Because of the advance of genomic sequencing, we can now ascertain the existence of true RecBCD homologues (i.e. containing three subunits showing extensive amino acid homology to RecB, RecC and RecD) in several genera, encompassing a large part of the eubacterial tree (Fig. 3). Despite this prevalence, however, it appeared that a functional interaction between the E. coliχ sequence and other RecBCD enzymes was preserved only in enteric bacteria that are closely related to E. coli (McKittrick and Smith, 1989). This suggested either that the RecBCD–χ interaction was restricted to a small group of microbes or that functionally equivalent sequences unrelated to the canonical E. coliχ sequence were used in different organisms. Subsequent work showed that this latter view was correct. Short specific DNA sequences (5–8 bp) that protect dsDNA from exonucleolytic degradation in the bacteria Lactococcus lactis (Biswas et al., 1995), Haemophilus influenzae (Sourice et al., 1998) and Bacillus subtilis (Chédin et al., 1998) have been identified. In all cases, those putative χ sequences were believed to interact with, and regulate, an enzyme that possesses both helicase and exonuclease activity. The H. influenzae enzyme is a clear homologue of RecBCD. However, for the two Gram-positive bacteria L. lactis and B. subtilis, the corresponding enzyme seems to belong to a novel class of enzymes that are functionally, but not structurally, homologous to the RecBCD class of enzymes. Distribution of the RecBCD and AddAB enzymes in eubacterial genomes. Groups containing RecBCD homologues are indicated in blue, while groups containing AddAB homologues are indicated in green. The schematic tree was based upon 16S ribosomal RNA sequence comparisons. Numbers in parentheses indicate the number of species containing a RecBCD or AddAB homologue in each group. The species containing a homologue are given. It has long been known that a strong ExoV activity is present in crude extracts of B. subtilis (Chestukhin et al., 1972; Ohi and Sueoka, 1973). several for this ATP-dependent activity were et al., to the proteins for this but that this protein the for an and a with other ExoV enzymes and et al., The and of two genes B. subtilis and were by and et al., and two genes are as an are during as part of a et al., 1998) that also the et al., In B. subtilis, and the two subunits of the AddAB enzyme and The subunit of homology to the subunit of the RecBCD enzyme. of are by a large of DNA helicases and an and other helicase nuclease present on the of the E. coli protein et al., 1998) is also present at the of et al., the between and In the subunit does not homology to the other known proteins, RecC or for a putative and a of the nuclease by homologues of the B. subtilis AddAB complex have now been by sequence homology in at different three other B. and B. Lactococcus and and various and (Fig. 3). of organisms belong to the of the Gram-positive which seems to be a the AddAB class of enzymes It is interesting to however, that in subunits a level of homology to each other with other proteins such as both genetic and clearly that at the AddAB enzymes of B. subtilis and L. lactis are composed of only two subunits of the B. subtilis the of a putative homologue of the E. coli RecD protein by the This more than from the is also in L. lactis However, the of this be unrelated as it that a of RecD homologues are present in organisms that are clearly of RecBC or AddAB such as by the or the by the it is a RecD homologue or this is involved in a different of DNA B. subtilis and L. lactis are the only two organisms for which is information the role of their AddAB enzyme in cellular DNA the homologue of the B. subtilis AddAB enzyme in L. (for recombination et al., 1998) will be referred to in a as In both B. subtilis and L. of or to to DNA-damaging and of homologous recombination, as by phage or et al., 1993; et al., 1997; et al., 1998). closely those for E. coli In B. subtilis, AddAB one of the main of homologous recombination et al., 1991, of B. subtilis or L. lactis AddAB in E. coli of a for UV and conjugal recombination et al., 1993; et al., that AddAB and RecBCD enzymes are functions and that other are for their a potent dsDNA exonuclease to initiate DNA recombination, a process that is to repair and the of the genetic requires that the exonucleolytic activity be regulated in some described the E. coli RecBCD enzyme and to a specific DNA sequence while translocating through DNA. This interaction the RecBCD enzyme from a degradative nuclease to a nuclease by both and the of DNA of evidence that B. subtilis and L. lactis also specific DNA sequences that are homologues of the E. coliχ sequence will be referred to as and sequences, present on protect DNA from exonucleolytic degradation by the corresponding RecBCD or AddAB enzyme in vivo et al., 1992; et al., et al., 1998). of their and in their some to the E. coli sequences are short for and for are in are strongly and with to DNA replication et al., 1997; et al., 1998; et al., 1999). evidence for an interaction between AddAB and was in vivo for L. the L. lactis in E. coli the degradation of dsDNA the DNA carries a et al., 1998). It was also that functions as an recombination et al., thus the between the E. coliχ and its In B. subtilis, the interaction between the AddAB enzyme and its cognate sequence has been in in vitro (Chédin et al., 2000). In the B. subtilis AddAB enzyme activities that are of the RecBCD it is both a and processive DNA helicase, a potent dsDNA exonuclease and an et al., et al., 1997; and Kowalczykowski, interesting between the AddAB enzyme and the RecBCD enzyme, however, is that while degradation by the RecBCD enzyme a strong for the 3′-terminated strand at the (Dixon and Kowalczykowski, the degradation of DNA by AddAB is on the two strands (Fig. 2, right). This behaviour is with the that the AddAB enzyme two putative nuclease as in while the RecBCD enzyme only and recognition of by the AddAB enzyme only in one the enzyme the sequence from the 3′ written for recognition to occur (Chédin et al., 2000). This is also found for L. lactis a of et al., and is in with the behaviour of the E. coli RecBCD enzyme. In B. subtilis, recognition of the the potent nuclease activity is while the nuclease activity is This one before the sequence, of the of the nuclease activity before which that the recognition event is very After recognition and have unwinding The AddAB enzyme then as a 5′ to 3′ thus for the of a 3′-terminated single-stranded DNA a of that is used universally to initiate the process of homologous DNA recombination. The AddAB enzyme carries two homologous nuclease of the nuclease found in the E. coli protein with homologous found in various proteins. The indicate the three that were to be critical for nuclease activity et al., 1998; et al., 2000). B. of the putative nuclease present at the of various proteins. The indicate homologous to that the of the is not known in and was out the sequence were from The for at The most between the AddAB and the RecBCD enzymes is the their biochemical the two enzymes have a very different subunit The based on is a homologue of the which by ssDNA and short et al., 1997). It is believed that the the is the helicase of the the from to DNA and to catalyse its unwinding and Emmerson, of the or the subunit that bind or are for DNA recombination and 1992; et al., that the AddAB enzyme both strands of the DNA duplex and that each of its subunits carries a nuclease it is to that each strand is by one Because of its homology with RecB, which with the 3′-terminated strand at the and Smith, 1993; and Smith, and which is known to in a 3′ to 5′ et al., 1997; Bianco and Kowalczykowski, it is that the subunit is for the 3′ to 5′ an activity that can be by recognition of on the other is to the opposite with a 5′ to 3′ this however, is not subject to regulation by this it that the subunit functions normally by the RecC and RecD This of the DNA helicase activity of in a to of the helicase activity of the subunit by the RecC subunit by the RecD subunit et al., 1997). the of the complex to and to respond to its cognate χ sequence, which that the subunit is a functional homologue of both the RecC subunit in χ et al., 1998) and the RecD subunit in the to Churchill et al., 1999). recognition of χ by the AddAB enzyme is translated into a change of enzymatic activity is of the discovered functions of the RecBCD enzyme is to the loading of the RecA protein onto the 3′-terminated ssDNA (Anderson and Kowalczykowski, a that is also by the RecBC enzyme et al., 1999). The of this loading to the that the RecA protein is at a strong with the protein for binding to ssDNA. that this in of RecA protein have and an important role in the initiation of DNA recombination. exist in E. coli, the and proteins. proteins RecA protein to with the protein for binding to ssDNA and the of the nucleoprotein filament to the et al., 1993; and 1994; et al., 1997). the co-ordinated loading of RecA by the AddAB enzyme has not been for the B. subtilis and Kowalczykowski, However, a combination of E. coli RecA protein and B. subtilis AddAB enzyme was used of the proper which have the It is known that of recombination the of RecA protein and RecBCD enzyme from the species et al., and this were to be a homologous combination of proteins, it strongly that other as the and the in B. subtilis. In this it is interesting to that the pathway also in B. subtilis, and it seems to a much more role in this as by of for the or genes et al., 1991, 1993; et al., 1999). such as the RecBCD or the AddAB enzyme are now to be among are to an important role in the initiation of DNA recombination and DNA can however, this which involves a dsDNA nuclease and the to them and Stahl, has been have been used to the for recombination (Fig. 1). to this from an to the cells by the of a strong double-stranded DNA exonuclease activity. Indeed, in E. coli, the and the activity of the ExoV of is critically for a large number of double-stranded DNA breaks are et al., 1997; et al., 1998). such which be created by of the recombination upon or in a genetic be by the exonuclease activity of RecBCD cells are to and 1993; et al., and Stahl, 1997). of this is found in a bacterium ability to at is on the that ExoV activity is and 1999). Another from the of ExoV is the of DNA recombination between Homologous recombination to in a as a of the to recombination and 1997; and 1998). it is known that ExoV provides for a dsDNA molecules et al., although it is that in this is from It that although the RecBC enzyme is able to catalyse the initiation phase of DNA recombination, the of the RecD subunit enable the cells to with the with a or to to such as large of double-stranded DNA breaks are This out to be true in for RecBCD their in and as a of their to the of et al., et al., 2001), a of DNA strand breaks et al., 2000). In this it is interesting to that are of bacteria for which clear homologue of either RecBCD or AddAB enzyme can be the this for and However, most of organisms the of the which the known to homologues of the and proteins, that the pathway of recombination has been well throughout It be interesting to the species of a functional RecBCD pathway are more in their ability to to are with a of DNA and this is by a species that a functional RecBCD or AddAB however, the for a able to the exonuclease activity is have to The of the E. coli phage are from the RecBCD enzyme by the of a and the phage use of the protein and The lambda phage the which directly the activity of RecBCD while the phage to its the activity of the RecBCD complex 2000). Bacteria, use of sequences present in their genomes. to this is the χ sequences are in their et al., which their with a can use to from (Biswas et al., and Smith, It is however, that this has strong in the event that a of DNA possesses the χ sequence, this DNA will be for homologous recombination and effectively into the a that has been used to in E. coli and Smith, 1997). The paradigm can only on short DNA or in the DNA has homologous in the of the In the of can to of DNA, this on its own most be In the of the main for the of the exchange is the repair et al., 1989; et al., The can DNA formed by DNA recombination and its or its are found and 1998). This role that has been preserved throughout the repair also to DNA recombination between sequences in and 1998; et al., more for the use of χ sequences as a means to regulate enzymes such as RecBCD or AddAB is found in the co-ordination by this regulation. all the found in at the of the enzyme, the use of a sequence the cells to the nuclease function, while on the initiation of DNA recombination in a process that can be as as The of this has been in recent studies of the repair of dsDNA breaks in E. coli. to a considerable number of dsDNA breaks during each replication as a of the and collapse of DNA replication in E. coli 1997; et al., 1997; Kowalczykowski, 2000). The RecBCD enzyme, with its for DNA and its unique ability to initiate effectively the repair of lesions, the main by which cells can with this et al., 1998). This also to the with the of RecBCD nuclease activity regulation by the of χ sites in the and their with the of DNA replication this repair more It is that this for other B. subtilis and L. Indeed, their χ sequences are in their at for B. subtilis and L. with DNA replication et al., 1999). It that regulated than of cells with to with a of such as dsDNA that the E. coli RecBCD enzyme and the B. subtilis AddAB enzyme two of regulated the functions of enzymes, in the mechanism by which their is regulated in to specific sequences, will some into our understanding of the process of DNA recombination in the of the Kowalczykowski for their critical of the Seitz and This work was by from the of