INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In bacteria, double-stranded DNA breaks (DSB) are frequent events that may be provoked, for example, by pauses in the replication fork (36, 43). Such genomic disruptions are lethal in the absence of DNA repair. In Escherichia coli DSB repair requires the activity of a large enzyme complex, known as RecBCD, that has ATP-dependent helicase and exonuclease activities (see reference 32 for a review). The enzyme degrades both strands, starting from the DNA break until it reaches an octanucleotide sequence, known as Chi, that attenuates degradation and stimulates recombination (44, 46). The enzyme exhibits helicase activity and residual exonuclease activity with an altered polarity after Chi (4, 16); the remaining activity provides a single-stranded DNA substrate for recombination enzymes to mediate repair.

Organization of the three-subunit exonuclease/helicase (exo/hel) RecBCD. Structure-functional studies of RecBCD have revealed some of the roles of each subunit. RecB seems to possess two key activities of the enzyme. The N-terminal 929 amino acids (out of 1,180 total) have confirmed ATPase and helicase activities (13, 54); this region is similar to that of UvrD helicase. RecBCD helicase activity was recently proposed to function via a mechanism similar to that determined for UvrD (6). Nuclease activity was recently localized to the C-terminal 251 amino acids of RecB and is associated with the presence of a conserved motif, G-i-i-D-x(12)-D-Y-K-t-d (amino acids in small letters show less conservation) (51, 53, 54). This motif is present in numerous bacterial and eukaryotic enzymes (5). RecBCD was shown to have a single nuclease catalytic center in RecB that works on both DNA strands (51). Little is known about the roles of RecC, except that it appears to greatly enhance activities and processivity of RecB (11, 38); mutations in the RecC gene can also result in loss or modification of Chi recognition, as do mutations in genes of all subunits (1). RecD is an ATPase with similarity to a helicase involved in conjugational transfer of an enteric bacterial plasmid; its homologues seem to be broadly distributed in bacteria (determined by BLAST comparisons; http://www.ncbi.nlm.nih.gov/BLAST/). As part of RecBCD, RecD appears to regulate exonuclease activity. Recent data suggest that RecD maintains RecBCD incompetent for homologous recombination prior to Chi; at Chi, RecD is suggested to undergo a conformational change that attenuates exonuclease activity and stimulates recombination (2, 3, 12, 33, 48). A swing model was proposed in which RecD assures proximity of the RecB nuclease with both DNA strands prior to Chi and a repositioning of the nuclease after Chi (51, 54).

Organization of the two-subunit exo/hel enzymes. To date, models of exo/hel activities are based on those of RecBCD. Numerous RecBCD homologues have been identified in gram-negative enterobacteria and in the high-guanine-plus-cytosine-content mycobacteria. However, the functional RecBCD analogue in the low-guanine-plus-cytosine-content branch of gram-positive bacteria is structurally distinct from RecBCD. Using Lactococcus lactis as a model, a two-subunit enzyme called RexAB (comprising 1,073- and 1,099-amino- acid subunits, respectively) is necessary and sufficient to confer exo/hel activity and interacts with the L. lactis Chi site (22). RexAB bears homologues in at least six other gram-positive low-guanine-plus-cytosine-content bacteria as well as in the gram-negative bacterium Porphyromonas gingivalis (determined by BLAST comparison; http://www.ncbi.nlm.nih.gov/BLAST/). As studied in L. lactis or in Bacillus subtilis (AddAB), the two-subunit exo/hel enzymes display biological and/or biochemical activities equivalent to those of RecBCD (i.e., ATP-dependent exonuclease, helicase, exonuclease blocking at Chi, and Chi-stimulated recombination; 8, 9, 10, 22, 23, 28, 30, 31). RexA and its analogues in other gram-positive bacteria have homology with PcrA helicase (determined by BLAST comparison; http://www.ncbi.nlm.nih.gov/BLAST/), whose mechanism has been deduced from structural determinations (49). In addition, the nuclease motif described above for RecB is conserved in L. lactis RexA and in all known two-subunit exo/hel enzymes (5, 28, 53). This similarity has lead to the hypothesis that all exo/hel enzymes function via similar mechanisms.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and plasmids. The E. coli strains used were TG1 [F' traD36 lacI^q (lacZ)M15 proA⁺B⁺/supE (hsdM-mcrB)5(r_K m_K mcrB) thi (lac-proAB)], AB1157 (argE3 his-4 leuB6 proA2 thr-1 ara-14 galK2 lacY1 mtl-1 xyl-5 thi-1 rpsL31 tsx-33 supE44), and KM21 (AB1157 isogenic strain containing [recC ptr recB recD]::kan, referred to here as recBCD) (37). The rexAB genes were derived from L. lactis strain MG1363 (25) and were cloned on low-copy-number plasmid pGB2 (confers spectinomycin resistance) to generate pRexAB (22).

Media. E. coli strains were grown in Luria broth at 30 or 34°C, as specified below. Antibiotics were used in E. coli as follows: ampicillin at 100 µg/ml, spectinomycin at 50 µg/ml, tetracycline at 15 µg/ml, kanamycin at 40 µg/ml, and chloramphenicol at 15 µg/ml.

rexAB mutant constructions. To generate mutants affected in the nuclease motifs of RexAB, fragment mutagenesis was performed using modified primers and pRexAB as template DNA. RexAB^D910A was constructed by replacing the SfcI-ClaI fragment that contains the 3' end of rexB with a corresponding PCR fragment that was modified by a point mutation. The SfcI-end primer was 5'-CTTTCTACAGATTACTTAGGGGCGATTGCGTATA-3' (the SfcI site is underlined; the GAC aspartate codon, RexB position 910, is replaced by the GCG alanine codon; changes are in bold). The ClaI-end primer was 5'-TCGACAAATCGATTTGAGAGGACAATATCGACA-3' (the ClaI site is underlined). The resulting fragment was first subcloned onto an intermediate vector and then was cloned to replace the wild-type segment in pRexAB. The RexAB^DYK mutant was constructed essentially in the same way except that the SfcI-end primer was 5'-CCAACTTTCTACAGATTACTTAGGGGCGATT//TCAAGTGCTCATTCATT-3' (the 9-codon deletion, RexB amino acid positions 910 to 912, is indicated by double slashes).

DYK

DYK

771-1063

UV sensitivity tests for E. coli. The E. coli recBCD strains containing mutated or wild-type rexAB alleles were maintained at 30°C. Note that exonuclease activity of RexAB is thermosensitive in E. coli, possibly reflecting an optimal growth temperature of L. lactis of 30°C. Tests to determine UV resistance were performed as described previously (18).

T4g2 test for exonuclease activity in E. coli. The bacteriophage T4g2 amber mutant (kindly provided by W. Wackernagel, University of Oldenburg, Oldenburg, Germany) was used to evaluate exonuclease activity. The gene 2 product encodes a protein which protects phage DNA extremities from RecBCD degradation (34). The phage stock was prepared on a recBC strain not containing a supE mutation, so DNA ends of this phage mutant are exonuclease sensitive. Therefore, the number of PFU is low when this phage is titrated on strains expressing RecBCD. Lactococcal exonuclease activity expressed from pRexAB wild-type and mutated alleles was evaluated in the E. coli recBCD strain and compared to that of plasmid-free AB1157 (wild type) and E. coli recBCD strains essentially as described (52), except that cultures and plates were incubated at 30°C.

Detection of HMW. High-molecular-weight linear plasmid multimer (HMW) accumulation was detected on whole-cell lysates after agarose gel electrophoresis (14). Plasmid DNA, labeled by chemiluminescence using the ECL system (Amersham), was used as a probe. Southern blot hybridization was performed as recommended by the kit supplier.

Recombination test. We used a previously described strategy to measure Chi-stimulated homologous recombination using short dsDNA substrates (15). The plasmid target (named pBla) is a pBR322 derivative with a 111-bp deletion in the -lactamase gene (bla). The intact bla gene is restored via a double-exchange event with a linear DNA fragment (see reference 22 for details of its construction). In brief, primers were designed to PCR amplify a bla gene internal fragment covering the DNA deleted from pBla plus an additional 360-bp flanking homology with bla. Primer couples generating double Chi_Ll sites or no Chi sites (Chi_Ll⁰) were as described previously (22). Chi_Ll sites are located about 300 bp from heterologous dsDNA ends. Linear DNA used for experiments was recovered by PCR using the primers 5'-GTTGGGAAGGGCGATCGGTG-3' and 5'-CACTCATTAGGCACCCCAGGC-3'. The final fragment sizes were ~1.3 kb.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Conserved regions of two- versus three-subunit exo/hel enzymes. Alignments of several two-subunit exo/hel enzymes reveal that these enzymes are poorly homologous, even in closely related species. They bear very little homology with the three-subunit exo/hel. Nevertheless, a short nuclease motif was previously revealed in the AddA subunit of the B. subtilis AddAB enzyme and in the RecB subunit of RecBCD (28, 53). In RecBCD, this motif corresponds to the sole nuclease activity of the exo/hel enzymes (51). We found that this motif is actually present in both subunits of the two-subunit enzymes (Fig. 1). In the RexA subunit, a consensus is G-i-i-D-x(12)-D-Y-K-t-d (amino acids in small letters show some variation); in RexB it is G-r-i-D-R-i-D-x(9-12)-v-D-Y-K-S-s. The striking similarities between these motifs lead us to ask whether RexAB may bear two active nuclease sites, one in each subunit.

View larger version (36K): [in this window] [in a new window]   FIG. 1.   Alignment of nuclease motifs present in each subunit of the two-subunit exo/hel enzymes. (A) RexA subunit homologue alignments are presented for the region surrounding the nuclease motif (corresponding to positions 1085 to 1164 of the 1,173-amino acid L. lactis RexA subunit). (B) RexB subunit homologies in the region surrounding the nuclease motif (corresponding to positions 871 to 949 of the 1,099-amino acid L. lactis RexB subunit). Highly conserved motifs are enclosed in a rectangle, and amino acids that are totally conserved are in bold. The black bar over amino acids DYK corresponds to the region deleted in lactococcal RexA or RexB mutants; the arrow over position 910 in RexB indicates a point substitution from aspartic acid to alanine generated in RexB (see the text). Llac, L. lactis; Spyo, Streptococcus pyogenes; Spnu, Streptococcus pneumoniae; Efae, Enterococcus faecalis; Cdif, Clostridium difficile; Bsub, B. subtilis; Saur, Staphylococcus aureus.

Site-specific mutagenesis of putative nuclease motifs in RexA and RexB. The rexAB operon (rexB is followed by rexA) was previously cloned on a low-copy-number plasmid and shown to fulfill the biological roles of RecBCD in an E. coli recBCD strain (22). The putative RexA nuclease motif (RexA^Nuc) was modified by a 3-amino-acid deletion removing the tripeptide DYK in positions 1114 to 1116 (called RexA^DYKB) (Fig. 1). The putative RexB nuclease motif (RexB^Nuc) was modified by alteration of D⁹¹⁰ to A (called RexAB^D910A) or by a 3-amino-acid deletion, removing the tripeptide DYK in positions 910 to 912 (called RexAB^DYK) (Fig. 1). In addition, a 293-amino-acid in-frame deletion of RexB that removed C-terminal amino acids 771 to 1063 (called RexAB^771-1063) was constructed. The cloned rexAB mutated genes gave rise to pRex plasmids bearing the name of the mutation and were established in an E. coli recBCD strain (37).

DYK

771-1063

DYK

View this table: [in this window] [in a new window]   TABLE 1.   Capacity of RexAB nuclease mutants to complement a recBCD strain

Changes in RexA or RexB nuclease motifs affect RexAB exonuclease activities. Phage T4g2 is susceptible to exo/hel degradation. As nuclease activity inhibits plaque-forming ability, plaque formation would reflect diminished nuclease activity in the rexAB mutants (Table 1). The recBCD mutant containing pRexAB efficiently inhibits phage multiplication, whereas this strain lacking rexAB genes is totally permissive (22). The strain containing pRexA^DYKB allowed efficient phage multiplication. Thus, inactivation of the nuclease motif of RexA essentially abolishes all DNA degradation activity by RexAB.

DYK

771-1063

RexB^Nuc mutants recognize Chi. We previously developed an in vivo test to detect Chi attenuation of exo/hel exonuclease activity by using an RC plasmid as substrate. A -formed replication intermediate of RC plasmids provides a dsDNA end as an entry point for exo/hel enzyme. If the RC plasmid contains a Chi site in the orientation recognized by exo/hel, degradation is attenuated and -form replication results in accumulation of HMW (14, 26). In the absence of Chi on the plasmid, HMW do not accumulate as long as exo/hel is active. However, if exo/hel is nuclease defective, any RC plasmid will accumulate HMW (e.g., RC plasmids accumulate large amounts of HMW in an E. coli recD mutant; our unpublished data).

DYK

DYK

771-1063

View larger version (53K): [in this window] [in a new window]   FIG. 2.   HMW accumulation in the presence of wild-type or mutant RexA or RexB subunits. The upper portion shows the schema of HMW accumulation. RC plasmid replication may result in formation of a -shaped intermediate with a dsDNA tail. This tail is susceptible to exonuclease (Exo) degradation, and a monomeric circle is restored (left). If the strain is exonuclease defective, or if Chi is present on the RC plasmid, the  form is extended and HMW accumulates (right). The bottom portion shows E. coli recBCD derivatives containing plasmid pRC2 with ChiLl (marked Ll) or without (marked Ll0) as well as a plasmid that carries a rexAB allele or no rexAB, as indicated above the wells. Cells were grown to mid-logarithmic phase at 34°C, and whole-cell lysates were prepared. HMW were detected by Southern hybridization using a pRC2 DNA fragment as the probe. Positions of HMW and supercoiled monomer plasmid (sc) migration are indicated. Note that results shown for RexABDYK are from a separate gel.

DYK

RexB^Nuc but not RexA^Nuc mutants display a Chi-stimulated hyperrecombination phenotype. Gene replacement with linear DNA fragments is stimulated if correctly oriented Chi sites are present in the linear DNA flanking regions of homologies (15, 22, 23). Using this criterion, it was previously demonstrated that the presence of Chi_Ll stimulates homologous recombination (22, 23). We examined the ability of mutant RexAB exo/hel to mediate Chi_Ll-stimulated homologous recombination by using linear fragments with or without double Chi_Ll sites at the ends (Fig. 3 and Table 2). The E. coli recBCD recipient contained the recombination target plasmid (pBla) together with a plasmid encoding a mutated rexAB allele. As expected, very few recombinants were obtained in the recBCD host, regardless of whether Chi_Ll were present on the linear dsDNA ends. Recombination via wild-type RexAB was stimulated 30-fold by the presence of Chi on incoming linear fragments, in keeping with previous results (22, 23). In the presence of pRexA^DYKB, recombination was at background levels, further confirming that a change in the RexA nuclease motif results in total inactivation of RexAB biological activities in vivo.

View larger version (37K): [in this window] [in a new window]   FIG. 3.   Strategy to test ChiLl effect on RexAB mutant-mediated homologous recombination. The recombination target plasmid pBla bears an internal deletion of bla (bla). Linear transforming DNA contains an internal fragment of bla which spans the bla deletion (black rectangle) plus an additional 360 bp of flanking homology (dark grey rectangle) with pBla (called blaint). Where present, double ChiLl sites on the linear fragments are oriented for recognition to enhance recombination and are represented by (RexAB recognizes the arrow tail). Wavy lines represent heterologous dsDNA tails. Double-crossover homologous recombination is required to convert cells to being ampicillin resistant (bla+). Hatched rectangles represent bla DNA outside the regions present on linear DNA. The figure is essentially the same as one published in reference 22, with permission from the National Academy of Sciences.

DYK

771-1063

771-1063

DYK

DYK

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

RexAB exo/hel enzyme function involves two active nuclease sites. The E. coli RecBCD-Chi couple has served as the prototype for bacterial DSB repair. Indeed, exo/hel-Chi couples identified in other bacteria were confirmed to fulfill the biological or biochemical functions established in E. coli (9, 22, 42, 52). However, RecBCD and the L. lactis exo/hel enzyme, RexAB, differ operationally: RecBCD relies on a single nuclease to effect DNA degradation (51, 53). In contrast, we have shown that RexAB contains two nuclease motifs, one in each subunit, both of which are required for full nuclease activity. Inactivation of the RexA nuclease motif results in total loss of exo/hel functions in vivo, whereas inactivation of the RexB nuclease motif reduces degradation while retaining Chi activity. As these two nuclease motifs are present in all identified (or predicted) two-subunit exo/hel enzymes, we predict that these enzymes will have properties similar to those described here.

View larger version (35K): [in this window] [in a new window]   FIG. 4.   Model for RexAB activity. (A) Normal RexAB. RexAB advances from an end on its dsDNA linear substrate via the RexA-driven helicase. RexB is proposed here to increase processivity, as does RecC of RecBCD (38). The RexA nuclease motif degrades from the 3' end until the exo/hel enzyme reaches a Chi site. The conformational change at Chi attenuates RexA nuclease activity and thereby stimulates homologous recombination. The RexB nuclease degrades the dsDNA substrate from the 5' end (bottom strand). At ChiLl, degradation is essentially unchanged or possibly enhanced, as reported, in vitro for the bottom strand after RecBCD encounters an E. coli Chi site (3). (B) Mutant affected in the RexB nuclease motif. Activities are as in panel A, except that the RexB nuclease is inactive. As such, although the 3' end is degraded by RexA nuclease, a 5' end is generated that can act as a substrate in homologous recombination, as shown from previous in vitro and in vivo data (35, 39). This can explain elevated levels of recombination in the assay using linear DNA fragments lacking ChiLl (Table 2). Upon a ChiLl encounter, RexA nuclease is attenuated, thus making both DNA strands accessible for recombination. A and B refer to RexA and RexB subunits, respectively. ANuc and BNuc correspond to the nuclease domains surrounding the motifs presented in Fig. 1. An open mouth has an active nuclease, while a barred mouth represents an inactive nuclease. AHel corresponds to the RexA helicase domain (deduced from BLAST homologies with PcrA helicase).

DYK

DYK

DYK

DYK

Why is exo/hel so poorly conserved? The primordial need for an intact genome suggests that DNA genome repair mechanisms have been present early in evolution. Accordingly, generalized homologous recombination proteins such as RecA, SSB, and RecF are rather highly conserved (19) (comparisons analyzed using BLAST; http://www.ncbi.nlm.nih.gov/BLAST/).