DNA double‐strand breaks caused by replication arrest

Bénédicte Michel, S.Dusko Ehrlich, Marilyne Uzest

Author Affiliations

  1. Bénédicte Michel1,
  2. S.Dusko Ehrlich1 and
  3. Marilyne Uzest1
  1. 1 Laboratoire de Génétique Microbienne, Institut National de la Recherche Agronomique, F‐78352, Jouy en Josas, Cedex, France
View Full Text


We report here that DNA double‐strand breaks (DSBs) form in Escherichia coli upon arrest of replication forks due to a defect in, or the inhibition of, replicative DNA helicases. The formation of DSBs was assessed by the appearance of linear DNA detected by pulse‐field gel electrophoresis. Processing of DSBs by recombination repair or linear DNA degradation was abolished by mutations in recBCD genes. Two E.coli replicative helicases were tested, Rep, which is essential in recBC mutants, and DnaB. The proportion of linear DNA increased up to 50% upon shift of rep recBTS recCTS cells to restrictive temperature. No increase in linear DNA was observed in the absence of replicating chromosomes, indicating that the formation of DSBs in rep strains requires replication. Inhibition of the DnaB helicase either by a strong replication terminator or by a dnaBTS mutation led to the formation of linear DNA, showing that blocked replication forks are prone to DSB formation. In wild‐type E.coli, linear DNA was detected in the absence of RecBC or of both RecA and RecD. This reveals the existence of a significant amount of spontaneous DSBs. We propose that some of them may also result from the impairment of replication fork progression.


DNA double‐strand breaks (DSBs) can have multiple origins including exogenous factors, such as radiation or chemical DNA‐damaging agents (Krasin and Hutchinson, 1977; d'Andrea and Haseltine, 1978), or site‐specific endonucleases (Murialdo, 1988; Haber, 1995). Since they represent a lethal lesion, all organisms have developed repair pathways. The most economical is a simple ligation with another available DNA end. This pathway is preponderant in higher eukaryotes but has serious disadvantages, since eventually it is accompanied by the loss of genetic material and may even lead to gross chromosomal rearrangements (Roth and Wilson, 1989; Lübrich et al., 1995). The most elegant repair pathway is conservative and uses recombination with an intact homologous copy. These recombination events require RecA and RecBCD in Escherichia coli and proteins of the Rad 52 pathway in yeast (reviewed in Clark and Sandler, 1994; Kowalczykowski et al., 1994; Myers and Stahl, 1994; Haber, 1995). The RecBCD protein presumably acts at the initiation stage of homologous recombination, combining degradation (exonuclease V) and helicase activities to render the DNA single‐stranded, and therefore accessible to the RecA protein (reviewed in Kowalczykowski et al., 1994). It is essential for the repair of DNA DSBs induced by DNA‐damaging agents (Wang and Smith, 1983, 1986; Sargentini and Smith, 1986) and binds to double‐stranded DNA ends in vitro (Ganesan and Smith, 1993).

It has frequently been observed that only a minority of recB or recC mutant cells in culture form colonies when plated (Capaldo‐Kimball and Barbour, 1971). This was thought to result from a defect in the repair of spontaneously occurring DSBs, the origin of which remained mysterious. Recently, several observations suggested that DSBs may occur during replication. Two classes of models were proposed, depending on whether the break precedes or follows the passage of the replication fork. DSBs may result from the encounter of a replication fork with a pre‐existing nick or gap present in the template strand of the DNA (Kuzminov, 1995a). Such events are probably rare in wild‐type cells, since E.coli synthesizes enough ligase to repair a large number of nicks (Heitman et al., 1989). However, they could take place following the action of specific E.coli endonucleases, such as the SbcCD endonuclease, which cuts hairpin structures formed at palindromic DNA sequences (Connelly and Leach, 1996), or in ligase‐ or polA‐deficient strains (Cao and Kogoma, 1995; Kuzminov, 1995a). In addition, Howard et al. (1994) observed in vitro that a T4 topoisomerase II cleavage complex formed in the presence of a topoisomerase inhibitor is converted to a DSB following the action of a helicase.

DSBs could also result from the breakage of arrested replication forks. Progression of the replication fork may be impeded by DNA secondary structures, DNA damage or DNA‐bound proteins, which inhibit the polymerases or the helicases. These replication arrests could lead to DNA breakage (Bierne et al., 1991; reviewed in Bierne and Michel, 1994; Kuzminov, 1995b). Experiments supporting the existence of a link between a defect in a replicative helicase and the occurrence of DSBs were reported for two E.coli helicases, DnaB and Rep. DnaB is the only helicase essential for chromosome replication, and two independent observations suggest that DnaB inactivation or blockage creates entry sites for the RecBCD complex: (i) in several dnaBTS mutants, recB‐dependent DNA breakdown has been observed upon shift to restrictive temperature (Buttin and Wright, 1968; Fangman and Russel, 1971; Weschler and Gross, 1971); (ii) the replication terminator sites of E.coli, Ter sites, which arrest replication by inhibiting DnaB action (Khatri et al., 1989; Lee et al. 1989), are hotspots for RecBCD‐dependent homologous recombination (Horiuchi et al., 1994; Horiuchi and Fujimura, 1995). A second replicative helicase in E.coli is the Rep helicase. Although not essential for E.coli viability, it is required for the ‘normal’ replication of the chromosome (Lane and Denhart, 1974). About 80 min are required for the replication of the entire chromosome in rep mutants, versus 40 min in wild‐type strains (Lane and Denhart, 1975; Cosalanti and Denhart, 1987). Based on in vitro data, it was proposed that Rep might facilitate the progression of replication forks by removing DNA‐bound proteins (Yancey‐Wrona and Matson, 1992; Matson et al., 1994). We recently reported that rep recB and rep recC double mutants are inviable and proposed that, in rep strains, DSBs occur at a higher rate than in wild‐type strains, thus rendering the RecBCD‐dependent recombination repair essential (Uzest et al., 1995). Since chromosome replication is slowed down in rep cells, breakage could result from frequent replication pausing. The breaks could either be repaired or lead to cell lethality, as described in the model shown in Figure 1.

Figure 1.

Model for lethality of rep recBC and rep recA recD strains. Arrest of a replication fork due to the lack of the Rep helicase leads to breakage of the fork. In the presence of RecBCD and RecA (a) the DSB is repaired and replication can restart. In rep recA strains (b), the broken arm of the chromosome is degraded by RecBCD, and replication can restart from the chromosomal origin. When RecBC or both RecA and RecD are defective (c), no degradation or repair can take place, and a second breakage will lead to the formation of a linear chromosome (adapted from Uzest et al., 1995).

In the present work, the occurrence of chromosome breakage upon inactivation of replicative helicases was investigated by pulse‐field gel electrophoresis (PFGE) of bacterial chromosomes. In contrast to linear DNA which migrates in pulse‐field gels, E.coli chromosomes do not enter the gels, probably due to their large size and circular structure (Birren and Lai, 1993). Forked DNA, which is Y‐shaped or σ‐shaped, also remains in the wells (Petit et al., 1992; Nakayama et al., 1994). Consequently, PFGE allows the detection and quantification of intracellular linear DNA. Our data confirm the occurrence of spontaneous DSBs in strains defective for the Rep helicase. DSBs are detectable only in the absence of the repair enzymes RecA and/or RecBCD. They are associated with ongoing replication since a chromosomal initiation block, such as a dnaATS mutation, abolishes breakage. Blockage of the replication forks by a dnaBTS mutation or insertion of an improperly oriented Ter site in the chromosome also results in increased breakage.


Linear DNA can be detected in recBC and recA recD E.coli mutants

In order to study the effect of helicase mutations on the breakage of E.coli chromosomal DNA, a method allowing the direct detection of DSBs has been developed. The presence of linear DNA in vivo was investigated by using PFGE combined with cell lysis in agarose plugs. Cells of E.coli grown in the presence of [3H]thymidine were lysed in agarose plugs. The plugs were subjected to PFGE, and the gels were stained with ethidium bromide to visualize the DNA and then sliced. The tritium label present in each slice was counted, and the proportion of migrating DNA was determined. Under the conditions used, the majority of chromosomal DNA from wild‐type cells remained in the well (Figure 2; Table I). When the same plugs were treated with a restriction enzyme, nearly all the DNA entered the gel (not shown). This indicates that chromosomes remain intact upon gentle lysis in plugs and that only linear DNA migrates in these experiments, as reported previously (Birren and Lai, 1993).

Figure 2.

The proportion of linear DNA is increased in recB and in recA recD mutants. Overnight E.coli cultures were diluted in minimal medium supplemented with 0.4% glucose, 0.2% casamino acids and 5 μg/ml of deoxyadenosine at 37°C at OD650 0.01 or 0.02 and grown until OD650 0.2‐0.5. Agarose plugs were prepared and used for PFGE as described in Materials and methods. Gels were cut in 3 mm slices and the amount of [3H]thymidine present in each slice was measured. The total amount of tritium per lane varied from ∼45 000 to 55 000 c.p.m., hence, in order to render the different lanes directly comparable, the values shown in the figure were corrected for a total amount of 50 000 c.p.m. in each lane. The gel origin is not shown. The position of the size markers is indicated. Under the conditions used, S.cerevisiae and H.wingei 0.2‐1 Mb chromosomes migrated as individual bands, while 1.8‐3 Mb chromosomes co‐migrated as a diffuse band. Results from a representative experiment are shown.

View this table:
Table 1. Linear DNA in recBCD cells

recBC mutants are defective for both the RecBCD‐mediated recombination repair of DSBs and the degradation of linear DNA by exonuclease V (exoV) activity. In recBC mutants, linear DNA was detected, mostly migrating as 2‐3 Mb controls (Figure 2). The total amount of linear DNA reached 15‐20% (Table I). recD strains are devoid of exoV activity but proficient for homologous recombination (Amundsen et al., 1986). They showed low amounts of migrating DNA (Table I). However, in the recA recD mutant, where both recombination repair and degradation of linear DNA are inhibited, a significant proportion (25%) of DNA entered the gel (Figure 2; Table I). These experiments indicate that upon inactivation of RecBC or RecA and RecD, broken chromosomes accumulate in vivo.

rep recBTS recCTS mutants accumulate linear DNA upon incubation at the restrictive temperature

It was proposed previously, on the basis of genetic evidence, that DSBs form frequently in cells lacking the Rep helicase (Uzest et al., 1995). To follow directly the amount of broken DNA produced in rep mutants, it appeared necessary to prevent the degradation and repair of these DSBs by controlling the level of RecBCD activity. recBTS recCTS mutations, which inactivate the RecBCD protein at 42°C, were used for this purpose. Δrep::kan recBTS recCTS cells were grown at 30°C in the presence of [3H]thymidine. After two generations, part of the culture was shifted to 42°C, where inactivation of the RecBC protein leads to the loss of cell viability (Uzest et al., 1995; Figure 3A and B). Aliquots were removed periodically for PFGE analyses (Figure 4). An increasing amount of linear DNA, migrating as 2‐3 Mb linear controls, appeared upon incubation at 42°C. The proportion of linear DNA increased to as much as 50% compared with 15% in the recBTS recCTS mutant (Figure 5). Similar results were obtained in rep71 recBTS recCTS cells (not shown). These results show that the chromosomes of rep‐defective cells suffer more breaks than those of wild‐type cells, and the resulting linear DNA can be detected in the absence of RecBCD.

Figure 3.

Loss of viability in various strains upon shift to restrictive conditions. (A) JJC505 (rep recBTS recCTS), OD650; other panels, colony‐forming units (c.f.u.). (B) JJC505. (C) JJC213 [pSF117] (rep). (D) JJC356 [pSF117] (rep recA). (E) JJC507 (dnaB8 recB). (F) JJC438 [pBadTus] (InvTer recB). Overnight cultures were diluted in minimal medium supplemented with 0.4% glucose, 0.2% casamino acids and 5 μg/ml of deoxyadenosine at 30°C at OD650 0.01 or 0.02, with the exception of JJC507, grown in LB, and JJC438 grown at 37°C. For plasmid‐containing cultures, 100‐200 μg of ampicillin were added. When the cultures reached OD650 ∼0.1, they were divided in two, and one part was incubated at 42°C (time 0); for JJC438 (F), 0.2% of arabinose was added at time 0. Aliquots were taken every hour to measure the OD650 and c.f.u. Plates were incubated for 24‐48 h at 30°C; for JJC438 at 37°C. Results from representative experiments are shown. □, 30°C; ▪, 42°C. In (F), JJC438, open symbols show no arabinose and closed symbols 0.2% arabinose. For the rep recA strain (D), although the OD650 increased 24‐fold in 5 h at 30°C (not shown), the c.f.u. only increased 4‐fold during that time, reflecting the low plating efficiency of the strain (Uzest et al., 1995). Growth of all control strains was as expected (see Uzest et al., 1995).

Figure 4.

DNA breakage in rep recBTS recCTS cells. JJC505 cells were grown as described in the legend to Figure 3 in the presence of 5 μCi of [3H]thymidine. After 3, 4 or 5 h of growth at 30°C, at OD650 ranging from 0.05 to 0.2, cells were incubated for 3, 2 or 1 h at 42°C respectively. The control culture was grown at 30°C for 6 h. DNA was prepared and analysed as described in Figure 2. Results are the mean of two independent experiments.

Figure 5.

The proportion of linear DNA in recBTS recCTS strains increases with incubation at 42°C. JJC330 (rep+) and JJC505 (Δrep::kan) cells were grown and treated as described in the legend to Figure 4. The proportion of linear DNA was calculated. Results are the mean of 2‐5 independent determinations. □, JJC330; □, JJC505.

Little or no DSBs were detected in rep single mutants (Table II), indicative of efficient DSB repair and/or linear DNA degradation by the RecBCD enzyme. As expected, in rep recA strains, which are proficient for exoV activity, or in a rep recD mutant, which is proficient for homologous recombination, the amount of migrating DNA was unchanged compared with the isogenic rep strains (Table II). The rep recA recD triple mutants could not be tested, as they are non‐viable (Uzest et al., 1995) and thus DSBs in rep mutants can only be detected in recBTS recCTS mutants, where both exoV and recombination can be inactivated.

View this table:
Table 2. Linear DNA in rep cells

Generation of linear DNA in rep strains in the presence of the Gam protein

Induction of the λ Gam protein, which inactivates RecBCD (Unger and Clark, 1972), is also lethal for rep mutants (Uzest et al., 1995), although residual RecBCD‐dependent recombination activity is detectable in Gam‐containing cells (Friedman and Hays, 1986; Murphy, 1991). We therefore used Gam to investigate further the formation of Rep‐dependent DSBs. Cultures of cells containing a temperature‐inducible gam gene (plasmid pSF117, Friedman and Hays, 1986) were grown at 30°C (Gam). An aliquot of the culture was shifted to 42°C (Gam+) for 3 h and the chromosomes were analysed by PFGE (Table III). The proportion of migrating DNA was at most 2‐fold higher upon Gam induction in rep than in wild‐type strains, in spite of a clear loss of viability (Figure 3C).

View this table:
Table 3. The effects of Gam on linear DNA

To check that residual recombination is responsible for the relatively low amount of linear DNA in rep Gam+ cells, rep recA gam+ cells, where homologous recombination is totally inactivated by the recA mutation, were used. The lethality upon Gam induction is increased compared with that of rep single mutants (compare Figure 3C and D). The proportion of linear DNA in the rep recA gam+ strain reached 50‐60% of total DNA (Table III). The migration properties of linear DNA in the rep recA gam+ strain were similar to those observed in the rep recBTS recCTS strain (see Figure 4).

In recA mutants, both SOS induction and homologous recombination are abolished. To determine which of the two effects caused the accumulation of linear DNA, we used a lexA ind strain, which is proficient for recombination but defective in the SOS response. Inactivation of only the SOS response did not lead to a significant increase in the amount of linear DNA (compare rep lexA ind gam+ and rep gam+ strains, Table III). We conclude that the accumulation of linear DNA in a rep recA gam+ strain results from the inactivation of the homologous recombination function of RecA.

In rep Gam+ cells, recombination could proceed by the RecF pathway or result from the residual RecBCD activity. The amount of linear DNA was similar in rep gam+ and rep recF gam+ cells (Table III). This suggests that, in the rep gam+ cells, the conversion of linear DNA to a non‐migrating form is mediated by RecA and residual RecBCD activity.

The Gam protein also inactivates the SbcCD protein (Kulkarni and Stahl, 1989), which exhibits endo‐ and exonucleolytic activities in vitro (Connelly and Leach, 1996). Both the presence of a high level of linear DNA in rep recA gam+ cells and the lack of effect of sbcCD mutations in the rep71 recBTS recCTS strain (strains JJC484 and JJC498, Table IV; data not shown) indicate that the SbcCD nuclease does not have a major role in the formation of DSBs in rep cells.

View this table:
Table 4. Strains and plasmids

The accumulation of linear DNA in rep recA gam+ cells is DnaA dependent

In a previous report, we proposed that these breaks could occur upon blockage of the replication forks (Uzest et al., 1995; Figure 1). To test directly whether rep‐dependent formation of linear DNA requires the presence of replication forks, a dnaATS mutation was used. The DnaA protein is essential for the initiation of chromosomal replication but not for its propagation. Therefore, after shifting a dnaATS mutant to the restrictive temperature, ongoing replication is not affected while no new replication rounds are initiated at the origin (Kornberg and Baker, 1992). As we could not construct a rep recBTS recCTS dnaATS mutant (not shown), we took advantage of the Gam‐inducible system to test the involvement of replication in DSB formation. The dnaATS rep recA strain, (strain JJC481, Table IV) carrying the plasmid encoding Gam (pSF117) was used. In this strain, incorporation of tritiated thymidine was arrested after 1‐2 h at the restrictive temperature, in contrast to the dnaA+ control strain (not shown). During the first 1.5 h at 42°C, while replication still took place, the amount of linear DNA increased. When replication stopped, the amount of linear DNA ceased to increase (Figure 6). As expected, in the control dnaA+ strain, the amount of linear DNA continuously increased during 3 h (Figure 6). This shows that formation of linear DNA requires the presence of replication forks.

Figure 6.

Effect of a dnaATS mutation on the proportion of linear DNA in rep recA [pSF117]. Cells were grown as described in the legend to Figure 3. An aliquot was grown at 30°C (time 0) and two aliquots were shifted to 42°C for 1.5 and 3 h respectively. Results are the mean of two independent determinations for rep recA [pSF117] (▪) and three independent determinations for rep recA dnaATS[pSF117] (□).

Inactivation or inhibition of the DnaB helicase leads to the formation of DSBs

DnaB is the only essential helicase in E.coli, as it catalyses the unwinding of replicating chromosomes. Several lines of evidence, including RecB‐dependent DNA degradation in dnaBTS mutants (Buttin and Wright, 1968) and elevated rates of RecBCD‐dependent recombination in the region next to replication arrest sites (Horiuchi et al., 1994), suggested that inhibition of the DnaB helicase might also induce DSB formation. A recB::Tn10 mutation was introduced in a dnaB+ and a dnaBTS isogenic strains (JJC506 and JJC507, Table IV). The amount of linear DNA was measured at different times after shift to 42°C by PFGE (Figure 7). The percentage of linear DNA increased rapidly in the dnaBTS mutant at the restrictive temperature, confirming the hypothesis that replication block effectively leads to DSB.

Figure 7.

The proportion of linear DNA increases upon incubation of recB dnaBTS cells at 42°C. Cells were grown in LB medium plus 5 μg of [3H]thymidine. Part of the culture was grown at 30°C for 6 h. Aliquots were shifted to 42°C for 0.5, 1, 1.5 and 3 h for the dnaBTS mutant and for 1.5 and 3 h for the DnaB+ control. Results are the mean of two independent determinations. □, JJC506; ▪, JJC507.

Replication in E.coli is naturally arrested at sites named Ter, located in the terminal region of the chromosome. Ter sites act by inhibiting DnaB action when bound by their cognate protein Tus. To investigate the effects of fork arrest at Ter sites on DSB formation, we made use of a strain carrying two inverted Ter sites in the terminus of E.coli (Inv‐Ter structure) in which Y‐structures typical of arrested replication forks accumulate in the presence of Tus (Sharma and Hill, 1995). As expected, overproduction of Tus, encoded from an arabinose‐dependent promoter (plasmid pBadTus, Table IV), was lethal in the Inv‐Ter strain carrying a recB mutation (Figure 3F). In the absence of Tus, the proportion of linear DNA was 17%, as in other recB cells. This proportion increased up to 30% in 3 h of Tus induction. This confirms that DSBs are generated efficiently upon a replication fork block due to the inhibition of the main replicative helicase, DnaB.


In this work, we examined the susceptibility of the chromosomes to DSB formation under conditions in which different replicative helicases were impaired. In the absence of exoV and recombination repair, a significant amount of linear DNA is detected, revealing a surprisingly high level of spontaneous DSBs. Inhibition of replicative DNA helicases further increased the amount of linear DNA, indicating that arrested replication forks are susceptible to breakage. This suggests a role for replication blockage in spontaneous DSBs and points to a role for helicases in the prevention of replication‐associated DSBs, possibly by a limitation of replication pauses. The main phenotype of cells lacking such a helicase is a dramatic need for efficient DSB repair.

DSB repair

This study also allowed the investigation of DSB repair in the absence of any DNA‐damaging agent. Linear DNA was only detected in significant amounts in recBC or recA recD mutants (see Table I). The elevated proportion of linear chromosomes found in these mutants correlates with their low colony‐forming ability. This confirms the central role of the RecBCD pathway in the repair of linear DNA in E.coli, via its dual action as a double‐stranded DNA exonuclease and recombinase.

It is interesting to note that the amount of linear DNA increased with temperature in a recB mutant (Figure 7). This could reflect an increased number of replication forks at high temperature, and consequently more opportunities for fork breakage. DSB repair in the rep mutant proceeds via the same pathway, since (i) the proportion of linear DNA increases upon inactivation of RecBC in the rep recBTS recCTS strains and (ii) rep recD and rep recA strains which are viable do not accumulate linear DNA, while unrepaired DSBs may be responsible for the rep recA recD triple mutant inviability.

Results in the λ Gam‐containing cells revealed more complex repair pathways. In rep recA gam+ cells, there is a clear correlation between lethality and DSB formation. This is not the case for rep Gam+ cells which contain only 11‐17% linear DNA. The linear DNA could be converted to a non‐migrating form by RecA‐dependent homologous recombination. Alternatively, RecA could protect the chromosomes specifically from DSB formation in the rep gam+ cells. Neither mechanism, however, restores cell viability, possibly because proteins remain bound to the DNA, leading to structures which do not allow replication restart. A similar ‘lethal repair’ may take place in rep recBCD sbcBC and rep recBCD sbcA strains, which would account for the intriguing observation that these mutants are inviable, although they are recombination proficient via the alternative RecF and RecE recombination pathways, respectively (Uzest et al., 1995). Repair of DSBs by the wild‐type RecBCD enzyme appears to be the only way to restore viable chromosomes in rep strains.

Among the proteins proposed to be involved in restarting stalled or repaired replication forks are the primosome assembly protein PriA and the RecF protein (Kogoma et al., 1996; Sandler, 1996). PriA and RecF mutations, therefore, were tested for their effect in rep strains. However, as PriA is essential for RecBCD‐dependent homologous recombination (Kogoma et al., 1996; Sandler et al., 1996), it is expected to be essential in rep strains. Effectively we could not construct a rep3, priA::kan double mutant (our unpublished result). RecF does not seem to play a major role in rep strains, as we detected no effect of the recF::Tn5 mutation in either rep, rep recD or rep gam+ strains. In addition, the high level of replication‐dependent breakage reported here in cells which are proficient for both PriA and RecF indicates that neither of these two proteins allows replication restart in the absence of Rep.

DSB formation and helicase inhibition

The present work shows that impairment of the progression of replication forks leads to DSB formation. Inhibition of the E.coli helicase DnaB leads to chromosome breakage and replication‐dependent DSBs occur in rep mutants. Studies on the progression of replication forks in rep mutants revealed a slow but constant rate of replication (Lane and Denhart, 1975; Cosalanti and Denhart, 1987). This suggests that replication pauses occur all over the chromosome. Replication pauses due to collisions with the transcription apparatus or DNA‐bound proteins were observed in vivo and in vitro (French, 1992; Vilette et al., 1992; Liu and Alberts, 1995), and the Rep protein removes DNA‐bound proteins in vitro (Yancey‐Wrona and Matson, 1992). Therefore, the Rep protein could prevent the formation of DSBs by limiting replication pauses.

The mechanism of chromosomal breakage is unknown. Are the breaks at arrested replication forks due to enzymatic cleavage of DNA, and in this case are there specific nucleases involved, or do DNA breaks occur spontaneously, as a result of tensions consequent to cell growth? The two best characterized single‐stranded endonucleases in E.coli are RecBCD and SbcCD. However, if these endonucleases are responsible for the introduction of breaks at arrested replication forks, they are dispensable, since a significant amount of breakage occurred in the rep recBTS recCTS sbcCD mutants at restrictive temperature. Analysis of appropriate mutants should allow the identification of functions that promote breakage at arrested replication forks.

A rep homologue has been identified in Staphylococcus aureus, where this gene is essential (Iordanescu, 1993). Whether the absolute requirement for the rep gene in this species correlates with a limited capacity for DSB repair is unknown. The role proposed here for the Rep helicase is reminiscent of that proposed for the T4 Dda helicase which shares no homology with Rep (Bedinger et al., 1983). Recently, a correlation between the defect in a putative helicase and increased DSBs was described in humans. Two genetic disorders, Bloom's syndrome and Werner's syndrome, were found to result from a defect in proteins sharing homologies with prokaryotic and yeast helicases (Ellis et al., 1995; Yu et al., 1996). Indicators of defective DNA metabolism in these syndromes include chromosomal instability and hyper‐recombination phenotype, resulting from increased DSBs. In Bloom's syndrome cells, the rate of nascent DNA chain elongation is retarded (Hand and German, 1975; Gianelli et al., 1977). This and several additional observations led to the proposal that the helicase defective in Bloom's syndrome is involved in DNA replication (Ellis et al., 1995). The present work illustrates how, in E.coli, a defect in a replicative helicase can lead directly to increased DSB formation.

Materials and methods

Strains and plasmids

The E.coli K12 strains and plasmids used are listed in Table IV. All strains constructed during this work were made by P1 transduction. P1 stocks were prepared, and transductions performed as described (Miller, 1992). Transductions with stocks grown on strains carrying a malF3089::Tn10, fuc3154::Tn10kan or a malF::cam allele were used as controls for the TetR, KanR and CamR markers respectively. Unless otherwise indicated, all transductions described in this work gave the expected amount of clones with the expected phenotype. Compared with JJC482 (rep71 recBTS recCTS), strain JJC336 (Δrep::kan, recBTS recCTS; Uzest et al., 1995) displays better growth ability and less linear DNA at 30°C. P1 transduction of the rep71 mutation in JJC336 revealed that this could not be attributed to the difference in the rep alleles, and was probably due to the presence of a compensatory mutation in JJC336, selected during growth at 30°C. For this reason, a new Δrep::kan, recBTS recCTS strain, JJC505, was constructed and used in the present work. This strain grows as JJC482.

The Rep phenotype was verified by transforming CaCl2‐competent cells by M13mp2 RF DNA on a lawn of JJC41 (F′) indicator cells. The phenotype of recA, recB and recC mutants was verified by measuring UV sensitivity. Dilutions of cultures in late exponential phase (∼108 cells per ml) were plated on LBAT (Luria broth agar thymine medium) and irradiated at 254 nm. Plates were irradiated with an energy of 20‐120 J/m2 at a rate of 2‐4 J/m2/s. Dose rate was measured with a Latarjet dosimeter. The exo phenotype of recB, recC and recD mutants was verified by measuring the plating efficiency of T4 gene 2 phage, which is increased ∼2000‐fold in these mutants.

Rich medium was Luria broth (Miller, 1992) supplemented with 25 μg/ml thymine (LBT). Minimal medium was medium M63 (a medium devoid of NaCl, Miller, 1992), supplemented with 0.4% glucose and 0.2% casamino acids. Antibiotics were used at the following concentrations: ampicillin (Amp), 40 μg/ml; tetracycline (Tet), 15 μg/ml; kanamycin (Kan), 45 μg/ml; chloramphenicol (Cam), 25 μg/ml; and spectinomycin (Spc), 60 μg/ml. Overnight cultures were grown routinely in 5 ml of LBT broth. Cells were made competent by CaCl2 treatment and transformed as described (Dagert and Ehrlich, 1979).

Preparation of plugs and PFGE migration

Overnight cultures were grown in LBT medium. The cells were diluted to OD650 0.01 or 0.02 in M63 supplemented with 0.4% glucose, 0.2% casamino acids, 100 μg/ml deoxyadenosine and 5 μCi/ml of [3H]thymidine (sp. act. 87 Ci/mmol, New England Nuclear). For JJC338 (thyA deoC), 5 μg/ml of cold thymidine were added, and, due to poor growth in minimal medium, JJC506 and JJC507 (thyA deoC) were grown in LB plus 5 μCi/ml of [3H]thymidine. When 3 or 4 h of incubation at restrictive temperature were required, the culture was divided into two when the OD650 reached 0.07‐0.08. One part was incubated further at 30°C, the other was shifted to 42°C. For shorter times of incubation at restrictive temperature, the shift was performed at a higher OD. Cells were collected (they reached OD650 0.3‐0.7), washed in cold minimal medium and resuspended in Tris 10 mM, EDTA 100 mM, EGTA 10 mM (TEE buffer) at a final OD of 0.7. After 10 min of incubation at 37°C, the suspension was mixed with an equal volume of agarose InCert 1.5% in TEE and allowed to solidify in plug molds. The plugs were then incubated for 2 h at 37°C in 20 ml of TEE buffer containing 0.05% sarcosyl, 5 mg/ml lysozyme solution, and overnight at 52°C in 10 ml TEE buffer containing 1 mg/ml proteinase K, 1% SDS. Plugs were kept in 10 ml of TEE buffer at 4°C and used within 2 weeks.

The 1% agarose gels were run for 40 h at 160 V (6‐7 V/cm) in 0.3× TEB, using a PFGE ramp from 3 ms forward/1 ms backward to 180 ms forward/60 ms backward.

After migration, gels were stained with ethidium bromide and photographed. Plugs containing the chromosomes of the yeast Saccharomyces cerevisiae and plugs containing the chromosomes of Hansenula wingei were used routinely as molecular weight standards (Bio‐Rad). The lanes were cut into 3 mm slices using a razor blades slicer. A total of 36‐42 slices representing 12‐14 cm of migration were usually made to allow the recovery of all fragments from the migration origin to 30‐50 kb fragments. The amount of tritium in the slices corresponding to the smaller fragments was generally low or null. The agarose slices were melted at 90°C, and diluted 5‐fold in 0.1 M HCl at the same temperature. Counting was performed in 10 ml of scintillation fluid (Beckman ready safe) in a Beckman scintillation counter. The fraction of migrating DNA was calculated by dividing the amount of tritium found in a lane by the total amount of tritium found in the lane plus the migration origin. This total amount generally did not differ significantly from the tritium present in an intact plug counted as a control.


We are very grateful to Drs R.D'Ari, R.Devoret, C.Gross, A.Gruss, S.Kushner, D.Leach, R.Lloyd, T.Lohman, S.Sommers and W.Wackernagel for sending strains and plasmids. We are very grateful to Dr T.Hill for providing the Inv‐Ter strain and manuscripts prior to publication, and to Dr D.Leach for providing manuscripts prior to publication. We are particularly grateful to A.Gruss for her enthusiasm and constant support during this work and for essential participation in the redaction of this manuscript. We thank E.Cassuto and M.A.Petit for helpful discussions and corrections to the manuscript, A.Kuzminov for helpful reading of the manuscript, and F.Haimet for her help in producing the artwork. B.M. is on the CNRS staff. This work was supported, in part, by a CEE grant number BIOT CT91‐0268.


View Abstract