Chromosomal damage was detected previously in the recBCD mutants of the Antarctic bacterium Pseudomonas syringae Lz4W, which accumulated linear chromosomal DNA leading to cell death and growth inhibition at 4°C. RecBCD protein generally repairs DNA double-strand breaks by RecA-dependent homologous recombination pathway. Here we show that ΔrecA mutant of P. syringae is not cold-sensitive. Significantly, inactivation of additional DNA repair genes ruvAB rescued the cold-sensitive phenotype of ΔrecBCD mutant. The ΔrecA and ΔruvAB mutants were UV-sensitive as expected. We propose that, at low temperature DNA replication encounters barriers leading to frequent replication fork (RF) arrest and fork reversal. RuvAB binds to the reversed RFs (RRFs) having Holliday junction-like structures and resolves them upon association with RuvC nuclease to cause linearization of the chromosome, a threat to cell survival. RecBCD prevents this by degrading the RRFs, and facilitates replication re-initiation. This model is consistent with our observation that low temperature-induced DNA lesions do not evoke SOS response in P. syringae. Additional studies show that two other repair genes, radA (encoding a RecA paralogue) and recF are not involved in providing cold resistance to the Antarctic bacterium.
Cold adaptation of psychrophilic bacteria includes many genetic and physiological changes that have evolved to overcome the growth barriers inherent to permanently cold environments (Russell, 1998; Cavicchioli et al., 2002; Feller and Gerday, 2003; Chattopadhyay, 2006; Ray, 2006). These adaptive changes are largely shaped by lower thermal energy and low molecular motion, apart from the nature and abundance of nutrients that are available in the surrounding environment. Broadly, psychrophilic proteins and enzymes are adapted to retain molecular motion in and around the ligand-binding and catalytic sites (Cavicchioli et al., 2002; Feller and Gerday, 2003), cytoplasmic membrane lipids contain higher amounts of unsaturated and branched chain fatty acids to enhance molecular disorder and maintain homeoviscosity (Russell, 1998), and the lipopolysaccharides in Gram-negative bacteria show higher amount of hydroxylated acyl chains and lower amount of phosphates for outer membrane functions (Seshukumar et al., 2002). Adaptive changes in other important class of macromolecules, such as DNA and RNA, which can adopt various stable secondary structures at low temperatures, are relatively less studied and little known. Stable RNAs (e.g. tRNAs and rRNAs) in psychrophilic bacteria and Archeae have been noted to have base modifications or altered base compositions which are thought to give structural flexibility at low temperature (Dalluge et al., 1997; Saunders et al., 2003; Khachane et al., 2005). It is generally believed that low temperature stabilized DNA and RNA structures are metabolically modulated by DNA/RNA-binding proteins, and/or DNA/RNA helicases which allow the nucleic acid metabolizing enzymes to overcome the structural barriers (Lim et al., 2000; Purusharth et al., 2005; 2007). However, adaptations of these nucleic acid metabolic pathways in psychrophilic bacteria have not been investigated closely.
Maintaining genome integrity is the first and foremost requirement for psychrophilic bacteria to survive and grow at 0°C or near 0°C. Importance of this aspect of cold adaptation came from our study with the Antarctic bacterium Pseudomonas syringae Lz4W, which displayed cold-sensitivity upon recD inactivation (Regha et al., 2005). RecD functions as a component of the DNA repairing RecBCD complex, and P. syringae mutants lacking any one of the components of RecBCD machinery accumulate linear chromosomal DNA and short DNA fragments in cells, leading to cell death at the low temperature (Pavankumar et al., 2010). Interestingly, although cold stress-induced DNA damage that threatens genomic integrity of psychrophilic bacteria was not quite obvious before our study, inactivation of nucleoid-associated protein H-NS in the mesophilic Escherichia coli was reported to cause cold-sensitive phenotype (Dersch et al., 1994). In Bacillus subtilis, cold shock-induced protein CSPs were observed to localize around the nucleoid concomitant with increased compaction (Weber et al., 2001). Low temperature was also reported to induce change in DNA topology (increased supercoiling) which impacted gene expression. Novobiocin, an inhibitor of DNA gyrase, specifically abolished the cold-induced expression of unsaturated fatty acid synthesis in B. subtilis (Grau et al., 1994). In the Antarctic bacterium P. syringae Lz4W, we observed that supercoiled DNA was preferred template for transcription at low temperature (0–4°C) when transcribed in vitro by cold-active RNA polymerase (Uma et al., 1999). All these studies indicated that bacteria have adapted to maintain proper chromosomal structure and functions at low temperature.
We reported that recBCD mutants of P. syringae are not only cold-sensitive, but also sensitive to many DNA-damaging agents, like UV and mitomycin C (Regha et al., 2005; Satapathy et al., 2008; Pavankumar et al., 2010). More importantly, we observed that in contrast to the model bacterium E. coli, P. syringae recD null mutation abolishes recombinational repair of DNA double-strand breaks (DSBs) as much as a recB or a recC mutations, and consequently specific recB point mutants have to be used to abolish specifically the exonuclease function of RecBCD (Pavankumar et al., 2010). Additionally, by pulsed field gel electrophoresis (PFGE), we showed that recBCD mutants accumulate linear chromosomal DNA as well as shorter DNA fragments, when cells are shifted from 22°C to 4°C. These DNA fragmentations can be correlated to the loss in cell viability and growth defects at low temperature (Regha et al., 2005; Pavankumar et al., 2010). Apart from its role in this cold resistance, RecBCD pathway is also known to protect cells from exogenous stressors, such as NO, H2O2 and reactive oxygen species (ROS)-generating molecules (Carlsson and Carpenter, 1980; Spek et al., 2001; Stohl and Seifert, 2006), and it is essential for the Salmonella enterica virulence in mice (Cano et al., 2002). RecBCD machinery is primarily responsible for repairing the double-strand DNA breaks (DSBs) by homologous recombination (Kuzminov, 1999). The dependence of cell survival on RecBCD has been generally attributed to the repair via recombinational pathway or the suppression of DSBs by linear DNA degradation as happen in UV-exposed cells (Miranda and Kuzminov, 2003; Khan and Kuzminov, 2012). In the recombinational pathway, RecBCD binds to blunt or nearly blunt ends of double-strand DNA (ds-DNA) and processes them with the help of its ATP-dependent helicase and nuclease activities to generate substrates for homologous recombination. Generally, RecBCD degrades both strands of linear DNA very fast till it encounters regulatory Chi (cross-over hotspot instigator) sequence. Chi (χ) in the correct orientation (5′-GCTGGTGG-3′ in E. coli) switches RecBCD degradation to the 5′-ending strand only and slows it down. RecBCD also protects the remaining 3′-ending strand and loads RecA onto it. The RecA bound ss-DNA (RecA nucleoprotein filament) aligns with the homologous sequence of intact chromosome, and RecA catalyses strand exchange between them. The strands exchange reaction leads to formation of Holliday junctions (HJ) which are resolved by RuvABC complex to produce recombinants (Kuzminov, 1999; Dillingham and Kowalczykowski, 2008). Thus, RecA and RuvABC proteins play a crucial role in the recombinational repair pathway. However, it is becoming increasingly clear that, in some cases of replication arrest, the nascent daughter strands of replication forks (RFs) can self-anneal leading to replication fork reversal (RFR) (Michel et al., 2004; Atkinson and McGlynn, 2009). Irrespective of whether RFR occurs spontaneously due to DNA topology or by enzyme catalysis ahead of RFs (Postow et al., 2001; Baharoglu et al., 2006; Fierro-Fernandez et al., 2007; Atkinson and McGlynn, 2009), the four-way DNA junctions at the reversed replication forks (RRFs) are cleaved by RuvABC resolvase and result in chromosome linearization (Seigneur et al., 1998). For rescuing the RRFs, either RecBCD should degrade the linear open end emanating from the four-way junction, or RecA, with the help of RecBCD, should homologously invade this end into the parental duplex (Michel et al., 2004; Atkinson and McGlynn, 2009). Thus RecBCD function is crucial for both pathways.
In P. syringae Lz4W, although RecBCD mediated DNA repair is essential for low-temperature growth, nothing is known in respect to why DNA damage would occur so critically only at low temperature, or which critical RecBCD activity is important for cell survival and growth, or whether recF pathway is involved in the cold stress-induced DNA damage repair. The present study addresses these questions, and examines the consequences of cold stress on P. syringae Lz4W in vivo by analysing different DNA repair gene mutations (recA, radA, ruvAB, recF) in the presence and absence of RecBCD machinery, in an effort to understand the mechanism of RecBCD mediated protection of chromosomal integrity, allowing psychrophilic bacteria to grow in cold environments.
Cold stress-related DNA damage does not induce SOS response
Previous work has shown that recB, recC and recD mutants of P. syringae Lz4W are cold-sensitive, fail to grow at 4°C, and accumulate linear chromosomal DNA and short DNA fragments, as a result of chromosomal breaks that remain unrepaired at low temperature (Pavankumar et al., 2010). As bacteria in general are capable of sensing DNA breaks or other lesions to induce a co-ordinated genetic response, called SOS response, for the repair/bypass of DNA lesions (Friedberg et al., 1995; Michel, 2005) we sought to examine whether P. syringae mounts this response at low temperature. RecA induction is a hallmark for SOS response, and therefore we analysed the cellular levels of RecA by Western analysis following the shift of P. syringae cultures from 22°C to 4°C. As shown in Fig. 1, the low temperature exposure did not induce RecA, either in wild-type (Fig. 1A) or in LCBD (ΔrecBCD) mutant (Fig. 1B). In contrast, RecA levels increased following UV exposure of the cells as a mark of the response at both temperatures (Fig. 1A–C). Thus, evidently cold stress-induced DNA lesions do not induce SOS response in P. syringae.
RecA inactivation does not cause cold sensitivity in P. syringae
The lack of RecA induction led us to ask whether repair of the cold-induced DNA damage involves at all the RecA-dependent recombination pathway. If this pathway is involved, recA mutants would be cold-sensitive. To examine this, we created a RecA-null mutant (ΔrecA) of P. syringae by replacing chromosomal copy of recA gene with a kanamycin resistance gene cassette (recA::kanR), as detailed in Fig. S1 (Supporting information). Using identical strategy we also created ΔrecA ΔrecBCD double mutant, in which recA was disrupted in LCBD (ΔrecBCD) background.
We then compared the growth of ΔrecA with wild-type and LCBD at both high (22°C) and low (4°C) temperatures. To our surprise, ΔrecA mutant grew almost like the wild-type at both temperatures (Fig. 2A and B). On the other hand, ΔrecA ΔrecBCD double mutant (LCBD–ΔrecA) exhibited growth inhibition at 4°C, similar to LCBD parent. The cell survival in 4°C cultures, as measured by cfu analysis (Fig. 2C) was consistent with the growth profiles of recA and recBCD mutants. These experiments suggested that RecA protein does not participate in the repair of cold-induced DNA lesions and therefore does not affect growth of P. syringae at low temperature.
RadA does not protect cells from cold-induced DNA lesions
Many bacteria possess a paralogue of RecA, called RadA (Beam et al., 2002), which might provide RecA-like activity for supporting the growth at low temperature. To address this possibility we created a radA knockout mutant (ΔradA) of P. syringae, using kanR cassette-insertional mutagenesis (Fig. S2A). radA inactivation was also created in LCBD background to create LCBD–ΔradA mutant for checking the effect of double mutations (ΔrecBCD ΔradA). The growth profiles of these mutants (ΔradA and LCBD–ΔradA) at the two temperatures (22°C and 4°C) were comparable to their respective parental strains (Fig. 2A and B). radA single mutant grew similar to wild-type, and the double mutant (ΔrecBCD ΔradA) exhibited cold-sensitivity like the parental LCBD mutant (Fig. 2C). This result suggested that RadA is not involved in the repair of cold-induced DNA lesions, nor participates in the growth of P. syringae at low temperature.
Lethality of ΔrecBCD mutant at low temperature is suppressed by ruvAB inactivation
We also examined the role of RuvAB which is important for DNA repair in the recombinational pathway. RuvAB binds Holliday junctions and resolves them in association with RuvC endonuclease at late stage of homologous recombination to produce the recombinants. In addition, RuvAB plays a role in replication forks reversal (RFR) when replication is arrested in certain DNA polymerase mutants of E. coli (Baharoglu et al., 2006). To investigate the RuvAB role in P. syringae, we created RuvAB-null mutant (ΔruvAB) by kanR insertion, as shown schematically in Fig. S2B. RuvAB inactivation was made in both wild-type and LCBD (ΔrecBCD) backgrounds to create single (ΔruvAB) and double (ΔruvAB ΔrecBCD) mutants, and the effects on growth were examined.
Growth profiles of these mutants at 22°C and 4°C are shown in Fig. 3A and B respectively. The ΔruvAB single mutant was resistant to cold stress like wild-type. However, the double mutation, i.e. ruvAB inactivation in LCBD background, restored the growth of LCBD at low temperature to considerable extent (Fig. 3B). Cell viability assays (Fig. 3C) were also consistent with the growth in cultures of the mutant. However, why LCBD–ΔruvAB double mutant shows a lag of about 72 h before the visible growth (Fig. 3B and C) is not clear to us and needs further investigation on whether additional suppressor mutation occurs. Importantly, introduction of a functional copy of ruvAB on the plasmid pMMruvAB into the double mutant (ΔruvAB ΔrecBCD), reversed the ΔruvAB effect by restoring cold sensitivity (Fig. 3D). At the same time, the pMMruvAB plasmid did not inhibit growth of WT or ΔruvAB at 4°C (data not shown). Thus a role of RuvAB in the suppression of ΔrecBCD growth defect was established in P. syringae. This suppressor effect of ruvAB inactivation is akin to (in terms of partial recovery or viability) what has been reported for replication fork reversal phenotype in some replication mutants of E. coli (Baharoglu et al., 2006), and led us to believe that cold-induced DNA lesions occur largely due to replication arrest, and that RuvAB plays a role in the replication fork reversal in P. syringae under cold stress.
PFGE analysis of genomic DNA suggests that DSBs are produced by a RuvAB-dependent process
Cold stress-induced chromosomal DSBs were apparent in RecBCD-depleted cells of P. syringae at low temperature (Pavankumar et al., 2010). The loss of chromosomal integrity was reflected in accumulation of linear chromosomal DNA and shorter (∼ 30–50 kb) DNA fragments in LCBD mutant (ΔrecBCD), as detected by pulsed-field gel electrophoresis (PFGE). Therefore, we examined the levels of these lesions in ΔrecA, ΔruvAB and LCBD–ΔruvAB mutants in this study. Results (Fig. S4) showed that, at 4°C the levels of DNA lesions in ΔrecA mutant is comparable to wild-type, which is consistent with genetic data that recA inactivation does not affect the low-temperature growth of P. syringae. More importantly, the levels of linear chromosomal DNA and shorter DNA fragments were diminished by additional inactivation of ruvAB in LCBD mutant (Fig. S4). This implies that RuvAB plays a role in the generation of DNA lesions in LCBD under cold stress. The ruvAB inactivation alone however did not lead to DNA damage, as the levels of linear chromosomal DNA were comparable to wild-type at 4°C and consistent with its cold-resistant phenotype. To quantify the levels of DSBs/linear chromosomal DNA in different mutants we used the 32P-orthophosphate labelling method of cellular DNAs (Khan and Kuzminov, 2012) and quantified the PFGE separated chromosomal DNA species. The results (Fig. 4 and Table S2) indicate that the amounts of linear chromosomal DNAs do not exceed 12–14% of total DNA in the wild-type, both at 4°C and at 22°C. However, the increased amount of linear DNA (38% of total) observed in LCBD mutant at 4°C was considerably reduced in the ΔruvAB background. LCBD–ΔruvAB produced ∼ 27% linear DNA (Fig. 4) which is statistically significant (P < 0.0078). On the other hand, linear chromosomal DNA in LCBD–ΔrecA mutant remained high (42%) like in parental LCBD, suggesting that the additional inactivation of RecA does not much influence the DSBs in the LCBD cells under the cold stress. To rule out the possibility that the trappings of RecA mediated DNA recombination intermediates (unresolved HJs) in the wells of PFGE is responsible for the lowering of linear DNA in LCBD–ΔruvAB mutant, we created a LCBD–ΔruvAB–ΔrecA triple mutant (Fig. S3A); this mutant produced ∼ 22% linear DNA, which is statistically similar (P = 0.287) to the value (∼ 27%) observed in LCBD–ΔruvAB. The LCBD–ΔruvAB ΔrecA triple mutant was however cold-sensitive, unlike the LCBD–ΔruvAB double mutant (Fig. S3C and D). We also noted that shorter linear DNAs which appeared to decrease in the LCBD-ruvAB mutant, compared with LCBD, in the ethidium bromide-stained gels (Fig. S4) did not show the reduction in the 32P-labelled experiments (Fig. S5C), presumably due to lesser sensitivity of ethidium bromide-based detection of DNA.
RuvAB dependence of elevated chromosome linearization in P. syringae at low temperature suggests replication fork reversal (RFR), similar to what happens in the rep null mutants of E. coli (Seigneur et al., 1998). In the case of P. syringae, it would be important to know whether RF arrest occurs at specific locations of chromosomes, and how the short DNA fragments are produced from the arrested RF sites, which are possibly cleared by RecBCD exonuclease activity. While the answers to these questions are not known at this point of time, we examined chromosomal distribution of the DNA fragments (∼ 30–50 kb) accumulating in LCBD cells, by probing PFGE separated DNAs in Southern hybridization analysis. Using 32P-labelled DNA probes of four genes (recA, radA, recF and 16S rRNA gene which are likely to occur in different regions of chromosome as per different Pseudomonad genomes) we observed that both circular and linear chromosomal DNA and shorter DNA fragments hybridize to these probes without any noticeable difference (Supplementary Fig. S7). This suggests that the sites of replication arrest, if responsible for generating short DNA fragments in LCBD mutant cells, are not probably restricted to any particular loci of the chromosome.
RecBCD nuclease activity is critical for repair of chromosomal DSBs
Reversed forks are reset either by RecA-dependent recombination pathway or by DNA degradation involving RecBCD nuclease activity in E. coli (Seigneur et al., 1998). Inactivation of RecD alone (causes loss in exonuclease activity of RecBCD) or RecA alone (eliminates homologous recombination) did not affect cell viability, but simultaneous inactivation of both RecA and RecD resulted in accumulation of DSBs and caused cell death of several E. coli replication mutants. To check whether this is the case with P. syringae, we examined the ability of nuclease deficient RecBCD protein (RecBD1118ACD) to support the growth of LCBD lacking RecA (LCBD–ΔrecA), by introducing the plasmid pGCBDAD (Table 2). RecBD1118ACD lacks nuclease activity due to alanine substitution of the catalytic aspartate at 1118 position of RecB nuclease active site (Pavankumar et al., 2010). It is important to note that nuclease deficient RecBD1118ACD protein is capable of supporting the low-temperature growth of LCBD which harbours recA+ allele, probably aided by other exonucleases as shown by RecJ overexpression (Pavankumar et al., 2010). Here we show (Fig. 5) that LCBD–ΔrecA expressing the nuclease deficient RecBD1118ACD protein is cold-sensitive. This implies that RecBCD nuclease activity is critical in recA-null background of P. syringae and that RecA contributes to the repair of arrested RFs by homologous recombination at low temperature.
recF pathway is not involved in the cold resistance of P. syringae
Although our results show that RecBCD plays an essential role in repairing DSBs at the arrested RFs, recF pathway is also known to rescue the UV-induced RF arrests in E. coli (Courcelle and Hanawalt, 2001; Courcelle et al., 2003). Hence, we examined the importance of recF in the cold resistance of P. syringae. We inactivated the chromosomal copy of recF gene by inserting a kanamycin resistance gene cassette (Fig. S2C) in both wild-type and LCBD (ΔrecBCD) backgrounds of P. syringae. The resultant ΔrecF single and ΔrecBCD ΔrecF double mutants were then examined for growth defects (Fig. 6A and B) and sensitivity to UV radiation (Fig. 6C). recF single mutant was observed to be cold-resistant and grew like wild-type at 4°C, but displayed UV sensitivity. On the other hand, ΔrecBCD ΔrecF double mutant displayed cold-sensitivity like the parental LCBD strain and was highly sensitive to UV, much higher (> 100-fold) than the parental single mutants. These results suggest that recF pathway of P. syringae is involved in the repair of UV-induced DNA damage, but does not play any role in cold-resistance, or in rescuing the growth defects of LCBD (ΔrecBCD) at low temperature.
Cold stress and UV radiation produce different type of DNA lesions in P. syringae
A clue to the nature of DNA damage produced by different damaging agents can be drawn from commonality and/or variability in the selection of DNA repair pathways that bacteria employ to protect cells against the agents. Here, we tested the resistance of cold-sensitive and cold-resistant mutants of P. syringae against UV radiation. We observed that the cold-resistant recA mutant was highly sensitive to UV. The ΔrecBCD ΔrecA double mutant (LCBD–recA), which displayed cold sensitivity identical to LCBD (Fig. 2), displayed very high sensitivity to UV radiation, higher than any of the individual mutants (Fig. 7A). The results suggest that the psychrophilic P. syringae uses the RecA-dependent recombination pathway to protect cells from UV-induced DSBs as in other bacteria, and UV-induced DSBs are different from the DSBs occurring at arrested RFs under cold stress.
We also analysed the UV sensitivity of ruvAB deletion mutant individually or in combination with ΔrecBCD mutation of LCBD (Fig. 7B and Fig. S6B). The ΔruvAB mutant displayed modest UV sensitivity. The LCBD–ΔruvAB double mutant displayed negligibly higher sensitivity than any individual mutants, consistent with the observation that E. coli single and double mutants display similar degree of sensitivity (Lloyd et al., 1984; McGlynn and Lloyd, 2000). However, the point to note here is that susceptibility of P. syringae recA or ruvAB mutants to UV (Fig. 7) contrasts markedly to their resistance against cold stress (Fig. 2B and C). More importantly, the UV sensitivity of recBCD mutant was markedly enhanced by additional mutation in recA (Fig. 7A), unlike its effect on cold sensitivity. The cold sensitivity of LCBD either was ameliorated by the additional ruvAB inactivation (Fig. 3A and B), or remained unaffected by recA deletion (Fig. 2B and C). Thus, UV-sensitive but cold-resistant property of recA and ruvAB mutants clearly indicates that DNA damage produced at the arrested replication forks under cold stress or upon exposure to UV are of different kinds, and P. syringae employs separate mechanisms to repair or overcome them. RuvAB-dependent activity causes DNA damage to the arrested RFs at low temperature, while it participates in the repair of UV damaged DNA requiring both RecA and RecBCD.
We also tested the role of radA in protecting cells against UV (Fig. 7A). The radA single mutant displayed UV resistance almost similar to wild-type. recBCD–radA double mutant showed also similar degrees of sensitivity to UV as did the recBCD single mutant. As radA mutant was not also cold-sensitive (Fig. 2), it is inferred that RadA participates in the repair of neither UV nor cold stress-induced DNA damage in P. syringae.
Growth inhibitory DNA lesions at low temperature possibly include very little of oxidative damage
Cold stress and oxidative stress have a strong association (Smirnova et al., 2001). Oxidative DNA damage does occur due to production of many ROS in E. coli (Keyer et al., 1995). It has also been suggested that increased production of ROS at low temperature might be responsible for DNA damage in Antarctic bacteria (Chattopadhyay et al., 2011). We therefore speculated that paraquat, a known inducer of ROS and DNA damage, would also affect the survival and growth of LCBD mutant. If cold stress- and paraquat-induced DNA lesions are similar in nature, the paraquat-induced lesions also can be rescued by RuvAB inactivation, as seen with the alleviation of cold-sensitive phenotype in LCBD–ΔruvAB mutant (Fig. 3B and C). Accordingly, we tested paraquate resistance of wild-type, LCBD, ΔrecA, ΔruvAB and LCBD–ΔruvAB mutants (Fig. 8). LCBD showed much higher (approximately twofold) sensitivity to paraquat compared with recA and ruvAB mutants suggesting that RecBCD plays greater role than RecA or RuvAB, in providing resistance against paraquat. More importantly, additional inactivation of ruvAB was not able to rescue the paraquat sensitivity of LCBD (Fig. 8), unlike that of cold-sensitive phenotype. Taken together, these data indicate that oxidative DNA lesions are repaired by RecBCD–RecA-dependent recombination pathway. The ROS-induced lesions leading to RF collapse are not aggravated by RuvAB activity, and hence ruvAB inactivation fails to offer paraquat resistance to LCBD. The difference in the requirement for RecA, RecBCD and RuvAB in protecting cells against ROS and cold stress-induced DNA lesions suggests that replication fork problems leading to cell death at low temperature are unlikely to contain those that are produced by oxidative stress.
In the present study we have addressed the issue of maintaining genomic integrity in psychrophilic adaptation, because we noted earlier that one of the problems in Antarctic P. syringae Lz4W growing at frigid temperatures lie in the occurrence of DNA damage (Regha et al., 2005). The problem is not normally detected due to efficient RecBCD activity in the bacterium (Pavankumar et al., 2010). Although the role of RecBCD protein complex in the repair of chromosomal DSBs is well established (Dillingham and Kowalczykowski, 2008), the essentiality of RecD subunit in RecBCD activity and in growth of the bacterium at low temperature was unique, and unconventional to our knowledge on the RecD from mesophilic bacterium E. coli (Amundsen et al., 2000). Nevertheless, we demonstrated that RecD ATPase activity, which could be as little as 10% of wild-type, is essential for RecBCD functions in supporting the low-temperature growth of P. syringae (Satapathy et al., 2008). We proposed that RecBCD-dependent recombination pathway is responsible for repairing the cold-induced DSBs and preventing cell death (Pavankumar et al., 2010). In this study we examined the mechanism by which DSBs are produced under cold stress and whether these DSBs are similar to those that are produced by UV radiation and ROS-generating agent paraquat, and whether RecBCD repairs them in collaboration with other DNA repair proteins (e.g. RecA, RadA, RuvAB and RecF) to protect the cells. We primarily used genetic analysis of cell survival and growth of the mutants which were created by inactivation of the above DNA repair genes in the bacterium, and correlated the growth and repair defects of the mutants with chromosomal damage in DNA separated by PFGE. We show that RecA-dependent recombination pathway, which is responsible for repairing UV-induced DNA lesions and provides UV resistance to cells, plays a lesser role in protecting cells from cold-induced DNA damage or supporting the growth at low temperature. The RecA-independent role of RecBCD is critical for repairing the cold stress-induced DNA lesions, which are distinct from the damage caused by UV irradiation. The latter is generally repaired in E. coli by homologous recombination or by UmuCD-dependent translesion DNA synthesis, both of which are RecA-dependent process (Kuzminov, 1999; Courcelle and Hanawalt, 2001; Courcelle et al., 2003).
The requirement for RecBCD and not RecA in resistance to cold stress can be explained by a mechanism of indirect DNA damage resulting from replication fork arrest and fork reversals (Fig. 9) which was originally proposed in case of E. coli rep and dnaBts mutants (Seigneur et al., 1998). DNA double-strand ends of the reversed RFs formed by the self-annealing of nascent DNA strands can be processed/repaired to restart replication by one of the two possible pathways (Seigneur et al., 1998). In the first case, the ds-DNA ends of replicating fork are processed by RecBCD and RecA mediated homologous recombination to reassemble RF for reinitiating replication (step 5 to 1, Fig. 9). This pathway depends upon the presence of Chi (χ) like regulatory sequence on self-annealed DNA strands, in which Chi facilitates RecBCD nuclease activity and RecA loading (Dillingham and Kowalczykowski, 2008). Alternatively, in the second pathway (Fig. 9, step 4 to 1), DNA ends of the reversed forks at stalled sites can be simply degraded by the processive exonuclease (ExoV) activity of RecBCD without using RecA functions, thereby re-establishing RFs to facilitate replication restart. In RecBCD-depleted cells RuvAB proteins would bind to HJs produced by RFR, which in association with RuvC nuclease (RuvABC complex) resolve them, leading to chromosome linearization (Michel et al., 2004; 2007). Our results are consistent with the RFR model of chromosome linearization, as the levels of linear DNA reduced drastically in LCBD–ΔruvAB mutant, compared with cold-sensitive LCBD (ΔrecBCD) parent (Fig. 4), and LCBD–ΔruvAB double mutant regained the ability to grow at low temperature (Fig. 3). Altogether, our results suggest that replication arrest is a threat to the growth of P. syringae at low temperature, and RecBCD activity is important for chromosomal integrity and RF re-establishment at the arrested sites.
Replication fork arrest and damage occur under normal growth conditions, in the absence of any external stressors (Mirkin and Mirkin, 2007). Arrested forks are very much susceptible to DNA damage mediated by RuvAB and RuvC activities, which leads to cell death if unrepaired. Although RuvAB has been proposed to be responsible for the fork reversals in E. coli mutants of an accessory replicative helicase (e.g. rep) and replicative DNA polymerase (dnaEts) (Baharoglu et al., 2006), the role of RecA in fork reversal has been shown only in dnaBts-induced replication arrest (Seigneur et al., 2000; Robu et al., 2001; Michel et al., 2004). Our results with P. syringae suggest that RecA does not play a role in cold-induced RFR, as recA inactivation neither rescues nor aggravates the cold-sensitivity of recBCD mutant. Thus, cold stress effect on RFs in first instance appears to be similar to one exhibited by rep and dnaEts mutants in E. coli. Hence, it is likely that reduced activities of Rep and DNA polymerase at low temperature, in addition to stabilized DNA secondary structures and DNA–protein complexes might be responsible for replication fork arrests and fork damage, limiting bacterial growth in cold environments. An in vitro measurement of the DNA polymerization activity in P. syringae cell extracts using [α-32P]-dATP indicated that the polymerization rate (∼ 2 × 105 CPM min−1 mg−1 protein) at 4°C is about ∼ 2.4-fold lower than at 22°C (∼ 4.8 × 105 CPM min−1 mg−1 protein). This reduced polymerization activity at 4°C is only an indication of the lower DNA polymerase activity at low temperature, which would be further affected in vivo by many cellular factors including template structures.
Our results suggest that DNA degradation activity (ExoV) of RecBCD preempts the RuvAB action at reversed RFs and prevents chromosome linearization, and thus plays an important role in maintaining chromosomal integrity of P. syringae at low temperature. However, this function was not obvious from our earlier study that demonstrated that RecBCD protein with point mutation in the nuclease active site (RecBD1118ACD) or lacking the RecB nuclease domain (RecBΔnucCD) were able to provide cold resistance when expressed in LCBD (Pavankumar et al., 2010). It was proposed that exonuclease activity perhaps provided by RecJ participated in the in vivo repair by homologous recombination for recovery of cells at low temperature. This was supported by the observation that slow growth at 4°C exhibited by recA+ LCBD mutant expressing RecBΔnucCD enhanced to wild-type level when a plasmid borne RecJ nuclease was additionally expressed (Pavankumar et al., 2010). However, in the light of suggestion that RecBCD-dependent degradation of ds-DNA ends of reversed RFs prevent RuvABC mediated chromosomal linearization, the relative importance of RecA mediated recombination pathway at low temperature came into question. In this study we show that, while recA+ LCBD mutant expressing nuclease deficient RecBD1118ACD protein is capable of growing at 4°C, LCBD–ΔrecA expressing RecBD1118ACD failed to grow at the low temperature (Fig. 5). This suggests that a complete lack of RecBCD activity (as in LCBD) or a combination of RecA depletion and nuclease inactivation (as in LCBD–ΔrecA expressing RecBCDD1118A) eliminate both the repair mechanisms (by degradation and by recombination) leading to cold-sensitivity. However, a distinction can be made in respect to the possible roles of RecA at arrested replication forks in LCBD and in E. coli dnaBts mutant. In the latter, RecA was proposed to cause replication fork reversal, as additional recA mutation prevented RuvB-dependent fork breakage in dnaBts recB mutant under non-permissive condition (Seigneur et al., 2000). In the case of LCBD mutant, recA inactivation did not much alter the amount of DSBs, and did not rescue the cold-sensitive phenotype as expected for the fork reversal role. RecA possibly participates in the recombination-dependent pathway of fork re-establishment as observed in LCBD mutant expressing nuclease deficient RecBCDD1118A protein from plasmid borne genes. RecA was also observed to be crucial for the growth of LCBD–ΔruvAB cells at low temperature as LCBD–ΔruvAB ΔrecA triple mutant was unable to grow and died at 4°C (Fig. S3C and D).
Additionally, our results indicate that RecFOR recombination pathway does not protect cells from cold stress-induced DNA damage, as recF mutant of P. syringae was cold-resistant like the wild-type. Also, RadA which has been proposed to play a role in fork reversal during replication arrest in E. coli (Beam et al., 2002; Lovett, 2006) appears to have no role in cold tolerance. Inactivation of radA alone or in combination with recBCD mutation did not cause any effect in terms of growth phenotype.
All together, the critical RecBCD requirement for growth and viability of P. syringae at low temperature lies more in the prevention of DNA damage occurring at arrested RFs than in repairing them by homologous recombination per se. The RRF degradation role of RecBCD eliminates the recruitment of RecA to the arrested RFs and explains why cold stress-induced DNA damage does not induce SOS response in the psychrophilic bacterium. RecA binding to damaged DNA and RecA-dependent cleavage of LexA repressor is an absolute requirement for the SOS induction, which help bacteria to protect chromosomal integrity in case of sudden heavy DNA damage (Friedberg et al., 1995). Our observation that P. syringae exhibits SOS response against UV-induced DNA damage, but not against the cold stress-induced lesions, has evolutionary implication. Bacteria living in permanently cold environment are likely to encounter replication arrest-induced DNA damage regularly, and therefore might have evolved mechanisms to keep RecA protein away from arrested RFs. This would allow these bacteria to avoid chronic induction of SOS response, which can be lethal and energetically costly.
A protective effect of antioxidant systems in cold shock has been documented in several studies. It has been shown that cold shock-induced response of bacteria overlaps with the response to oxidative stress as the two include many common inducible genes and their products (Smirnova et al., 2001). Oxidative stress causes DNA lesions leading to formation of DSBs, and homologous recombination pathway is essential for repairing them (Carlsson and Carpenter, 1980; Keyer et al., 1995; Stohl and Seifert, 2006). The cryosensitivity observed in rep recA mutants of E. coli has also been linked to increased sensitivity to oxidative damage, as the defects could be the result of additive effect of replication arrest and oxidative stress-induced DNA damage (Bredeche et al., 2001). In P. syringae, the high sensitivity to paraquat observed in recBCD mutant also suggests the important role of RecBCD complex in providing resistance to oxidative stress. Surprisingly, recA and ruvAB mutants of P. syringae were almost twofold more resistant, compared with recBCD mutants against paraquat-induced oxidative challenge. The recA mutants of E. coli displayed sensitivity to oxidative stress greater than recBCD single mutants (Carlsson and Carpenter, 1980; Keyer et al., 1995) while the Neisseria gonorrhoeae recA mutants were as sensitive as the recBCD mutants (Stohl and Seifert, 2006); however, in all these cases the organisms were challenged by H2O2 for oxidative stress. On the other hand, the rice pathogen Xanthomonas oryzae pv. oryzae recA mutant and its parental strain exhibited similar level of resistance to both H2O2 and paraquat (Mongkolsuk et al., 1998). However, the important point here is that, although recBCD mutants of P. syringae are highly sensitive to paraquat, the additional ruvAB inactivation fails to alleviate this sensitivity (Fig. 8). Therefore, we conclude that most of the replication arrest and DNA damage caused by cold stress are not directly linked to the oxidative stress-induced lesions.
In summary, we propose that during low-temperature growth of Antarctic P. syringae, replication fork arrests occur frequently, leading to the fork regression at the arrested sites. We further propose that the regressed forks are degraded by RecBCD nuclease activity, which allows restoration of the RFs and continuation of the replication process. Any delay in this restoration of RFs probably allows RuvABC to resolve the HJ structures at RRFs, leading into linearization of chromosomes, loss of chromosomal integrity and cell death. Although RecBCD nuclease activity precludes the RuvAB-dependent activity at RRFs, a certain fraction of DSBs in chromosomes are repaired by RecA–RecBCD-dependent recombination pathway which are detectable in RecA depleted RecBCD nuclease-deficient cells at low temperature. It is also evident that the lesions in arrested RFs produced by UV and cold stress are different in nature requiring different set of proteins for recovery. RecF pathway does not play any role at the arrested RFs or recovery from DNA damage at low temperature. We are currently investigating the roles of other players that modulate RecBCD requirement in P. syringae at low temperature using suppressor analysis of the cold-sensitive LCBD mutant.
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Tables 1 and 2. The psychrophilic P. syringae Lz4W was isolated from a soil sample of Schirmacher Oasis, Antarctica (Shivaji et al., 1989) and routinely grown at 22°C or 4°C (for high and low temperatures respectively) in Antarctic bacterial medium (ABM) composed of 5 g l−1 peptone and 2.0 g l−1 yeast extract, as described earlier (Regha et al., 2005). E. coli strains were cultured at 37°C in Luria–Bertani (LB) medium, which contained 10 g l−1 tryptone, 5 g l−1 yeast extract and 10 g l−1 NaCl. For solid media, 15 g l−1 bacto-agar (Hi-Media) was added to ABM or LB. When necessary, LB medium was supplemented with ampicillin (100 μg ml−1), kanamycin (50 μg ml−1), gentamicin (15 μg ml−1) or tetracycline (20 μg ml−1) for E. coli. For P. syringae, the ABM was supplemented with tetracycline (20 μg ml−1), kanamycin (50 μg ml−1), when required.
Table 1. Bacterial strains
Escherichia coli strain
F− pro recA1 (r− m−) RP4-2 integrated (Tc::Mu) (Km::Tn7) [Smr Tpr]; used as a plasmid-mobilizing strain
General molecular biology techniques including isolation of genomic DNA, polymerase chain reactions (PCR), restriction enzyme digestion and ligation, electroporation, Southern hybridization etc. were performed as described (Sambrook et al., 1989). All restriction enzymes, T4 DNA ligase, T4 polynucleotide kinase, and other enzymes used in this study were from New England Biolabs (NEB, MA, USA). pMOSBlue blunt end cloning kit was from Amersham Biosciences (Uppsala, Sweden). Polymerase chain reactions were carried out using proof reading pfx DNA polymerase from Invitrogen (San Diego, CA, USA). PCR products were purified by Qiagen PCR purification kit (Qiagen, Hilden, Germany). DNA Sequencing reactions were carried out using ABI PRISM Dye terminator cycle sequencing method (Perkin-Elmer, Boston, USA) on an automated DNA Sequencer (ABI model 377). Oligonucleotides were purchased from a commercial source (Bioserve Biotechnology, Hyderabad, India). The conjugal transfer of plasmid into P. syringae was carried out by a biparental mating method using the E. coli strain S17-1, as described (Simon et al., 1983). Transconjugants were selected by their resistance to appropriate antibiotics.
Generation of ΔrecA, ΔradA, ΔruvAB, ΔrecF, ΔrecBCD–ΔrecA, ΔrecBCD–ΔradA, ΔrecBCD–ΔruvAB and ΔrecBCD–ΔrecF mutants of P. syringae Lz4W
Chromosomal target genes were inactivated by gene replacement method using homologues recombination of plasmid-borne disrupted-gene constructs as described earlier (Pavankumar et al., 2010). A common strategy was adopted for gene/s disruption, in which kanamycin resistance gene (kanR) cassette was first inserted into the middle portion of target gene/s cloned on the suicidal plasmid, pJQ200SK (Quandt and Hynes, 1993). Adequate length of homologous DNA sequence (> 500 bp) was provided on either side of the kanR cassette for double-cross-over recombination to occur between the suicidal plasmid constructs and P. syringae chromosome. In this study, we made four suicidal plasmid constructs pJQrecA::kan, pJQradA::kan, PJQruvAB::kan and pJQrecF::kan for the disruption of the recA, radA, ruvAB and recF genes respectively. All four constructs was prepared by a common method. Briefly, sets of suitable forward and reverse primers (Supplementary Table S1) were designed for amplification of each of the above genes by taking conserved nucleotide sequences among Pseudomonas species into consideration, and the genes were amplified from P. syringae Lz4W genomic DNA. Amplified genes were confirmed by nucleotide sequencing, and cloned into pMOSBlue blunt-end vector. Then, kanR cassette (1550 bp) was taken out as a PvuII DNA fragment from pUC4K plasmid (Vieira and Messing, 1982) and cloned into HpaI site, created by site-directed mutagenesis within pMOSBlue borne target genes. Finally, the kanR cassette disrupted target genes were amplified from pMOSBlue plasmid constructs and cloned into SmaI site of the suicidal plasmid pJQ200SK to create the final gene disruption vectors. The descriptions of these constructs are given in Table 2. The disruption vectors were introduced into P. syringae by biparental mating with E. coli S17-1 harbouring the plasmid constructs. Ampicillin-resistant property of P. syringae helped in the counter selection of S17-1 parent which was kanamycin resistance owing to kanR gene on the plasmids. Exconjugants containing the putative gene knockouts were selected on ABM-agar plates supplemented with ampicillin, kanamycin and 5% sucrose. pJQ200SK borne sacB gene which kills cells in the presence of sucrose (Quandt and Hynes, 1993) helped to select gene knockouts that were created by double cross-overs into the chromosomes. For generation of the double mutants, such as ΔrecBCD–recA, ΔrecBCD–radA, ΔrecBCD–ruvAB and ΔrecBCD–recF the suicidal plasmid constructs (pJQrecA::kan, pJQradA::kan, PJQruvAB::kan and pJQrecF::kan) were introduced individually into LCBD (ΔrecBCD) mutant strain (Pavankumar et al., 2010), and the mutants generated by double cross-over were selected as described above. All mutants were screened and selected initially by colony PCR method using gene-specific primers, and subsequently confirmed by Southern hybridization analysis.
For Southern hybridization, genomic DNA digests were separated on 1% agarose gel and transferred onto Hybond N membrane (Amersham Biosciences) by capillary transfer method (Sambrook et al., 1989). 32P-labelled hybridization probes were prepared by labelling the gene-specific DNAs with [α-32P]-dATP using random primer labelling kit (Jonaki, BARC, India). Hybridization was carried out in 0.5 M sodium phosphate buffer containing 7% SDS and 1 mM EDTA for 8–10 h at 65°C. The membrane blots were washed twice (20 min each) at 65°C in the 40 mM sodium phosphate buffer containing 1% SDS and 1 mM EDTA. Radioactive signals were developed using a phosphorimaging device (Fuji FLA3000).
Construction of ΔrecBCD ΔruvAB ΔrecA triple mutant
To create a ΔrecBCD ΔrecA ΔruvAB triple mutant of P. syringae, we employed a partial DNA fragment (65–857 bp) of recA gene in pJQrecA′ suicide plasmid (Table 2) which contains gentamicin resistance (gmR) selectable marker. pJQrecA′ was introduced into LCBD–ΔruvAB double mutant containing tetR kanR markers, and mutants with tetR kanR gmR markers arising by insertion of the plasmid into chromosomal recA by single cross-over were selected on the antibiotics-agar plate. Southern hybridization analysis confirmed the recA gene disruption in LCBD–ΔruvAB mutant (Fig. S3). The lack of RecA protein in this mutant was also confirmed by Western analysis using anti-RecA antibodies (data not shown).
Antibodies and Western analysis
Polyclonal antibodies were raised in rabbit against the His-tagged RecA protein of P. syringae using standard protocol (Sambrook et al., 1989). Production of anti-RecB, anti-RecC and anti-RecD antibodies has been described (Pavankumar et al., 2010). Anti-His antibodies were bought from commercial source (Santa Cruz Biotechnology). For Western analysis, proteins were separated by SDS-PAGE, transferred onto Hybond C membrane (Amersham Biosciences), and probed with appropriate antibodies. The immunoreactive protein bands were detected by alkaline phosphatase-conjugated anti-rabbit goat antibodies (Bangalore Genie, India). For quantification the blots were scanned with a HP scanjet and band intensities were measured using Image J software (rsbweb.nih.gov/ij/).
UV sensitivity test
For qualitative spot assays, cultures were grown at 22°C to ∼ 0.6 OD600. Then, serially diluted cultures (10 μl) were spotted on ABM-agar plates, and plates were exposed to UV light of desired dose, at a rate of 3 J m−2 s−1. Plates were incubated at 22°C in dark for 48 h, and cell growth on the spots were examined and photographed. For quantitative analysis, cultures with appropriate dilutions were spread on plates, and surviving cells were counted as cfu (colony-forming unit) following their growth on plates as described (Regha et al., 2005). Colony numbers on unirradiated plates were taken as 100% for calculation of surviving cells.
Cell viability and cfu analysis of growth
Cell viability in cultures by cfu analysis of cells was performed on ABM-agar plates as described earlier (Regha et al., 2005). Briefly, growing cultures were diluted and spread onto ABM-agar plates for colony formation at 22°C. The relative viability was calculated as the number of cells (cfu) per ml divided by OD600 value of the cultures. Cell viability at low temperature was measured by shifting 22°C grown cultures to 4°C and enumerating the viable cells as above at 24 h intervals.
Pulsed field gel electrophoresis (PFGE)
PFGE was performed as described earlier (Regha et al., 2005). Typically, bacterial cells were harvested from the cultures of exponential phase (OD600 ∼ 0.5), and embedded in agar blocks (1% LGT agarose, FMC-Bioproducts, Rockland, ME). Cells were lysed by incubating the agar blocks in buffer containing lysozyme and RNase for 1 h at 37°C followed by protease K treatment for 4 h at 50°C. Each agar block contained ∼ 0.5 × 107–108 cells. Electrophoresis was performed using CHEF-DRII (Bio-Rad) at a constant voltage (6 V cm−1) and with increasing pulse time of 60–120 s over a period of 21 h at 14°C. PFGE separated DNAs were visualized by staining with ethidium bromide.
For 32P analysis of chromosomal fragmentation DNAs were labelled by the method of Khan and Kuzminov (2012) as described in Supporting information. Phosphorimages were scanned by Typhoon Trio variable mode Imager (GE Health Care, USA) and data were analysed using Image Gauge software V4.0 (Fuji). Signal intensities were measured primarily from three zones: wells containing circular chromosomes, segment of compressed zones representing linear chromosomes below the wells, and the gel segment containing shorter (30–50 kb) DNA fragments. Linear chromosomal DNAs (%) in each sample was determined from the values of compressed zones divided by the total value representing circular (wells) and linear chromosomes, and multiplied by 100.
Paraquat sensitivity assays
Paraquat sensitivity assay was performed as described (Ma et al., 1998). Briefly, bacteria were grown for 22–24 h at 22°C in ABM broth with respective antibiotics. Aliquots (50 μl) of these cultures were added to 5 ml of the same medium containing increasing concentrations of paraquat and incubated shaking at 150 r.p.m. for 20–22 h at 22°C. Surviving cells were then enumerated on ABM-agar plates by spreading of appropriately diluted cultures, as described above. The percentage survivors were calculated by considering the cfu values of untreated cells as 100%.
DNA polymerization activity assay
In vitro DNA polymerization activity was measured by incubating cell extracts in the presence of [α-32P]-dATP at low (4°C) and high (22°C) temperatures. The cell extracts were prepared by the method as described (Villani et al., 1978) from exponentially growing cells of P. syringae in ABM at 22°C. Each assay (50 μl) contained 1 μg of heat denatured template DNA (sheared E. coli genomic DNA), random hexanucleotide primers, 0.8 mM each of dCTP, dGTP, dTTP and 40 μCi of [α-32P]-dATP (Jonaki, BARC, India), and 100 μg of cell lysate proteins (pre-incubated at 22°C or 4°C). Reactions were stopped at different time points (0, 15, 30, 60, 90 and 120 s) by 5% TCA and precipitated samples were collected by centrifugation, and radioactivity was measured in a Packard Tricarb scintillation counter. The radioactivity (count per minute or CPM) incorporation into DNA or polymerization rate was expressed as CPM per minute per milligram of protein.
We thank Dr J. Gowrishankar (CDFD, India), Dr Manjula Reddy (CCMB), and members of M.K.R. laboratory for discussion and suggestions, and acknowledge Dr M. Hynes (University of Calgary) for plasmid pJQ200SK. Research in M.K.R. laboratory is supported by the Council of Scientific and Industrial Research (CSIR), Government of India. A part of the work was supported by a grant to M.K.R. from Department of Science and Technology (DST), Government of India. A.K.S., T.L.P. and P.M. acknowledge CSIR for research fellowships. S.K. was a summer trainee at CCMB.