Biochemical characterization of RecBCD enzyme from an Antarctic Pseudomonas species and identification of its cognate Chi (χ) sequence

Pseudomonas syringae Lz4W RecBCD enzyme, RecBCDPs, is a trimeric protein complex comprised of RecC, RecB, and RecD subunits. RecBCD enzyme is essential for P. syringae growth at low temperature, and it protects cells from low temperature induced replication arrest. In this study, we show that the RecBCDPs enzyme displays distinct biochemical behaviors. Unlike E. coli RecBCD enzyme, the RecD subunit is indispensable for RecBCDPs function. The RecD motor activity is essential for the Chi-like fragments production in P. syringae, highlighting a distinct role for P. syringae RecD subunit in DNA repair and recombination process. Here, we demonstrate that the RecBCDPs enzyme recognizes a unique octameric DNA sequence, 5′-GCTGGCGC-3′ (ChiPs) that attenuates nuclease activity of the enzyme when it enters dsDNA from the 3′-end. We propose that the reduced translocation activities manifested by motor-defective mutants cause cold sensitivity in P. syrinage; emphasizing the importance of DNA processing and recombination functions in rescuing low temperature induced replication fork arrest.


Introduction
The RecBCD enzyme-mediated homologous recombination is a DNA repair pathway that ensures genome integrity by faithful repair of broken DNA in E. coli and many Gram-negative bacteria. The heterotrimeric RecBCD enzyme complex, comprised of RecB, RecC, and RecD subunits, is essential for double-strand breaks (DSBs) repair via homologous recombination and protects host cells from foreign DNAs and invading phages [1][2][3][4]. DSBs are generated in cells by various exogenous and endogenous factors including the running of replication forks into preexisting lesions [5]. RecBCD enzyme is a highly processive helicase and nuclease, a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 The appearance of GroEL as a contaminant during purification of RecBCD protein is also evidenced in E. coli [18]. Nonetheless, the presence of RecB, RecC, and RecD subunits was further confirmed by Western analysis using the polyclonal antibodies specific to these proteins (Figure A in S1 File).

ATP hydrolysis activity of the wild-type and the mutant RecBCD enzymes
ATP hydrolyzing activity of wild-type and mutant RecBCD enzymes was measured by thin layer chromatography (TLC) as described in Materials and methods. The wild-type RecBCD enzyme displayed dsDNA dependent ATPase activity in the presence of linear pBR322 double-stranded DNA (dsDNA). At 22˚C, RecBCD hydrolyzed ATP with the maximum velocity (V max ) of 245.5 μmol ATP per μmol enzyme s -1 and a K m of 57.8 μM for ATP (Fig 2A and Table 1). The mutant RecB D1118A CD enzyme (nuclease defective mutant) showed ATPase activity similar to the wild-type enzyme. In contrast, the ATP-binding defective mutant enzymes such as RecB K28Q CD (mutation in the consensus ATP binding site of RecB) and RecBCD K229Q (mutation in the consensus ATP binding site of RecD) showed a 10-fold decrease in ATP hydrolyzing activities (Table 1).
However, the ATP hydrolyzing activities of RecBCD enzymes measured at three different temperatures (37, 22 and 4˚C) indicated no apparent effect of temperature on ATP hydrolyzing activity on RecBCD and its mutants. (Fig 2C). The slopes obtained from the Arrhenius plots for wild-type and mutants were similar (Fig 2C). The activation energy calculated by respective slopes (E A ) of wild-type RecBCD, RecB D1118A CD, RecB K28Q CD and RecBCD K229Q enzymes were 24.8, 25.0, 26.6 and 26 kJ mol -1 respectively. This suggests that there is no obvious temperature dependent effect on ATP hydrolyzing ability of RecBCD and mutant enzymes.

DNA unwinding and degradation activities of the RecBCD enzyme at different concentration of Mg ++ and ATP
The DNA unwinding and degradation activity of E. coli RecBCD enzyme (RecBCD Ec ) are free [Mg 2+ ] ion dependent. An increase in free [Mg 2+ ] increases nucleolytic cleavage by RecBCD enzyme [19]. To understand the DNA unwinding and degradation behavior of P. syringae RecBCD enzyme (RecBCD Ps ), we performed experiments at various concentrations of magnesium and ATP as described in Materials and methods. In the first set of reactions, the molar concentration of magnesium was kept constant (2 mM) and the ATP concentration was varied (0-10 mM); and in the second set of reactions, the ATP concentration was kept constant (2 mM) and magnesium concentration was varied (0-10 mM). The results indicate that the DNA unwinding and degradation properties of RecBCD Ps enzyme are also modulated by the ratio of Mg 2+ and ATP. When the molar concentration of Mg 2+ is lesser than ATP, RecBCD Ps enzyme unwinds the dsDNA substrate, and the degradation of DNA is not much pronounced ( Fig 3A, lane [4][5][6]. However, when the molar concentration of Mg ++ exceeds the ATP concentration, RecBCD Ps degrades unwound DNA more vigorously (Fig 3B, lane 4-10). The ratio of [Mg 2+ ]: [ATP] thus affects the kinetics of DNA unwinding and degradation properties of RecBCD enzyme. We also observed that an increase in ATP concentration more than three folds over Mg ++ inhibits the DNA unwinding activity of RecBCD Ps enzyme (Fig 3A, lane [7][8][9][10]. The inhibition of DNA unwinding activity is possibly due to sequestration of a Mg 2+ ion by ATP leading to the depletion of free magnesium in the reaction. Subsequently, the DNA unwinding and degradation experiments were performed under specified reaction conditions. The limiting magnesium reaction condition (5 mM ATP, 2 mM Mg ++ ) was chosen to observe the DNA unwinding activity and, the excess magnesium condition (2 mM ATP, 6 mM Mg ++ ) to observe the DNA unwinding and degradation activities of RecBCD enzyme. Interestingly, under excess magnesium conditions, we observed three shorter discrete DNA fragments (Fig 3B, Lane 5-10). These DNA fragments production by RecBCD enzyme is possibly due to Chi-like sequence on a DNA substrate and is discussed later in the results section.
Calcium has been shown to inhibit the nuclease activity of E. coli RecBCD enzyme [20]. We also studied the effects of calcium on P. syringae RecBCD by increasing the Ca ++ ion concentration in a reaction mixture that contained fixed amounts of ATP and Mg 2+ (2 and 6 mM respectively) ( Figure B in S1 File). We found that similar to RecBCD Ec , the high concentration of calcium inhibits both helicase and nuclease activity of RecBCD Ps enzyme.
temperature is plotted and the linear slope of each enzyme is indicated. There is no apparent difference in slope (E A /R, where E A is activation energy and R is the gas constant) is observed despite reduced ATP hydrolysis in ATPase mutants of RecBCD enzyme. The ATP hydrolysis data presented are the results obtained from three independent experiments.  The reaction mixture contained the fixed amount of ATP (2 mM) and concentration of Mg-acetate was varied (0 to 10 mM) as indicated. The reactions were performed for 5 min and analyzed on a 1% agarose gel containing 1X TBE buffer at a 25-30 Volts for 15 hrs. Agarose gels were dried, exposed to phosphor imaging plates and quantified using Phosphor Imager (Fuji-3000). The data were analyzed using image gauge software. The lanes C 1 and C 2 contain Temperature-dependent DNA unwinding and nuclease activity of P. syringae RecBCD enzyme ATPase RecBCD mutants affect P. syringae Lz4W growth in a temperature-dependent manner. Hence, we sought to assess the effects of temperature on the DNA unwinding and degradation properties of RecBCD Ps enzyme. We measured the enzyme activities at 22˚C, and 4˚C, using NdeI linearized [5 0 -32 P] labeled pBR322 plasmid as a substrate. Under the limiting magnesium reaction condition (5 mM ATP, 2 mM Mg 2+ ), the DNA unwinding activity (i.e., production of full-length ssDNA) of RecBCD Ps enzyme was detected only at 22˚C and not at 4˚C (Fig 4A). However, under excess magnesium reaction condition (2 mM ATP, 6 mM Mg ++ ) the DNA degradation activity of RecBCD Ps enzyme was observed at both 22˚C and 4˚C (Fig 4B).
We calculated the rate of DNA unwinding by measuring the decreased band intensities of 5 0 -[γ-32 P]-labeled dsDNA substrates at different temperatures. The wild-type RecBCD Ps unwound the DNA at the rate of 35.1 ± 1.6 bp/sec at 22˚C when ATP and Mg ++ ratio was 5:2 (limiting magnesium condition) ( Table 2). However, the enzyme could unwind the DNA even much faster, when the ATP and Mg ++ ratio was changed to 2:6 (excess magnesium condition). Under the latter condition, wild-type RecBCD Ps enzyme unwound (also degraded) the DNA at the rate of 101.5 ± 3.3 bp/sec (Table 2). Notably, under the limiting magnesium reaction condition at low temperature (4˚C), the RecBCD enzyme failed to produce detectable unwound DNA products ( Fig 4A). However, under excess magnesium conditions, RecBCD enzyme could unwind and degrade the DNA at the rate of 55.8 ± 6.9 bp/s ( Table 2, Fig 4B), which is about 54% of the activity observed at 22˚C. The apparent unwinding rates obtained under excess magnesium conditions, based on the disappearance of 32 P-end-labeled DNA substrate could be an over-estimation. It is possible that some enzyme molecules could just nucleolytically remove the end-labeled nucleotide, but couldn't fully unwind the dsDNA substrate.

DNA unwinding and nuclease activities of mutant RecB K28Q CD, RecBCD K229Q , and RecB D1118A CD enzymes
We have shown that P. syringae cells carrying recB K28Q CD or recBCD K229Q mutants are sensitive to cold temperature, UV irradiation and Mitomycin C (MMC) [14]. These two mutant enzymes also display very weak ATPase activity. To understand the impact of these mutations on the DNA unwinding and degradation properties of RecBCD complex, we analyzed RecB K28Q CD and RecBCD K229Q enzymes in vitro at 22˚C and 4˚C. At 22˚C, under limiting and excess magnesium conditions, RecB K28Q CD with the ATPase-defective RecB subunit displayed weak helicase activity ( Fig 5A). The rate of DNA unwinding was 1.5 ± 0.7 bp/sec and 12.9 ± 3.4 bp/sec at 22˚C, under limiting and excess magnesium conditions respectively ( Fig  5A and Table 2). At 4˚C, on the other hand, RecB K28Q CD showed no detectable DNA unwinding activity under the excess magnesium condition ( Fig 5A and Table 2). These results suggest that RecB K28Q CD is a poor helicase-nuclease enzyme.
Compared to the RecB K28Q CD enzyme, RecBCD K229Q enzyme with the ATPase-defective motor RecD subunit displayed higher DNA unwinding and nuclease activities. At 22˚C, the RecBCD K229Q enzyme with the ATPase-defective RecD subunit displayed substantial DNA unwinding activity, 27.2 ± 6.1 bp/sec and 92.6 ± 14.1 bp/sec, under limiting and excess magnesium conditions respectively ( Fig 5B and Table 2). However, at 4˚C under excess magnesium conditions, the mutant enzyme unwound the DNA at the rate of 17.3 ± 9.6 bp/sec (Fig 5B and  Table 2). The combined DNA unwinding and degradation activity of the RecBCD K229Q enzyme (under excess magnesium) were~90% and~30% of the wild-type RecBCD Ps activity at 22˚C and 4˚C respectively. Interestingly, the mutant RecBCD K229Q enzyme failed to produce discrete DNA fragments ( Fig 5B). We also tested DNA unwinding and degradation by the nuclease defective RecB D1118A CD enzyme. (Fig 5C). The mutation in the nuclease catalytic site of RecB subunit (RecB D1118A CD) led to a complete loss of in vivo nuclease activity in RecBCD Ps complex, without affecting the recombination proficiency and cold adaptation of bacteria [14,15]. At 22˚C, under limiting magnesium reaction condition, RecB D1118A CD enzyme produced single-stranded pBR322 DNA at the rate 30.4 ± 1.7 bp/sec, which is 85% of the rate observed with the wild-type enzyme. Under excess magnesium condition, the rate of DNA unwinding increased to 109.7 ± 17.5 bp/sec, which is similar to wild-type. At 4˚C, this enzyme could unwind the pBR322 DNA at the rate of 41.0 ± 8.1 bp/sec, which is about 75% of wild-type enzyme ( Table 2). More importantly, under both limiting and excess magnesium conditions, this enzyme as expected, produced only the full-length ssDNA of pBR322 and did not degrade the DNA (Fig 5C). From these data, it is clear that RecB D1118A CD enzyme is nuclease deficient in vitro, but its DNA unwinding activity is comparable to wild-type enzyme at both 22 and 4˚C ( Fig 5C and Table 2).

The endonucleolytic activity of P. syringae RecBCD enzyme is not apparent on circular ssDNA substrate
The ATP-dependent endonuclease activity of wild-type RecBCD enzyme of P. syringae was examined using a circular M13 ssDNA as substrate as previously described for RecBCD Ec [21][22][23]. We tested the endonuclease activity under three different conditions of ATP and magnesium concentrations as described in Figure C in S1 File. Under all the three conditions, wildtype RecBCD and mutant proteins failed to degrade M13 phage ssDNA (Figure C in S1 File). It is possible that unlike RecBCD Ec , the RecBCD Ps may have a very weak affinity to circular ssDNA and thus, no endonucleolytic cleavage was detected under the reaction conditions tested.

A specific DNA sequence on pBR322 plasmid modulates nuclease activity of P. syringae RecBCD
We noticed that, under the excess magnesium reaction conditions, RecBCD Ps produced a discrete DNA fragments shorter than the full-length unwound DNA substrate (Fig 3B, Lanes 5-10), and the shortest DNA fragment showed a higher intensity than the other ones ( Fig 3B). This interesting observation led us to hypothesize that the P. syringae RecBCD enzyme recognizes a specific DNA sequence. This specific DNA sequence could potentially be a Chi (χ) like sequence; which alters the nuclease activity of RecBCD Ps complex, allowing 3 0 -ended ssDNA  to escape from DNA degradation as observed earlier [2,8,24,25]. To further confirm that these DNA fragments are indeed single-stranded, we performed degradation assays in the presence of Exonuclease-I (ExoI), which specifically degrades ssDNA by its 3 0 !5 0 exonuclease activity [26]. The RecBCD Ps -generated short DNA fragments disappeared in the presence of ExoI ( Figure D(i) in S1 File) suggesting that they are indeed ssDNA, similar to Chi-specific DNA fragments produced by E. coli RecBCD enzyme after Chi-recognition [8,24,25,27].
Using genetic experiments, we have previously shown that RecBCD Ps enzyme does not recognize E. coli Chi sequence (5 0 -GCTGGTGG-3 0 ) [14,16]. We performed biochemical experiments using pBR322 χ+3F3H (a pBR322 derivative, containing three tandem E. coli χ sequences) [28] and pBR322 (lacking E. coli χ sequence) as a DNA substrate. Interestingly, ssDNA fragments produced by RecBCD Ps enzyme were identical in size with both the substrates, confirming that RecBCD Ps does not recognize E. coli Chi sequence but recognizes an unknown DNA sequence of pBR322 plasmid (Figure D(ii) in S1 File). To locate putative Chi sequence of RecBCD Ps enzyme, we amplified 3.6 kb segment of pBR322 using primer OROPI and OROPII ( Table B in S1 File), which excludes rop region of the plasmid (Figure E(i) in S1 File). Amplified fragments were 5 0 -labeled, and the assays were performed in the presence of excess magnesium. Apparently, all the three shorter ssDNA fragments were also observed with the 3.6 kb DNA substrate, indicating the presence of putative Chi sequence within 3.6 kb of the plasmid (Figure E(iii) in S1 File).
Chi dependent protection of pBR322 DNA fragments are strand specific E. coli RecBCD enzyme recognizes Chi sequence in a specific orientation, 5 0 -GCTGGTGG-3 0 , when enzyme enters through 3 0 -end [29] and the 3 0 -ended ssDNA is being protected from RecBCD nuclease activity after Chi recognition [8,30,31]. Here, we examined which strand of the linearized pBR322 is being protected to produce these discrete DNA bands ( Fig 3B). Hence, we performed DNA degradation assay with RecBCD Ps , in which only one DNA strand was labeled at a time. We individually labeled the OROPI and OROPII primers with γ 32 P at 5 0end and amplified the 3.6 kbp region of pBR322 plasmid with one unlabeled and another 32 Plabeled primer or with both labeled primers. In our assay, we always used sub-saturating concentration of RecBCD Ps enzyme compared to DNA substrate, so that RecBCD Ps enzyme enters through either one or the other end, but not through both the ends. DNA degradation assays using these DNA substrates produced all the three bands when top strand (OROPII end) was alone labeled, or when the both strands were end-labeled ( Fig 6A). We did not observe any ssDNA band when the bottom strand (OROPI end) was labeled (Fig 6A), suggesting that RecBCD Ps recognizes a Chi-like sequence in a specific orientation and has a polarity for Chi sequence recognition. These results also suggest that all the Chi-like sequences are on the top strand.
We then performed DNA degradation assays using two other DNA substrates amplified from the pBR322 plasmid. A PCR amplified 2.8 kbp DNA fragment (using OROPII and pBRS1R primers) and a 2.5 kbp PCR amplified fragment (using OROPII and pBRB1R primers set ( Figure F(i) in S1 File). The DNA band of the lowest size (~2.4 kb) and the highest intensity and degradation properties. But, interestingly, the discrete DNA bands are absent (C) DNA unwinding by nucleasedeficient RecB D1118A CD enzyme under limiting and excess magnesium conditions at 22 and 4˚C. RecB D1118A CD enzyme is unable to degrade DNA at both 22 and 4˚C and notably, the DNA unwinding seems to be faster in excessmagnesium reaction condition (2 mM ATP:6 mM Mg ++ ) compared to limiting magnesium reaction condition (5 mM ATP:2 mM Mg ++ ). Each reaction mixtures contained 0.5 nM of enzyme and 10 μM (nucleotides) linear [5 0 -32 P] labeled pBR322 dsDNA. The lanes C 1 and C 2 contain [5 0 -32 P] labeled NdeI linearized double-stranded and the heat-denatured ssDNA of pBR322 respectively.
https://doi.org/10.1371/journal.pone.0197476.g005  (Figure F(ii) in S1 File). This suggests that one of the Chi-like sequences has the strongest RecBCD-inhibitory activity and this site (we designate it as Chi Ps ) is located proximal to the 2.5 kbp substrate end (as the protected ssDNA was about~2.4 kb). Two upper DNA bands with less intensity could be due to variants of Chi Ps , which might have a weak inhibitory effect on RecBCD Ps nuclease activity (see below).

Identification of Chi Ps as an 8-mer (5 0 -GCTGGCGC-3 0 ) sequence that modulates P. syringae RecBCD nuclease activity and protects DNA from further degradation
To identify the precise location of Chi Ps site in pBR322 DNA substrate, we generated several internally deleted constructs of pBR322 plasmid by site-directed segment-deletion as described in Materials and methods. The 3.6 kbp DNA region of pBR322 plasmid with deleted region/s were PCR amplified using OROPI and OROPII primers. DNA degradation assays were then performed with RecBCD Ps enzyme using the DNA fragments with internal deletion as substrates ( Figure G in S1 File).
The localization of Chi Ps on the plasmid was first based on two hypotheses: the sequence might be GC rich and should be located close to the right end of 2.5 kbp DNA fragment. From the pBR322 nucleotide sequence analysis, it appeared that a GC-rich region spanning the region between 273-450 nucleotides within tet R gene of pBR322 might contain the Chi Ps . Accordingly, 3.6 kbp DNA containing two deletions (Δ273-381 and Δ400-450) were initially tested by the Chi-protection assays.
The 3.6 kbp DNA deleted for 273-381 region produced the Chi-specific high-intensity fragment (Figure G in S1 File), while 400-450 nucleotides deletion failed to produce it (Fig 6B). This indicated that the Chi Ps sequence is located within 400-450 nucleotide region of the pBR322 plasmid. The three additional deletions within the 400-450 region of the plasmid segment were made; Δ401-419, Δ421-439, and Δ441-449 (Fig 6C and 6D). DNA degradation assays with the fragments in which these sequences were deleted revealed that Chi is located within the 421-439 nucleotides segment, as these deletions did not produce a protected prominent DNA fragment (Fig 6C). we further made two more deletion constructs (Δ421-429 and Δ431-439) within the 421-439 nucleotides region and tested for Chi-specific fragment production. Interestingly, deletion of base-pair from 431-439 bp abolished the prominent DNA band (Fig 6D) indicating that the deleted region between 431-439 bp, 5 0 -GCTGGCGCC-3 0 contains the putative Chi sequence of P. syringae. The schematic representation of all the constructs with deleted nucleotide regions and, the presence or absence of protected prominent DNA band (putative Chi sequence) are also shown in Fig 6E. The identification of putative Chi Ps sequence further prompted us to investigate the reason for producing an apparent three distinct DNA fragments by RecBCD Ps enzyme (Fig 3B). We speculated that pBR322 DNA substrate might have sequences that are similar but not identical to Chi Ps sequence and might act like Chi Ps sequence with weaker recognition property under the in vitro assay conditions. The 9-mer (5 0 -GCTGGCGCC-3 0 ) sequence appears only at one Pseudomonas RecBCD enzyme and its Chi recognition place on the pBR322 substrate. Considering the E. coli Chi sequence is an octamer, we first looked in to 5 0 -GCTGGCGC -3 0 (first 8 nucleotides of the 9-mer from the 5 0 -end), and 5 0 -CTGGCGCC-3 0 (eliminating first G of the 5 0 -end) as possible Chi-like sequences. These two octamer sequences appear at a single location on the pBR322 substrate. However, among the 7-mer combinations we looked into, the first 7 nucleotide sequence, 5 0 -GCTGGGC -3 0 , was located in three reasonable positions on pBR322 substrate which could potentially produce discrete DNA bands as observed in DNA degradation assays. Hence, we further focused on 5 0 -GCTGGCGC -3 0 sequence as a putative Chi Ps sequence. We found that the putative Chi Ps sequence (5 0 -GCTGGCGC -3 0 ) is located at 431-438 bp position of pBR322 and, the similar 7-mer sequences at three different locations. The first one at 964-971 position (5 0 -GCTGGCGT-3 0 ); the second one at 1493-1500 position (5 0 -GCTGGCGG-3 0 ) and the third one at 2525-2532 position (5 0 -GCTGGCGT-3 0 ) ( Figure H in S1 File). All the three-bands show 7 bases identical to the 8 bp Chi Ps sequence (5 0 -GCTGGCGC-3 0 ), and occur in the same orientation (5 0 !3 0 ) as the Chi Ps . The RecBCD Ps enzyme might recognize 8 mer sequences (5 0 -GCTGGCGC-3 0 ) as well as the 7+1 mer sequences (5 0 -GCTGGCG+T/G-3 0 ) resulting in multiple DNA bands. Although it is expected to observe four discrete DNA bands, we observed only three DNA bands. The fourth-expected DNA band with a size of~230 bp (when RecBCD Ps enzyme enters from 3'-side of the NdeI linearized pBR322 substrate as depicted in Figure H in S1 File was not apparent in in-vitro experiments performed on agarose gels. Possibly, it is the last Chi-like sequence to recognize by RecBCD enzyme and also being the shortest one with 230 bp (Figure H in S1 File).
To better define the Chi Ps sequence, we mutated the T to C at the 971 st position of pBR322, which creates a sequence identical to the putative 8-mer Chi Ps sequence and performed the Chi-protection assay with RecBCD Ps enzyme. Interestingly, the intensity of the Chi-protected second DNA fragment (after T to C mutation) appeared stronger, and the intensity was similar to the prominent DNA band (Fig 7A), which establish the strong recognition of octamer sequence (5 0 -GCTGGCGC-3 0 ) as Chi Ps .

Cloning of Chi Ps in pBKS plasmid and confirmation of ectopic Chi Ps activity
To further confirm that the octamer sequence (5 0 -GCTGGCGC-3 0 ) is indeed the Chi sequence in P. syringae and can function as an ectopic Chi Ps sequence, we cloned the octanucleotide sequence into pBKS which is devoid of Chi Ps sequence (Table A in S1 File). RecBCD Ps reactions under excess magnesium were performed with XbaI digested linear pBKS (Chi 0 ) plasmid and the Chi Ps containing pBKS (pBKS-Chi Ps ) plasmid. As expected, no Chi-protected DNA bands were observed using the linear pBKS (Chi 0 ) DNA as substrate. In contrast, the expected Chi specific DNA fragment was observed in the case of XbaI digested pBKS-Chi plasmid (Fig 7B). These experiments confirm that RecBCD Ps recognizes an 8-mer sequence (5 0 -GCTGGCGC-3 0 ) as Chi Ps sequence during resection of double-stranded DNA ends, and thus Chi Ps attenuates the nuclease activity of RecBCD enzyme and promotes DNA recombination and repair.

Discussion
We have earlier shown that RecBCD protein complex is essential for cold adaptation in Antarctic P. syringae Lz4W [14][15][16]. In this study, we have analyzed the biochemical properties of the wild-type (RecBCD Ps ) and three mutant enzymes (RecB K28Q CD, RecBCD K229Q , and RecB D1118A CD) to analyze the role of RecBCD during growth at low temperature. We also report for the first time, Pseudomonas octameric Chi-sequence (Chi Ps ), (5 0 -GCTGGCGC-3 0 ), and its ability to attenuate the nuclease activity of P. syringae RecBCD enzyme in vitro.

Role of ATPase activity of RecB and RecD subunits
The analyses of wild-type and mutant enzymes of P. syringae have revealed that mutation in the critical ATP binding sites of RecB (RecB K28Q ) and RecD (RecD K229Q ) motor subunit reduces the ATPase activity of trimeric RecBCD complex by~10 fold. The mutation in the nuclease active-site does not affect the ATPase activity. P. syringae cells carrying the alleles of RecB K28Q or RecD K229Q are sensitive to cold temperature, UV irradiation and MMC [14]. However, the temperature dependent effect on ATP hydrolysis doesn't seem to be an apparent cause for the P. syringae survival at low temperature.

DNA unwinding and degradation activities of RecBCD and mutant enzymes
Our data suggest that the DNA unwinding and degradation properties of RecBCD enzyme depend on ATP and Mg ++ concentrations, and on the temperature in vitro. Under limiting magnesium reaction condition, at 22˚C, the RecB K28Q CD, RecBCD K229Q and RecB D1118A CD enzymes have retained about 4%, 77% and 85% of the wild-type DNA unwinding activity respectively, while at 4˚C, we could not detect the DNA unwinding activity of both wild-type and mutant RecBCD Ps enzymes under the limiting magnesium reaction condition. In contrast, when magnesium was in excess over ATP, the DNA unwinding and degradation activity of the RecBCD enzymes could be measured 22˚C and 4˚C). These results suggest that the excess of magnesium over ATP is favorable for the helicase and nuclease activities of RecBCD Ps enzyme.
The RecB K28Q CD enzyme is a poor helicase and with no detectable nuclease activity in vitro even at 22˚C. However, at 4˚C, RecB K28Q CD enzyme failed to unwind and degrade the DNA under both limiting and excess magnesium reaction conditions. Hence, low-temperature growth sensitivity of the P. syringae recB K28Q CD mutant can be attributed to the lack of RecBCD unwinding and/degradation activities at 4˚C.
The RecBCD K229Q enzyme with defective RecD ATPase unwinds and degrades the dsDNA, albeit at the reduced rate. The DNA unwinding/degradation rate of RecBCD K229Q mutant enzyme is reduced to one-third of the wild-type enzyme at 4˚C. Therefore, we propose that its inability to support the DNA repair process and growth at low temperature is possibly due to its decreased DNA unwinding/degradation activity, particularly at low temperature. Interestingly, the RecBCD K229Q enzyme shows lack of discrete DNA fragments (putative Chi-specific fragments) production. However, the E. coli RecD ATPase defective RecBCD enzyme produces the Chi-specific fragments in vitro and confers DNA repair proficiency in vivo [32]. This converse observation suggests that the RecD subunit of RecBCD Ps enzyme has a distinct role in DNA repair and recombination function in P. syringae. Despite both RecBCD K229Q and RecB K28Q CD mutant enzymes being similarly defective in ATP hydrolyzing activity, their DNA unwinding activity was largely varied. This possibly be due to the selective motor dependency of the enzyme complex, similar to observation made in E. coli RecBCD enzyme [32], in which, the RecB motor is the absolute requirement for Chi recognition and the motor activity of RecBCD complex.
The RecB D1118A CD enzyme is a processive helicase with no detectable nuclease activity in vitro, at both 22˚C and 4˚C. Interestingly, P. syringae cells carrying recB D1118A allele are capable of growing at low temperature, suggesting nuclease activity is not essential for its growth at low temperature [14].

Identification and characterization of novel octameric Chi Ps sequence
One of the novel findings in this study is the identification of P. syringae Lz4W Chi-sequence (Chi Ps ), 5 0 -GCTGGCGC-3 0 . This sequence has not been identified so far in any Pseudomonas species. Most of the identified bacterial Chi (χ) sequences are GC rich sequences (Table 3) and the number of nucleotides in χ sites vary from 5-mer (in Bacillus subtilis) to 8-mer (in E. coli, L. lactis and H. influenza) [33]. A study using cell lysate from Pseudomonads indicated that Pseudomonas species do not recognize E. coli χ sequence [25]. We have confirmed this observation earlier by genetic experiments [14], and in the present study, we have biochemically identified the Chi Ps (5 0 -GCTGGCGC-3 0 ) sequence; which is identical up to 5 bases from the 5 0 -end to the E. coli Chi (Chi Ec ) (5 0 -GCTGGTGG-3 0 ) sequence. 7-mer (5 0 -GCTGGCG-3 0 ) sequence, with change in the last 8 th nucleotide, are partially recognized. The mutation of 7-mer sequence to make a perfect 8-mer Chi Ps sequence enables it to be strongly recognized by RecBCD Ps . Interestingly, appearance of three ssDNA fragments indicates that RecBCD Ps enzyme can sometimes bypass Chi Ps and recognize the next putative Chi-like sequence. Similar observations were made in E. coli, where the probability of Chi recognition by E. coli RecBCD enzyme and nicking the DNA is about 30-40% [29,34].
Interestingly, the difference between E. coli and P. syringae RecBCD enzymes are confined to the last three nucleotides (5 0 -GCTGGTGG -3 0 vs 5 0 -GCTGGCGC -3 0 ). The recent study on the molecular determinants responsible for the Chi recognition by RecBCD enzyme has revealed the importance of RecC channel in Chi recognition. Among the 35 amino acid residues of RecC channel examined, the Q38, T40, Q44, L64 W70 D133, L134, D136, D137, R142, R186 and D705 residues of E. coli RecC subunit have shown to affect the Chi recognition property of RecBCD Ec enzyme [35]. Surprisingly, all these residues are well conserved in the P. syringae RecC subunit. Therefore, the RecC amino acids responsible for recognition of the last three nucleotides of are still elusive. Further analysis of E. coli and P. syringae RecC subunits could shed more insight on the molecular determinants responsible for the recognition of last three nucleotides of Chi sequence in E. coli. E. coli contains 1,008 Chi sequences [36]. They are four-to eightfold more frequent than expected by chance and appear on average once every 4.5 kbp. 75% of Chi sites are skewed towards the replicative leading strand in E. coli [36] keeping with their function in stimulating double-strand break repair upon replication fork collapse. These observations suggest a role for the RecBCD enzyme as a repair factor functioning towards re-establishment of DNA replication fork in case of collapse. However, this skewed nature is not applicable for B. subtilis, S. aureus and H. influenzae, where the skew is statistically insignificant, and the Chi sequence (of the respective species) is distributed all over the genome [33]. The search for Chi Ps sequence (5 0 -GCTGGCGC-3 0 ) in the draft genome sequence of P. syringae Lz4W (4.98 Mb in 42 contigs, Accession no. AOGS01000000) revealed that it contains 1541 Chi Ps sequences (5 0 -GCTGGCGC-3 0 ) and 4564 of 7-mer Chi Ps sequences (5 0 -GCTGGCG-3 0 ). The Chi Ps sequence appears once in every~3 kb and is overrepresented compared to other random octamer sequences. No skewed Chi Ps sequence distribution was observed in P. syringae Lz4W genome (A. Pandiyan and M. K. Ray, unpublished observation). Analysis of the closely related P. fluorescens Pf0-1 genome (Accession no. NC_007492.2) [37] revealed that it contains 2241 Chi Ps sequences and the Chi Ps sequence appears once in every 3 kb and, is almost equally distributed on both strands (1119 Chi-Ps vs. 1122 complementary Chi-Ps sequences). Thus the Staphylococcus aureus Streptococcus agalactiae Streptococcus thermophilus Pseudomonas syringae Lz4W 5 0 GCTGGCGC 3 0 RecBCD [14] https://doi.org/10.1371/journal.pone.0197476.t003 Pseudomonas RecBCD enzyme and its Chi recognition pattern of Chi-distribution is not universal, and although Chi Ps is over-represented in Pseudomonas genome, the orientation bias is not observed.

Biochemical properties of RecBCD enzyme and its role in P. syringae growth at low temperature
This study has revealed the biochemical properties of Pseudomonas RecBCD enzyme. The biochemical properties of RecBCD Ps enzyme, compared to the RecBCD Ec , are particularly associated with the RecD subunit. In E. coli, RecD is dispensable for DNA repair process. The recD null E. coli strain is hyper-recombinogenic [38] and RecBC Ec enzyme (without RecD) unwinds dsDNA and loads RecA constitutively in a Chi-independent manner [31]. Also, the E. coli RecBCD enzyme with a mutation in the ATP binding site of RecD subunit produces Chi-specific fragments and cells expressing the mutant enzyme are UV resistant [32]. In contrast, P. syringae, RecD is essential for the RecBCD's function [14] and ATP hydrolyzing activity of RecD motor is an absolute requirement for Chi Ps fragments production. Also, cells expressing RecD ATPase mutant enzyme are cold sensitive, UV and MMC sensitive (13). Importantly, Chi-like octameric sequence (5 0 -GCTGGCGC-3 0 ) attenuates nuclease activity of RecBCD Ps enzyme producing Chi Ps containing ssDNA fragments and thus, can act as a Chi sequence for RecBCD enzyme of Pseudomonas species. Based on our results we propose a model (Fig 8) which explains the role of RecBCD pathway in rescuing the replication fork arrests in P. syringae Lz4W at low temperature. In this model, RecBCD enzyme can rescue replication fork from the arrest by linear chromosomal DNA degradation in a recA-independent manner. We propose that the motor activity of RecBCD enzyme is essential for rescuing P. syringae cells from low temperature induced replication fork arrest.

Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table A in S1 File. The psychrophilic P. syringae Lz4W was isolated from a soil sample of Schirmacher Oasis, Antarctica [39] and routinely grown at 22 or 4˚C (for high and low temperatures respectively) in Antarctic bacterial medium (ABM) composed of 5 g peptone and 2.0 g yeast extract per liter, as described earlier [16]. E. coli strains were cultured at 37˚C in Luria-Bertani (LB) medium, which contained 10 g tryptone, 5 g yeast extract and 10 g NaCl per liter. For solid media, 15 g bacto-agar (Hi-Media) per liter 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 ) as needed.
pBR322 plasmid DNA (4.3 Kb) and χ+-3F3H dsDNA (a pBR322 derivative, containing two sets of three tandem repeats of χ sequences [28]) were purified using a Qiagen midi kit. Plasmids were linearized with NdeI restriction endonuclease, dephosphorylated with SAP. The dephosphorylated linear dsDNA was 5 0 -labeled with T4-PNK and [γ-32 P] ATP as per the manufacturer guidelines. DNA concentrations were determined by absorbance at 260 nm using molar extinction co-efficient of 6500 M -1 cm -1 (in nucleotides). All restriction enzymes, DNA ligase T4 Polynucleotide Kinase (T4 PNK), Shrimp alkaline phosphatase (SAP) and E. coli SSB were purchased from New England Biolabs (MA, USA). Accuprime Pfx DNA polymerase was purchased from Novagen (WI, USA).

Antibodies and western analysis
Production of anti-RecB, anti-RecC and anti-RecD antibodies has been described [14]. 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/).

Over expression and purification of recombinant proteins
The LCBD (ΔrecBCD) strain harboring pGHCBD, pGHCB K28Q D pGHCBD K229Q pGHCB D1118A D plasmids [14] were initially grown in 10 ml ABM broth containing kanamycin (50 μg/ml) for 24 hrs at 22˚C. 1% of above culture was inoculated into a 2-liter conical flask containing 500 ml ABM broth with kanamycin (50 μg/ml). The culture was then incubated at 22˚C with aeration for 24 hrs. Later, the bacterial cells were harvested by centrifugation at 4˚C, 6000 rpm for 10 min. The bacterial cell pellets were stored at -80˚C. The cells pellet was removed as and when required for the purification of proteins. All the recombinant proteins in this study were expressed with His-tag on N-terminus of RecC and purified by nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography as described in the manufacturer's protocol (Qiagen, New Delhi, India). In brief, cell lysate of overexpressed strains containing His-tagged proteins were prepared by dissolving the cell pellet in 10 ml lysis buffer (50 mM NaH 2 PO 4 (pH 7.4), 300 mM NaCl, 10 mM imidazole and 10% glycerol) and lysed by mild sonication. The sonicated cell lysate was then centrifuged with 14,000 rpm at 4˚C, for 30 min to remove insoluble cellular debris. The supernatant was passed through a pre-equilibrated column containing 1 ml slurry of Ni-NTA agarose beads and column was allowed to bind Ni-NTA agarose beads with His-tagged proteins. Further, column was washed with 4-5 volumes of wash buffer (50 mM NaH 2 PO 4 (pH 7.4), 300 mM NaCl, 20 mM imidazole, and 10% glycerol). Finally, the bound proteins were eluted with 1-2 ml elution buffer (50 mM NaH 2 PO 4 (pH 7.4), / 300 mM NaCl / 300 mM imidazole / 10% glycerol).
Gel filtration technique (size-exclusion column chromatography) was employed for the purification of His-tagged RecBCD complex and the mutant protein complexes. For this, Superose-12 gel filtration column (Pharmacia Fine Chemicals) was used in the fast protein liquid chromatography (FPLC) (Pharmacia Fine Chemicals). Initially, the column was pre-equilibrated with buffer contained 20 mM Tris HCl (pH 7.5), 0.1 mM EDTA, 150 mM NaCl, 0.1 mM PMSF and 10% glycerol. Later, 0.5 ml of Ni-NTA purified protein solution was injected to the column and allowed to pass through the column at a flow rate of 0.4 ml/min. Optical density at 280 nm was recorded for the eluted protein fractions and the fractions were collected in separate microcentifuge tubes. The protein fractions were then analyzed on SDS-PAGE stained with coomassie brilliant blue or silver nitrate. The gel filtration protein fractions of interest were further membrane dialyzed in 50% glycerol containing gel filtration buffer. RecBCD enzyme concentrations were determined by measuring OD 280 and using molar extinction co-efficient 4.7 X 10 5 M -1 cm -1 as determined by ExPASy-ProtParam tool (https:// web.expasy.org/protparam).

Thin-layer chromatography based assay
ATPase activity of RecBCD and mutant proteins was assayed at different temperatures by thin layer chromatography (TLC) on polyethylene-imine (PEI)-cellulose plates (E-Merck, Germany). The assay was performed as described earlier [40,41]. Assays were carried out at indicated temperatures in a reaction volume of 20 μl containing 25 mM Tris acetate (pH 7.5), 1 mM Mg acetate, 1 mM DTT, 100mM NdeI linearized pBR322-dsDNA and 200 μM ATP as a substrate with 2 nM RecBCD and mutant enzymes. One μl of 100 times diluted 10 mCi ml -1 stock of [γ-32 P] ATP (specific activity 3000 Ci mmol -1 ) was used as a tracer in each reaction to measure the rate of ATP hydrolysis. Following 0, 1, 2, 3, 5 10 minutes of reaction, 0.5 μl aliquots of the samples were spotted on TLC plate, air-dried and were allowed to develop in a mobile phase containing 0.5 M formic acid and 0.5 M lithium chloride for 15 minutes. The TLC plate was dried and exposed to the Phosphor imaging plate for 4-6 hrs. The Imaging plates were scanned in a Phosphor Imager, and the amounts of 32 Pi and [γ-32 P] ATP were quantified using Image gauge software (Fuji-3000). Further, data were analyzed using Graph-Pad Prism 4.0 software.

DNA unwinding assay
Plasmid DNAs were linearized with appropriate restriction enzymes in the presence of shrimp alkaline phosphatase and 5'-end-labeled by T4 polynucleotide kinase and [γ-32 P] ATP. Subsequent purification of labeled DNA was accomplished by passage through a MicroSpin S-200 HR column (Amersham biosciences-GE healthcare, Buckinghamshire, UK). The reaction mixtures contained 25 mM Tris acetate (pH 7.5), 2 mM magnesium acetate (as indicated), 1 mM DTT, 10 μM (nucleotides) linear pBR322 dsDNA P 32 -labeled at 5 0 -end, 5 mM ATP, 2 μM E. coli SSB protein and 0.5 nM RecBCD Ps or mutant enzymes. DNA unwinding reactions were started with the addition of either enzyme or ATP, after pre-incubation of all other components at 22 or 4˚C for 5 min. Assays were stopped at the indicated times by addition of proteinase K to a final concentration of 0.5 mg/ml, which was dissolved in sample loading buffer (125 mM EDTA, 40% glycerol, 2.5% SDS, 0.25% bromophenol blue, and 0.25% xylene cyanol). After 5-min incubation with proteinase K at room temperature, the reaction products were run on a 1% (w/v) agarose gel in a 1X TBE (45 mM Tris borate (pH 8.3) and 1 mM EDTA) buffer at 25-30 constant volts for 15 hrs. Agarose gels were dried, exposed to phosphor imaging plates and quantified using Phosphor Imager (Fuji-3000). Further, data were analyzed using image gauge software.
The DNA unwinding rates of were measured by using the following formula nM bp s À 1 ðnM helicaseÞ Where C is the concentration of linear dsDNA substrate in base-pairs in nM (i.e., 5000 nM), t s is the time in seconds, and E is the enzyme concentration in nM.

DNA degradation assay
The assays were performed as described above in DNA unwinding assay, except that the reaction mixtures contained 6 mM magnesium acetate and 2 mM ATP.

Single-stranded DNA endonuclease assay
The endonuclease activity of RecBCD enzyme on ssDNA was examined using a circular M13 ssDNA substrate as described previously [21,23]. In brief, endonuclease activity was tested in 3 different buffer conditions. The first reaction mixture contained 50 mM MOPS (pH = 7.

Site directed deletion of regions from plasmid pBR322
Different internal regions of plasmid pBR322 were deleted using site directed deletion. In short, primers were designed consists of sequences flanking the region to be deleted. PCR was performed to amplify whole Plasmid DNA using Accuprime Pfx DNA polymerase (Invitrogen). After PCR, reaction mixture was subjected for the overnight DpnI digestion. 5 μl of reaction mixture was then transformed into DH5α ultra-competent cells. All selected regions, which had to be deleted was in tetracycline resistance gene of the plasmid. Therefore, for primary screening only those colonies were selected which were Amp R and Tet S . Deletion was further confirmed by PCR and sequencing. All primer list and corresponding deleted regions are shown in Table B in S1 File.