|Home | About | Journals | Submit | Contact Us | Français|
Bacterial nucleoid associated proteins play a variety of roles in genome maintenance and dynamics. Their involvement in genome packaging, DNA replication and transcription are well documented but it is still unclear whether they play any specific roles in genome repair. We discovered that untwisting of the DNA double helix by bacterial non-specific DNA binding proteins stimulates the activity of a repair endonuclease of the Nth/MutY family involved in abasic site removal during base excision repair. The essential Bacillus subtilis primosomal gene dnaD, coding for a protein with DNA-untwisting activity, is in the same operon with nth and the promoter activity of this operon is transiently stimulated by H2O2. Consequently, dnaD mRNA levels persist high upon treatment with H2O2 compared to the reduced mRNA levels of the other essential primosomal genes dnaB and dnaI, suggesting that DnaD may play an important role in DNA repair in addition to its essential role in replication initiation. Homologous Nth repair endonucleases are found in nearly all organisms, including humans. Our data have wider implications for DNA repair as they suggest that genome associated proteins that alter the superhelicity of the DNA indirectly facilitate base excision repair mediated by repair endonucleases of the Nth/MutY family.
Molecular crowding, supercoiling and genome-associated architectural proteins, collectively known as nucleoid associated proteins (NAPs), fold the bacterial chromosome into a compact structure by bridging, bending or wrapping the DNA (1–5). NAPs are implicated in initiation of DNA replication (6), chromosome segregation (7) and transcription (8–13), but their potential roles in DNA repair have not been explored in detail. When DNA repair pathways are hindered, the bacterial nucleoids are organized into morphologies that promote DNA repair and protection (14). The roles of various NAPs in these responses and their links to DNA repair are not fully understood. They may, directly through protein–protein interactions or indirectly by affecting the superhelical properties of the DNA substrates, modulate the functions of repair enzymes. Alternatively, by locally increasing or decreasing local compaction they may affect access to damaged sites by repair enzymes.
In Bacillus subtilis, and other low G+C content gram positive bacteria, the essential primosomal proteins DnaD and DnaB are involved in loading the replicative helicase DnaC at the replication origin, oriC (15–17), and at sites of replication fork collapse across the genome (17,18). The two proteins are structurally related (19,20) and multi-functional, exhibiting significant DNA remodelling activities consistent with NAPs (21–24). In B. subtilis, DnaD comprises two domains DDBH1-DDBH2 (DnaD DnaB Homologies 1 and 2) whereas DnaB has a central redundant DDBH2 domain spanned by a DDBH1 domain at the N-terminus and a DDBH2 domain at the C-terminus (19). The DDBH1 domains mediate DNA-independent oligomerization while the DDBH2 domains mediate DNA-binding and a second DNA-dependent oligomerization activity (19,20,23).
DDBH1 contains a WH (Winged Helix) motif with a helix-strand-helix inserted at its N-terminus and an extra helix at its C-terminus (19,25). DDBH2 is a helical bundle with a highly conserved YxxxIxxxW sequence that contributes to DNA binding together with a solvent exposed somewhat unstable helix and a mobile C-terminal unstructured region (19). Proteolysis of this unstructured C-terminal region in vivo appears to regulate its oligomerization and localization at the oriC (20). B. subtilis DnaD forms large scaffolds that ‘open up’ supercoiled DNA by increasing negative twist (untwisting the double helix) and at the same time eliminating writhe, while DnaB laterally compacts DNA without affecting DNA supercoiling (22,24,26). The DDBH1 and DDBH2 domains must be physically linked for DnaD to exhibit its effective DNA remodeling activity (23). Binding of the DDBH2 to DNA can cause some untwisting of the DNA duplex, albeit not as effective as native DnaD (24,26).
The dnaB gene is juxtaposed with the helicase loader dnaI gene in the same operon, consistent with its proposed essential function as co-loader of the replicative helicase, DnaC (27–29). The dnaD gene is juxtaposed with the nth gene coding for an endonuclease III of the Nth/MutY family of DNA repair glycosylases (30). They belong to a wider HhH (Helix hairpin Helix) superfamily of glycosylases and are highly conserved across bacteria, archaea and eukaryotes (31). Nth is involved in BER (Base Excision Repair). It targets abasic AP (Apurinic Apyrimidinic) lesions and cleaves the C-O-P bond 3′ to the AP site by a β-elimination reaction, leaving a nick with a 3′-terminal unsaturated sugar and a terminal 5′ phosphate (32,33). The nick is then processed by repair enzymes to eventually eliminate the lesion. It is well established that functionally linked bacterial genes are often clustered together in the same operon for coordinated, efficient and rapid responses to environmental and nutritional stimuli. The juxtaposition of the dnaD and nth genes in the same operon in B. subtilis raises the possibility of functional cooperation.
Here, we investigated the effect of DnaD on Nth activity and found that DnaD stimulates the endonuclease activity of Nth on AP-containing supercoiled DNA substrates in vitro. We propose that this stimulatory effect is mediated indirectly via the duplex untwisting activity of DnaD. In support of this, we established that B. subtilis YonN, a homolog of the bacterial NAP HBsu coded for by the SPβ prophage yonN gene, and HBsu also untwist the DNA duplex and similarly stimulate the Nth activity. By comparison, DnaB, which does not alter DNA supercoiling, does not affect the Nth activity. We conclude that untwisting of the DNA duplex enhances the endonuclease activity of Nth at AP sites. We show that deletion of the nth gene results in a strain sensitive to H2O2 exposure, suggesting that Nth plays a prominent role in the oxidative damage response. Furthermore, we show that the dnaD-nth operon is constitutively active and transiently stimulated by H2O2. Furthermore, the dnaD mRNA levels persist at relatively high levels compared to the reduced mRNA levels of the other two primosomal proteins dnaB and dnaI. These data suggest that DnaD may play an important role in DNA repair in addition to its essential role in the initiation of DNA replication. We conclude that NAP proteins with DNA untwisting activities facilitate Nth-mediated BER.
All bacterial strains and plasmids used in this work are listed in Supplementary Figure S1.
The nth gene was cloned by PCR from B. subtilis strain 168 genomic DNA into the NcoI and XhoI sites of pET28a (Novagen) to create the pET28a-nth plasmid. This was transformed into BL21 (DE3) Escherichia coli and 10ml overnight LB (Luria Bertani) cultures were obtained from single colonies. An overnight culture was used to inoculate a 1l of LB containing kanamycin (30µgml−1) and incubated at 12°C with vigorous shaking. At an OD595=0.6, 1mM IPTG was added to induce expression of the Nth protein (Native MW 24853Da) and the culture was left to grow for ~86h. Cells were harvested at 5000g for 15min, and suspended in 40ml of binding buffer (50mM phosphate buffer, pH 8.0, 5mM imidazole and 250mM NaCl), sonicated in the presence of a protease inhibitor cocktail (Sigma p8849), and the suspension was clarified by centrifugation for 30min at 40000g. The supernatant was filtered before loading onto a 5ml HisTrap HP column (GE Healthcare) equilibrated in binding buffer. Protein was eluted over 150ml with a 0–100% gradient of elution buffer (50mM phosphate buffer, pH 8.0, 1mM DTT and 500mM imidazole). Nth-containing fractions were pooled and loaded onto a 5ml HiTrap heparin column (GE Healthcare), equilibrated in TED0 (50mM Tris, pH 7.5, 1mM EDTA, 1mM DTT), keeping the conductivity at <8ms during loading. Protein was eluted over 120ml with a 2–40% gradient of elution buffer TED2M (50mM Tris, pH 7.5, 1mM EDTA, 1mM DTT, 2M NaCl). Nth-containing fractions were pooled and the volume reduced using 10kDa cut-off concentrator spin columns (Amicon) to 5ml which was then loaded onto a HiLoadTM SuperdexTM S75 gel filtration column (GE Healthcare), equilibrated in TED250 (50mM Tris, pH 7.5, 1mM EDTA, 1mM DTT, 250mM NaCl). Nth-containing fractions were pooled, made up to 10% v/v glycerol, the concentration determined spectrophotometrically at 280nm using the extinction coefficient 0.17M−1cm−1 and aliquots were flash frozen in liquid nitrogen for storage in −80°C.
Nth was highly pure with no other bands visible on SDS-PAGE gels even when grossly overloaded samples of 400µM were analysed (Supplementary Figure S2). The E. coli endonuclease III homolog of Nth contains an [4Fe-4S]+ cluster that plays primarily a structural role (34). Predictably, B. subtilis Nth exhibited a deep brown colour characteristic of [4Fe-4S]+ cluster containing proteins. Solvent-exposed [4Fe-4S]+ clusters in proteins are sensitive to oxidation resulting to univalent demetallation to [3Fe-4S]+ which can lead to protein instability and functional inactivation [(35) and references therein)]. To establish whether the B. subtilis Nth [4Fe-4S]+ cluster is prone to air oxidation, Nth was buffer exchanged into TE300 buffer (50mM Tris, pH 7.5, 1mM EDTA, 300mM NaCl) and incubated at room temperature for up to 5h. UV-visible light absorption spectra in the 235–748nm range were obtained from samples at 0, 1.5, 2.5 and 5h (Supplementary Figure S3). The characteristic absorption maximum at 420nm was observed. All spectra were identical with no discernible difference in absorption at 420nm. Furthermore, the colour of the Nth solution did not fade over time. Collectively these data indicate that the [4Fe-4S]+ cluster is not prone to air oxidation. Therefore, under our experimental conditions the Nth protein was considered to be stable.
The yonN gene was cloned by PCR from B. subtilis strain 168 genomic DNA into the NdeI and HindIII sites of pET28b to create the pET28b-yonN plasmid for expression of a C-terminal hexahistidine tagged YonN. This was transformed into BL21 (DE3) E. coli and 10ml overnight LB cultures were obtained from single colonies. An overnight culture was used to inoculate 1 Lt of LB containing kanamycin (30µgml−1) and incubated at 37°C with vigorous shaking. At OD595=0.6 YonN (Native MW 9714Da) expression was induced with 1mM IPTG for 3h. Cells were harvested at 5000g for 15min and suspended in 40ml binding buffer (50mM phosphate buffer, pH 8.0, 1mM DTT, 5mM imidazole and 500mM NaCl). Around 1mM of Phenylmethyl Sulfonyl Fluoride (PMSF) was added prior to sonication. The suspension was clarified by centrifugation at 40000g for 30min and the supernatant loaded onto a 5ml HisTrap HP column equilibrated in binding buffer. YonN was eluted over 150ml from 0–100% of elution buffer (50mM phosphate buffer, pH 8.0, 1mM DTT, 500mM imidazole and 500mM NaCl). The appropriate fractions were pooled and the protein precipitated with ammonium sulphate (29g per 100ml−1). The protein was centrifuged at 40000g for 30min and the pellet suspended in 8ml TED0. The sample was filtered and injected onto a HiLoad™ Superdex™ S75 26/60 gel filtration column equilibrated in TED0. YonN-containing fractions were pooled and made up to 10% v/v glycerol. YonN does not contain tryptophans, tyrosines or cysteines and therefore does not absorb light at 280nm. It does not give a proportionally increasing signal in a Bradford assay and hence its concentration was obtained by comparison of serial dilutions against a known control (Lysozyme) in an SDS-PAGE gel. Aliquots were flash frozen in liquid nitrogen for storage in −80°C.
The hbs gene was amplified from the plasmid pHM9 (gift from Alan Grossman, MIT) using the primers 5′-AGG AGA CCA CAT ATG AAC AAA ACA-3′ and 5′-TGC GGC CGC AAG CTT TTA ATG ATG-3′ and cloned into the NdeI and HindIII sites of pET22b (Novagen) creating the pET22b-HBsu plasmid coding for a C-terminal His-tag HBsu. This was transformed into BL21 (DE3) pLysS E. coli and 10ml overnight LB cultures were obtained from single colonies. An overnight culture was used to inoculate 1l of 2×YT containing chloramphenicol (30µgml−1) and ampicillin (100µgml−1) and incubated at 37°C with vigorous shaking. At OD595=0.6 HBsu (Native MW 9746Da) the temperature was lowered to 30°C and protein expression induced by IPTG (0.2mM) for 4h. Cells were harvested at 5000g for 15min, suspended in 40ml buffer B (50mM Tris–HCl, pH 7.5, 0.5M NaCl) and lyzed by sonication. The suspension was clarified by centrifugation at 40000g for 30min and the supernatant loaded onto a 5ml HisTrap HP column equilibrated in buffer B supplemented with 0.03M imidazole. HBsu was eluted with buffer B supplemented with 0.25M imidazole. The appropriate fractions were pooled and loaded onto a Superdex™ S75 26/60 gel filtration column equilibrated in buffer A (50mM Tris–HCl, pH 7.5, 1mM EDTA, 1mM DTT) supplemented with 100mM NaCl. HBsu containing fractions were pooled and loaded to a 5ml heparin HiTrap HP column (GE Healthcare) equilibrated in buffer A. HBsu was eluted with buffer A supplemented with 0.8M NaCl. Like YonN, HBsu does not contain tryptophans, tyrosines or cysteines and does not give a proportionally increasing signal in a Bradford assay. Its concentration was determined in the same manner as that of YonN.
DnaD (Native MW 27490Da) and DnaB (Native MW 54722Da) were purified and quantified as described previously (21,22). The proteins were stored in −80°C in TED300 (50mM Tris, pH 7.5, 1mM EDTA, 1mM DTT, 300mM NaCl) supplemented with 10% v/v glycerol for DnaD, and TED100 (50mM Tris, pH 7.5, 1mM EDTA, 1mM DTT, 100mM NaCl) supplemented with 10% v/v glycerol for DnaB.
Abasic sites were generated on supercoiled pBSK plasmid as described elsewhere (36). Briefly, pBSK (100nM) was incubated for 8min with AP citrate buffer (0.1mM NaCl, 0.01mM citrate, pH 5.0) at 70°C. Under these conditions on average two AP sites per plasmid molecule are generated (37). Gel solubilisation buffer (Sigma GenElute Gel Extraction kit) was added along with isopropanol and the extraction procedure was followed according to the manufacturer's instructions. The AP-containing pBSK was finally eluted in 10mM Tris–HCl, pH 9.0. The presence of AP sites was verified by fluorescence using the FARP (Fluorescent Aldehyde Reactive Probe), Alexa Fluor® 488 C5-aminooxyacetamide bis(triethylammonium) salt (Alexa Fluor® 488 hydroxylamine) (Invitrogen), as described elsewhere (38). Briefly, the fluorophore was suspended in 1ml DMSO (Dimethyl Sulphoxide) for storage and freshly diluted as appropriate for experimental use. AP-containing and control pBSK of varying concentrations were incubated separately with 100-fold excess of the fluorophore and incubated at room temperature for 4h in the dark under gentle rotation. The plasmid DNA was then isolated by two successive ethanol precipitations, which removed excess unbound fluorophore, and finally suspended in 10mM HEPES and 1mM EDTA, pH 7.4. To verify the creation of AP lesions, equimolar concentrations of AP-containing and control pBSK were run on a Perkin Elmer LS 55 luminescence spectrophotometer with an emission slit width of 5nm, excitation slit width of 2.5nm, excitation wavelength of 488nm and emission detection between 500 and 600nm.
Nth nicking assays were carried out with AP-containing and control supercoiled pBSK (7.14nM) incubated at 37°C with Nth (0.05–5nM) for 1–15min in 20mM Tris–HCl, pH 7.0, 2mM EDTA, 1mM DTT and 250mM NaCl. Proteinase K was added for 30min and pBSK was resolved through 1% w/v agarose gels in TAE (40mM Tris-base, pH 8.0, 1mM EDTA and 20mM acetic acid). Under these conditions Nth-mediated nicking was detected rapidly (<1min), by the conversion of supercoiled to open circular pBSK, and the percentage nicking did not increase significantly after that time point (data not shown). The assay essentially detects only the first nick at an AP site as further nicks at additional sites on the same plasmid molecule will not be detected. The DNA was stained with ethidium bromide and quantification was carried out using a G:BOX gel imaging system with its associated Gene Tools software (SynGene). All experiments were carried out in triplicate.
AP-containing pBSK (7.14nM) was incubated with DnaD (10.0µM), YonN (9.7µM) or DnaB (53µM) for 10min at 37°C, followed by the addition of 0.5–1nM Nth for 1–15min. The reactions were terminated by addition of proteinase K for 30min and pBSK was resolved through 1% w/v agarose gels in TAE. Control reactions were set up with only Nth and the appropriate buffers added instead of the other proteins. Further experiments with control pBSK plasmid (without AP sites) were carried out in a similar manner. All experiments were carried out in triplicate.
Assays to establish the untwisting activities of YonN and HBsu were carried out as described before for DnaD (26).
The B. subtilis Δnth(cat) strain was created by double crossover integration of the chloramphenicol acetyl transferase (cat) gene at the nth locus, and selecting for chloramphenicol resistance. The upstream and downstream flanking regions (~1000bp on either side) of the nth gene were amplified from genomic DNA using the PSnthFO (5′-ATCGCGTCTTAAAGAACATGGACTGGATTCTG-3′) and PSnth/catRE (5′-TACCGCACAGATGCGTAAGGAGGATTGTCACCTTTTCACTTTATTGTTCAAGCC-3′, the bold underlined sequence is complementary to the beginning of the cat gene) for the upstream region, and the PSnthRE (5′-AACAGCCGTCAGTGATGATGGCAAAAGTACAG-3′) and PSnth/catFO (5′-TAATATGAGATAATGCCGACTGTACTTTGACGCAAGCAAAAGAAGTGTTGGCTTCCTTAT-3′, the bold underlined sequence is complementary to the end of the cat gene) for the downstream region. The products from these amplifications were used as primers in a PCR with the pGEMcat plasmid (39) as substrate to splice by overlap the nth upstream and downstream regions at the front and back of the cat gene, respectively. The product of this PCR was used as substrate with the PSnthFO and PSnthRE primers to amplify it further by PCR, and the final product was used to transform competent wild type B. subtilis cells. Selection of positive transformants was carried out on chloramphenicol plates. Single colonies were used to inoculate LB/Chloramphenicol cultures (5ml) which were grown to late logarithmic phase. Cells were harvested and genomic DNA prepared from these cells was used to back cross with wild type B. subtilis to select the final Δnth(cat) strain by chloramphenicol resistance. Deletion of nth was finally confirmed by PCR using the PSnthFO–PSnthRE and OMM (5′-CTCTATTCAGGAATTGTCAGATAGG-3′, annealing near the end of the cat gene)-PSnthRE primer pairs to amplify the entire cat-containing insertion and a fragment containing part of the cat gene and the downstream nth region (data not shown). Control reactions with genomic DNA from the wild type B. subtilis strain 168 were used for comparison (data not shown).
The wt B. subtilis 168 and Δnth(cat) strains were grown from fresh single colonies in LB at 30°C to an OD600 of 0.12 and 0.14, respectively. Samples (1ml) were removed and H2O2 was added (22, 45, 60, 90, 120, 150 and 180mM). Distilled water instead of H2O2 was added to the control samples. All samples were incubated for 30min at 30°C with vigorous shaking, before serial dilutions were spotted on LA plates and left to grow at 30°C overnight.
To generate the lacZ reporter construct for the dnaD-nth promoter (PdnaD), a 600bp DNA fragment immediately upstream from the ATG start codon of dnaD was isolated by PCR using genomic DNA from B. subtilis strain 168 and the primers proFO (5′-AGACACGAATTCGTCAAAACTTGAGCAAATC-3′, the bold underlined sequence is the EcoRI site) and proRE (5′-TTTTCAGGATCCTCGTCACACCTCTTACTGT-3′, the bold underlined sequence is the BamHI site), and then directionally cloned in the EcoRI and BamHI sites in front of the promoterless lacZ-spoVG fusion of the pDG268 plasmid (40) with a cat marker. The entire cassette with the cat-PdnaD-spoVG-lacZ was inserted in the amyE locus by double cross over integration and the sequence was confirmed by sequencing. Positive transformants were isolated by chloramphenicol selection and genomic DNA from these was transformed back to the wt B. subtilis strain 168 to obtain the final PdnaD-lacZ strain by chloramphenicol selection. The correct insertion at the amyE locus was confirmed by PCR (data not shown). As an additional control, the promoterless cat-spoVG-lacZ fragment was also integrated at the amyE locus in exactly the same manner to obtain the final B. subtilis lacZ control strain.
β-Galactosidase reporter assays were carried out as described elsewhere (41). Briefly, a 15ml LB culture was grown at 30°C overnight from a single colony. At an OD600=0.6 H2O2 (0.1mM) was added, with the equivalent volumes of dH2O added to the respective control cultures. Aliquots (0.5ml) were taken at 2, 5, 7, 10, 15 and 20min. The cells were collected by centrifugation at 13000r.p.m. for 10min, suspended in 1ml of 100mM sodium phosphate, pH 7.0, 10mM KCl, 1mM Mg2SO4 and 4mM β-mercaptoethanol, and lysed at 30°C for 2min by adding 50µl of chloroform and 25µl of 0.1 % v/v SDS. The reaction was initiated by the addition of 200µl of ONPG (ortho-nitrophenyl-β-galactosidase) (4mgµl−1) in each sample and vortexing. After 30min at 30°C the reaction was stopped with 0.5ml 1M Na2CO3. Absorbance values at 410, 550 and 650nm were recorded and the β-galactosidase activity was determined using the equation
where 1.775ml is the total reaction volume, T=time of the reaction in minutes, V=volume (ml) of cells added, 0.045µM−1cm−1 is the extinction coefficient of o-nitrophenol at OD420.
Bacillus subtilis strain 168 was cultured in LB. An overnight culture was diluted into fresh LB and incubated at 30°C. At an OD600=0.3–0.4 the culture was divided in two and 80mM H2O2 was added to one of the media with the equivalent volume of sterile dH2O added to the control. After 30min incubation, 4ml samples were collected for RNA extraction, quantification and quality control check as described before (20).
Primers for the RT–PCR were designed with Primer Express 3.0 software (PE Applied Biosystems) for dnaB, dnaD, dnaI genes and the endogenous control rrno-16s ribosomal RNA (42): dnaB-forward (5′-CTGCCCGATTTGGTGGAA-3′), dnaB-reverse (5′-CCGGTTGTTCGCCTTCAC-3′), dnaD-forward (5′-CCGTTATCGCCTTTGGAGTGT-3′), dnaD-reverse (5′-TTGTGCGTCATGCTGATCCT-3′), dnaI-forward (5′-CGGAAGATGAAAACTGCAACAA-3′), dnaI-reverse (5′-AGACCTCCCATTGACAACAAGCT-3′), 16S-forward (5′-AGCGATGTGCGTAGTCAGTCAA-3′) and 16S-forward (5′-TGCGGCGACACCTGTGT-3′). RT–PCR reactions were carried out as described before (20). Relative quantification was used, as described before (20), to determine change in expression of dnaB, dnaD and dnaI genes compared to r16S (ΔCT value), followed by comparison of the change in expression of the three genes incubated with H2O2 compared to the control sample (ΔΔCT value). All results were expressed as the mean of triplicate assays±standard deviation. Statistical analysis was carried out with the StatsDirect 2.5.8 software package, applying a parametric paired t-test to test for the mean difference in ΔCT value for each gene in the presence versus absence of H2O2 in the three experiments (data not shown).
Treatment of plasmid DNA for 8min with a citrate buffer at 70°C has been shown before to create AP sites in a controlled manner (on average two AP sites per plasmid molecule) [(36,37) and references there in]. However, this method does not control the uniform distribution of AP sites across all plasmid molecules. Inevitably, the population of AP-containing plasmid molecules will consist of a mixture of molecules some containing 0, 1, 2 or even more AP sites giving a total average of two AP sites per plasmid. Following this procedure we treated pBSK with citrate buffer for different times over a period of 30min to generate AP sites and, at the same time, to establish the best experimental conditions for preserving the superhelicity of the plasmid and minimize acid-induced nicking (Supplementary Figure S4A). The optimal experimental conditions were established to be identical to those published before [(36,37) and references there in], 8min treatment at 70°C. The presence of AP lesions was verified by fluorescence using a FARP. The AP-containing pBSK exhibited greater fluorescence than the control pBSK, characteristic of FARP binding to AP sites (Supplementary Figure S4B). Using such AP-containing supercoiled pBSK plasmids we were able to assay the activity of Nth. The enzyme acted specifically on the AP-containing pBSK to nick it hence converting the supercoiled pBSK to an open circular form that migrated slower through agarose gels (Figure 1). Under the conditions of our assay, Nth did not exhibit any significant nicking activity on control pBSK without AP sites (Figure 1). Our assay reports on Nth-mediated nicking at an AP site by conversion of supercoiled plasmid to open circular but even though the enzyme is ‘turning over’ additional nicks on the same plasmid molecule are silent and undetected. The presence of a mixed population of plasmid molecules some of which contain no AP sites as explained above is consistent with the failure of Nth to nick 100% of the AP-containing plasmid indicating that a proportion of the plasmid substrate contains no AP sites. The exceptionally high purity and high concentration of purified stock Nth (Supplementary Figure S2) makes it highly unlikely that the observed nicking activity and its stimulation by DnaD are the result of a contaminant protein in our assays. For a contaminant activity to be detected it should be extremely and unusually hyperactive as well as exclusively targeting AP sites and stimulated by DnaD. This is a highly unlikely and improbable scenario. It is, therefore, safe to conclude that the AP-specific nicking observed is the result of Nth activity.
Agarose gel shift assays were used to determine the optimum DnaD concentration at which pBSK (7.14nM) starts to shift, indicative of extensive binding of DnaD [Supplementary Figure S5A and (22,23,26)]. We have shown before that multiple DnaD molecules bind to supercoiled plasmid forming a large scaffold that shifts higher up in agarose gels during electrophoresis (20–25). Treatment of the pBSK-DnaD binding mixtures with proteinase K prior to gel electrophoresis digested the DnaD, and the pBSK plasmid reverted to its original supercoiled state (evident from its migration position through the agarose gel during electrophoresis) confirming previous reports that DnaD does not nick supercoiled DNA (Supplementary Figure S5A and (22,23,26)]. The optimum concentration under our experimental conditions was established to be 10µM of DnaD which is compatible with previously published data (22,23,26). This concentration was used to investigate whether DnaD affects the activity of Nth. Using nicking assays we found that binding of DnaD to pBSK stimulates the nicking activity of Nth on AP-containing supercoiled pBSK (Figure 2).
In order to establish whether this stimulatory effect is mediated by a direct protein-protein interaction analytical gel filtration, Isothermal Titration Calorimetry (ITC), Surface Plasmon Resonance (SPR) and glutaraldehyde fixing experiments were carried out with purified DnaD and Nth proteins. Despite our extensive efforts no direct DnaD-Nth interaction was detected with any of these experimental techniques (data not shown). In the absence of evidence for an Nth-DnaD interaction, an alternative explanation for the stimulation of the Nth activity was considered. We hypothesized that DnaD, via its DNA remodelling activity, alters the DNA substrate rendering it more susceptible to the Nth activity. DnaD has been shown to form large scaffolds with DNA and to eliminate writhe of supercoiled plasmids by increasing the negative twist (22,24,26). We further hypothesized that if Nth stimulation is mediated by the untwisting of the DNA duplex, other proteins that also untwist the DNA duplex could have similar effects on the activity of Nth.
YonN is a small protein located in the SPβ prophage of the B. subtilis genome (43). It is homologous to HBsu, both proteins being of equal length (92 amino acid residues) and sharing 68.5% identity (Supplementary Figure S6). HBsu is a homodimeric homolog of the E. coli heterodimeric HUαβ. HU proteins from E. coli, Thermotoga maritima, Borrelia burgdorferi and Anabaena have been widely reported to bind to supercoiled DNA and increase negative supercoling (44–47). Consistent with these studies, we established that YonN and HBsu also increase negative supercoiling. Incubation of pBSK with increasing concentrations of YonN or HBsu stimulated the activity of topoisomerase I (Figure 3A). As more negative supercoiling was introduced, by increasing concentrations of YonN or HBsu, topoisomerase I progressively relaxed the additional negative supercoils by introducing positive linking number changes. Subsequent removal of all the proteins by proteinase K treatment left behind supercoiled pBSK with progressively higher positive linking number changes resulting in the appearance of more relaxed plasmid higher up the gel (Figure 3A). Neither YonN nor HBsu possessed any nicking activity since incubation with high concentrations of these proteins with pBSK and their subsequent removal by proteinase K treatment resulted in fully supercoiled plasmid without any detectable nicking [Supplementary. Figure S5 and (22,23,26)]. From these data we conclude that both YonN and HBsu bind to supercoiled DNA and untwist the DNA double helix.
We then investigated whether YonN or HBsu can affect the nicking activity of Nth. Agarose gel shift assays were first used to determine the optimum YonN and HBsu concentrations at which pBSK (7.14nM) starts to shift, indicative of extensive protein binding (Supplementary Figure S5B and C). Treatment with proteinase K digested the YonN or HBsu in the reaction mixtures, and the pBSK plasmid reverted to its original supercoiled state, confirming that YonN and HBsu do not nick supercoiled DNA (Supplementary Figure S5B and C). The optimum concentrations under our experimental conditions were established to be 9.72 and 10µM for YonN and HBsu, respectively. These concentrations were used to investigate whether YonN and HBsu affect the activity of Nth in assays similar to those carried out for DnaD. We found that binding of YonN or HBsu to pBSK stimulated the nicking activity of Nth on AP-containing supercoiled pBSK (Figure 3B). By comparison, the primosomal protein DnaB, which is involved with DnaD in the initiation of DNA replication in B. subtilis and shown previously to form large complexes with supercoiled DNA without altering DNA superhelicity (22,26), formed large complexes with supercoiled pBSK [Supplementary Figure S7C and (22)] and did not stimulate the Nth nicking activity in our assay (Figure 3C). From the combined data we conclude that non-specific DNA binding proteins such as DnaD, YonN and HBsu that increase negative supercoiling stimulate the nicking activity of Nth, whereas non-specific DNA binding proteins such as DnaB that do not affect DNA supercoiling do not affect the activity of Nth.
In order to verify the putative role of Nth in oxidative damage response, the nth gene was deleted by a double cross over insertion of a fragment carrying the cat gene flanked by 1-bp sequences from the upstream and downstream regions of nth. The deletion was verified by PCR (data not shown). In comparative experiments the resultant Δnth B. subtilis strain grew marginally slower at 37°C compared to the isogenic wt B. subtilis strain 168 but at 30°C the two strains grew at similar rates (data not shown). Treatment with H2O2 (22–180mM) revealed that Δnth was markedly more sensitive than the wt strain. Significant bacterial death was observed at >45mM H2O2 (Figure 4). In comparison, the wt strain exhibited good tolerance to H2O2 and significant death was only apparent at concentrations >120mM. These data confirm that Nth is an important contributor of the oxidative damage DNA repair in vivo.
Since the nth gene was shown to be important in oxidative damage repair, an important question to answer is whether the dnaD-nth operon is responsive to H2O2. In the absence of precise information about the absolute location of the promoter region (PdnaD) of the dnaD-nth operon, we isolated by PCR a 600bp fragment immediately upstream from the start codon of dnaD which we assumed included the entire promoter region. This fragment was cloned in the EcoRI–BamHI sites of the pDG268 plasmid immediately up-stream of (and co-directionally with) the promoterless spoVG-lacZ fusion gene. An entire cassette containing the 600bp-promoter, spoVG-lacZ and cat was inserted at the amyE locus by a double cross-over event. The resultant PdnaD-lacZ catR strain was assayed for β-galactosidase activity after H2O2 treatment (0.1mM) and the data compared to control untreated cells, as well as to the isogenic wt strain and to a control strain containing the promoterless spoVG-lacZ fragment at amyE (Figure 5A). High β-galactosidase activity was detected in the PdnaD-lacZ catR strain compared to no detectable activity in the wt and promoterless lacZ control strains, indicating that the PdnaD promoter was active (compare 0mM and wt, promoterless strains in Figure 5A). Exposure to 0.1mM H2O2 for 2, 5 and 7min resulted in ~30–40% stimulation of the promoter activity compared to the control (compare the first three bars in 0.1 and 0mM in Figure 5A), whereas at 10, 15 and 20min the promoter activity returned to that of the control. From these data we conclude that the PdnaD promoter is constitutively active and transiently stimulated by H2O2.
This was consistent with comparative RT–PCR experiments showing that the primosomal dnaD mRNA levels were consistently high in the presence or absence of H2O2 in the growth medium, compared to markedly reduced levels of the dnaB and dnaI mRNA levels after H2O2 exposure (Figure 5B). Treatment of cells at mid-logarithmic growth with 80mM H2O2 for 30min resulted in considerable reduction of the dnaB, dnaI mRNA levels (reduced to ~30% of the control levels) but only marginal (statistically insignificant) decrease of the dnaD mRNA level. These data are consistent with a constitutively active PdnaD promoter whose activity is maintained at high levels during H2O2 exposure. Although the mRNA levels of dnaB, dnaI were markedly reduced in the presence of H2O2, it is not clear whether this is the result of decreased promoter activity of the dnaB-dnaI operon, a consequence of increased mRNA degradation as part of the general replication stress induced by H2O2, or indeed a combination of the above. Further experiments, beyond the scope of this work, will be required to distinguish between these possibilities.
Bacillus subtilis is a soil bacterium exposed to a variety of oxidizing agents from external (environmental) and internal (metabolic) sources. The PerR and OhrR proteins coordinate the oxidative stress response while the general stress response factor σB and the RNA polymerase-interacting thiol-based sensor Spx also contribute to the protective response via the general stress responses [(35) and references therein]. During the general stress response the expression of the stationary phase NAP Dps, controlled by σB, likely compacts the nucleoid to protect it against damaging agents (14), whilst during vegetative growth the PerR-controlled expression of the Dps-like protein MrgA is part of the protective oxidative stress response (35). Despite the protective roles of such NAPs lesions still arise and need to be repaired by DNA glycosylases. The substrate specificities and excision kinetics of DNA glycosylases differ significantly with the nature of the DNA substrate and with the accessibility of the targeted DNA lesions in vivo (14,30). With their abilities to alter the structure of the bacterial nucleoid, NAPs have the potential to participate and/or modulate DNA repair.
Here we show that the B. subtilis DnaD, an essential primosomal protein with NAP-like properties, HBsu (a de facto NAP) and YonN (a homolog of HBsu) stimulate the nicking activity of Nth in supercoiled AP-containing plasmid DNA in vitro. We have not been able to detect direct protein-protein interactions between the Nth and DnaD or YonN proteins despite using a range of biophysical techniques including ITC, SPR, glutaraldehyde fixing and analytical gel filtration with purified proteins. Although we cannot unequivocally eliminate weak, transient DNA-mediated physical interactions (negative data do not constitute definitive proof) or the presence of an adaptor protein that could bridge DnaD, HBsu or YonN and Nth in vivo, it is evident that the functional stimulation is a consequence of changes in the superhelicity of the DNA double helix mediated by the interaction of these non-specific DNA binding proteins with the DNA substrate. DnaD was previously shown to untwist supercoiled DNA by increasing the negative twist and at the same time eliminating writhe (22,26). We show here that HBsu and its YonN homolog also untwist supercoiled DNA. All three proteins were found to stimulate the Nth nicking activity. The set up of the in vitro nicking assay reported here gives an underestimate of the overall stimulatory effect but even relatively modest (2–3-fold) stimulation in vivo could have a major impact on BER on a genome-wide scale. By comparison DnaB, an essential primosomal protein with NAP-like activity but unable to untwist supercoiled DNA, did not stimulate the Nth protein. We propose that it is the DNA helix untwisting by DnaD, YonN or HBsu that indirectly stimulates the nicking activity of Nth.
The molecular mechanism of this stimulatory effect likely relates to the recognition of the estranged base by Nth. The crystal structure of the Bacillus stearothermophilus EndoIII (78% identical and 89% similar to the B. subtilis Nth) bound to a lesion-containing DNA revealed significant bending, distortion and melting of the DNA double helix at the lesion site (48). Many glycosylases exhibit some lesion specificity determined mainly by specific interactions with the lesion bases but also partly by differences in the local DNA strain and base-stacking energies. Unwinding of the DNA double helix by non-specific NAPs lowers such energy barriers to enhance binding at the site of lesion and could potentially provide a global DNA repair regulation mechanism. NAPs coordinate the regulation of superhelicity at global and local levels which affects transcription (13) and as shown here they have the potential to regulate DNA repair too.
The E. coli HU, a homolog of HBsu and YonN, was shown to inhibit the rate of removal of dihydrouracil by endonuclease III when another nick was present in the opposite strand in close proximity but did not inhibit subsequent DNA synthesis and ligation during BER (49). This was suggested to be a protective mechanism to reduce the formation of double strand breaks during the repair of closely opposed lesions. HU exhibits high affinity for DNA nicks (50,51) and binds directly to them preventing access of repair nucleases and enhancing repair by DNA polymerase I and ligase. This is fundamentally different than the stimulation of the Nth nicking activity during AP repair observed here as there are no pre-formed nicks at AP sites for HU to bind and direct the Nth.
The genetic co-localization of the dnaD and nth genes in the same operon observed in B. subtilis is widely conserved in many gram positive Firmicutes that contain dnaD-like genes (Supplementary Figure S7). The list contains a range of Bacillus, Lactobacillus, Staphylococcus, Acholeplasma, Enterococcus, Streptococcus species and also Clostridium perfringens and Clostridium botulinum. Usually bacterial genes in the same operon co-operate in the same cellular functions. This is likely the case with the dnaD-nth operon too. A global transcriptome microarray analysis of Staphylococcus aureus in response to H2O2 reported transient up-regulation of the dnaD expression upon H2O2 exposure (52). Exposure to 10mM H2O2 resulted in about 3× to 4× higher levels of dnaD mRNA after 10min of exposure but expression returned to control levels after 20min. This transient up-regulation of dnaD-nth operon in response to H2O2 exposure suggests that it is an integral part of the cellular response of S. aureus to H2O2-induced oxidative stress.
We found that the promoter of this operon in B. subtilis is constitutively active during exponential growth. This is consistent with the constant mRNA levels of dnaD throughout asynchronous growth detected before (20). Exposure to 0.1mM H2O2 for 2, 5 and 7min resulted in transient up-regulation of the promoter activity which returned to the control levels at 10, 15 and 20min. This is somewhat analogous to the transient H2O2-mediated stimulation of dnaD expression observed in S. aureus by transcriptome microarray analysis (52). RT–PCR experiments revealed comparable levels of dnaD mRNA in H2O2 exposed and control cells. By comparison, the mRNA levels of the other two essential primosomal genes dnaB and dnaI, co-localized in a different distant operon, were significantly reduced upon H2O2 exposure. Given that the mRNA levels of dnaB and dnaI were reported to be comparable to that of dnaD throughout asynchronous growth (20), this down regulation may be a consequence of reduced activity of the dnaB–dnaI promoter or increased mRNA degradation or part of the replication stress response to H2O2 exposure or a combination of these. Down regulation of replication initiation under stress conditions during H2O2 exposure, implied by the down regulation of primosomal genes, may be beneficial in order to avoid catastrophic replication fork collapse and double strand breaks at H2O2-induced lesions across the replicating genome. It may also allow additional time for the DNA repair mechanisms to spring into action and repair lesions before the chromosome is fit again to be replicated. Indeed AP sites form preferentially at regions undergoing DNA replication and exposure to H2O2 increased further their occurrence in newly replicated DNA (53).
The persistence of relatively high levels of the dnaD mRNA upon H2O2 exposure in comparison to dnaB and dnaI suggests that DnaD plays an additional role in a cellular function other than its cooperative role, with DnaB and DnaI, in the initiation of DNA replication. Normally there are only a handful of DNA replication initiation protein molecules per cell but DnaD is rather atypical as relatively large intracellular concentrations (3000–5000 molecules per cell) have been reported (54). This is suggestive that DnaD participates in other intracellular functions in addition to replication initiation. Our data show that an additional DnaD function is the stimulation of Nth during the cellular response to H2O2-induced oxidative damage. Its role is likely via non-specific DNA binding in the vicinity of the lesions to introduce negative twist that assists Nth binding to the lesions during BER. At the reported intracellular concentrations it is unlikely that DnaD would be able to coat the whole genome to facilitate BER. It could, however, operate at the level of distinct topological domains as suggested before (22). DnaD binding to a topological domain containing AP sites may alter its local topology by converting the plectonemic DNA into a paranemic form to facilitate BER. How DnaD targets AP sites is unknown. It may be that an unidentified adaptor protein, as suggested above, bridges Nth and DnaD and acts as a recruiter of DnaD to AP- sites. An alternative possibility is that structural deformations of the DNA double helix in the vicinity of an AP site may act as DnaD nucleation sites. Further work is required to determine the targeting mechanism.
DNA supercoiling has been proposed to act at the apex of a regulatory hierarchy that responds to environmental stress. Responsiveness of the superhelical linking number to environmental stress in collaboration with NAPs offers an attractive mechanism for the global adjustment of the gene expression profile of the cell [(55–57) and references therein]. In addition to regulation of transcription, our data suggest that this mechanism can also accommodate direct tuning of the activity of certain DNA repair enzymes such as Nth involved in BER. The genetic linkage (both genes are juxtaposed in the same operon) and functional cooperation of DnaD with Nth offers an additional H2O2-mediated regulatory level, above and beyond that exerted by the global superhelical density, exclusively modulating oxidative damage induced BER. The common PdnaD promoter is transiently stimulated by H2O2 resulting in upregulation of DnaD and Nth that then target specifically AP sites formed by H2O2-induce oxidative damage.
Nth is conserved across all species including eukarea. Loss of the eukaryotic Caenorhabditis elegans nth-1, a homolog of the bacterial nth and the only BER enzyme known to initiate oxidative DNA damage repair in this organism, leads to oxidative stress and additional global gene expression changes that lead to upregulation of endogenous stress genes and downregulation of insulin-like signalling (58). The human homologs NTH1 and NTHL1 play instrumental roles in removing potentially mutagenic oxidation products of 5-methylcytosine (59). A double knock-out strain of mice (Nth1−/−Neil−/−) lacking both NTH1 and NEIl1, the human homolog of the prokaryotic Nei endonuclease VIII (60,61), exhibited a high incidence of pulmonary and hepatic tumours in comparison to the single knock-out strains Nth1−/− or Neil−/− (62). Although there is no eukaryotic homolog of DnaD it will be interesting to establish whether other DNA binding proteins that alter the superhelicity of the DNA may play analogous roles in human BER by stimulating the activities of human NTH1, NTHL1 and NEIL1 proteins.
Supplementary Data are available at NAR Online: Supplementary Figures S1–S7. Supplementary Reference .
Wellcome Trust grant (091968/Z/10/Z to P.S.). C.C. was supported by a PhD studentship from the School of Chemistry, University of Nottingham. W.K.S. is supported by Veni grant 016.116.043 from the Netherlands Organisation for Scientific Research and a Gisela Thier Fellowship from the Lieden University Medical Center. Funding for open access charge: The Wellcome Trust.
Conflict of interest statement. None declared.
The authors thank Alan Grossman, Catherine Lee and Houra Merrikh for useful discussions, advice and help with the genetics experiments.