Deinococcus radiodurans PriA is a Pseudohelicase

Reactivation of repaired DNA replication forks in bacteria is catalyzed by PriA helicase. This broadly-conserved bacterial enzyme can remodel the structure of DNA at a repaired DNA replication fork by unwinding small portions of duplex DNA to prepare the fork for replisome reloading. While PriA’s helicase activity is not strictly required for cell viability in E. coli, the sequence motifs that confer helicase activity upon PriA are well-conserved among sequenced bacterial priA genes, suggesting that PriA’s duplex DNA unwinding activity confers a selective advantage upon cells. However, these helicase sequence motifs are not well-conserved among priA genes from the Deinococcus-Thermus phylum. Here, we show that PriA from a highly radiation-resistant member of that phylum, Deinococcus radiodurans, lacks the ability to hydrolyze ATP and unwind duplex DNA, thus qualifying D. radiodurans PriA as a pseudohelicase. Despite the lack of helicase activity, D. radiodurans PriA has retained the DNA binding activity expected of a typical PriA helicase, and we present evidence for a physical interaction between D. radiodurans PriA and its cognate replicative helicase, DnaB. This suggests that PriA has retained a role in replisome reloading onto repaired DNA replication forks in D. radiodurans despite its lack of helicase activity.


Introduction
PriA helicase catalyzes reloading of the DNA replication machinery (the replisome) at repaired DNA replication forks in bacteria, giving bacterial cells the ability to restart DNA replication following disruptive encounters of the replisome with DNA damage [1]. This pathway, known as DNA replication restart, is broadly conserved among diverse bacteria and likely represents an indispensable component of bacterial genome maintenance. Indeed, some bacteria have evolved multiple DNA replication restart pathways that are catalyzed by different sets of primosome proteins and that operate on different types of DNA structures [2,3]. In E. coli, a PriA-dependent DNA replication restart pathway catalyzes replisome reloading onto forked DNA structures that lack a gap at the three-way fork junction, while a PriC-dependent DNA replication restart pathway facilitates replisome reloading onto DNA structures that contain a leading strand gap at the fork junction [3]. The presence of multiple DNA replication restart pathways likely confers upon bacterial cells a greater ability to respond to the varied assortment of DNA structures that could be encountered at a stalled or collapsed DNA replication fork [1].
Given the multitude of ways in which DNA can be damaged and the myriad structures that can result from failure of a replisome to fully replicate a chromosome, it seems inevitable that some DNA structures that are created from the premature dissociation of a replisome would require remodeling before a functional replisome could be reloaded [4,5]. The main DNA replication restart initiator protein, PriA, contains a series of well-conserved helicase motifs that confer upon it the ability to remodel the structure of DNA at a repaired DNA replication fork by unwinding short stretches of duplex DNA at the three-way fork junction [6][7][8]. This helicase activity requires the coupling of ATP binding and hydrolysis to the translocation of the enzyme along the nucleic acid lattice. Translocation occurs in the 3'-5' direction, which would result in the displacement of a portion of the nascent lagging strand were PriA to unwind duplex DNA at a repaired DNA replication fork [6]. This DNA unwinding activity is thought to create a binding site for reloading the replicative helicase, DnaB, back onto the DNA as part of the fork reactivation process. In addition to unwinding duplex DNA, PriA catalyzes the assembly of additional primosome proteins onto a repaired DNA replication fork. In the PriAdependent DNA replication restart pathway in E. coli, these additional primosome proteins include PriB and DnaT [3,[9][10][11][12]. The combined activities of PriA, PriB, and DnaT result in reassembly of a functional replisome at a repaired DNA replication fork which allows DNA replication to resume.
In E. coli, mutations in priA have been generated that abolish PriA's ATPase activity and render it incapable of unwinding duplex DNA, but do not affect its primosome assembly function on model DNA substrates [13]. Interestingly, E. coli strains carrying these mutant priA alleles remain viable and are capable of undergoing recombination, indicating that PriA's ATPase and duplex DNA unwinding activities are not strictly required for cell survival or assembly of a functional primosome [13][14][15]. However, despite the fact that PriA's helicase activity is dispensable under laboratory growth conditions, it seems likely that its helicase activity would play an important genome maintenance role in nature by expanding the cell's ability to reactivate repaired DNA replication forks on a wider subset of DNA structures, such as those that contain duplex DNA at the three-way junction of a repaired DNA replication fork. This hypothesis is bolstered by the observation that PriA's helicase motifs are well-conserved among sequenced bacterial priA orthologs (Fig 1).
Members of the Deinococcus-Thermus phylum are notable exceptions to the evolutionary tendency to conserve PriA's helicase motifs. These bacteria encode a PriA helicase that lacks amino acid residues that are otherwise well-conserved among known bacterial PriAs and are critical for catalyzing ATP hydrolysis and facilitating duplex DNA unwinding. In particular, the Deinococcus-Thermus phylum members have poor matches to the consensus sequences of helicase motifs I and II, also known as the Walker A and Walker B boxes, respectively (Fig 2). In E. coli PriA and other well-studied helicases, the Walker A box (also known as a phosphatebinding loop, or P-loop) is involved in ATP binding, while the Walker B box is involved in binding an active site Mg 2+ . Both motifs are therefore critical for helicase-catalyzed duplex DNA unwinding [16][17][18]. Within the Walker A box of Deinococcus-Thermus PriAs, the catalytic lysine residue in the GKT sequence of the motif has been substituted for an arginine residue (Fig 2). This amino acid substitution has been constructed and studied in E. coli PriA (the K230R PriA variant), and it has been shown to be sufficient to abolish E. coli PriA's ATPase and helicase activities [13]. Taken together, these observations led us to hypothesize that Deinococcus-Thermus PriA enzymes would be deficient at hydrolyzing ATP and unwinding duplex DNA, effectively qualifying them as pseudohelicases-enzymes that have clearly evolved from an ancestral helicase but have lost their catalytic activity.
To test this hypothesis, we cloned the priA gene from Deinococcus radiodurans R1, expressed and purified recombinant D. radiodurans PriA protein, and tested it for its ability to catalyze ATP hydrolysis and duplex DNA unwinding. Our findings reveal that D. radiodurans  Evolutionary conservation of PriA helicase motifs. Sequence logos were generated for helicase motifs I and II (the Walker A and Walker B boxes, respectively) from a multiple sequence alignment of 100 PriA helicases using the Weblogo 3 sequence logo generator [27,28]. Each PriA sequence represents a unique genus, and 18 phyla are represented. A multiple sequence alignment of 17 members of the Deinococcus-Thermus phylum appears below each sequence logo. These sequences show significant departure from the otherwise well-conserved helicase motifs. PriA is indeed a pseudohelicase that is unable to hydrolyze ATP or facilitate duplex DNA unwinding. Yet, D. radiodurans PriA is capable of binding to single-stranded DNA (ssDNA) and fork DNA, and we show evidence for a physical interaction between D. radiodurans PriA and its cognate replicative helicase, DnaB, using native agarose gel electrophoresis. Collectively, these data suggest that D. radiodurans PriA has retained its function in DNA replication restart in Deinococcus-Thermus bacteria despite the absence of helicase activity.

Materials and Methods
Molecular cloning of priA and dnaB genes The priA gene of D. radiodurans was amplified from strain R1 genomic DNA by polymerase chain reaction (PCR) using primers 5'-GCG TAT TCC ATA TGA CTC TCT GGA CCG TTT C and 5'-GTC ACG GAT CCT CAG GTC GAG CGC GGG TTC AC. The PCR-amplified product was cloned into the pET28b expression vector (Novagen) using NdeI and BamHI restriction sites. The resulting plasmid contains a six-histidine tag and thrombin cleavage site fused to the 5' end of D. radiodurans priA. The dnaB gene of D. radiodurans was amplified from strain R1 genomic DNA by polymerase chain reaction (PCR) using primers 5'-GCG TAT TCC ATA TGG AAA CGA CTC CGC GTG TG and 5'-GTC ACG GAT CCT CAC ATC CCC TCC GGC GCC. The PCR-amplified product was cloned into the pET28b expression vector (Novagen) using NdeI and BamHI restriction sites. The resulting plasmid contains a six-histidine tag and thrombin cleavage site fused to the 5' end of D. radiodurans dnaB. The fidelity of the cloned genes was confirmed by DNA sequencing. The priA gene of E. coli strain K12 was cloned as previously described [19].
The concentrations of recombinant proteins were determined by measuring their absorbance at 280 nm in 6.0 M guanidine-HCl using molar extinction coefficients of 184,370 M -

Nucleic acids
For the DNA unwinding assays and DNA binding assays, a fully-duplex DNA fork substrate with a 25 bp parental duplex arm, a 25 bp nascent leading strand arm, and a 25 bp 3'-fluorescein-labeled nascent lagging strand arm was constructed by annealing the following partially complementary DNA oligonucleotides: (template leading strand) 5'-GTC GGA TCC TCT AGA CAG CTC CAT GAT CAC TGG CAC TGG TAG AAT TCG GC; (template lagging strand) 5'-AAC GTC ATA GAC GAT TAC ATT GCT ACA TGG AGC TGT CTA GAG GAT CCG AC; (nascent leading strand) 5'-GCC GAA TTC TAC CAG TGC CAG TGA T; (nascent lagging strand with 3' fluorescein) 5'-TAG CAA TGT AAT CGT CTA TGA CGT T. The oligonucleotides were annealed in 10 mM TrisÁHCl pH 8, 50 mM NaCl, 1 mM EDTA at a 2:1 molar ratio of non-labeled DNA to fluorescein-labeled DNA. The DNAs were incubated at 95°C for 5 min, slow-cooled to 70°C and incubated at that temperature for 60 min, and slow-cooled to 25°C. The duplex DNA was gel-purified through a 6% polyacrylamide gel using 100 mM TrisÁborate pH 8.3, 2 mM EDTA as the electrophoresis buffer. The DNA was excised from the polyacrylamide gel, electroeluted using the same electrophoresis buffer, dialyzed against 10 mM TrisÁHCl pH 8, 5 mM MgCl 2 , aliquoted, and stored at -20°C. An 18-base mixedsequence ssDNA oligonucleotide (5'-AAG CAC AAT TAC CCA CGC) labeled at its 3' end with fluorescein was also used for the equilibrium DNA binding assays. A homopolymer dT 36 ssDNA oligonucleotide was used for steady-state enzyme kinetics experiments to measure DNA-dependent rates of PriA-catalyzed ATP hydrolysis. All oligonucleotides were purchased from Integrated DNA technologies, Inc (Coralville, IA).

DNA unwinding assays
The DNA fork substrate was diluted to 1 nM in 20 mM TrisÁHCl pH 8, 50 mM NaCl, 3 mM MgCl 2 , 1 mM 2-mercaptoethanol, 1 mM ATP. Indicated concentrations of D. radiodurans PriA or E. coli PriA were added to the reaction mixtures and incubated at 37°C for 10 min to facilitate duplex DNA unwinding. Reactions were stopped by addition of SDS to a final concentration of 1% (w/v). The amount of duplex DNA unwound was determined by measuring the fluorescence anisotropy of the samples after addition of SDS using a Beacon 2000 fluorescence polarization system (Invitrogen). Fluorescence anisotropy values were compared to the fluorescence anisotropy of the DNA fork substrate incubated in buffer alone (representing the fully intact DNA fork substrate) and the fluorescence anisotropy of the DNA fork substrate after being heated to 95°C and rapidly cooled to 25°C (representing the fully denatured DNA fork substrate) to calculate the fraction of DNA unwound.

ATP hydrolysis assays
PriA-catalyzed ATP hydrolysis was measured using a coupled spectrophotometric assay. This assay uses an ATP regeneration system that converts ADP to ATP in a reaction that is coupled to the oxidation of NADH to NAD + [20]. The coupled reaction can be monitored spectrophotometrically by measuring the decrease in absorbance at 340 nm due to NADH oxidation. Concentrations of D. radiodurans PriA or E. coli PriA ranging from 0-100 nM were incubated with 1 μM dT 36 and 1 mM ATP in 20 mM Hepes pH 8, 50 mM NaCl, 7 mM 2-mercaptoethanol, 2 mM phosphoenolpyruvate, 0.1 mM NADH, 7 unitsÁmL -1 pyruvate kinase, 10 unitsÁmL -1 lactate dehydrogenase, 0.1 gÁL -1 BSA at 37°C. Steady-state Δ[NADH]/Δt rates were calculated using the molar extinction coefficient 6,220 M -1 Ácm -1 for NADH, and these rates are equivalent to Δ [ATP]/Δt.

Native agarose gel electrophoresis
Indicated amounts of D. radiodurans PriA and D. radiodurans DnaB were mixed and incubated on ice for 10 min prior to the addition of gel loading buffer consisting of 20% (v/v) glycerol, 0.2% (w/v) bromophenol blue, 0.12 M Tris base. Samples were added to wells of horizontal 0.6% (w/v) agarose gels that were prepared with 25 mM TrisÁHCl pH 8.5 and 19.2 mM glycine. We obtained the best results when the wells were cast in the middle of the gel as opposed to one end. The gels were submerged in the same buffer solution and the samples were electrophoresed at 50 V for approximately 4 h at 4°C. The gels were stained with 0.12% (w/v) Coomassie Brilliant Blue R-250 in 45% (v/v) methanol and 10% (v/v) acetic acid and destained with 45% (v/v) methanol and 10% (v/v) acetic acid.
Protein unfolding assay D. radiodurans PriA at a final concentration of 0.1 μM was incubated with 0-6 M guanidine-HCl in a buffered solution consisting of 50 mM Hepes pH 7 and 1 mM 2-mercaptoethanol at 25°C. Fluorescence measurements were made at 25°C using a Cary Eclipse fluorescence spectrophotometer using an excitation wavelength of 280 nm and a 5 nm excitation filter slit width. Emission was recorded from 300 nm -400 nm using a 5 nm emission filter slit width and a scan rate of 30 nmÁmin -1 . Spectra for the protein samples were baseline subtracted using the spectra of the buffered solutions without PriA at each concentration of guanidine-HCl. The fluorescence intensities at 324 nm were used to calculate the fraction of D. radiodurans PriA folded at each concentration of guanidine-HCl, where the 6 M guanidine-HCl sample represents the unfolded spectrum and the sample with no guanidine-HCl represents the folded spectrum [21]. The fraction of PriA folded as a function of guanidine-HCl was modeled using a four-parameter function [y = (a + bÁx)/(1 + cÁx + dÁx 2 )] where a is the upper asymptote (corresponding to the folded protein), c is the lower asymptote (corresponding to the unfolded protein), and b and d influence the shape of the curve (CurveExpert 1.3). This model was used to interpolate the concentration of denaturant at which one-half of the protein is unfolded, [D] 1/2 .

Results and Discussion
To test the hypothesis that D. radiodurans PriA is a pseudohelicase, we cloned the D. radiodurans priA gene, expressed the recombinant PriA protein in an E. coli host, and purified the PriA protein to approximately 95% purity as estimated by SDS-PAGE analysis and Coomassie Brilliant Blue-R250 staining (data not shown). To determine if D. radiodurans PriA has helicase activity, we employed a fluorescence anisotropy-based duplex DNA unwinding assay using a fork DNA with a fluorescein-labeled 25 bp nascent lagging strand arm as the substrate, as described previously [22]. In brief, the fluorescein-labeled fork DNA substrate was incubated with either E. coli PriA or D. radiodurans PriA in the presence of ATP for 10 min at 37°C, the reactions were terminated by addition of SDS, and the fluorescence anisotropy of the samples was measured. The degree of unwinding of the fork DNA substrate was determined by comparing the fluorescence anisotropy of the samples to that of the fork DNA substrate incubated in buffer alone (representing the fully intact fork DNA substrate) and to samples heated briefly to 95°C and fast-cooled back to 25°C (representing the fully denatured fork DNA substrate). This allowed us to measure the fraction of the fork DNA substrate that is unwound by various concentrations of the PriA proteins. As expected, E. coli PriA is capable of unwinding the fork DNA substrate in a PriA-dependent manner with 56% unwinding of the fork DNA occurring at 10 nM PriA (Fig 3). D. radiodurans PriA shows negligible levels of unwinding on the same fork DNA substrate. At 10 nM D. radiodurans PriA, only 4% of the fork DNA is unwound (Fig 3).
Given the lack of conservation of catalytic amino acid residues in the Walker A and Walker B boxes of D. radiodurans PriA, it seems likely that the inability of D. radiodurans PriA to unwind the fork DNA substrate in the duplex DNA unwinding assays stems from an inability of D. radiodurans PriA to hydrolyze ATP. Therefore, we measured the ATPase activity of D. radiodurans PriA using a coupled spectrophotometric assay that relates ATP hydrolysis to the oxidation of NADH to NAD + . We then compared the ATPase activity of D. radiodurans PriA to that of E. coli PriA. This assay uses a homopolymer dT 36 oligonucleotide to stimulate the ATPase activity of PriA since PriA's ability to catalyze ATP hydrolysis is known to be strongly dependent upon the presence of a DNA lattice [22,23]. Under these experimental conditions, 100 nM E. coli PriA catalyzes ATP hydrolysis at a rate of 369 ± 9 nM ATPÁsec -1 (Fig 4). Under these same conditions, D. radiodurans PriA does not catalyze ATP hydrolysis at a measurable rate (Fig 4).
To verify that our recombinant D. radiodurans PriA protein is not unfolded, we measured its intrinsic fluorescence as a function of increasing concentrations of the denaturing agent, guanidine-HCl. The fluorescence emission spectra of D. radiodurans PriA in the presence and absence of 6 M guanidine-HCl clearly differ from one another (Fig 5). The spectra show the greatest difference in fluorescence intensity at 324 nm, so fluorescence intensity measurements  at this wavelength were used to calculate the fraction of PriA folded at various concentrations of denaturant. The resulting unfolding curve lacks a strong folded-state baseline (Fig 5), which precludes an estimation of the free energy of the unfolding reaction. The denaturant concentration at the midpoint of the unfolding curve, [D] 1/2 , is 1.2 M for guanidine-HClinduced unfolding. Therefore, regardless of the mechanism by which PriA unfolds or the presence and number of any unfolding intermediates, these data indicate that the recombinant D. radiodurans PriA protein that we expressed in E. coli, purified, and tested for duplex DNA unwinding activity, DNA binding activity, and ATPase activity is not unfolded.
Given the lack of helicase activity, we sought to determine if D. radiodurans PriA possesses any other activities that would be expected for a functional PriA. Thus, we tested for its ability to bind to an 18-base mixed-sequence ssDNA oligonucleotide using a fluorescence anisotropybased equilibrium DNA binding assay. For these experiments, D. radiodurans PriA at various concentrations was incubated with a fixed concentration of fluorescein-labeled ssDNA and the fluorescence anisotropy of the mixtures was measured. We observed a D. radiodurans PriAdependent increase in the fluorescence anisotropy of the fluorescein-labeled ssDNA, indicating that D. radiodurans PriA is capable of binding to ssDNA (Fig 6A). These equilibrium DNA binding assays were also conducted using E. coli PriA and the same ssDNA substrate. As expected, we observed an E. coli PriA-dependent increase in the fluorescence anisotropy of the fluorescein-labeled ssDNA, confirming many previously published reports that E. coli PriA binds ssDNA (Fig 6A) [24][25][26]. The calculated apparent dissociation constants are 129.2 ± 11.9 nM for the E. coli PriA-ssDNA interaction, and 215.4 ± 17.0 nM for the D. radiodurans PriA-ssDNA interaction. These apparent dissociation constants differ only slightly from one another, suggesting that the structural features of PriA that are required for binding to ssDNA have been largely conserved between these two PriAs throughout evolution. This also indicates that the lack of ATPase activity observed for D. radiodurans PriA is not due to a defect in binding ssDNA. We next tested the ability of D. radiodurans PriA to bind to a fork DNA substrate which mimics the type of DNA structure that PriA would presumably bind in vivo during replisome reactivation. As with the ssDNA substrate, we observed a D. radiodurans PriA-dependent increase in the fluorescence anisotropy of the fluorescein-labeled fork DNA, indicating that D. radiodurans PriA is capable of binding to a fork DNA structure (Fig 6B). Likewise, E. coli PriA binds to the same fork DNA substrate and does so with an affinity that is similar to that of D. radiodurans PriA (Fig 6B). The calculated apparent dissociation constants are 52.1 ± 1.0 nM for the E. coli PriA-fork DNA interaction, and 108.5 ± 1.7 nM for the D. radiodurans PriA-fork DNA interaction. We have previously used this same equilibrium DNA binding assay and the same fork DNA substrate to measure the interaction between Neisseria gonorrhoeae PriA and fork DNA and determined the apparent dissociation constant to be 134 ± 22 nM for the N. gonorrhoeae PriA-fork DNA interaction [22]. Thus, the affinity of D. radiodurans PriA for the fork DNA substrate lies between that of the E. coli and N. gonorrhoeae PriAs, both of which have well-documented helicase activities.
Given that D. radiodurans PriA binds ssDNA and fork DNA, we next wanted to determine if it physically interacts with any other proteins that would be expected to be present at a repaired DNA replication fork in the context of a replisome reloading pathway. Since D. radiodurans lacks clear homologs of priB and dnaT in its genome, we tested for the ability of D. radiodurans PriA to bind to its cognate replicative helicase, DnaB. While a PriA-DnaB physical interaction has not been reported in the well-studied E. coli system, it is likely that E. coli PriA leads to the recruitment of DnaB to a repaired replication fork through some type of physical interaction, perhaps mediated indirectly through other primosome proteins with which PriA physically interacts. We used native agarose gel electrophoresis to test for a physical interaction between D. radiodurans PriA and DnaB proteins. When 50 pmol of D. radiodurans PriA is resolved by itself through an agarose gel under native conditions, Coomassie Brilliant Blue R250 staining reveals the presence of a single band (Fig 7, lane 1). Likewise, resolution of 50 pmol of D. radiodurans DnaB by itself under the same electrophoresis conditions produces a single band (Fig 7, lane 2). These individual PriA and DnaB bands do not comigrate through the gel. However, when the PriA and DnaB proteins are mixed together and the mixture is resolved through a native agarose gel, only a single band is observed (Fig 7, lanes 3-5). This band migrates approximately midway between the positions of free PriA and free DnaB in the gel, suggesting that it represents a complex of the PriA and DnaB proteins.
In conclusion, we have demonstrated that PriA from D. radiodurans is a pseudohelicase that has retained the DNA binding activity of a typical bacterial PriA but can neither hydrolyze ATP nor unwind duplex DNA. We have also shown evidence for a physical interaction between D. radiodurans PriA and its cognate replicative helicase, DnaB. Collectively, these data suggest that D. radiodurans PriA has retained a function in DNA replication restart pathways despite its inability to unwind duplex DNA at a repaired DNA replication fork. Whether D. radiodurans uses another DNA helicase for this function, or catalyzes replisome reloading onto DNA structures that do not require remodeling by a helicase remains unclear.