Strand Displacement by DNA Polymerase III Occurs through a τ-ψ-χ Link to Single-stranded DNA-binding Protein Coating the Lagging Strand Template*

In addition to the well characterized processive replication reaction catalyzed by the DNA polymerase III holoenzyme on single-stranded DNA templates, the enzyme possesses an intrinsic strand displacement activity on flapped templates. The strand displacement activity is distinguished from the single-stranded DNA-templated reaction by a high dependence upon single-stranded DNA binding protein and an inability of γ-complex to support the reaction in the absence of τ. However, if γ-complex is present to load β2, a truncated τ protein containing only domains III–V will suffice. This truncated protein is sufficient to bind both the α subunit of DNA polymerase (Pol) III and χψ. This is reminiscent of the minimal requirements for Pol III to replicate short single-stranded DNA-binding protein (SSB)-coated templates where τ is only required to serve as a scaffold to hold Pol III and χ in the same complex (Glover, B., and McHenry, C. (1998) J. Biol. Chem. 273, 23476–23484). We propose a model in which strand displacement by DNA polymerase III holoenzyme depends upon a Pol III-τ-ψ-χ-SSB binding network, where SSB is bound to the displaced strand, stabilizing the Pol III-template interaction. The same interaction network is probably important for stabilizing the leading strand polymerase interactions with authentic replication forks. The specificity constant (kcat/Km) for the strand displacement reaction is ∼300-fold less favorable than reactions on single-stranded templates and proceeds with a slower rate (150 nucleotides/s) and only moderate processivity (∼300 nucleotides). PriA, the initiator of replication restart on collapsed or misassembled replication forks, blocks the strand displacement reaction, even if added to an ongoing reaction.

All cellular replicases are tripartite assemblies, consisting of a replicative polymerase (Pol 2 III in bacteria, Pol ␦ or ⑀ in eukaryotes), a sliding clamp processivity factor (␤ 2 in bacteria, proliferating cell nuclear antigen in eukaryotes), and a clamp loader composed of a five-protein core of AAA ϩ -like proteins that assembles the sliding clamp around DNA (DnaX complex in bacteria, RFC in eukaryotes) (1)(2)(3). The Escherichia coli DnaX complex comprises three copies of DnaX and one each of ␦, ␦Ј, and (4 -6). E. coli and many other bacteria contain two forms of DnaX, the full-length translation product and a shorter protein, ␥, that results from translational frameshifting (7)(8)(9). Both and ␥ contain the three domains that are required for ATP-dependent ␤ 2 loading onto DNA (6). The third domain of and ␥ is responsible for binding other DnaX subunits as well as ␦, ␦Ј, and (5,10,11). contains two additional domains that interact with the DnaB replicative helicase (domain IV) and Pol III (domain V) (12,13).
The primary determinant of processivity of the E. coli replicase is the interaction of Pol III with ␤ 2 (14,15). However, other interactions stabilize the interaction of Pol III with the replication fork. Two protomers bind the DnaB helicase, further stabilizing the replicase at the fork (16,17). Pol III alone is unable to replicate single-stranded DNA (ssDNA) coated by SSB. To accomplish this feat, , , and must be present if ␤ 2 is absent (18). does not serve its prototypical role as the clamp loader in this minimal system but apparently only serves as a bridge, tethering in the same complex with Pol III, enabling an otherwise weak Pol III-template interaction to be stabilized by a -SSB contact (18).
During processive replication of long single-stranded templates, the Pol III HE typically stops synthesis upon encountering a duplex (19). The elongation reaction is very rapid (400 -700 nt/s) and exhibits processivity that may enable replication of the entire E. coli chromosome without dissociation (1). However, a more feeble strand displacement activity of the DNA polymerase III holoenzyme has been observed under a variety of conditions (20 -23).
Canceill and Ehrlich (20) observed that Pol III HE could replicate through a 30-nt stem. A mechanism was proposed where the enzyme could non-processively elongate a few nucleotides when base pairing is transiently disrupted. Stephens and McMacken (21) observed more extensive strand displacement synthesis on flapped templates using native Pol III HE in a reaction that was dependent upon SSB. Xu and Marians (22), in studies of replicative resolution of recombination intermediates, observed a background strand displacement reaction in the absence of helicase. O'Donnell and co-workers have also observed strand displacement of oligonucleotides containing large internal secondary structures (23).
Observations of the strand displacement activity of Pol III HE have occurred peripheral to studies conducted for other pur-poses. No investigation has been made aimed at understanding the reaction, its protein requirements, and how it might differ from the well characterized reaction that occurs on singlestranded templates. This study was directed toward remedying these deficiencies in knowledge. We found that an interaction of the leading strand polymerase with the lagging strand template, mediated by a Pol III----SSB bridge, was essential for efficient strand displacement. This interaction network probably stabilizes the replicase at the fork in addition to the characterized Pol III-␤ 2 and -DnaB interactions.
DNA Templates-M13Gori ssDNA, M13Gori template with a 30-nt primer, and activated calf thymus DNA were prepared as described previously (25,32,33). pUCNICK tail DNA was generated from plasmid pUCNICK (2,716 bp) (34) and purified by Qiagen plasmid Maxi prep. Purified plasmid DNA (30 g, 1 g/l) was nicked at the single recognition site with Nb.BbvCI nicking enzyme (300 units; New England Biolabs) at 37°C for 4 h. Nb.BbvCI was thermally inactivated (80°C, 20 min). A 1% agarose gel showed that the nicking reaction was Ͼ90% complete. The nicked plasmid was precipitated by the addition of 0.5 volumes of 5 M ammonium acetate and 1.5 volumes of isopropyl alcohol, and the pellet was washed with 70% ethanol and dissolved in 10 mM Tris-HCl buffer (pH 8) to a 0.2 g/l final concentration. Glycerol was added to a 13% final concentration. This solution was incubated with 200 M dATP, dGTP, and dTTP and Klenow fragment (exo ϩ ; 2 units/g of DNA) at 37°C for 30 min, so that the 61-nt-long cytosineless DNA flap was formed. EDTA was added to a final concentration of 20 mM. Tailed pUCNICK (2,777 nt) was precipitated with isopropyl alcohol and dissolved in 10 mM Tris-HCl (pH 8) at a concentration of 1 g/l. Tailing efficiency, determined by PvuII digestion and 1.8% agarose gel electrophoresis, was Ͼ90% complete.
Strand Displacement Assays-4 nM pUCNICK tail, 100 M ATP, 0.75 M SSB 4 , 25 nM ␤ 2 , 51 nM Pol III, 17 nM -complex, 300 M dNTPs, and ϳ130 cpm/pmol [ 3 H]TTP were incubated for 5 min at 30°C in 25 l. The strand displacement buffer was the same as the single-stranded replication buffer, except it contained 20 mM NaCl. Incorporation was expressed in terms of total nucleotide by multiplying the total dTTP incorporated by 4 (pUCNICK contains 24.7% T).
Determination of the Rate of Strand Displacement-Initiation complexes between Pol III HE and flapped templates were formed by incubating 5 nM pUCNICK tail, 125 M ATP, 0.94 M SSB 4 , 31 nM ␤ 2 , and 25 nM Pol III* at 30°C for 1 min. Then 375 M dNTPs and 5 Ci/reaction [␣-32 P]dTTP were added to initiate strand displacement. Aliquots of the 20-l reaction were quenched by 100 mM EDTA (final concentration) at various time points. The samples were loaded along with alkaline agarose gel loading buffer (30 mM NaOH, 2 mM EDTA, 1% glycerol, and 0.02% bromphenol blue) to 0.5% alkaline agarose gels in a running buffer of 30 mM NaOH and 2 mM EDTA. Gels were run at 22 V for 18.5 h, fixed in 5% trichloroacetic acid, dried, and scanned by a PhosphorImager. The molecular weights of the longest reaction products were determined by measuring their relative mobilities (compared with 32 P-labeled DNA size marker) on the alkaline agarose gel.
Determination of the Processivity of Pol III HE on a Shortflapped Template-Challenge assays were performed to determine processivity. In these assays, a large excess of activated calf thymus DNA (challenge) was added to elongating complexes to trap dissociated polymerase so that products represented a single processive association-elongation-dissociation event. Solutions containing strand displacement buffer (125 M ATP, 0.31 M SSB 4 , 31 nM ␤ 2 , and 10 nM Pol III * ) were prewarmed and mixed with 1.5 nM ␥-32 P-labeled 32/91, 32/50/91, or 67/50/91 at 30°C for 15 s to form the initiation complexes. To block ⑀-catalyzed primer degradation, 10 M dGTP was added for 32/91 and 32/50/91 templates, and 10 M dTTP and dCTP was added for 67/50/91 during incubation. Then 375 M dNTPs (final concentration) and 20 g of activated calf thymus DNA were added to allow the reaction to occur. After 10 s, 55% formamide, 50 mM EDTA, 0.0083% bromphenol blue, and 0.0083% xylene cyanol (final concentrations) were added to quench the 20-l reaction. The solution was fractionated on 12% polyacrylamide gel with 8.2 M urea for 3 h at 95 watts. The gel was scanned with a PhosphorImager and quantified with Image-Quant version 5.2 software (Amersham Biosciences). Controls to show that the challenge was effective followed the same procedure, except activated calf thymus DNA was added with the templates before the addition of Pol III HE.
Determination of the Processivity of Pol III HE on a Rolling Circle Template-In this assay, prewarmed 1 nM pUCNICK tail, 100 M ATP, 0.75 M SSB 4 , 25 nM ␤ 2 , and 10 nM Pol III* were assembled to form holoenzyme initiation complexes for 20 s at 30°C. Then 300 M dNTPs (final concentration) and 5 g of activated calf thymus DNA were added to initiate the reaction. This 25-l reaction was carried out at 30°C for various time periods and stopped by 83 mM EDTA (final concentration). Each reaction sample was precipitated by the addition of 0.5 volume of 5 M ammonium acetate, 20 g of glycogen, and then 2.5 volumes of 95% ethanol. The pellet was washed by 70% ethanol and dissolved in 20 l of 10 mM Tris-HCl buffer (pH 8). DNA was then digested by 75 units of EcoRI at 37°C for 18 h. Unchallenged strand displacement assays followed the same procedure, except activated calf thymus DNA was added just prior to EcoRI digestion to ensure uniformity in sample workup. After digestion, all samples were treated at 95°C for 5 min with alkaline agarose gel loading buffer and then loaded onto 1.5% alkaline agarose gels in a running buffer of 30 mM NaOH and 2 mM EDTA. Gels were run at 100 V for 2 h and then fixed in 5% trichloroacetic acid, dried, and scanned by a Phosphor-Imager. The product bands were quantified by ImageQuant 5.2 software (Amersham Biosciences). Because the reaction products were diffuse, the center of the product in each lane (r avg ) was defined as the average length of the product. The whole product was sliced into pieces about every 100 nucleotides. r avg ϭ ͚(r i ϫ P i )/͚P i , where r i represents the average length of product slice i, and P i is the pixel of product slice i. The processivity of Pol III* was calculated by subtracting the primer length, 67 nt, from r avg .
PriA Inhibition Assay-To test whether PriA could inhibit strand displacement before the initiation complex formed, 10 nM pUCNICK tail, 0.94 M SSB 4 , 31 nM ␤ 2 , 125 M ATP, 375 M dNTPs, and 100 cpm/pmol [ 3 H]TTP were mixed first and then incubated with 10 nM Pol III* and varying amounts of PriA for 5 min at 30°C in 20 l. To test whether PriA could inhibit strand displacement after initiation complex formation, the same procedure was followed, except that protein components were first incubated with pUCNICK tail in the presence of ATP at 30°C for 2 min, and then dNTPs were added with PriA. For assays on ssDNA templates, PriA was added following the same sequence as described above either before or after the initiation complex formed. Every other component contained the same concentration as described under "Single-stranded Replication Assay," except that 10 nM Pol III* replaced Pol III and -complex.

RESULTS
An initial characterization of the strand displacement activity of Pol III HE was performed on a circular template with a 61-nt flap. Protein components were titrated and compared with the requirement for standard processive assays on long single-stranded templates (Fig. 1). Approximately twice as much Pol III was required for the strand displacement reaction, probably a consequence of the decreased processivity of Pol III HE in the strand displacement reaction (see below). Both -complex and ␤ 2 were required to support the strand displacement reaction at protein levels approximately the same as required on single-stranded templates. Strikingly, the ␥ form of DnaX could not be substituted for in the strand displacement reaction, in contrast to reactions on single-stranded DNA templates. Another significant difference was observed in the SSB requirement. As is typically observed, SSB stimulated the Pol III HE marginally (ϳ2-fold) on preprimed single-stranded templates; the dependence was nearly absolute for strand displacement.
Another profound difference became apparent upon varying dNTP concentrations (Fig. 1F). In the ssDNA reaction, a low micromolar K m was measured. However, nearly 100-fold higher concentrations of dNTPs were required to drive the strand displacement reaction. To obtain a comparison of the specificity differences of the two reactions, we calculated k cat /K m that revealed a 330-fold preference of Pol III HE for synthesis on an ssDNA template compared with a duplex, during strand displacement (Table 1). A possible explanation for the high K m for strand displacement could be the need for a rapid second-order nucleotide association reaction being required to trap an intermediate that lies on the elongation pathway and competes with steps that lead to dissociation. It is also possible that association of the polymerase with a displaced strand or loss of ssDNA template contacts might place it into an alternative conformation with a distorted active site that interacts with dNTPs less favorably.
Next, we asked whether the requirement was a manifestation of a unique role for in loading ␤ 2 onto DNA or whether performed another function, separate from ␤ 2 loading. To address this issue, we loaded ␤ 2 onto the DNA templates with ␥-complex and determined the contribution of various forms of . Adding alone had little effect, but adding a complex of --, which alone is inactive in ␤ 2 loading, stimulated the strand displacement reaction significantly ( Fig. 2A). and ␥, in the absence of other proteins, exchange very slowly. The presence of ␦, ␦Ј, and blocks exchange, eliminating the possibility that , added briefly to reactions, exchanges into the ␥-complex (35). Furthermore, if the result obtained was due to such an exchange reaction, would not be required, since it is already present in the ␥-complex.
In earlier work, we observed an effect of in enabling Pol III to replicate ssDNA coated with SSB, which required only its function of binding . In this example, held in the same complex with Pol III, enabling it to bind template-bound SSB and stabilizing an otherwise weak interaction (18). Thus, we added a protein that comprises domains III-V of DnaX, III-V . DnaX domain III binds , and domain V binds the ␣ subunit of Pol III (5,12). If the only function of is to link Pol III and , this truncated protein should suffice to stimulate strand displacement when ␥-complex is present to load ␤ 2 . Indeed, III-V stimulated strand displacement the same amount as full-length bound to , although higher concentrations were required (Fig. 2B). The addition of an equimolar mixture of IV-V and did not stimulate the reaction. IV-V binds the ␣ subunit of Pol III but not .
The above results are consistent with a critical -SSB contact required to stabilize the strand displacement reaction. interacts with the C-terminal tail of SSB (36,37). To further test the existence and importance of this interaction in the strand displacement reaction, we replaced wild-type SSB with two C-terminal SSB proteins that had 8 and 42 amino acids deleted from their C termini (30,31). We observed that neither supported strand displacement (Fig. 3A), consistent with our hypothesis. As a control experiment, we tested the effect of both on the Pol III HE single-strand templated reaction and observed an inhibition, although there is not much of a requirement for SSB in the normal reaction (Fig. 3B). Thus, deletion of the C terminus of SSB creates a gain of an inhibitory function for SSB, presumably because the protein-interacting tail is not available for modulation of binding state (30) or displacement from the template.
We also determined the rate of elongation for the strand displacement reaction by determining the length of the longest products visible on a denaturing gel starting with labeled primer (Fig. 4). We observed a rate of 150 nt/s, slower than the FIGURE 1. Requirements for Pol III HE components for strand displacement. Pol III HE protein subunits were titrated in single-stranded (ss) replication reactions (ࡗ) and strand displacement reactions (F) in the presence of other protein subunits at optimal saturating levels as described under "Experimental Procedures." A, titration with Pol III. B, titration with ␤ 2 . C, titration with -complex. D, titration with ␥-complex. E, titration with SSB 4 . In this single experiment, an M13Gori template primed by annealing a 30-nt primer was used instead of DnaG primase-primed template, so the elongation requirement of SSB could be determined separately from the absolute requirement for DnaG-catalyzed primer formation. F, titration with dNTPs.

TABLE 1 Steady state kinetic parameters for Pol III HE in single-stranded replication and strand displacement
The dNTP titration curves were fit to the Michaelis-Menton kinetic function, 1/v ϭ (K m /v max ͓S͔) ϩ (1/v max ), by nonlinear least-squares regression to determine both v max and K m . ͓S͔ was the concentration of dNTPs, and v was the rate of dNTP incorporation (pmol/min). The value for k cat at 30°C for the strand displacement reaction was taken from the data reported in Fig. 4; that for the ssDNA reaction was taken from Ref. 43. Based on an initial expectation of low processivity, we made a series of synthetic templates with a common 91-nt segment (Fig. 5). One was simply primed at the 3Ј-end of the template with no other oligonucleotides annealed (template a). Two contained flapped blocking oligonucleotides, one with a 35-nt gap between the primer terminus and the flap (template b) and the other with the primer terminus abutting the flap junction (template c). Processivity was determined by first forming initiation complexes and then adding an excess of challenge DNA to capture any polymerase that dissociated during elongation. The efficacy of the challenge template was demonstrated by complete inhibition if added with the template before the addition of enzymes (Fig. 5, lanes 4 -6). In the absence of a challenge, all templates were nearly completely elongated to the expected full-length 91-nt product (lanes 7-9). If initiation complexes were formed prior to the addition of the challenge template concomitant with dNTPs, again most of the primer was elongated to full-length product (lanes 10 -12), indicating a processivity greater than 24 for strand displacement. However, 30% of the elongated product on template b terminated when the Pol III HE encountered the flap (lane 11), although little product of a length intermediate between 67-mer and 91-mer was detected (Ͻ2%).

Single-stranded replication Strand displacement
Because the processivity was too high to estimate on short, linear templates, we turned to longer flapped templates. Initial experiments failed to resolve the product of processive synthesis from the 2,777 nt starting material, indicating limited processivity. To permit resolution, we labeled the primer for the template during creation of the flap, limiting the position of radioactive nucleotides to the 3Ј-terminal 61 nt. After elongation, this permitted cleavage with restriction endonuclease EcoRI, generating a product of 67 nt plus the number of nt added during the elongation event. This permitted better product resolution. Using this longer template, a challenge experiment was conducted, similar to the one performed on the short, synthetic template above. The presence of a challenge template mixed with the labeled template prior to enzyme addition inhibited the elongation reaction (Fig. 6, lanes 8 and 9). Preformation of initiation complexes followed by the addition of the  challenge template with dNTPs limited synthesis (Fig. 6, lanes  10 -13) relative to the unchallenged controls (lanes 4 -7), indicating limited processivity. Since processivity is an intrinsic property of an enzyme, the length of a processive product should not be affected by incubation time. We observed that the product length remained unchanged beyond the initial 5 s time point. Thus, the 10 -20 s products from the elongation experiment (Fig. 6, lanes 11-13) were used to calculate processivity. The population of products as a function of length was quantified, and an average processivity of 280 nt was calculated.
Xu and Marians (22) have observed helicase-independent strand displacement by Pol III HE in complex reactions where recombination intermediates are resolved replicatively. We investigated whether PriA could block the strand displacement reaction by Pol III HE in the simpler system we use, where SSB is the only other protein present. We observed that PriA blocked the strand displacement reaction whether or not an initiation complex was formed between the Pol III HE and DNA prior to the PriA addition. In contrast, PriA had no effect on the Pol III HE-catalyzed reaction on ssDNA templates (Fig. 7A). We also investigated whether PriA could halt an ongoing elongation reaction (Fig. 7B). Initiation complexes were formed on flapped templates, and 45 s after initiation, PriA was added, resulting in an immediate block in the presence of 56 nM PriA.

DISCUSSION
We observed that the Pol III HE has an intrinsic strand displacement reaction that has properties markedly different from those of the well studied synthesis reaction catalyzed on ssDNA templates. Both reactions require Pol III, ␤ 2 , and a clamp loader. However, unlike the ssDNAtemplated reaction, SSB is nearly absolutely required for strand displacement. It only modestly stimulates reactions on ssDNA templates when the complete Pol III HE is present. Even more striking is the observation that ␥-complex alone is not effective; must be present for strand displacement to occur. However, the unique requirement for is not a consequence of its ␤ 2 clamp loading activity. If ␥-complex is provided to load ␤ 2 , a truncated protein that lacks the critical domains required for ATP binding and hydrolysis (domains I and II) will serve to drive strand displacement. The truncated must contain domain III, the binding domain (5). The C-terminal tail of SSB that is involved in a variety of protein interactions (38) is required.
These observations are reminiscent of the minimal Pol III that is required for modest replication on SSB-coated ssDNA templates, Pol III---, where only serves as a tether to hold and Pol III in the same complex (18). The explanation of activity in this system was that contacted SSB when bound to   7, 8, 10, and 11, respectively; percentages of 67-mer were 0.2, 7.6, 26, and 19% for lanes 8,9,11, and 12, respectively; and percentages of 91-mer were 80, 92, 92, 66, 61, and 81% for lanes 7-12, respectively. Right, the three DNA templates (a-c) with 32 P-labeled primers (indicated by an asterisk) made as described under "Experimental Procedures." SA, positions of streptavidin attachment.
ssDNA, increasing binding of the polymerase. The molecular interactions behind this protein network are well understood. Pol III binds in an interaction between the C terminus of Pol III and domain V of (12,39,40). One protomer binds a trimeric assembly of DnaX proteins through their domain III (5,11). binds (41), and binds to SSB through its C-terminal domain (18,36,37).
In the strand displacement reaction, however, the only ssDNA available for SSB binding is the displaced strand. Thus, we propose the model depicted in Fig. 8, where a Pol III----SSB interaction stabilizes the interaction of Pol III with the template sufficiently to permit moderately processive strand displacement. It is interesting that the ␤ 2 -Pol III interaction alone is inadequate to stabilize the Pol III-template interaction sufficiently to enable strand displacement. Perhaps limited ssDNA template-Pol III contacts make additional stabilizing interactions necessary. The interaction network shown in Fig. 8 could be important for stabilizing interaction of the leading strand polymerase at the replication fork, through an interaction of the DnaX complex with the lagging strand template. Normally, a 2 -DnaB 6 interaction will further stabilize the replisome, but there may be situations (e.g. when difficult structures are encountered during mismatch repair) where the Pol III----SSB interaction network becomes critical for function. In unusual cases, such as blockage of the leading strand polymerase by a lesion or other obstruction when the helicase continues to progress, single-stranded DNA would be created on the leading strand template and bound by SSB, enabling stabilization of Pol III HE interactions, perhaps providing a tether to localize Pol III HE during polymerase switching.
The Pol III HE strand displacement reaction exhibits only modest (ϳ300 nt) processivity as compared with the proposed megabase processivity of the Pol III HE on ssDNA and the Ͼ100,0000-base processivity observed on reconstituted replication forks. Thus, even a combination of the Pol III-␤ 2 interaction and the -mediated -SSB contact cannot provide FIGURE 6. Pol III HE adds 280 nucleotides processively during strand displacement. Challenge experiments were performed to determine the processivity of Pol III* on the pUCNICK tail as described under "Experimental Procedures." Left, lane 1, 32 P-labeled pUCNICK tail without EcoRI digestion; lane 2, 32 P-labeled pUCNICK tail and unlabeled activated calf thymus DNA with EcoRI digestion; lane 3, 32 P-labeled pUCNICK tail, unlabeled activated calf thymus DNA, EDTA, Pol III HE mixture, and dNTPs were mixed sequentially, and then DNA was precipitated with ethanol and digested with EcoRI; lanes 4 -7, strand displacement reactions of designated times; lanes 8 and 9, reactions where the challenge DNA was added before initiation complex formation; lanes 10 -13, reactions where the challenge DNA was added after initiation complex formation. Right, pUCNICK tail strand displacement and EcoRI digestion. The red segment indicates the positions of 32 P-labeling. Two brackets illustrate the unextended 32 P-labeled 67-nt primer and the elongation product after EcoRI digestion.  sufficient stability for the highest levels of processivity. It is interesting that approximately one-third of Pol III HE dissociates upon encountering a flap while actively polymerizing yet appears to dissociate at the very low frequency required for 300-nt processivity at other positions. The enzyme presumably encounters the same structure prior to the addition of each nucleotide. It is possible that formation of the -SSB contact with the lagging strand template (displaced strand in Fig. 8) is on the same order as the rate of nucleotide addition and that, upon encountering a flap, a portion of the enzymes fails to form the contact in adequate time and dissociates due to weak interactions.
As first pointed out by Xu and Marians (22), a Pol III HE-catalyzed DnaB-independent strand displacement reaction could present problems for the cell and, under some circumstances, is negatively regulated by PriA. They proposed that PriA could act by binding to the 3Ј-end of a primer juxtaposed to a fork and block binding by the Pol III HE. Interestingly, gp59, a T4 bacteriophage-encoded protein that has multiple functions, some of which overlap with PriA, can block the action of T4 DNA polymerase by forming a ternary complex with it on DNA, where gp59 site-specifically contacts the polymerase and locks it into a conformation where exonuclease and polymerase activities are inhibited (42). Our studies cannot yet resolve which of these two mechanisms are used by E. coli PriA in blocking the Pol III HE strand displacement reaction.
Our studies have an additional practical benefit. Rolling circle DNA replication systems that mimic the action of the replisome at an in vivo replication fork typically employ the PriA protein to initiate a series of interactions that result in the biologically relevant assembly of an active DnaB helicase at a replication fork. However, fork systems are sometimes assembled that lack the natural helicase loaders, and researchers instead add large excesses of DnaB to drive self-assembly. Such systems, lacking the PriA checkpoint protein, could be complicated by the strand displacement activities of Pol III HE, especially if conducted in the presence of and high Pol III and dNTP concentrations. The understanding of the properties of the intrinsic Pol III HE strand displacement reaction will permit artifacts, driven by its action, to be avoided in such systems in the future.