Atomic force microscopy–based characterization of the interaction of PriA helicase with stalled DNA replication forks

In bacteria, the restart of stalled DNA replication forks requires the DNA helicase PriA. PriA can recognize and remodel abandoned DNA replication forks, unwind DNA in the 3′-to-5′ direction, and facilitate the loading of the helicase DnaB onto the DNA to restart replication. Single-stranded DNA–binding protein (SSB) is typically present at the abandoned forks, but it is unclear how SSB and PriA interact, although it has been shown that the two proteins interact both physically and functionally. Here, we used atomic force microscopy to visualize the interaction of PriA with DNA substrates with or without SSB. These experiments were done in the absence of ATP to delineate the substrate recognition pattern of PriA before its ATP-catalyzed DNA-unwinding reaction. These analyses revealed that in the absence of SSB, PriA binds preferentially to a fork substrate with a gap in the leading strand. Such a preference has not been observed for 5′- and 3′-tailed duplexes, suggesting that it is the fork structure that plays an essential role in PriA's selection of DNA substrates. Furthermore, we found that in the absence of SSB, PriA binds exclusively to the fork regions of the DNA substrates. In contrast, fork-bound SSB loads PriA onto the duplex DNA arms of forks, suggesting a remodeling of PriA by SSB. We also demonstrate that the remodeling of PriA requires a functional C-terminal domain of SSB. In summary, our atomic force microscopy analyses reveal key details in the interactions between PriA and stalled DNA replication forks with or without SSB.

A stable DNA replication system is essential for cell viability because DNA replication frequently encounters damages or replication blocks that need to be repaired or removed as soon as possible (1)(2)(3)(4)(5). In bacteria, the restart of the replication machinery is mediated by a series of enzymes, including the PriA helicase (6,7). It is known that PriA is required for replication restart and that the helicase activity is not required for restart (8). The strong dependence on PriA indicates that a replisome needs to be reassembled at the inactivated forks by a mechanism that can be distinguished from the original initiation by DnaA at oriC (9).
PriA is a primosome assembly protein with 3Ј-to 5Ј-DNA helicase activity (10). It was discovered because of its requirement for the synthesis of the complementary strand of bacteriophage X174 ssDNA in vitro (11)(12)(13)(14). PriA has a two-domain architecture: an N-terminal DNA-binding domain and a C-terminal helicase domain (15)(16)(17)(18)(19)(20). The cooperation among each domain preserves the recognition and binding activity of PriA to various DNA constructs, as well as the interaction with other proteins (6,21,22).
In addition to structure-specific DNA binding, PriA interacts with the ssDNA-binding protein (SSB) 2 (23)(24)(25). SSB, an essential protein in Escherichia coli, binds to and thereby stabilizes the ssDNA that occurs in DNA replication, recombination, and repair (26,27). PriA recognizes abandoned DNA replication forks with either duplex DNA or SSB-coated ssDNA and processes these to expose ssDNA as a binding site for primosome components (PriB and DnaT), followed by the reloading of the replicative helicase DnaB (22,28). It was also shown that the PriA helicase activity could be stimulated by the binding of SSB onto the initial DNA substrate (15,23). Recent work has shown that the SSB interaction with PriA involves the oligosaccharideoligonucleotide binding fold (OB-fold) within the N-terminal domain of the helicase and the linker domain of SSB (29). The mechanism of binding is identical to that between the RecG OBfold and the linker domain of SSB (30 -32).
To understand how PriA interacts with forks and how binding might be influenced by SSB, we used atomic force microscopy (AFM). The experiments were performed in the absence of ATP to separate the PriA-DNA binding properties of the protein from its helicase activity. AFM studies revealed the role of the fork type on the efficiency of PriA binding. Furthermore, SSB protein interacts with PriA and changes the protein conformation, allowing for the binding of PriA to the DNA duplex. Experiments with an SSB mutant revealed the role of the C-terminal segment in this remodeling activity of SSB. These findings are consistent with the ability of SSB protein to remodel the DNA helicase RecG that we discovered and described previously (33,34). This work was supported by NIGMS, National Institutes of Health Grants R01 GM118006 (to Y. L. L.) and R01 GM100156 (to P. R. B. and Y. L. L.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S5. 1 To whom correspondence should be addressed. E-mail: ylyubchenko@ unmc.edu.

Binding preference of PriA to various DNA constructs
DNA constructs- Fig. 1 shows a set of four DNA constructs used in this work. Constructs T3 and T5 are DNA duplexes with ssDNA tails of opposite polarities. These two substrates were used to elucidate the effect of ssDNA polarity in the assembly of complexes with PriA. The T3 DNA substrate consists of a 244-bp duplex region and a 3Ј-end, 69-nucleotide singlestranded region. The T5 DNA substrate has a 5Ј-ssDNA region of the same length, but the duplex size is 376 bp. The fork substrates F3 and F5 also differ in the polarity of 69-nt ssDNA. The interaction of PriA with these substrates will allow us to evaluate the role of fork orientation at the junction on the assembled complexes with PriA. F3 construct has a 673-bp duplex region, with a 69-nt gap in the nascent leading strand. The duplex length of F5 is 675 bp, with a 69-nt gap in the nascent lagging strand. These forks contain a 69-nt ssDNA arm asymmetrically placed within the DNA duplex region. Consequently, for F3, the DNA duplex regions are a 280-bp parentalstrand duplex and a 393-bp lagging-strand duplex, although in F5, the parental-strand duplex has 416-bp residues and the other duplex region is a 260-bp leading-strand duplex. The assembly of each DNA substrate was verified by contour length measurements (Fig. S1).
Interactions of SSB protein with the DNA substrates-To characterize the substrates and evaluate the accessibility of ssDNA, SSB protein was incubated with each of the substrates separately. AFM images are shown in Fig. S2 (A and B), which demonstrate that SSB binds to only one of the two ends on tailed DNA substrates. On fork DNA substrates, the position of SSB was measured from the end of the shorter arm toward the center of the blob in each DNA-SSB complex (Fig. S2E). The data shown as histograms in this figure were approximated by single-peak Gaussians. The peak values, 276 Ϯ 14 bp (S.D.) on F3 DNA and 260 Ϯ 16 bp on F5 DNA, correspond to the fork positions, which are 280 and 260 bp for the F3 and F5 constructs, respectively. Therefore, as expected, SSB binds only to the ssDNA regions of each substrate.
PriA binding to the tailed DNA substrates-When mixed with T3 and T5 DNA substrates, PriA binds to only one end of the tailed DNA substrates, regardless of the relative concentration of the protein (8:1 PriA-to-DNA molar ratio, Fig. 2, A and  B; large-scale images shown in Fig. S3, A and B). The analysis of over 500 PriA-DNA complexes showed that binding of PriA to blunt DNA ends was less than 0.5% of the observed binding events. This suggests that PriA binds poorly to either blunt ends or dsDNA. Furthermore, the binding yield of PriA on tailed DNA substrates, collected from three independent experiments, was 9.2 Ϯ 0.3% on T3 DNA and 7.9 Ϯ 0.7% on T5 DNA, pointing to a minor preference for PriA binding to the substrate with 3Ј-ssDNA tail.
PriA binding to the fork DNA substrates-To assess the interactions of PriA with fork DNA substrates, PriA was incubated with F3 and F5 fork DNAs separately and imaged (Fig. 2, C and D). The results show that protein was bound exclusively to a site embedded within the duplex region. To determine whether the binding site corresponds to the position of the fork, we ascertained the position of PriA in each complex by measuring the distance from the end of the shorter arm to the center of the blob (protein). The histograms shown in Fig. 2 (E and F) were fitted by single-peak Gaussians with a bin size of 20 bp. The peak of the histogram for PriA position on F3 was found to be centered at 295 Ϯ 20 bp (S.D.). For F5-PriA complexes, the peak was centered at 254 Ϯ 23 bp (S.D.). These peak values match the designed fork position, which supports our previous assumption that the fork provides the recognition and binding site for PriA to load onto DNA replication forks.
The yields of complexes for both substrates were measured, and the data revealed different results compared with tailed DNA substrates. The yield of PriA was 13.0 Ϯ 1.2% on F3 DNA and 8.2 Ϯ 1.3% on F5 DNA, based on the results of three independent experiments. Given a small difference in PriA binding to ssDNA tails with different polarities, the difference between F3 and F5 DNA suggests that structural features of the fork substrates must be involved.

Interactions between the SSB protein and PriA on fork DNA substrates
Studies show that SSB binds to the RecG and PriA helicases both in vivo and in vitro (23)(24)(25). Previously, we demonstrated that SSB remodeled RecG during DNA loading (33,34). To determine whether SSB modulates PriA loading, a similar analysis was performed. Here, PriA was preincubated with SSB at a 2:1 molar ratio for 10 min on ice. The mixture was added to fork DNA substrates in a molar ratio of 2:1 (protein mixture: DNA substrate), incubated for 10 min at room temperature, and imaged.

SSB loading PriA onto duplex regions of fork DNA substrates
The resulting images show that for both F3 and F5, doubleblob complexes were observed (Fig. 3, A and B). The double-

AFM of PriA-DNA complexes
blobs correspond to PriA and SSB bound to the same DNA molecule, with the larger one being SSB and the smaller being PriA, as explained below. Furthermore, in some complexes, the two blobs are located far from each other, whereas in others, the two blobs colocalize on the DNA (Fig. 3, enlarged images). These data suggest that the interaction of PriA with SSB leads to a change of PriA conformation, allowing for the protein to bind to the DNA duplex. We termed this property of SSB "remodeling," which was initially identified for SSB-mediated loading of RecG protein on the DNA fork.
To determine the identity of each blob, PriA and SSB were bound to each DNA substrate separately. These samples were imaged, and the volume values of the complexes were determined. As measured, the SSB volume was averaged to 163 Ϯ 31 nm 3 (Fig. S2). In contrast, the average volume value for PriA was 115 Ϯ 30 nm 3 (Fig. S3). Therefore, and as observed previously for RecG, the large blobs correspond to SSB and the smaller ones to the DNA helicase. By applying these quantitative analyses, the position of SSB and PriA could be determined using the same measurement approach used for PriA only. The results show that for forks F3 and F5, SSB is bound at the fork (  AFM of PriA-DNA complexes (Fig. 4). Both proteins differ in sizes and can be unambiguously identified on the AFM images. In Fig. 4, the SSB position (green squares) is set to 0 to mark the fork position. For fork F3, the negative values correspond to the parental duplex arm, and the positive values correspond to the lagging strand. PriA positioning is depicted as red dots. PriA was found to colocalize with SSB at the fork ϳ40% of the time, whereas it was found to be positioned on the parental arm and lagging strand 28 and 32% of the time, respectively. Therefore, in the presence of SSB, PriA is loaded preferentially at the fork. In addition, it can be found in the duplex regions with no clear preference for the type of flanks. In contrast, on F5, PriA was found to colocalize with SSB at the fork ϳ28% of the cases, whereas it was found to be positioned on the parental arm and leading strand with frequencies of 24 and 48%, respectively. Therefore, when the fork has a gap in the nascent lagging strand, PriA is preferentially loaded onto the leading strand arm of the fork.

SSB-PriA assembly at the fork
At the fork position, and in addition to the colocalized twoblob complexes, single-blob complexes were also observed. The sizes of these single blobs varied; the single blob could be PriA or SSB only or SSB-PriA complexes, which are of larger sizes. Volume analysis by cross-section was performed to identify the possible component of each single-blob complex. The volume distribution fitted with multipeak Gaussian for single-blob complexes in F3 DNA sample is shown in Fig. 5A. Peak 1 is centered at 137 Ϯ 37 nm 3 , which is close to the measured bound SSB volume on F3 DNA (X c ϭ 159 Ϯ 29 nm 3 in Fig. S2H). Peak 2 is approximated at 244 Ϯ 28 nm 3 , corresponding to the complexes of PriA and SSB (free PriA volume was 58 Ϯ 12 nm 3 , shown in Fig. S5A; DNA-bound SSB volume was 159 Ϯ 29 nm 3 , shown in Fig. S2H). The population of the larger blobs is ϳ24%, approximated by the area under the Gaussian.
A similar analysis was done for F5 substrates, with the data shown in Fig. 5B. The volume distribution was fitted with a single-peak Gaussian centered at 170 Ϯ 45 nm 3 . This peak value shows a negligible difference from the volume of bound SSB on F5 (168 Ϯ 27 nm 3 , shown in Fig. S2I). The p value from Kolmogorov-Smirnov test for these two sets is larger than 0.05, which indicates no interaction between SSB and PriA at the fork region of F5.

The role of the C-terminal domain of SSB in PriA loading
It is known that the C-terminal domain of SSB is required for partner binding (23,30). To determine whether the C-terminal domain of SSB is required for PriA loading, we used the SSB⌬C8 protein, which has the last eight residues or acidic tip removed. First, the binding of the SSB mutant to forks F3 and F5 was assessed. The results show that the yield of SSB⌬C8 bind-

AFM of PriA-DNA complexes
ing onto each DNA substrate was 83.7% for F3 DNA and 81.8% for F5 DNA (Fig. S4). This is within experimental error, the same as that observed for WT. Thus, as observed previously, ssDNA binding for this mutant is unaffected (29,35,36).
In mixing experiments, PriA and SSB⌬C8 were mixed as described for the WT, bound to the DNA, and imaged. The yield of double-blob complexes formed on F3 and F5 was reduced to less than 5%, which is lower than that of WT SSB and PriA (Table 1). Furthermore, analysis of the single-blob complexes revealed that these proteins are SSB⌬C8 (shown in Fig.  6). Collectively, these results show that SSB⌬C8 does not facilitate the loading of PriA onto the DNA.

Interaction between SSB and PriA in the absence of DNA
To characterize the assembly of WT SSB-PriA complexes in the absence of DNA, we analyzed the sizes of free proteins in the samples of fork DNA substrates mixed with WT SSB and PriA. The volume distribution is shown in Fig. 5C. The histogram shows three peaks based on multipeak Gaussian fitting, which are 59 Ϯ 12, 103 Ϯ 22, and 233 Ϯ 20 nm 3 . Peak 1 matches the volume of free PriA (58 Ϯ 12 nm 3 ), shown in Fig. S5A. Peak 2 is close to the volume of free SSB (91 Ϯ 21 nm 3 in Fig. S5B). According to the narrow single-peak distribution of each volume measurement in our control experiments (shown in Fig. S5, A and B), neither tetrameric SSB nor monomeric PriA aggregate with themselves. Based on this, we deter-  is centered at 137.2 Ϯ 36.5 nm 3 , and peak 2 is approximated at 243.7 Ϯ 28.1 nm 3 . The population approximated by the area under the curve is 76.1% for peak 1 and 23.9% for peak 2. B, the volume distribution for single blobs on F5 DNA. It was fitted with single-peak Gaussian with a centered peak at 169.8 Ϯ 44.6 nm 3 . C, volume distribution for free proteins in samples of fork DNA mixed with SSB and PriA. The histogram shows three peaks based on multipeak Gaussian fitting, which are 58.7 Ϯ 12.4, 102.9 Ϯ 22.3, and 233.1 Ϯ 20.4 nm 3 . The population approximated by area under curve is 59.9% for peak 1, 27.5% for peak 2, and 12.6% for peak 3.

AFM of PriA-DNA complexes
mined peak 3 to be SSB-PriA complexes. Note that peak 3 has a low probability of formation of 12.6%, evaluated by the area under Gaussian, whereas the probabilities of the other two are 59.9 and 27.5%, for peaks 1 and 2, respectively. Therefore, although the population of the SSB-PriA complexes is low, the interaction between SSB and PriA in the absence of DNA does take place. We also characterized the role of the C terminus on the interaction of SSB and PriA in the absence of DNA. The analysis was performed on the free proteins present in the protein-DNA samples characterized above. Size analysis for free proteins in the sample of fork DNA substrates mixed with SSB⌬C8 mutant and PriA is shown in Fig. 6E. The volume distribution was fitted by Gaussian with a single peak centered at 49 Ϯ 17 nm 3 , which is a departure from the multipeak distribution of free proteins in samples of fork DNA substrates with WT SSB and PriA mixture (Fig. 5C). Thus, the C terminus of SSB is needed for SSB-PriA interaction. A 2-fold difference in the volume values for free SSB⌬C8 and WT SSB can be explained by a possible dissociation of SSB⌬C8 tetramer during its preparation for AFM.

Discussion
The major function of PriA, a DNA helicase, is to restart the replication process by facilitating the loading of the replicative helicase DnaB onto the stalled replication fork. Once bound, PriA exposes ssDNA for the replicative DNA helicase DnaB by unwinding the lagging strand or by remodeling SSB-coated ssDNA (22,37). This priming process requires the binding of PriA to various DNA structures (6). This was the focus of this paper, because it remains unclear how PriA differentiates among various DNA structures and then plays the role needed for the restart of the stalled replication fork. To uncouple the DNA-binding property of PriA from its helicase activity, we performed studies in the absence of ATP. The major findings are discussed below.

The effects of ssDNA polarity and the fork structure on the interaction between PriA and DNA
On the fork DNA substrates, PriA binds to both substrates at the fork position, with a preference for the fork DNA substrate with a gap in the nascent leading strand (F3), which is an unexpected finding. According to other publications (18, 22, 37, 38), is no significant difference between these two data sets. This analysis suggests that these single-peak events belong to SSB⌬C8. D, single-peak Gaussian fitted volume distribution for single blobs on F5 DNA. The centered peak is approximated at 110.9 Ϯ 30.7 nm 3 . The Kolmogorov-Smirnov test for the volume of the single-blob complexes on F5 substrates and the bound PriA (112.9 Ϯ 23.4 nm 3 ) led to a p value of 0.57, suggesting that there is no significant difference between these data sets. Similarly, there is no significant difference (p value, 0.22) between the volume values for the single-blob complexes and the bound SSB⌬C8 volume (122.6 Ϯ 27.9 nm 3 ). However, based on the yields of protein-DNA interaction, 81.8% for SSB⌬C8 with F5 and 8.2% for PriA with F5, we evaluate the percentage of each protein in the single-blob complexes as ϳ9% for PriA and ϳ91% for SSB⌬C8, suggesting that primarily, the single-blob complexes on F5 substrate are SSB⌬C8. E, the volume distribution for free proteins in the sample of fork DNA substrates mixed with SSB⌬C8 mutant and PriA. The histogram was fitted by Gaussian with a single peak centered at 48.6 Ϯ 17 nm 3 .

AFM of PriA-DNA complexes
PriA should bind preferentially to fork F5, because F5 has a 3Ј-OH group at the fork position as the recognition site for the 3Ј-binding domain in PriA, whereas F3 does not. Our finding on PriA's preference for binding suggests the participation of the other domains in PriA-DNA binding and is in line with the previous finding that PriA binds to the arrested replication fork in a manner independent of the 3Ј terminus as well (18). In addition to the independence of polarity, the bend at the fork position may play an important role in PriA-binding activity, as demonstrated in Refs. 15 and 39. This emphasizes the essential role of the fork structure and the presence of a nascent lagging strand in the binding of PriA onto the stalled replication fork.
Our model of the role of other PriA domains in the assembly of complexes with the replication fork is supported by the findings that PriA does not show binding preference to the ssDNA of tailed DNA substrates (Fig. 2, A and B, and Table 1). Nurse et al. (39) demonstrated that PriA bound with high affinity to duplexes with 3Ј tails, whereas it did not bind to duplexes with 5Ј tails at all. However, they detected stable binding to the 3Ј extension when the ssDNA tail exceeded 12 nt, and high affinity binding resulted when the tails were in excess of 16 nt in length. In our studies, the size of the ssDNA region is 69 nt. The excess of the ssDNA region might facilitate the multidomain-binding activity, which stabilizes the interaction between PriA and DNA substrates. Note as well that PriA has over 100-fold higher binding yield to the ssDNA region compared with the blunt end on the tailed DNA substrates. This property contributes to the high efficiency of targeting at the stalled replication fork where the restart is needed.

Remodeling of PriA by SSB on fork DNA substrates
Although interactions between SSB protein and PriA have already been characterized (15,23,24,40,41), our results revealed a novel role for SSB in the interaction of PriA with fork DNA substrates. Shown in Fig. 3 (A and B), both proteins were colocalized at the fork position (inset ii), where PriA was bound to the fork, whereas SSB was bound to the rest of the ssDNA region. Interestingly, we also identified complexes, shown as inset i on both images, in which SSB and PriA are well-separated. On both substrates, SSB locations produce narrow positional distributions (histograms C and D in Fig. 3). However, the positional distributions of PriA in the SSB-PriA complexes assembled on both substrates are very broad (Fig. 3, D and E). These data are in contrast with the data obtained from PriA-DNA complexes in the absence of SSB, which has a narrow distribution on each histogram and peak positions that coincide with the location of the fork (Fig. 2, E and F). We hypothesize that after remodeling, PriA is either capable of binding to the DNA duplex and/or spontaneously translocates over DNA duplexes.
The mapping of protein positions on the fork substrates in Fig. 4 shows that, for F3, there is no clear preference of PriA to position on the flanks of the fork. However, on fork F5, which has a gap in the lagging strand, remodeled PriA showed a preference of binding to the leading strand. These data suggest that the interaction of PriA with SSB changes the helicase conformation in such a way that PriA becomes capable of binding to the DNA duplex. Note that in the absence of SSB, such a change in DNA-binding conformation of PriA has not been identified.
We previously identified a similar remodeling property of SSB for the RecG protein (34), and the translocation mobility of RecG has been demonstrated by direct visualization of RecG mobility using a time-lapse high-speed AFM approach (33). It was hypothesized that the remodeling of RecG by SSB allows RecG to translocate along the duplex strands in an ATP independent way so that the helicase can be recruited rapidly to accomplish its fork regression role (33,34). Recently, the protein-protein interaction of SSB with partner proteins has been reported for RecQ (42)(43)(44) and RecOR (45,46), suggesting that the remodeling of components of the DNA replication machinery is a common property of SSB. Remodeling of PriA by SSB in the absence of ATP increases the ability of PriA to bind onto duplex DNA, which was detected as a property of the C-terminal domain of PriA (22). Therefore, the binding and/or translocation of PriA along the DNA duplex may stimulate the loading of PriA onto the stalled replication fork in an ATPindependent way, facilitating the restart process once the ATP is available for PriA helicase activity.
Studies showed that most of the interactions between SSB and its partner protein are mediated by the C terminus of SSB (24,47). The C-terminal domain of SSB, corresponding to residues 117-178, can be subdivided into an intrinsically disordered linker (amino acids 117-170) and a highly conserved acidic tip (amino acids 171-178) (32). The linker mediates protein-protein interactions, whereas the acidic tip is required to maintain the structure of the C terminus of SSB, so that it does not bind to SSB itself (29). Thus, when the acidic tip is mutated or deleted, the C terminus binds to the DNA-binding domain of SSB, thereby eliminating binding to target proteins. Consequently, SSB-partner interactions are lost. Experiments with SSB⌬C8, in which the acidic tip of SSB was removed, showed that the yield for double-blob complexes dropped significantly. These findings are in line with the effect of the C terminus on the remodeling of RecG (34), suggesting that the C terminus plays an important role in SSB remodeling. Intriguingly, the volume of SSB⌬C8 is 3-fold smaller than that of WT SSB (Fig. S5). This result is consistent with a model proposing that when the acidic tip is mutated, SSB C termini collapse back onto the tetramer, occlude the OB-fold, and prevent the linker form binding to interactome partners (29).

Protein-protein interactions in the colocalized SSB-PriA complexes
PriA interacts with SSB both in the absence and presence of fork DNA substrates. According to published data, the stimulation effect/co-localization activity of SSB-PriA interaction requires an excess of SSB over the helicase (15,23,24,40). In our experiments (Fig. 3, A and B), colocalized PriA-SSB complexes on DNA were detected at a molar SSB-to-PriA ratio of 1:2. Importantly, the concentration of PriA was 5 nM. The volume analysis of colocalized SSB and PriA complexes (Fig. 5) was clearly seen for the F3 substrate in peak 2. The yield of such complexes was 24%, compared with the single-blob complexes (76%) that correspond to complexes with SSB binding only. Considering that the single-blob complexes count for 78% of AFM of PriA-DNA complexes binding events in the mixture sample (with the rest of events being 7% of free DNA substrates and 15% of double-blob complexes), the overall binding yield of PriA-SSB on F3 DNA should be 0.24 ϫ 0.78 ϭ 0.187, which is 18.7%. No such clearly identified peak appears for the F5 substrate (Fig. 5B), suggesting that the colocalization depends on the substrate type. In our case, colocalization is clearly higher for the F3 substrate, which is shown to be a better substrate for binding PriA only as well (Table 1). In this way, the PriA-SSB interaction increased the binding of PriA to fork DNA substrates and improved the selectivity of PriA to a more favorable substrate, which is F3 substrate in our study. This suggests that the SSB stimulation effect, as it is known on PriA helicase activity, also plays a role in the binding of PriA onto DNA substrates.
In addition to the protein-protein interaction found in protein-DNA complexes, the volume analysis of free protein (Fig. 5C) revealed a minor peak (peak 3), which has a larger volume than each protein itself, corresponding to the volume of the complex formed from SSB-PriA assembly. So, the SSB-PriA interaction is independent of the presence of DNA or ATP. This property helps SSB locate or direct PriA to the place it is needed in a more efficient way.

Conclusions
Our AFM studies revealed several novel properties of interactions between PriA and stalled DNA replication forks, with or without SSB. In the absence of ATP, we observed that PriA binds preferentially to the forked DNA with a gap in the nascent leading strand, compared with the other forked DNA substrate, which has a gap in the nascent lagging strand. Because PriA showed no clear preference toward the polarity of ssDNA in tailed DNA substrates, it is the fork structure that plays a more essential role in PriA binding. The interactions between SSB and PriA revealed a remodeling of PriA by SSB, which loaded PriA onto the duplex DNA. This property of SSB can be attributed to its C-terminal segment. This ATP-independent loading of PriA and its translocation over the DNA duplex can facilitate the recruiting of PriA to the stalled replication fork.
SSB protein was purified from strain K12⌬H1⌬trp, as described previously (27,48). The concentration of the purified protein was determined at 280 nm using ⑀ ϭ 30,000 M Ϫ1 cm Ϫ1 (35). Similar to the WT, the his-SSB⌬C8 mutant protein was purified using nickel column chromatography, followed by step elution from ssDNA-cellulose (29).

DNA substrates preparation
Each tailed DNA substrate (T3 or T5) was assembled from a duplex-DNA segment and a tail-DNA segment. The fork DNA substrates (F3 and F5) were assembled from two duplex-DNA segments and a core fork segment as described previously (33,34). The two duplex-DNA segments (224-bp segment and 356-bp segment) were precisely the same as the segments we had in Refs. 33 and 34. To assemble the tail-DNA segment for T3, the ssDNA oligonucleotides O42 (5Ј-TCATGACTCGCT-GCGCAAGGCTAACAGCATCACACACATTAACAATTC-TAACATCTG-3Ј) and O43 (5Ј-CCTTGCGCAGCGAGTCA-3Ј) were phosphorylated, mixed in an equal molar ratio, and then annealed by heating to 95°C in the hot boiled water and then cooling down slowly to room temperature. The tail-DNA segment was then ligated with the 224-bp duplex-DNA segment in the molar ratio of 1:1 to assemble the T3 DNA substrate. Similar to the assembly of T3 DNA substrate, the tail-DNA segment for T5 was annealed from the 5Ј-phosphorylated oligonucleotides O36 (5Ј-TACGTGTAGGAATTATATTA-AAGAGAAAGTGAAACCCAAAGAATGAAAAAGAAGA-TGTTAGAATTGTAAGCGGTATCAGCTCACTCATA-3Ј) and O37 (5Ј-GCTTATGAGTGAGCTGATACCGC-3Ј) in the equal molar ratio and then was ligated together with the 356-bp duplex-DNA segment in 1:1 molar ratio to obtain the T5 DNA substrate. The core fork segment of F3 was assembled by annealing the 5Ј-phosphorylated oligonucleotides in the same molar ratio (O30, 5Ј-TCATCTGCGTATTGGGCGCTCTTC-CGCTTCCTATCT-3Ј; O31, 5Ј-TCGTTCGGCTGCGGCGA-GCGGTATCAGCTCACTCATA-3Ј; O32, 5Ј-GCTTATGAG-TGAGCTGATACCGCTCGCCGCAGCCGAACGACCTTG-CGCAGCGAGTCAGTGAGATAGGAAGCGGAAGAGCG-CCCAATACGCAGA-3Ј; and O33, 5Ј-CACTGACTCGCTG-CGCAAGGCTAACAGCATCACACACATTAACAATTCT-AACATCTGGGTTTTCATTCTTTGGGTTTCACTTTCT-CCAC-3Ј), while the core fork segment of F5 was annealed from phosphorylated O30, O31, O32, and O34 (O34, 5Ј-CTA-ACAGCATCACACACATTAACAATTCTAACATCTGGG-TTTTCATTCTTTGGGTTTCACTTTCTCCACCACTGAC-TCGCTGCGCAAGG-3Ј), in the same molar ratio. The two duplexes and core fork segment were ligated together at the molar ratio of 1:1:1 at 16°C overnight. The final products were purified with HPLC using a TSKgel DNA-STAT column. All oligonucleotides were bought from Integrated DNA Technologies, Inc. (Coralville, IA).

Protein-DNA complex and AFM sample preparation
PriA-DNA complex was prepared by mixing the PriA monomer (molar concentration, 100 nM) with DNA substrates (molar concentration, 45 nM) in a molar ratio of 8:1. The mixture was incubated in 10 l of binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 5 mM MgCl 2 , 1 mM DTT) for 10 min at room temperature. After incubation, the complex was then diluted to achieve lower DNA concentration (ϳ1 nM), which was ready for deposition onto the APS functionalized mica.

AFM of PriA-DNA complexes
The SSB-PriA-DNA complex was prepared by mixing the proteins first: the SSB tetramer (molar concentration, 50 nM) was mixed with PriA monomer in a molar ratio of 1:2, and the mixture was kept on ice for 30 min before use. The mixture of proteins was added to fork DNA substrates in a 1:2:4 (DNA substrates:SSB:PriA) molar ratio and then incubated in 10 l of binding buffer for 10 min at room temperature. After incubation, the complex was then diluted to achieve lower DNA concentration (ϳ1 nM) for AFM imaging using APS functionalized mica procedure.
1-(3-Aminopropyl) silatrane (APS) functionalized mica was used as the AFM substrate for all experiments (49). Fresh cleaved mica was incubated in 4 ml of APS (167 M) in a cuvette for 30 min and then was rinsed with double-distilled H 2 O thoroughly as described before (33,34). 10 l of the sample were deposited onto the APS functionalized mica for 2 min. After 2-min incubation, the mica was rinsed with double-distilled H 2 O and then was dried with a gentle argon-gas flow.

AFM imaging and data analysis
The images were acquired using tapping mode in the air on a MultiMode 8, Nanoscope V system (Bruker, Santa Barbara, CA) using TESPA probes (320-kHz nominal frequency and a 42 N/m spring constant) from the same vendor. The dry sample AFM images were analyzed using the FemtoScan online software package (Advanced Technologies Center, Moscow, Russia). The positions of each protein were measured from the end of the shorter arm on the DNA substrates toward the center of the protein. The contour lengths of the DNA were then continuously measured from the center of the protein toward the other end of the DNA substrate. The yield of protein-DNA complexes was calculated from the number of complexes dividing by the total number of DNA molecules. The histograms were approximated by Gaussian distribution, and the mean values and errors (S.D.) were calculated by using Origin software (OriginLab Corporation, Northampton, MA). The protein height and volume were measured with the cross-section option. The volume was calculated by applying the measured data to the formula: V ϭ 3.14 ϫ H/6 ϫ (0.75 ϫ D 1 ϫ D 2 ϩ H 2 ), in which D 1 and D 2 are the diameters of the protein, which were measured twice, and H is the highest height of the two measurements of the protein (34,50,51).