Direct analysis of Holliday junction resolving enzyme in a DNA origami nanostructure

Holliday junction (HJ) resolution is a fundamental step for completion of homologous recombination. HJ resolving enzymes (resolvases) distort the junction structure upon binding and prior cleavage, raising the possibility that the reactivity of the enzyme can be affected by a particular geometry and topology at the junction. Here, we employed a DNA origami nano-scaffold in which each arm of a HJ was tethered through the base-pair hybridization, allowing us to make the junction core either flexible or inflexible by adjusting the length of the DNA arms. Both flexible and inflexible junctions bound to Bacillus subtilis RecU HJ resolvase, while only the flexible junction was efficiently resolved into two duplexes by this enzyme. This result indicates the importance of the structural malleability of the junction core for the reaction to proceed. Moreover, cleavage preferences of RecU-mediated reaction were addressed by analyzing morphology of the reaction products.


INTRODUCTION
A four-way deoxyribonucleic acid (DNA) junction [Holliday junction (HJ)] is a central intermediate of homologous recombination that constitutes a ubiquitous pathway in the repair of double-stranded DNA breaks (1) and the restart of stalled replication forks (2,3). The junction is ultimately resolved into two nicked-duplex species by the action of junction-resolving enzymes (resolvases) (4,5). These enzymes exhibit high structural selectivity for HJ DNA. However, paradoxically, resolvases impose significant torsional distortion on the junctions that they recognize (6).

One of the remarkable examples is the Bacillus subtilis
RecU resolvase (5), where the junction adopts an open structure after RecU binding (7). RecU acts as a homodimer and binds the four-way DNA junction with a dissociation constant of ∼0.5 nM (7). Although RecU cleaves on either side of the junction at a specific sequence (5 -G / T G↓C A / C -3 ), recognition of and binding to the junction occur based on the structure rather than sequence preference (7). Crystallographic study of dimeric RecU revealed a mushroomlike morphology, with a cap and stalk (8,9), the stalk region of which is essential for HJ recognition (10) ( Figure  1a). Biochemical studies suggest that the central hole of the HJ accommodates the stalk region, resulting two arms of the junction positioned near the active sites of the two RecU monomers (8) (Figure 1b). Thus, recognition of the junction and nicking of the DNA would be structurally connected by the distortion of the HJ. This hypothesis is supported by the characterization of RecU mutants that poorly distort the junction and fail to cleave the HJ (10). It is therefore of considerable interest to explore the role of structural malleability of the junction on the reactivity of the enzyme.
Here, we created two different framed HJ structures, flexible and inflexible, by employing a DNA origami nanoscaffold (DNA frame) described in the previous reports (11)(12)(13)(14)(15). The DNA frame (80 nm × 90 nm) has a vacant rectangle area (40 nm × 40 nm), in which four connection sites were introduced for hybridization of the helices of a fourway DNA junction (Figure 1c). Two antiparallel four-way junctions were introduced into the DNA frame: one with arms of 36-nucleotides (nt) and another with 41-nt arms (each assembled structure was termed as DF-HJ 36 and DF-HJ 41 , respectively). The 36-nt junction just fitted as a tensed state in the frame, while the 41-nt junction is in a more relaxed state that can allow the flexible motion of the junction core (Figure 1d). Using these fabricated nanostructures, the effect of architectural malleability of the junction on its recognition and resolution by RecU was examined. A junction-resolving reaction was also successfully monitored using high-speed atomic force microscopy (HS-AFM).

Design of DNA frame
DNA frame was designed as previously reported (11). The DNA sequence design of the DNA frame structure followed the rules of the DNA origami method. The sequence of the M13mp18 was used, and the staple strands (most of them are 32-nt) were assigned for the formation of the designed DNA frame. The DNA frame lacks the right bottom corner for the identification of the orientation of the frame (orientation marker). The sequences and positions of the staple strands are listed in the supporting information of ref (11).

DNA frame formation and Integration of the Holliday Junction structures
The DNA frame and the Holliday Junction structure were assembled in a 25 l solution containing 10 nM M13mp18 single-stranded DNA (New England Biolabs), 50 nM staple strands (226 strands), 20 mM Tris-HCl (pH 7.6), 1 mM EDTA, and 10 mM MgCl 2 . The mixture was annealed from 85 to 15 • C at a rate of −1.0 • C/min. The four-way junctions, which contain a 13-bp homologous core, were constructed by annealing four strands HJ 36 -a, HJ 36 -b, HJ 36 -c and HJ 36d (for 36-nt junction) or HJ 41 -a, HJ 41 -b, HJ 41 -c and HJ 41 -d (for 41-nt junction). The sequences of the oligonucleotides used for constructing four-way junctions are seen in Supplementary Figure S1. The HJ structures were subsequently incorporated into the DNA frame by annealing the mixture from 40 to 15 • C at a rate of −0.5 • C/min using thermal cycler. The sample was purified by gel-filtration chromatography (GE sephacryl-400, GE Healthcare Japan, Tokyo, Japan) to remove excess amounts of the staple strands and the HJ structures (Supplementary Figure S2).

RecU protein purification
Escherichia. coli BL21(DE3) [pLysS] cells containing pCB210-borne recU gene (7) were resuspended in buffer A (50 mM Tris-HCl, pH 7.5/1 mM DTT/5% glycerol) containing 100 mM NaCl and lysed with a French press. RecU was purified as previously described (7) to >99% purity, as judged by 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and matrix-assisted laser desorption ionization time-of-flight analysis. The RecU concentration was determined by using a 24 900 M −1 cm −1 molar extinction coefficient and is expressed as mol of protein monomers. Tris-HCl (pH 7.6), 1 mM EDTA, 50 mM NaCl and 10 mM MgCl 2 ] for 30 min at 30 • C. For AFM imaging, 2 l of sample was deposited onto a freshly cleaved mica disc immediately after the above described incubation. After 1 min, the sample was rinsed with 10 l of the Tris-buffer A (for binding assay) or B (for cleavage assay) and imaged in the same buffer. The appearance frequency of each distinct structure was calculated from the obtained images. The broken structures were not taken into account for the yield calculations.

Time-lapse HS-AFM imaging of individual reactions
The resolution reactions were imaged as follows: 30 nM of RecU was pre-incubated with 10 nM of the DNA frame carrying 41-nt junction at 30 • C for 5 min in a Tris-buffer A. After the incubation, 2 l of the sample was immediately deposited onto a freshly cleaved mica disc. After 30 s incubation, the sample was rinsed with 10 l of the Tris-buffer B and imaged in the same buffer.

AFM imaging
The imaging was performed using prototype high-speed AFMs (Olympus Corporation, Tokyo, Japan) (16,17). The sample was imaged in the imaging buffer solution at ambient temperature using small cantilevers with dimensions (L × W × H) of 10 m × 2 m × 0.1 m (Olympus Corporation, Tokyo, Japan). These cantilevers have a spring constant of 0.1-0.2 N/m with a resonant frequency of 400-1000 kHz in water. A sharp probe was deposited on each cantilevers using electron beam deposition by Nanotools (Munich, Germany). The 320 × 240 pixels images were obtained at a scan rate of 0.2-0.5 frame/s.

Preparation and analysis of the framed Holliday Junctions
AFM images of the assembled structures revealed a clear X-shaped junction in the DNA frame (Figure 2a-d). The flexibility of the junction core was evaluated by measuring the angle between arms (Figure 2e). In the case of the 36-nt junction, frequency distribution of the angle θ AB and angle θ BD between arms showed a sharp peak at 120 ± 4 • and 61 ± 4 • , respectively (Figure 2f and g). These values agreed well with expected angles when the junction is attached to the frame. In the case of the 41-nt junction, the frequency distribution of these angles showed a much broader peak at 81 ± 19 • and 99 ± 18 • (Figure 2h and i). Note that the values of standard deviation were about 5-times larger than those for the 36-nt junction, allowing us to assume that the 41-nt junction in the DNA frame was flexible enough to change the junction structure, unlike the 36-nt junction. Time-lapse AFM imaging of DF-HJ 41 also showed this flexible feature (Figure 3a, See also Supplementary Movie S1). It is noted that the DNA frame was stably attached on the mica surface, while an inside HJ was mobile enough on the surface. Figure 3b shows the variation with time of θ AB and θ BD of the DF-HJ 41 , together with histograms of the measured angles. The histograms provided mean values of θ AB = 84 ± 12 • and θ BD = 98 ± 18 • which are similar to the distribution in Figure 2h and i, showing the flexible motion of each arm with the extent of ±12-18 • around the junction core. This type of flexible motion of the framed HJ was restricted when its arms were shortened to 36-nt ( Supplementary Figure S3).

Binding of RecU to the framed Holliday Junctions
The binding of RecU to these HJs incorporated into the DNA frame was examined by AFM. RecU requires high concentrations of Mg 2+ (10-15 mM) to resolve the junction (7). Therefore, the binding reaction was performed in the presence of a lower concentration of MgCl 2 (3 mM) to avoid the cleavage. Assembled DF-HJs were incubated with RecU at 30 • C for 30 min, and then the mixture was imaged (Figure 4a and b). The height of the RecU bound junction was about 4.9 nm, which was ∼3 nm higher than that of the center of the junction in the absence of protein (Supplementary Figure S4). The enzyme was seen to specifically recognize and bind to the four-way junction. The binding efficiency of the 36-nt junction and the 41-nt junction under this condi- Appearance of X-shape, Type I, Type II and RecU bound complex were counted. The data were collected from three independent measurements.  Table  S1). Interestingly, even though HJ was tense and inflexible in the DF-HJ 36 , 21% of them were bound with RecU, implying the capability of RecU of distorting this junction to allow its binding.

Reactivity of RecU with flexible/inflexible Holliday Junctions
To examine whether differences in core flexibility affect the reaction efficiency, each DF-HJ was incubated with RecU in the presence of 10 mM of MgCl 2 for 30 min, and imaged in the same buffer (Figure 4c and d). Note that both junctions contain the same 13-bp homologous junction core, which carries two pairs of symmetric cleavage sites (Figure 4e, blue and red arrows). Therefore, it is expected that the nicking reaction will result in two possible structural patterns: two separate parallel DNA strands (Type I) or two separate double-looped DNA structures (Type II) (Figure 4e and Supplementary Figure S5). In the case of DF-HJ 36 , incubation with RecU resulted in ∼39% of the HJ resolved (Type I: 24%, Type II: 15%), with the protein-unbound DNA frames retaining their X-shape structure (50%) (Figure 4f and Supplementary Table S1). On the other hand, when RecU was incubated with DF-HJ 41 , 60% of the HJ was resolved (Type I: 36%, Type II: 24%) and only 30% retained the X-shape structure (Figure 4g and Supplementary Table S1). HS-AFM imaging of the junction-resolving reaction in the individual DF-HJ 41 directly provided the evidence for both reaction patterns. The sample was pre-incubated with RecU in the buffer containing 3 mM MgCl 2 and then imaged in the buffer containing 10 mM of MgCl 2 . Figure 5 shows the snapshot images of the reaction (see also Supplementary Movie S2). At the initial frame of Figure 5a, RecU protein was imaged bound to the center of the junction core, which is located at around 40-bp from each connection point. The protein dissociated at 4 s, and two separate parallel double stranded DNAs were observed in the DNA frame. The other type of the reaction, production of the double-loop was also successfully visualized (Figure 5b, see also Supplementary Movie S3). The complex was seen until 2 s. At 4 s, the protein was seen to dissociate leaving double-looped structure in the DNA frame. These results show that the 41nt HJ is malleable enough to allow not only the binding but also the nicking reaction, even though DNA frames were placed on the mica surface. It is noteworthy that type I configurations (due to cleavage on the exchanging strands) were Nucleic Acids Research, 2014, Vol. 42, No. 11 7427 more frequently observed as reaction products in our DNA frame system (Figure 4f, g and Supplementary Table S1). Previous in vitro studies showed that RecU has no apparent preferred orientation for cleavage (7). However, on the other hand, RecU resolves HJs with non-crossover preference in vivo (18) by symmetrically cleaving continuous strands (19). Our results obtained with framed HJs strongly suggest that DNA topology determines the in situ outcome of the reaction.
The results obtained here also show that the DF-HJ 36 was not as effectively resolved as the DF-HJ 41 , and the relaxed junction is a better substrate for RecU than the tensed HJ structure, although the sequence of the junction core was the same in both the DF-HJ 36 and DF-HJ 41 junctions. Considering that both structures can be bound by RecU, the reactivity seems to be affected by whether the junction can accommodate the subsequent conformational changes which are required for the DNA cleavage. The tensed junction in DF-HJ 36 might be less capable of adopting this requisite changes, resulting in the poor resolution of the junction. In contrast, the relaxed four-way junction in DF-HJ 41 is malleable enough to allow RecU to complete the reaction. A four-way junction is a structurally polymorphic system that can adopt various conformers (20)(21)(22). Structural dynamics of the four-way junctions have been investigated by both biochemical (20,23) and single molecule analyses (22,24). However, there has been incomplete information on how those dynamic features of the HJ contribute to the reactivity of resolvases. DNA origami-based single-molecule approaches illustrated here could provide previously unattainable clues to the interaction/working mechanisms of junction binding enzymes in the context of HJ topology and flexibility.

CONCLUSION
In conclusion, we here report the direct single-molecule analysis of the HJ-resolving reaction mediated by RecU. The flexibility of the substrate HJ was regulated in the defined nanoscaffold, a DNA frame, which was constructed by the DNA origami method. This method allowed us to examine the role of junction malleability in the reaction. The results obtained demonstrated the strong capability of RecU to distort the junction and the importance of junction malleability especially in the nicking reaction of the junction. The presented method should be applicable to investigate the mechanisms used by other junction binding enzymes which are involved in branch migration and junction resolving reaction.

SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.

ACKNOWLEDGMENT
We acknowledge Dr John B. Rafferty from University of Sheffield (UK) for his help in preparing the RecU-HJ model.