DNA Origami Catenanes Templated by Gold Nanoparticles

Mechanically interlocked molecules have marked a breakthrough in the field of topological chemistry and boosted the vigorous development of molecular machinery. As an archetypal example of the interlocked molecules, catenanes comprise macrocycles that are threaded through one another like links in a chain. Inspired by the transition metal-templated approach of catenanes synthesis, the hierarchical assembly of DNA origami catenanes templated by gold nanoparticles is demonstrated in this work. DNA origami catenanes, which contain two, three or four interlocked rings are successfully created. In particular, the origami rings within the individual catenanes can be set free with respect to one another by releasing the interconnecting gold nanoparticles. This work will set the basis for rich progress toward DNA-based molecular architectures with unique structural programmability and well-defined topology.


Introduction
Mechanically interlocked molecular architectures are molecules, which are connected as a consequence of their chemical topology. Over the last decades, a variety of interlocked molecules have been created, ranging from Borromean rings [1][2][3] , knots [4,5] , catenanes [6] , rotaxanes [7] , pretzelanes [8,9] to Solomon-links [10] and daisy-chains [11,12] . A catenane is a molecule with two or more topologically linked macrocycles [13,14] , whereas a rotaxane consists of a dumbbell-shaped axis threaded through at least one macrocycle [15] . The first catenane synthesis was realized by Edel Wasserman in 1960 at the Bell Laboratories [13,14] . Later in 1967, Harrison et al. reported the first synthesis of rotaxanes. [15] For quite some time, low efficiencies and complicated procedures had been a bottleneck for the synthesis of interlocked molecules. The situation was changed after Jean-Pierre Sauvage introduced the transition metal-templated approach in 1983. This approach utilizes a transition metal cation to arrange two bidentate ligands in a tetrahedral array as a prelude to cyclization and catenane formation. [16,17] Figure 1 shows the two representative routes. In strategy A "gathering and threading", a crescent-shaped molecule is threaded through a ring-shaped molecule via a transition metal ion and subsequently closed by another crescent-shaped molecule in a single cyclization reaction. In strategy B "entwining", two crescent-shaped molecules are linked together via a transition metal ion, followed by a double cyclization reaction to close them at once. In each case, the transition metal ion can be removed after serving its purpose, enabling the free relative movements of the two rings.

Figure 1 | Transition metal-templated synthesis of catenanes.
In strategy A "gathering and threading", a crescent-shaped molecule and a pre-formed ring molecule are linked via a transition metal ion and subsequently closed by another crescent-shaped molecule. In strategy B "entwining", two crescent-shaped molecules are linked together via a transition metal ion, followed by a double cyclization reaction. Removal of the transition metal ion allows for the free relative movements of the two rings.
In 2010, DNA rotaxanes were reported. [51] Subsequently, chemically more rigid [52] , lightswitchable [53] and Daisy-chain rotaxanes [54] , DNA-nanoparticle rotaxanes [55] as well as catalytically active rotaxanes [56] were demonstrated. Despite being topologically well-defined, the as-fabricated DNA catenanes and rotaxanes were in general mechanically rather flexible and floppy, because they were made of single-or double-stranded DNA. In contrast, DNA origami-based interlocked architectures possess much better structural rigidity. The first attempt to realize such catenanes was carried out by Yan et al., who followed an innovative scheme of molecular kirigami. [57] In 2016, Simmel et al. reported switchable DNA origami rotaxanes with long-range on-axis motion. [58] In this work, we experimentally demonstrate the hierarchical assembly of DNA origami catenanes templated by gold nanoparticles (AuNPs), taking inspirations from the transition metal-templated synthesis of catenanes developed by Jean-Pierre Sauvage. DNA origami catenanes, which contain two, three or four interlocked rings are successfully created. In particular, the origami rings within the individual catenanes can be set free with respect to one another by releasing the interconnecting AuNPs through toehold-mediated strand displacement reactions. Figure 2 shows the schematic of the DNA origami ring formation. A long single-stranded DNA scaffold (M13) is folded by ~210 short staple strands through hybridization in a self-assembly process to form an origami monomer, i.e., a quarter of a ring (see Figures 2a, S1, and S2). Each monomer contains 24 curved DNA helices bundled in a honeycomb lattice with two distinct junction regions at its ends. In total, four different sets of staple strands are utilized in order to differentiate the junction regions (I, green; IIa, blue; IIb, red; III, orange) for achieving a controlled connectivity as shown in Figure 2b (see also Figure S3). An origami ring with an inner diameter of ~120 nm can be created by linking four monomers. More specifically, junction I consists of 10 head-to-head connector strands with 3-base long sticky ends. An AuNP binding site is positioned in the region of junction I (see Figure 2b). At the binding site, there are 4 capture strands of the same sequence extended from each monomer to assemble one AuNP (15 nm in diameter; for design details see Figures S4, and S5). Junction III is designed similar to Junction I, but it comprises 12 head-to-head connector strands and lacks the AuNP binding site. Junctions IIa and IIb form a pair. Junction IIa consists of 11 connector strands with 9-base long sticky ends, whereas junction IIb does not contain sticky ends (for design details, see Figures S6 and S7). Such a design scheme ensures the precise control of the origami ring connection as well as the bound AuNP number. Importantly, it also inhibits the generation of undesired side-products. (a) Curved DNA origami monomer is formed by folding a M13 scaffold and staple strands through a hierarchical assembly process. The monomer has two different junction regions at its ends. (b) Four monomers are linked together to form an origami ring with a diameter of 120 nm by DNA hybridization. Four different sets of staple strands are used to differentiate the junction regions (I, green; IIa, blue; IIb, red; III, orange) for achieving a controlled connectivity. Capture strands (green) extended from junctions I are used to anchor a AuNP.

Assembly of the DNA-[2]-catenanes
In resemblance to the transition metal-templated synthesis of catenanes [16] , two different assembly strategies, A ("gathering and threading") and B ("entwining") are explored to create DNA-[2]-catenanes, respectively. As shown in Figure 3a, the assembly steps for strategy A are as follows. More specifically, the gel images in Figure 3b show weak bands (see the black-dashed frames) for the 1.5-ring complexes (iv) and the DNA-[2]-catenanes (v). In contrast, the gel images exhibit clear bands for the entwined half-rings with AuNPs (vi) and the DNA- [2]-catenanes (v) as shown in Figure 3c. It is noteworthy that for each strategy the step right before the DNA-[2]catenane formation is particularly crucial to ensure a successful assembly. As presented by the TEM images in Figure 3b, the intermediate structures show relatively good yields (i, ~62%; ii, ~54%), whereas the yield of the 1.5-ring complexes (iv, ~24%) is quite low. This is likely because threading a half-ring through a complete ring is sterically hindered. As a result, creation of the DNA-[2]-catenanes by further cyclization of the 1.5-ring complex with another half-ring becomes challenging. On the other hand, the TEM image of the interlocked half-ring complexes in Figure 3c reveals a good yield (vi, ~50%), which leads to the successful assembly of the DNA-[2]-catenanes (v, ~23%). When taking all the assembly steps into account, the overall yields of strategies A and B are 4.8% and 23%, respectively. Strategy B achieves a higher yield than strategy A also due to the associated fewer assembly and purification steps (see Figure 3a). The enlarged TEM images of the interlocked half-ring complex and the DNA- Figure 3c clearly shows the presence of the interconnecting AuNP between the two origami components in the individual structure.

Unlocking the DNA-[2]-catenanes
To achieve the unlocked state, in which the two origami rings within a DNA-[2]-catenane can move freely relative to one another, the interconnecting AuNP is released from the structure by toehold-mediated strand displacement reactions as shown in Figure 4a. Each capture strand extended from the origami contains two functional segments (see Figure   4b). One (red) is a 21-base segment for binding the AuNP and the other (blue) is a 9-base toehold segment. The removal strand completely hybridizes with the capture strand, releasing the interconnecting AuNP from the DNA-[2]-catenane. Figure 4c presents the TEM image of the DNA- [2]-catenanes after the dissociation of the AuNPs. The enlarged TEM image reveals the absence of the AuNP in between the two origami rings. When taking all the assembly steps of strategy B into account, an overall yield of 20% of the DNA- [2]-catenanes after the removal of the AuNPs has been achieved.

Assembly of the DNA-[3]-catenanes and DNA-[4]-catenanes
Taking a further step, we demonstrate the assembly of the DNA-

Conclusion
We have demonstrated a AuNP-templated approach to hierarchically assemble DNA origami AuNPs could be assembled on the origami rings so that the removal of the interconnecting AuNPs could be optically monitored in real time as a result of the plasmonic coupling changes.
Also, through smart designs controlled ring rotation within catenanes could be envisioned for the creation of DNA origami-based nanomachines and motors. This work will enrich the toolbox of DNA-based functional nanodevices and stride a step further towards advanced DNA architectures with programmable and well-controlled topologies. To assemble the origami half-rings, the staple strands were mixed with the p7560 scaffolds in a 1 × TE-Mg 2+ buffer (40 mM Tris, 2 mM EDTA, 20 mM MgCl2, pH 8). The molar stoichiometric ratio between the core mixture staples and the scaffolds was 10:1, the ratio between the connector strand staples and the scaffolds was 20:1 and the ratio between the AuNP capture strands and the scaffolds was 40:1. The final concentration of the scaffolds was adjusted to 20 nM. The assembly mixture was then annealed in a thermal cycler at 65 °C for 10 min and cooled with a linear ramp from 60 °C to 40 °C in 21 hours. The temperature was set to 25 °C in a final step to store the assembled origami half-rings.

Supporting Information
To assemble the origami half-rings incapable of AuNP binding, the core mixture, connector strands III, and connector strands IIa or IIb, respectively, were mixed. To assemble the origami half-rings capable of AuNP binding, the core mixture for AuNP-binding, connector strands I, AuNP capture strands, and connector strands IIa or IIb, respectively, were combined.
The DNA origami rings were assembled from purified half-rings. The half-rings modified with connector strands IIa and IIb, respectively, were mixed and assembled in a thermal cycler with a linear ramp from 40 °C to 25 °C (10 min/°C) for 10 cycles.  Polysciences, Inc. Data processing was performed using the Fiji for ImageJ software. [60] II    The TEM images show that the yield of the AuNPmodified origami half-rings is higher when using the extended anchoring segments. Almost all half-rings are modified with AuNPs. When using shorter anchoring segments, half-rings are only partly modified with AuNPs. The short anchoring segment may lead to dissociation of the AuNP capture strands due to lower binding forces resulting in half-rings without the AuNP capture strands and therefore unable to bind the AuNPs. Yields are shown in Table S1. The TEM grid of the 8 AuNP capture strand sample shows a dense distribution of correct assembled interlocked half-rings, whereas the 2 AuNP capture strand sample only reveals a significantly lower density of correct assembled structures. Using only two AuNP capture strands, incorrect assembly or dissociation can lead to half-rings unable to bind AuNPs. The extended number of AuNP capture strands ensures that every half-ring contains AuNP capture strands. Moreover, the higher binding forces may guarantee more stable interlocked half-ring structures. Yields are shown in Table S1. The gel image reveals that many side products occur (red-dashed box) when using the short anchoring segments. Short anchoring segments accompanied by lower binding forces may lead to dissociation of connector strands what enables unspecific binding and therefore the formation of side products. The characteristic structures can be observed in the TEM image. Using extended anchoring segments prevents the connector strands from dissociation due to higher binding strength. This thus inhibits the formation of the side products and only give rise to the correctly assembled origami structures. Yields are shown in Table S1. after purification (scale bar: 100 nm). The gel and TEM images show that the 9-base long sticky ends lead to the highest yield of the origami rings. Almost all half-rings were assembled to rings, whereas many half-rings are not able to form rings when using 2-or 4-base long overhangs. When using short overhangs, the binding forces may be too low to generate stable origami rings. Yields are shown in Table S1.  Table S1.  Table S1.  Table S1.

Figure S11 | Influence of the assembly time and temperature for the interlocked half-rings formation.
Samples: Origami half-rings to AuNPs mixture in a ratio of 1 to 0.5. Half-rings contain connector strands I and AuNP capture strands. The assembly times and temperatures in the different lanes are described as follows: (1) Assembly at room temperature for 24 hours.
(2) Four-step annealing program at 34 °C/3h, 30 °C/3h, 28 °C/3h and 26 °C/3h. (3) Annealing from 35 °C to 33 °C with a temperature decrease of 2h/°C followed by a temperature interval from 32 °C to 28 °C with a decrease of 4h/°C. The gel images show that 3 leads to the highest yield of the interlocked origami half-rings, indicated by the most intense interlocked half-ring band (green-dashed box). 2 shows a similar result, but the yield of 1 is significantly lower. Unreacted free half-ring (red-dashed box) is highlighted accordingly. The initial high annealing temperature results in higher meeting probabilities of AuNPs and AuNP-binding sites. Lowering the temperature then leads to the binding process through DNA hybridization. Yields are shown in Table S1.  Table S1 | Yield chart of the assembly steps. For Figure S4, a statistical analysis was done by counting the correct and false structures on the TEM grid. The yields of Figures S5 to S11 were determined by gel analysis using the ImageJ and Magicplot (©2016, Magicplot Systems, LLC) software. Analysis was done by plotting the relative density of the gel lanes. Each peak of the plot represents a DNA origami band in the gel. After peak fitting followed by calculation of the peak integrals, the product yield was determined. Figure S4 AuNP