Up till now, DNA single strands have been generally applied to produce topological macrocycle units or catenated systems composed of 7 rings  at most. Artificial DNA networks always appear in the form of DNA tiles or origami, and no DNA catenane has been used to produce topologically interlocked network structure. This summer, enlightened by the natural kDNA network and previous studies towards DNA catenanes, we intended to build a novel ssDNA catenane: DNA hauberk with controlled topology (the linking number, Lk) (see figure 1).
Figure 1. DNA hauberk
DNA hauberk has some unique features compared to DNA tiles. First, the DNA hauberk is more stable. Excessively high temperature and improper concentration of ions can cause DNA double helix to unwind, splitting DNA tiles apart, while catenanes are still topologically linked together based on their mechanical bonds. Second, units of DNA hauberk are joined together one by one unlike the single-step mixing and annealing when constructing DNA tiles. Since the shape and size of products are determined by the concentrations of raw materials, the stepwise reaction allows us to control the addition of reagents. Therefore, we can control the properties of products more accurately and avoid unnecessary wastes.
However, it is not easy to forge the DNA hauberk. Before synthesizing the complex catenane, we designed a series of experiments to explore the topology and the influence of different sequences in the enzymatic ligation of catenated DNA rings.
·We designed hinges of different lengths (3T and 5T) to explore their impacts on enzymatic interlinks between DNA ring and ssDNA.
·We designed staple strands to control the topology of catenanes and simulated the synthesis of part of the hauberk.
·We designed linking domains of different lengths (10nt,12nt and 14nt) to explore their impacts on enzymatic ligation between the long splint which we called ‘scaffold’ and ssDNA strands.
·We tried several methods to connect the rings to improve our yield.
To lay the foundation for the DNA hauberk, we synthesized the simplest catenane first (see figure 2). Sequences of the same color are either identical or complementary to each other. The hinges are used to provide flexibility for the construction of DNA hauberk, yet it’s not clear how long they should be to make for the cyclization. So we tried both hinges made of 3 and 5 thymines (see figure 3) to verify which one was better. Sequences are demonstrated below (see table 1).
Figure 2. A two-ring catenane (Lk =1)
Figure 3.Catenanes made of circular strand A and linear strand C with (a) 3T and (b) 5T hinges. Staples in both figures contribute to the stabilization of the topological structures.
Table 1. Sequences of DNA strands employed in part 1 and 2.
(A-strand A, C-strand C, 3-TTT hinge, 5-TTTTT hinge, l-linear, sp-splint, ACl3-the left staple between circle A and strand C with TTT hinge, ACr3-which on the right)
To control the Lk of our DNA hauberk, we devised staples and conducted an experiment to explore their impacts on the topology of catenanes (see results).The linking domain was designed to be 10nt to restrict the Lk to 1.
In CA5+, strand A bind to circle C and get cyclized with the help of spA, while others use circle A as splint. Two staples, ACl5 and ACr5, are aimed to control the topology of them, and spA can do also this in AC5+. As staples are added, they can not only bind the two circles together, but also prevent them from incorrect intertwining. Meantime the splint added can reduce the mispairing by-product like this(see figure5).
Figure 4. Catenanes made of different DNA strands, splints and staples.
(CA-circle C and strand A, AC-circle A and strand C, 5-TTTTT hinge, +-both staples and splint, 2-2 staples, 1-1 staple, 0-no staple and splint)
Figure 5.The by-product when there is no staple and splint.
The LD (linking domain) between the scaffold and single strands play a major role in enzymatic ligation by T4 DNA ligase. We designed three kinds of scaffold (the lengths of LDs are 10nt, 12nt and 14nt) to see which one was more suitable for DNA hauberk.(see figure 6)(see results). We didn't try five T hinge, for the formation of catenane not only needs the cyclization of a single strand, but also requires them to bind "hand in hand". Indeed, a long hinge can reduce steric hindrance,while reducing the chance of intertwining between A and B at the same time.Furthermore, LD cannot be too long as well since we need to control the topology and keep the Lk to be one DNA sequences used in this process are demonstrated below (see table 2). We used DNA-mfold server to check the secondary structures of strand A, B, C and D to ensure that excessive mismatches had been avoided (see figure 7). The folding temperature was set to 25℃, and the ionic strength was set to 0 mM Na+ and 10 mM Mg2+ according to the T4 ligase buffer.
Figure 6. The first layer of hauberk with linking domains of (a)10nt, (b)12nt and (c)14nt between strand A, strand B and the scaffold.
Table 2. Sequences of DNA strands employed in part 3, 4, 5.
Figure 7.Secondary structures for (a) strand A,(b) strand B,(c) strand C and (d) strand D in Mfold. The folding temperature was set to 25℃.
To form the desired DNA hauberk, we tried several methods to improve our yield, like by changing the order of adding substrates to the reaction system (cyclizing strand A and B together or separately), using cycles or ssDNA as splints, etc. We choose two methods with significative results to show our idea.
We tried two methods, namely, using the scaffold and an oligonucleotide as the splint. For strand A and B, only the strands that bound to L3 can be cyclized, thus the catenane could not be formed by more than four rings. However, if strand A bound to L3 before intertwining with strand B, it would become an independent ring and lose the chance to form the catenane. As for A5ls and B5ls, their tail ends were free, enabling them to intertwine with others after binding toL3. They had less steric hindrance for T4 DNA ligase. Unfortunately, as they didn’t need L3 to get cyclized, by-products may increase.
In AB, both A and B use L3 as splint. In AsB, A use spAs (another type of splint we used, 12nt) as splint, and B use L3 as splint. In ABs, A use L3 as splint and B use spBs as splint. In AsBs, A use spAs as splint, and B use spBs as splint.
Fig 8.Four structures in Part 4.1 (A-A5l, As-A5ls and spAs, B-B5l. Bs-B5ls and spBs)
Table 3. Sequences of DNA strands employed in part 4. (The different between A5l and A5ls is that A5l use the scaffold as splint and A5ls use spAs as splint to get cyclized, and so on.)
We used the method in step two, that we add splints to control the topology. The presence of spAs and spBs prevents A5l and B5l from incorrect intertwining and reduces the by-product.(see figure 9)
In AsBs, spAs and spBs is aimed to help the cyclization of A and B. In AB, spAs and spBs is aimed to control the topology. In AB, only spAs is used to control the topology. In AB, only spBs is used to control the topology. In AB, L3 is used as splint and with no strand to control its topology.
Fig 9.Five structures in part 4.2 (A-only A5l, As-A5ls and spAs, B-only B5l. Bs-B5ls and spBs, ，A-spAs is added, B-spBs is added)
Table 4.Sequences of DNA strands employed in Part 5.(The different between C5l and C5ls is that C5l use the bottom of the first layer as splint and C5ls use spCs as splint to get cyclized, and so on.)
The ends of C5p and D5p will bind to the bottom of circle A and B, and get cyclized. We also try using the staple to promote the formation of target product (see figure 10). In R2CD, both C and D are added. In R2CD, both C and D are added, and ACl5 is used to help the formation of circle C. In R2C, Only C is added. In R2C, only C is added, and ACl5 is used to help the formation of circle C. In R2D, only D is added
Figure 10.Five structures in part 5.1 (R2-this is the second row of DNA hauberk, C-C5l, D-D5l, C-C5l and ACl5)
C5ls and D5ls will bind to circle A and B first, then pair with splint, and finally get cyclized. For the cyclization is not bound by the circle above and C5ps can intertwine freely with D5ps, the yield may be higher (see figure 11).
In R2CsDs, both C and D are added. In R2Cs, only C is added. In R2D, only D is added.
Figure 11.Five structures in part 5.2 (R2-this is the second row of DNA hauberk, Cs-C5ls, D-D5ls)
Lu C H, Cecconello A, Qi X J, et al. Switchable Reconfiguration of a Seven-Ring Interlocked DNA Catenane Nanostructure[J]. Nano Letters, 2015, 15(10):7133-7137.
Zuker M. Mfold web server for nucleic acid folding and hybridization prediction.[J]. Nucleic Acids Research, 2003, 31(13):3406-3415.