Self-assembly of DNA double multi-arm junctions (DMaJs)

Abstract

This article develops a class of DNA nanomotifs: double multi-arm junction (DMaJ) nanomotifs or tiles. Each motif consists of two multi-arm, branched, DNA junctions (either 6- or 8-arm-junctions). In the DMaJ motifs, two, long, parallel, DNA duplexes allow the tiles to interact with each other to form large nanostructure frameworks. In addition, they contain extra helical domains that could provide easy anchors for additional functional/structural elements or opportunities for complex, inter-motif connectivity to construct complex DNA-based nanomaterials. We have thoroughly characterized the formation of the new motifs by polyacrylamide gel electrophoresis (PAGE) and their assembly behaviors by atomic force microscopy (AFM) imaging.

---

Introduction

Tile-based self-assembly, complementary to DNA origami and the single-strand tile approach,^1,2 is a common strategy in DNA nanotechnology and has been successfully used to construct a wide range of DNA nanostructures and nanodevices.^3–8 Such structures have great potential for organizing nano-objects,^6,9–11 templating nanofabrications,^12,13 transporting materials,¹⁴ serving as biosensors,¹⁵ probing biological processing,¹⁶ and performing molecular computations.¹⁷ To increase building efficiencies and structural complexities, it is highly desirable to expand the list of available DNA nanomotifs. Herein, we report a class of DNA nanomotifs that offer a potential advantage of high connectivities for DNA nanoconstruction.

The DNA motifs reported here are constructed based on a hypothesis that an even-numbered, branched DNA junction (e.g. 4-, 6-, or 8-arm motif) will adopt a stacked conformation in the presence of divalent cations, such as Mg²⁺ (Fig. 1). DNA duplex branches pairwise stack onto each other at the center of the molecule to form pseudo-continuous DNA duplexes.^18–21 In this conformation, the number of free duplex ends is reduced, leading to lower the free energy of the junction. It is true for 4-arm junctions (Fig. 1a and b). A free 4-arm junction, however, has two possible stacking patterns and possesses a widely variable inter-helix angle; thus, is not suitable for nanoconstruction. When two 4-arm junctions are coupled into a double-crossover motif (DX) molecule, both the stacking pattern and the inter-helix angle become fixed (Fig. 1c).²² The resulting DX motif is rigid and its invention has led to a great leap in structural DNA nanotechnology.⁴ In this study, we reason that we can generalize the strategy to build DNA nanomotifs for nanoconstruction.

Fig. 1 Construction of double multi-arm junction motifs (DMaJ). (a) & (b) Single 4-arm junction. (c) Double 4-arm junction or double crossover (DX) motif. (d) & (e) Single 6-arm junction. (f) Double 6-arm junction (D6aJ) motif. (g) & (h) Single 8-arm junction. (i) Double 8-arm junction (D8aJ) motif. (a), (d), and (g) Single junctions in open conformation. (b), (e), and (h) Single junctions in stacked conformation. All structures are drawn by Tiamat program.¹⁸ Colored, thick lines represent DNA strand backbones and gray, thin lines represent base pairs. Note that a two-fold rotational axis (indicated by a pair of black arrows) runs vertically in plane through each DMaJ motif. Due to this symmetry, each four-stranded motif contains only two different strands: a long strand (L, colored red/orange) and a short strand (S, colored green/blue).

A pair of double multi-arm DNA junctions (DMaJs) nanomotifs, double 6-arm junction (D6aJ) and double 8-arm junction (D8aJ), have been reported (Fig. 1). Inspired by the 4-arm junction,¹⁸ which models Holliday junction – an intermediate molecule of genetic recombination,^19,20 other branched junctions [e.g. 6-arm junction (Fig. 1d) and 8-arm junction (Fig. 1g)] have been constructed with the motivation of building materials of connectivity other than four.^23,24 For example, an icosahedron (the basic architecture for many viral capsids) exhibits a connectivity of five²⁵ and many 3D crystal lattices have a connectivity of six (two along each of x, y, and z axes). Though the detailed structures are not known for those multi-arm junctions, the hydroxyl footprinting data indicates that base-stacking exists around the junction points as those in the 4-arm junction, suggesting that they may adopt stacked structures (Fig. 1e and h) as well.^23,24 Each pair of helical arms stacks onto each other at the junction point to form a pseudo-continuous duplex. In a free multi-arm junction, the stacking patterns are free to change, meaning one arm could stack onto different arms in different junction molecules and change stacking partners in the same junction molecule. To overcome this problem, we have coupled two even-number branched DNA junctions into a DMaJ, such as D6aJ (Fig. 1f) and D8aJ (Fig. 1i). To simplify the structures, we have introduced hairpins to the middle duplex(es) to reduce the number of component strands to four (colored differently). The two DMaJs in this study have two-fold rotational symmetries. Thus, a sequence-symmetry technique was further applied to reduce the number of unique DNA strands. The two, long L strands (colored red and orange) can be the same and the two, short S strands (colored green and blue) can be the same.

Experimental

Oligonucleorides

All DNA strands were purchased from IDT, Inc. and purified by denaturing PAGE. The strands were listed below:

S₆ (28 nt): GTGCATCACGCGttttCGCGCTAAGTCG;

S₆-1 (33 nt): GTGCATCACGCGttttCGCGCTAAGTCGCTCTG;

S₆-2 (36 nt): GGCCGTGCATCACGCGttttCGCGCTAAGTCGGCGC;

S₈ (40 nt): GTGCATCACGCGttttCGCGCCGGttttCCGGCTAAGTCG;

S₈-1 (45 nt): GTGCATCACGCGttttCGCGCCGGttttCCGGCTAAGTCGCTCTG;

S₈-2 (48 nt): GGCCGTGCATCACGCGttttCGCGCCGGttttCCGGCTAAGTCGGC GC;

(52 nt): CCGGGTGCATCAGAGGCttttGCCTCGCGTGttttCACGCCTAAGTCGG GCC;

L₆ (64 nt): CGACTTAGACTTCAGGCCTGAAGTGTTCAAGGCCTTGAACGGTT ATCCGGATAACC TGATGCAC;

L₆-1 (69 nt): CGACTTAGACTTCAGGCCTGAAGTGTTCAAGGCCTTGAACGGTT ATCCGGATAACC TGATGCACCAGAG;

L₈ (80 nt): CGACTTAGACTTCAGGCCTGAAGTAGTGTTGGCCAACACTGTTC AAGGCCTTGAACGGTTATCCGGATAACCTGATGCAC;

L₈-1 (85 nt): CGACTTAGACTTCAGGCCTGAAGTAGTGTTGGCCAACACTGTTC AAGGCCTTGAACGGTTATCCGGATAACCTGATGCACCAGAG;

L_8a (80 nt): CGACTTAGCTAGTACTCCTCCGAGTAATCCATGATCCTACGGCTA CAGACTGCATTAGATCCTATCCGTTCGTGATGCAC;

L_8b (80 nt): CGACTTAGCTCGGAGGAGTACTAGGTAGGATCATGGATTAAATGCAGTCTGTAGCCCGAACGGATAGGATCTTGATGCAC.

Strand composition of each nanostructure

D6aJ motif: S₆ + L₆ (1 : 1);

D6aJ 1D array: S₆-1 + L₆-1 (1 : 1);

D6aJ 2D array: S₆-2 + L₆ (1 : 1);

D8aJ motif: S₈ + L₈ (1 : 1);

D8aJ 1D array: S₈-1 + L₈-1 (1 : 1);

D8aJ 2D array: S₈-2 + L₈ (1 : 1);

D8aJ* motif: S₈ + L_8a + L_8b (2 : 1 : 1);

D8aJ* 2D array: + L_8a + L_8b (2 : 1 : 1).

Assembly of blunt-ended, individual motif

DNA strands were combined in TAE/Mg²⁺ buffer and incubated at 95 °C/5 min, 65 °C/30 min, 55 °C/30 min, 37 °C/30 min, 22 °C/30 min, then kept at 4 °C. TAE/Mg²⁺ buffer contained 40 mM Tris base (pH 8.0), 20 mM acetic acid, 2 mM EDTA, and 12.5 mM magnesium acetate.

Assembly of DNA 1D or 2D arrays

DNA strands (50 nM for motif) were dissolved in TAE/Mg²⁺ buffer and a freshly cleaved mica plate (Ted Pella, Inc.) was immersed into the solution. Then the DNA solution (with the mica substrate) was slowly cooled from 95 to 22 °C over 48 hours in a 2 liter water bath.

Denaturing PAGE

20% denaturing PAGE gel was prepared with the 19 : 1 acrylamide/bisacrylamide gel, 8 M urea, and TBE buffer [89 mM Tris base (pH = 8.0), 89 mM boric acid, 2 mM EDTA]. The gel was run at 55 °C for ∼1.5 h at 650 V on Hoefer SE 600 electrophoresis system and then stained with ethidium bromide (Sigma). The major band was cut under UV light and eluted out.

Native PAGE

8% native PAGE gel was prepared with 19 : 1 acrylamide/biacrylamide gel and TAE/Mg²⁺ buffer. The gel was run at 22 °C for Fig. 2 and at 4 °C for Fig. 4, stained with Stains-All (Sigma) and scanned by an HP scanner (Scanjet 4070 Photosmart).

Fig. 2 Native polyacrylamide gel electrophoresis (PAGE) analysis of blunt-end DMaJ motifs at room temperature. (a) D6aJ motif and (b) D8aJ motif. The DNA composition for each sample is indicated above the gel image and the chemical identity of each band is suggested beside the gel images. Note that all DMaJ motifs here are symmetric. Each motif is composed two identical L strands (red and orange) and two identical S strands (green and blue). Different colors are used to clarify the overall structures.

Quantification of the assembly yield

The intensities of all bands in a given lane were measured by ImageJ, an image processing software. The assembly yield of a given band was calculated by dividing the intensity of this band by the sum of the intensities of all bands in the entire lane.

AFM imaging

After array assembly, the mica was taken out from the DNA solution and washed by 25 μL 2 mM Mg(Ac)₂ and dried by compressed air. Imaging was performed on a Bruker Multimode 8 AFM at SCANASYST-AIR mode with ScanAsyst-air Nitride probe (Bruker). Or added 25 μL TAE/Mg²⁺ buffer onto the mica surface and imaged at SCANASYST-FLUID mode with ScanAsyst-fluid+ probe (Bruker). Fourier transformation was done by Gwyddion, a modular program for scanning probe microscopy.

Results and discussion

Formation of DMaJ motifs

The DMaJ complexes could efficiently form and were stable as evidenced by native polyacrylamide gel electrophoresis (PAGE, Fig. 2). DNA strands were dissolved in an Mg²⁺-containing, neutral, aqueous buffer (TAE/Mg²⁺ buffer) at a concentration of 1 μM. The DNA complexes formed upon slowly annealing DNA solution from 95 °C to 22 °C in 2 hours. Upon formation, the DNA samples were analyzed by native PAGE. DNA molecules migrated in the gel with different mobilities according to their molecular weights. Individual, component strands were used as references. In addition, strands L and S at the ratio of 2 : 1 were purposely used to form fragments of the target DMaJ complexes. At the desired stoichiometry (2 : 2), a single, dominant band appeared in each gel. This band could be confidently assigned to the target, four-stranded complex (L₆)₂(S₆)₂ or (L₈)₂(S₈)₂ based on the reference bands. Note that the L strands could form homodimer and higher oligomers (corresponding to bands with lower mobilities) even from thermally quenched solutions. Such phenomenon was more serious for D8aJ motif than for D6aJ since L₈ was longer than L₆ with more segments of self-complementary sequences. It also resulted in a lower assembly yield (41%) for the targeted (L₈)₂(S₈)₂ complex than the yield (78%) for (L₆)₂(S₆)₂.

Assembly of periodic arrays of DMaJ motifs

Both D6aJ and D8aJ motifs could assemble into large periodic arrays (Fig. 3). To program inter-motif association, we introduced sticky-ends to both the top and bottom duplexes of the motifs. By designing the sequence complementarities among the sticky-ends, the motifs could be programmed to specifically assemble into either one-dimensional (1D) periodic chains (Fig. 3a and c) or two-dimensional (2D) periodic arrays (Fig. 3b and d). We followed a protocol of surface-mediated self-assembly to directly assemble the DNA arrays on mica surface. Briefly, freshly cleaved mica substrates were immersed into buffered DNA solutions. DNA assembly was conducted by slowly cooling the mica-DNA solutions from 95 °C to 22 °C in two days.^26–28 After assembly, the DNA samples on mica surfaces were directly imaged by AFM without any further treatment. As are shown in the images, the surface coverage for each design is quite high. Both D6aJ and D8aJ motifs formed 1D chain-like structures as design up to a few hundred nanometers, corresponding to tens of copies of the motifs (each motif is ∼12 nm long). While the D6aJ motif formed periodic 2D arrays (Fig. 3b), the D8aJ motif formed only random networks on the mica surface (Fig. 3d) due to the high complexity of L₈ strand.

Fig. 3 Self-assembly of DMaJ motifs into periodic arrays as imaged by atomic force microscopy (AFM). (a) 1D chain and (b) 2D array assembled from D6aJ. (c) 1D chain and (d) random networks formed by D8aJ. For each case, a pair of AFM images at different magnifications is shown. Height scale bars are shown on the right. Note that all DMaJ motifs here are symmetric. Each motif is composed two identical L strands (red and orange) and two identical S strands (green and blue). Different colors are used to clarify the overall structures.

Asymmetric D8aJ motif

The symmetric D8aJ motif did not behave well. In PAGE analysis, there were multiple bands in addition to the correct D8aJ complex (Fig. 2b); indicating that the symmetric D8aJ motif was not a dominant, stable complex. In 2D assembly, only random networks, instead of regular 2D arrays, were observed. We suspected that the sequence symmetry (particular the long strand L₈, which contains four 16-base-long palindromes) led to formation of other alternative complexes other than the desired D8aJ complex and caused the failure of the formation of 2D arrays. To overcome this problem, an asymmetric D8aJ* was designed. In the revised motif, the secondary structure-rich L₈ strand was replaced by two simple strands L_8a and L_8b. They associated together to form the core of the motif, and further associated with the two outside, short strands (S₈) to form the D8aJ* motif. The revised asymmetric motif formed well (assembly yield: 95%) as analyzed by native PAGE (Fig. 4b). Individual strands L_8a and L_8b both appeared as one single band in their lanes. At the correct ratio (1 : 1 : 2), strands L_8a, L_8b and S₈ associated together to form a stable D8aJ* motif, suggested by a dominant, sharp band in the lane. When the D8aJ* motif was revised to contain sticky ends, the modified D8aJ* motif, as expected, successfully self-assemble into desired 2D arrays (Fig. 4c). The regularity was clearly indicated by the Fourier transform pattern of the AFM images.

Fig. 4 Asymmetric D8aJ* motif. (a) Structural scheme. (b) Native PAGE analysis of the formation of the blunted, asymmetric D8aJ* motif. (c) Scheme and (d) AFM images of 2D arrays assembled from the sticky-ended, asymmetric D8aJ* motif. A pair of AFM images at different magnifications is shown. Height scale bar is on right. Inset: Fourier transform pattern of the array. Note that the asymmetric DM8J motif is composed of two different L strands (red L_8a and orange L_8b) and two identical S strands (green and blue).

The DMaJ motifs developed here share some common features with other motifs, the so-called weave tile in particular.²⁹ In the weave tile, two DNA single strands weave together to form multiple, 1.5-helical turn-long, helical domains. The overall architecture resembles the homo-/hetero-dimers formed by the L strands in the present study. The weave tiles have been demonstrated to be able to assemble into 2D arrays. The weave tiles, however, expose a large amount of DNA base-pair surfaces, which align with each other on two sides. In aqueous solutions, the weave tiles will have strong tendency to randomly base-stacking onto each other and bury quite large hydrophobic surfaces.^30–34 One way to prevent the base-stacking is to add single-stranded DNA (ssDNA) loops at the ends of each helical domain. However, the ssDNA loops will introduce undesired and uncontrollable flexibility to the tiles. In the present study, all the end basepairs in the dimers of the L strands are protected by the end basepairs of the multiple helical domains of the S strand. And there is no undesired, flexible ssDNA involved. Thus, the DMaJ motif would be expected to be more geometrically defined. In addition, it is worthy pointing out that multi-arm junctions in such stacked conformation have been observed before, but in the content of a very large DNA origami structure.³⁵ Multi-arm junctions in open conformations have also been used in a DNA icosahedron and DNA origami structures.^36–38

Conclusions

In conclusion, we have developed a class of DNA nanomotifs (DMaJs). Each motif contains multiple, pseudocontinuous DNA duplexes and can be regarded as two multi-arm junctions connected together. The proposed structural models for DMaJs are consistent with the AFM imaging study. Though the present study only exploits four sticky ends from the DMaJ motif, it is straightforward to replace the hairpin loops on the S strand to sticky-ends, making the DMaJ motifs to contain more sticky ends for construction of multiple-connected networks in both 2D and 3D. Such networks are expected to be highly stable. In addition, this study hints on the limitation of exploiting sequence symmetry approach⁷ for reducing the number of unique DNA strands. As the structural complexity increases, symmetric sequences potentially increase the risk of forming undesired structures as in the case of the M8aJ motif. Removing the sequence symmetry effectively avoids the chance of mis-assembly and promotes the formation of the designed structures.

Acknowledgements

We thank NSF, NSFC (#81429001), and ONR (N00014-15-1-2702) for support of this work.

References

1. P. W. Rothemund, Nature, 2006, 440, 297–302.
2. B. Wei, M. Dai and P. Yin, Nature, 2012, 485, 623–626.
3. C. Lin, Y. Liu, S. Rinker and H. Yan, ChemPhysChem, 2006, 7, 1641–1647.
4. E. Winfree, F. Liu, L. A. Wenzler and N. C. Seeman, Nature, 1998, 394, 539–544.
5. S. H. Park, R. Barish, H. Li, J. H. Reif, G. Finkelstein, H. Yan and T. H. LaBean, Nano Lett., 2005, 5, 693–696.
6. H. Yan, S. H. Park, G. Finkelstein, J. H. Reif and T. H. LaBean, Science, 2003, 301, 1882–1884.
7. Y. He, Y. Tian, Y. Chen, Z. Deng, A. E. Ribbe and C. Mao, Angew. Chem., Int. Ed., 2005, 117, 6852–6854.
8. Y. He, T. Ye, M. Su, C. Zhang, A. E. Ribbe, W. Jiang and C. Mao, Nature, 2008, 452, 198–201.
9. D. N. Selmi, R. J. Adamson, H. Attrill, A. D. Goddard, R. J. C. Gilbert, A. Watts and A. J. Turberfield, Nano Lett., 2011, 11, 657–660.
10. Y. He, Y. Tian, A. E. Ribbe and C. Mao, J. Am. Chem. Soc., 2006, 128, 12664–12665.
11. J. Sharma, R. Chhabra, A. Cheng, J. Brownell, Y. Liu and H. Yan, Science, 2009, 323, 112–116.
12. Z. Deng and C. Mao, Angew. Chem., Int. Ed., 2004, 43, 4068–4070.
13. S. P. Surwade, S. Zhao and H. Liu, J. Am. Chem. Soc., 2011, 133, 11868–11871.
14. G. Zhu, J. Zheng, E. Song, M. Donovan, K. Zhang, C. Liu and W. Tan, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 7998–8003.
15. C. Lin, Y. Liu and H. Yan, Nano Lett., 2007, 7, 507–512.
16. R. Chhabra, J. Sharma, Y. Liu, S. Rinker and H. Yan, Adv. Drug Delivery Rev., 2010, 62, 617–625.
17. R. D. Barish, R. Schulman, P. W. Rothemund and E. Winfree, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 6054–6059.
18. S. Williams, K. Lund, C. Lin, P. Wonka, S. Lindsay and H. Yan, in DNA Computing, Springer, Berlin Heidelberg, 2008, pp. 90–101.
19. J. H. Chen, M. E. A. Churchill, T. D. Tullius, N. R. Kallenbach and N. C. Seeman, Biochemistry, 1998, 27, 6032–6038.
20. B. F. Eichman, J. M. Vargason, B. H. Mooers and P. S. Ho, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 3971–3976.
21. J. P. Cooper and P. J. Hagerman, Proc. Natl. Acad. Sci. U. S. A., 1989, 86, 7336–7340.
22. T. J. Fu and N. C. Seeman, Biochemistry, 1993, 32, 3211–3220.
23. Y. Wang, J. E. Mueller, B. Kemper and N. C. Seeman, Biochemistry, 1991, 30, 5667–5674.
24. X. Wang and N. C. Seeman, J. Am. Chem. Soc., 2007, 129, 8169–8176.
25. C. Zhang, M. Su, Y. He, X. Zhao, P. Fang, A. E. Ribbe, W. Jiang and C. Mao, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 10665–10669.
26. S. Hamada and S. Murata, Angew. Chem., Int. Ed., 2009, 121, 6952–6955.
27. J. Lee, S. Hamada, S. U. Hwang, R. Amin, J. Son, S. R. Dugasani, S. Murata and S. H. Park, Sci. Rep., 2013, 3, 2115.
28. X. Sun, S. H. Ko, C. Zhang, A. E. Ribbe and C. Mao, J. Am. Chem. Soc., 2009, 131, 13248–13249.
29. M. N. Hansen, A. M. Zhang, A. Rangnekar, K. M. Bompiani, J. D. Carter, K. V. Gothelf and T. H. LaBean, J. Am. Chem. Soc., 2010, 132, 14481–14486.
30. H. Yan, T. H. LaBean, L. Feng and J. H. Reif, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 8103–8108.
31. R. Wang, A. Kuzuya, W. Liu and N. C. Seeman, Chem. Commun., 2010, 46, 4905–4907.
32. T. Gerling, K. F. Wagenbauer, A. M. Neuner and H. Dietz, Science, 2015, 347, 1446–1452.
33. S. Woo and P. W. Rothemund, Nat. Chem., 2011, 3, 620–627.
34. C. Tian, C. Zhang, X. Li, Y. Li, G. Wang and C. Mao, J. Am. Chem. Soc., 2012, 134, 20273–20275.
35. X. Bai, T. G. Martin, S. H. Scheres and H. Dietz, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 20012–20017.
36. E. Benson, A. Mohammed, J. Gardell, S. Masich, E. Czeizler, P. Orponen and B. Högberg, Nature, 2015, 523, 441–444.
37. M. Matthies, N. P. Agarwal and T. L. Schmidt, Nano Lett., 2016, 16, 2108–2113.
38. D. Bhatia, S. Mehtab, R. Krishnan, S. S. Indi, A. Basu and Y. Krishnan, Angew. Chem., Int. Ed., 2009, 48, 4134–4137.

---

Footnote

† These authors contributed equally.

---