DNA self-assembly from 2D to 3D

DNA self-assembly from 2D to 3D

(Parte 1 de 3)

DNA self-assembly: from 2D to 3D

Chuan Zhang,a Yu He,a Min Su,b Seung Hyeon Ko,a Tao Ye,a Yujun Leng,b Xuping Sun,a Alexander E. Ribbe,a Wen Jiangb and Chengde Mao*a

Received 16th March 2009, Accepted 6th April 2009 First published as an Advance Article on the web ????? DOI: 10.1039/b905313c

This paper describes our recent efforts on the self-assembly of three-dimensional (3D) DNA nanostructures from DNA star motifs (tiles). DNA star motifs are a family of DNA nanostructures with 3, 4, 5, or 6 branches; they are named as 3-, 4-, 5-, 6-point-star motifs, respectively. Such motifs are programmed to further assemble into nanocages (regular polyhedra or irregular nanocapsules) with diameters ranging from 20 nm to 2 m. Among them, DNA nanocages derived from 3-point-star motif consists of a group of regular polyhedra: tetrahedra, hexahedra (or cubes), dodecahedra and buckyballs (containing 4, 8, 20, and 60 units of the 3-point-star motif, respectively). An icosahedron consists of twelve 5-point-star motifs and is similar to the shapes of spherical viruses. 6-point-star motifs can not assemble into regular polyhedra; instead, some sphere-like or irregular cages with diameters about 1–2 m will form. Similar large cages can also assemble from the 5-point-star motif when the DNA concentrations are higher than those for assembling regular icosahedra. In our study, we have identified several important factors for assembly of well-defined 3D nanostructures, including the concentration, the flexibility, and the arm length of the DNA tiles and the association strength between the DNA tiles.

1. Introduction

Beyond the genetic interest, DNA has been shown as a superb molecular system in self-assembly towards bottom-up nanofabrication.1–3 It has many attractive properties: the excellent capability of molecular recognition, the well-defined secondary structure (duplex), reasonable chemical stability, and readily commercial availability. These properties together endow DNA molecules with a great potential to serve as ‘‘building blocks’’ for the preparation of nanostructures.

In structural DNA nanotechnology, DNA motifs (tiles) are designed and investigated as building blocks to fabricate nanostructures. For example, a range of DNA motifs including double crossover (DX) and multiple-crossover motifs, star motifs, triangle and DX triangle motifs, and parallelogram motifs, have been introduced to construct various DNA two-dimensional (2D) arrays and non-periodic 2D patterns.4–18 However, the investigation of DNA 3D nanostructures is quite limited.19–27 In the early 90s, DNA 3D nanostructures with the connectivities of a cube19 and of a truncated octahedron20 were constructed by Seeman and his coworkers. Limited by characterization methods (denaturing gel electrophoresis), such 3D structures had to be covalently linked, which required the authors to useDepartment of Chemistry, Purdue University, West Lafayette, Indiana, 47907, USA. E-mail:

mao@purdue.eduMarkey Center for Structural Biology and Department of Biological Sciences, Purdue University, West Lafayette, Indiana, 47907, USA

PAPER 143/14 w.rsc.org/faraday_d | Faraday Discussions 1 stepwise synthesis via enzymatic ligation. Successive steps of ligation and purification, unfortunately, greatly increased the workload and resulted in a very low overall synthetic yield. With the introduction of new characterization tools [cryogenic electron microscopy (cryoEM) and atomic force microscopy (AFM)], one was able to visualize non-covalently associated DNA structures by direct imaging. Starting from 2004,21 strategies of one pot self-assembly were used to assemble DNA 3D nanostructures. For example, Turberfield and his coworkers used four oligonucleotides to assemble DNA tetrahedra;2 the resulting structures were characterized by AFM. In another elegant example, Shih and his coworkers folded a long single DNA strand into pre-designed, branched, secondary structures, which further folded into an octahedon through intra-complex paranemic interactions with the help of a number of short DNA helper strands.21 Essentially, this is related to the so-called ‘‘DNA origami’’ approach.16 The resulting octahedral structures were clearly shown by cryoEM characterization. This work together demonstrated that DNA polyhedra could be readily assembled from synthetic DNA molecules.

Though tetrahedra, cubes, octahedral, and truncated octahedra are highly symmetric, those reported DNA 3D objects do not really possess these symmetries if viewed at the level of DNA sequences. Consequently, every object requires many unique DNA strands. As the objects become larger and more complicated, the number of DNA strands with unique sequences quickly increases. It is not easy to apply the above mentioned methods to construct complicated polyhedra, which require too many unique DNA strands. Thus, while these studies are elegant and exciting, they do not provide a general and simple route to fabricate DNA 3D nanostructures. In biology, many complex nanostructures with various biological functions assemble in different ways. For instance, the outer protein shells (capsids) of spherical viruses have highly symmetric, icosahedral structures. They are composed of many copies of identical protein subunits, which associate with each other via non-covalent bonding. The complexity and robustness of viral capsids indicate

Fig. 1 DNA nanocages self-assembled from DNA star motifs. ART B905313C

5 2| Faraday Discuss., 2009, 143, 1–13 This journal is ª The Royal Society of Chemistry 2009 that this architectural strategy is very effective. Despite the difficulty to fully understand and mimic the self-assembly of viral capsids, it should be possible to rationally design DNA nanomotifs (tiles) that can behave in a similar way to capsid component proteins to self-assemble into highly symmetric DNA 3D nanostructures. Recently we have successfully used symmetric DNA 3-point-star motifs to assemble three DNA polyhedra (tetrahedra, dodecahedra, and buckyballs) and proved their structures by cryoEM and single particle 3D reconstruction.28 Since then, a number of other DNA polyhedra and DNA cages have been assembled from a series of DNA star motifs (Fig. 1).

In this paper, we will discuss our efforts on the self-assembly of discrete, 3D DNA nanostructures from symmetric DNA star motifs. The DNA 3D assemblies will be classified by their component units, the nanomotifs. We will start from a 3-point-star motif and discuss the design and parameter controls for fabrication of different types of DNA nanocages; 5- and 6-point-star DNA cages will be discussed in the same way. At the end, we will present our thoughts on future directions.

2. DNA nanocages assembled from 3-point-star motifs

From the view of geometry, a number of regular polyhedra can be considered as assemblies of 3-branched tiles and these polyhedra differ from each other by the number of the component building blocks (tiles). Such polyhedra include tetrahedra, hexahedra (cubes), dodecahedra and buckyballs with increasing sizes. Each 3- branched motif can be represented by a DNA 3-point-star motif. Similar to the self-assembly of DNA 2D arrays, the assembly of DNA nanocages can be carried out by a one-pot hierarchical self-assembly process (Fig. 2): as the solution temperature decreases, individual single DNA strands assemble into sticky-ended, 3-pointstar motifs (tiles), and then the tiles further assemble into 3D nanocages through sticky-end association between the tiles.

DNA tetrahedra, dodecahedra, and buckyballs were the first three examples assembled by 3-point-star motifs with component numbers of 4, 20 and 60.28 This work has clearly demonstrated this biomimetic, hierarchical assembly strategy. The design is shown in Fig. 2 and the key parameters that determine the final structures are the flexibility and the concentration of the 3-point-star DNA tiles. In

Fig. 2 Self-assembly of DNA polyhedra. Three different types of DNA single strands (L, M, and S) stepwise assemble into symmetric 3-point-star motifs (tiles) and then into polyhedra in a one-pot process. Note that there are three single-stranded loops (red) in the center of the 3- point-star tile. The final structures (polyhedra) are determined by the loop length and the DNA concentration. In each of the final polyhedral structures, one of the component 3-point-star tiles is highlighted in gold.

general, DNA tiles with less flexibility and at higher concentrations will result in larger assemblies. The three-point-star motif contains a three-fold rotational symmetry and consists of 7 strands: a long repetitive central strand (L), three identical medium strands (M), and three identical short peripheral strands (S). Three single-stranded loops (colored red) are introduced at the center of the DNA tile. The tile flexibility can be easily adjusted by varying the loop length. Increasing the loop length increases the tile flexibility. At the termini of each branch of the tile, there are two self-complementary, 4-base-long, single-stranded overhangs, or sticky ends. Sticky-ends association (base-pairing) between the tiles allow finite numbers of tiles to assemble into 3D nanostructures.

Agarose gel electrophoresis was used for the initial characterization of those DNA cages and for estimation of the assembly yields. It was found that small structures generally had higher assembly yields than the larger cages did. For example, the assembling yield for the tetrahedra was over 90% as estimated from the gel by an image processing software, ImageJ (developed at the National Institute of Health, NIH), but it became lower and lower for the dodecahedra and the buckyballs. The yield of the properly assembled DNA buckyballs was only 60%. Non-uniform growth and deformation of the larger assemblies might be responsible for the observed, low assembly yields.


A tetrahedron is the smallest closed structure assembled from the 3-point-star DNA tiles. Each vertex is one 3-point-star tile and each edge comprises two associated branches from two tiles. The tiles in a tetrahedron adopt a significantly bent conformation when compared with the free, flat tiles. To accommodate such bending, the tiles have to be quite flexible; such flexibility can be provided by using long central loops (5 bases long for the red segments). At a sufficiently low DNA concentration (75 nM), tetrahedra are the dominating complexes assembled. Dynamic light scattering (DLS) shows that the apparent radius of the DNA complexes is 10.3 0.6 nm, agreeing well with the radius (10.9 nm) of the circumscribed sphere of the DNA tetrahedral model if taking the parameters of the standard B-DNA conformation (pitch: 0.3 nm base pair 1; diameter: 2 nm). The DNA assemblies also appear to be uniformly-sized, triangular particles when imaged by AFM in air. The observed structures in the AFM images are probably collapsed DNA tetrahedra due to the strong interaction between the DNA samples and the substrates.

The tetrahedral structure is more clearly shown by direct cryoEM imaging (Fig. 3, top panel). In cryoEM images, most visible particles have tetrahedral shapes of the expected size. The observed edges are 16 nm long, nicely matching with the designed model (16.2 nm). With experimentally observed particles, a DNA tetrahedral structure can be modeled at 2.8 nm resolution through a well-established technique of single particle 3D reconstruction.


A dodecahedron consists of twenty 3-point-star tiles, 12 pentagonal faces, and 30 edges. Compared to those in a tetrahedron, the 3-point-star tiles in a dodecahedron are less bent and require much less flexibility. Thus, the length of the central singlestranded loop is reduced to 3 bases long, which stiffens the star tiles. The concentration of DNA tiles is also critical to dodecahedron assembly. The dodecahedral complex is the main product only at low DNA concentration (50 nM). The apparent radius of the assembled objects in DLS measurement is 24.0 1.8 nm, agreeing with the calculated value (23.6 nm) from the designed structural model. Under AFM imaging (in air), the assembled objects show circular features and uniform sizes. Pentagonal structures are visible at the center of the top layer, representing a reasonable, collapsed 2D projection of a dodecahedron.

5 4| Faraday Discuss., 2009, 143, 1–13 This journal is ª The Royal Society of Chemistry 2009

Fig. 3 Visualization of the self-assembled DNA polyhedra by cryogenic transmission electron microscopy (cryoTEM). (top panel) tetrahedra, (middle panel) dodecahedra, and (bottom panel) buckyballs. (a) A representative image. White boxes indicate the DNA particles. (b) Comparison between raw images of individual particles (left) and the corresponding

CryoEM (Fig. 3, middle panel) showed dodecahedral-shaped objects of the expected size (in the boxed area). All particles with the right size were picked up for single particle 3D reconstruction. The icosahedral symmetry was imposed during the reconstruction, resulting in a well-defined dodecahedral structure (at a resolution of 2.5–3.0 nm). The individual, raw cryoEM particles matched with the computergenerated model projections very well.


3-point-star tiles can also assemble into Buckminsterfullerenes (or buckyballs), a type of highly symmetric polyhedron. A buckyball contains 60 vertexes, 90 edges, and 32 faces (12 pentagons and 20 hexagons). Assembling DNA buckyballs and dodecahedra require exactly the same DNA strands. To selectively assemble these structures, the DNA concentration has to be carefully controlled. At a high DNA concentration (500 nM), the 3-point-star DNA tiles (the central, single-stranded loops are 3 bases long) readily assemble into the large buckyball structures rather than small dodecahedra. The DLS measurement indicates that the DNA assemblies have an apparent hydrodynamic radius of 42.2 4.0 nm, close to the calculated radius of the buckyball model (41 nm). However, the polydispersity of the assembled DNA buckyballs is significantly higher than those of DNA tetrahedra and dodecahedra. Under AFM imaging, the DNA assemblies appear as uniform-sized, round, collapsed DNA nanostructures with a diameter of 110 nm. Round-like DNA structures can be seen in the raw CryoEM images (Fig. 3, bottom panel) and a buckyball structure has been obtained from reconstruction based on hand-picked particles. Compared to tetrahedra and dodecahedra, the size distribution of the particles in the cryoEM images is much worse than that from the AFM images. This phenomenon is likely due to larger assemblies being easier to deform or break up, which lowers the quality of the reconstructed map.

The first three 3D objects have a common feature in their design. Each pseudocontinuous DNA duplex of the edge in the final polyhedra is 4 turns (42 bases) long after sticky end hybridization. In such a design, all component DNA tiles are facing the same direction and any intrinsic curvatures of the tiles would add up toward the same direction to promote the formation of closed structures. Meanwhile, the DNA concentration plays a key role during the self-assembly. Selfassembly relies on inter-tile interactions. High DNA concentrations favor large assemblies in the same design such as buckyballs, and low DNA concentrations will promote small assemblies such as dodecahedra. The desired products dominate in self-assembly only in a certain range of DNA concentration.

(Parte 1 de 3)