The importance of being modular

DNA is the material of choice for making custom-designed, nanoscale shapes and patterns through self-assembly. A new technique revisits old ideas to enable the rapid prototyping of more than 100 such DNA shapes. See Letter p.623

Carpenters have been turning trees into furniture and dwellings for thousands of years, and so the discipline of woodworking features well-established techniques for joining pieces of wood to achieve a desired form. Nanotechnologists similarly try to use DNA as a material for crafting nanometre-scale shapes, but 'DNA-working' has been in development for a mere three decades. Because our picture of how DNA self-assembles is incomplete, DNA-working techniques are still evolving. The latest development is reported on page 623 of this issue, where Wei et al.1 present a method whose intrinsic modularity enables arbitrary DNA shapes to be constructed with striking speed.

The practice of building nanoscale structures from DNA2 once required creativity, intimate knowledge of DNA geometry and considerable synthetic effort. For example, in the late 1980s, the design and multistep synthesis of a 7-nanometre cube3 from ten DNA strands, and its subsequent characterization, took about 2 years (N. C. Seeman, personal communication).

In 2006, DNA origami4 emerged as a simple method that allows non-experts to rapidly design and synthesize complex DNA structures of approximately 100 nanometres in diameter, with reaction yields that often exceed 90%. In this technique, a single long strand of DNA is folded in one step by approximately 200 short DNA strands called staples, to create whatever shape is desired (Fig. 1a). In less than a week, one can accomplish all the steps required to make a DNA object: the computer-aided design and chemical synthesis of the staples; the formation of the object; and the final characterization of the product by atomic force microscopy (AFM). The original origami method4 made only two-dimensional shapes, but was quickly extended to enable the construction of three-dimensional architectures5,6 and curved geometries7,8.

Figure 1: More emoticons for your money.

a, In DNA origami, a set of short DNA strands (known as staples) is used to fold a long strand into a shape. For each different shape, a new set of staples must be synthesized, at a cost of roughly US$1,000. Here, a set of 237 staples (yellow; not all are shown) folds a long DNA strand (red) to create a smiley face4, whereas another completely different set of 204 staples (blue) creates a dolphin14. Scale bar, 100 nanometres; colours have been added to the micrographs of the DNA shapes on the right. b, In Wei and colleagues' approach1 for making DNA shapes, a set of 310 different strands acts as a library of tiles that, when mixed together, self-assemble to form a rectangular molecular 'canvas'. To make any other shape, an appropriate subset of these strands is selected and mixed together — the tiles shown in yellow on the left are the subset that self-assembles into the corresponding emoticon at right. An extra set of 1,396 strands (not shown) is also required to seal the shapes' edges and prevent them from aggregating. A fixed set of 1,706 strands, costing roughly $7,000, can therefore make an astronomical 2310 = 2×1093 potential shapes, such as the emoticons shown.

Because of its modularity, DNA origami provides a general platform for arranging other nanoscale objects — from electronic components to enzymes — as required. For example, DNA has been used to make a 'pegboard' onto which carbon nanotubes were organized into transistors9; a 'picture frame' into which individual DNA-repair enzymes were mounted so that their motion could be captured by AFM as they processed a substrate10; and a 'clamshell' that was programmed to respond to specific cancer cells in vitro by popping open to deliver a potentially therapeutic payload11. In each case, the pattern of components on top of the origami can be quickly and inexpensively reconfigured to perform a different task.

The modularity of DNA origami makes it a highly efficient technique for generating patterns, and its efficiency can be quantified. Let us consider how patterns might be added to a DNA origami rectangle. An experimenter can purchase a set of N staples (we'll call these 'white' staples) to fix the long strand into shape, and a second set of N 'black' staples that is identical to the first except that each staple carries a special chemical group. Each white staple, and equivalently the corresponding black staple, specifies a unique position in the final rectangle. By choosing the colour of the staple for each position, a set of just 2N strands can be used to create any of 2 N possible black and white patterns. If the chemical group on the black staples can act as a point of attachment for a small piece of a nanowire, for example, then any of 2 N possible patterns of wires can be made.

The most fundamental limitation of DNA origami is that this trick for obtaining an exponential increase in the number of patterns from a linear increase in the number of DNA strands does not generalize to shapes — for each new shape, one must design a new fold for the long strand and purchase another set of staples. Wei and colleagues' technique1 dispenses with the long strand and so allows different shapes to be generated highly efficiently.

The authors' approach returns to a previously used paradigm for DNA-working, that of DNA tiles12. In their system, each tile is a single DNA strand with four different binding domains that specify which four other tiles can bind to it as neighbours. The authors' general scheme specifies a set of N tiles that self-assemble to form a rectangle, within which each tile adopts a particular position. By mixing together appropriate subsets of tiles and allowing them to self-assemble, arbitrary DNA shapes can be prepared (Fig. 1b; N = 310).

The DNA strands on the edges of each shape have free binding domains, which can cause the shapes to clump together. To render the edges non-sticky, the authors added edge-protector strands where necessary. Because each of the four domains on N different tiles might need to be protected, a set of 4N additional strands was required. So, to access any of 2 N potential shapes, the single-stranded tile technique requires just 5N different strands. This efficient and modular architecture allowed Wei et al. to construct 107 shapes by hand, spending just a few hours on each shape. By using a robot to select and mix strands, the authors reduced the time required to make a shape to one hour. In this way, they constructed 44 shapes in about 44 hours. This advance truly brings DNA nanotechnology into the rapid-prototyping age, and enables DNA shapes to be tailored to every experiment.

Wei and colleagues' technique is the large-scale realization of a concept known as uniquely addressed tiling, which was first formally described13 12 years ago. So why is this advance happening only now? One answer is that, according to the predominant thinking about DNA self-assembly, such a technique should not work well — making the concentrations of tile strands perfectly equal is experimentally difficult, and relatively small departures from equality were expected to result in low yields of target structures. This idea followed from the common assumption that many DNA structures would begin self-assembling simultaneously, and then get stuck as partially complete shapes when tiles present at lower concentrations were exhausted. This potential problem was so compelling that DNA origami was invented expressly to avoid it. But the yields of Wei and colleagues' structures are surprisingly high: up to 40% for some shapes.

The success of the method cries out for explanation. The authors suggest that, if the nucleation of self-assembly is rare and the subsequent growth of a DNA shape is fast, then complete structures will form in preference to partial ones. Another possibility is that more-complete structures can gain strands from less-complete ones through a mechanism called Ostwald ripening, in which strands fall off less-stable structures and rejoin more-stable ones. Wei and colleagues' choice of single DNA strands as tiles — rather than the more complex, multistranded tiles used previously12 — could have a crucial role, because more-complete structures might steal single strands from less-complete structures directly, without any tiles falling off, by strand displacement.

More generally, both the single-stranded-tile method1 and DNA origami violate several other previous intuitions about what should and should not work. In both cases, careful studies of yields, kinetics and mechanism will be required to circumscribe the conditions under which each method works best and determine whether the single-stranded tile method will supplant DNA origami in practical applications. Wei and colleagues' findings remind us that we are still just apprentice DNA carpenters, and will embolden others to mix hundreds of DNA strands together against prevailing wisdom. The results will probably surprise us.


  1. 1

    Wei, B., Dai, M. & Yin, P. Nature 485, 623–626 (2012).

    CAS  ADS  Article  Google Scholar 

  2. 2

    Seeman, N. C. J. Theor. Biol. 99, 237–247 (1982).

    CAS  Article  Google Scholar 

  3. 3

    Chen, J. & Seeman, N. C. Nature 350, 631–633 (1991).

    CAS  ADS  Article  Google Scholar 

  4. 4

    Rothemund, P. W. K. Nature 440, 297–302 (2006).

    CAS  ADS  Article  Google Scholar 

  5. 5

    Andersen, E. S. et al. Nature 459, 73–76 (2009).

    CAS  ADS  Article  Google Scholar 

  6. 6

    Douglas, S. M. et al. Nature 459, 414–418 (2009).

    CAS  ADS  Article  Google Scholar 

  7. 7

    Dietz, H., Douglas, S. M. & Shih, W. M. Science 325, 725–730 (2009).

    CAS  ADS  Article  Google Scholar 

  8. 8

    Han, D. et al. Science 332, 342–346 (2011).

    CAS  ADS  Article  Google Scholar 

  9. 9

    Maune, H. T. et al. Nature Nanotechnol. 5, 61–66 (2010).

    CAS  ADS  Article  Google Scholar 

  10. 10

    Endo, M., Katsuda, Y., Hidaka., K. & Sugiyama, H. Angew. Chem. Int. Edn 49, 9412–9416 (2010).

    CAS  Article  Google Scholar 

  11. 11

    Douglas, S. M. et al. Science 335, 831–834 (2012).

    CAS  ADS  Article  Google Scholar 

  12. 12

    Winfree, E., Liu, F., Wenzler, L. A. & Seeman, N. C. Nature 394, 539–544 (1998).

    CAS  ADS  Article  Google Scholar 

  13. 13

    Rothemund, P. W. K. & Winfree, E. Proc. 32nd Annu. Symp. Theor. Comput. 459–468 (2000).

  14. 14

    Andersen, E. S. et al. ACS Nano 2, 1213–1218 (2008).

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding authors

Correspondence to Paul W. K. Rothemund or Ebbe Sloth Andersen.

Ethics declarations

Competing interests

P.W.K.R. is the inventor on a patent that describes the method of ‘scaffolded DNA origami’. For certain nanopatterning applications, the method in the paper described by Yin et al. is a competitor with the method of DNA origami.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Rothemund, P., Andersen, E. The importance of being modular. Nature 485, 584–585 (2012).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.