8/3/2019 Thomas H. LaBean- Another dimension for DNA art

1/2

ba

NANOTECHNOLOGY

Another dimension for DNA artThomas H. LaBean

Many of natures intricate nanostructures self-assemble from subunits. Efforts to mimic these assembly

processes enter a new phase with a method to design and build three-dimensional DNA nanostructures.

Through the ages, some of the most iconicandlasting artefacts of human ingenuity have beensculptures and carvings, created from a widevariety of materials. But until now, a general-purpose material from which nanometre-scale,three-dimensional shapes could be made has

been lacking. On page 414 of this issue, Douglaset al.1introduce a clever method for fabricat-ingnanometre-scale objects from DNA, andreport the construction of several such objects.The authors describe their method as analo-gous to sculpture from a porous crystallineblock, except that the structure of their blockconsists of tubes DNA double helices arranged in a regular honeycomb lattice. Thedesired shapes are not literally carved into thestarting material, but instead form from DNAthat has been designed to self-assemble into asupramolecular complex.

The use of DNA as a construction material for

making nanometre-scale objects began morethan 25 years ago2, and has since developed intothe field of structural DNA nanotechnology3,4.The field relies on the fact that molecularrecognition and assembly of DNA can beprogrammed so that it forms designed nanos-tructures. Such programming is enabled by ourunderstanding of WatsonCrick base pairing:for any DNA base sequence, we can immedi-ately determine the complementary sequence,and know that the two molecules will find andbind to one another in water under appropri-ate conditions. Well-developed synthesis tech-niques allow DNA strands of any desired base

sequence to be easily prepared.In 1998, DNA nanotechnology was trans-formed by the introduction of the tile andlattice strategy5. Tiles are nanometre-scalebuilding blocks that fold independently, andtypically contain domains of DNA doublehelices tethered by strand-exchange points.These points model naturally occurring junc-tions that form in genetic-recombination com-plexes when DNA strands are traded betweenhelices. Short, single-stranded DNA segmentshang off the tiles at strategic locations. On cool-ing in solution, the single-stranded segmentson different tiles bind to each other, so that

the tiles assemble into larger, predominantlytwo-dimensional lattices.

Another conceptual leap occurred in 2006,with the demonstration of DNA origami6.This strategy uses a long, single-strandedDNA for example, the genome of the M13virus as a scaffold molecule that is foldedback and forth on itself to form a planar raft

of double helices.

The resulting structure isknitted together by a few hundred short, syn-

thetic DNA molecules that act as staples, link-ing together the helices at appropriately spacedstrand-exchange points. The raster-like routingof DNA scaffolds through origami struc-tures provides a general system for makingnanometre-scale, two-dimensional sheets

of any shape, and with any desired surfacepattern. But given the flat raft architecture,

Figure 1 | DNA sculpture. Douglas et al.1 report a method for designing and constructing three-dimensional nanostructures from DNA. a, The computer-aided design process begins with a blockof tubes arranged in a honeycomb lattice. b, A template for the desired DNA structure is designed byremoving sections of the tubes, just like carving a sculpture from a block. The remaining tubes willbecome DNA duplexes in the final object. The DNA structure is designed by routing a single-strandedscaffold DNA (a virus genome) through every section of the tube template. Hundreds of short strandsof DNA are then designed to bind to the folded scaffold, cross-linking between different tubes andstapling together the overall structure. When the staple molecules are synthesized and mixed with thescaffold DNA in solution under appropriate conditions, they direct the folding of the scaffold into thedesired nanostructure. The structure shown here is more complex than those prepared by the authors(see Fig. 2 on page 416).

331

Vol 459|21 May 2009

NEWS & VIEWS

2009 Macmillan Publishers Limited. All rights reserved

8/3/2019 Thomas H. LaBean- Another dimension for DNA art

2/2

it is not easily used to make intrinsicallythree-dimensional objects.

Douglas and colleagues work1 represents athird revolution in DNA nanotechnology. Theyhave extended the DNA origami technique byshowing how a DNA scaffold strand can formlayers of helices arranged in a honeycomblattice, thus providing a general-purpose, crys-talline material from which three-dimensional

objects can be constructed. In principle, anyshape can be made from this DNA material, aslong as it can be carved out from a block of thehoneycomb lattice.

To design their nanostructures, the authorsdevised a computer-aided process that beginswith a template block composed of tubes(Fig. 1); each tube becomes a DNA duplex inthe final structure. Once a target shape hasbeen defined by removing sections of the block,a single-stranded scaffold DNA (the M13 virusgenome, as in flat DNA origami) is routedthrough every part of the structure, and com-plementary staple strands are designed to bind

to the scaffold and thus create duplexes. Finally,strand-exchange points are defined betweenneighbouring double helices. Enough of thesejunctions must be used to stabilize the overallstructure, while still maintaining enough flex-ibility in the system to allow the desired shapeto assemble. Having drawn up plans for theirtarget structures, Douglas et al.1 heated, thenvery slowly cooled, a solution of the scaffoldDNA and its hundreds of staples. Under theseconditions, the staples directed the folding ofthe scaffolds into the desired shapes.

Douglas and colleagues approach can becompared with a recently published procedure

for three-dimensional DNA origami7

, in whicha hollow box (42 36 36 nanometres) wasassembled. Two-dimensional DNA origamiwas used to construct all six flat walls of thebox on a single scaffold strand, and then inter-wall staple strands directed the assembly of thefinal three-dimensional form. The box designis highly innovative it even includes a lid thatcan be opened and closed but the box gainsits three-dimensionality by orienting intrinsi-cally two-dimensional subunits against oneanother in space. By contrast, the honeycomblattice technique1 is inherently three-dimen-sional from the start of the design process.

Of course, the primary goal of DNAnanotechnology is not to create aestheticallypleasing sculptures, but to make functionaldevices and materials. For practical applica-tions, structures generated using Douglas andcolleagues method will probably need to beintegrated with other nanomaterials that haveelectronic, photonic or catalytic propertiessuperior to those of DNA. There are currentlyalso other limitations to the technique. Forexample, the self-assembly process results inlow product yield (providing only about 744%of the theoretical yield), proceeds very slowly(taking about a week), and generates products

that have an unfavourably high charge density(because the charged DNA backbone is packed

tightly in space). Furthermore, the upper limitson the total size of the products and the lowerlimits on their feature resolution have yet to bedetermined. The shapes that have been madeso far are also somewhat blocky (see Fig. 2 onpage 416); the sculpture depicted in Fig. 1b ofthis article would require either a larger scaf-fold strand than is currently used, or severalsuch strands.

Nevertheless, the potential of Douglas andcolleagues technique is clear. Hierarchicalstructures, constructed from several repeat-ing subunits, are a much-sought-after goalof nanotechnology, and the authors presentthree examples in their paper, including astunning icosahedron assembled from three

M13 genome scaffolds (see Fig. 4 on page 418).This successful move into three dimensionsheralds a new era for the fie ld of structuralDNA nanotechnology. Thomas H. LaBean is in the Departments of

Computer Science, Chemistry and Biomedical

Engineering, Duke University, Durham, North

Carolina 27708, USA.

e-mail: thomas.labean@duke.edu

1. Douglas, S. M. et al. Nature459,414418 (2009).2. Seeman, N. C.J. Theor. Biol.99,237247 (1982).3. Seeman, N. C. Nature421,427431 (2003).4. LaBean, T. H. & Li, H. Nano Today2,2635 (2007).5. Winfree, E., Liu, F., Wenzler, L. A. & Seeman, N. C.Nature

394,539544 (1998).6. Rothemund, P. W. K. Nature440,297302 (2006).7. Andersen, E. S. et al. Nature459, 7376 (2009).

COMPUTATION

The edge of reductionismP.-M. Binder

Research at the frontier between computer science and physics illustrates

the shortcomings of the reductionist approach to science, which explains

macroscopic behaviour using microscopic principles.

In his 1972 paper More is different, PhilipAnderson1 claimed that multi-componentphysical systems canexhibit macroscopicbehaviour that cannot be understood from thelaws that govern their microscopic parts afeature known as emergent or complex behav-iour. Andersons position is at odds with that of

Stephen Hawking, who once suggested2

that,as soon as all fundamental laws of the Universeare understood, we will in principle be able toexplain all macroscopic phenomena. Writingin Physica D, Gu and colleagues3 provide abeautiful illustration of a physical system thatcannot be easily reduced, and of the develop-ing symbiosis between theoretical physics andcomputer science4.

To address the understandable, StephenWolfram5 examined the relation betweencomputation and the unfolding of the physicalworld. He defined as reducible those systemsfor which there is a computational shortcut

that allows their behaviour to be efficientlypredicted rather than reproduced step by step.For example, the motion of a simplependulumis described by a cosine function that can becomputed using a rapidly converging math-ematical series, rather than simulating eachand every pendulum oscillation. Such short-cuts do not usually exist for chaotic systems,for example.

Wolfram made an additional, importantpoint. Many systems are irreducible, butamong them only a few are undecidable: theyhave properties that cannot be formally cal-culated, as stated in Kurt Gdels and Alan

Turings theorems

6

.Undecidability is a propertyof universal computers or Turing machines.

Macs, PCs and DNA computers7 with unlimitedmemory would qualify as such machines.And this is where the notion of different(or complex) systems can be made more pre-cise those with undecidableglobal prop-erties despite having well-understood local(microscopic)governing laws.

As a first example of undecidability, considera cellular automaton (CA) a lattice of cells,each of which can take on a finite number of

values (states) and evolves over time accordingto the configuration of a set of neighbouringcells. This is the microscopic transition rule.Forthe one-dimensional CA known as elementaryrule 110,two states are allowed (0 or 1), andany cell will evolve to 0 if either its state and thatof its right-neighbour cell are 0, or if its stateand those of both its immediate neighbours are1 otherwise it will evolve to 1.Thus, the localgoverning law is fully understood.

But the global dynamics of a CA is a different

matter, as can be seen in Figure 1. Each rowdisplays the lattice at a different time step, thusproviding a full spatiotemporal record of thedynamics of the system.Cells far apart act inconcert to sustain particles8: structuresthatmove and interact, and in doing so, compute.The result is an intricate and undecidableglobal dynamics.

It is not easy to demonstratethat rule 110can simulate a universal computer9. Suchproofsoften involve the construction of a fewlogic gates and information channels that allowuniversal computation to be implemented, andcould well be argued to be reductionist. But

once these elements have been constructed, thestep that shows that a system has undecidable

332

NATURE|Vol 459|21 May 2009NEWS & VIEWS

2009 Macmillan Publishers Limited. All rights reserved