Beyond Genetics: DNA in Nanotechnology

The discovery of DNA's role as bearer of heredity and biological information shook biology to its core and ushered in the age of molecular biology¹, a revolution whose coattails we are still riding. The ability to reprogram cells with recombinant DNA is a powerful tool for basic researchers and bioengineers alike. More easily overlooked is the fact that DNA has a very simple and predictable chemistry, compared to the likes of proteins, giving it great potential for applications in nanotechnology ranging from miniature computers to capsules for delivering drugs.

Building with DNA

In his famous book What is Life?², physicist Erwin Schrödinger prophetically speculated that the molecular basis of genes is an aperiodic solid, with atoms arranged in a code that builds organisms. DNA was later identified as this very molecule. This "variations on a theme" chemistry, where each part of the DNA polymer is chemically very similar, except for the choice of one of four bases (A, C, T, or G), allows easy and increasingly cheap assembly of custom DNA. What's more, DNA's strict base-pairing rules, where A binds to T and C to G, makes it possible to design molecules that fold into shapes or interact in predictable ways. By thinking in three dimensions (or programming a computer), you can create a one-dimensional DNA sequence that will spontaneously fold into three-dimensional shapes stabilized by A:T and G:C base pairing.

An entire field has emerged from this collision: the predictability of DNA chemistry, computational design (biological CAD), inexpensive DNA synthesis, and in vitro DNA technologies like PCR and DNA sequencing. DNA origami³, a particular approach to building nanoscale DNA structures, is an easily understandable approach that has produced dramatic results. The method relies on a long piece of single-stranded DNA (ssDNA) called the scaffold and many short pieces of ssDNA called staples. Staples hybridize with the scaffold at multiple places, pulling distant pieces of the scaffold together and forming stable double helices. The staples can be created very inexpensively and in great abundance through DNA synthesis, but the much longer scaffold is more problematic. The answer is to use the genome of a ssDNA virus such as the M13 phage, which infects E. coli and can be propagated in great abundance.

Engineering at the level of individual DNA molecules demands a close attention to detail. A single 360° twist in the double helix contains 10.5 DNA bases and is 3.6 nm long and 2 nm wide. That's an astonishingly small 1/35,000 of the width of a human hair. In addition to considering spacing, it is crucial to consider the rotation of the helix. Leaving parts of the scaffold single-stranded allows for unique shapes formed by the weaving of the scaffold back and forth. Fail to account for the twisting of the DNA helix itself, and you lose control of the planes in which the scaffold goes single-stranded, a twisted knot instead of an elegant pattern. Keeping these properties in mind, and with the assistance of CAD tools, a nearly endless collection of nanoscale DNA shapes are possible (see the figure!).

DNA origami is a simple and elegant method, but it is also fundamentally restricted. To make large, even more complicated shapes requires a longer scaffold, but we are limited by the sizes of ssDNA viral genomes. Additionally, each DNA origami project requires going back to the drawing board and designing new staples. In contrast to the DNA origami approach, human-scale construction projects use many small pieces (think bricks, shingles, boards, etc.) arranged to construct large designs. This modular approach has already been miniaturized in the form of DNA tiles, where DNA helices fold into small rectangles of dsDNA with ssDNA overhangs⁴. By carefully designing the overhangs you can determine how a collection of tiles will connect together.

This two-dimensional modular tile design has recently been thrust into three dimensions with a field-shifting publication⁵ from researchers at Harvard's Wyss Institute. The researchers created a collection of 32-base DNA blocks. Each block is divided into four 8-base pieces which can bind to other blocks. An entire toolbox of DNA blocks can be created relatively inexpensively since the price of synthesizing short pieces of ssDNA is cheap and still falling. Amazingly, the paper includes over 100 unique three-dimensional shapes created by this method, from cubes to boxes, letters of the alphabet, numbers, and pegs. This flexibility could revolutionize DNA nanotechnology devices.

DNA Nanocomputers

From my previous examples, it might seem like DNA nanotechnology's utility begins and ends with clever structures like smiley faces and letters of the alphabet. However, DNA nanotechnology, like synthetic biology, has a strong computer science undercurrent.

In the 1990's, Leonard Adleman founded the field of DNA computing, which uses the unique physical properties of DNA to build miniature computers. In the field's inaugural publication, Adleman used short pieces of ssDNA to model the Hamiltonian Path Problem (HPP)⁶. The HPP is an important problem in computer science because there are no known efficient algorithms to solve it, and because it is entangled in the P vs. NP problem (MIT News has an informative and entertaining explanation), which has a $1 million bounty on its head. Here's the gist of the HPP: A traveler wants to see a huge list of sites around the world, but transportation from city to city is limited. E.g., he can fly from Rome to London, and London to Beijing, but not from Rome to Beijing. The question is, given a list of cities, can our traveler get from his starting city to his ending city (maybe the same place), visiting every city in the list exactly once? If so, then there is said to be a Hamiltonian path. There might be many Hamiltonian paths for a given scenario.

Adleman approached this problem by encoding the HPP into DNA. Each city is assigned a 20-base DNA sequence, and each connection between cities is created by concatenating the last 10 bases of the sequence from the origin city and the first 10 bases of the destination. When mixed together, the DNA molecules hybridize to form longer strands. If there is a Hamiltonian path, then on an agarose gel a band of size 20(n + 1) bp will appear. The diagram shows how Adleman's method can be applied to solve a trivial HPP: travel from Rome to Beijing, visiting London, and with permissible routes as Rome to London and London to Beijing. Encoding the problem requires seven 20-base ssDNA molecules. Four encode paths, and three the cities. The pieces of DNA anneal to form a 20(n + 1) = 80 bp product. If, say, the London-->Beijing edge is removed, then the trip is not possible and only shorter bands will appear on a gel, indicating no Hamiltonian path. Adleman's method is elegant in its simplicity, but its scope is limited. Yet it demonstrated for the first time that computation is possible at the scale of individual biological molecules.

A completely different method of computing with DNA has recently emerged in prominence, which again relies on DNA's ability to base pair, but unlike Adleman's method relies on the dynamic rearrangement of DNA strands. This approach is called strand displacement^7,8, because it relies on one ssDNA wedging into a double helix, forcing one of the strands to become a new ssDNA. This creates an input/output system, where ssDNA inputs create ssDNA outputs. Fluorescent reporters can even turn terminal ssDNA inputs into an optical output that can be read with an instrument. Researchers have created very complex networks of ssDNAs binding to double helices and freeing new ssDNAs, which go on to bind ssDNAs or interrupt other double helices. This approach has even been used to compute square roots⁷ and encode neural networks⁸, feats that I admit escape my understanding.

It's not clear what role DNA computing might play in future technologies. At the time of the publication of Adleman's paper, the fastest supercomputer in the world operated at 10¹² operations per second. He calculated that his method could plausibly scale to roughly 10²⁰ DNA operations. Therefore, in one experiment he could perform so many calculations that it would take a supercomputer 76 years to do the same. Today's supercomputers, however, have gotten much better. The best operate on the order of 10¹⁶ operations per second, which means that 10²⁰ calculations take a mere 2.8 hours. Since Moore's Law shows no signs of slowing down, it is hard to imagine a scenario where DNA computers would rival traditional computers in terms of sheer computational power.

DNA and Nanomedicine

As we get better at building with DNA, the number of applications of DNA nanotechnology will grow. One intriguing possibility is DNA nanomedical devices, which could carry small payloads of drugs, antibodies, or other proteins. DNA is relatively stable in the blood, and DNA nanodevices could take off from established assembly techniques like DNA origami and modular DNA building blocks.

The central goal of nanomedicine is selective drug delivery. Chemotherapy is so taxing on cancer patients because the chemotherapy drugs flood the entire body. Delivering the drugs to cancer cells only would mitigate side effects and allow for a stronger, more directed attack against the offenders. It is not immediately obvious how this kind of selectivity could be programmed into DNA, since in cells changes in DNA are caused by other molecules, usually proteins. However, there is extensive literature of modifying the shapes of RNAs with small molecules or even proteins. These shape changes can be coupled to the regulation of gene expression. In these applications, the RNAs are called aptamers, and the same techniques can be applied to DNA aptamers.

New aptamers can be created with rapid directed evolution in the laboratory⁹. Aptamers that bind well to a substrate, say caffeine, are separated from a large pool of RNAs or DNAs (depending on the application), and are rewarded with being copied. The copying has a high rate of mutation, allowing for the creation of a pool of new aptamers, some of which might be better than the initial. Repeating this process through many iterations yields aptamers that bind the substrate with high efficiency.

Aptamers might have applications in medical diagnostics, by generating aptamers that bind to pathogens or diagnostically informative molecules. But, returning to the problem posed above, they enable DNA nanodevices to sense their environments.

These concepts have already been put together into a prototype. A DNA nanorobot has been built using DNA origami, which releases a drug payload only if its aptamer sensors recognize pre-designated small molecules or proteins¹⁰. The robot is a hinged box which, when closed, has the shape of a hexagonal prism. The box is latched closed with two aptamer hinges, and inside are 12 docking sites for therapeutics (the researchers used gold nanoparticles or antibodies). The therapeutics are kept inactive by virtue of being locked inside the closed robot. In the paper, the authors tested various combinations of antigens produced by a collection of cancer cell lines. The aptamer sensors in general did an excellent job controlling the release of cargo in response to the cancer antigens. Above at right are electron micrographs of the nanorobot carrying payloads. Nanomedical devices like this might someday enable much more effective cancer treatments with reduced side effects.

Summing Up

My earliest memories of the concept of nanotechnology conjure up images of manipulating single atoms to make new devices. DNA brings a priceless addition to this enterprise: self-assembly. DNA's regularity in base pairing rules makes it a feasible building material for self-assembling nanodevices, constructed by simply mixing together their components in the correct ratio. An array of computational tools, a variety of assembly strategies, and the ability to make DNA functional by imbuing computational ability on the DNA or employing aptamers could make DNA a powerful tool for nanotechnology.

Image Credits:

DNA structure: Madprime (via Wikimedia Commons)

DNA origami schematic: Eric Sawyer (free to use with attribution)

DNA origami shapes and micrographs: Fig. 2 from Ref. 3

HPP Sample Problem: Eric Sawyer (free to use with attribution)

Nanorobot: Adapted from Fig. 1F from Ref. 10

References:

1. Brenner, S. The Revolution in the Life Sciences. Science 338, 1427–1428 (2012).

2. Schrödinger, E. What is Life? Cambridge University Press, 1944. [Free PDF]

3. Rothemund, P. W. K. Folding DNA to Create Nanoscale Shapes and Patterns. Nature 440, 297–302 (2006).

4. Rothemund, P. W. K., Papadakis, N., & Winfree, E. Algorithmic Self-Assembly of DNA Sierpinksi Triangles. PLoS Biology 2, e424 (2004). [Open Access]

5. Ke, Y. et al. Three-Dimensional Structures Self-Assembled from DNA Bricks. Science 338, 1177–1183 (2012).

6. Adleman, L. M. Molecular Computation of Solutions to Combinatorial Problems. Science 266, 1021–1024 (1994).

7. Qian, L. & Winfree, E. Scaling Up Digital Computation with DNA Strand Displacement Cascades. Science 332, 1196–1201 (2011).

8. Qian, L., Winfree, E., & Bruck, J. Neural Network Computation with DNA Strand Displacement Cascades. Nature 475, 368–372 (2011).

9. Ellington, A. D. & Szostak, J. W. In Vitro Selection of RNA Molecules that Bind Specific Ligands. Nature 346, 818–822 (1990).

10. Douglas, S. M., Bachelet, I., & Church, G. M. A Logic-Gated Nanorobot for Targeted Transport of Molecular Payloads. Science 335, 831–834 (2012).