Nucleic acid‐based nanotechnology

Abstract

DNA's ability to form Holliday junctions and double cross‐overs can be used to build complex two‐ and three‐dimensional nano‐scale structures for biotechnological applications and high‐precisions electronics.

DNA has often been described as the building block of life, a description that refers mainly to its role in storing and processing information. But DNA, as well as RNA, are also genuine building blocks per se that can form increasingly complex structures—loops, junctions, hairpins and so on. In fact, one of the advantages of DNA over RNA is that huge molecules can be compacted into a super‐dense storage medium that houses an enormous amount of information in each cell without consuming too much space needed by other functions. Its capability to self‐assemble makes DNA an ideal candidate for building nanostructures for applications both within and outside biology. Furthermore, the high precision of this assembly process based on nucleotide pairing lends itself to generating accurate molecular scaffolds with potential nanoscale applications, as well as for investigating the dynamics of biological reactions at atomic scale.

Its capability to self‐assemble makes DNA an ideal candidate for building nanostructures for applications both within and outside biology

The challenge for nanotechnology is harnessing these properties for crafting artificial structures, assembled by the same principles. Such attempts started in the early 1980s when Nadrian Seeman, a crystallographer at New York University, USA, who then came to specialize in DNA nanotechnology, conceived the idea of using DNA to build molecular scaffolds. His motivation was to develop a faster, more reliable and accurate way of crystallizing organic molecules than the existing trial‐and‐error approach. Seeman's idea was to build a simple DNA framework that would confine the molecule of interest in an exact position, effectively forming a single‐molecule crystal. He created DNA sequences comprising branched junctions with four arms, which became the foundation for making ever more complex DNA structures 1.

A crucial element for DNA nanotechnology is the Holliday junction, discovered in 1964 as an important intermediate structure during genetic recombination and in DNA repair

Seeman's second advance came in 1983 when he constructed the first 3D DNA lattices with six‐arm junctions, which, he claimed, were inspired by an M.C. Escher woodcut. Since, he and others have taken DNA assembly to ever‐greater levels of geometrical complexity focusing as much on topology as on molecular biology.

2D and 3D origami

A crucial element for DNA nanotechnology is the Holliday junction, discovered in 1964 as an important intermediate structure during genetic recombination and in DNA repair. It comprises four double‐stranded arms joined together around a central point. It was followed by the discovery of a related structure, the DNA double‐crossover molecule, another intermediate in DNA recombination, that contains two crossover links between helical domains. Seeman's team constructed a synthetic version of the DNA double‐crossover molecule, which became a harbinger of more complex DNA nanostructures 2. In particular, it led to DNA origami, developed by Paul Rothemund in 2006, which plays a major role in DNA nanotechnology today.

… scalable production is needed for preclinical studies and at this stage it looks like that could only be achieved by emulating nature

Rothemund's original invention was 2D origami, which he applied to create simple images about 100 nm in diameter 3 (Fig 1). One image, a smiley face taken with an atomic force microscope, graced the front cover of Nature. The process began with the scaffold strand, a long piece of single‐stranded DNA from a bacteriophage, which was stabilized by adding more than 200 short complementary oligonucleotides called staple strands. The resulting DNA molecule can be folded into a variety of shapes—hence the name origami—and fixed by annealing the staple strands to the scaffold. Rothemund and his team created different 2D shapes such as squares, discs and five‐pointed stars with a spatial resolution of 6 nm; in effect, each oligonucleotide is a 6 nm pixel for inscribing complex patterns, words and images.

Figure 1. 2D DNA origami shapes.

Top row, folding paths. (A) square; (B) rectangle; (C), star; (D), disc with three holes; (E), triangle with rectangular domains; (F), sharp triangle with trapezoidal domains and bridges between them (red lines in inset). Dangling curves and loops represent unfolded sequence. Second row from top, diagrams showing the bend of helices at crossovers (where helices touch) and away from crossovers (where helices bend apart). Colour indicates the base‐pair index along the folding path; red is the 1^st base, purple the 7,000^th. Bottom two rows, AFM images. White lines and arrows indicate blunt‐end stacking. White brackets in (A) mark the height of an unstretched square and that of a square stretched vertically (by a factor > 1.5) into an hourglass. White features in (F) are hairpins. All images and panels without scale bars are the same size, 165 × 165 nm. Scale bars for lower AFM images: (B) 1 μm; (C–F), 100 nm. From 3 with permission.

In 2009, William Shih's group at Harvard's Wyss Institute in the USA extended DNA origami to create simple three‐dimensional shapes from pleated layers of helices held together by a honeycomb lattice 4 (Fig 2). As with 2D origami, a key point was that the dimensions, in the range from 10 to 100 nm, were precisely controlled with the structures assembled on a hierarchical basis. The authors anticipated that this would offer a general method for manufacturing devices on the nano scale after further optimization.

Figure 2. 3D DNA origami shapes.

The first and second rows show perspective and projection views of cylinder models, with each cylinder representing a DNA double helix. (A) Monolith. (B) Square nut. (C) Railed bridge. (D) Slotted cross. (E) Stacked cross. Rows three to seven show transmission electron microscope (TEM) micrographs of typical particles. Bottom, expanded view of boxed area from above. (G) Left, typical monolith particle. Right, integrated‐intensity profile (red) of line orthogonal to the longitudinal axis of typical monolith particle, with expected profile (grey) assuming a simple homogeneous cylinder model. (H) Left, Gaussian‐fitted mean peak positions (circles) in such integrated‐line profiles for 20 different monolith particles as a function of peak index. (I) Analysis as in (G) repeated for the square‐nut shape. (J) Histogram of Gaussian‐fitted peak‐to‐peak distances as found for the square‐nut particles. Scale bars: (A–E) 20 nm; (F) 1 μm (top), 100 nm (bottom). From 4 with permission.

As Shih explained, the functional difference between single‐layer or 2D and multilayer or 3D origami can be compared with building structures out of paper or corrugated cardboard. “You can make a lot of useful macroscopic tools from paper, such as wrapping gift boxes, but sometimes you need materials that are more rigid, like corrugated cardboard, for example to make the gift box itself strong enough to protect the contents inside”, he said. “Alternatively, if you need to build a precisely shaped cavity or surface, you probably need to carve this out of a thick slab, therefore multilayer origami is a much more convenient starting material”.

In 2015, Hao Yan et al at Arizona State University, USA, created porous structures 5 by exploiting the DNA structure for novel properties rather than folding it into 3D shapes. While the nanotech structures in Shih's 2009 paper comprised tightly packed parallel helices, Yan's structures resemble wire frames connected to each other as required with a precision not attainable using traditional techniques, including lithography. It was this achievement that suggested the potential for integrated circuits with improved precision in locating components.

Folding control

But the time and effort involved in constructing these complex shapes remained a hurdle until Björn Högberg's group at the Karolinska Institute in Sweden developed a method to automate the sophisticated folding procedures. They combined mathematical graph theory to analyse the connections with simulation of the dynamics involved in folding towards a final stable state (known as relaxation in physics) 6 and developed a computer algorithm to calculate the origami‐folding pathway for the scaffold strand to create any desired shape in a mesh‐like pattern.

Hogberg's team then used DNA origami for studying the dynamics of biological pathways. “The patterning capability allows us to also put conjugated molecules, for example protein ligands, in very precise locations at the surface of our structures”, Hogberg explained. “We are currently using this to investigate how spatial organization of ligands might effect juxtacrine (cell‐cell contact) signalling systems. In a sense, one can imagine many of these systems acting as a kind of cell based braille system for blind reading, but instead of reading dots on a paper, cells ‘feel’ patterns of ligands in neighbouring cells”. They applied this to study the ephrin/Eph system, which is involved in regulation of various processes, including stem cell differentiation and cancer and comprises membrane‐bound proteins that require direct cell–cell interactions to activate signalling. “We found that the exact spacing of ligands had a significant impact on both receptor activation and functional outcome. In the breast cancer cell‐line we tested, their invasive properties were drastically reduced by just decreasing the distance between ligands on the origami from 100 to 40 nm”, Hogberg commented. “This way of looking at effects of ligand spacing is quite unique, because we display the patterns directly from solution and can compare distance effects decoupled from overall protein concentration. […] Of course, this type of patterned structure could also potentially be used therapeutically, possibly in combination with a similar type of drug loading […]”.

However, scalable production is needed for preclinical studies and at this stage it looks like that could only be achieved by emulating nature. “Relying on synthetic DNA is fine for most demonstrator studies, but for animal experiments for example, we are going to need a lot more material and the requirements for purity of the DNA are going to be increasingly stringent”, Hogberg explained. “Biology has evolved extremely sophisticated methods for producing sequence controlled polymers that are very hard to rival for synthetic chemistry. For most researchers who mainly use DNA oligos for PCR, the quite common errors in synthetic oligos are generally not a problem, but for us, to bring these techniques out of the test‐tube, we need new methods that harness the power of biological production both for quality and scale”.

Another problem is making DNA nanostructures that are stable in physiological conditions, which again is essential for animal studies. “We have in my lab for example worked on alternative ways to build origami using a looser polygonal web of DNA”, Hogberg said. “Because of the way these are designed, they are stable in physiological salt conditions as opposed to earlier versions of origami. […] Protection from nucleases and clearance in‐vivo are other issues and here some of the more recent techniques, such as coating with PEG (poly‐ethylene glycol)‐polylysine molecules, seem very promising for future applications that potentially could go into live animals”.

Scaling up production

Hendrik Dietz at Technical University in Munich, Germany, developed a method to address the production problem; as he explained, using nanostructures as research tools, complex materials or as novel therapeutics would require amounts of materials beyond the scope of current production methods. Their process is based on bacteriophage‐derived production of single‐stranded precursor DNA in which target strand sequences are interleaved with self‐excising DNAzyme cassettes. DNAzymes or deoxyribozymes are the DNA analogue of protein or RNA enzymes that cleave specific selected sequences, including bonds. Dietz applies them to create temporary supports while building DNA structures, which are then removed. This has enabled end‐to‐end production and final assembly of macroscopic amounts of a DNA origami nanorod in a litre‐scale stirred‐tank bioreactor. Dietz is convinced this work is well on the way to solving the mass production problem, but this still leaves other challenges, especially for ex vivo applications. “While there is certainly potential for non‐biological applications, the challenge is to stabilize DNA nanostructures for use in non‐native conditions, in air, in organic solvents and so on”, he explained.

Another major focus with biomedical potential is precise insertion of molecules on the surfaces or in the pores of DNA nanostructures. A team under Kurt Vesterager at Aarhus University in Denmark recently demonstrated covalent immobilization of antibodies in cavities in both 2D and 3D DNA origami structures, with the ability to precisely control the orientation of the molecules 7. “By locking the antibody in the cavity, the origami could serve as an engineering template to introduce specific chemical groups to specific sites of the antibody at other positions than the coordination sites”, Vesterager commented. This has immediate diagnostic potential through creation of homogeneous surfaces of immobilized antibodies for both diagnostic and therapeutic applications. Yet, it would still need to increase stability of DNA origami structures in vivo, Vesterager added. “This is being intensively investigated by several laboratories and the first potential solutions here are appearing”.

“I think we'll see some disruptive single‐molecule diagnostic and analysis tools within the next decade”, Shih said. “For therapeutics, it will take longer for most applications because interacting with living systems is so complicated and unpredictable. The biggest challenge is to decrease the incidence of self‐assembly errors, or else to create more error‐tolerant architectures. Cells have had to evolve many layered strategies for dealing with similar errors, such as proofreading polymerases, excision repair, recombination‐based repair and proteasome garbage collection. By analogy we'll have to expand our bag of tricks accordingly. Getting progressively better at this will keep us busy for decades”.

RNA nanostructures

Similar to DNA, RNA origami was pioneered by Rothermund in partnership with Ebbe Sloth Andersen et al at Aarhus University, Denmark 8. “RNA behaves very differently from DNA as a building material”, Andersen explained the rationale for this work. “RNA origami differs by folding a single strand into a well‐defined nanoscale shape, whereas DNA origami uses a large amount of helper strands, so called staple strands. One of our main reasons for going with the dogma of folding a single strand of RNA is that it allows RNA nanostructures to be folded during synthesis by the RNA polymerase, called co‐transcriptional folding, which in turn will allow the RNA nanostructures to be produced in cells. Co‐transcriptional folding further allows higher yield than assembly from multiple parts and is also the way nature does it when producing complicated shapes such as the ribosome. DNA origami on the other hand uses heat annealing for self‐assembly, which is not compatible with cells and if produced in vitro provides a delivery challenge if required to work in cells”.

DNA nanowires could, for instance, be used for computing devices even denser than what is achievable by conventional optical lithography…

RNA origami does bring challenges too, such as designing nanostructures that are compatible with the co‐transcriptional folding process. “Here we needed to consider kinetics of folding and topological issues”, Andersen explained. “In RNA origami we mainly have hairpins, junctions and kissing‐loops, the formation of which needs to happen in the right order not to ‘tangle up’ and cause a topological problem, that is a non‐covalent knot”. It does offer biomedical applications too. “RNA nanostructures can be used for nanomedicine by creating RNA nanoparticles with different functionalities. One of the main ways is to incorporate si (small interfering) RNAs that knockdown a certain disease‐causing gene”, Andersen commented. “However, I am more interested in the synthetic biology applications, where we use the RNA nanostructures to provide new properties to cells or to reprogram properties. In synthetic biology, possible applications are scaffolds for enzymes that may improve production of a valuable chemical, or for regulating gene expression in a controlled fashion”.

Application in electronics

Non‐biological applications focus on both optical and electronic properties of DNA nanoparticles to produce sensing devices, or components for digital storage or processing. DNA nanowires could, for instance, be used for computing devices even denser than what is achievable by conventional optical lithography, the size of which is constrained by wavelength of the illuminating light. “In the 1990s and 2000s it became clear that lithography will hit its limits at some point and computer chips can't get smaller anymore”, explained Tim Liedl at Ludwig‐Maximilians‐Universität in Munich, Germany. “One approach was to explore the use of DNA‐assembled wires”. The idea was to use DNA wire meshes as templates for assembling silicon chips of gold or other metal nanowires. However, the promise of greater densities has not been realized, Liedl acknowledged, given the difficulties of fine‐tuning chemical growth of gold, or metals in general, below dimensions of 10 nm. “Current lithography manages control to basically that same length scale, so DNA‐templating of metal wires is not to become better than state‐of‐the‐art lithography”, Liedl said.

However, DNA‐based methods allow greater precision, which could lead to improved performance at a given chip size. “We are able to position nanoparticles with very high precision, better than lithography, and also the quality of the gold crystals is better if self‐assembled instead of deposited on surfaces”, Liedl explained. “We have recently shown that we can even position different material particles with single‐nanometre‐precision and in this way build a loss‐less plasmonic passage, as an alternative to electronic wiring” 9.

DNA is potentially a viable replacement for existing archival storage mechanisms, such as tape backup

This could lead to all‐optical chips with a fundamental leap in performance as light travels faster than electrical signals and dissipates less energy as heat, thereby avoiding the problem of cooling along with saving power. This is based on plasmonics, which arises from the interaction between an electromagnetic field and free electrons in the orbitals of metal atoms or molecules, to generate collective oscillations that can be exploited for optical effects or to transfer energy with high efficiency. Plasmons can be pictured as waves of electron charge density, which can be excited by incident light. Liedl's team has generated a nano assembly of two gold nanoparticles either side of a tiny island of silver. Under light stimulation, plasmons on the gold nanoparticles exhibit resonant coupling and form a coherent system within which energy can be transferred at approaching the speed of light.

While it will be some time before plasmonic devices make any impact in computation, dense information storage on DNA is closer, according to Andrew Phillips, Head of Microsoft's Biological Computation Group. “DNA is potentially a viable replacement for existing archival storage mechanisms, such as tape backup”, he commented. “DNA is extremely high density, and since it is present in every living organism, technology for reading and writing DNA will not become obsolete. If the costs of manipulating DNA continue to decrease at current rates, it is likely that archival DNA storage will become commercially viable in the next few years”. The current challenge is the slow speed of reading and writing data to a DNA device, which will restrict applications to archive systems where capacity is more important than performance. Yet, it may lead to applications that actually run faster by exploiting the ability to address the DNA molecule directly. “Once information is stored in DNA, there may be opportunities to conduct elementary computation on this information directly at the molecular level”, Phillips explained. “For example, one could envisage molecular circuits that perform search queries or that execute machine learning on data encoded in DNA directly, though this will depend on the specific ways in which the information is encoded and on the form in which the DNA is stored”.

High‐precision sensors

In the near term though, the most promising applications are high‐precision detection of molecular compounds in disease diagnosis, which requires accuracy rather than processing speed. Phillips described how so‐called DNA domino circuits, comprising DNA molecules positioned at regular intervals on a DNA origami tile, could be adapted to detect combinations of RNA strands indicative of a given disease in patient samples. “This would require changes to the DNA domino design to accommodate binding of RNA signals rather than DNA, and modifications to the circuits to decrease their susceptibility to degradation by enzymes”, he explained. “This could be achieved by replacing the flat DNA origami tile with an enclosed DNA structure, to allow small molecular signals to flow through while offering some protection against enzyme degradation, or potentially by enclosing the molecular circuits inside vesicles that are porous to the molecular signals being detected. […] This could ultimately enable selected killing of cells that possess a distinct molecular signature, for example indicative of certain types of cancers, without harming nearby healthy cells”.

DNA self‐assembled nanostructures have clear potential for a wide variety of applications; the immediate challenge is now to identify those that are commercially or clinically viable in a reasonable timeframe.