DNA, the medium of life, is so deeply associated with the biochemical world that considering its nonbiological applications may seem far-fetched. However, for researchers in the 1980s and 1990s working in the fledgling field of DNA nanotechnology, it was more than a flight of fancy.
Nadrian “Ned” Seeman, the father of DNA nanotechnology and formerly a biochemist at New York University, first proposed the idea that DNA is not only a genetic material but also a construction material in his seminal 1982 Journal of Theoretical Biology paper.1 A crystallographer by training, Seeman struggled to crystalize proteins. He wanted to build DNA cages that were strong enough to hold the protein in place long enough to take a great picture. En route to tackling this problem, Seeman created the field of DNA nanotechnology. “Ned had such a huge body of literature that he’s sort of everyone’s go-to inspiration source,” said Erik Winfree, a computer scientist and bioengineer at the California Institute of Technology (Caltech).
[Rothemund’s] demonstration was so elegant and thorough that it really opened people’s eyes to how powerful the idea was.
—Erik Winfree, California Institute of Technology
In the decades that followed, grand ideas transformed into even grander demonstrations as scientists repurposed nucleic acids to store nonbiological information, such as all 154 of William Shakespeare’s sonnets, and build microscopic robots for drug delivery.2,3
DNA computers
In the early 1990s as an undergraduate student at Caltech, Paul Rothemund, now a computer researcher there, took a computer science class that introduced him to the potential of DNA. The professor discussed an idea from Charles Bennett, then a physicist at IBM, that a computer built from DNA could simulate a Turing machine wherein hypothetical enzymes could read the information stored in DNA and use that information to alter DNA bases.4 “He said, ‘you know, someday, somebody who knows something about computer science and biology or chemistry will come up with a way to compute using DNA.’ At that moment, I said, ‘Well, maybe I can do that,’” recalled Rothemund.
He didn’t have to wait long. Rothemund had the opportunity to further explore the idea of building a molecular computer for a project in another class. There, Rothemund met Winfree, then a graduate student, and introduced him to Seeman’s work.
Erik Winfree (left), Bernard Yurke (middle), and Paul Rothemund (right) explore ways to expand the use of DNA.
Erik Winfree
As Rothemund neared the end of his studies, he wanted to continue working on the problem of building DNA computers, so he shopped the project around. Computer science professors felt ill-equipped to advise on the life science aspects of the project, so he turned to life scientists. “The biology professors at Caltech and elsewhere told me that I was crazy, and they had no idea what I was talking about,” said Rothemund. He hit a dead end, and following graduation in 1994, he joined a geobiology lab as a technician.
Later that year, Leonard Adleman, a computer scientist at the University of Southern California (USC), published a seminal paper in Science in which he used DNA to compute an algorithm.5 “I simultaneously felt sort of scooped and validated that there was something to the idea of encoding information in DNA molecules and doing computing,” said Rothemund.
Winfree recalled Rothemund’s class project on a similar topic and hunted him down to see if he wanted to attend the inaugural one day workshop on DNA-based computers that would be held at Princeton University in the spring of 1995. They scrounged up the money to attend the conference. Winfree’s talk touched on Seeman’s work on the self-assembly of DNA structures: the spontaneous organization of molecules due to attractive forces.
After returning to his seat, Winfree felt Seeman tug on his arm. Later that evening, Adleman, Seeman, Rothemund, and Winfree huddled over a red and white checkered tablecloth in a pizza parlor and reflected on the day’s events. “That sort of seeded the next 25 years of my life,” said Rothemund. He joined Adleman’s lab at USC later that year as a graduate student, while Winfree ventured to the east coast to collaborate with Seeman at New York University on the self-assembly of DNA crystals.6
Folding DNA
DNA computing captured the imaginations of many entering the nascent field of DNA nanotechnology by demonstrating a nonbiological application for nucleic acids. However, by the turn of the century, many researchers shifted their focus. “Many in the field convinced ourselves that although this was intellectually stimulating, this was not going to compete with electronic computers,” said Winfree.
Rothemund focused his efforts on building nanostructures using DNA, or programmable approaches for DNA assembly. Specifically, he considered how self-assembly processes could be treated as algorithms and studied using computer science tools.
In 1993, Seeman detailed the construction of complex nanostructures via the self-assembly of molecules.7 He was interested in figuring out how to design molecules that self-assemble to form parallel DNA helices where strands cross over and become part of another double helical line, thus stitching together the helices. Over the next decade, this inspired others in the field to innovate, increasing the number of crossovers and helices.
Rothemund and Winfree’s paths crossed again in 2001. Winfree returned to Caltech as a professor, and Rothemund joined his lab as a postdoctoral researcher, where they continued their work on algorithmic self-assembly of DNA.8
DNA is a versatile building block, but Rothemund described DNA as optically, biochemically, and electronically dead compared to other molecules and materials like quantum dots, carbon nanotubes, or antibodies. “But what it can do is you can use the information in DNA sequences to build structures, and then you can use that structure to organize those other things,” said Rothemund. Since the available methods for creating shapes out of DNA were laborious and time consuming, Rothemund set out to develop an easier approach. “And really, that was the idea for DNA origami,” he said.
For the first DNA origami experiments, Rothemund created one-third of a square. By adding only one-third of the required staple strands, only one-third of the square folded, resulting in rectangular shapes captured using atomic force microscopy. The “ss” label denotes unfolded single-stranded scaffold; “s,m” denotes a stable monomer; and “u,m” denotes an unstable monomer. The scale bar is 100 nm.
Paul Rothemund
Rothemund’s DNA origami consisted of two main components: a long, single-stranded piece of bacteriophage DNA, which serves as the scaffold material, and a bunch of shorter strands of oligonucleotides, or staple strands, that fix the structure in place.9 He fed his design into a computer program that used principles of Watson-Crick base pairing to determine which sequences were needed to instruct the scaffold strand to fold into the desired shape or pattern. In his one-pot method, Rothemund mixed the scaffold strand with the custom-made staple strands and waited patiently as molecular self-assembly took shape. Then, to confirm the structures, he used atomic force microscopy.
At the time, Winfree granted his lab members the freedom to independently explore their interests. “There was a period in 2005 when I didn’t see [Rothemund] around very much,” said Winfree. Eventually, Rothemund re-emerged with something to share. “He showed me his results on the DNA origami, and to tell you the truth, my first reaction was ‘Blech! Where’s the algorithm? Where’s computer science?’” Uninterested in collaborating on the project, Winfree suggested that Rothemund publish it himself, and so he did.
Winfree eventually came around to DNA origami. “It’s fantastic. It’s revolutionized the field,” said Winfree. “[Rothemund’s] demonstration was so elegant and thorough that it really opened people’s eyes to how powerful the idea was.”
DNA origami’s spiritual successors
DNA origami isn’t the only approach for DNA assembly, but it’s robust and relatively easy.10 “It’s the ability to create a geometrically structured testbed for your experiment with each molecule in the right place, and now, hundreds to thousands of molecules,” said Winfree. “That was unprecedented. That opened up an ability to do experiments in all sorts of fields that people previously couldn’t do.” For Winfree, that was putting short DNA sequences in the correct order to trigger self-assembly. For others, it was putting enzymes in the right order to produce a cascade of enzyme reactions or quantum dots in a particular organization to control optics.
“One of the things that DNA origami has been able to do since that paper was published is not something that’s widely used in everybody’s cell phone or something like that, but it’s a research tool for other things,” said Rothemund. He noted that these custom instruments for biology allow researchers to begin asking questions about proteins or other biomolecules and even translate these ideas into therapeutics and molecular diagnostics.11,12
I saw his talk, and my jaw dropped because these images that he produced—nobody had seen anything like this.
—William Shih, Harvard Medical School
William Shih, a biochemist at Harvard Medical School, employs principles of DNA origami in his quest to build nanoscale objects, including molecular robots. In 2005, at a conference in Albany, Shih learned what Rothemund had been up to. “I saw his talk, and my jaw dropped because these images that he produced—nobody had seen anything like this,” said Shih. He recalled a particularly memorable image of DNA origami with “Ned” patterned on top in homage to Seeman. Still mesmerized by Rothemund’s creations, Shih returned to his lab, scrapped what he was doing, and pivoted to Rothemund-style DNA origami.
“To me, what’s most special about DNA origami is that you have an excess of building blocks that do absolutely nothing except when they see a copy of this master controller scaffold strand,” said Shih. Ten copies of the scaffold strand produce 10 DNA origami structures if there are sufficient staple strands.
Around the same time that Rothemund tinkered with DNA at Caltech, Shih worked as a postdoctoral researcher down the road at the Scripps Research Institute. In 2004, he published a paper in Nature demonstrating the construction of a nanoscale octahedron.13 However, Shih’s DNA folding approach required the assembly of a substantial number of short, cloneable DNA structures. Shih noted that without a master scaffold strand, the construction is harder to control. “It’s kind of like herding cats,” said Shih.
Recently, Shih’s research group has been busy developing what he referred to as the spiritual successors of DNA origami. The scale of DNA origami structures is limited by the length of the scaffold strands, which is typically on the order of 10,000 nucleotides. Therefore, building anything bigger than that requires forgoing the leading scaffold strand. To address this, Shih and his team recently developed crisscross DNA origami whereby a controller molecule on the scale of a single DNA origami directs the construction of a larger 1,000 DNA origami structure.14 “They’re more like well-trained dogs than cats,” said Shih.
Shih hopes that these advances will facilitate the construction of bigger nanorobots with the size and complexity of a bacterial or mammalian cell. He views this as a complementary technology advancement to the broader field of synthetic biology where scientists modify the genomes of living cells. “But they’re still living cells,” said Shih. They’re still surrounded by a membrane; they still have metabolism; and they still do DNA replication. “It’s important, technologically, to have alternate schemes that maybe don’t have a membrane, that are not beholden to the normal process of DNA replication and translation, that maybe can be deployed in environments that are hostile to living cells,” said Shih.
References
- Seeman NC. Nucleic acid junctions and lattices. J Theor Biol. 1982;99(2):237-247.
- Goldman N, et al. Towards practical, high-capacity, low-maintenance information storage in synthesized DNA. Nature. 2013;494:77-80.
- Douglas SM, et al. A logic-gated nanorobot for targeted transport of molecular payloads. Science. 2012;335(6070):831-834.
- Bennett CH. Logical reversibility of computation. IBM J Res Develop. 1973;17(6):525-532.
- Adleman LM. Molecular computation of solutions to combinatorial problems. Science. 1994;266(5187):1021-1024.
- Winfree E, et al. Design and self-assembly of two-dimensional DNA crystals. Nature. 1998;394(6693):539-544.
- Fu TJ, Seeman NC. DNA double-crossover molecules. Biochemistry. 1993;32(13):3211-3220.
- Rothemund PWK, et al. Algorithmic self-assembly of DNA Sierpinski triangles. PLoS Biol. 2004;2(12):e424.
- Rothemund PWK. Folding DNA to create nanoscale shapes and patterns. Nature. 2006;440:297-302.
- LaBean TH. Reminiscences from the trenches: The early years of DNA nanotech. In: Jonoska N, Winfree E, eds. Visions of DNA Nanotechnology at 40 for the Next 40. Natural Computing Series. Springer, Singapore; 2023:55-67.
- Andersen ES, et al. Self-assembly of a nanoscale DNA box with a controllable lid. Nature. 2009;459:73-76.
- Ochmann SE, et al. Optical nanoantenna for single molecule-based detection of Zika virus nucleic acids without molecular multiplication. Anal Chem. 2017;89(23):13000-13007.
- Shih WM, et al. A 1.7-kilobase single-stranded DNA that folds into a nanoscale octahedron. Nature. 2004;427:618-621.
- Wintersinger CM, et al. Multi-micron crisscross structures grown from DNA-origami slats. Nat Nanotechnol. 2023;18(3):281-289.