1 Discovering DNA Computing

DNA-based nanotechnology, for me, began in 1994 when Len Adleman’s paper “Molecular Computation of Solutions to Combinatorial Problems” came out [1]. I was at a meeting out west, and people were talking about this new Science paper at lunches and dinners. It was the buzz subject of social gatherings for the week. Adleman is the “A” in the RSA encryption standard, and here, he was dabbling in actual biochemistry in order to prototype the encoding and processing of non-biological information in DNA molecules for what seemed to be the first time ever. Adleman used a “generate and sort” strategy to solve an NP-complete problem (i.e., the Hamiltonian Path Problem) using a library of DNA molecules as the scratchpad upon which a library of possible solutions was written. Contrary to many popular press reports at the time, Adleman did not solve the Traveling Salesman Problem; he answered the question “is there a Hamiltonian path” through this specific 7-node, directed graph, not “what is the shortest Hamiltonian Path?” His system was designed such that biochemical laboratory steps could be used to sort through the molecules and reject those with incorrect answers recorded upon them (i.e., invalid or non-Hamiltonian paths through the graph). Through Adleman’s paper, molecular computing or DNA-based computing was introduced to me and to a broad scientific audience. Of course, by then, Ned Seeman had already conceived of and been toiling within the field of DNA nanotechnology for over a decade; it was just that he toiled mostly alone and outside the eye of a large audience, and certainly outside of my notice, up until then. DNA nanotech entered into my attention and seemingly into the wider public attention through the door of computer science (CS), through molecular computing.

I didn’t join the DNA nanotech community until 1998. At the time, I was a postdoc in Jane and Dave Richardson’s lab working on de novo protein engineering. We were testing our knowledge of the rules of protein folding by trying to design well-folding proteins from scratch. We were deeply involved with the Diversity Biotechnology Consortium (DBC) including Jane and Dave, plus Stu Kauffman (my Ph.D. advisor and pioneer of “self-organization in complex systems”), Mario Geysen (the father of “mimetopes”), Frances Arnold (world-famous protein engineer/evolver who would later be awarded a Nobel Prize), Andy Ellington (who actually coined the term “aptamer”), Pim Stemmer (who invented “sexual PCR”), and many other prominent folks with whom I had the good fortune to interact. The DBC was busy furthering the quite new field of combinatorial chemistry within the biological realm by designing clever selection techniques and applying them to libraries of random-sequence biopolymers (i.e., polypeptides and polynucleotides) in order to winnow the vast populations of possible polymers within sequence space and find/evolve functional, individual sequences for specific purposes. This focus was a direct, logical follow-on from my Ph.D. project in which I randomly searched amino acid sequence space for polypeptide strands that were able to fold into compact, globular 3D structures as do many evolved, biological proteins [2,3,4,5].

2 Connections to Broader Scientific Themes

The more general context of my work at the time was that combinatorial chemistry was sweeping the pharmaceutical industry and revolutionizing drug discovery. Then, 1990 was a very big year, in which new tools emerged that allowed combinatorics and in vitro evolution to successfully break through into biochemistry and structural biology. The various biopolymer sequence spaces could now be explored and mapped. Specifically, Ellington and Szostak [6] and Tuerk and Gold [7] described the formation of randomized nucleic acid libraries and in vitro selection of RNA molecules with specific sequences that provided self-folding and affinity binding to a variety of chemical targets, while Scott and Smith [8] did the same thing for polypeptides via the implementation of phage-display libraries. In the late 80s, we were pursuing similar goals in Stu Kauffman’s lab, but our early genetic strategies turned out to be too recombinogenic, and thus stable subclones could not be effectively isolated and propagated.

On a personal front, I was divorced in 1996 and had a clause added to the separation agreement that my sons remain in Durham for the indefinite future, so that my ex-wife could not try to move my sons away from me. While this worked fine for my family life, it put a severe crimp in my professional plans, since I was unable to perform a national search and pursue a tenure-track, Assistant Professor position with any geographic freedom. Moving into DNA nanotech enabled instead a slow but steady climb up through the Research Professor track at Duke. I was fortunate to finally land a tenure-track faculty job as Associate Professor at North Carolina State University in 2011, be awarded tenure in 2012 and be promoted to full professor in 2017. Professionally (as well as scientifically), DNA nanotech has been a great field to work in, and it has allowed me to follow an unconventional career path in which I have never been an Assistant Professor.

Back in 1998, the invitation to abandon de novo and library-based protein design and start designing novel nucleic acid structures came when John Reif walked across Research Drive in the middle of the Duke campus looking for some biochemist to teach him about DNA and molecular biology. Another postdoc in Jane’s and Dave’s group said there was a crazy, theoretical computer scientist (redundant (?)) looking around for someone to talk to. I taught John a lot of very basic stuff like what ‘ligase’ was, but he was a quick learner and was offering a decent salary to me if I were to dump Biochemistry and join the forces of Computer Science. I ended up making a deal with the Richardson’s, and we maintained our wet lab operations in the Duke Biochemistry Department for a number of years until our DNA team grew too large and started crowding Jane and Dave in their own space.

Shifting gears from protein engineering to nucleic acid engineering was not a difficult move. In fact, a researcher must be comfortable with the manipulation of DNA, through synthetic chemistry, molecular biology, and microbiology procedures in order to hope to do experimental protein design. Moving to DNA structural engineering actually increased the speed of work by eliminating many gene expression and protein purification headaches and other tedious steps in the macromolecular design cycle.

3 Ned Seeman: Founder of the Field

The official start of DNA nanotech (as is reflected in the title of this volume) was actually in 1982. Ned Seeman’s paper in the Journal of Theoretical Biology [9] challenged people to think about DNA as a structural material instead of as a genetic material. The concept was to assemble periodic matter from DNA for various uses such as guest–host systems for docking protein molecules to reliably form 3D crystals for use in X-ray diffraction studies to solve the atomic-scale structures of the guest proteins. The way Ned frequently explained it in later years was that he was trying to succeed as a crystallographer but was failing to produce high quality protein crystals; therefore, no crystals meant no crystallographer, so he had to try something novel. Probably most people, like me, did not know anything about the 1982 JTB paper when it first came out. I was an undergraduate at the time the paper was published, and it would not come to my attention until about sixteen years later. My first introduction to Ned’s work was learning about his closed, geometric-like, or topological objects such as the DNA cube and truncated octahedron, the family of double-crossover tiles, and finally the 2D DNA crystals imaged on mica by atomic force microscopy [10].

In 1998, John Reif organized a team that brought major DARPA funding to the nascent field of DNA nanotech. I signed on at that point. At my first DNA Computing conference, DNA4 in Philadelphia, I got to visit my old stomping grounds at UPENN and to meet many of the characters who were busy developing the discipline of DNA nanotech. Seeman’s lab at NYU was, of course, a hot spot that was producing the first rounds of young experimental pioneers in the field including Junghuei Chen, Chengde Mao, and Hao Yan. In those early days, people typically were coming into DNA nanotech through either chemistry/biochemistry/molecular biology or through CS and were joining the community to learn more about the opposite wing of the field. People who came in through the CS door included Grzegorz Rozenberg, Len Adleman, Anne Condon, Richard Lipton, Natasha Jonaska, and Lila Kari. They worked on widely varying topics including word/code design, encoding strategies, simulation or theoretical models of DNA computing strategies for addressing demanding CS problems (NP-hard or NP-complete problems), or simpler problems like molecular implementations of Boolean logic, databasing, cryptography, etc. It was also exciting to meet people like physicists Andrew Turberfield and Bernie Yurke who were fluent in both experimental and theoretical languages. These thinkers expanded the use of DNA in several directions including not only as a structural material but also to function as fuel for these new molecular machines [11].

At the DNA4 meeting in 1998, I first met Erik Winfree, mature for his age and obviously brilliant, he was a graduate student who understood all the scientific background implicitly, acted socially with the grace and ease of a senior academic, and was responsible for establishing the idea of algorithmic assembly at the center of the burgeoning field of DNA nanotech. Erik would soon continue “a family tradition” and receive a MacArthur Prize as did his father, Art Winfree (sidenote: Art was a friend of my Ph.D. advisor Stu Kauffman, so I knew his name and his work long before meeting his son, Erik. Also, I have had the privilege of working directly under two MacArthur fellows, Kauffman, as well as Jane Richardson). The application of algorithmic assembly to DNA computing was a revolutionary concept. Prior to that, the “generate and sort” strategy predominated, in which all or many possible solutions were created within a molecular library, and then, the set was sorted biochemically by discarding incorrect or suboptimal solutions to the problem, similar to what Adleman did in 1994. On the other hand, algorithmic assembly was a molecular implementation of Wang tiling, a theoretical computing model in which colored tile edges specify allowed assembly rules and a small tile set is capable of generating very large, complex, programmed patterns. Algorithmic assembly allowed implementations in which only correct answers would be assembled in the first place. This opened the field to a whole world of new possibilities and pushed creative thinking in diverse directions.

Directly following DNA4, John Reif and I rented a car and drove Ned back to New York where we visited his lab for a couple of days. I enjoyed the distinct privilege of going up to Ned’s apartment in NYU faculty housing and seeing exactly what you would expect from a dedicated bachelor/scientist in Manhattan: a room lined with overflowing bookshelves and a single reading chair in the middle of the room. I was amused to note that the well-worn reading chair had leaked a small pile of crumbled polyurethane foam that rested at the angle of repose right where it had fallen out of the back of the aging chair. It just looked like he didn’t waste time with unimportant distractions like decorating or cleaning. My other fun memory of Ned came a couple years later when he was talking at a meeting somewhere in Europe (I think), and he whipped his belt off from around his waist to use it as a visual aid for demonstrating DNA supercoiling. The funny thing was his joke announcement, something like: “Sorry to disappoint you, folks, but that’s as far as it goes.” There have been so many personal and professional memories surrounding lots of characters in the field of DNA nanotech, and I have very much enjoyed being part of this community.

At DNA4, I was just joining the DNA nanotech club, observing and learning, but by DNA5 in 1999, I was deeply embedded in the field and co-authored three papers at that conference. One of my contributions included the first use of the term “scaffold strand” to indicate a long strand around which other oligos would assemble and generate a structure larger than a standard “tile” [12]. I took Seeman’s concept of a “reporter strand,” a strand that is ligated together only within the context of the desired, assembled structure, and inverted it by preassembling the long “scaffold” so that it could act as a nucleation element for a larger superstructure. The concept of scaffold strands was further developed in [13] where a “fully addressable” 2D structure was illustrated and proposed. We also proposed another structural variant of a scaffolded assembly that same year [14]. Years later, Paul Rothemund mentioned to me that these early uses of scaffold strands and schematic proposals for scaffolded structures were used against him as pre-existing technology when the US patent office first turned down his patent application describing DNA origami, a big mistake, in my opinion, on the part of the patent examiner.

4 Personal Milestones

Among the milestones of DNA nanotech that I am proudest to have been a part of, I would include our 2000 demonstration of cumulative XOR computation using assembly of TX tiles [15]. This was the first published experimental demonstration of a molecular computation by algorithmic self-assembly using DNA tiles. Winfree and others would create more impressive control and computational scale soon enough [16], but it was fun to be on the cutting edge however briefly. This community has been highly collaborative and cooperative even while also being competitive.

While we were still using labs in the Biochemistry Department, I managed to lure “Ned’s best student,” Hao Yan to join us at Duke in 2001, and thus ensued a number of highly productive years, often centering around the “golden hands” of my first graduate student, Sung Ha Park. At that time, Sung Ha could get any experiment to work; he calmly solved many complex, sticky experimental problems that other people banged their heads against unsuccessfully. Hao was known around the group as “the finisher” because he could see, plan, and execute the final experiments necessary to finish a story and complete what was needed for the next publication. Hao then moved on in 2004 to set up his own research group at Arizona State, and somehow that felt like the “early years” of the field were starting to come to a close.

Another major event of 2004 was the publication of William Shih’s DNA octahedron [17]. This was the first time I had ever heard William’s name, but not surprisingly, thereafter, I would never forget it. A few extremely impressive things about the groundbreaking accomplishments of that report include: self-folding (via thermal annealing) of a long molecule with the assistance of a few “helper strands,” production of the nanostructure by DNA polymerase (rather than solid-phase chemical synthesis of short oligonucleotides), and surprisingly detailed cryo-EM structure evaluation. These attributes went against the grain of many long-held tenets and traditions of the field and presaged several revolutions to come.

In 2006, I was again proud to be at the cutting edge of the field by helping to create what was at the time, the “largest” (since we were trying to construct nanostructures of increasing size), fully addressable 2D array; these were 80 × 80 nm grids with 16 pixels that were “turned off and on” by programmable binding of avidin protein molecules [18]. We compared a couple of hierarchical assembly strategies and employed the tile-lattice method. Our accomplishment and “world record” held for only a couple of months and was soon shattered by the advent of DNA origami.

Paul Rothemund’s description of DNA origami in 2006 changed all the rules of DNA nanotechnology [19]. First of all, a single-author, cover-of-Nature paper was relatively anomalous but not too surprising once you got to know Paul and his body of work which was and is uniformly groundbreaking, intellectually deep, and widely divergent in topics. The design constraints that the field had been laboring under including: (1) Religious re-use of a small number of specific nucleotide sequences at Holliday-junction-like strand exchange points for crossovers (previously everyone only used variants of the J1 junction sequence first worked out by Seeman). Paul showed that essentially any sequence would work. (2) Exact strand stoichiometry matching during assembly reactions was no longer an issue; excess staple strands effectively folded “all” of the available scaffold. (3) Impure oligonucleotides (i.e., standard desalt from IDT instead of laborious, in-house, PAGE purification of each individual oligo) did not affect assembly yield. These changes may not seem like much from today’s perspective, but at the time, they arrived as a major, earth-shattering revolution. It also felt like almost everybody shifted gears and got into the origami game, including quite a few people who came into DNA nanotech on the wave of relative ease with which origami structures could be adapted to different uses, modified, and even designed anew from scratch. However, completely redesigning an origami cost several thousand dollars back then, so the basic cost of a new architectural design increased significantly versus the old repetitive, tile-and-lattice strategy. There also came an immediate echo from China that heralded the presence of another brilliant, self-motivated student when Lulu Qian rapidly and independently designed and assembled an origami in the shape of China [20]. Extension of DNA origami into all types of massive and fantastic 3D shapes has come from William Shih’s group, especially during the Douglas, Dietz, Liedl, Högberg years, and continued subsequently in their individual labs and those of many others.

5 The End of the Early years

In attempting to restrict this missive to a reasonable length and also to the “early years” of DNA nanotech as stated in the title, I will quickly point to a few more important developments and then conclude. Another key signpost for the field came when single-strand tile or the SST strategy was implemented and first published in 2008 by Peng Yin [21]. Peng had been gone from the Duke group for a while by then, but this work resulted from his time in Durham. We had been working on a project that Sung Ha Park and I referred to as “tile-less lattice” where individual oligos would assemble upon the growing lattice without first forming tile-like units. The most successful design using this concept was finalized by Peng while he was at Caltech and became the programmed tube circumference strategy which subsequently became the 2D [22] and finally 3D SST DNA bricks or DNA Lego architectures which are now a successful subfield of DNA nanotech. Major efforts are now under way to build ever larger self-assembled DNA materials including the incredible 2D multi-origami, hierarchical structures of Lulu Qian’s group and wireframe tessellated 3D structures from several groups. Progress has also been breathtaking on functional DNA hydrogels for soft robotics by Rebecca Schulman and others. Single-stranded RNA origami has been a thing since 2014 [23], and I have recently had the good luck to co-founded a startup called Helixomer, to commercialize an RNA origami-based anticoagulant (with reversal agent) for human health care. There are a number of startup companies now bringing DNA nanotechnology in various forms to the marketplace.

Following its founding by Ned Seeman in 1982, the field of DNA nanotechnology has grown up through the efforts of a large and expanding community of researchers. Ned also founded the International Society for Nanoscale Science, Computation, and Engineering (ISNSCE) (in which I am currently serving as President). The society has sponsored two conferences per year for a long time: DNA-X focuses on computational aspects of DNA nanotech while FNANO draws in a larger community of people studying self-assembly in nucleic acids but also in other molecular systems. I believe that there have already been 26 annual DNA-X meetings (of which, I have been to 11) and 18 FNANO meetings (of which I have been to 17). FNANO has always been held at the Snowbird resort in Utah, partially due to John Reif’s lifelong love of downhill skiing. DNA-X has been held alternately in North America, Europe, North America, Asia (repeat) for quite some time (until the pandemic forced it online in 2020). Traveling to the DNA-X conferences, as well as to work in the labs of collaborators in Denmark and China, has provided me with fantastic opportunities to explore the world. Memories of hanging out with the Italians in Japan, for example, still bring a smile to my face. These travels have led to lifelong friendships.

During this journey through the world with DNA nanotech, I have managed to amuse myself at various times by embedding personal symbols and souvenirs or unnecessary cultural references within the scientific literature. Some examples include a photograph of William S. Burroughs that I put into a cryptography paper, the phrase “sub-Lilliputian holiday” I tucked into a book review, and a molecularized likeness of the Venus de Milo placed in a news-and-views piece. I think people have to play small games like that just to keep things interesting and as minor acts of mischief even if the only one in on the joke is yourself. If someone else had written this type of piece, it would almost certainly highlight a different subset of people and events, but this is the way I remember the unfolding of DNA nanotech. I apologize if anyone is offended or hurt by anything I have written or failed to write in this short memoir-style paper. My intention is simply to record some of my memories for posterity, and I am not trying to cause trouble, attack/insult anyone, or grind any particular axes. When I received the invitation to contribute to this volume, the editors described a colorful array of possible essay types that they were imagining and hoping to include. This sort of personal history and reminiscence was the option that caught my interest. I am hoping to be able to read similar musings from other longtime members of the DNA nanotechnology community because, of course, my idiosyncratic points of view about events and developments through time are necessarily limited. Still, I hope this slim contribution adds to the overall historic document that the editors have bought into the world.

This volume is now imbued with added poignancy since Ned Seeman passed away during the month before the due date of this manuscript. His creativity, irreverence, and intelligence will be sorely missed in our community.