Imagine a sensor smaller than the tiniest cell. As it encounters a predefined set of molecular cues, it fires, releasing a drug or another small molecule.
It sounds like science fiction, but according to a new publication in Nature Nanotechnology, such devices are theoretically feasible.
Ido Bachelet of Bar-Ilan University in Israel, and colleagues describe a set of “DNA origami robots” that, when mixed in various combinations, function as biological logic circuits, capable of responding to the presence or absence of various molecular triggers. 
These “robots,” aren’t robots in the sense of Wall-E. They’re DNA molecules that fold into complex and intricate shapes. Upon receiving the proper signals, they change shape, producing a measurable output.
The robots were made in several forms. Some are “effectors” capable of releasing a “payload,” while others act as positive or negative regulators that modify the effectors’ behavior. When combined in specific stoichiometric ratios and combinations, these robots can mimic the behavior of any of seven discrete “logic gates”: AND, OR, XOR, NAND, NOT, CNOT, and a “half-adder.”
George Church, Professor of Genetics at Harvard Medical School, who was Bachelet’s postdoctoral advisor and was senior author on a 2012 paper by Bachelet describing the first version of these DNA robots , explained that such circuits could theoretically be used to build sensors that respond to complex molecular signals – for instance, an insulin release mechanism that responds not only to glucose but also to lipid levels, cytokines, and other cues.
In this case, the team’s circuits respond to either VEGF or PDGF (or both) on cockroach hemolymph cells by exposing a fluorescently labeled antibody. The AND gate (which fires if both of molecules are present) is the simplest example. The presence of both PDGF and VEGF (but not either molecule alone) causes the robot to open and reveal its payload, an event the authors monitored using FACS.
An XOR gate (which fires if one molecule is present, but not both) is more complex, containing an effector that can bind both PDGF and VEGF, positive regulator robots that supply keys to open the unopened half of the effector robot if only one signal is received, and a negative regulator that deactivates the effector in the event both signals exist.
The authors’ most complicated circuit comprises six robots—two effectors, three positive regulators, and one negative regulator—but more elaborate designs are feasible. They describe a hypothetical system that could produce any of four distinct outputs based on the presence or absence of four protein cues, and speculate that far greater scaling theoretically is possible.
Erik Winfree, Professor of Computer Science at the California Institute of Technology, warns that enormous engineering advances will be needed to realize such possibilities. “One should never oversimplify the challenges that need to be solved in order to get molecular circuitry working well in a scalable architecture, especially when unknown biological complexity is involved,” he wrote in emailed comments to BioTechniques.
Unlike electronic logic gates that open and close repeatedly, the molecular robots Bachelet describes can fire only once (that is, they cannot be reset), severely complicating system scaling. “The Commodore 64 had over 256,000 bits of memory and a CPU with something like 4,000 transistors operating at 1 MHz. Amir et al. demonstrated a system with two digital logic gates that can act only once. The gap is unspeakably large”—but not necessarily insurmountable, Winfree added.
Church says the new study represents a significant advance over his team’s earlier work. Where the 2012 study described only a single logic gate, the new study describes seven. Also the earlier study was done in cells, while the current study shows that these robots can function in a live animal. The authors injected their robots (and trigger molecules) into cockroaches, then extracted the hemolymph a short while later. FACS analysis of the isolated cells demonstrated that the robots retained function under those conditions.
“To get it to work in an animal, you have to make sure the DNA structures are stable, that they’re not neutralized by the complex internal components of that animal... So I think that’s a pretty big step,” Church said.
 Y. Amir, et al., “Universal computing by DNA origami robots in a living animal,” Nat Nanotechnol, April 6, 2014. doi:10.1038/nnano.2014.58.
 S.M. Douglas, et al., “A logic-gated nanorobot for targeted transport of molecular payloads,” Science, 335:831–4, 2012.