New software being developed at Ohio State University will allow creation of more complex DNA robots, at much faster speeds.

The software, called "MagicDNA ," will help designers take tiny strands of DNA, form small rotor and hinge parts, and combine them into complex structures that can move and complete a variety of tasks, including drug delivery.

Researchers have been designing DNA-based machines for a number of years — with slower tools and with tedious manual steps, said Carlos Castro, co-author of the study and associate professor of mechanical and aerospace engineering at Ohio State .

“Nanodevices that may have taken us several days to design before now take us just a few minutes,” Castro said.

The team's study was published in an April edition of Nature Materials .

What is a DNA Robot?

The “MagicDNA” software allows more complex design, according to the Ohio State team, by working primarily at the level of the links, joints, and ssDNA connections between bundle components.

A kind of “DNA robot” that supports tasks like drug delivery consists of one or more stiff parts. The shape and stiffness of these components can be precisely controlled depending on the number and arrangement of double-stranded, or dsDNA, helices.

To make dynamic DNA machines or robots, these stiff components can be connected together by hinges that allow rotation, or sliding joints that allow for translation. These joints connecting the components are made via flexible single-stranded DNA (ssDNA).

“For example, a hinge joint can be formed by connecting two stiff bundles by a few short ssDNA connections arranged along a line,” said Castro. “A sliding joint can be formed by fitting together two complementary bundle geometries such as a cylinder inside a hollow tube."

The hinge and slide components are similar to the rigid links and joints that are used to make complex machines.

In practice, one full structure is typically made up of two types of DNA strands, says Castro: 1) a long ssDNA scaffold strand (measuring about 7000-8000 nucleotides) and 2) many shorter ssDNA strands referred to as “staples.”

The staples pair with the scaffolds to form the bundles of double-stranded DNA, and the joints are formed from the parts of the scaffold that are left single-stranded.

MagicDNA allows researchers to carry out an entire DNA design in 3D, and the routing of underlying strands is automated.

"With this new approach and software tool, we envision establishing a library of parts that could be easily incorporated into new designs while maintaining the possibility of making custom parts and integrating many standard and custom parts into new designs as needed," Castro told Tech Briefs.

Previously, DNA-made devices could only feature about six individual components and they had to be connected with joints and hinges for complex motions, said study co-author Hai-Jun Su , professor of mechanical and aerospace engineering at Ohio State.

“With this software, it is not hard to make robots or other devices with upwards of 20 components that are much easier to control," said Prof. Su. "It is a huge step in our ability to design nanodevices that can perform the complex actions that we want them to do.”

Some of the complex devices created by the OSU lab included robot arms with claws that can pick up smaller items, and a hundred nanometer-sized "airplane" structure, 1000 times smaller than the width of a human hair.

“We want to be able to design robots that respond in a particular way to a stimulus or move in a certain way,” said Castro.

Castro expects that the MagicDNA software will be used at universities and other research labs for the next few years, but its use could expand in the future.

“There is getting to be more and more commercial interest in DNA nanotechnology,” said Castro. “I think in the next five to 10 years we will start seeing commercial applications of DNA nanodevices and we are optimistic that this software can help drive that.”

In a Q&A with Tech Briefs below, Castro reveals how both a "top-down" and "bottom-up" approach to DNA design will lead to exciting commercial applications.

Carlos Castro, OSU, Ohio State
Carlos Castro

Tech Briefs: To demonstrate the robotic complexity that is now possible, can you compare a previous DNA robot application with what you envision a robot with, say, 20 components able to do?

Carlos Castro: Current versions of robots typically have one or two moving components. With this approach we can make devices with many parts and overall more complex geometries. Applications of previous robots demonstrated manipulation of one or two components, for example, bringing two molecules together that can undergo a chemical reaction. With more components we could build devices that could manipulate (i.e. grasp, move, release, orient) many components, including molecules, nanoparticles, or other nanomaterials, for example, to build other nanoscale devices or systems. In addition, where prior robots carried out a single function, new robots with many components could combine many functions into one system such as sensing, information processing, actuation, and sequestration or release of molecules.

Tech Briefs: When you are designing with strands of DNA, can you treat it like any other material that you would simulate in a design program? Are there any challenges in simulating DNA strands in a design?

Carlos Castro: You can treat it like any other material from the standpoint that the physical properties and local structure basically determine the overall properties. It is still challenging because the physical models used in simulations are still not perfect. They can predict some aspects like structure very well, but there are still limitations in predicting for example, range of motion, or stimulus response for dynamic devices.

Furthermore, the process of connecting simulation results to design iteration is different, mostly because the design constrains are different. For example, in typical design if the stiffness of a structure is not sufficient, you could change the material. But for DNA devices the choices are more limited, basically dsDNA and ssDNA, so properties are often modulated through changing the geometry. In addition, we are often limited on the overall amount of material we use, so if we make a structure longer, it will likely have to be reduced in width. There is a lot of work ongoing to integrate DNA devices with other materials that could expand or improve the functionality.

Tech Briefs: Once you have the design, what is the process like to create something like a tiny robot arm with claws that pick up small items. How difficult is it to create a nanotechnology once you have the design?

Carlos Castro: Once we have a design, the fabrication is carried out by a process called self-assembly. That means we basically mix all the DNA strands that make up the design into a solution and we rely on thermal fluctuations (motion of the water molecules in solution driving random motions of the molecules) to drive interactions between the DNA strands.

The process is controlled by temperature. Usually the self-assembly reactions are heated up to 65-70 oC to melt all the interactions, and then the solution is cooled slowly to give the strands time to find their correct binding sites. Once we have a design, the process is quite robust, but the yields can vary greatly, from a few percent on the low end to near 100% on the higher end. In most cases, relatively simple structures assemble with high yield, and complex devices fold with lower yield. But even more complex devices can often get to yields of better than 50%, and there are methods to purify well-formed structures away from misfolded structures. There is still ongoing research in the field to better understand the details of the self-assembly process to be able to improve the yields and speed up the assembly process.

Tech Briefs: What kinds of commercial applications do you envision with DNA Nanodevices?

Carlos Castro: There are a large range of potential commercial applications. Some examples could include: drug delivery devices for cancer therapeutics or other disease applications; biosensors to aid in disease diagnostics; nanopores for new sequencing sensing applications; new targeted contrast agents for medical imaging; and new approaches to genetic engineering that are complementary to or build on technologies like CRISPR. Beyond biological applications, there is exciting recent work showing DNA nanostructures can be used to organize components that could be used for electronics, energy storage, and optics.

Tech Briefs: What have been the main limitations of DNA-robot design that MagicDNA software addresses?

Carlos Castro: There are a few main advances that this software and design process enable. One is it allows carrying out the full design process in 3D. In prior approaches, designs were mapped onto 2D strand routing diagrams. That is a very useful way to visualize and manipulate parts of the design, but it is very challenging to arrange and connect many components correctly in 3D.

In addition, there were prior tools that are either bottom-up (i.e., arranging and manipulating strands to build up a structure) or top-down (i.e., inputting a geometry and automating the strand routing). Here we combined the benefits of both approaches. Hence, we can build complex overall geometries following a top-down approach while maintaining fine control on local parameters that are important to control properties like flexibility or reconfiguration.

Furthermore, a key step was to connect our design process with simulation feedback. As we are increasing the complexity of DNA robots, it is harder to predict how they will behave or even some specific parameters of what they will look like. Hence, getting simulation feedback is very useful. This also open a door to incorporate consideration of more advanced function like motion, reconfiguration, mechanical properties, and interactions with other materials, into the design process, where previously design was based mostly on geometry.

What do you think about DNA-based machines and robots? Share your questions and comments below.

What do you think? Share your questions and comments below.