Introduction: Design, Assembly, and Verification of a 2D DNA Origami Nanostructure

About: My name is Aaron Berliner. I studied bioengineering, control theory, optimization theory, synthetic biology, systems biology, nanotechnology, artificial neural networks, and some microelectromechanical system …

DNA origami is a method of using DNA as a construction material to create two- and three-dimensional structures at the nanoscale. This method involves taking a long single strand of DNA as a base material and using many smaller "staple" strands to fold the long strand in specific places. This technique was published in 2006 by Paul W.K. Rothemund at Caltech, and has since grown into a popular and easy to use method for building structures at the nanometer scale.

The Autodesk A

The Bio/Nano Group at Autodesk Research are working with world-class researchers and industry experts to create the next generation of CAD tools to enable the design and manufacturing of these incredible nanoscale structures. In order to better understand the barriers faced by researchers trying to advance this field of science, the Bio/Nano Research group at Pier 9 set out to ‘Imagine, Design, and Create’ the Autodesk ‘A’ in the nanoscale.

  • For more information on DNA Nanotechnology, contact Dr. Joseph Schaeffer.
  • For more information on this specific DNA Origami project, contact Aaron Berliner.

Step 1: Design in Cadnano

Standard DNA Origami is designed in a program called cadnano.

cadnano simplifies and enhances the process of designing three-dimensional DNA origami nanostructures. Through its user-friendly 2D and 3D interfaces it accelerates the creation of arbitrary designs. The embedded rules within cadnano paired with the finite element analysis performed by cando, provide relative certainty of the stability of the structures.

In order to demonstrate the complete process of design to assembly and verification, we elected to construct a DNA Origami nanostructure of the Autodesk Logo.

The target design was constructed in cadnano version 2. The design is composed of 39 rows of a single scaffold arranged in 1 column on a square array.

The design was exported as a cadnano .json file containing the structure information and a .csv file containing the staple strand sequence data for ordering.


A scaffold length of 3818 nucleotides was needed for this design, so we used the M13mp18 viral ssDNA as the scaffold sequence. While this sequence is 7249 nucleotides, the unused portion should not interfere with the construction though we realized later that it would be better to use it completely. The scaffold was routed through the design using the mid-seam auto-scaffold method with the mid-seam located between position [135] and [136]. The length of the Autodesk logo ribbon was set to ~40-50 basepairs. The mid-seam of the design divided in half to create the division in the A and the scaffold of each ribbon is therefore composed in the raster style. Because the scaffold is composed in a single column, the only scaffold crossovers occur in the proceeding and following cell of the square grid.


111 staple strands were generated through the cadnano auto staple function and manual editing. The auto staple function is used initially and produced staples with long lengths. These staples were broken up manually using the cadnano nick function to produce staples between 20-50 basepairs with at least 4 bound basepairs adjacent to crossovers. After these operations, we ended up with staple strands with an average length of 36.5 basepairs with a minimum length of 18 basepairs and a maximum length of 50 basepairs. In order to avoid stacking interactions at the outer edges of the construct, four nucleotide basepair tails were created using a poly-T sequence. Due to the cadnano design lattice corresponding to a particular orientation of the DNA helices, it was necessary to adjust where to place the entire raster design on that lattice in order to optimize the number of staple cross overs and thus increasing the stability. We found that for this design, setting the midpoint of the construct to position [135] worked well.

Step 2: Molecular Modeling and Visualization

After designing in cadnano, we wanted to visualize our designs. We encountered problems when trying to convert the cadnano representation of the nanostructure into an atomic representation for visualization in the standard VMD software. As a result of this struggle, we built a small module to translate cadnano .json files into the .mmcif format, and then used that to extend the Autodesk Molecular Viewer to accept cadnano file uploads. These mmcif files are the new industry standard for atomic representation and can be used for visualization of large scale models which could not be handled by the old PDB models. The 'A' design was converted to a .mmcif file by simply uploading the cadnano .json file into the Molecular Viewer via the upload portal. Using the viewer was helpful in determining where each staple stand was located in relationship to the larger model itself.

Step 3: Simulation in CanDo

Once we had designed and visualized the nanostructure, we wanted to simulate the origami to verify that our design was robust. Since DNA origami is a very flexible material, the crossovers in the design are used to make the structure more rigid. There are a lot of different ways to place the crossovers, and in order to tell whether a structure will have rigid pieces we used the CanDo web app for simulating our design.

The 3D solution shape and flexibility of DNA nanostructures are predicted based on the assumption that the mechanical response of the DNA double-helix is well approximated by a homogeneous elastic rod with axial, twisting, and bending moduli that have been measured experimentally using a variety of direct, single-molecule, as well as ensemble techniques [8]. Double-strand crossovers are modeled as rigid links connecting neighboring helices that are initially positioned on either a honeycomb or square lattice, providing internal constraints that deform DNA from its straight, rod-like conformation to complex shapes as shown in detail in [4, 7-8]. The model also accounts for the effect of backbone nicks in DNA strands, entropic elasticity of single-stranded DNA used to design, for example, tensegrity structures [10], and distant crossovers required to model wireframe structures [4]. Each of these model features is approximate, however, and therefore remains an active area of research. Computational prediction of deformed DNA shapes is performed using the Finite Element Method implemented in the commercial software program ADINA (ADINA R&D, Inc.), which is a well established numerical technique for the analysis of complex structural mechanics and dynamics [11]. The thermally-induced fluctuations of DNA nanostructures are computed using the equipartition theorem of statistical mechanics and normal mode analysis, as shown for proteins in [12-13]. Atomic models of DNA nanostructures are generated from 3D solution shapes and thermal fluctuations, as shown for the design of light-harvesting nanodevices [14], DNA casting molds for inorganic structures [15], and lattice-free DNA structures [16].

The resultant cadnano files were uploaded to CanDo for solid-body simulation with following parameters.

-- CanDo submission --<br>
Date and time: 2015-06-29 16:23:49
Name: Aaron Berliner
University/Institute/Company: Autodesk
Filename: aNANO_2D_6_29_helixEnds.json
File type: square
Axial rise: 0.34
Helix diameter: 2.25
Crossover Spacing: 10.5
Axial stiffness: 1100
Bending stiffness: 230
Torsional stiffness: 460
Nick stiffness factor: 0.01
Model: fine
Movie: yes
Atomic model: yesWithMovie
Sequence filename: aNANO_2D_6_29_helixEnds.csv
Logged in: TRUE

CanDo returned solid-body simulations with a final displacement range of between 1.6 and 6.2 nanometers. The simulation was illustrated by both figures and movies. In the first iterations of this design, the displacement range exceeded ~50 nanometers (recall that DNA helices are about 2 nanometers in diameter, and the entire design here is about 120nm tall and 60nm wide at the A's join). This was reduced greatly by iterating the placement of the crossovers, and our resulting CanDo simulations demonstrate that no single helix or small helical subunit displays large displacements. Qualitatively, a large global twist is placed on the structure, but following Rothemund's results, the experimental results are expected to remain flat on a mica surface on AFM.

Step 4: Ordering DNA and Assembly Kit

Now that we had designed, visualized, and simulated the nanostructure, we wanted to build it! Before we could go into the laboratory, we needed to order the right chemicals and DNA. Nowadays, DNA Origami can be constructed through a kit which makes the process really easy.

Scaffold Strand DNA and Assembly Kit

We ordered the scaffold and assembly kit from Tilibit Nanosystems in Munich, Germany. We used the standardt M13mp18 scaffold DNA which is found in Folding kit basic, type p7249 and costs ~€140 ($150).

The contents of the Folding kit basic, type p7249 include

  • 500 μl 100 nM single-stranded scaffold DNA, length 7249 bases
  • 500 μl tilibit 10x folding buffer XM
  • 500 μl 200 mM MgCl2 stock solution
  • 600 μl tilibit 6x gel loading dye

The kit is good for 25 reactions. Note that the folding buffer here is essentially the same as TAE buffer.

Staple Strand DNA

We ordered the staple stands from IDT-DNA. For our design, 111 staple strands were ordered as 111x 25 nmole dried pellets in 2x 96-well plates for $770.07 on July 24, 2015 and arrived on July 29, 2015. These samples will need to be resuspended in MQ water during assembly. The DNA arrived and was stored at -20 °C. These 111 staple stands are found in the .csv file created by cadnano. Ordering is as simple as copy+pasting sequences from this file into the IDT portal.

Step 5: Stoichiometry

While we waited for our orders to arrive from IDT and Tilibit, we needed to do some stoichiometry to determine the exact amounts of chemicals and DNA to use in our experiments.

Given the reactants:

From Tilibit:

  • 500 μl 100 nM scaffold
  • 500 μl 10X folding buffer
  • 500 μl 200 mM MgCl2

From IDT

  • 111x 25 nmol staples in 2x 96 well plates

Assume a final reaction of

  • 1X buffer
  • 12 mM MgCl
  • 20 nM scaffold
  • 200 nM each staple

Assume that 1 uL of each resuspended staple will be added to a solution with a final volume of 200 uL.

Calculate the volume of MgCl2 needed to produce a 12 mM concentration in a 200 uL solution given 500uL of a 200 mM solution.

M1V1 = M2V2<br>(12 mM)(200 uL) = (200 mM)(V2)
(12e-3 M)(200e-6 L) = (200e-3 M)(V2)
V2 = (12e-3 M)(200e-6 L) / (200e-3 M) = 1.2e-5 L = 12 uL

Calculate the volume of scaffold needed to produce a 20 nM concentration in a 200 uL solution given 500 uL of a 100 uM solution.

M1V1 = M2V2<br>(20 nM)(200 uL) = (100 nM)(V2)
(20e-9 M)(200e-6 L) = (100e-9 M)(V2)
V2 = (20e-9 M)(200e-6 L) / (100e-9 M) = 4e-5 L = 40 uL

Calculate the volume of buffer needed to produce a 1x concentration in a 200 uL solution given 500 uL of a 10X buffer solution.

Remember that 1X buffer is composed of

  • 5 mM Tris
  • 1 mM EDTA
  • 5 mM NaCl

and thus the 10X buffer is composed of

  • 50 mM Tris
  • 1 mM EDTA
  • 5 mM NaCl

This helps us calculate the volume. So take any of the components of buffer and calculate the volume of Tris needed to produce a 5 mM concentration in a 200 uL solution given 500uL of a 50 mM solution.

M1V1 = V1V2<br>(5 mM)(200 uL) = (50 mM)(V2)
(50e-3 M)(200e-6 L) = (50e-3 M)(V2)
V2 = (50e-3 M)(200e-6 L) / (50e-3 M) = 2e-5 = 20 uL
Or we could just realize that 10X to 1X is a dilution of 1/10!

Now calculate the final reaction components given a final volume of 200 uL:

<p>1X buffer = 20 uL<br>12 mM MgCl = 12 uL
20 nM scaffold = 40 uL 
200 nM each staple = 1 uL each staple
20 + 12 + 40 + (111 * 1) = 183 uL
we need 200 - 183 = 17 uL water</p>

Now calculate the staple resuspension volume and concentration given the 1 uL distribution volume and 25 nmol mass of each staple.

<br>25 nmoles / X L water = Y concentration
M1V1 = M2V2
(200 nM)(200 uL) = (1 uL)(M2)
(200e-9 M)(200e-6 L) = (1e-6 L)(M2)
M2 = (200e-9 M)(200e-6 L) / (1e-6 L) = 4e-5 M = 40 uM
M2 = Y concentration = 40 uM
X water = 25 nmol / 40 uM = 25e-9 mol / 40e-6 M = 6.25e-3 = 625 uL water

Step 6: Assembly

Once all the chemicals and DNA arrived and the stoichiometry was complete, it was time to assemble the nanostructures. We used the following methods.


  1. Create 1X Buffer Add 20 μl 10X buffer to 1.5 ml tube
  2. Resuspend staples Add 625 μl water to each staple in 96 well-plate
  3. Pool staples Add 1 μl per staple to 1.5 ml tube.
  4. Pool scaffold Add 40 μl scaffold to 1.5 ml tube
  5. Add Water Add 17 uL water to 1.5 ml tube.
  6. Add MgCl Add 12 μl of MgCl
  7. Thermocycle Anneal from 95 °C to 20 °C in a PCR machine at 1 °C/minute in 0.1 °C steps

Important Notes

  • It was important to add the MgCl last because of the interactions between the cations and the DNA
  • It was important to heat the top of the thermocycler to 105 °C to avoid temperature gradiants in the system. If the temperature of the bottom of the thermocycler is high and the temperature of the top of low, the system will evaporate on the bottom and the salts will plate out of solution, leading to an unwanted white crystalline substance appearing on the bottom of the tube

Untested Alternative Methods

While we used an off-the-shelf thermocycler, there are several other ways you could do the annealing step. One would be to assemble your own and use it, keeping in mind the heating note above to make sure you don't get evaporation. Or, it is probably possible to just assemble the origami by annealing the DNA in hot water! Mix the DNA and kit chemicals as described above but instead of using a thermocycler to anneal the DNA, boil water (100 °C) and add the tubes. Make sure they are completely submerged and turn the heat off on stove. Let the water cool down with the tubes submerged. This should mimic the thermocycling process, but it will be less precise about the exact temperature ramp.

Step 7: Gel Electrophoresis

Once we had finished our assembly process, we needed to verify if we had successfully created the nanostructures. We used a gel electrophoresis technique to verify the existence of our origami.

Creating Gel

We created a 2% agarose gel by adding 2% agarose to the buffer media and microwaving it for 60 seconds on high. After microwaving, we poured the gel into a gel box and added the gel comb to produce wells for loading samples.

Loading and Running Gel

First we mixed 5 uL of 10-kilobase ladder, staples, scaffold, rx1, rx2 (two different aliquots of annealed origami) and a control origami each with 5 uL of the Tilibit loading dye. The gel was loaded with 2 uL of the [10-kilobase ladder, staples, scaffold, rx1, rx2, and a control origami] + dye solution. The gel was run for 60 minutes. The resultant gel was then imaged under UV light and captured as a photograph. We saw strong bands for both the rx1 and rx2 annealed origami, right where we'd expect for a fully assembled origami with some staples remaining (remember, they're in 10X excess so that's expected as well).

Step 8: Atomic Force Microscopy

Once we had verified the existence of the nanostructures through gel electrophoresis, we were very happy that our process worked, but we wanted to see our results. And so we used an atomic force microscope.

Atomic-force microscopy (AFM) or scanning-force microscopy (SFM) is a very high-resolution type of scanning probe microscopy (SPM), with demonstrated resolution on the order of fractions of a nanometer, more than 1000 times better than the optical diffraction limit.

Substrate Preparation

We first needed to prepare the slide for the AFM which is made of smooth, polished mica. First, using the magical tool, scotch tape, we cleaned the surface of the mica by sticking the adhesive to the mica surface and promptly ripping it off. The surface was visible more shiny after several cleaving processes. Upon the surface we added 20 uL of solution composed of 2 uL of the assembled DNA origami and the 18 uL of the buffer.

Using the Atomic Force Microscope

Once the substrate was loaded onto the AFM, we calibrated the system. The AFM was used in tapping mode under liquid phase. We used a large window of ~6 um to get a large field view of the nanostructures. Many many many 'As' became more and more visible as we focused the machine. Once we had a grasp on the system, we dived deeper into the details to resolve higher resolution images of single nanostructures.

Image Processing

The atomic force images were saved by the software as Nanoscope AFM data files. As it happens, there is an open source program Gwyddion that can load and process these data files. Once in Gwyddion, we were able to export the images as png (or other image) files, as well as measure features on the surface, etc.

Alternative to purchasing an AFM

Atomic force microscopes are expensive to buy and maintain, but there are several options that are possible alternatives: One is to contacting a contract laboratory service capable of doing AFM, e.g. through the Assay Depot portal. You could also find a local university with AFM equipment that would be willing to help. If there's enough interest, convincing your local bio hacker space to acquire one might be approachable. And finally, there's also an instructable to make one; while it doesn't appear to support the tapping mode or liquid phase needed to image DNA origami structures, you could try to add those features!

Step 9: Print and Feel the Data

Once we had finished the process, we wanted a way to better understand the data. We wanted to feel the data, and so we developed a way to 3D print the atomic force images as a topographic map. This allowed us to touch and feel the height of the A on the surface. It helped us illustrate the noisiness of atomic force microscopy and also helped us grasp of the AFM worked by measuring a height profile.

Processing the Image Data for Good Conversion

While just the raw TIF data from Gwyddion can be used for conversion, it has a lot of height variation due to the small pixel count which can make the result very spiky. We decided to do a blur step with a small window size on the image to smooth out the data while maintaining the overall feel. Also, it's very common to have large height artifacts in the AFM data, for example when salt crystals land on the surface they can be quite a bit taller than the origami, so we also did a max height filter to cut off those pieces and provide better contrast for the resulting print.

These two steps were accomplished using some basic Python programming and the PIL image processing library, though they could be easily done on a single image using any popular image editing software.

Converting Images to STL

First, we converted the grey-scale image to STL using the Heightmap2STL.jar script. We downloaded the script and placed in in the appropriate directory. Using terminal we ran the command

java -jar heightmap2stl.jar 'path to imagefile' 'height of model' 'height of base'

and we played with 'height of model' and 'height of base' until we had turned the parameters to an acceptable standard for 3D printing. We used 'height of model'=50 and 'height of base'=1. The conversion was very fast (on the order of 2 seconds).

3D Printing

Once we had a valid STL, we loaded the mesh file into Autodesk Meshmixer where we scaled and repaired the model. With the repaired STL file, we successfully printed the file on a Stratasys Objet Connex500 using Vero Clear resin.

It is also possible to use Meshmixer to send the print directly to Shapeways where it can be configured with many material choices.