DNA Design

  • Using DNA origami we designed a spring-loaded box structure capable of capturing molecular cargo unbound. The box consists of two hollow DNA lids joined by a hinge, as well as a brace which holds the lids apart.
  • By altering the sequences of the DNA in the hinge we predict the box will be held tightly closed under thermal fluctuations. The box is then braced open with DNA containing a cut-site specific to a particular endonuclease and we hypothesise that the endonuclease can cut the brace and be caught inside the cavity.
  • Bulk of the design was done in caDNAno but various other optimisation tools such as Nupack were also used. Behaviour of designs were predicted using finite element modelling software called Cando. Various permutations of hinge and brace types were designed as well as structures which have been latched or sealed close for use as controls in the following experiments.

Can I provide you with some supplementary material? I CanDo.


In the past the technology and methods used to build nanoscale structures required enormous amounts of energy and lacked precision and control. The recent development of the technique DNA Origami back in 2006 by Paul Rothemund has changed the game.

DNA origami uses DNA solely as a structural material rather than a biological carrier of information. The technique makes use of the predictable base pairing rules and programmability of DNA sequences, it involves designing strands in such a way that when they come together in the right conditions they self-assemble into the desired structure with sub-nanometer precision.

DNA Origami structures consist of a single long strand of DNA called a scaffold (shown in white), folded into shape by hundreds of shorter strands called staples (shown in yellow).


To show that the concept worked, in 2006 Paul synthesised several arbitrary shapes such as smiley faces, triangles and stars. Since then DNA origami structures have become much more complex, especially since the release of open source DNA design software such as caDNAno and vHelix.

The unprecedented control that DNA origami offers over the shape and size of nanostructures is incredibly appealing and forms the bulk of our project. To understand and appreciate DNA origami and our project some degree of technical knowledge is necessary, and so the following section is some background information to introduce the concept.


We wanted to design a DNA origami structure that was capable of reliably capturing small molecular cargo. To do so, we wanted to design a structure in which a conformational change could be triggered by a non-covalent interaction with the target molecule.

To read more about the motivation behind our design, see our project page.


DNA has a natural helical twist of 10.5 base pairs per turn, which is equivalent to 240 degrees of rotation per exactly 7 base pairs.



If the twist of two adjacent helices line up, we create the opportunity for strands of DNA to crossover to neighbouring helices. This doesn’t happen spontaneously, it has to be preferential for the strand of DNA to crossover and bind to the neighbouring helix. However DNA sequences are programmable and therefore can be designed to occur at specific locations in order to build complex structures image-center

By spacing crossovers every 7 base pairs along a helix (and therefore every 240 degrees around the helix if looking down the axis of the DNA helix) we can see that potential crossover positions occur at three equally spaced directions. The image below shows the helical rotation of the DNA with the white dashed line as we move along the strand. The yellow lines represent the potential positions for crossovers.


By routing strands of DNA in and out of several adjacent helices we can extend this branching pattern to create a hexagonal DNA lattice with crossovers occurring every 7 base pairs. We call this a honeycomb lattice, and it forms the basis for our design.



To achieve our aim, we designed a DNA Origami box which could be both wound up like a spring and braced open. The structure in this configuration is in a high energy but stable state, and as soon as the brace is cut by a restriction enzyme (see protein engineering) the structure snaps closed into its low energy state.


The box is made of two symmetrical halves comprised of DNA arranged in a honeycomb lattice. Each half contains a cavity approximately the size of a large protein, and the two halves are joined only by a set of staple strands that crossover between two specific helices, which we call the hinge region (shown in red below).



Two braces are used to hold apart the two halves.


The design of the box is broken into several distinct parts, including the two lids (yellow), the hinge (red) and positions for the braces to extend from (blue).


The design of the structure was done in caDNAno using the 7249 base pair m13 plasmid. We designed the two halves separately in caDNAno so that we could use the autostaple function to staple the scaffold into the lid shape without stapling the halves to each other. We then manually edited the scaffold and staples of the hinge helices 35 and 36 to crossover using the imaginary helices 72 and 73 as a guide.



The staples in the box here are colour coded into different sets: the top half (orange), the bottom half (green) and the hinge (purple).

Each staple in the entire structure was then inspected and if necessary adjusted to obey a set of rules in order to optimise the kinetics of the boxes folding and formation. The set of rules include:

  • Staple length averages 40 base pairs, no shorter than 18 or longer than 48 base pairs
  • Each staple must contain at least one uninterrupted 14 base pair ‘seed’ between crossovers to aid formation
  • Edge staples include 5 thymine bases overhanging to prevent base-stacking effects between structures
  • No staple crossovers within 5 base pairs of a scaffold crossover
  • The scaffold routing was also optimised by splitting it into two ‘modules’ and therefore avoiding large folds.


The predicted 3D shape was tested using finite element modelling implemented in the software CanDo, three angles of the model are shown below.


Movies were also generated, embedded below is a movie showing the fluctuation of the box around the hinge, revealing the cavity.

The Hinge

The hinge helices were designed to behave as under-wound springs, applying a restoring force onto the box and holding the halves tightly closed. We could do this by altering the number of base pairs between crossover strands, and have produced different degrees of angular displacement of the halves.

In the background section I mentioned that DNA has a helical twist of 10.5 base pairs per turn, this corresponds to approximately 35 degrees per base as illustrated in the image below (again, looking down the axis of double helix.) image-center

If each base pair in the DNA helix corresponds to a 35 degree rotation, then displacing a crossover by a single base pair will correspond to an angular displacement of 35 degrees. The hinge is made up of two strainable helices, and so if we zoom in on the cross-section of the box we can see the different combinations of angular displacement we can apply.



The following screenshots are of the hinge region in the caDNAno files for each of these designs (a-e).

Each of these caDNAno designs were tested in CanDo to visualise how the skips and insertions in the design above would affect the behaviour of the structure. The movies showing the fluctuation around the hinge are shown below.

Control box (natural hinge position, no angular displacement)


35 degree angular displacement


70 degree angular displacement


105 degree angular displacement


140 degree angular displacement


The CanDo models do not account for electrostatic or Van De Waals repulsion between DNA helices and hence the halves overlap while fluctuating, which in reality cannot happen. Instead, we expect that by increasing the angular displacement, we expect the hinge to be more wound up like a spring, pushing the two halves closed. As we increased the angular displacement of the hinge the videos show that the halves do not flap open as far which is an indication that our design is predicted to behave as expected.

Based on the CanDo predictions we decided to synthesise boxes with 0, 35, 70 and 105 degrees of angular displacement (we excluded 140 as it shows significant hinge deformation.) We also optimised our hinge staples so that there we no single stranded nics present in helices 35 or 36 which would reduce the rigidity and robustness of the hinge which will be under torsional strain.

The sealed box

After finalising our hinge design we also isolated a set of staples on helices 25 and 46 which could be replaced with an alternative set of staples which would run between the two helices and seal the box closed. The logic of this sealed design was to give us a completely closed standard to which we could compare our boxes under torsional strain (which we expect to remain tightly closed.)

Using imaginary helices 74 and 75 (shown below), we manually edited the staples in helices 25 and 26 to seal the box closed. The sealed caDNAno design is shown below with the sealing staples coloured in black.


Non-sealed (orange staples are in the top half, green staples are in the bottom half)


Sealed box (edited staples are shown in black)


The braces

The final piece of the puzzle was the design of the braces. The braces are formed by the hybridisation of two extended staple strands which reach out from the either edge of the cavity from both lids. The two brace positions we chose to experiment with are located on the hinge side of the cavity so that they would not interfere with the entry of the payload The positions for the two pairs of braces are shown below.


In order for the staple extensions to form a brace rather than a latch, the extensions from each half must be extending in anti-parallel.


Because we wanted our braces to form from staple extensions on the inside of the cavity, there is only one choice as to which staple we can extend per helix. As shown in the caDNAno screenshot below extending staples from 33 to 38, while symmetrical and therefore aesthetically appealing, yields a latch as opposed to a brace. For this reason we have designed our braces off-kilter, and hence designed our staples to extend from helices 33 to 39 and 31 to 41.




The brace staple extensions had a calculated number of overhanging bases designed to achieve a particular length of dsDNA and therefore an intentional angle of openness of the halves.

We designed braces for 45, 90, 135 and 180 degrees of openness between the two halves, and designed staple overhangs to achieve those angles for two brace positions (helix 33 to 39 and 31 to 40).

Given our two brace positions (helix 33 to 39 and 31 to 41) we then used the size parameters of DNA (0.34nm per base) and triangular geometry to calculate the number of base pairs required to produce the target angle of cleavage.


See the supplementary material for cool calculations

Our design contains two braces as opposed to one which makes our structure less likely to capture the payload as the endonuclease component is required to sequentially cut before being trapped (see protein engineering for the details of the payload.) However this was an intentional decision made with the short time frame of BIOMOD in mind, and still allows us to prove the concept of a payload triggering the closing mechanism of the box by first binding inside (see our project page for more information about our aims for BIOMOD.)

Because we designed our box with two braces there are four staples which extend from the cavity and hybridise to their respective partners. The sequences of those extensions were designed using ‘GAGA’ algorithms and were analysed and optimised in Nupack to minimise likelihood of secondary structures.

Within each of those sequences is also an EcoRI recognition site so that the payload can cut the brace and therefore trigger the closing mechanism once inside the cavity.


Our brace sequences, their positions, and the corresponding angle of opening can be found in our supplementary material.

Latch Design

Results from FRET indicate that the hinge torque in our designs isn’t sufficient to close the lips of the box (as compared with results from chemically sealing the box with staple strands.) For a final experiment we designed single stranded latch extensions for the lips of a braced box to seal it closed in absence of a brace.

The latches are designed to join helix 25 and helix 46 by extending 8 base pairs of single stranded DNA at several points along the lips of the halves complementary to extensions in the opposite half. The number and the positions of the latches are limited to replacing our non-sealing staples. The latch staples are highlighted in pink below.



Latch formation requires that the staples in opposite halves be modified at opposite ends, for example if the top half has a 5’ extension then the bottom half should have a 3’ extension. If the extension is on the same end of the staples in the opposite halves a brace will form rather than a latch. See the image below.


The latch staple sequences along with their starting and ending positions in the caDNAno design are in the table below.

Start End Sequence Latch end mod Title for Nupack
46[20] 46[14] TAATCCCCGAGTGGA End 4

Each of the overhanging sequences forming the latch are 8 nucleotides long – this way there is a significant off-rate for unwanted hybridisation with excess complement staples in solution. Each of the latch sequences were optimised in Nupack and checked for secondary structures. Nupack analysis shows that no structures form between the strands other than the desired latch hybridisation (strand 1 with 5, 2 with 6 and so on). Nupack optimisation results are in the supplementary notes.

Hinge Extensions

To determine if the braces were functioning as predicted we extended two staple strands in the hinge region of our control (0 degree angular displacement) with sequences complementary to a Nickel Nitrilotriacetic acid (Ni-NTA) modified strand and a red acceptor fluorophore (ATTO565N) strand.

The hypothesis is that while the box is braced open a green fluorescent protein (GFP) can enter the box and bind to the Ni-NTA via a histidine tag (his-tag) interaction. Upon binding to the Ni-NTA strand the GFP is co-localised within the Forster radius with the red acceptor and Forster Resonance Energy Transfer (FRET) will be observable.


An existing staple in the hinge (start 37[63] end 58[70]) was cut at 35[64] and the 3’ end of the staple ending at 35[64] was modified with a 20 base pair extension complementary to a 5’ NTA-modified strand. The staple was cut at 35[64] rather than the natural crossover position 35[62] so that the strand would extend directly into the cavity. The sequences of the new staples and an image of the design are shown below (the extension is marked with italics.)

Start End Sequence


In a similar manner another hinge staple (start 34[48] end 13[41]) was split at 36[47] and the new staple ending at 36[47] was 3’ modified with 20 nucleotides complementary to an ATTO565N 5’ modified strand. Like above, the extension began 2 base pairs from the natural crossover position to project the strand directly into the cavity. The new staples and their sequences are listed below with the extension marked in italics.

Start End Sequence

Materials and Methods

Origami Synthesis Protocol

All staples are suspended in water and were separated into staple mixes including general (common), hinge, brace position one, brace position two, and sealing. This gives us the most flexibility for synthesising various structures with different braces and different hinge variations also in different buffers. Any given box needs a staple set from each of those mixes; the choices are outlined in the table below.

General General      
Hinge 0 degrees 35 degrees 70 degrees 105 degrees
Sealing Staples Sealing Non-sealing    
Brace Position 1 No brace 45 degrees 90 degrees 135 degrees
Brace Position 2 No brace 45 degrees 90 degrees 135 degrees

In addition to these, however, some special staple mixes have also been developed for specific applications.

Type Name Function
General Female Biotinylated, Fluorophore Contains acceptor fluorophores and biotin complementary
  Male Biotinylated, Fluorophore Contains acceptor fluorophores and biotin complementary
  Non-Biotinylated, Fluorophore Contains acceptor fluorophores
Sealing staples Sealing, Fluorophore Contains donor fluorophores, while sealing the box
  Non-Sealing, Fluorophore Contains donor fluorophores, without sealing the box
  Non-Sealing, Latching Latches the box closed
  Non-Sealing, Latching, Fluorophores Contains donor fluorophores, while latching the box closed
Hinge Staples 0 degree NTA Conjugates to Ni-NTA
  0 degree ATTO Conjugates to Ni-NTA and ATTO 565n

Our standard synthesis protocol involves a 50ul reaction volume with 20nM m13 plasmid, 200nM of the appropriate staples, 20mM MgCl, 1mM EDTA and 5mM Tris buffer. The samples are annealed using the following temperature cycle:

  • Denature at 65C for 15 minutes
  • Slow ramp to 45C at 0.1C per 7 minutes
  • Hold at 20C

To purify the samples they are mixed with PEG precipitation buffer (15% w/v PEG 8000, 5mM Tris, 1mM EDTA, 505mM NaCl) in 1:1 ratio. The mixture is then spun hard for 25 minutes at room temperature and the supernatant is discarded. The DNA origami pellet is then left to resuspend at room temperature for 24 hours at 20nM in DNA origami buffer 20mM MgCl, 1mM EDTA and 5mM Tris.)


Douglas SM, Marblestone AH, Teerapittayanon S, Vazquez A, Church GM, et al. Rapid prototyping of 3D DNA-origami shapes with caDNAno. Nucleic Acids Res. [2009]

J. N. Zadeh, C. D. Steenberg, J. S. Bois, B. R. Wolfe, M. B. Pierce, A. R. Khan, R. M. Dirks, N. A. Pierce. NUPACK: analysis and design of nucleic acid systems. J Comput Chem, 32:170–173, 2011. (pdf)

P. W. K. Rothemund, Folding DNA to create nanoscale shapes and patterns. Nature 440, 297 (2006).

Douglas SM, Bachelet I, Church GM. A logic-gated nanorobot for targeted transport of molecular payloads. Science. 2012;335(6070)

H. Dietz, S. M. Douglas, W. M. Shih, Folding DNA into twisted and curved nanoscale shapes. Science325, 725 (2009).

DN Kim, F Kilchherr, H Dietz, M Bathe. Quantitative prediction of 3D solution shape and flexibility of nucleic acid nanostructures. Nucleic Acids Research, 40(7):2862-2868 (2012).

CE Castro, F Kilchherr, DN Kim, EL Shiao, T Wauer, P Wortmann, M Bathe, H Dietz. A primer to scaffolded DNA origami. Nature Methods, 8: 221-229 (2011).

Ranjbar R1, Hafezi-Moghadam MS2. Design and construction of a DNA origami drug delivery system based on MPT64 antibody aptamer for tuberculosis treatment.

Wang JC (1979). “Helical repeat of DNA in solution”. PNAS. 76 (1): 200–203.

Stahl E, Martin TG, Praetorius F, Dietz H. Facile and Scalable Preparation of Pure and Dense DNA Origami Solutions. Angewandte Chemie (International Ed in English). 2014;53(47):12735-12740. doi:10.1002/anie.201405991.

Andersen, Ebbe S., et al. “Self-assembly of a nanoscale DNA box with a controllable lid.” Nature 459.7243 (2009): 73-76.

Benson, E., Mohammed, A., Gardell, J., Masich, S., Czeizler, E., Orponen, P., & Högberg, B. (2015). DNA rendering of polyhedral meshes at the nanoscale. Nature, 523(7561), 441-444.