Optimization of Hinge Assembly Conditions


The aim of this experiment was to optimize hinge assembly by determining optimal conditions for staple strands to anneal to the scaffold. The impact of  MgCl2 concentrations and thermal ramp conditions on hinge assembly was inferred by the degree of staple strands that were able to anneal to the scaffold.


The self-assembly of DNA origami structures follows a process of denaturation and cooling at respective varying rates. In this regard, the annealing temperature and the rate of heat removal are crucial to the yield of the formed structure. During the cooling phase of a thermal annealing ramp, the staple strands anneal to the scaffold strands by complementary base pairing, with faster rates being directly related to the affinity strength of their complementary sequences. The role of adding the MgCl2 salt is to stabilize the negatively charged DNA backbone (1), which is known to have negatively charged repulsion (2). Increasing the MgCl2 concentration results in increased folding, therefore, compacting the structure, whereas decreasing the concentration will result in fragmentation of the backbone due to the destabilization of charges.


A combination of MgCl2 concentrations from 0 to 30 mM at intervals of 5 mM and three thermal annealing ramps (27-, 37-, and 67-hour) were used to prepare hinges.  Samples were first prepared in triplicates according to Table 1. 

Final concentration of MgCl2 (mM) C0 C5 0 5 10 15 20 25 30
100 nM p8064 (μL) 9 9 9 9 9 9 9 9 9
500 mM Staple master mix (μL) 0 0 9 9 9 9 9 9 9
10X Folding Buffer (μL) 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5
100 mM MgCl2  (μL) 0 2.25 0 2.25 4.5 6.75 9 11.25 13.5
dH2O (μL) 31.5 29.25 22.5 20.25 18 15.75 13.5 11.25 9
Total Volume (μL) 45 45 45 45 45 45 45 45 45

Table 1: Preparation of hinge samples under different concentrations of MgCl2. C0 and C5 represent negative controls containing no staple strands.

Each set of samples was heated and cooled according to one of the thermal cycles outlined below. Cycle 1 and cycle 2 were adapted from the work of Lauback et al. and Douglas et al. respectively (3, 4).

Cycle 1: 67 Hours

  1. 65 °C → 62 °C at a rate of 1 °C/hour
  2. 62 °C → 59 °C at a rate of 1 °C/2 hours
  3. 59 °C → 46 °C at a rate of 1 °C/ 3 hours
  4. 46 °C → 40 °C at a rate of 1 °C/hour
  5. 40 °C → 25 °C at a rate of 1 °C/30 min
  6. 25 °C → 4 °C at a rate of 1 °C/min

Cycle 2: 37 Hours

  1. 80 °C → 60 °C at a rate of 4 °C/min
  2. 60 °C → 24 °C at a rate of 1 °C/hour
  3. 24 °C → 4 °C at a rate of 1 °C/min

Cycle 3: 27 Hours

  1. 65 °C → 60 °C at a rate of 1 °C/30 min
  2. 60 °C → 40 °C at a rate of 1 °C/hour
  3. 40 °C → 25 °C at a rate of 1 °C/20 min
  4. 25 °C → 4 °C at a rate of 1 °C/min

A 1 % agarose gel was prepared, loaded, and run as described in Hinge Assembly.


Figure 1. Effect of varying salt concentrations and thermal ramps on staple and scaffold annealing. The thermal annealing ramp protocol used to assemble V2 and V3 is denoted as 27-hour, 37-hour, or 67-hour. MgCl2 concentrations range from 0 mM to 30 mM. A. Salt and thermal screen of V2 hinge. 30* is the 30 mM sample for the 27-hour annealing ramp. p8064 scaffold control supplemented with 5 mM MgCl2 is denoted as C5. B. Salt and thermal screen of V3 Hinge. An additional control of C0 represents the p8064 scaffold without MgCl2 supplementation. The 1 Kb DNA ladder used in B was from FroggaBio, whereas A and C were from Thermo Fisher. C. Comparison of  V2 and V3 assembly to the Lauback hinge (L). V2 and V3 were assembled with 15 mM MgCl2 using the 37-hour annealing ramp. All gels were composed of 1 % agarose, 11 mM MgCl2, and run at 90V for ~ 1 hour.

As salt concentration was increased, the DNA migrated further across all thermal ramp conditions and samples. MgCl2 concentrations higher than 25 mM increased the amount of intermediate folding and the formation of undesirable aggregates. Based on the salt screen results for V2, the 27-hour and 67-hour thermal ramp did not show a difference between concentrations ranging from 5-25 mM MgCl2 and 10-15 mM MgCl2, respectively. For the 37 hour ramp, 25 mM produced the highest average band intensity. With regards to the results for V3, no difference was observed between MgCl2 concentrations ranging from 5-25 mM MgCl2 for the 27-hour and 37-hour ramp, and 10-25mM for the 67-hour ramp. 

The 20 mM MgCl2 concentration and 37-hour thermal annealing ramp were chosen for subsequent single hinge assemblies as they provided the most consistent agarose gel band expected for our single hinges.


At the time of this experiment, earlier iterations of the H1 and H2 hinge denoted as V2 and V3 respectively were used. These hinges were designed in the same manner as H1 and H2 and differed in how the pivot point of the hinges was formed. Instead of the scaffold connecting two bricks together into a hinge shape, V2 and V3 consisted of two 36 helix bundles connected to each other by 8 single-stranded DNA linkers. Because this experiment assessed which conditions were more favorable for staples to anneal to the scaffold, the results of the experiment were considered applicable to the assembly of H1 and H2.

  1. Hu Y, Chen Z, Hou Z, Li M, Ma B, Luo X, Xue X. 2019. Influence of Magnesium Ions on the Preparation and Storage of DNA Tetrahedrons in Micromolar Ranges. Molecules. 24:2091.
  2. Xiao S, Zhu H, Wang L, Liang H. DNA conformational flexibility study using phosphate backbone neutralization model. Soft Matter. 2014;10(7):1045–55. 
  3. Lauback S, Mattioli KR, Marras AE, Armstrong M, Rudibaugh TP, Sooryakumar R, et al. Real-time magnetic actuation of DNA nanodevices via modular integration with stiff micro-levers. Nat Commun. 2018;9(1446).
  4. Douglas SM, Dietz H, Liedl T, Högberg B, Graf F, Shih WM. Self-assembly of DNA into nanoscale three-dimensional shapes. Nature. 2009;459(7245):414–8.



Optimization of Scaffold to Staple Strand Ratio


This experiment aims to assess the impact of the molar ratio of staple to scaffold on the assembly of H1 and H2 hinges.


DNA nanostructures assembled through the use of DNA origami are typically assembled at a molar ratio of 5:1 of staple to scaffold strands. The inclusion of excess staples strands increases the likelihood that a staple strand will bind to its intended location on the scaffold. However, a greater excess of staples included during the assembly step also increases the probability of residual unbound staples strands remaining after hinges are purified. These residual staples could affect downstream experiments such as when the strands of the locking mechanism are introduced.


H1 and H2 hinges were assembled as described in Hinge Assembly. To create the samples that contained different ratios of staple to scaffold, staples strand master mix was added to a final concentration of 40 nM, 60 nM, and 100 nM to produce samples with a ratio of 1:2, 1:3, and 1:5 padlock to staple strand, respectively. These samples were annealed under the 37-hour thermal ramp and visualized on 1% agarose gel as described in Hinge Assembly.


Figure 1. Impact of scaffold to staple strand ratio on hinge formation. Lad indicates lanes that contain GeneRuler Ultra Low Range DNA Ladder. H1 and H2 were assembled with the following ratios of scaffold to staple strands, 1:5, 1:4, and 1:3.

As the ratio of scaffold to staple strand decreases, so too does the amount of H1 and H2 hinge assembled. This was inferred based on the decreasing band density of the H1 and H2 hinge as the amount of excess staple strand decreased.


Decreasing the concentration of excess staple strands, in general, led to a decrease in H1 and H2 hinge assembly. Although H1 and H2 hinges were able to form under all conditions, a higher concentration of H1 and H2 hinges during assembly will help ensure that there will be more hinges available for downstream experiments. Therefore, we decided to maintain hinge assembly at a 1:5 scaffold to staple ratio and introduce further wash steps to remove excess staples during hinge purification.


Validation of Lock and Padlock Hybridization


This experiment aims to further validate the ability of the lock strand to recognize and bind to the padlock strand of the locking mechanism.


The lock strand which is 24 bp in length binds to the padlock strand to force H1 and H2 hinges into a closed conformation. As observed in Locking Mechanism 1: Target Recognition, lower molecular weight strands such as the lock strand did not appear as clearly in the gel compared to other strands in the locking mechanism that had a higher molecular weight. Therefore we conducted further experimentation to demonstrate that the lock strand is able to bind and recognize the padlock.


Lock and H2 padlock 1 were combined at an increasing molar ratio of padlock to lock (1:0.5, 1:1, 1:1.5, 1:2, 1:2.5, and 1:3). These samples were prepared according to Table 1.

Padlock: Lock 1 uM Padlock (μL) 1 uM Lock (μL) ddH2O (μL) 100 mM MgCL2 (μL)
1:0.5 1.09 0.65 7.01 1.25
1:1 1.09 1.31 6.35 1.25
1:1.5 1.09 1.96 5.70 1.25
1:2 1.09 2.61 5.05 1.25
1:2.5 1.09 3.27 4.39 1.25
1:3 1.09 3.92 3.74 1.25

Table 1: Preparation of lock and padlock strands at increasing molar ratios.

The negative controls which consisted of either lock strand or padlock strand were made as follows. Each control was made to ensure that the sample contained 30 ng of the control strand. To prepare the lock strand control, 4.04 μL of the lock strand (1 μM) was mixed with 5.96 μL of TE buffer. The padlock strand control was prepared by mixing 1.09 μL of the padlock strand (1 μM) with 8.91 μL of TE buffer.

The samples and controls were then incubated at 30 °C for 25 min and visualized on a 20% polyacrylamide gel. The samples and controls (each 10 μL) were mixed with 2 μL 6X loading dye and loaded into the gel. The gel was prepared, run, and stained in the same condition as described in Locking Mechanism 1. To serve as a reference for relative band migration, 1 μL of  GeneRuler Ultra Low Range DNA Ladder (ThermoFisher) at a concentration of 100 ng/μL.


Figure 1. Increasing ratio of padlock to lock demonstrates lock and padlock hybridization. GeneRuler Ultra Low Range DNA ladder is represented by lanes annotated Lad. Lanes L and P contain the lock and padlock negative control respectively. Samples were incubated with increasing ratios of padlock to lock strand from 1:0.5 to 1:3 in increasing increments of 0.5.

As the molar ratio of the lock strand to the padlock strand increased, the band representing the unbound lock strands also increased in intensity. This demonstrated that the lock strand is able to bind to and recognize the padlock strand.


As the number of lock strands increases relative to the number of padlock strands, there are fewer padlock strands available for the lock strands to bind to. As such it was expected that if the lock strand was binding to the padlock strand we would see an increasing gradient of band intensity for the band representing unbound lock strand, which was shown in Figure 1. 

H2 padlock 2 was arbitrarily chosen to conduct this experiment. As all padlock strands are identical aside from their overhangs to bind to the hinges, it is expected that the remaining padlocks would produce similar results. These overhangs are not complementary to the locking mechanism strands and are not expected to influence their ability to bind to their targets.


Detection of Kinetic Traps


To optimize the folding of H1 and H2 hinges, python scripts were made to assess the presence of kinetic traps in cadnano outputs.


Kinetic traps result from excess mechanical strain which refers to the energy stored in bent or twisted structures programmed (internal energy) to occur by engineering staple crossovers at certain helix locations. The graph shown below shows how this folding process works.

Figure 1. Energy Coordinate Diagram Approximation. The local energy minima occurs in the circle in red. Structures, if not designed properly can result in a geometry that is different than intended or be stuck in the red circular stage as noted.

As shown above, we need to ensure that the structure in question is the one with the most minimal Gibbs Free Energy. This is not so easy with DNA origami structures. The structure can be caught in smaller energy minima as it seeks to minimize its internal energy and maximize its entropy. (1). Kinetic traps are an issue because if a scaffold strand bends to make a crossover, a staple strand will have to circumvent or bend around the scaffold to ensure that the crossover occurs. Additionally, there is an “angular strain” on the staple strand and a “steric clash” between the staple and the scaffold strands (2–4). This is shown in the figure below in which is a kinetic trap shown from the cross-section of the crossover’s point of view. Overall, the “tight bending” and “angular strain” cause issues when binding.

Figure 2.  Cross-section of a crossover showing that the staple strand bend or Θstaple is less than the scaffold strand bend or Θscaffold creating a steric strain for the tightly wound staple strand and a steric clash between the scaffold and the staple due to the fact the staple circumvents around the staple.


Based on Stephanie Lauback’s MATLAB scripts, python scripts were created to detect kinetic traps. Python 3.0 Scripts required to analyze the presence of kinetic traps can be found here: under the name “KT modified version-ER.ipynb” 

After detection, kinetic traps were manually resolved in cadnano by adjusting staple strands adjacent to the location of the kinetic trap.

  1. Lauback S. Magnetic Actuation of Biological Systems. The Ohio State University; 2017.
  2. Glaser, M.; Deb, S.; Seier, F.; Agrawal, A.; Liedl, T.; Douglas, S.; Gupta, M. K.; Smith, D. M. The Art of Designing DNA Nanostructures with CAD Software. Molecules (Basel, Switzerland). 2021, 26(8), 2287. 
  3. Fern. J,; Lu, J.; Schulman, R. The Energy Landscape for the Self-Assembly of a Two-Dimensional DNA Origami Complex. ACS Nano. 2016 10(2), 1836-44. 
  4. Garcia, H. G.; Grayson, P.; Han, L.; Inamdar, M.; Kondev, J.; Nelson, P. C.; Phillips, R.; Widom, J.; Wiggins, P. A. Biological consequences of tightly bent DNA: the other life of a macromolecular celebrity. Biopolymers. 2007, 85(2), 115–130. 


Detection of Sandwich Strands


To ensure proper annealing of staple strands to the H2 and H1 strands, python scripts were created that interpreted cadnano inputs and identified coordinates in the helices.


Sandwich strands occur due to the relative length of staple strands that bind to multiple domains on the scaffold or have crossovers that connect multiple helices together. They are generally characterized as type 1 or type 2 sandwich strands (1), as shown in Figure 1.

Figure 1. Sandwich Strand Types. Type 1 sandwich strands have a long binding domain followed by a crossover, then a shorter binding domain relative to the previous binding domain, followed by another crossover that has a longer binding domain relative to the previous binding domain.  In contrast, type 2 sandwich strands have the same length binding domains separated by a single crossover.

Type 1 sandwich strands present a structural issue due to the fact that the staple DNA and the scaffold DNA need to form a double helix or “wrap” around each other. When one long binding domain binds first, it is difficult kinetically for the shorter binding domain to bind as shorter sequences of staple strands need to wrap around a shorter stretch of scaffold followed again by a long stretch. In other words, if the two longer binding domains bind first in a thermocycler, it will be kinetically impossible for the shorter binding domain to properly bind (2). Type 2 strands are also an issue as there is no sequential binding of the staple to the scaffold. It is improbable for both binding domains to bind at the same time. Difficulty regarding sandwich strands can leave the user feeling frustrated as they can often result in the shape of the scaffold deformed, relative to the intended conformation (3).


Python 3.0 scripts were also developed to identify the relative locations of these sandwich strands. The final result would be an output that determines the sandwich strands at the helice and positions in which the sandwich strands exist as well as the type of sandwich strands identified. The GitHub Repository can be found at under the name Sandwich-Checker-Latest-Version.ipynb

After detection, the sandwich strands were carefully identified and manually fixed in cadnano.

  1. Jung, W. H.; Chen, E.; Veneziano, R.; Gaitanaros, S.; Chen, Y. Stretching DNA origami: effect of nicks and Holliday junctions on the axial stiffness. Nucleic acids research. 2020, 48(21), 12407–12414. 
  2. Glaser, M.; Deb, S.; Seier, F.; Agrawal, A.; Liedl, T.; Douglas, S.; Gupta, M. K.; Smith, D. M. The Art of Designing DNA Nanostructures with CAD Software. Molecules (Basel, Switzerland). 2021, 26(8), 2287. 
  3. Fern. J,; Lu, J.; Schulman, R. The Energy Landscape for the Self-Assembly of a Two-Dimensional DNA Origami Complex. ACS Nano. 2016 10(2), 1836-44.