The NanoFilter is a chain-like DNA origami-based structure with an in-built mechanism that allows it to fold into a spiral shape in response to stimuli. The concept targets the atherosclerosis aspect of CVD. It has been designed to target electronegative LDL cholesterol molecules that are present in the bloodstream. The structure would bind to arterial endothelium using an aptamer with a binding affinity for specific surface markers on the cells of the inner arterial wall. The nanofilter is flow-activated. Therefore, the above-mentioned aptamer itself would be integrated into an attachment point on one end of the nanofilter that would help the structure orient itself so as to be opened using blood-flow.
The force of blood-flow would be strong enough to unhinge the locks that keep the structure in its closed state thereby exposing the cholesterol binding sites. The binding would be a trigger for the refolding of the nanofilter into its spiral form essentially trapping the cholesterol molecules. The chain and spiral duality of the structure allows flexibility in controlling the surface area to volume concept. The spiral form would provide more mobility to the nanofilter as its smaller size would prevent drag resistance due to blood flow and faster approach towards target locations i.e. arterial walls. Contrastingly, the foldability allows multiple cholesterol specific sites to be integrated into the structure, making it more effective in terms of units captured per filter.
Figure 1c): Binding of cholesterol triggers refolding
Figure 1d): Structure returns to locked state
Figure 1e) Video of the NanoFilter unfurling
MAIN STRUCTURE: THE NODE-LINKER SYSTEM
CADNANO (separate box/back ground)
The main structure was designedusing the software, Cadnano. It is a design software developed by Dr Shawn Douglas which provides a simplified interface that allows the user to visualise and build DNA origami structures by accounting for each base, more specifically every 7th base, in a single stranded DNA strand and linking it all together using a cross-section lattice which guides scaffold and staple cross-overs. The main interface has three approaches:Slice Panel (Cross-section: hexagonal or square lattice), Paths Panel (Directs scaffold and staple paths and cross-overs) and a 3D panel which provides a 3-dimensional view of the structure in the form of bundled cylinders (helices).
Figure 2: Cadnano: Slice Panel, Path Panel and 3D Panel
To begin a design on the Cadnano, across-section must be decided on. An example of a cross-section is seen in figure []. Once this is done, scaffold strands appear on the path panel which can be extended to either end. The square end represents the 5’ end while the arrowhead represents the 3’ end of the scaffold strands.
Figure 3: Sample cross-section, scaffold, staple and 5' and 3' ends in the Cadnano interface
Now, to begin constructing, the scaffold must be linked together until it forms one continuous loop. Cadnano suggests possible connection points or ‘cross-overs’ independently. Therefore, cross-overs must be added cautiously ensuring we still have a continuous scaffold (This may vary if the structure is made of two scaffolds). Once the cross-overs are placed at relevant locations, staples maybe added to the structure (manually or auto-stapled). Cadnano also displays relevant staple cross-over points, however, cross-over density over a region of the structure and the length of the staple strand must be considered. If too many cross-overs are condensed into a small region, the electronegative charges of the DNA molecule may cause instability and irregular folding when done experimentally. The length of the staple must be limited to approximately 60-90 bases; longer staples are energetically unfavourable.
Figure 4: 6 Helix Bundle, Molecular model (tacoxdna), Cadnano paths view and folding of the origami
The main structure was designed with Cadnano, using the segmented chassis in the paper Cargo rigidity affects sensitivity of dynein ensembles to individual motor pausing by Driller-Colangelo et al. (2016) as a reference and starting point.
The general structure comprises 7 nodes and 6 linkers in between. The cross-section features a 12-helix bundle with helix 10 and 11 acting as the backbone. It is built using the p8064 scaffold and 239 staple sequences.
The nodes provide structural integrity ensuring the formation of a robust spiral and preventing entanglement of the open-chain form, in addition to surface area for placement of target (cholesterol) specific molecules.
The linkers are designed in a way that provides complete control over structural rigidity. An aspect of this study that we were most interested in was the difference in rigidity that stemmed from having double or single stranded linkers. According to the abovementioned study, double-stranded DNA displayed more rigidity than single-stranded DNA. Therefore, by direct implication, double-stranded linkers would be more rigid than single-stranded linkers. Here, double stranded directly corresponds to the presence of staple strands in the linker region while single stranded linkers imply that no staples are present in the linker region (only scaffold).
This related directly to our intention of having a structure folding into a spiral as having a control over the rigidity would imply that we can dictate the radius of the spiral by controlling how tightly or loosely it folds i.e. a mechanism may be developed where the ratio of double stranded regions to single stranded regions can be varied to achieve an intermediate rigidity.
The structure is built from a single scaffold of DNA. We remodelled the chassis from the paper with the same cross-section and generated staple sequences by adding a breakpoint in the scaffold two bases away from the end of helix 11 on node 6.
After close examination of the staple sequences provided in the supplementary information with the paper, we realised that, despite inducing the same break-point to generate sequences, Cadnano started mapping the p8064 scaffold from a different start-point. This meant that different staple sequences would be generated compared to the ones in the supplementary document.
Therefore, a custom scaffold, sourced from the supplementary document, was added to the program which matched the scaffold start and end points present in the chassis in the paper. This was done to ensure folding in the first attempt and enabling the lab sub-team to observe differences in linker rigidity by reproducing the experimental results from the paper.
It is modular in nature i.e. the size and cross-section of the nodes and linkers can be varied. More nodes can be added and the radius of the spiral can be controlled through rigidity which expands its area of application as, by varying the target specific molecule and appropriate chain parameters (node size and number, linker size and number), the nanofilter can be applied to a range of target molecules.
THE LOCKING MECHANISM: HANDLES
As previously illustrated, the structure goes through four stages through its entire cycle comprising its initial unfurled stage, a locked stage that is introduced into the bloodstream, unfurling due to flow and rolling up in response to cholesterol binding. This is made possible by the handles, which enable the locking mechanism.
The handles are essentially a continuation of staple strands allowed to extend above and below the structure. They are made of Spacer T nucleotides and a unique 20 nucleotide long sequence which is added to the 5’ or 3’ end of a structural staple. What needs to be considered is the direction of the exiting handle. If left unaccounted for, the handles may extend into the structure depleting it functionality and causing possible entanglement.
Figure 8: Direction of handles in helices 4 and 5
The direction can be evaluated using the slice panel of the Cadnano interface. Using the hexagonal lattice, the interface shows where one helix bind to another. Therefore, for instance, if we wish to attach a handle to helix 4, we’d have to ensure that at least one of the ends of any staple on helix 4 must be in a position that is exactly 180 degrees from the crossover point between helices 4 and 5. This implies half a turn of the helix. One turn is approximately 10.5 nucleotides; hence, half a turn would be approximately 5 nucleotides. Therefore, the staples on helix 4 need to be placed in a way that any one of their ends are positioned 5 bases before or after a crossover point between helix 4 and 5.
Figure 9: Comparison of staple positions in the two versions of the chassis
Handles on helix 10 require a different approach. From the cross-section, it can be noted that helix 11 prevets any handles exit points that are directly above helix 10. Therefore, i) handles cannot be attached to helix 10 in the node region due to high structural density, ii) in the linker region, handles on helix 10 would have the ability to extend around helix 11 and therefore, have functional exit points that are simply the crossover points between helix 10 and 11. The latter point mentioned above is valid as, according to the cross-section, the crossover between helices 10 and 11 implies that a staple on helix 10 has to extend upwards and outwards towards helix 11 which is the direction in which we want handles to proceed towards.
Figure 8: Direction of handles in helices 10 and 11
The handle attachments were categorised based on where in the structure they were to be placed. We sorted them into two categories: catalyst handles and locking handles. The handles were developed using the online interface known as Nupack. The procedure has been described below.
NUPACK
Nupack (Nucleid Acid Package) is a software package used for the design and analysis of nucleic acid sequences (Zadeh et al., 2010). It can be run online through their web server, or downloaded and compiled using their source code. It has a range of functionalities, including design, analysis and utilities:
Design: Nupack is able to design DNA and RNA sequences in order to form a target secondary structure at equilibrium.
Analysis: The interactions between nucleic acid strands in dilute solutions are analysed thermodynamically.
Utilities: The equilibrium properties of nucleic acid strand complexes are calculated and displayed.
For each complex, Nupack calculates the lowest-energy conformations of RNA and DNA secondary structures, as well as the equilibrium base-pairing probabilities and the concentrations of RNA and DNA complexes that would be present at equilibrium in a dilute solution.
On the design page, the target structure can be input into the program in dot-parens-plus or DU+ notation. We used dot-parens-plus notation, in which dots represent unpaired bases, matching parentheses represent base pairs, and plus signs represent nicks (Zadeh et al., 2011). We can also choose to prevent certain sequences by writing out the sequence of nucleotides, for example, GGGG and NNNNN, where G is Guanine and N is any nucleotide.
We then run the design job to obtain the DNA sequences. We can select the number of trials, i.e. the number of appropriate sequence designs the program will output, and then choose the one with the lowest normalized ensemble defect to analyse.
On the analysis window, we can choose the number of strand species we wish to have in solution and their sequences, as well as the maximum size a complex can form.
We then run the analysis job. The time the program takes to process the request can take from seconds to hours, depending on the complexity of the sequences, number of sequences and maximum complex size.
Once the job is done, it shows what strands have hybridised to each other and in what concentrations these complexes are found.
We can obtain further information, such as graphs relating to the nucleotides and configuration energies, by clicking on each complex.
The structures we obtain aren’t a very accurate portrayal of our original input. This is due to the 11.2% normalized ensemble defect, caused by the complexity of the secondary structure, and also because we used two strands, making more configurations possible. The strands used in our project, with normalized ensemble defects of around 0.4-0.6%, had a much higher accuracy.
THE CATALYST SITE
To ensure the spiral rolls up in the right direction, and to prevent the formation of entanglements and unwanted secondary structures, a catalyst mechanism was developed. The catalyst consists of the two final nodes (5 and 6) being arranged to bind specifically to each other, by making the two handles on either node complementary to the handles on the other node. We call these nodes the catalyst nodes, and the four handles attached to these are called the catalyst handles.
Each of the catalyst handles was 30 nucleotides long. Due to the flexibility of the structure, the last node (node 6) is able to bend onto node 5, allowing the strands to hybridise, causing the first and second nodes to remain bound to each other. Due to the length of the strands (30 nt), the bond between the catalyst handles is strong and won’t break due to blood flow.
Figure 9: Catalyst sequences
Figure 10: The folded catalyst
Catalyst handle placement
The two complementary pairs of catalyst handles (A/A’’ and B/B’) were attached on alternate staples on helix 4 of both nodes (helix 11 had locking handles). The handles were added to the 5’ end of the staples on both nodes. Since they are complementary strands, they need to align so that the 5’ end of a handle on node 5 begins to hybridize with the 3’ end of the handle on node 6. T-spacers were also present at the handle attachments to add more flexibility to the bond, as double stranded DNA is quite rigid.
When attaching these handles to the scaffold, the staple and handle exit points had to be considered. The catalyst handles are attached to helix 4, therefore, the staple ends were aligned in the paths panel to be 5 bases ahead or before the crossover points between helix 4 and 5.
Catalyst handle design
The catalyst handles were designed using the Nupack software, which was described earlier.
The sequences obtained, including toeholds, were as follows:
A: CCAGGCCAGATTGCTAGCCTCGCCGATCCGTTT
A’: CGGATCGGCGAGGCTAGCAATCTGGCCTGGTTT
B: ACTAAACGAGCTGTGGTGGCAAACGCGCGCTTT
B’: GCGCGCGTTTGCCACCACAGCTCGTTTAGTTTT
The normalized ensemble defect of this design was 0.4%. The analysis results showed that these sequences were very reliable for the configuration we wanted, and effectively 100% of the complementary strands bound to each other in solution and didn’t create any unwanted links or structures.
Thanks to this catalyst design, the last two nodes are bound to each other. This means that the tip of the scaffold has started to roll up, making it easier for the next step to begin: the rolling up of the entire structure.
THE LOCKING HANDLES
For the locking mechanism, we used two types of handles: “top” handles and “bottom” handles, each 20 nucleotides long, which were placed on the nodes and on the backbone respectively. Due to the arrangement of the staples, we were able to place 4 “top” handles on top of the nodes, and 3 “bottom” handles on the backbone side of the nodes, as well as 6 “bottom” strands on each linker, facing downwards.
Importantly, “top” and “bottom” handles are not complementary, and hence do not hybridise directly to each other like the catalyst strands did. Instead, each pair of “top” and “bottom” strands are held together by a single-stranded DNA sequence which we call the “locking strand”. This strand, of 20 nucleotides in length, is complementary to a section of both the “top” and “bottom” handles (10 nucleotides of each to be precise, that is, half of each handle), forming a segment of double stranded DNA. As more and more handles get bound together by locking strands, the entire structure gradually rolls up, until it is finally in the “locked” state.
Figure 11: The locking handle sequences
Figure 12: Locking handles hybridised with short complementary lock strand
Lock handle placement
The handles on helix 4 form a continuous strand with the handles on helix 11 i.e. when put together they have one common complementary toe-hold and close strand. Therefore, to ensure that the handles form a continuous 5’ to 3’ sequence, the handles were attached to the 3’ end of staples on the backbone - helix 11 on nodes and helix 10 and 11 in the linker region - and 5’ end of staples on helix 4 of the nodes. Now, when the close strand is introduced, it would be able to hybridize with the handle essentially locking the structure in its spiral state until flow is applied.
Lock handle design
To ensure structural stability efficient binding of handle + staple complex to the scaffold, we restricted the length of the staple and handle combination to 60 nucleotides. However, the handles themselves need to be at least 20 base pairs long to ensure successful binding with the complimentary close and open strands. For example, a staple that is 21 base pairs long can have one handle and three T-spacers attached to it. Similar to the catalyst case, the T spacers allow flexibility in the spiral folding and increase the probability of aligning correctly by allowing more configurations.
The lock handle sequences obtained from Nupack were as follows:
Top Handle: TTTGCGTGGTTGCCACCTTAATT
Bottom Handle: CGACGACCCGATTTGCCACGTTT
Lock Strand: CGGGTCGTCGAATTAAGGTG
Fig: Equilibrium concentrations of the top/bottom pair bound to the lock strand, and analysis results.
The normalized ensemble defect was 0.5%. Nupack’s analysis shows that the lock strand will bind to the top/bottom pair with effectively 100% probability, without forming unwanted secondary structures.
