Self-assembly and characterization of the structure
The basic structure was designed as a scaffolded origami that forms an octahedron with edges consisting of double helices 27 bps long. The scaffold was a 370 nt RNA strand designed to run through the whole structure without crossing or base pairing with itself. Figure 1 shows a graphic representation of the structure. The scaffold is held together by eight staple strands. The four ’Z’ staples (blue in Figure 1) have a central 27-mer domain and two flanking 13-mer domains. The central domain spans a whole edge and together the four Z staples form a central square in the octahedron. The remaining edges going up or down, in respect to the plane of the square, is partly formed by the flanking 13-mer domains. The rest of these edges are formed by the ’V’ staples (green in Figure 1), which consist of two 14-mer domains. The Z staples form four vertices and the V staples form the remaining two. A linker of uracils (U-linker) between the individual domains was chosen to provide the flexibility needed to assemble the structure. Because the length of the edges was fixed and the vertices were not well-defined junctions such as Holiday or point-stars, the structure was expected to be somewhat wobbly. To make a fairly rigid structure and still assure the ability to assemble correctly, U-linkers of four nucleotides were chosen. For the experiments where dsRNA was not needed we used DNA staples to save time and/or cost.
Band shift assays
To show the self-assembly of the scaffold and the staples we made band shift assays. The very first self-assembly was done with a six-fold excess of staples (Figure 2). The gel shows a band shift of the self-assembly, which indicates the association of scaffold and staple strands. Whether the structure of the self-assembly is the expected octahedron or not, has not been proven by the experiment, but the band shift and the significant reduction in staples affirm the self-assembly. The two distinct bands above the band of the structure indicate that the structure might have been subject to multimerization. The multimerization can be amplified by an insufficient excess of staples, creating the possibility for staples to bind two scaffolds together, which is why the following self-assemblies were done using a 10-fold excess of staples (Figure 3).
Distance measurements using Förster Resonance Energy Transfer
Structures with fluorophores attached to two of the staples were assembled and the FRET efficiencies were calculated (Figure 4). The four structures (F1-F4) are identical except for the positioning of the fluorophores. The exact positioning of the fluorophores (Cy3 and Cy5) can be found in the Supplemetary. The transfer efficiencies are approximately 0.3 with F2 having the lowest (0.25) and F4 having the highest (0.31). Based on a simple geometrical evaluation the F1 structure was expected to have the highest efficiency, because the placement of the fluorophores were at the end of the staples closest to the vertex. The F4 structure was expected to have the lowest efficiency because the fluorophores were placed 4 nt from the vertex. The F2 and F3 structure were expected to have efficiencies between those of F1 and F4. The difference in efficiencies between F2 and F3 indicates a non-symmetrical position of the fluorophores, which was expected due to the turn of the helices and the helices ability to rotate due to the flexibility of the U-linkers. The change in distance was expected to be greater for F1 than for F4, which it is. Owing to the uncertainties introduced by the linkers and unknown orientation of the fluorophores the distances are only to be seen as an estimate. The center axes at the end of the helices were expected to be approximately 5 nm apart. Depending on the rotation of the helices and the rigidness of the structure, the distance between the fluorophores in the closed structure, including the length of the linkers, should be between 2 and 8 nm. Information about the linkers can be found in the Supplemetary. The calculated distances indicate proper assembly of the structure.
Small Angle X-ray Scattering
The SAXS measurements were preformed on DNA/RNA hybrid octahedrons in concentrations of 1,0 µg/µL, 0.5 µg/µL and 0.3 µg/µL. All samples had a volume of 60 µL in buffer (1 x TAE, 100 mM KCl, 0,025 % NaN3, 12 nM MgCl2, pH 7.2) and measurements were carried out at room temperature. The experimental data can be seen in Figure 5. The experimental curves for the different concentrations are similar. This similarity suggests that structure and amount of aggregation of the particles is not affected by diluting the sample to one third of the original concentration and the analysis is therefore continued with c = 1,0 µg/µL.
To obtain general information of our structure we made an indirect Fourier transform of the scattering data. This gives a size distribution function p(r), which is a histogram over internal pair-distances in the structure. This can for the c = 1,0 µg/µL sample be seen in Figure 6. The first bump at approximately 20 Å comes from the well-defined diameter of the hybrid structure. The maximum is located around 100 Å, indicating that this distance is the most abundant. The function goes to zero at 300 Å, thus this is the maximum distance observed. Distances this large, must be ascribed to the occurrence of oligomers, since the theoretical size of our structure is in the magnitude of 10 nm. This analysis and our expectations of how the structure should fold, gives us a starting point for designing a model for further study of the SAXS data.
To determine whether our SAXS data could confirm the formation of DNA/RNA hybrid octahedrals a model of the structure was fitted to the data. The model is an octahedron with sides consisting of helices with length corresponding to 27 base pairs. The fitting and the model is shown in Figure 7. It was possible to make a very good fit with this model to our experimental data. The χ2 value, a control parameter for the fit, was 4.2 which is very good for this type of fitting. The octahedron has the size 41.2±0.6 Å measured from the middle of the structure to the centre of the strands. However a few corrections to the model were included to obtain a good fit. These corrections can be seen as defects or deviations from the perfect octahedron structure and include a background from misfolded species that introduce flexible components and a small fraction of di- and trimers.
The inclusion of di- and trimers to the models does not necessarily mean that we have true di- and trimers, but gives us the possibility to include the contribution of larger oligomers in the model. These oligomers can form if the stable strands connect to more than one scaffold strand during the self assembly. This will lead to octahedrons joint together in larger structures. However the fraction of oliogomers is rather small with a contribution of ~ 3 % for dimmers and ~ 8 % for trimers, leading to a majority of monomers, with 89 %.
The SAXS data supports the formation of a stable DNA/RNA hybrid octahedron with a well-defined size with a small number of larger aggregates. The size we get from the SAXS analysis is matching the theoretical size of the octahedron and suggests thereby, that we have succeeded in folding the RNA/DNA according to our design.
Based on the band shift assays it is concluded that the scaffold and staple strands self-assemble into a structure of a well-defined size. The FRET studies further support the self-assembly of the structure. The FRET efficiencies indicate a positioning of the fluorophores that correspond to the designed structure. This is interpreted as a testament to the folding of the scaffold into the expected octahedron structure, which is further supported by the SAXS results.
Gene knockdown using the RNAi pathway
Small interfering RNA
Believed to be an antiviral defense pathway, the siRNA pathway is triggered when foreign doublestranded RNA is introduced into the cell. The enzyme Dicer initially degrades the longer RNA into short dsRNA oligonucleotides of 21-23 nt in length called small interfering RNAs (siRNAs). These siRNA have a 2 nt single-stranded overhang in each 3' end and there is perfect base-pairing between the two strands. The processed siRNA is recognized by the RNA-induced silencing complex (RISC). Only one strand, the antisense or guide strand, is incorporated into RISC. The sense, or passenger, strand is discarded and degraded. Following incorporation, the guide strand can anneal to an mRNA with a sequence exactly complementary to its own, when this happens the mRNA is cleaved by an RNA endonuclease within RISC, while the guide strand is left intact. This mechanism allows for a quite small amount of siRNA to cause major down regulation of their target gene. It has been shown that synthetic siRNAs can also be incorporated into RISC, and trigger mRNA cleavage. This is very promising from a medical point of view because it allows for the design of siRNAs against virtually any gene endogenous or foreign.
It is also believed that Dicer plays a roll in the loading of siRNA into RISC as it has been shown that a 27 bp RNA Dicer substrate yields a more efficient knockdown than mature siRNA.
The purpose of this assay is similar to the band shift assay. When the structure is degraded into siRNAs by Dicer, a band in the 20-25 nt area should appear. An RNA/RNA structure with an internally radioactively labeled scaffold was made. The octahedron had five 3’-overhangs, which should make it possible for Dicer to cleave the structure into 5 siRNAs. The gel shows that only the structure, which was supposed to be substrate for Dicer, is cut into pieces similar in size to siRNAs (Figure 8). The fact that Dicer treatment degrades at least parts of the structure to siRNA-sized fragments is a very strong indicator that the structure not only assembles as intended in the design, but also that the design is working as intended, and that the structure could be used in experiments to specifically silence targeted genes. However, because no probing experiments were done there is no way of knowing if it is only the desired bits that are cut out of the structure or if the Dicer cleavage is less specific.
The purpose of the designed dsRNA octahedron was to cause a knockdown effect of Renilla Luciferase as a proof of concept. To investigate the knockdown effect of our structure, we transfected it into a stable H1299 lung cancer cell line, and afterwards conducting a dual Luciferase assay.
The results are presented in Figure 9. The three positive controls all gave a knockdown around 96% (P=0.01441 and P=0.01450), which is significant knockdowns. Transfection with purified structure gave a knockdown around 40% (P=0.1425) compared to the siEGFP mismatch negative control. Cells transfected with a non-prurified sample of our structure also containing high staple strand concentration, obtained a knockdown around 80% (P=0.0263) compared to the siEGF mismatch control. The difference in knockdown between these two samples can be caused by the stociometric difference in the transfections. It was not possible for us to determine the concentration of the structure after self-assembly. The estimated concentration of the structure in non-purified sample is 16nM, so lower than the 40nM in the positive controls. The concentration of the purified structure must be even lower, because of the purification process. From the band intensity on an agarose gel assay we estimated it to be around 2nM. For the two knockdown assays the statistical p-value is calculated to p=0.1425 and p=0.0263 for purified and non-purified respectively. For a knockdown we accept p-values below 0.05. From the calculated values it is therefore not statistically significant to claim that the purified structure forsake a knockdown effect, but oppositely for the non-purified we see a significant effect. As it is seen that siEGFP at concentration 2nM gives a clearly significant knockdown, the low concentration of structure cannot be fully blamed for the lacking knockdown, but it might also be the cleavage by Dicer which is not fully effective.
We did not obtain a statistically significant knockdown from our purified structure, but we did obtain one from the non-purified. An explanation could be that we had a higher concentration of the actual structure in the non-purified sample, meaning this one could cause more knock-down. The purification process meant we got rid of the majority of the excess staples, but also the structure. This lower concentration of the structure was non-the-less able to cause some knockdown. We need to increase the concentration of our structure in order to obtain better results and better be able to compare them with the strandard RNAi assay of siEGFP. The purified structure had a concentration at least 40 times lower than the siEGFP, and the non-purified had one of 10 times less. Generally this study leads us to conclude that there is a clear reason for further investigating the knockdown effect of our structure, but have not found proof for the knockdown effect.
Controlled opening of the structure
To show that the structure can be used as a responsive dynamic nanocarrier we modified the octahedron with a lock, which can be opened with a specific key (Figure 9). The lock consists of two staple strands that form a double helix of 14 bps and a 8 nt toehold. The key (ssDNA) is expected to bind to the toehold and use strand displacement to open the structure. Fluorophores were attached to the modified staples making it possible to visualize the opening of the structure by FRET. Four structures were made, each with a different combination of the fluorophores. A detailed 2D blueprint with the exact placement of the fluorophores, along with a list of the sequences, can be found in the Supplementary.
Upon addition of the key the fluorescence intensity of donor increases concurrent with a decrease in the fluorescence intensity of the acceptor (Figure 10). The normalized fluorescence spectra before and after addition of the key (Figure 11) also show an increase in the intensity of the donor after addition of the key, which is caused by a lower energy transfer between the donor and acceptor. The calculated distances (Figure 12) confirm that we have successfully designed a responsive dynamic nanostructure.