Biomod/2012/Harvard/BioDesign/large canvas SST
Large SST Canvas
Rationale of Approach
The large canvas approach was the basis for our final design, which can be found here. The large canvas design was created from what we had learned from our Small Canvas Approach and includes the same modifications that we performed on the template to bias the handles to one side (see modifications). With a large canvas, we could template just an interior portion. This prevents any type of aggregation we saw of our templated layer when we templated a the entire canvas. Also, by having part of the structure templated, we would see a clear juxtaposition of tempalted vs. non-templated area, making it easier to determine our success from AFM (which feels the contours of the structure).
Templating D-DNA on canvas
We initially tried to template the entire structure out of D-DNA (rather than L-DNA) on top of the D-DNA template. This experiment acted as a proof-of-concept of the final experiment, since D-DNA is much cheaper and ships much faster than the corresponding L-DNA.
Differences in Procedure
We used two different methods for annealing the first tiled strand with the template canvas. For the first method, we first annealed the template and the first ribbon in two steps. Since this method has 2 successive annealing steps, we call it the 2-pot reaction. The other method, which we call the 1-pot anneal, involved annealing the D-DNA canvas and the first L-DNA strands in a single annealing reaction.
When added in one pot, the D-DNA could have interacted with the template strands, potentially lowering our yield. This would not be the case with L-DNA, as L-DNA strands cannot interact with D-DNA.
Following either the 1-pot or 2-pot reaction, we added the 2nd L-DNA strand and annealed from 40°C down.
The following AFM images show the results of the final annealing reaction. As expected, a raised square (bright white indicates a higher surface) is visible in the center of many of the large canvases. The solutions shown are unpurified, which explains the relatively high concentration of incorrectly-formed structures.
Design 3 D-DNA, Full Templating, at 5μm
Design 3 D-DNA, Full Templating, at 1μm
Design 3 D-DNA, Full Templating, at 1μm, Repositioned
The structures seem to template in a high yield, though getting a perfectly templated layer happens at a much lower yield. In the above images, all of the structures are oddly clumped. They are not aggregated - as confirmed by a gel, so we explain the strange grouping as a failure to mix thoroughly when diluting our sample prior to imaging.
Validating Design of Template Modifications
Our template modifications (recall here) had the goal of trying to bias the handles towards one side. If our handles were not all on the same face, then our final structure would not template completely. We tested adding a 1 base pair gap - called "g", a TT, 2 base pair linker to provide space for the handle - abbreviated "tt", and both modifications - shortened as "gtt".
For large canvas, we attached our 3rd handle design (optimized against all of the large canvas template strands, labelled as H3. The 15-mer sequence of H3 is 'TATACATAACAATCA'.
All of the templates, themselves, formed normally to the extent that we could not distinguish them through the gel or through AFM imaging - unlike the small canvas. We carried on templating D-DNA ribbons above each of the modified templates, through the procedure described in the section above. The images from the above section are all of gttH3 - that's with a 1 base pair gap, a TT link sequences and then the handle to the modified strands. Compare those to the following two images of having just a gap and having just TT linker within the template:
Design 3 D-DNA, Gap without TT Linker, Full Templating, at 5μm
Design 3 D-DNA, TT Linker without Gap, Full Templating, at 2μm
The collection of these images convinced us that gtt was the best modification. We believe that the gap does geometrically favor one surface of the sheet, as we had expected. Adding another the TT base pair, by "filling in the gap" with a non-interacting base pair (the TT linker), we help the stability of the structure while still preserving the geometric freedom of the handle. Notice the image of modification with just the gaps, showed broken structures (destabilized) while the image of having just the linker sequence without the gaps produced a lower yield of templating (handles were not geometrically biased).
Supporting our hypothesis was a structure we formed by accident. In annealing this below template, we had forgotten to add in the modified interior strands. They formed as one would expect, without a rectangular center. But then, upon purification to add in the second strand (we didn't realize the middles were missing), the structure with a hole was more likely to rip apart during the slightly harsher conditions of centrifugation invovled in our gel purification protocol.
Templating L-DNA on canvas
We repeated the method above with L-DNA strands to achieve the final product of L-DNA templated on D-DNA. Other than the replacement of D-DNA ribbons with L-DNA, the methods remained the same.
As with the D-DNA proof-of-concept above, we analyzed the results using the AFM. Note that these AFM images are unpurified - to show percent yield. Notice that he broken structures are rarely templated and the ones that did maintain the full structure have a high rate of templating.
L-DNA Design 3, Full Template, 1-pot reaction, at 1.67 μm
L-DNA Design 5, Full Template, 1-pot reaction, at 2.5 μm
Using the nanoscope tools, we also showed a measure of the profile, to get specific data on the height, rather than using the color scaling of the image.
Vertical Profile of L-DNA Design 5, Full Template, at 500nm
Vertical Profile of L-DNA Design 5, Full Template, at 500nm
We also ran a sawtooth gel to compare the relative sizes of the original template canvas, the final product, and the intermediate template with the first L-DNA strand attached. The gel below shows the sawtooth gel comparing the original canvas (Orig), the canvas with L-DNA strand 1 (L3a/L5a) and the final canvas with both L-DNA strands (L3ab/L5ab) for L-DNA ribbon designs 3 and 5. As expected, the template with one strand annealed is significantly larger than the original, and the final template is larger than both the previous structures. Each of these differences resulted in a detectable gel shift.
By seeing the relative darkness of the gel band, we see that L-DNA Design 5 seems to produce a slightly higher yield than Ribbon Design 3, but both work (see L-DNA layer details. From this gel, we could have also purified our structures for imaging - but we did not pursue this - wanting to show what the general yield would look like overall.
|<< Small SST Canvas||DNA Origami >>|