Biomod/2012/Harvard/BioDesign/small canvas SST
Small SST Canvas
Basis of Design
As a starting point for our strands, we used the 10 helix by 10 turns SST canvas published by Wei et al., in 2012. However, we soon found out that the 10, 11, 11, 10 42-mer motif had a curvature to it.
The strand diagram as well as the Nanoengineer 3D model of the original SST design. Notice the large curvature even with just a few helices.
But they had done recent work on SST motifs (work not yet published) which involved alternating the lengths of the domains (9, 12, 10, 11) to get flatter structures. We decided to use "Motif I', shown by the diagrams below, which formed a flatter sheet than the original.
The strand diagram as well as the Nanoengineer 3D model of Motif 1. Notice the decreased curvature relative to the previous model.
Original Canvas Assembly
We first assembled the canvas without any modifications, as presented in the paper. The figures below show the results of small canvas assembly at 2μm, 1μm, and 667nm.
Original Small Canvas Unmodified (as in Wei et al., 2012), at 2μm
Original Small Canvas Unmodified (as in Wei et al., 2012), at 1μm
The first template we designed was a small (10 helix by 10 turn) canvas designed by Bryan Wei et al (2012). This canvas is made of 66 single stranded tiles (SSTs), 12 of which are protector strands bordering the top and bottom rows. We add handles (shown in red below) to rows 3, 5, 7, and 9 as attachment sites for L-DNA SSTs.
Since the handle is also made of D-DNA, we had to optimize its nucleotide sequence to reduce interference with the formation of the SST canvas.
We chose to test 2 different handles; the handle used by Bryan Wei et al. for biotin attachment (H1), and a new handle sequence optimized against the SST strands using [DyNAMiC Workbench http://yin.hms.harvard.edu/workbench/] (H2).
The sequence of H1 was GGAAGGGATGGAGGA; the sequence of H2 was AAGATTAAGATAGTT. Both were designed off 3 letter alphabets intentionally to help reduce unwanted mispairing with the template strands
We ultimately chose to work with H2 as this was an optimized handle to the Motif 1 Strands that we were using discussed above. We saw that the non-optimized handle actually interfered with our structure's formation seen further below
The images below show that the formation of the canvas is not affected by either the modification of the template or the handles that connect to it. The handles are just single strands of DNA and thus cannot be directly observed from the AFM image.
Images of the 10x10 SST structure (at 2μm and 500 nm) with both gap and TT modifications
For our project to work with the handle connections, all of the handles needed to be oriented on one side of the sheet, otherwise we would not be able to template a full layer. To bias this geometry, we experimented with various modifications to the template at the positions we wanted to attach the handles to.
Here are our modifications: (we used a similar strategy in designing the large canvas):
- No modification: Handle sequence was connected directly to the 3' end of the D-DNA SST.
- Gap: The 3' end of the D-DNA SST was shortened by 1 base pair before the handle sequence was attached, to bias the geometry of the handles to point in one direction, as seen by the nanoengineer diagram below and how the one missing base pair at the top left points into the screen, favoring connections on that side.
- TT spacer: Two excess thymine nucleotides added between the 3' end of the D-DNA SST and the handle sequences to provide some flexibility for the attached handles.
- Gap + TT: Both the gap and TT modifications described above were used
We predicted that the gap would help bias the geometry of the handles to point to one face and the TT linker would give enough flexibility to allow maximum templating.
We wanted to make sure that the modifications and handles did not affect the formation of our template structure. We used a gel to test this.
Note: the gel is labelled correctly, however, our naming convention has since changed - what we called TH is now called H. So the gel lanes test all of the template modifications for both H1 and H2.
You can see that in most cases, the modification did not affect the integrity of the structure (most of the gel bands are equal). However, we the one that is unique would be the gap-only test which formed a structure that moved further on the gel. This was a source of concern but wound up not having an effect (see analysis of modifications in Large Canvas It turns out that the gap only is a problem for structural integrity but as long as there is another random base there (such a thymine with the TT linker), the structure stabilizes). From the below gel, we also saw that TH1 interferred with the structure folding, causing some aggregates and less distinct bands where we expected them.
Templating First L-DNA Strands
We tried 2 different techniques to anneal the first L-DNA strand to the template. For the first technique, we first annealed the template canvas strands. We then added the first L-DNA strand in a second annealing reaction, stepping down from 40°C to prevent the template canvas from dissociating. Since this technique requires 2 annealing reactions, we called it the 2-pot reaction.
For the second technique, we combined the template strands with the first L-DNA strands in one annealing reaction. We called this the 1-pot reaction.
Following the 1-pot or 2-pot reaction, we purified the canvas structure to remove any excess L-DNA strands. Finally, we added the second L-DNA strands and annealed from 40°C down, again to avoid dissociation of the pre-annealed structure. Following this reaction, we obtained a rectangular L-DNA canvas attached to a D-DNA template of the same size.
The following AFM images show the template canvas with the first L-DNA strand annealed. With both 1 pot and 2 pot reactions, we do not see the structure's integrity affected by the extra strands (they still form the normal 10x10 structure)
1-pot and 2-pot anneals (respectively) with D-DNA Strand 3a attached at 2μm
2-pot anneals with D-DNA Strand 5a attached at 2μm
Templating Second L-DNA Strands
Following the annealing of the first L-DNA strands to the template and purification, we annealed the second L-DNA strands to the canvas
The images below show the AFM results for the final canvas with both L-DNA strands templated on it after an additional round of gel purification. We first added the strands at a 1:1 ratio of the L-DNA type B strands with the number of sites we expected. We expected these canvases to be twice as thick as the original canvas, since an L-DNA canvas of equal size should be templated on top. What we saw, was some aggregation of the L-DNA layer, which we attribute the type B strands, connecting two different templated portions together, in an aggregate.
A 5μm broad view of the L-DNA Ribbon 3, Fully Tempalted at a 1:1 Ratio of strand type A:B - notice the aggregated templated layers
To get around this aggregation, we tried adding in the second strand at a 1:100 less ratio than the number of sites. This would give us a much lower yield of the final product, but we hoped to see that our design would actually template. We noticed that by adding in the second strands at a much lower concentration, we got fully templated structures (though at the extremely low yield we expected as you can see from the below images). Because the yield was so low, it was hard to see whether the different L-DNA designs made any difference.
Note: the long white strands on the top 2 images are bits of agarose gel remaining from purification - another difficulty of imaging very small (10x10) nanostructures.
2-pot, Fully Templated, Small Canvas, with L-DNA Design 3, with 1/100 less of strand B at 2μm and 1 μm respectively
2-pot, Fully Templated, Small Canvas, with L-DNA Design 5 with 1/100 less of strand B at 2μm and 1 μm respectively
Following these results, we decided to re-approach our project using a different template - either larger SST canvas or using DNA origami. Follow the links below to these other approaches. A large canvas would allow us more flexibility to design around the issues we encountered here: L-DNA layer aggregation as well as difficulty of the AFM imaging process for small structures. DNA Origami, being 3D, would spatially limit where the handles could point (away from the dense structure).
|<< DNA Origami||Large SST Canvas >>|