A DNA-Imaging Bottleneck

by Eric Drexler on 2008/12/04

The Battelle-led roadmap, my recent talks, and Nanorex (the molecular-CAD company I’ve been advising) all emphasize structural DNA nanotechnology as a basis for developing large, complex, easily reconfigured frameworks for building composite nanosystems. This gives me a strong interest in the difficulties that hamper SDN research.

AFM has been used to image of DNA origami, as shown on a cover of the journal Nature
AFM images of flat DNA structures


“Folding DNA to create nanoscale shapes and patterns”
PWK Rothemund, Nature, 440:297–302 (2006). [pdf]

Cryo-electron tomography has been used to image a DNA octahedron as shown on a cover of the journal Nature
Cryo-electron tomography of a 3D DNA structure


“A 1.7-kilobase single-stranded DNA that folds into a nanoscale octahedron”
WM Shih, JD Quispe, GF Joyce, Nature 247:618-621(2004).

Paul Rothemund tells me that the scarcity of cryo-electron tomography capability is a problem. He and others have focused on 2D origami structures, not because DNA origami is an inherently 2D technology, but because flat structures can be laid down on mica surfaces and imaged with ordinary AFMs. 3D structures require 3D imaging, and with a resolution better than 3 nm, cryo-electron tomography [+pdf] (barely) does the job.

It’s important to increase capacity in cryo-electron tomography and to make more of the existing capacity available to SDN researchers. It would be great if some of this capacity were in user facilities (including DOE facilities like the Molecular Foundry, for example).

Full-structure images are cool, informative, and make great journal covers, but are they always necessary? Other techniques can answer many questions. One is simply whether a structure has failed to fold, or whether it has defects of a particular kind. This question can often be answered by checking a few distance measurements. Conventional TEM can be informative if the structures in a sample are decorated with (relatively opaque) gold nanoparticles at known locations.

Using AFM to measure the height of a structure and image a small patch on top of it can be highly diagnostic. Turberfield’s group used this method to characterize a family of DNA tetrahedra. Poking a structure and measuring its mechanical properties will often be adequate to detect defects; in many instances, a defect would make a structure weak and soft in a predictable way. (This is another measurement used by Turberfield’s group.)

Electron microscopists have developed a range of techniques for preparing imaging nanoscale biostructures, or rather, for imaging what remains after (for example) exposing them to vacuum, “shadowing” them with obliquely deposited metal vapor, and looking at the metal piles and shadows while the original structures are toasted and obliterated by the electron beam. Ugly, but useful, and there are other methods (for example, the uranyl formate stain technique applied to a 3D origami in this paper [pdf]). The craft of TEM sample preparation rests on a broad and ancient body of lore.

It is a general pattern that, as a technology for design and fabrication advances, relevant questions change with it. If a non-imaging measurement shows that the result has all its parts in very approximately the right place — assembled and folded, not scattered, aggregated, or unfolded — then it’s a good bet that the parts fit pretty much as intended. The design itself fills in the probable details. Add a few distance measurements, and confidence may be high enough to proceed to the next step in development. This may be entirely adequate when the product will become part of a system, and returning for more study is always an option.

Finally, when development leads to a product with a practical application, the most important test will become quite simple: Does it work?

Previous post:

Next post: