We know what DNA looks like and have been looking at it for nearly 60 years. So why has a new analysis of DNA structure been reported so poorly?
by Stephen Curry
I’m not an angry man but a new analysis of the structure of DNA using electron microscopy made me cross yesterday. It wasn’t the fault of the scientists involved, but the sloppy way the result was reported that got my scientific goat.
The structure of DNA was first determined almost 60 years ago by Watson’s and Crick’s famous analysis of the scattering patterns recorded by Maurice Wilkins and Rosalind Franklin as they fired beams of X-rays at narrow fibres of the stuff. We have had a long time to refine and digest this result so I was surprised to run across so much inaccurate information in the internet digests of the new finding, reported in the journal Nano Letters by an Italian group led by Enzo di Fabrizio.
The web-site io9.com headlined George Dvorsky’s piece “Scientists snap a picture of DNA’s double helix for the very first time.” No, they hadn’t. The accompanying article interspersed fact with fancy before finally concluding that the new imaging technique would enable us to see “how it interacts with proteins and RNA”. No, it won’t.
I’ll explain why in a minute but first let’s look at New Scientist’s coverage of the same paper. This was a more measured and more accurate account of the new result but the piece got off to a bad start. Roland Pease’s article claimed that “an electron microscope has captured the famous Watson-Crick double helix in all its glory.” But it clearly hadn’t. The accompanying image was fuzzy and did not show a double helix that resembled the one described by Watson and Crick
Model and electron micrograph of a DNA fibre
Pease followed this up with the same ill-founded claim that the new method would allow researchers to see how other biomolecules interact with DNA. I’m not sure where this claim has come from because it’s not in the paper. A faulty press release perhaps?
Finally Alex Wild blogged about the article at Scientific American. His post, riskily titled “What DNA actually looks like” claimed that the paper reports “the first ever microscope image of an isolated DNA molecule”. If he had take the nanotrouble to type “electron micrograph DNA” into a Google search, he would have seen there are plenty of earlier microscope images of DNA. If he had looked more carefully at the abstract of the paper — helpfully illustrated in this instance (see above) — he would have seen that the sample was in fact a bundle of DNA molecules, not an isolated one. Sheesh.
Why make a fuss about this? OK, in part because I use X-ray crystallography rather than electron microscopy to look at the structures of interesting biological molecules in my research and the exaggerated claims made on behalf of electron microscopy by the science writers were fist-clenchingly annoying. We scientists are a territorial lot, you know.
I shouldn’t get so worked up because the problem is largely due to the seductive power of the image, something I’ve puzzled over before. All three writers focused on the fact that the new paper reported a picture that you can see; in contrast, although the X-ray method yields a much higher level of detail, its results are produced indirectly through mathematical analysis of the way that molecules scatter an X-ray beam. Dvorsky, Pease and Wild may not have fully grasped that the indirectness of X-ray crystallography in no way diminishes the quality of the information obtained — admittedly not something I would expect a non-specialist to know — but the allure of the image nevertheless seems to have dulled their vision. What frustrated me is that, in spite of having an image served up to them, they didn’t look at it properly, and that allowed errors to creep in.
What is actually new in the paper is that the authors have been able to take a high-contrast image of a DNA fibre (made up of a bundle of DNA double-helices) using electron microscopy. They did this by drying out a drop of DNA dissolved in water over a layer of silicon that had been micro-fabricated to have an array of tiny pillars across its surface. As the water evaporated, strands of DNA were left stretched between the pillars. Because they are suspended above the silicon base, it was possible to get a good image of the DNA fibres (you get poorer contrast if the DNA is lying on a solid surface). In a nice touch, the authors note that their sample preparation is similar to the method used by Wilkins, but they got fibres that were about a thousand times finer than he was able to achieve.
And what do they see? There is certainly some fine structure in the image. There are repetitive features of the size expected for the helical structure in DNA. But it was clear to the Italian researchers and should have been clear to anyone looking at the picture in their online abstract, that the image is not of a single molecule of DNA but a bundle of them. Di Fabrizio and colleagues modelled the structure as a bunch of seven parallel DNA double-helices since that generated a structure with the same thickness as the imaged fibre.
However, is their model correct? If you look at the inset detail in the figure, you see that the indentations on the underside are much deeper than those on the DNA model (middle panel). Perhaps this is an artefact of the way that electron microscopes make images. I don’t know because I am not an expert and the authors don’t comment on the discrepancy.
The bundled nature of the DNA samples prepared for these experiments also helps to explain why the microscopy technique will be unsuitable for analysing the interactions of protein molecules with DNA, contrary to the claims of Dvorsky and Pease. DNA bundles do not occur naturally; in living cells when DNA is being manipulated by proteins — to be copied or used to produce instructions for cellular processes — the double-helix has to be prised apart into separate strands so that the genetic code can be read. We are not likely to be able to investigate these processes using samples composed of tightly packed bundles of DNA double-helices.
Even if a single DNA strand could be isolated and imaged by electron microscopy, the fact that the method relies on largely drying out the sample makes it unsuitable for analysing any proteins bound, since these molecules depend critically on being immersed in water to work properly.
What all this tells you is that Nature is a bitch who likes to make life hard for scientists. Fair play to the Italians who have refined the techniques for preparing DNA fibres to a new level, but it remains to be seen whether their technique will reveal any interesting new biology. I wouldn’t bet on it just yet. Scientists have more work to do, and so too, do science writers.
Read more: www.guardian.co.uk