Introduction
After more than 50 years of study, the structural properties of duplex DNA are very well understood—or so you might think. And for this reason, it frequently is pressed into service as a test bed for evaluating new biophysical methods. A number of these techniques provide distance information in solution, in the 1–10 nm range, and are sometimes called 'molecular rulers'1. Spectroscopic methods such as resonance energy transfer between donor and acceptor fluorophore pairs2, 3, or coupling between the unpaired electrons of spin labels4, 5, can fill in the distance 'gap' that exists beyond the range of nuclear Overhauser enhancements between protons in NMR. Another approach that is being applied more and more is small-angle X-ray scattering6. Mathew-Fenn et al.7 have developed a variation of this approach in which the scattering is used to measure distances between two gold clusters attached to a macromolecule. In a new paper in Science, they have applied the method to a series of DNA duplexes8 and found that double-stranded DNA is rather more stretchy than previously thought.
Gold nanocrystals are clusters of 75 gold atoms9 that can be attached to DNA via a 3' thiol linker, and because they are so electron dense (79 electrons per gold atom), they scatter X-rays very strongly10. When clusters are tethered to each end of a DNA duplex, interference between X-rays scattered from the two centers generates a modulation of the scattering profile. Fourier transformation of the extracted modulation generates a distribution of the center-center distances between the gold clusters. This provides two pieces of information: the mean separation and its spread.
In the new paper, Mathew-Fenn et al.8 have made a series of gold-labeled DNA duplexes ranging from 10 to 35 base pairs in length in steps of 5 base pairs. The length distribution for each duplex species was fitted to a Gaussian curve, from which mean center-center distances were calculated. This gave an average rise per base pair that was very close to that expected for a standard B-form helix. The distributions have a significant width, with a half-width of 
5 Å for a 20-base-pair duplex, for example. Some of this may be attributable to variation in cluster position relative to the DNA, but because the width increases with helix length, a significant proportion must be due to inherent stretching fluctuation in the DNA (see Fig. 1). The distributions are surprisingly symmetrical, as DNA is plainly not a rubber cylinder but has a molecular structure that is subject to conformation constraints and steric clash.
Figure 1: Models of double-stranded DNA.
Shown are DNA structures of normal geometry (center), and subject to compression (left) and extension (right), each by 10%. These images were generated using coordinates provided by R. Das (Stanford University).
Full size image (53 KB)However, in another respect, the behavior of the DNA was found to differ significantly from that of an elastic rod. The distribution widths for the different species were found to be quadratically related to helix length, which indicates that stretching variations are cooperative along the DNA. Perhaps this is not too surprising. It seems self-evident that stretching fluctuations would have to be coupled to changes in the helical twist, and thus variation in helical conformation might propagate over a turn or more of helix. A possible functional implication is that this variation might contribute to modulation of protein binding via indirect readout effects. This result is consistent with very recent FRET experiments involving resonance energy transfer between terminally stacked cyanine fluorophores, where a lateral averaging process is likely to be due in part to torsional flexibility in the DNA duplex11.
The analysis of the distribution widths also provided an estimate of the stretching modulus of DNA. This puts a value on the 'stretchiness' of DNA and is the force required to extend the helical length by a factor of two. The scattering data give a value of 91 pN. This can be compared with values measured in single-molecule force-extension experiments12, 13 that are an order of magnitude greater than this, which perhaps suggests that the application of stretching force stiffens the DNA significantly.
Mathew-Fenn et al. have provided us with a powerful new addition to the armory of methods for investigating nucleic acid structure. In so doing, they have also revealed some interesting new insights into DNA structure. The double helix is still capable of springing the odd surprise now and again.
