EXPERIMENTS AND COMPUTER SIMULATION
Most molecules are much to small to be seen using an ordinary light microscope, which can only resolve objects larger than half the wavelength of light, about 250 nm. Thus it might seem unlikely that it would be possible to manipulate or feel individual molecules. However, the scanning tip microscopes can do just that. This family of instruments all have a sharp tip, just like a phograph needle but much smaller, which can be moved along a surface to either determine its height profile or to measure some property of the surface at the atomic level. The atomic force microscope, a member of this group of instruments, was used by a team at the Naval Research Labs in Washington DC in 1994 to measure the force required to pull one DNA strand from the other in a helix 12 base pairs long [Lee et al., Science 266: 771, 1994].
In 1996, GeneVue was developing a DNA diagnostic instrument that stretched DNA fragments in order to determine their lengths. It was thus a relief to find that the force required to pull these DNA molecules apart was about 1000 fold greater than the forces required by our diagnostic proceedure. This would insure that the process we were using to measure the length of the DNA wouldn't destroy them.
However, we were interested in the behaviour of DNA molecules in a variety of environments with different lengths and nucleotide sequences. In a separate project we were using computer simulations of molecular dynamics of DNA to study drug binding, and thus we tried the same approach to explore the response of DNA to tension. We started with the same double stranded 12-mer used by the Lee et al. and were pleased to see the strands separate in our simulation at the same tension as seen by Lee's group when actually doing the experiment. However, before strand separation, as increasing tension caused the DNA to stretch to about twice it's length, we were surprised to see a sudden increase in length and a transition to a new flat ribbon structure.
Reading the literature (always a good idea, even if it's after the fact) revealed that two years before Watson and Crick published the helical structure of DNA, Wilkins et al. described an extended form of DNA [Nature 167:759, 1951], while the year before the use of the atomic force microscope by Lee's group, Bensimon et al. [Physical Rev Lett, 74:4754, 1995] showed that DNA could be stretched on the surface of a liquid to about two times its original length before breaking.
Since DNA molecules can be thousands or millions of base pairs long, their length can be much longer that the wavelength of light. As an example, the DNA genome of the bacterial virus Lambda is a single DNA hellix containing 50,502 base pairs. At 0.34 nm per base pair, it has an extended length of 17 microns. If one end of the Lambda DNA is attached to a fixed surface and a flourescent bead is attached to the other end, the length can be determined using an optical microscope while it is stretched by increasing tension. Two groups, in back to back papers [Cluzel et al., Science 271:792 and Smith et al., ibid 271:795, 1996] reported results of just this type of experiment. They each determined extension versus tension curves for this DNA, and found a sharp 2X length extension at the same value of tension as we saw in our computer simulations. We published our results a few months later in the November 13 th issue of the Journal of the American Chemical Society (of course it would have been more dramatic if we had published our results before those of the experimental groups, but better late than never).
Double stranded DNA is not the only molecular complex that is greatly elongated as one part is pulled from the other. The computer simulation by Grubmuller et al. [Science 271:997, 1996] of the separation of the biotin-streptavidin complex, shows a high tension over a distance of at least 0.9 nm (9 A) on biotin as it is removed from the binding pocket. As with DNA, this long dissociation path is possible because the complex deforms as the biotin is removed.
However, the formation of a molecular complex usually takes place over a smaller distance, and is not associated with large, energy requiring deformations of it's two parts. Thus, association and disassociation, in these cases, occurs by different pathways. In order to predict the stability of such a complex, it is necessary to know details of association and dissociation, knowledge of the energy of the complex alone is not sufficient. In order to model dissociation processes, it desirable to have powerful molecular dynamics software, which we are developing,