Ultrafast Biomolecule Imaging

Due to interaction with x-rays, radiation damage presents a fundimental limitation for x-ray crystallography: it is simply impossible to expose a sample indefinately since the sample is eventually destroyed. For this reason a practical limit has emerged for the minimum crystal size from which useful x-ray diffraction data can be recorded, & this limit equates to crystals being about 5 um in all directions. While a member of the laboratory of Janos Hajdu, we explored the possibility that short pulsed x-ray sources such as emerging x-ray free electron lasers (eg. in Stanford or Hamburg) could enable one to circumvent the radiation damage barrier. In essence we used molecular dynamics simulations to ask the question: "Can useful x-ray diffraction data be recorded from nm-scale samples using femtosecond x-ray pulses before the sample is destroyed by radiation damage?"

In our simulations we took a single molecule of lysozyme & exposed it in silico to a highly focussed extreme intensity x-ray pulse. The result was that the atoms of the sample became highly ionised during the pulse & electrostatic repulsion drove the rapid explosion of the sample. The figure on the right illustrates this event graphically. While it is obvious that the sample had little resemblance to its starting lysozyme structure by the end of the simulation, it is less clear exactly how much exploitable structural information is contained within the scattered x-ray data.

To explore this question we calculated the x-ray scattering from the sample over the entire x-ray pulse & compared this with that from a molecule which suffered no radiation damage. To quantify the extent to which the "radiation damage experiment" differed from a "radiation damage free experiment" we used an R-factor, similar to that used in crystallography. For conditions under which a relatively low value of the R-factor was recovered (somewhat arbitarily chosen as 15 %) it was judged that the radiation damage suffered was "acceptable" & there was a reasonable chance to record useful structural information.

The results of this analysis are shown above: the x-axis gives the pulse-duration in femtoseconds & the y-axis gives the number of 12 keV x-ray photons which passed through a 0.1 um focal spot. As would be expected, for a given x-ray pulse intensity a shorter pulse duration yeilded better quality x-ray diffraction data, reflected by the fact that the R-factor is lower in the bottom left corner of the diagram. The up-shot of our analysis was that, for x-ray pulses significantly shorter than 100 fs, it appears to be possible to record useful x-ray diffraction data from single-shot experiments on crystals of only a few nm in dimensions, & potentially from regular arrays like a virus capsid without the need for crystals. One would need to record & average data from numberous samples in order to build up complete data, but the result opens the possibility of a radically different approach to data-collection from very small protien samples. Since the publication of these simulations the work has formed a component of the scientific case for the developing x-ray free electron lasers, & Janos Hajdu in particular has been active in building collaborations aimed towards realising these long term goals.