paper review

BB-581 Term Project (Fall 2013)
Introduction: The purpose of this paper is to show us a relatively new method called Serial Femtosecond Crystallography (SFX). The motivation of this work is to use smaller sized crystals and to decrease the radiation damage on the crystal. Normally growing a protein crystal is pretty hard, and growing bigger crystals is even harder. Moreover, radiation damage is a problem for all methods related to any imaging and diffraction methods[1-3]. The researchers use hen egg-white lysozyme (HEWL)[4] as their sample which is the first enzyme determined by X-ray crystallography. They have taken several millions of individual snapshots of the diffraction pattern from nanocrystals of the protein contained in small water droplets. Their data are in reasonable agreement with those from standard X-ray crystallography.
Key methods: This paper is to introduce a new method SFX, which is based on X-ray crystallography. X-ray diffraction is based on the Bragg's Law which provides the relationship between the scattering angle and resolution:
λ = 2*d*sin(θ/2)
In standard X-ray crystallography, we need to first grow a crystal, which may take several months or several years, and then we need to collect the X-ray diffraction data and solve the phase problem by molecular replacement, perturbation methods or direct methods. After that, we need to interpret the electron density map built from the data and do some refinements. In X-ray diffraction, it is the electrons in the molecules that are responsible for scattering, The interaction of electrons with the X-ray photons could cause chemical bonds to be broken or rearranged, hence damaging the sample.
SFX uses X-ray free-electron lasers (XFEL) which could provide sufficient intensity in a short time duration. This short duration time limites the chemical damage. They also use Coherent X-ray Imaging (CXI) instrument[5] which could provide hard X-ray pulses. Hard X-ray means short wavelength which gives a higher resolution. A liquid microjet is used to spray out hydrated crystals.
In this paper, they use molecular replacement[6] to solve the phase problem. This method is to replace the unknown structure with a known model which is similar. The method is to place the model in every possible position and then calculate the expected pattern. If that matches the real data then the phase decided by the model is assumed to be a good approximation of the the true phase. In this particular experiment, the authors use turkey egg white lysozyme as the know model to simulate the structure of the actual hen egg white lysozyme.
Key results: This experiment use two duration times 40 fs and 5 fs. First, the authors collect 1.5 million diffraction patterns with the longer duration time. About 4.5% of their data is marked as crystal hit, 18.4% of which is integrated with the CrystFEL software to solve the structure. This gives them a high resolution of 1.9 Å. Their R-factor and R-free are 0.196 and 0.229, which are acceptable. Then the authors
collect 2 million diffraction pattern with the shorter 5 fs pulse, resulting in a 2% hit rate. With 26.3% of the crystal hit data integrated, the resolution is 1.9 Å and the R-factor/R-free are 0.189/0.227.
Interpretation and Conclusions : The authors have compared the results using two different pulse durations with previous low dose x-ray diffraction data. The R-factor is higher in SFX which means there exist a systematic difference. Both the electron density map and the Wilson B factor are similar in all cases, showing no significant radiation damage. The authors think that their data demonstrate the effectiveness of SFX. Under the exposure time used, the radiation damage is under control and hence SFX could solve the crystal size problem. Thus SFX could allow lots of difficult-to-crystallize molecules to be analyzed.
Critique: I do think this is a promising method which needs to be developed. I think lots of biochemists are suffering from the difficult process of crystal growth and the size limit of X-ray experiments. This method give us a way to use small crystals and reduce the radiation damage. That is really liberating. Equipments like XFEL and CSPAD would help biochemists and biophysists collect more useful data. In this work, the authors use hen egg white lysozyme, a structure well known from standard crystallography, to demonstrate their method. I think the sample is a very good choice for them.
The weakness I think is their result. Although the resolution could be high and radiation damage could be under control, both the R-factor and the R-free are higher than those from low-dose experiments. There may be systematic differences in structures obtained from SFX and from standard crystallography. And this experiment may have lower reproducibility. I hope the authors could explain in more detail the differences and solve this problem. A reader like myself would then have more faith in their experiment.
Actually this work is similar to what I am doing in the laboratory. In our experiment, we do not even use small crystals: we just use single molecules. Another advantage we have is that we use a high energy laser to align the non-oriented molecules. This would give us diffraction patterns with known orientations. However our data now is worse than theirs, so I do think they have a head-start. I am interested in their oncoming results and whish to learn some lesson from th em so to develop our own equipment.
References:
1. R. Neutze, R. Wouts, D. van der Spoel, E. Weckert, J. Hajdu,
Nature 406, 752 (2000).
2. A. Barty et al., Nat. Photonics 6, 35 (2011).
3. L. Lomb et al., Phys. Rev. B 84, 214111 (2011).
4. C. C. F. Blake et al., Nature 206, 757 (1965).
5. S. Boutet, G. J. Williams, New J. Phys. 12, 035024 (2010).
6. W. Kabsch, Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. J. Appl. Cryst. 26, 795 (1993).。