Protein structure determination by single-wavelength anomalous diffraction phasing of X-ray free-electron laser data

Abstract

Serial femtosecond crystallography (SFX) at X-ray free-electron lasers (XFELs) offers unprecedented possibilities for macromolecular structure determination of systems that are prone to radiation damage. However, phasing XFEL data de novo is complicated by the inherent inaccuracy of SFX data, and only a few successful examples, mostly based on exceedingly strong anomalous or isomorphous difference signals, have been reported. Here, it is shown that SFX data from thaumatin microcrystals can be successfully phased using only the weak anomalous scattering from the endogenous S atoms. Moreover, a step-by-step investigation is presented of the particular problems of SAD phasing of SFX data, analysing data from a derivative with a strong anomalous signal as well as the weak signal from endogenous S atoms.

Keywords: SAD phasing; SFX; X-ray free-electron lasers; XFELs; serial femtosecond crystallography; single-wavelength anomalous diffraction.

Figure 1

Histograms showing the distribution of the unit-cell parameters a, b and c (a) before and (b) after optimization of the detector distance using lysozyme–gadolinium images. The dashed lines are Gaussian fits to the individual histograms. Before optimization, broad, multimodal distributions are observed. After optimization, much narrower, single-Gaussian-like distributions are obtained. It is important to note that a lower number of images could be indexed after detector-distance optimization (171 909 before optimization and 155 605 after optimization), suggesting that the highest number of indexed images may not necessarily result in the best quality of the data set.

Figure 2

Relationship between the number of images used and (a) the anomalous signal strength S _ano and (b) the map quality CC_map both before (filled squares) and after (open circles) detector-distance refinement, as well as after the selection of images with a CC_min of >0.83 (grey triangle).

Figure 3

(a) Electron-density maps from different stages of the sulfur SAD phasing process for thaumatin using 125 000 SFX images contoured at 1σ. The map calculated from the S-SAD phases calculated by SHARP is shown on the left (‘sharp’), the map after solvent flattening in the middle (‘sol’) and the final refined map (2mF _o − DF _c) on the right (‘ref’). (b) As (a) but using all (363 300) available images. (c) Anomalous difference density map calculated using 125 000 images and phases from the final refined model (shown as a ribbon model; cysteine residues and the single methionine residue are depicted as stick models). The map (orange mesh) is contoured at 5σ.