Experimental phase determination with selenomethionine or mercury-derivatization in serial femtosecond crystallography

Abstract

Serial femtosecond crystallography (SFX) using X-ray free-electron lasers (XFELs) holds enormous potential for the structure determination of proteins for which it is difficult to produce large and high-quality crystals. SFX has been applied to various systems, but rarely to proteins that have previously unknown structures. Consequently, the majority of previously obtained SFX structures have been solved by the molecular replacement method. To facilitate protein structure determination by SFX, it is essential to establish phasing methods that work efficiently for SFX. Here, selenomethionine derivatization and mercury soaking have been investigated for SFX experiments using the high-energy XFEL at the SPring-8 Angstrom Compact Free-Electron Laser (SACLA), Hyogo, Japan. Three successful cases are reported of single-wavelength anomalous diffraction (SAD) phasing using X-rays of less than 1 Å wavelength with reasonable numbers of diffraction patterns (13 000, 60 000 and 11 000). It is demonstrated that the combination of high-energy X-rays from an XFEL and commonly used heavy-atom incorporation techniques will enable routine de novo structural determination of biomacromolecules.

Keywords: SAD phasing; XFELs; mercury soaking; selenomethionine derivatization; serial femtosecond crystallography.

Figure 1

Data quality and phasing statistics as a function of the number of patterns. FOM is reported by SHELXE for the correct hand. Map CC is the real-space CC between the model built by Buccaneer and the final refined 2mF _o − DF _c map. ‘Anode’ is the maximum peak height of the anomalous difference Fourier map calculated by ANODE (Thorn & Sheldrick, 2011 ▸) with the refined model. was calculated with F_A and σ(F_A) in the output of SHELXC (Sheldrick, 2010 ▸). The high-resolution cutoffs for Stem-Se, ACG-Se and LRE-Hg are 1.4, 1.5 and 1.5 Å, respectively. Note that the reason why the overall multiplicities do not increase in the same way despite the same Laue symmetry (for Stem-Se and LRE-Hg) is (i) a different resolution cutoff, (ii) a per-pattern resolution cutoff in merging, and (iii) different reciprocal-lattice point sizes determined for each pattern. This figure was prepared using ggplot2 (Wickham, 2009 ▸) in R (R Development Core Team, 2008 ▸).

Figure 2

Initial and final maps and models of Se-Met Stem. (a) An experimentally phased map and traced polyalanine model. (b) A 2mF _o − DF _c map and refined model. 13 000 indexed patterns of SeMet-derivative crystals were used for the calculation. Electron-density maps are contoured at 1.0σ.

Figure 3

Initial and final maps and models of ACG. (a) An experimentally phased map and traced polyalanine model. (b) A 2mF _o − DF _c map and refined model. 60 000 indexed patterns of SeMet-ACG crystals were used for the calculation. Electron-density maps are contoured at 1.0σ.

Figure 4

Initial and final maps and models of LRE-Hg. (a) An experimentally phased map and traced polyalanine model. (b) A 2mF _o − DF _c map and refined model. A total of 11 000 indexed patterns of Hg-derivative crystals were used for the calculation. Electron-density maps are contoured at 1.0σ.

Figure 5

Effect of the high-resolution cutoff and number of patterns in the case of LRE-Hg. The real-space CC of the model built by Buccaneer and the final refined 2mF _o − DF _c map are indicated by colors, which were calculated using phenix.get_cc_mtz_pdb (Adams et al., 2010 ▸). Success (CC 0.65) and failure of phasing are represented as circular and triangular symbols, respectively. This figure was prepared using ggplot2 (Wickham, 2009 ▸) in R (R Development Core Team, 2008 ▸).