Research Article: Towards the spatial resolution of metalloprotein charge states by detailed modeling of XFEL crystallographic diffraction

Date Published: February 01, 2020

Publisher: International Union of Crystallography

Author(s): Nicholas K. Sauter, Jan Kern, Junko Yano, James M. Holton.


Electronic configurations at distinct metal centers within a metalloprotein may be characterized by inspecting the scattering factors at the X-ray absorption edge. Such experiments may be feasible at XFEL sources, using Bayesian data analysis.

Partial Text

For proteins containing transition metal sites, a complete understanding of function requires not only the atomic structure, but also the electronic structure and chemical environment of the metal atoms (Kern et al., 2015 ▸). X-ray absorption spectroscopy has been highly informative, with the extended X-ray absorption fine structure (EXAFS) offering a sensitive measurement of metal–metal and metal–ligand distances, whereas the X-ray absorption near-edge structure (XANES) classically reveals the oxidation state and coordination geometry (Yano et al., 2005 ▸; Glatzel & Bergmann, 2005 ▸). Fundamentally, the K-absorption edge, corresponding to the removal of a core 1s electron, is shifted to a slightly higher energy when a transition metal is oxidized, as the loss of a valence electron increases the interaction between core electrons and the nucleus (Fig. 1 ▸; Sherrell, 2014 ▸).

A total of 100 000 simulated diffraction patterns were processed with dials.stills_process (Method 3, Fig. 6 ▸). The 67 936 patterns with ≥3 DIALS-identified Bragg spots in the region of interest (Fig. 4 ▸) yielded 305 777 ‘shoeboxes’ (rectangular boxes each containing a Bragg spot plus background, Fig. 8 ▸), representing 100% of the 8241 unique Miller indices in the C2 asymmetric unit that span the 2.1–2.5 Å resolution range, implying an average 37-fold multiplicity of observation. These contained a total of 106 628 830 pixels (both background and Bragg spot) to be used for maximum-likelihood estimation of the energy-dependent anomalous scattering parameters at the two iron centers in ferredoxin.

The maximum-likelihood analysis presented above offers a path for using XFEL diffraction as a spatially resolved spectroscopic method. Anomalous scattering has the potential for distinguishing the electronic environment at metalloprotein metal sites (Einsle et al., 2007 ▸), but such a measurement has yet to be achieved under the time-resolved, physiologically relevant conditions that are possible with XFELs. Several-atom cofactors such as the [4Mn:5O:Ca] oxygen-evolving complex of photosystem II have been investigated using X-ray emission spectroscopy at the K-edge, but this does not distinguish among the multiple Mn sites (Kern et al., 2018 ▸). There are certainly many practical challenges: the anomalous scattering contribution is small compared with the overall diffraction (Fig. 7 ▸), the XFEL pulse’s broad bandpass smears out the energy-dependence of the signal (Figs. 1 ▸ and 8 ▸), and it has been notoriously difficult to scale XFEL-measured Bragg spots into self-consistent structure factor amplitudes. However, the consideration of simulated data (Table 3 ▸; Fig. 9 ▸) suggests that the anomalous scattering technique is possible with present XFEL instrumentation, provided that the incident X-ray spectra are measured to normalize the energy dependence (Zhu et al., 2012 ▸; Fig. 3 ▸), the high-resolution Bragg diffraction is imaged by a pixel array positioned far enough back to spread out the energies (Fig. 8 ▸) and detailed physical modeling (such as nanoBragg) is applied to the signals from each pixel using sufficiently large datasets that are best analyzed by current petascale supercomputers.

The availability of XFEL beamlines has facilitated the study of proteins under physiological conditions free from radiation damage. For metalloenzymes in particular, time resolution has also been key for the study of catalytic mechanisms. In order to fully exploit the potential of time-resolved measurements, we have previously developed multimessenger techniques, simultaneously combining the results from X-ray diffraction for reporting the atomic structure, and X-ray emission spectroscopy for reporting the electronic state of active site transition metals (Kern et al., 2013 ▸, 2018 ▸; Young et al., 2016 ▸; Fuller et al., 2017 ▸; Fransson et al., 2018 ▸). Now, based on the current results, there is the potential of adding a third reporter to follow the time-dependence of the spatially resolved anomalous scattering factors and the underlying metal chemistry over the course of the reaction cycle. This information can be obtained without additional experiments, provided that the X-ray diffraction is collected at the metal absorption edge, hence avoiding the problems of normalization and comparability between different separate measurements. We hope that this approach will be a driver for future experimental design, and with respect to detectors and beam spectrometers, for XFEL endstation development.

The program nanoBragg is available as a standalone C program at In this work, nanoBragg was ported into the open-source Python/C++ framework of CCTBX and can be downloaded at All scripts for reproducing this work are at, and in particular see the README file under paper1.




Leave a Reply

Your email address will not be published.