Whole pattern analysis of serial protein X-ray crystallography diffraction data

Abstract

Serial X-ray crystallography has quickly developed as an approach for the determination of crystal structures from proteins that do not form crystals of sufficient size or quality for conventional X-ray crystallography experiments. The opening of X-ray free-electron lasers sources has spurred developments in experimental and analytical techniques in this field and has motivated the use of similar approaches at microfocused synchrotron sources. In these experiments, final data-sets are typically formed from collections of diffraction images from single exposures of individual protein micro- and nano- crystals. Variations in crystal and beam characteristics and in crystal orientations may be expected from image to image. Data processing and analysis methods have developed in response to the unique requirements posed by the experimental approach of serial X-ray crystallography. This thesis presents and examines a new analysis approach for the extraction of structure factor amplitudes from serial X-ray crystallography data.

Crystal size and crystal disorder can affect the distribution of intensities between Bragg peak positions in the regime of crystals as finite diffracting objects. This thesis examines the diffracted intensity distributions formed from the merging of data from such protein crystals. A general procedure for the extraction of structure factor amplitudes from these distributions is presented that is based on an approach of whole-pattern fitting analysis. This approach holds similarities to intensity extraction techniques used in powder diffraction, yet is given a higher dimensionality in this work due to the collection of diffraction data from individual protein crystals that is achieved in serial X-ray crystallography experiments.

This thesis demonstrates that the modeling and whole-pattern fitting analysis of continuous diffracted intensities distributions can improve the accuracy of extracted structure factor amplitudes. Simulation studies are presented to examine the feasibility of whole-pattern fitting for the analysis of merged diffraction data from finite protein crystals under several conditions, including two simple models of crystal disorder. Significant stages in the treatment of serial X-ray crystallography data occurs prior to the extraction of structure factor amplitudes. These include the sorting of diffraction images, the indexing of images from unconstrained orientations and the merging of images to form a single data-set for the determination of the crystal structure. These steps are shown in this thesis for two separate serial X-ray crystallography experiments performed using synchrotron radiation. Finally, the application of the whole-pattern fitting analysis to an experimental data-set is presented and protein crystal structures are determined.

A general framework for the processing, merging and analysis of serial X-ray crystallography data is presented in this thesis. The applicability of this analysis framework can be regarded to be most relevant for the nanoscale protein crystals, for which extended intensity distributions around Bragg locations may be expected. In such cases, a three-dimensional powder diffraction analysis approach as presented here may be valuable for the determination of new protein structures.