Practical approaches to macromolecular X-ray structure determination

Abstract

The X-ray structure determination of macromolecules is unsurpassed in precision and detail. However, such structures cannot be derived directly from X-ray diffraction data. Phases and a molecular model are needed for the interpretation of the measured reflections; the structure needs to be solved.In this dissertation, practical approaches for the structure solution finding the phases and refining the model - are given. Two unknown protein structures were solved and a new tool for the analysis of anomalous and heavy-atom density in structures was developed. The structure of human RNase T2 was solved by molecular replacement (MR) with the program PHASER in combination with SHELXE. It is shown that in cases where MR cannot clearly solve a structure, or the correct solution is not clearly indicated, density modification and auto-tracing with SHELXE can provide additional phase information and clearly point to the correct solution.The X-ray structure of Hellethionin D from Helleborus purpurascens was solved with the NMR structure ensemble as starting point. The program ARCIMBOLDO allowed for a multi-solution MR approach, and the phase information was combined with weak phase information from anomalously scattering atoms in the structure. Again, density modification and auto-tracing in SHELXE served to improve the phases, to minimize model bias and to validate MR solutions. Why the structure could not be determined by conventional sulfur-SAD (single wavelength anomalous diffraction) phasing remains unclear. Comparison with artificial data suggests that better understanding of the nature of the disordered solvent regions in combination with the anomalous signal might lead to an answer. The evaluation of the anomalous signal led to the development of the program ANODE, which estimates anomalous (or heavy-atom) density. This tool allows for even very weak anomalously scattering atoms to be located in the structure. It could be shown that ANODE can serve for structure validation, to visualize radiation damage and for MR-SAD phasing. The program also gives a good indication of the mean phase error of the model employed and the quality of long-wavelength data sets. ANODE supplements experimental phasing in SHELX and can be easily automatized.To make macromolecular refinement in SHELXL more feasible at medium resolution, two rigid-bond restraints for the refinement of macromolecular structures in SHELXL were developed and tested. For a highly automatized and robust test procedure, several scripts and a test structure library were prepared. Also, a new constraint has been introduced to SHELXL, which can be used to prohibit ADPs from becoming non-positive definite. While the restraints could not be proven to enhance the macromolecular model in all cases, the constraint will be implemented in the distributed version of SHELXL.