Fitting small molecules to cryo-electron microscopy data

Sweeney, Aaron Patrick (2022) Fitting small molecules to cryo-electron microscopy data. PhD thesis, Birkbeck, University of London.

Abstract

Recent innovations in the field of cryogenic-electron microscopy (cryo-EM) has enabled the visualisation of biological systems at atomic resolutions that rival that of X-ray crystallography. This is increasing the relevance of cryo-EM in the field of drug discovery, as it is now possible to solve high-resolution structures of biological complexes that may not have been amenable to crystallisation [1] and also in a more “native-like” state. However, it is not always possible to obtain structures to atomic resolutions with cryo-EM, currently only 16.28 % of structures deposited in the electron microscopy database [2] are at resolutions better than 3.0 Å, with the majority (45.05 %) at resolutions between 3.0 and 4.0 Å (correct as of December 2021). A vast body of work has been conducted with the aim of fitting biological macromolecules into cryo-EM at various resolutions [3–6]. However significantly less has been reported regarding the fitting of small molecules into cryo-EM maps. The work presented in this thesis aimed at developing methodologies that enable the fitting of small molecules to cryo-EM maps at resolutions from near atomic to 4.5Å. First, I used a fitting methodology that utilised consensus docking [7] in conjugation with a local difference mapping technique [8] to model the complex of the Eg5 kinesin motor domain with a novel inhibitor (GSK-1) in the presence of tubulin, into a 3.8 Å cryo-EM map (Chapter 2). The arrangement of structural elements within the protein allowed inferences to be made as to the mechanism of action of the drug [9]. Next, I present a new empirical molecular docking score for identifying correct ligand conformations within protein ligand complexes (Chapter 4). This score was integrated with goodness-of-fit scores commonly used for assessing the fit of biological molecules to cryo-EM maps [10]. Furthermore, we assessed the utility of this integrated score for fitting small molecules using simulated full maps and density difference maps (Chapter 4). This integrated score was then developed into a full methodology for fitting small molecules into cryo-EM maps, where its effectiveness was evaluated with experimental data at high (≤ 3.0 Å) and low (3.0 to 4.5 Å) resolution (Chapter 5). The accurate identification of protein ligand interactions from atomic models is an important consideration for drug discovery. To this end, a new software is presented that predicts protein ligand interactions using geometric parameters (Chapter 3). This software was benchmark using 35 high resolution protein-ligand complexes and compared to current state-of-the-art available software [11, 12]. Finally, I present the refined protein model of a Torpedo nicotinic acetylcholine receptor including the MX helix in a 6.6 Å cryo-EM map (Chapter 6). A combination of fitting software and bioinformatics identified the position of the MX helix relative to the cellular membrane. Our investigation suggested that the MX may function to entrap cholesterol, imposing rigidity to the receptor around the narrowest point of the central pore.

Metadata

Statistics

DownloadsShow export options

Export as XMLJSONCSV

Activity Overview

6 month trend

145Downloads

6 month trend

277Hits

Additional statistics are available via IRStats2.

Archive Staff Only (login required)

Edit/View Item