Maximizing Electron Signal

The field of cryogenic electron microscopy (cryoEM) has greatly benefitted from improvements upon the integration of complementary structural determination methods and the development of high-throughput processing workflows aimed at producing high-resolution structures of proteins. Herein I will present a study integrating cryoEM with mass spectrometry and atomic modelling followed by two studies covering the development of a high-throughput cryogenic electron tomography (cryoET) workflow aimed at high-resolution cryoET. The development and continued adoption of direct electron detectors (DEDs) over the past ten years has opened up new avenues for software development. DEDs have brought a substantial increase in detection efficiency over the entire frequency range relative to previous generation cameras, increasing the signal-to-noise ratio (SNR) of electron interference patterns generated by thin, frozen-hydrated biological specimen. In addition to the increase in SNR, the rapid pixel readout of DEDs allows for multiple frames to be recorded per exposure. With these advances has come the ability to align frames together to account for beam induced specimen movement and the ability to account for structural degradation over the course of an exposure due to beam damage. The technique of tilt-series alignment by cross correlation in cryoET where the sample is tilted at increments during a single exposure then iteratively aligned had previously been limited for most samples by the relatively low SNR of CCD cameras. The increase in SNR of DEDs has allowed for robust tilt-series alignment by cross correlation, defocus estimation, and frame alignment. These hardware advances have paved the way for the development and optimization of cryoET processing workflows using cross-correlation based tilt-series alignment. High throughput, high-resolution cryoET workflow development will likely broaden the types of samples whose native or near-native structures have not yet been solved that are not amenable to single particle cryoEM often due to structural or compositional heterogeneity. Facilitating the determination of near-native protein structure and interactions with an improved cryoET workflow potentially allows for a better rational understanding of diseases and for more rational drug development. In Chapter 1 of this dissertation, I will introduce the theoretical and practical aspects of cryoEM. In Chapter 2, I use the highest resolution cryoEM structure of the conformationally heterogeneous COPII cage, collected with a CCD camera, in conjunction with known crystallography structures of its constituents and differential hydrogen/deuterium exchange mass spectroscopy results to create a pseudo-atomic model of the whole cage. In Chapter 3, I introduce the integration of a semi-automated implementation of the Protomo correlation-based tilt-series alignment software into the Leginon/Appion software suite, allowing for high-throughput and standardized cryoET processing. In Chapter 4, I introduce a complete, streamlined protocol for sub-nanometer sub-tomogram alignment from grid preparation to complete structure. In Chapter 4, I also present an application of this protocol to the determination of the structure of Adeno-Associated Virus serotype DJ with a bound drug, Arixtra, to sub-nanometer resolution, exemplifying the accuracy of correlation-based alignment as being comparable to fiducial-based alignment. Videos are included as supplementary files and a List of Videos with their descriptions is included on page xii.