Single particle 2D Electron crystallography for membrane protein structure determination

Abstract

Proteins embedded into or attached to the cellular membrane perform crucial biological functions. Despite such importance, they remain among the most challenging targets of structural biology. Dedicated methods for membrane protein structure determination have been devised since decades, however with only partial success if compared to soluble proteins. One of these methods is 2D electron crystallography, in which the proteins are periodically arranged into a lipid bilayer. Using transmission electron microscopy to acquire projection images of samples containing such 2D crystals, which are embedded into a thin vitreous ice layer for radiation protection (cryo-EM), computer algorithms can be used to generate a 3D reconstruction of the protein. Unfortunately, in nearly every case, the 2D crystals are not flat and ordered enough to yield high-resolution reconstructions. Single particle analysis, on the other hand, is a technique that aligns projections of proteins isolated in solution in order to obtain a 3D reconstruction with a high success rate in terms of high resolution structures. In this thesis, we couple 2D crystal data processing with single particle analysis algorithms in order to perform a local correction of crystal distortions. We show that this approach not only allows reconstructions of much higher resolution than expected from the diffraction patterns obtained, but also reveals the existence of conformational heterogeneity within the 2D crystals. This structural variability can be linked to protein function, providing novel mechanistic insights and an explanation for why 2D crystals do not diffract to high resolution, in general. We present the computational methods that enable this hybrid approach, as well as other tools that aid several steps of cryo-EM data processing, from storage to postprocessing.