A novel mirror-symmetry analysis approach for the study of macromolecular assemblies imaged by electron microscopy

Multivariate statistical symmetry analysis is widely employed in single-particle electron-microscopy studies for the detection of symmetry components within a set of noisy two-dimensional images. So far, this technique has been used to retrieve information from the analysis of end-on view oriented particles only. Here, we propose a method to detect symmetry components from side- and tilted-view oriented particles. This method is validated using a number of in silico generated as well as real datasets, can be used to analyze stoichiometrically heterogeneous datasets, and is useful for separating particle datasets with respect to their symmetry components. Additionally, translational components in lock-washer ring configurations can be detected. Most relevantly, this method represents a powerful tool for the characterisation of distinct symmetry components within multi-layered protein assemblies, and any putative symmetry mismatch between layers. Such configurations have often been postulated, though rarely observed directly, and are thought to have a crucial role in conferring dynamicity to molecular machineries like nucleic acid packaging motors, ClpAP/ClpXP proteases, flagellar motors and the F1/F0 ATPase.