Methods for cryo-electron microscopy single-particle analysis

We develop methods for single particle electron microscopy (SPEM). Our SPEM developments are motivated by our interest in transcriptional control because transcription regulatory complexes are among the most difficult SPEM targets. They are asymmetric and their composition and conformation are inherently dynamic as a consequence of their functional roles in the regulation of gene expression. Therefore, the structural and biochemical knowledge about higher order transcription regulatory assemblies is fragmental. Our research aims at overcoming this deficit in understanding and methodology to enable high-resolution structure determination of these and other similarly challenging targets for SPEM. The objectives of our SPEM developments are to (1) remove the requirement for a priori structural knowledge and open the method to the study of particles with a novel structure. Robust algorithms for ab initio 3D reconstruction are particularly important for the analysis of small particles with low symmetry. (2) enable the study of extremely heterogeneous particle populations with SPEM. Macromolecules are dynamic entities that move to exert their function. As we pointed out in our review published in Annual Reviews of Biochemistry current methods for so-called “heterogeneity analysis” are immature and suffer from serious shortcomings. A robust ensemble 3D reconstruction method would open for the study of biological systems currently intractable to structure determination by SPEM. (3) allow image processing orders of magnitudes faster than currently possible. This will provide the capability for immediate evaluation of data quality and adjustment of data acquisition parameters in real-time. For the researcher undertaking a SPEM investigation, this means increased insight into the nature of the data, improved success rate and enhanced turnover. (4) eliminate bias due to model, noise, and human. The problem of bias is central in SPEM and we must strive to develop approaches capable of overcoming the bias of all kinds. We have taken the first steps toward an approach free of bias, but the methods need to be generalized to extremely heterogeneous SPEM image populations and incorporate a more sophisticated image restoration methodology.