MACHINE LEARNING FOR STRUCTURE DETERMINATION IN SINGLE-PARTICLE CRYO ELECTRON MICROSCOPY: A SYSTEMATIC REVIEW
Abstract : Traditionally, X-ray crystallography and NMR spectroscopy represent major workhorses of structural biologists, with the lion share of protein structures reported in protein data bank (PDB) being generated by these powerful techniques.Despite their wide utilization in protein structure determination, these two techniques have logical limitations, with X-ray crystallography being unsuitable for the analysis of highly dynamic structures and with NMR spectroscopy being restricted to the analysis of relatively small proteins In recent years, we have witnessed an explosive development of the techniques based on Cryo-electron microscopy (Cryo-EM) for structural characterization of biological molecules In fact, single-particle Cryo-EM is a special niche as it is a technique of choice for the structural analysis of large, structurally heterogeneous, and dynamic complexes. Here, sub-nanometer atomic resolution can be achieved (i.e., resolution below 10 Å) via single-particle imaging of non-crystalline specimens, with accurate 3D reconstruction being generated based on the computational averaging of multiple 2D projection images of the same particle that was frozen rapidly in solution.
? Sample heterogeneity and biomolecular flexibility remains a great challenge in crystal formation, hence predisposing the need for the development of novel techniques and multi-disciplinary approaches that are able to surmount existing limitations of the technique.
? High contrast is observed due to existing relative electron density between the sample and the surrounding stain.
? The emergence of single particle cryo-EM was a ray of hope for structural determination of biologically relevant macromolecules considered unamenable to existing methods.
? When faced with a problem, a reduction to its component parts is generally a common practice in science. Oftentimes, fundamental insights are readily obtained by piecing the individual parts back together.
? A deep understanding of the overall structure of a problem informs potential strategies that could be undertaken in solving the underlying constraint. This rationale particularly applies to the growing field of structural biology, recognized approximately 70 years ago following structural elucidation of the atomic structure of DNA and other globular proteins
? Therefore, new phase plates, based on high-intensity laser technology, are currently under development.
? This new technology is likely to replace VPP as it proposes no loss in image information and a constant phase shif
? Built on the collective efforts of researchers in this field, it is now possible to determine the structural architecture of biologically relevant macromolecules and assemblies (proteins and nucleic acids) that play very important roles in the overall coordination of human physiology.
? cryo-EM as a powerful tool for the reduction of sample heterogeneity
? So far in cryo-EM, various cryogenic substances, such as liquid helium, liquid ethane, and liquid nitrogen, have been used at some point.
? Thin-section electron microscopy was the first method used to have a glance at the structural organization of the RyR channel.
? Negative stain Cryo-EM was first used to model the 3D structure of the RyR channel at 40 Å resolution, giving insight into its mushroom-shaped morphology
|