Abstract

Cryo-EM and artificial intelligence: A marriage made in cellular fractions

Advances in electron microscopy have provided unprecedented access to the structural characterization of large, flexible, and heterogeneous complexes. Until recently, cryo-electron microscopy (cryo-EM) has been applied to understand molecular organization in either highly purified, isolated biomolecules or in situ. An emerging field is developing, bridging the gap between the two approaches, studying molecular organization in native cellular fractions. Here, I will present structural insights into previously elusive megadalton protein complexes uncovered via integrative analysis of cellular fractions. I will also introduce an AI-assisted, cryo-EM image analysis, model identification, and reconstruction pipeline for studying endogenous protein complexes. After integrating traditional biochemical sample characterization and external database information, it is feasible to obtain structural insights from multiple, simultaneously resolved atomic models of higher-order protein complexes. Overall, our methods and data provide a framework for further understanding cellular organization by systematically mapping the structure of endogenous biomolecular assemblies within cellular fractions.