Integrative Modeling Approaches to Interpret High-Resolution cryo-EM Reconstructions

Abstract

Modeling 3D-EM reconstructions with computational tools currently enables the interpretation at near-atomic resolution of different functional states of macromolecules, thereby deciphering the functional mechanism of biologically relevant complexes. Recent advances in cryo-EM, such as direct electron detectors, specimen preparation, image processing, and data automation, are increasing the number of determined structures, particularly at high (<5 Å) resolutions. Here, several new integrative approaches are presented to retrieve structural information from these accurate reconstructions by incorporating modeling constraints from complementary biophysical techniques (crystallography, SAXS, FRET, etc.) or any other source of structural information (cross-linking, mutagenesis, prediction data, etc.). First, a two-step integrative approach was developed to unravel the topology of helical bundles using cryo-EM maps, distance restraints, and secondary structure predictions. This method unambiguously localized all helices of a key unassigned proteasome helical bundle and provided a topologically correct model that was later confirmed by crystallography. Second, our normal mode based flexible fitting algorithm, iMODFIT, was accelerated and adapted to deal with high resolution cryo-EM maps and other experimental constraints. Third, a fast loop-closure algorithm (RCD) was combined with integrative fitting strategies for modeling loops into unfilled densities. We strongly believe that these tools will facilitate the interpretation of the incoming high-resolution maps.