Enhanced Sampling Assisted Flexible Fitting of Atomic Structures into Electron Microscopy Maps

Abstract

For many problems in structural biology, flexible-fitting computational algorithms are often useful in interpreting low-resolution electron microscopy (EM) maps of macromolecular assemblies. A widely used atomistic simulation technique is molecular dynamics flexible fitting (MDFF), which has been applied to generate structural models of large complexes. All-atom explicit-solvent MDFF simulations are not only computationally demanding, but also can be sensitive to the resolution of the target EM map. Moreover, functional movements of many biomolecules require large-scale conformational reorganization elicited via domain translations/rotations, where methods such as MDFF may be limited in capturing the rotations of structural elements. To decrease the computational cost and alleviate the limitations stemming from domain orientations, one can combine MDFF with an enhanced sampling technique to accelerate the conformational search in a single atomistic simulation. In this work, we judiciously combine MDFF with temperature-accelerated molecular dynamics (TAMD), an enhanced sampling method, and carry out TAMD-assisted MDFF (TAMDFF) simulations of proteins and nucleic acids. We find that TAMDFF simulations can achieve target structures of similar quality as MDFF on short timescales. In some cases, only TAMDFF simulations are able to capture conformational changes likely because MDFF simulations are unable to overcome the underlying free-energy barriers. We suggest that TAMDFF may be a viable strategy for structural refinement of large ribonucleoprotein complexes such as the ribosome.