AFMfit : Deciphering conformational dynamics in AFM data using fast nonlinear NMA and FFT-based search

Abstract

Atomic Force Microscopy (AFM) offers a unique opportunity to study the conformational dynamics of proteins in near-physiological conditions at the single-molecule level. However, interpreting the two-dimensional molecular surfaces of multiple molecules measured in AFM experiments as three-dimensional conformational dynamics of a single molecule poses a significant challenge. Here, we present AFMfit, a flexible fitting procedure that deforms an input atomic model to match multiple AFM observations. The fitted models form a conformational ensemble that unambiguously describes the AFM experiment. Our method uses a new fast fitting algorithm based on the nonlinear Normal Mode Analysis (NMA) method NOLB to associate each molecule with its conformational state. AFMfit processes conformations of hundreds of AFM images of a single molecule in a few minutes on a single workstation, enabling analysis of larger datasets, including high-speed (HS)-AFM. We demonstrate the applications of our methods to synthetic and experimental AFM/HS-AFM data that include activated factor V and a membrane-embedded transient receptor potential channel TRPV3. AFMfit is an open-source Python package available at https://gricad-gitlab.univ-grenoble-alpes.fr/GruLab/AFMfit/.