Introduction: Atomic force microscopy (AFM) makes it possible to visualize the dynamics of individual biomolecules during their functional activity. All observations, however, are limited to regions accessible by a large enough probe tip when scanning. Therefore, AFM only images the biomolecular surface with limited spatial resolution, missing important information necessary for a detailed understanding of observed phenomena.
To facilitate the interpretation of experimental imagery, Romain Amyot and Holger Flechsig of NanoLSI have now developed the mathematical framework and computational methods to reconstruct 3D atomic structures from AFM surface scans. In their recent research paper which is a collaboration with experimental researchers, they provide applications to high-speed AFM images ranging from single molecular machines, protein filaments, even to large-scale assemblies of protein networks, and demonstrate how comprehensive atomistic information advances molecular understanding beyond topographic images.
The methods developed use AFM simulation, which was previously established by Amyot and Flechsig, and distributed in the free software package BioAFMviewer (PLoS Comput Biol, 2020). Simulation AFM computationally emulates the experimental biomolecule scan to translate structural data into topographic simulation AFM images that can be compared to real AFM images. An automated fitting procedure was then implemented to identify the high-resolution molecular structure behind an experimental limited-resolution AFM image. It is therefore possible to recover complete 3D atomic information from a simple scan of the protein surface obtained under AFM observations. To illustrate the explanatory potential of this realization, Flechsig describes: “Imagine that instead of just seeing the tip of an iceberg, you can now see everything that lies under the sea, to the extent that you can even detect impurities or density differences inside. its structure, helping you explain the coloring of icebergs.
To share developments with the large Bio-AFM community, all calculation methods are integrated into the user-friendly interactive BioAFMviewer software interface. The new methods are already being applied in many interdisciplinary collaborations to help understand experimental observations.
Figure. 1 Illustration of the method to reconstruct the 3D atomistic protein structure from an experimental AFM surface scan.
Figure. 2 From the HS-AFM image of the membrane channel array (left), the best-match simulation-AFM array (middle) provides the atomic-resolution arrangement of the proteins (right). This information allows for a molecular-level explanation of how these protein channels are activated.
This work was supported by the World Premier International Research Center Initiative (WPI), MEXT, Japan, and JSPS KAKENHI 21K03483.
Title: Simulation atomic force microscopy for the atomic reconstruction of biomolecular structures from experimental images with limited resolution.
Review: PLOS Computational Biology 18(3): e1009970 (2022)
Authors: Romain Amyot, Arin Marchesi, Clemens M Franz, Ignacio Casuso, Holger Flechsig
Posted: March 16, 2022