multiSMD - A Python Toolset for Multidirectional Steered Molecular Dynamics

Abstract

Molecular forces govern all biological processes from cellular mechanics to molecular recognition events. Understanding the direction-dependence of these forces is particularly critical for elucidating fundamental interactions, such as protein-protein binding, ligand dissociation, and signal mechanotransduction. While steered molecular dynamics (SMD) simulations enable the study of force-induced transitions, conventional single-direction approaches may overlook anisotropic mechanical responses inherent to biomolecular systems. Therefore, probing the mechanical stability of molecular systems with respect to a director of an external force may provide critical information. Here, we present multiSMD, a Python-based tool that automates the setup and analysis of multidirectional SMD simulations in NAMD and GROMACS. By systematically probing forces along multiple spatial vectors, multiSMD captures direction-dependent phenomena, such as changing energy barriers or structural resilience, that remain hidden in standard SMD. We demonstrate the utility of our approach through three distinct applications: (i) anisotropic unbinding in a protein-protein complex, (ii) search for ligand dissociation pathways dependent on the pulling direction, and (iii) force-induced remodeling of intrinsically disordered regions in proteins. multiSMD streamlines the exploration of nanomechanical anisotropy in biomolecules, offering a computational framework to guide experiments (e.g., atomic force microscopy - AFM or optical tweezers) and uncover mechanistic properties inaccessible to single-axis methods.

1

(a) Flowchart illustrating the operation of a multiSMD. (b) Angles describing the successive generated pulling force vectors. (c) Directions of the external force application in parallel SMD simulations of the test system of the S-protein-ACE2 (angiotensin-converting enzyme) model.

2

Multidirectional analysis of SARS-CoV-2 S - ACE2 unbinding. (a) Schematic of nine pulling directions applied to the complex. (b) Simulation system of truncated S - ACE2 complex shown in green and gray with mutated residues shown as black sticks. (c) Maximum rupture forces across pulling directions for wild-type (WT) versus mutant (MUT) complexes. (d) Number of hydrogen bonds during S-protein pulling. For panels (c) and (d), the solid lines represent mean values, and the light-blue shaded areas (outline) represent the standard deviation (SD), both calculated from five independent replica simulations for each pulling direction.