Abstract:Glycosaminoglycans (GAGs) are negatively charged polysaccharides found on cell surfaces, where they regulate transport pathways of foreign molecules toward the cell. The structural and functional diversity of GAGs is largely attributed to varied sulfation patterns along the polymer chains, which makes understanding their molecular recognition mechanisms crucial. Molecular dynamics (MD) simulations, thanks to their unmatched microscopic resolution, have the potential to be a reference tool for exploring the patterns responsible for biologically relevant interactions. However, the capability of molecular dynamics force fields used in biosimulations to accurately capture sulfation-specific interactions is not well established, partly due to the intrinsic properties of GAGs that pose challenges for most experimental techniques. In this work, we evaluate the performance of molecular dynamics force fields for sulfated GAGs by studying ion pairing of Ca2+ to sulfated moieties─N-methylsulfamate and methylsulfate─that resemble N- and O-sulfation found in GAGs, respectively. We tested available nonpolarizable (CHARMM36 and GLYCAM06) and explicitly polarizable (Drude and AMOEBA) force fields, and derived new implicitly polarizable models through charge scaling (prosECCo75 and GLYCAM-ECC75) that are consistent with our developed "charge-scaling" framework. The calcium-sulfamate/sulfate interaction free energy profiles obtained with the tested force fields were compared against reference ab initio molecular dynamics (AIMD) simulations, which serve as a robust alternative to experiments. AIMD simulations indicate that the preferential Ca2+ binding mode to sulfated GAG groups is solvent-shared pairing. Only our scaled-charge models agree satisfactorily with the AIMD data, while all other force fields exhibit poorer agreement, sometimes even qualitatively. Surprisingly, even explicitly polarizable force fields display a notable disagreement with the AIMD data, likely attributed to difficulties in their optimization and possible inherent limitations in depicting high-charge-density ion interactions accurately. Finally, the underperforming force fields lead to unrealistic aggregation of sulfated saccharides, which qualitatively disagrees with our understanding of the soft glycocalyx environment. Our results highlight the importance of accurately treating electronic polarization in MD simulations of sulfated GAGs and caution against over-reliance on currently available models without thorough validation and optimization.

GCMe: Efficient implementation of the Gaussian core model with smeared electrostatic interactions for molecular dynamics simulations of soft matter systems

Grand-canonical molecular dynamics simulations powered by a hybrid 4D nonequilibrium MD/MC method: Implementation in LAMMPS and applications to electrolyte solutions

A quantum mechanical computational method for modeling electrostatic and solvation effects of protein

Electronic coarse graining enhances the predictive power of molecular simulation allowing challenges in water physics to be addressed

A fast, convenient, polarizable electrostatic model for molecular dynamics

Development of an Effective Polarizable Bond Method for Biomolecular Simulation.

Seamless integration of GEM, a density based-force field, for QM/MM simulations via LICHEM, Psi4 and Tinker-HP

A New Coarse-Grained Model for Water: the Importance of Electrostatic Interactions

GPU-Accelerated Molecular Dynamics Simulation to Study Liquid Crystal Phase Transition Using Coarse-Grained Gay-Berne Anisotropic Potential

Molecular dynamics in a grand ensemble: Bergmann–Lebowitz model and adaptive resolution simulation

Accurate interatomic force field for molecular dynamics simulation by hybridizing classical and machine learning potentials

Towards Exact Molecular Dynamics Simulations with Machine-Learned Force Fields

sGDML: Constructing Accurate and Data Efficient Molecular Force Fields Using Machine Learning

Developing and Benchmarking Sulfate and Sulfamate Force Field Parameters via Ab Initio Molecular Dynamics Simulations To Accurately Model Glycosaminoglycan Electrostatic Interactions

Data-driven coarse-grained modeling of polymers in solution with structural and dynamic properties conserved

Incorporating surface polarization effects into large-scale coarse-grained Molecular Dynamics simulation

Implicit-solvent coarse-grained modeling for polymer solutions via Mori-Zwanzig formalism

Completely Multipolar Model as a General Framework for Many-Body Interactions as Illustrated for Water

Simulate Time-integrated Coarse-grained Molecular Dynamics with Geometric Machine Learning

Systematic Improvement of Empirical Energy Functions in the Era of Machine Learning