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 Ca
2+
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 Ca
2+
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.