Reassessing DMSO-lipid interactions: Improved AMBER force fields emphasize solvent rather than bilayer effects in...

The interaction of dimethyl sulfoxide (DMSO) with lipid membranes has been extensively studied using molecular dynamics (MD) simulations, yet discrepancies with experimental findings persist. Here, we re-evaluate the effects of low DMSO concentrations (1.5-10 vol%) on dimyristoyl phosphatidylcholine (DMPC) membranes using updated AMBER force fields (LIPID17, OPC, GAFF2) to assess its cryoprotective role. Simulations were performed in both fluid (330K) and gel (260 K) phases as well as under ice-forming conditions. Across three independent replicas, no statistically significant effects of DMSO were detected on membrane thickness, area per lipid, hydration, or acyl-chain order, indicating that low levels of DMSO do not alter bilayer structure. This represents an improvement over earlier force-field descriptions, which often exaggerated DMSO-lipid interactions, and provides results more consistent with experimental evidence. DMSO showed mild enrichment at the hydrophobic-hydrophilic interface, particularly near carbonyl and glycerol groups, but most molecules remained in the solvent. The strongest effects therefore emerged in the solvent phase: DMSO slowed ice crystal growth by about a factor of five, was excluded from the ice lattice, and accumulated at the ice-membrane boundary forming an ice-free layer. Surprisingly, even without DMSO, ice formation in contact with the bilayer did not cause structural disruption, suggesting that cryoinjury involves additional membrane components beyond lipids. DMSO also strongly inhibited the temperature-driven variation of water density in the amorphous state. These findings suggest that at low concentrations, DMSO's cryoprotective action arises mainly from modulation of water and ice behavior rather than direct bilayer perturbation.