Molecular Dynamics-Informed Prediction of Per- and Polyfluoroalkyl Substances (PFAS) Partitioning into Phospholipid Bilayer Environments

Abstract The incredible diversity of per- and polyfluoroalkyl substances (PFAS) creates a difficult challenge for predicting the steady-state partition ratios indicative of cellular health and function. Many PFAS are amphiphilic surfactants, often with ionic or polar headgroups driving their aqueous solubility and nonpolar tails driving the molecules to interfacial environments. Molecular dynamics (MD) simulations generate potential of mean force (PMF) curves that integrate the energy of molecular interactions along a defined pathway, offering a means to understand how PFAS interact with and pass through structured phospholipid bilayer environments. However, it is unclear whether MD results calculated over such short time scales (∼ns) can be extrapolated to inform the steady-state conditions relevant to metrics of phase partitioning. Here we propose that PFAS water-bilayer partition ratios can be extrapolated from MD-derived PMF parameters derived from DOPC and POPC phospholipid bilayers via models of damped Brownian motion. Predictions of this approach are consistent with observations that PFAS become rapidly absorbed into bilayers yet are slow to emerge, resulting in large partition ratios that we validate against existing measurements. This model predicts that although water-bilayer partition ratios depend exponentially on the number of fluorinated carbons in the alkyl tail, the exponent differs based on the headgroup between either sulfonate or carboxylate PFAS. We find that MD-derived energy profiles accurately inform tail group contributions to water-bilayer partition ratios, whereas PFAS headgroup chemistry regulates the extent of intrabilayer binding, which, taken together, creates a rich and complex bilayer partitioning profile.