Abstract
The phospholipid membrane-water partition coefficients (K(MW)) and equilibrium binding affinities for human serum albumin (HSA) of 60 structurally diverse perfluoroalkyl and polyfluoroalkyl substances (PFAS) were evaluated through laboratory measurements and modeling to enhance our understanding of PFAS distribution in organisms. Per- and polyfluoroalkyl carboxylic acids exhibited a 0.36 ± 0.01 log-unit increase in K(MW) as the fluorinated carbon chain length increased from C(4) to C(16), while per- and polyfluoroalkyl sulfonates showed a 0.37 ± 0.02 log-unit increase. The highest HSA affinity range was observed between C(6) and C(10), with the following structural subclass order: per- and polyfluoroalkyl sulfonates ≈ ether sulfonic acids > polyfluoroalkyl carboxylic acids > fluorotelomer unsaturated carboxylic acids > phosphate diesters ≈ per- and polyfluoroether carboxylic acids. A comparison between association rate constants (K(A)) and HSA-PFAS molecular docking predictions with AutoDock Vina indicated that modeling could effectively predict the affinity of PFAS to HSA, especially for PFAS carbon chain lengths from C(4) to C(10). Based on in vitro results, exposure-dependent PFAS partitioning in organisms was modeled by comparing distribution coefficients between PFAS in phospholipid membranes and HSA at different PFAS concentrations and demonstrated that at lower concentrations, PFAS had higher partitioning in HSA, while with increasing concentration, the proportion of binding relative to the aqueous phase shifted toward the phospholipid membrane. Few studies have compared the bioaccumulation of PFAS in phospholipid membranes and HSA. This research reports that protein-water distribution coefficients are higher than membrane-water partitioning coefficients at lower PFAS concentrations, which may have implications for interpreting exposure data and toxicity experiments.