Integrative Computational Approaches to Inform Relative Bioaccumulation Potential of Per- and Polyfluoroalkyl Substances Across Species

Abstract

Abstract Predictive toxicology is increasingly reliant on innovative computational methods to address pressing questions in chemicals assessment. Of importance is the evaluation of contaminant impact differences across species to inform ecosystem protection and identify appropriate model species for human toxicity studies. Here we evaluated 2 complementary tools to predict cross-species differences in binding affinity between per- and polyfluoroalkyl substances (PFAS) and the liver fatty acid-binding protein (LFABP): the Sequence Alignment to Predict Across Species Susceptibility (SeqAPASS) tool and molecular dynamics (MD). SeqAPASS determined that the structure of human LFABP, a key determinant of PFAS bioaccumulation, was conserved in the majority of vertebrate species, indicating these species would have similar PFAS bioaccumulation potentials. Level 3 SeqAPASS evaluation identified several potentially destabilizing amino acid differences across species, which were generally supported by DUET stability change predictions. Nine single-residue mutations and 7 whole species sequences were selected for MD evaluation. One mutation (F50V for PFNA) showed a statistically significant difference with stronger affinity than wild-type human LFABP. Predicted binding affinities for 9 different PFAS across 7 species showed human, rat, chicken, and rainbow trout had similar binding affinities to one another for each PFAS, whereas Japanese medaka and fathead minnow had significantly weaker LFABP-binding affinity for some PFAS. Based on these analyses, the combined use of SeqAPASS and MD provides rapid screening for potential species differences with deeper structural insight. This approach can be easily extended to other important biological receptors and potential ligands.