Publications : 2024

Lin H, Sakolish C, Moyer HL, Ferguson SS, Stanko JP, Carmichael PL, Hewitt P, Hoffmann S, et al. Predicting renal clearance of PFAS with a human kidney proximal tubule tissue chip and a novel physiologically-based kidney model. Abstract 3508, Society of Toxicology 63rd Annual Meeting, Salt Lake City, UT, March 2024.

Abstract

OPEN ACCESS

Background and Purpose: Per- and polyfluoroalkyl substances (PFASs) have become an integral part of various industries due to their unique chemical properties. However, the pervasive use of PFAS in modern society has led to a concerning ubiquity of these substances in the environment and in population biomonitoring data across the world. Some PFAS exhibit extremely long half-lives on the order of years in humans, believed to be due in part to high reabsorption by renal proximal tubules, but little is known about the renal clearance (CLrenal) for the vast majority of these substances. Microphysiological systems (MPS) have the potential to overcome some of the limitations of traditional 2D cultures due to their ability to recapitulate tissue-like microenvironments (e.g., dynamic flow). This study aims to enhance PFAS CLrenal predictions by combining in vitro testing with models of varying complexity (96-well plate, and static and fluidic Transwells®) with physiologically-based proximal tubular modelling. Methods: In this study, RPTEC/ TERT1-OAT1 cells were used to establish traditional 2D cell culture (96-well plate), static and fluidic human proximal tubule systems by using Transwells® without or with the dynamic flow (2uL/s, CNBio PhysioMimix™), to predict CLrenal of three PFAS: one with a long half-life (PFOS) and two with shorter half-lives (PFBS and PFHxA). The exposure included single (5 uM) or mixture (2 uM for each PFAS) settings. We conducted the bidirectional transporter studies in Transwells® under static or dynamic flow, and showed that these systems had good barrier formation as measured by transepithelial electrical resistance. We then measured both the flux of each PFAS across the barrier at 2, 4, 24, and 48h in cell culture media, as well as the uptake into cells by measuring cell lysates at 48 hr (LCMS/MS testing). Bayesian methods were used to fit the toxicokinetic data from the cell testing with the compartmental model and to obtain the in vitro permeability regarding reabsorption and secretion. As a simple comparator, we also measured PFAS in cell fractions of RPTECs in the 96-well plate. Then, we used these permeability estimates in a physiologically-based kidney model to extrapolate the overall in vivo CLrenal. Results: Predictions for human renal clearances of PFAS were highly correlated with available values from in vivo human studies, regardless of which in vitro system was used, indicating a strong ability to distinguish between low- and high-clearance PFAS. We also found similar predictions using both static and fluidic systems, as well as between single and mixture exposures. Predictions of CLrenal from Transwell® data overlapped with in vivo values reported in the literature, but their relative difference between low- and high-clearance PFAS was smaller in vitro than in vivo. Using data from 2D culture, the relative values of CLrenal were similar to the corresponding in vivo values, but these predictions were systematically lower by an order of magnitude. Conclusions: Our in vitro-in silico framework can strongly discriminate between low and high clearance PFAS, with 2D cell culture data having better performance for predictive relative CLrenal, and Transwell® data having better performance for predictive absolute CLrenal. Overall, we conclude these methodologies are useful for prioritizing PFAS with longer half-lives and thereby greater potential for human health concerns.