A New CSRML Structure-Based Fingerprint Method for Profiling and Categorizing Per- and Polyfluoroalkyl Substances (PFAS)

Identification of new PFAS for severe interference with thyroid hormone transport: A combined in vitro/silico approach

A tiered in vitro/in silico approach was developed to screen 12,654 per- and polyfluoroalkyl substances (PFAS) for their potential to disrupt the thyroid hormone transport. Initially, a set of 45 PFAS was tested using TTR-TRβ-CALUX bioassay, which was subsequently employed to develop a classification model, distinguishing active and inactive PFAS. The model fulfills all good practices for QSAR model validation and can predict whether a given PFAS can disrupt plasma transport of the thyroid hormone (T4). Subsequently, active compounds were used to develop two regression approaches: (i) multiple linear regression MLR, and (ii) second approach aimed at identifying multiple valid QSAR models based on different data-splitting strategies. Finally, a comprehensive virtual screening of a large PFAS dataset was conducted to assess their potency in disrupting thyroid hormone transport. The predictions indicated that more than 7500 compounds were active with over 100 PFAS potentially causing even greater adverse effects than PFOA. These findings highlight the critical role of integrating New Approach Methodologies (NAM)-based in vitro toxicity testing with multifaceted molecular modeling in assessing the risks associated with PFAS contamination in environmental matrices.

Skin sampling as a proxy for screening per- and polyfluoroalkyl substances (PFAS) exposures in endangered sea turtles

Per- and polyfluoroalkyl substances (PFAS) are widespread in the environment, including the ocean. However, information on PFAS exposures in protected and endangered sea turtle species is scarce because sampling liver or blood is invasive and challenging. We assessed, for the first time, whether skin tissue, which is less invasive to collect, can be used to screen for PFAS in sea turtles. We measured concentrations of 20 PFAS in the skin and serum from deceased juvenile green turtles. Perfluorooctane sulfonic acid (PFOS) was the predominant PFAS and was detected in most of the skin (85 %) and serum (100 %). For individuals with paired skin and serum, PFOS concentrations were not significantly different, and thus suggest that skin samples have promising use for screening PFAS in sea turtles. This method may be adapted and optimized in the future to increase data coverage and improve our understanding of sea turtle exposures to PFAS.

Selective Quantification of Charged and Neutral Polyfluoroalkyl Substances Using the Total Oxidizable Precursor (TOP) Assay

Perfluoroalkyl acid (PFAA) precursors are a diverse subclass of per- and polyfluoroalkyl substances (PFASs) that can be transformed into PFAAs of public health concern. Unlike strongly acidic PFAAs, precursors can be anionic, cationic, neutral, or zwitterionic. Precursor charge affects the environmental fate, but existing quantification techniques struggle to ascertain the abundance of compounds within each charge group. To fill this gap, we developed and validated a solid-phase extraction procedure that separates precursors by charge and quantifies the sum of the precursors in each fraction with the total oxidizable precursor (TOP) assay. Method performance was demonstrated by spiking known concentrations of ten precursors into aqueous film-forming foam (AFFF)-impacted groundwater, municipal wastewater, and soil samples. Precursor fractionation and recovery were greater in groundwater and soil samples than in wastewater. Use of the method provided results that were consistent with expectations based on precursor transport properties. In surficial soils near an AFFF source zone, anionic precursors with five or fewer perfluorinated carbons accounted for about 95% of PFASs, but less than half of PFASs in the underlying groundwater. In municipal wastewater influent, the sum of precursors exceeded the sum of PFAAs and was approximately equally distributed among all charge fractions.

Specific and potent inhibition of steroid hormone pre-receptor regulator AKR1C2 by perfluorooctanoic acid: Implications for androgen metabolism

Per- and polyfluoroalkyl substances (PFAS) are ubiquitous environmental pollutants that are highly stable synthetic organofluorine compounds. One congener perfluorooctanoic acid (PFOA) can be detected in nearly all humans and is recognized as an endocrine disrupting chemical (EDC). EDCs disrupt hormone synthesis and metabolism and receptor function. One mechanism of steroid hormone action is the pre-receptor regulation of ligand access to steroid hormone receptors by aldo-keto reductases. Here we report PFOA inhibition of AKR family 1 member C2 (AKR1C2), leading to dysregulation of androgen action. Spectrofluorimetric inhibitor screens identified PFOA as a competitive and tight binding inhibitor of AKR1C2, whose role is to inactivate 5α-dihydrotestosterone (5α-DHT). Further site directed mutagenesis studies along with molecular docking simulations revealed the importance of residue Valine 54 in mediating AKR1C2 inhibitor specificity. Binding site restrictions were explored by testing inhibition of other related PFAS chemicals, confirming that steric hinderance is a key factor. Furthermore, radiochromatography using HPLC and in line radiometric detection confirmed the accumulation of 5α-DHT as a result of PFOA inhibition of AKR1C2. We showed that PFOA could enhance the transactivation of AR in reporter genes assays in which 5α-DHT metabolism was blocked by AKR1C2 inhibition in HeLa cells. Taken together, these data suggest PFOA has a role in disrupting androgen action through inhibiting AKR1C2. Our work identifies an EDC function for PFOA not previously revealed.

Improved prediction of PFAS partitioning with PPLFERs and QSPRs

Per- and polyfluoroalkyl substances (PFAS) are chemicals of high concern and are undergoing hazard and risk assessment worldwide. Reliable physicochemical property (PCP) data are fundamental to assessments. However, experimental PCP data for PFAS are limited and property prediction tools such as quantitative structure-property relationships (QSPRs) therefore have poor predictive power for PFAS. New experimental data from Endo 2023 are used to improve QSPRs for predicting poly-parameter linear free energy relationship (PPLFER) descriptors for calculating water solubility (SW), vapor pressure (VP) and the octanol-water (KOW), octanol-air (KOA) and air-water (KAW) partition ratios. The new experimental data are only for neutral PFAS, and the QSPRs are only applicable to neutral chemicals. A key PPLFER descriptor for PFAS is the molar volume and this work compares different versions and makes recommendations for obtaining the best PCP predictions. The new models are included in the freely available IFSQSAR package (version 1.1.1), and property predictions are compared to those from the previous IFSQSAR (version 1.1.0) and from QSPRs in the US EPA's EPI Suite (version 4.11) and OPERA (version 2.9) models. The results from the new IFSQSAR models show improvements for predicting PFAS PCPs. The root mean squared error (RMSE) for predicting log KOWversus expected values from quantum chemical calculations was reduced by approximately 1 log unit whereas the RMSE for predicting log KAW and log KOA was reduced by 0.2 log units. IFSQSAR v.1.1.1 has an RMSE one or more log units lower than predictions from OPERA and EPI Suite when compared to expected values of log KOW, log KAW and log KOA for PFAS, except for EPI Suite predictions for log KOW which have a comparable RMSE. Recommendations for future experimental work for PPLFER descriptors for PFAS and future research to improve PCP predictions for PFAS are presented.

Prediction of PFAS bioaccumulation in different plant tissues with machine learning models based on molecular fingerprints

Due to the wastewater irrigation or biosolid application, per- and polyfluoroalkyl substances (PFASs) have been widely detected in agriculture soil and hence crops or vegetables. Consumption of contaminated crops and vegetables is considered as an important route of human exposure to PFASs. Machine learning (ML) models have been developed to predict PFAS uptake by plants with majority focus on roots. However, ML models for predicting accumulation of PFASs in above ground edible tissues have yet to be investigated. In this study, 811 data points covering 22 PFASs represented by molecular fingerprints and 5 plant categories (namely the root class, leaf class, cereals, legumes, and fruits) were used for model development. The Extreme Gradient Boosting (XGB) model demonstrated the most favorable performance to predict the bioaccumulation factors (BAFs) in all the 4 plant tissues (namely root, leaf, stem, and fruit) achieving coefficients of determination R2 as 0.82-0.93. Feature importance analysis showed that the top influential factors for BAFs varied among different plant tissues, indicating that model developed for root concentration prediction may not be feasible for above ground parts. The XGB model's performance was further demonstrated by comparing with data from pot experiments measuring BAFs of 12 PFASs in lettuce. The correlation between predicted and measured results was favorable for BAFs in both lettuce roots and leaves with R2 values of 0.76 and 0.81. This study developed a robust approach to comprehensively understand the uptake of PFASs in both plant roots and above ground parts, offering key insights into PFAS risk assessment and food safety.

In Vitro Hepatic Clearance Evaluations of Per- and Polyfluoroalkyl Substances (PFAS) across Multiple Structural Categories

Toxicokinetic (TK) assays and in vitro-in vivo extrapolation (IVIVE) models are New Approach Methods (NAMs) used to translate in vitro points of departure to exposure estimates required to reach equivalent blood concentrations. Per- and polyfluoroalkyl substances (PFAS) are a large chemical class with wide-ranging industrial applications for which only limited toxicity data are available for human health evaluation. To address the lack of TK data, a pooled primary human hepatocyte suspension model was used with targeted liquid chromatography-mass spectrometry to investigate substrate depletion for 54 PFAS. A median value of 4.52 μL/(min x million cells) was observed across those that showed significant clearance, with 35 displaying no substrate depletion. Bayesian modeling propagated uncertainty around clearance values for use in IVIVE models. Structural evaluations showed the fluorotelomer carboxylic acids were the only PFAS carboxylates showing appreciable clearance, and per- and polyfluorosulfonamides were more readily metabolized than other PFAS sulfonates. Biotransformation product prediction, using the chemical transformation simulator, suggested hydrolysis of PFAS sulfonamides to more stable sulfonic acids, which is an important consideration for exposure modeling. This effort greatly expands the PFAS in vitro toxicokinetic dataset, enabling refined TK modeling, in silico tool development, and NAM-based human health evaluations across this important set of emerging contaminants.

Species Difference? Bovine, Trout, and Human Plasma Protein Binding of Per- and Polyfluoroalkyl Substances

Per- and polyfluoroalkyl substances (PFAS) strongly bind to proteins and lipids in blood, which govern their accumulation and distribution in organisms. Understanding the plasma binding mechanism and species differences will facilitate the quantitative in vitro-to-in vivo extrapolation and improve risk assessment of PFAS. We studied the binding mechanism of 16 PFAS to bovine serum albumin (BSA), trout, and human plasma using solid-phase microextraction. Binding of anionic PFAS to BSA and human plasma was found to be highly concentration-dependent, while trout plasma binding was linear for the majority of the tested PFAS. At a molar ratio of PFAS to protein ν < 0.1 molPFAS/molprotein, the specific protein binding of anionic PFAS dominated their human plasma binding. This would be the scenario for physiological conditions (ν < 0.01), whereas in in vitro assays, PFAS are often dosed in excess (ν > 1) and nonspecific binding becomes dominant. BSA was shown to serve as a good surrogate for human plasma. As trout plasma contains more lipids, the nonspecific binding to lipids affected the affinities of PFAS for trout plasma. Mass balance models that are parameterized with the protein-water and lipid-water partitioning constants (chemical characteristics), as well as the protein and lipid contents of the plasma (species characteristics), were successfully used to predict the binding to human and trout plasma.

An artificial intelligence platform for automated PFAS subgroup classification: A discovery tool for PFAS screening

Since structural analyses and toxicity assessments have not been able to keep up with the discovery of unknown per- and polyfluoroalkyl substances (PFAS), there is an urgent need for effective categorization and grouping of PFAS. In this study, we presented PFAS-Atlas, an artificial intelligence-based platform containing a rule-based automatic classification system and a machine learning-based grouping model. Compared with previously developed classification software, the platform's classification system follows the latest Organization for Economic Co-operation and Development (OECD) definition of PFAS and reduces the number of uncategorized PFAS. In addition, the platform incorporates deep unsupervised learning models to visualize the chemical space of PFAS by clustering similar structures and linking related classes. Through real-world use cases, we demonstrate that PFAS-Atlas can rapidly screen for relationships between chemical structure and persistence, bioaccumulation, or toxicity data for PFAS. The platform can also guide the planning of the PFAS testing strategy by showing which PFAS classes urgently require further attention. Ultimately, the release of PFAS-Atlas will benefit both the PFAS research and regulation communities.

Baseline Toxicity Model to Identify the Specific and Nonspecific Effects of Per- and Polyfluoroalkyl Substances in Cell-Based Bioassays

High-throughput screening is a strategy to identify potential adverse outcome pathways (AOP) for thousands of per- and polyfluoroalkyl substances (PFAS) if the specific effects can be distinguished from nonspecific effects. We hypothesize that baseline toxicity may serve as a reference to determine the specificity of the cell responses. Baseline toxicity is the minimum (cyto)toxicity caused by the accumulation of chemicals in cell membranes, which disturbs their structure and function. A mass balance model linking the critical membrane concentration for baseline toxicity to nominal (i.e., dosed) concentrations of PFAS in cell-based bioassays yielded separate baseline toxicity prediction models for anionic and neutral PFAS, which were based on liposome-water distribution ratios as the sole model descriptors. The specificity of cell responses to 30 PFAS on six target effects (activation of peroxisome proliferator-activated receptor (PPAR) gamma, aryl hydrocarbon receptor, oxidative stress response, and neurotoxicity in own experiments, and literature data for activation of several PPARs and the estrogen receptor) were assessed by comparing effective concentrations to predicted baseline toxic concentrations. HFPO-DA, HFPO-DA-AS, and PFMOAA showed high specificity on PPARs, which provides information on key events in AOPs relevant to PFAS. However, PFAS were of low specificity in the other experimentally evaluated assays and others from the literature. Even if PFAS are not highly specific for certain defined targets but disturb many toxicity pathways with low potency, such effects are toxicologically relevant, especially for hydrophobic PFAS and because PFAS are highly persistent and cause chronic effects. This implicates a heightened need for the risk assessment of PFAS mixtures because nonspecific effects behave concentration-additive in mixtures.

Building Chemical Intuition about Physicochemical Properties of C8-Per-/Polyfluoroalkyl Carboxylic Acids through Computational Means

We have predicted acid dissociation constants (pK a), octanol-water partition coefficients (K OW), and DMPC lipid membrane-water partition coefficients (K lipid-w) of 150 different eight-carbon-containing poly-/perfluoroalkyl carboxylic acids (C8-PFCAs) utilizing the COnductor-like Screening MOdel for Realistic Solvents (COSMO-RS) theory. Different trends associated with functionalization, degree of fluorination, degree of saturation, degree of chlorination, and branching are discussed on the basis of the predicted values for the partition coefficients. In general, functionalization closest to the carboxylic headgroup had the greatest impact on the value of the predicted physicochemical properties.

High Throughput Read-Across for Screening a Large Inventory of Related Structures by Balancing Artificial Intelligence/Machine Learning and Human Knowledge

Read-across is an in silico method applied in chemical risk assessment for data-poor chemicals. The read-across outcomes for repeated-dose toxicity end points include the no-observed-adverse-effect level (NOAEL) and estimated uncertainty for a particular category of effects. We have previously developed a new paradigm for estimating NOAELs based on chemoinformatics analysis and experimental study qualities from selected analogues, not relying on quantitative structure-activity relationships (QSARs) or rule-based SAR systems, which are not well-suited to end points for which the underpinning data are weakly grounded in specific chemical-biological interactions. The central hypothesis of this approach is that similar compounds have similar toxicity profiles and, hence, similar NOAEL values. Analogue quality (AQ) quantifies the suitability of an analogue candidate for reading across to the target by considering similarity from structure, physicochemical, ADME (absorption, distribution, metabolism, excretion), and biological perspectives. Biological similarity is based on experimental data; assay vectors derived from aggregations of ToxCast/Tox21 data are used to derive machine learning (ML) hybrid rules that serve as biological fingerprints to capture target-analogue similarity relevant to specific effects of interest, for example, hormone receptors (ER/AR/THR). Once one or more analogues have been qualified for read-across, a decision theory approach is used to estimate confidence bounds for the NOAEL of the target. The confidence interval is dramatically narrowed when analogues are constrained to biologically related profiles. Although this read-across process works well for a single target with several analogues, it can become unmanageable when, for example, screening multiple targets (e.g., virtual screening library) or handling a parent compound having numerous metabolites. To this end, we have established a digitalized framework to enable the assessment of a large number of substances, while still allowing for human decisions for filtering and prioritization. This workflow was developed and validated through a use case of a large set of bisphenols and their metabolites.