Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Feb 23;58(13):5727–5738. doi: 10.1021/acs.est.3c09950

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

Weiping Qin †,‡, Luise Henneberger †, Juliane Glüge †,§, Maria König †, Beate I Escher †,‡,*
PMCID: PMC10993398  PMID: 38394616

Abstract

graphic file with name es3c09950_0006.jpg

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.

Keywords: PFAS, membrane concentration, liposome-water distribution ratio, potential adverse outcome pathways, peroxisome proliferator-activated receptors

Short abstract

A baseline toxicity prediction model for cell-based bioassays demonstrates that PFAS act as baseline toxicants in many in vitro bioassays.

Introduction

The threat that per- and polyfluoroalkyl substances (PFAS) pose to human health has been a great concern for society. The concern has expanded from perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) to thousands of PFAS with diverse structures and unclear toxicological effects. The OECD broadened the definition of PFAS to at least one perfluorinated carbon,1 which implies that there are now more than 14,000 PFAS chemicals in the CompTox Chemistry Dashboard.2 Traditional methods can no longer cope with the risk assessment of such large numbers of PFAS. New approach methodologies (NAM), especially those based on high-throughput screening (HTS) with cellular assays, provide a strategy of extensive screening for molecular initiating events and key events in adverse outcome pathways (AOP) because cell responses can serve as early warning signals. PFAS may trigger cell responses through several cellular toxicity pathways, including reactive (oxidative stress),3 specific (receptor-mediated signaling pathways),4 and nonspecific toxicity.

Baseline toxicity, also known as narcosis, is a common, nonspecific effect caused by an accumulation of chemicals in cell membranes that interfere with membrane function and destroy membrane integrity.5 Chemicals that act as baseline toxicants can trigger specific effects at critical membrane concentrations. Baseline toxicity does not equate to low toxicity; very hydrophobic chemicals can still be very potent through baseline toxicity, i.e., act in low concentrations. Hence, baseline toxicity may serve as an anchor or reference state to determine the degree of specificity of PFAS in cell-based bioassays because specific effects occur at even lower concentrations.

Baseline toxicity occurs when critical membrane burdens are exceeded. Escher et al.6 derived the critical membrane concentration of baseline toxicants that causes 10% of cytotoxicity IC10,membrane,baseline as 69 mmol/Llip from eight different cell lines. IC10,membrane,baseline was independent of the cell line and constant for all organic chemicals. As in vitro bioassay responses are reported as nominal (i.e., dosed) concentrations in bioassay medium, one needs to link IC10,membrane,baseline and freely dissolved concentrations (IC10,free,baseline) to nominal concentrations (IC10,nom,baseline), which can be accomplished by mass balance models (MBM).7 While IC10,membrane,baseline is constant, IC10,free,baseline is dependent on the partitioning of chemicals into membrane lipid bilayers, which can be simulated by phospholipid vesicles, so-called liposomes. The IC10,free,baseline can vary over many orders of magnitude. Hydrophobic chemicals have much lower IC10,free,baseline values than hydrophilic and charged chemicals. IC10,nom,baseline is determined by the experimental conditions and binding affinities of PFAS to proteins and lipids in the medium and cells. An overview of these metrics can be found in Figure 1.

Figure 1.

Figure 1

Open in a new tab

An overview of the dose-metrics and baseline toxicity prediction models for cell-based bioassays. Abbreviation: Vmedium orVcell = volume of medium or cells, Cnom = nominal concentration, Cfree = free concentration, SPME = solid-phase microextraction, MBM = mass balance model, IC = inhibitory concentration, EC = effect concentration, Dlip/w = distribution ratio between liposomes and water, TR = toxic ratio, and SR = specificity ratio.

The inhibitory concentration at 10% cytotoxicity (IC10,nom) can be derived from the concentration–cytotoxicity curves using nominal concentrations in the bioassay. The ratio between IC10,nom,baseline and experimental IC10,nom is called the toxic ratio (TRnom).8,9 At lower, noncytotoxic concentrations, many in vitro bioassays trigger specific effects, such as the activation of receptors or adaptive stress responses. The effect concentration at 10% effect (EC10,nom) can be compared to the cytotoxicity IC10,nom,baseline to derive the specificity ratio (SRnom).9 If TRnom or SRnom are close to 1, cell death or response may be caused by baseline toxicity. A positive artifact due to the so-called cytotoxicity burst may appear when cells are close to death. High TRnom or SRnom values suggest that chemicals specifically trigger cytotoxicity or defined targets at lower concentrations than baseline toxicity.

The specific effects refer to the signals from defined targets, which can be distinguished from the nonspecific baseline toxicity. Here are several cases among the numerous toxicological studies. Oxidative stress is a common mechanism of PFAS-induced cytotoxicity.3 For example, acute exposure to PFOS induced excessive production of reactive oxidative species, resulting in the apoptosis of mouse islet β-TC-6 cells10 and human neuroblastoma SH-SY5Y cells.11 As a cellular defense, PFOS was found to trigger the nuclear factor erythroid 2-related factor 2 (Nrf2)-antioxidant response element (ARE) pathway, which plays a critical role in cellular protection against toxicity and oxidative stress from chemical stressors.12 Besides, PFAS may disrupt endocrine homeostasis by interacting with receptor-mediated signaling pathways.4 For example, PFOS promoted adipogenesis via peroxisome proliferator-activated receptors (PPARs) in mouse preadipocyte 3T3-L1 cells;13 PFDA and PFDoA might disrupt the function of the thyroid hormone system via aryl hydrocarbon receptor (AhR) in rat pituitary GH3 cell;14 PFOS stimulated insulin secretion via membrane G-protein coupled receptor (GPR) 40 in mouse islet β-TC-6 cells.15 Neurotoxicity is also of concern; for example, PFOS enhanced nerve growth factor-induced neurite outgrowth in rat pheochromocytoma PC12 cells.16 To relate the observed effects to baseline toxicity is a way to allow comparison between different test systems and end points and to identify effects that are of particular concern due to high TRnom or SRnom.

Tox21 and ToxCast are programs of multiple federal agencies, including the U.S. Environmental Protection Agency (EPA) and the National Toxicology Program,17 aiming at efficiently identifying potential AOPs for specified chemicals at molecular, cellular, and organ levels through HTS to facilitate risk assessments of chemicals. Currently, in vitro effect data for 160 PFAS from the EPA have been published.18−21 Transactivation assays encompassing 81 diverse transcription factors were screened with 142 PFAS to describe the activation of nuclear receptor-mediated signaling pathways.18 148 biomarkers relevant to the immune system were measured with 147 PFAS to inform mechanisms of immunotoxicity.19 A NAM battery for developmental neurotoxicity was developed to evaluate the effects of 160 PFAS on neural network formation and function.20 Radioactive iodide uptake high-throughput assay was used to test 149 PFAS for potential thyroid disruption.21 However, data evaluation and ranking scores were done independently with diverse conclusions, which makes it difficult to determine which PFAS are of the highest concern and which in vitro end points are relevant for human health.

In the present study, we systematically assessed the role of PFAS distribution in bioassay systems and how this affected the different dose-metrics in bioassays (Figure 1). The free concentrations of 10 anionic and one partially charged PFAS in the bioassays run in a 96-well plate format were measured by solid-phase microextraction (SPME) to derive IC10,free from a free concentration–cytotoxicity curve. Experimental IC10,free values were compared with IC10,free predicted by the MBM. A validated MBM was used to link the known critical membrane concentrations for baseline toxicity IC10,membrane,baseline to IC10,nom,baseline.

Four cell-based HTS were selected as experimental batteries, including three reporter gene assays targeting oxidative stress (AREc32 assay), nuclear receptors of PPARγ and AhR, as well as an image-based neurotoxicity assay that quantifies inhibition of neurite outgrowth of differentiated SH-SYSY cells, because these targets may be relevant to PFAS according to previous studies.11−14,16 24 PFAS were then tested in a 384-well plate format. The cytotoxicity IC10,nom and the effect concentrations EC10,nom from these assays were compared with the IC10,nom,baseline to derive TRnom and SRnom, which were used to determine if the effects of PFAS are specific. We also demonstrated the applicability of the approach to other assays by evaluating the bioassay responses of 16 PFAS from the literature for their degree of specificity.

Theory

Free Concentrations of PFAS in Bioassays

The nominal concentration (Cnom) in a bioassay is the total amount of chemicals (ntot) dosed divided by the total volume of the bioassay system (Vtot), which is composed of medium (Vmedium) and cells (Vcell) .

graphic file with name es3c09950_m001.jpg 1

Fetal bovine serum (FBS) in medium serves as a reservoir of reversibly bound chemical and also facilitates a faster equilibrium of chemical between medium and cells,22 and thus, the free concentration of PFAS in medium (Cfree,medium) is usually lower than Cnom. An MBM that accounts for the binding of chemicals to proteins and lipids in the medium and cells (eq 2) has been developed and validated experimentally for chemicals.23

graphic file with name es3c09950_m002.jpg 2

Cfree,medium is related to the Cnom by inserting eq 1 in 2 yielding eq 3 with distribution ratios between the medium and water (Dmedium/w) and between the cells and water (Dcell/w), as well as volumes of protein and lipid (Vprotein+lipid) in cells and medium. The detailed derivation of eq 3 is in Supporting Information Text S1.

graphic file with name es3c09950_m003.jpg 3

Dmedium/w and Dcell/w can be measured experimentally or predicted by MBMs with the assumption that proteins and lipids are the main sorption phases in the medium and cells. Bovine serum albumin (BSA) serves as a surrogate for protein binding in medium (DBSA/w) and liposomes for partitioning to membrane lipids (Dlip/w). Then, Dmedium/w can be predicted by eq 4.

graphic file with name es3c09950_m004.jpg 4

The most abundant proteins in cells are structural proteins (SP), for which muscle proteins are better surrogates than BSA.24 Analogously, Dcell/w can be predicted by eq 5 with DSP/w referring to the distribution ratio between SP and water.

graphic file with name es3c09950_m005.jpg 5

Note that in the previous baseline toxicity model,7 we had defined the Dmedium/w and Dcell/w by including the water, protein, and lipid volumes in medium and cells, but here only protein and lipid volumes were defined as the sorption phases. Cfree,medium can be calculated from Cnom by inserting eqs 4 and 5 in eq 3.

graphic file with name es3c09950_m006.jpg 6

Membrane and Free Concentrations Related to Baseline Toxicity

Chemicals inevitably accumulate in the membrane during the cellular uptake from the medium into the cytosol. The partitioning to membrane lipid bilayers can be simulated by phospholipid vesicles, so-called liposomes. The distribution ratio (Dlip/w (pH = 7.4)) is the ratio of the concentration of a chemical bound to liposomes (Clip) divided by the free concentration in the water phase (Cw).

graphic file with name es3c09950_m007.jpg 7

Some ionized PFAS may be actively transported into cells via ion channels or transport proteins, but due to their high membrane permeability, passive diffusion has been proven to dominate the cellular uptake of even the anionic PFAS.25 There is no pH gradient under typical bioassay conditions; thus, even charged organic chemicals reach a steady state in cells within hours.22 It is safe to assume that free concentrations in cells (Cfree,cytosol) and Cfree,medium are equal at the steady state.

If Clip is related to a constant cell membrane concentration of baseline toxicants, namely, IC10,membrane,baseline of 69 mmol/Llip,6 the Cw presents the free concentration of baseline toxicants in the aqueous phase of both medium and cells, IC10,free,baseline, which can be calculated by transforming eqs 7 to 8.

graphic file with name es3c09950_m008.jpg 8

Nominal Concentrations Related to Baseline Toxicity

Combining eqs 6 and 8 yields eq 9, which predicts the nominal baseline toxicity.

graphic file with name es3c09950_m009.jpg 9

There is a linear relationship between proteins (log DBSA/w or log DSP/w) and lipids (log Dlip/w) for nonspecific binding, as shown in eq 10.

graphic file with name es3c09950_m010.jpg 10

Inserting eqs 10 to 9 yields an equation that is dependent only on the log Dlip/w and system parameters of the bioassay. The model can be fitted by an empirical exponential equation to mathematically simplify the equation to reduce it to three adjustable parameters (eq 11). Previously, Lee et al.7 developed an empirical baseline toxicity prediction model for neutral and cationic chemicals. As anionic PFAS bind stronger to proteins than comparable neutral chemicals, the model will need to be updated for anionic PFAS, and there will be a separate baseline toxicity QSAR (quantitative structure–activity relationship) for neutral and anionic PFAS.

graphic file with name es3c09950_m011.jpg 11

Toxic Ratio and Specificity Ratio

The ratio of cytotoxic concentrations (IC10) between predicted baseline toxicity and experimental cytotoxicity is the toxic ratio TR (Figure 1),8,9 which can be derived from the free (TRfree) or the nominal effect concentrations (TRnom) according to eq 12.

graphic file with name es3c09950_m012.jpg 12

In theory, TRfree and TRnom are the same, but the empirical values might differ due to different points of departure for their calculations. Chemicals with a TR < 10 are classified as baseline toxicants. TR > 10 suggests that there may be some specific mode of action triggering the toxic effects before baseline toxicity occurs.8,9

Any effect concentration (EC) for reporter gene activation or other defined targets can also be related to baseline toxicity by defining a SR, as shown in eq 13.

graphic file with name es3c09950_m013.jpg 13

F is typically 10%, and EC10 is defined as the effective concentration triggering 10% of the maximum effect. For antagonism, the suppression ratio SPR of 20% is often used, and in the present study, ECSPR20 is the suppression concentration leading to 20% inhibition of the background signal of rosiglitazone on PPARγ.23 SR < 1 suggests that the effects on defined targets may be nonspecific. One ≤ SR < 10 is considered as moderate specificity with uncertainty. SR > 10 is specific.9

Materials and Methods

Materials

Eleven PFAS (perfluorobutanoic acid (PFBA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluoroundecanoic acid (PFUnA), perfluoro-2-methyl-3-oxahexanoic acid (HFPO–DA), perfluorohexahexanesulfonic acid (PFHxS), perfluorooctanesulfonic acid (PFOS), 6:2 fluorotelomer sulfonic acid (6:2 FTSA), and perfluorooctane sulfonamide (PFOSA)) were investigated in detail. Their structures are shown in Figure S1. Additional 13 PFAS (10 per- and three polyfluorinated chemicals) were evaluated only in the cell-based assays (Table S1). All PFAS were dissolved in methanol (1428, Chemsolute) as a stock solution (Table S2).

Free Concentration and Cytotoxicity of PFAS in PPARγ-GeneBLAzer Reporter Gene Assays

A detailed description is in Supporting Information text S2, and an experimental workflow is shown in Figure S2. Briefly, on day 1, cells were seeded in a 96-well plate (655946, Greiner) for 24 h of incubation. On day 2, PFAS stock solutions in methanol were pipetted into dosing vials (2214340, Labsolute), and the methanol was blown down gently with nitrogen. The PFAS precipitate at the bottom of each vial was dissolved again with 1000 μL assay medium and serially diluted in a 96-deep-well plate (7696548, Labsolute) with a dilution factor of 2 to obtain 10 concentration points. Then, 120 μL of PFAS dosing medium was transferred from a 96-deep well plate to a cell plate for 24 h of exposure. The concentration exposed to cells is shown in Table S2. On day 3, the cell plate was imaged with the IncuCyte S3 live cell imaging system (Essen BioScience, USA). The cytotoxicity was determined by comparing the confluency of exposed cells and unexposed cells. Then, 200 μL of the supernatant from each well in the cell plate was transferred to a 96-deep-well plate (P-DW-500-C, Labsolute) for SPME to measure the free concentrations of PFAS in the medium. The experimental conditions for Supelco BioSPME 96-Pin Devices (59683-U, Sigma-Aldrich) are listed in Table S3a, and the calculation details for free concentrations of PFAS are as Henneberger et al.26

PPARγ-GeneBLAzer Medium Binding of PFAS

A serial dilution of PFAS in a 96-deep well plate (7696548, Labsolute) was prepared as above. A second 96-well plate (655946, Greiner) was filled with 100 μL of fresh assay medium. 120 μL of PFAS dosing medium was transferred from the 96-deep well plate to the second 96-well plate to make the concentrations of PFAS the same as those exposed to cells (Table S2). Then, 200 μL was transferred to the third 96-deep-well plate (P-DW-500-C, Labsolute) for SPME with a Supel BioSPME 96-Pin device. The distribution ratios of PFAS between medium components and water (Dmedium/w) were determined as described by Qin et al.23

Cell Binding of PFAS

The detailed description is in Supporting Information text S3, and the experimental workflow for cell binding experiments is shown in Figure S3. HEK293H cells (modified in the PPARγ-GeneBLAzer reporter gene assay) were homogenized by ultrasonic shattering (Sonoplus 2070, Germany). PFAS stock solutions were diluted with PBS. 100 μL of cell homogenate and 100 μL of PFAS solution were added and vortexed in a 1.5 mL HPLC vial with insert (7648146, 765116, Labsolute). Cell homogenates were derived from approximately 1.25 × 106 cells per experiment, and the concentrations of PFAS are listed in Table S2. We also studied the cell binding of PFAS with the other three cell lines (MCF7, H4lle, and SH-SY5Y), which were used in three cell-based HTS (AREc32, AhR-CALUX, and neurotoxicity). The distribution ratios of PFAS between cells and water (Dcell/w) were analyzed as described by Qin et al.23

Structural Protein Binding of PFAS

The detailed description is in Supporting Information text S4, and the experimental workflow is shown in Figure S4. The structural protein binding assay applied ground powder of chicken breast fillet, which was prepared as described previously.24 Chicken protein was suspended with PBS by a high-speed vortex. The pH value of the suspension was adjusted to neutral (pH = 7.4). 500 μL of protein suspension and 500 μL of PFAS solution were added and vortexed in a 1.5 mL HPLC vial (7654554, 7663230, Labsolute). The concentration of protein in each sample was 50 mg/mL, and the concentrations of PFAS are listed in Table S2. The binding affinity of PFAS to cells and structural proteins is considered weak,24 and thus, the C18-SPME fiber (57281-U, Sigma-Aldrich) with a larger volume of C18-particles embedded metal alloy was used. The experimental conditions for the C18-SPME fiber are in Table S3b. The distribution ratios of PFAS between structural protein and water (DSP/w) were analyzed according to Qin et al.23

Protein and Lipid Content of Medium and Cells

Protein concentrations of assay mediums and cell homogenates for four assays were determined by a Pierce BCA Protein Assay Kit (23228, Thermo Scientific). Their lipid concentrations were determined by the sulfo-phospho-vanillin reaction, as described previously.27 Units of protein and lipid were converted from mass concentration (mg/L) to volume concentration (mL/L) using a density of protein of 1.36 kg/L and a density of lipid of 1 kg/L.

High-Throughput Screening of PFAS in 384-Well Plates

An experimental workflow for four cell-based bioassays with different targets (PPARγ-GeneBLAzer, AREc32, AhR-CALUX, and neurotoxicity) is shown in Text S5 and Figure S5. These assays were performed in 384-well plates as described previously.7,28−30 Experimental conditions, including assay medium, cell lines, and cell number, are in Table S4, and PFAS concentrations are in Table S5. Briefly, on day 1, cells were seeded in a 384-well plate by a MultiFlo Dispenser (BioTek,Vermont, USA). On day 2, PFAS stock solutions were prepared with methanol. Defined volumes of PFAS stock were transferred to dosing vials (2214340, Labsolute) and blown down with nitrogen. PFAS were dissolved again in the assay medium. Then, serial concentrations of PFAS in a medium were prepared and dosed to cells by Hamilton Star Robot (Bonaduz, Switzerland). Cell plates were imaged by IncuCyte S3 at the start of exposure. On day 3, cell plates were imaged by IncuCyte S3 again after 24 h of PFAS exposure. The cytotoxicity of three reporter gene cell lines was determined by comparing the confluency of exposed cells and unexposed cells. The cell responses targeting PPARγ, AREc32, and AhR were quantified from the signals of reporter proteins with an Infinite M1000 plate reader (Tecan, USA). The neurite length of differentiated SH-SY5Y cells was quantified by phase-contrast imaging using an IncuCyte S3. Then, Nuclear Green LCS1 (ab138904, Abcam) and propidium iodide (81845, Sigma-Aldrich) were used to stain the total cells and death cells, which were quantified with the IncuCyte S3 to determine the cytotoxicity.

Results and Discussion

Binding to Medium Components, Cells, Liposomes, and Proteins

Dmedium/w and Dcell/w  of ten anionic PFAS and the partially anionic PFOSA were determined exclusively with PPARγ-medium and the HEK293H cell line. The Dmedium/w values of hydrophilic PFBA, PFHxA, and PFHpA and of the hydrophobic PFUnA were independent of concentration (Figure S6) and were used in the MBM (eq 3) to predict the Cfree,medium. In contrast, the Dmedium/w  of seven other PFAS (PFOA, PFNA, HFPO–DA, PFHxS, PFOS, 6:2 FTSA, and PFOSA) were found to be concentration-dependent (Figure S6). PFAS-specific regression equations of log Dmedium/w against log Cw were derived from the Freundlich-type model (Table S6).23

As the protein content in the medium was 64 times higher than the lipid content (Table S7), the concentration-dependent protein binding of anionic PFAS dominated the medium binding.23 IC10,nom or IC10,free (Table 1) were derived from the concentration–cytotoxicity curve at 10% cytotoxicity with a nominal or free concentration of PFAS in the medium (Figure S7). The log Dmedium/w at IC10,nom was calculated from the regression eqs (Table S6) with measured Cfree at IC10,free. log Dcell/w of HEK293H cells were measured at constant concentrations close to the IC10,nom (Table 1). As the protein and lipid are major sorption phases of the medium component and cells, DBSA/w, DSP/w, and Dlip/w were used as representatives of Dmedium/w and Dcell/w for the MBM (eq 6). Log DBSA/w were calculated with IC10,free from the regression equations listed in Table S6, and log DSP/w were measured at concentrations close to IC10,nom. log Dlip/w were collected from literature25,31 or predicted from other experimental descriptors (Figure S8, eqs S9–S11).

Table 1. Cytotoxicity of 11 PFAS in the PPARγ-GeneBLAzer Assay and Distribution Ratios of PFAS between Different Biomaterials and Watera.

  DTXSID IC10,nom [mol/L] IC10,free [mol/L] log Dmedium/w [Lw/Lprot+lip]b log Dcell/w [Lw/Lprot+lip]c log DBSA/w [Lw/Lprot]b log DSP/w [Lw/Lprot]c log Dlip/w [Lw/Llip] TRnom TRfree
PFBA DTXSID4059916 3.19 × 10–3 1.78 × 10–3 2.60 2.40 1.94 2.11 1.00d 2.36 3.88
PFHxA DTXSID3031862 9.95 × 10–4 4.72 × 10–4 2.78 3.45 2.70 2.45 2.32e 0.49 0.70
PFHpA DTXSID1037303 3.75 × 10–4 6.76 × 10–5 3.21 3.69 3.40 2.65 2.91e 0.86 1.26
PFOA DTXSID8031865 1.30 × 10–4 1.84 × 10–5 3.67 3.94 3.98 3.14 3.52e 1.32 1.13
PFNA DTXSID8031863 9.07 × 10–5 7.72 × 10–6 4.14 3.74 4.39 3.39 4.25e 1.68 0.50
PFUnA DTXSID8047553 2.16 × 10–5 3.83 × 10–7 4.74 3.78 4.75 3.97 4.54e 5.04 5.20
HFPO–DA DTXSID70880215 6.54 × 10–4 2.20 × 10–4 3.18 3.09 2.28 2.77 2.41e 0.47 1.22
PFHxS DTXSID3037709 1.46 × 10–4 4.22 × 10–5 3.14 3.56 3.52 2.95 4.13e 0.17 0.12
PFOS DTXSID8037706 4.78 × 10–5 3.89 × 10–6 4.04 4.32 4.52 3.94 4.89e 0.30 0.23
6:2 FTSA DTXSID6067331 3.07 × 10–4 4.25 × 10–5 3.45 3.90 3.71 3.39 3.87f 0.25 0.22
PFOSA DTXSID3038939 7.85 × 10–6 3.28 × 10–7 4.18 3.81 4.33 3.52 4.94f 2.25 2.43
Open in a new tab
a

Nominal and free inhibitory concentrations of PFAS triggering 10% cytotoxicity (IC10,nom and IC10,free). Distribution ratios of PFAS between PPARγ-medium and water (Dmedium/w), between HEK293H (PPARγ) cells and water (Dcell/w), between BSA and water (DBSA/w), and structural protein and water (DSP/w). Literature and predicted distribution ratios between liposome and water (Dlip/w). Toxic ratios (eq 12) with free (TRfree) and nominal (TRnom) concentrations in the PPARγ-GeneBLAzer assay.

b

The concentration-dependent distribution ratios Dmedium/w and DBSA/w at IC10,free were derived from the empirical equations given in Table S6.

c

log Dcell/w and log DSP/w were measured at PFAS concentrations near IC10,nom (Table S2).

d

log Dlip/w were from Droge et al.31

e

log Dlip/w were from Ebert et al.25

f

log Dlip/w of 6:2 FTSA and PFOSA was predicted from the linear relationship of experimental log Dlip/w from literature25,31 against the number of fluorinated carbons (Figure S8, eq S10).

Free and Nominal Concentration of PFAS in the PPARγ-GeneBLAzer Reporter Gene Assay

PFAS are considered proteinophilic and lipophilic,32 and thus, PFAS are prone to bind to components of medium and cells in bioassay systems, indicating Cfree,medium would be lower than the associated Cnom.33 As the numbers of carbons in the three perfluoroalkane carboxylic acids increased, the Cfree,medium deviated more from the 1:1 line to the corresponding Cnom, but the deviation was independent of the concentration (Figure 2a). The measured Cfree,medium values of the more hydrophilic PFHxA were close to Cnom, while the hydrophobic PFUnA had Cfree,medium values up to 90 times lower than Cnom. The relationship between log Cfree,medium and log Cnom was not linear, which was most pronounced for PFHxS and PFOS (Figure 2b). The difference can be explained because sulfonic acids were found to have specific binding to proteins at low concentration ranges, resulting in a higher deviation of Cfree,medium from Cnom at low concentrations.

Figure 2.

Figure 2

Open in a new tab

Relationship between nominal (Cnom) and free (Cfree,medium) concentrations measured in the PPARγ-GeneBLAzer reporter gene assay for (a) three perfluoroalkyl carboxylic acids and (b) three perfluoroalkyl sulfonic acids. (c) Experimental (Exp) distribution ratios between medium and water (Dmedium/w) and cell and water (Dcell/w) of 11 PFAS plotted against predicted Dmedium/w and Dcell/w. Dmedium/w was predicted by eq 4 with experimental distribution ratios between BSA and water (DBSA/w) and between liposome and water (Dlip/w) (Table 1). Dcell/w values were predicted by eq 5 with distribution ratios between structural protein and water (DSP/w) and Dlip/w (Table 1). (d) Inhibitory concentration IC10,free values were derived from concentration–cytotoxicity curves at 10% cytotoxicity with measured Cfree,medium. Experimental (Exp) IC10,free plotted against IC10,free predicted by the MBM (eq 6) with Dlip/w, DBSA/w, and DSP/w.

The observed differences between Cfree,medium and Cnom can be explained by the MBM. The Cfree,medium predicted by the MBM (eq 3) with Dmedium/w (Table S6) and Dcell/w (Table 1) agreed well with the experimental ones in a concentration-dependent way (Figure S9). Cfree,medium predicted by the MBM (eq 6) with Dlip/w, DBSA/w, and DSP/w were also consistent (Figure S9) because the experimental log Dmedium/w (Table 1) at IC10 and were represented well by log DBSA/w at IC10 and log Dlip/w (eq 4), as well as the log Dcell/w with log Dsp/w and log Dlip/w (eq 5) in Figure 2c. The volume fractions of proteins and lipids in the medium and cells are shown in Tables S7 and S8. Consequently, the experimental and predicted IC10,free values by the MBM (eq 6) also agreed well (Figure 2d).

Toxic Ratios: Baseline Toxicity and Cytotoxicity of PFAS

The toxic ratio TR (eq 12) allows an estimation if PFAS act as baseline toxicants or exert cytotoxicity due to a specific effect. All experimental IC10,free (Table 1) values were close to the IC10,free,baseline values (Table S9) predicted from the critical membrane concentration of 69 mmol/Llip (eq 8). TRfree of the 11 PFAS were within the range of baseline toxicants of 0.1–10, suggesting that the cytotoxicity of these PFAS is caused by baseline toxicity (diagonal lines in Figure 3a).

Figure 3.

Figure 3

Open in a new tab

TR (eq 12) of PFAS in the PPARγ-GeneBLAzer assay. (a) Measured free concentration at 10% cytotoxicity IC10,free (Table 1) plotted against the IC10,free,baseline predicted from the critical membrane burden of 69 mmol/Llip (eq 8). The diagonal line depicts the TRfree of 1, and the dotted lines represent the range for the baseline toxicity of 10 > TRfree > 0.1. (b) Nominal concentration at 10% cytotoxicity IC10,nom plotted against the IC10,nom,baseline predicted from IC10,free,baseline by eq 3 (filled circle) or eq 6 (empty circle). The diagonal line depicts the TRnom of 1, and the dotted lines represent the range for the baseline toxicity of 10 > TRnom > 0.1.

TRnom is more practical than TRfree because nominal concentrations are widely reported in the literature. IC10,nom,baseline (Table S9) were predicted from IC10,free,baseline by eqs 3 and 6 with the experimental parameters in Table 1. Consistently, the TRnom of the PFAS was still within the range of 0.1–10 (Figure 3b), confirming that PFAS behave like baseline toxicants in the cytotoxicity endpoint of the PPARγ-GeneBLAzer assay.

Development of a Baseline Toxicity Prediction Model for Nominal Cytotoxicity

Log DBSA/w and log DSP/w are both linearly correlated (eq 10) against log Dlip/w (Table 1) for the nine anionic PFAS (Figure S10a,b). 6:2 PFSA and PFOSA were excluded from the regressions in eqs 14 and 15 because their log Dlip/w were only predicted, but they are plotted in Figure S10a,b for comparison.

graphic file with name es3c09950_m014.jpg 14
graphic file with name es3c09950_m015.jpg 15

Previously, a relationship of log DBSA/w against log Dlip/w for neutral chemicals was derived by Endo and Goss34 (eq 16, rescaled from Kow). The higher affinity of anionic PFAS for proteins is evident by the larger slope and intercept in eq 14 than in eq 16. The binding of anionic PFAS to structural proteins (eq 15) was also stronger than that of neutral chemicals (eq 17)35 for log Dlip/w < 5.

graphic file with name es3c09950_m016.jpg 16
graphic file with name es3c09950_m017.jpg 17

Insertion of eqs 14 and 15 in eq 9 leads to eq 18 with sDlip/w as the sole input parameter to derive baseline toxicity predictions for anionic PFAS. The volume of cells is much lower than the volume of the medium, and the volume fraction of protein and lipid in the medium is <1% (Tables S7 and S8). Therefore, Vtot ≈ Vmedium ≈ Vw and the volume fraction (Vf) of protein and lipid in the medium and cells were used to present the distributions of protein and lipid in the baseline toxicity prediction model.

graphic file with name es3c09950_m018.jpg 18

For comparison, a baseline toxicity prediction model for neutral chemicals was developed by the insertion of eqs 16 and 17 into eq 9.

graphic file with name es3c09950_m019.jpg 19

Analogously to Lee et al.,7 empirical baseline toxicity prediction models for anionic PFAS were derived for all four cell lines by inserting bioassay-specific volumes of protein and lipids in cells and medium measured in the present study (Tables S7 and S8) into eq 18. The models differ because the protein and lipid contents in the assay systems are different. The resulting predictions for anionic PFAS are depicted in Figure 4a. It can be expected that any anionic chemicals behave similarly to anionic PFAS, but as the model was calibrated with anionic PFAS data only, we consider it a PFAS-specific model. The model holds for neutral chemicals (Figure 4b) in general and includes neutral PFAS. These bioassay-specific empirical baseline toxicity models for neutral chemicals and anionic PFAS for the four cell lines were derived by exponential fit (eq 11), and the adjustable parameters a, b, and c are listed in Table 2.

Figure 4.

Figure 4

Open in a new tab

Baseline toxicity prediction models for (a) anionic PFAS (eq 18) and (b) neutral chemicals (eq 19) in four cell-based bioassays: PPARγ-GeneBLAzer (magenta), AREc32 (green), AhR CALUX (gold), and neurotoxicity (blue). (c) Baseline toxicity prediction models with the generic cell model (eqs 18, 19). (d) Nominal inhibitory concentration at 10% cytotoxicity IC10,nom of 16 anionic PFAS in four cell-based bioassays: PPARγ (magenta circle), AREc32 (green triangle), AhR (gold diamond), and neurotoxicity (blue square) and color-matched bioassay-specific baseline toxicity prediction models as well as the generic model (black line). The gray area depicts the range for baseline toxicity of 10 > TRnom > 0.1. (e) IC10,nom of partially charged and neutral PFAS in four cell-based bioassays compared with the baseline toxicity prediction models as well as the generic model (black broken line). All fit parameters of the empirical baseline toxicity prediction model (eq 11) are listed in Table 2.

Table 2. Parameters of the Empirical Baseline Toxicity Prediction Model (eq 11) of Anionic and Neutral PFAS in Four Cell-Based Bioassays and the Generic Cell Assay.

    Anionic PFAS
 Neutral PFAS
assay Medium a b c a b c
PPARγ (HEK293H) OptiMEM +2 % FBS 1.25 4.76 0.251 1.24 5.47 0.235
AREc32 (MCF7) DMEM Glutamax + 10% FBS 1.22 3.78 0.263 1.26 4.43 0.278
AhR (H4IIe) DMEM Glutamax + 10% FBS 1.22 3.78 0.263 1.26 4.45 0.277
neurotoxicity (SH-SY5Y) neurobasal medium 1.22 4.07 0.247 1.23 5.61 0.209
generic cell (6% proteins, 0.1% lipids) generic medium (0.3% proteins, 0.001% lipids) 1.22 3.79 0.262 1.26 4.47 0.275
Open in a new tab

If the baseline toxicity models were used to evaluate literature bioassay data with undisclosed conditions, we assumed the volume fractions of protein and lipid in medium are Vfprotein,medium of 3 mLprotein/Lmedium and Vflip,medium of 0.07 mLlipid/Lmedium (Table S7), and Vfprotein,cell is 30 mLprotein/Lcell and Vflip,cell is 5 mLlipid/Lcell (Table S8). The volume of medium Vmedium and volume of cells Vcell are dependent on microtiter plates in 96-, 384-, or 1536-well formats, e.g., Vmedium of 40 μL and Vcell of approximately 30 nL for the 384-well plate format. The fit parameters of this generic bioassay model are also listed in Table 2.

These two models are compared in Figure 4c, where the two lines started to separate at log Dlip/w > 2. This is due to anionic chemicals having a higher affinity for proteins despite baseline toxicity occurring at concentrations where the protein binding is dominated by nonspecific binding. At log Dlip/w < 2 (e.g., PFBA), IC10,nom is close to IC10,free (Table 1 and Figure S9a). Therefore, eqs 18 and 19 are simplified to eq 20, which is similar to eq 8 and is valid for neutral, anionic, and cationic organic chemicals.

graphic file with name es3c09950_m020.jpg 20

Sensitivity analysis was performed to study how various parameters contribute to the baseline toxicity prediction model. The protein and lipid contents of these four cell lines were similar (Table S8), and thus, the Dcell/w of single PFAS (Table S10) among the four cell lines did not differ much. Besides, the ratio of Vcell/Vmedium is usually <0.001. Therefore, the differences between baseline toxicity prediction models (eqs 18 and 19) with cells and without cells are negligible (Figure S10c).

By contrast, medium components make a difference. Models for AREc32 and AhR CALUX assays were overlaying because the same medium was used in the assays, while less protein and lipid were in the media for PPARγ-GeneBLAzer and neurotoxicity assays (Table S7). The difference is more pronounced as the chemical hydrophobicity increases (Figure 4a,b). The lipid-bound fraction is smaller compared to protein, but it still influences the predicted IC10,nom,baseline. For example, if Vflip,medium changed from 0.07 mL/L to 0, the model for anionic PFAS did not change much because the concentration of anionic PFAS bound to protein is much higher than that bound to lipid, but the model for neutral chemicals deviates for chemicals with log Dlip/w > 4 (Figure S10d). We recommend to measure the volume fractions of protein and lipid in the medium used for bioassays on a routine basis as they are the main determinants of the model.

Is the Cytotoxicity of PFAS Merely Baseline Toxicity?

24 PFAS (Table S1) were measured in this study. These PFAS have diverse functional groups, including sulfonic acid, carboxylic acid, oxide dimer acid, fluorotelomer alcohol, and sulfonamide, as well as some fluorinated pesticides that contain individual CnF2n-groups. Their log Dlip/w ranges over 6 orders of magnitude (Table S11). Log Dlip/w values of these PFAS were used in the baseline toxicity prediction model to derive the IC10,nom,baseline (eq 11).

The IC10,nom of 24 PFAS were measured in four bioassays in 384-well plates (Table S12). The IC10,nom,baseline of 15 anionic PFAS were predicted with eq 18, while IC10,nom,baseline of the other four partially charged and five neutral PFAS were predicted with eq 19. As shown in Figure 4d,e, the measured IC10,nom were within a factor of 10 to the IC10,nom,baseline predicted by the four bioassay-specific models and all well within the gray band of 10 > TRnom > 0.1 of the generic model, indicating that these PFAS did not show specific effects but baseline toxicity was the cause of their cytotoxicity. The generic cell model is adequate to provide IC10,nom,baseline values as a reference for general experimental conditions.

Application of the Baseline Toxicity Prediction Model to Evaluate the Specificity of Effects

The measured EC10,nom values of the agonistic effects on PPARγ, AREc32, AhR, and neurotoxicity, as well as the ECSPR20,nom values of the antagonistic effects on PPARγ, are listed in Table S12, which were derived from the concentration–response curves of the 24 PFAS investigated experimentally in this study (Figure S11). No activation of the oxidative stress response was detected for all 24 PFAS in AREc32 before cytotoxicity started to kick in. A few PFAS showed weak activation of AhR and neurotoxicity. In contrast, PPARγ was a specific target for 19 of the 24 PFAS (Table S12), but the specificity was low. If the EC10,nom or ECSPR20,nom were compared with the IC10,nom,baseline from anionic or neutral baseline toxicity prediction models for PPARγ, AhR, and neurotoxicity (Table 2), as in Figure 5a, SRnom (eq 13) of most PFAS were within 0.1–10, indicating the effects of PFAS in the tested assays are nonspecific and caused by baseline toxicity. Only HFPO–DA had a SRnom > 10 for the agonistic mode of the PPARγ assay, indicating that HFPO–DA may be a specific PPARγ antagonist.

Figure 5.

Figure 5

Open in a new tab

Specificity ratios SRnom (eq 13) of PFAS in cell-based bioassays. (a) SR of 16 anionic (A.) and eight neutral (N.) PFAS in cell-based bioassays of PPARγ, AhR, and neurotoxicity were measured in this study. (b) SRnom of 11 anionic, one partially charged, and four neutral PFAS in cell-based bioassays of human (h) PPARγ, rat (r) PPARγ, hPPARα, rPPARα, and hERs from Evans et al.36

Evaluation of Literature Data for Specificity of Effects

Evans et al.36 studied the agonistic effects of 16 PFAS (Table S13) on five defined targets from humans or rats, including hPPARγ, rPPARγ, hPPARα, rPPARα, and human estrogen receptors (hER). The IC10,nom,baseline of 16 PFAS was calculated by the generic model (Table 2) from their Dlip/w (Table S11). Their EC10,nom were recalculated from a linear portion of concentration–response curves because most of the effects were lower than 50% and the linear model is more suitable to derive 10% effects than a four-parameter logistic model.37 hPPARα was more likely to be activated than rPPARα (Figure 5b). HFPO–DA, HFPO–DA-AS, and PFMOAA showed SRnom > 10, while the other 13 PFAS either had no effects or had 0.1 < SRnom < 10 and were therefore classified as baseline toxicants. The literature data set showed a clear tendency that SRnom decreased with increasing hydrophobicity (Figure 5b). This phenomenon can be explained by hydrophobic PFAS with strong affinity to membranes, which they may mostly accumulate in the cell membrane before reaching any intracellular-specific target sites. The FTOHs should be excluded from the analysis as they might have escaped from the well-plates or cross-contaminated neighboring wells because they are semivolatile.6,38,39

430 PFAS have been selected for a series of bioassay tests in the Tox21/ToxCast program with hundreds of targets (https://comptox.epa.gov/dashboard/chemical-lists/EPAPFASINV). The baseline toxicity prediction model of the present study can be used to evaluate the specificity of different targets once raw data from these bioassays becomes publicly available. Those PFAS with high values of TR and SR on defined targets may provide some hints on initial molecular events and key events, which will facilitate the in vitro to in vivo extrapolation to adverse outcomes and thus promote the development of AOP related to PFAS. However, current results of cell-based HTS from EPA suggest that a majority of 160 PFAS were inactive or equivocal.18−21

Although some PFAS showed specific effects on defined targets, their intrinsic specificities appear to be rather low. This does not mean that specific effects are absent. There are many reports about the specific effects of PFAS, but if they are not selective but occur at concentrations similar to baseline toxicity, these effects are easily predictable by the baseline toxicity prediction models derived here. Even if individual PFAS are merely baseline toxicants, their potency can still be of concern because baseline toxicants also cover several orders of magnitude in effect potency.

The biological effects of PFAS were usually detected at the micromolar level in toxicological studies. The blood concentrations of PFAS in some workers and residents living near fluorochemical plants have already reached such levels,40 even though that in the general population was at a nanomolar level.41 PFAS are highly persistent and have been termed “forever chemicals”. If their production is not stopped, they will continue to build up over time and may eventually reach levels where they cause effects, especially under chronic exposure, where PFAS are known to adversely affect the immune system42 and cause liver cancer,43 as well as other health impacts.44,45 Baseline toxicants act concentration-additive in mixtures, which means that not only the thousands of PFAS but also all other organic chemicals act together in mixtures. Mixture effects may lead to visible effects even if the individual PFAS’ concentrations are below their individual effect threshold.

Acknowledgments

We gratefully acknowledge access to the platform CITEPro (Chemicals in the Environment Profiler), funded by the Helmholtz Association for chemical analysis and bioassay measurements. We thank Julia Huchthausen for help with the BioSPME measurements, Jungeun Lee for help with the neurotoxicity assay, and Georg Braun for discussions about the dose–response modeling of literature data.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c09950.

  • Addition of information on chemicals, deviation of equations, experimental procedures and details, volume fractions of proteins and lipids in biomaterials, binding isotherms and binding constants of PFAS between biomaterials and water, a comparison of free and nominal concentrations of PFAS in the bioassay, all results from cell-based HTS assays, and concentration–response curves of 24 PFAS in the PPARγ-GeneBLAzer reporter gene assay (PDF)

W.Q. received funding through the Chinese Scholarship Council. This study was supported by the Helmholtz Association under the recruiting initiative scheme, which is funded by the German Ministry of Education and Research and was conducted within the Helmholtz POF IV Topic 9 and the Integrated Project “Healthy Planet-towards a non-toxic environment”.

The authors declare no competing financial interest.

Supplementary Material

es3c09950_si_001.pdf (5.1MB, pdf)

References

  1. Wang Z.; Buser A. M.; Cousins I. T.; Demattio S.; Drost W.; Johansson O.; Ohno K.; Patlewicz G.; Richard A. M.; Walker G. W.; White G. S.; Leinala E. A New OECD Definition for Per- and Polyfluoroalkyl Substances. Environ. Sci. Technol. 2021, 55 (23), 15575–15578. 10.1021/acs.est.1c06896. [DOI] [PubMed] [Google Scholar]
  2. Richard A. M.; Lougee R.; Adams M.; Hidle H.; Yang C.; Rathman J.; Magdziarz T.; Bienfait B.; Williams A. J.; Patlewicz G. A New CSRML Structure-Based Fingerprint Method for Profiling and Categorizing Per- and Polyfluoroalkyl Substances (PFAS). Chem. Res. Toxicol. 2023, 36 (3), 508–534. 10.1021/acs.chemrestox.2c00403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Wielsoe M.; Long M.; Ghisari M.; Bonefeld-Jorgensen E. C. Perfluoroalkylated substances (PFAS) affect oxidative stress biomarkers in vitro. Chemosphere 2015, 129, 239–245. 10.1016/j.chemosphere.2014.10.014. [DOI] [PubMed] [Google Scholar]
  4. Behr A. C.; Plinsch C.; Braeuning A.; Buhrke T. Activation of human nuclear receptors by perfluoroalkylated substances (PFAS). Toxicol. In Vitro 2020, 62, 104700. 10.1016/j.tiv.2019.104700. [DOI] [PubMed] [Google Scholar]
  5. Escher B. I.; Eggen R. I.; Schreiber U.; Schreiber Z.; Vye E.; Wisner B.; Schwarzenbach R. P. Baseline toxicity (narcosis) of organic chemicals determined by in vitro membrane potential measurements in energy-transducing membranes. Environ. Sci. Technol. 2002, 36 (9), 1971–1979. 10.1021/es015844c. [DOI] [PubMed] [Google Scholar]
  6. Escher B. I.; Glauch L.; Konig M.; Mayer P.; Schlichting R. Baseline Toxicity and Volatility Cutoff in Reporter Gene Assays Used for High-Throughput Screening. Chem. Res. Toxicol. 2019, 32 (8), 1646–1655. 10.1021/acs.chemrestox.9b00182. [DOI] [PubMed] [Google Scholar]
  7. Lee J.; Braun G.; Henneberger L.; Konig M.; Schlichting R.; Scholz S.; Escher B. I. Critical Membrane Concentration and Mass-Balance Model to Identify Baseline Cytotoxicity of Hydrophobic and Ionizable Organic Chemicals in Mammalian Cell Lines. Chem. Res. Toxicol. 2021, 34 (9), 2100–2109. 10.1021/acs.chemrestox.1c00182. [DOI] [PubMed] [Google Scholar]
  8. Maeder V.; Escher B. I.; Scheringer M.; Hungerbühler K. Toxic ratio as an indicator of the intrinsic toxicity in the assessment of persistent, bioaccumulative, and toxic chemicals. Environ. Sci. Technol. 2004, 38 (13), 3659–3666. 10.1021/es0351591. [DOI] [PubMed] [Google Scholar]
  9. Escher B. I.; Henneberger L.; Konig M.; Schlichting R.; Fischer F. C. Cytotoxicity Burst? Differentiating Specific from Nonspecific Effects in Tox21 in Vitro Reporter Gene Assays. Environ. Health Perspect. 2020, 128 (7), 77007. 10.1289/EHP6664. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Qin W.; Ren X.; Zhao L.; Guo L. Exposure to perfluorooctane sulfonate reduced cell viability and insulin release capacity of β cells. J. Environ. Sci. 2022, 115, 162–172. 10.1016/j.jes.2021.07.004. [DOI] [PubMed] [Google Scholar]
  11. Sun P.; Nie X.; Chen X.; Yin L.; Luo J.; Sun L.; Wan C.; Jiang S. Nrf2 Signaling Elicits a Neuroprotective Role Against PFOS-mediated Oxidative Damage and Apoptosis. Neurochem. Res. 2018, 43 (12), 2446–2459. 10.1007/s11064-018-2672-y. [DOI] [PubMed] [Google Scholar]
  12. Ojo A. F.; Peng C.; Ng J. C. Combined effects of mixed per- and polyfluoroalkyl substances on the Nrf2-ARE pathway in ARE reporter-HepG2 cells. J. Hazard. Mater. 2022, 421, 126827. 10.1016/j.jhazmat.2021.126827. [DOI] [PubMed] [Google Scholar]
  13. Li C. H.; Ren X. M.; Ruan T.; Cao L. Y.; Xin Y.; Guo L. H.; Jiang G. Chlorinated Polyfluorinated Ether Sulfonates Exhibit Higher Activity toward Peroxisome Proliferator-Activated Receptors Signaling Pathways than Perfluorooctanesulfonate. Environ. Sci. Technol. 2018, 52 (5), 3232–3239. 10.1021/acs.est.7b06327. [DOI] [PubMed] [Google Scholar]
  14. Long M.; Ghisari M.; Bonefeld-Jorgensen E. C. Effects of perfluoroalkyl acids on the function of the thyroid hormone and the aryl hydrocarbon receptor. Environ. Sci. Pollut. Res. Int. 2013, 20 (11), 8045–8056. 10.1007/s11356-013-1628-7. [DOI] [PubMed] [Google Scholar]
  15. Qin W. P.; Cao L. Y.; Li C. H.; Guo L. H.; Colbourne J.; Ren X. M. Perfluoroalkyl Substances Stimulate Insulin Secretion by Islet β Cells via G Protein-Coupled Receptor 40. Environ. Sci. Technol. 2020, 54 (6), 3428–3436. 10.1021/acs.est.9b07295. [DOI] [PubMed] [Google Scholar]
  16. Yadav A.; Amber M.; Zosen D.; Labba N. A.; Huiberts E. H. W.; Samulin Erdem J.; Haugen F.; Berntsen H. F.; Zienolddiny S.; Paulsen R. E.; Ropstad E.; Connolly L.; Verhaegen S. A human relevant mixture of persistent organic pollutants (POPs) and perfluorooctane sulfonic acid (PFOS) enhance nerve growth factor (NGF)-induced neurite outgrowth in PC12 cells. Toxicol. Lett. 2021, 338, 85–96. 10.1016/j.toxlet.2020.12.007. [DOI] [PubMed] [Google Scholar]
  17. Patlewicz G.; Richard A. M.; Williams A. J.; Grulke C. M.; Sams R.; Lambert J.; Noyes P. D.; DeVito M. J.; Hines R. N.; Strynar M.; Guiseppi-Elie A.; Thomas R. S. A Chemical Category-Based Prioritization Approach for Selecting 75 Per- and Polyfluoroalkyl Substances (PFAS) for Tiered Toxicity and Toxicokinetic Testing. Environ. Health Perspect. 2019, 127 (1), 14501. 10.1289/EHP4555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Houck K. A.; Patlewicz G.; Richard A. M.; Williams A. J.; Shobair M. A.; Smeltz M.; Clifton M. S.; Wetmore B.; Medvedev A.; Makarov S. Bioactivity profiling of per- and polyfluoroalkyl substances (PFAS) identifies potential toxicity pathways related to molecular structure. Toxicology 2021, 457, 152789. 10.1016/j.tox.2021.152789. [DOI] [PubMed] [Google Scholar]
  19. Houck K. A.; Friedman K. P.; Feshuk M.; Patlewicz G.; Smeltz M.; Clifton M. S.; Wetmore B. A.; Velichko S.; Berenyi A.; Berg E. L. Evaluation of 147 perfluoroalkyl substances for immunotoxic and other (patho)physiological activities through phenotypic screening of human primary cells. ALTEX 2023, 40 (2), 248–270. 10.14573/altex.2203041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Carstens K. E.; Freudenrich T.; Wallace K.; Choo S.; Carpenter A.; Smeltz M.; Clifton M. S.; Henderson W. M.; Richard A. M.; Patlewicz G.; Wetmore B. A.; Paul Friedman K.; Shafer T. Evaluation of Per- and Polyfluoroalkyl Substances (PFAS) In Vitro Toxicity Testing for Developmental Neurotoxicity. Chem. Res. Toxicol. 2023, 36 (3), 402–419. 10.1021/acs.chemrestox.2c00344. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Stoker T. E.; Wang J.; Murr A. S.; Bailey J. R.; Buckalew A. R. High-Throughput Screening of ToxCast PFAS Chemical Library for Potential Inhibitors of the Human Sodium Iodide Symporter. Chem. Res. Toxicol. 2023, 36 (3), 380–389. 10.1021/acs.chemrestox.2c00339. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fischer F. C.; Abele C.; Droge S. T. J.; Henneberger L.; Konig M.; Schlichting R.; Scholz S.; Escher B. I. Cellular Uptake Kinetics of Neutral and Charged Chemicals in in Vitro Assays Measured by Fluorescence Microscopy. Chem. Res. Toxicol. 2018, 31 (8), 646–657. 10.1021/acs.chemrestox.8b00019. [DOI] [PubMed] [Google Scholar]
  23. Qin W.; Henneberger L.; Huchthausen J.; Konig M.; Escher B. I. Role of bioavailability and protein binding of four anionic perfluoroalkyl substances in cell-based bioassays for quantitative in vitro to in vivo extrapolations. Environ. Int. 2023, 173, 107857. 10.1016/j.envint.2023.107857. [DOI] [PubMed] [Google Scholar]
  24. Henneberger L.; Goss K. U.; Endo S. Partitioning of Organic Ions to Muscle Protein: Experimental Data, Modeling, and Implications for in Vivo Distribution of Organic Ions. Environ. Sci. Technol. 2016, 50 (13), 7029–7036. 10.1021/acs.est.6b01417. [DOI] [PubMed] [Google Scholar]
  25. Ebert A.; Allendorf F.; Berger U.; Goss K. U.; Ulrich N. Membrane/Water Partitioning and Permeabilities of Perfluoroalkyl Acids and Four of their Alternatives and the Effects on Toxicokinetic Behavior. Environ. Sci. Technol. 2020, 54 (8), 5051–5061. 10.1021/acs.est.0c00175. [DOI] [PubMed] [Google Scholar]
  26. Henneberger L.; Muhlenbrink M.; Konig M.; Schlichting R.; Fischer F. C.; Escher B. I. Quantification of freely dissolved effect concentrations in in vitro cell-based bioassays. Arch. Toxicol. 2019, 93 (8), 2295–2305. 10.1007/s00204-019-02498-3. [DOI] [PubMed] [Google Scholar]
  27. Fischer F. C.; Henneberger L.; Konig M.; Bittermann K.; Linden L.; Goss K. U.; Escher B. I. Modeling Exposure in the Tox21 in Vitro Bioassays. Chem. Res. Toxicol. 2017, 30 (5), 1197–1208. 10.1021/acs.chemrestox.7b00023. [DOI] [PubMed] [Google Scholar]
  28. Neale P. A.; Altenburger R.; Ait-Aissa S.; Brion F.; Busch W.; de Aragao Umbuzeiro G.; Denison M. S.; Du Pasquier D.; Hilscherova K.; Hollert H.; Morales D. A.; Novak J.; Schlichting R.; Seiler T. B.; Serra H.; Shao Y.; Tindall A. J.; Tollefsen K. E.; Williams T. D.; Escher B. I. Development of a bioanalytical test battery for water quality monitoring: Fingerprinting identified micropollutants and their contribution to effects in surface water. Water Res. 2017, 123, 734–750. 10.1016/j.watres.2017.07.016. [DOI] [PubMed] [Google Scholar]
  29. Konig M.; Escher B. I.; Neale P. A.; Krauss M.; Hilscherova K.; Novak J.; Teodorovic I.; Schulze T.; Seidensticker S.; Kamal Hashmi M. A.; Ahlheim J.; Brack W. Impact of untreated wastewater on a major European river evaluated with a combination of in vitro bioassays and chemical analysis. Environ. Pollut. 2017, 220 (Pt B), 1220–1230. 10.1016/j.envpol.2016.11.011. [DOI] [PubMed] [Google Scholar]
  30. Escher B. I.; Dutt M.; Maylin E.; Tang J. Y.; Toze S.; Wolf C. R.; Lang M. Water quality assessment using the AREc32 reporter gene assay indicative of the oxidative stress response pathway. J. Environ. Monit 2012, 14 (11), 2877–2885. 10.1039/c2em30506b. [DOI] [PubMed] [Google Scholar]
  31. Droge S. T. J. Membrane-Water Partition Coefficients to Aid Risk Assessment of Perfluoroalkyl Anions and Alkyl Sulfates. Environ. Sci. Technol. 2019, 53 (2), 760–770. 10.1021/acs.est.8b05052. [DOI] [PubMed] [Google Scholar]
  32. Conder J. M.; Hoke R. A.; Wolf W. d.; Russell M. H.; Buck R. C. Are PFCAs bioaccumulative? A critical review and comparison with regulatory criteria and persistent lipophilic compounds. Environ. Sci. Technol. 2008, 42 (4), 995–1003. 10.1021/es070895g. [DOI] [PubMed] [Google Scholar]
  33. Henneberger L.; Muhlenbrink M.; Fischer F. C.; Escher B. I. C18-Coated Solid-Phase Microextraction Fibers for the Quantification of Partitioning of Organic Acids to Proteins, Lipids, and Cells. Chem. Res. Toxicol. 2019, 32 (1), 168–178. 10.1021/acs.chemrestox.8b00249. [DOI] [PubMed] [Google Scholar]
  34. Endo S.; Goss K. U. Serum albumin binding of structurally diverse neutral organic compounds: data and models. Chem. Res. Toxicol. 2011, 24 (12), 2293–2301. 10.1021/tx200431b. [DOI] [PubMed] [Google Scholar]
  35. Endo S.; Bauerfeind J.; Goss K. U. Partitioning of neutral organic compounds to structural proteins. Environ. Sci. Technol. 2012, 46 (22), 12697–12703. 10.1021/es303379y. [DOI] [PubMed] [Google Scholar]
  36. Evans N.; Conley J. M.; Cardon M.; Hartig P.; Medlock-Kakaley E.; Gray L. E. Jr In vitro activity of a panel of per- and polyfluoroalkyl substances (PFAS), fatty acids, and pharmaceuticals in peroxisome proliferator-activated receptor (PPAR) alpha, PPAR gamma, and estrogen receptor assays. Toxicol. Appl. Pharmacol. 2022, 449, 116136. 10.1016/j.taap.2022.116136. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Escher B. I.; Neale P. A.; Villeneuve D. L. The advantages of linear concentration-response curves for in vitro bioassays with environmental samples. Environ. Toxicol. Chem. 2018, 37 (9), 2273–2280. 10.1002/etc.4178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Goss K. U.; Bronner G.; Harner T.; Hertel M.; Schmidt T. C. The partition behavior of fluorotelomer alcohols and olefins. Environ. Sci. Technol. 2006, 40 (11), 3572–3577. 10.1021/es060004p. [DOI] [PubMed] [Google Scholar]
  39. Fischer F. C.; Henneberger L.; Schlichting R.; Escher B. I. How To Improve the Dosing of Chemicals in High-Throughput in Vitro Mammalian Cell Assays. Chem. Res. Toxicol. 2019, 32 (8), 1462–1468. 10.1021/acs.chemrestox.9b00167. [DOI] [PubMed] [Google Scholar]
  40. Gao Y.; Fu J.; Cao H.; Wang Y.; Zhang A.; Liang Y.; Wang T.; Zhao C.; Jiang G. Differential accumulation and elimination behavior of perfluoroalkyl Acid isomers in occupational workers in a manufactory in China. Environ. Sci. Technol. 2015, 49 (11), 6953–6962. 10.1021/acs.est.5b00778. [DOI] [PubMed] [Google Scholar]
  41. Gockener B.; Weber T.; Rudel H.; Bucking M.; Kolossa-Gehring M. Human biomonitoring of per- and polyfluoroalkyl substances in German blood plasma samples from 1982 to 2019. Environ. Int. 2020, 145, 106123. 10.1016/j.envint.2020.106123. [DOI] [PubMed] [Google Scholar]
  42. Ehrlich V.; Bil W.; Vandebriel R.; Granum B.; Luijten M.; Lindeman B.; Grandjean P.; Kaiser A. M.; Hauzenberger I.; Hartmann C.; Gundacker C.; Uhl M. Consideration of pathways for immunotoxicity of per- and polyfluoroalkyl substances (PFAS). Environ. Health 2023, 22 (1), 19. 10.1186/s12940-022-00958-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Costello E.; Rock S.; Stratakis N.; Eckel S. P.; Walker D. I.; Valvi D.; Cserbik D.; Jenkins T.; Xanthakos S. A.; Kohli R.; Sisley S.; Vasiliou V.; La Merrill M. A.; Rosen H.; Conti D. V.; McConnell R.; Chatzi L. Exposure to per- and Polyfluoroalkyl Substances and Markers of Liver Injury: A Systematic Review and Meta-Analysis. Environ. Health Perspect. 2022, 130 (4), 46001. 10.1289/EHP10092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Chang E. T.; Adami H. O.; Boffetta P.; Cole P.; Starr T. B.; Mandel J. S. A critical review of perfluorooctanoate and perfluorooctanesulfonate exposure and cancer risk in humans. Crit. Rev. Toxicol. 2014, 44 (sup1), 1–81. 10.3109/10408444.2014.905767. [DOI] [PubMed] [Google Scholar]
  45. Fenton S. E.; Ducatman A.; Boobis A.; DeWitt J. C.; Lau C.; Ng C.; Smith J. S.; Roberts S. M. Per- and Polyfluoroalkyl Substance Toxicity and Human Health Review: Current State of Knowledge and Strategies for Informing Future Research. Environ. Toxicol. Chem. 2021, 40 (3), 606–630. 10.1002/etc.4890. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es3c09950_si_001.pdf (5.1MB, pdf)

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES