Perfluorooctane sulfonate (PFOS) interferes with Cyp1a activity and dysregulates the aryl hydrocarbon receptor pathway in larval zebrafish (Danio rerio)

Abstract

Per- and poly- fluoroalkyl substances (PFAS) are ubiquitous, persistent toxicants. In the environment, PFAS co-occur with other compounds, including agonists of the aryl hydrocarbon receptor (AHR). In this study, in silico molecular docking, in vitro enzyme activity, and the in vivo zebrafish model were used to identify PFAS that may interact with the AHR pathway and assess impacts for developmental toxicity. Of the PFAS and proteins modeled, perfluorooctane sulfonate (PFOS) was predicted in silico by AutoDock Vina to have good affinity for CYP1A1. This prediction was supported by enzyme inhibition assays in vitro, where PFOS inhibited human CYP1A1 (estimated IC50 of 35 μM). For in vivo testing, two PFAS were selected: PFOS and the related perfluorohexane sulfonate (PFHxS), which did not inhibit CYP1A1 activity at tested concentrations in vitro. Zebrafish embryos were exposed from 24 to 100 hours post-fertilization to mixtures of PFAS (16 μM) with and without the AHR agonist β-naphthoflavone (βNF; 3.67, 36.7, and 183.6 nM). Co-exposures of βNF and PFOS caused severe morphological deformities not observed with either compound alone, accompanied by reduced Cyp1a-like enzyme activity, and cyp1a gene expression. In contrast, PFHxS did not interfere with Cyp1a activity in vivo and did not cause developmental toxicity. This suggests developmental toxicity of mixtures containing PFOS may be underestimated, and identification of other PFAS that interact with the AHR pathway is warranted.

Graphical Abstract

1). Introduction

Per- and poly-fluoroalkyl substances (PFAS) are ubiquitous environmental pollutants, detected in soil and water around the globe (Buck et al., 2011; Chen et al., 2012; Oliaei et al., 2013; Chen et al., 2018; Gobelius et al., 2018; Schulz et al., 2020). PFAS have adverse effects on aquatic life (Martin et al., 2004) and an ability to bioconcentrate in fish (Huang et al., 2010). PFAS concentrations have reached 72.9 μg/g in fish (Moody et al., 2004) and 120 μg/g in selected wildlife tissues (Cartron et al., 2025). In AFFF exposed fish and mice, linear PFOS has been found to be the majority of sulfonated PFAS by internal tissue concentration (Hill et al., 2024; McDonough et al., 2020). PFAS pose significant risk to human health, and their ability to cross the placenta leads to concerns regarding their developmental toxicity (Inoue et al., 2004; Toft et al., 2015; Yao et al., 2019). One of the most toxic and bioaccumulative PFAS is perfluorooctane sulfonate (PFOS), an 8-C fully fluorinated, sulfonated PFAS (Vogs et al., 2019). Another toxic PFAS is perfluorohexane sulfonate (PFHxS), a 6-C, fully fluorinated, sulfonated PFAS—frequently measured as one of the most common PFAS pollutants in the environment (Zafeiraki et al., 2015; Xie et al., 2020). Both PFOS and PFHxS are known to have longer half lives in humans of 3.4 and 5.3 years, respectively (Olsen et al., 2007; Zhang et al., 2013; Fu et al., 2016; Liu et al., 2017; Li et al., 2018).

The aryl hydrocarbon receptor (AHR) is a transcription factor (Kewley et al., 2004) with hundreds of exogenous and endogenous ligand activators (Rannug et al., 1995). Strength of binding to AHR and of its activation can vary among its ligands and be influenced by other pathways and small molecules. The mechanism of AHR regulation of its target genes is well studied (Pollenz et al., 1994; Eguchi et al., 1997; Lees and Whitelaw, 1999). The most inducible of those is cytochrome p450 1A1 (CYP1A1) in humans (Nebert et al., 2004; Saad et al., 2016) and cyp1a in zebrafish¹ (Jönsson et al., 2007).

Little is known about PFAS interactions with the AHR pathway. PFAS and polycyclic aromatic hydrocarbons (PAHs), a class of chemicals containing several AHR agonists, are ubiquitous and exist in environmental matrices including surface waters (Euripidou and Murray, 2004; Couderc et al., 2015), soils and sediments (Zhang et al., 2024; Jolankai et al., 2026), and in human samples, particularly in human ovarian follicular fluid (Young et al., 2025). This raises concerns about developmental toxicity in mixture exposures. Some AHR ligands (e.g., beta-naphthoflavone (βNF)) are substrates for biotransformation by CYP1A1 into more hydrophilic substances for excretion, while ligands that cannot be biotransformed by CYP1A1 can exert toxicity via prolonged AHR activation.

In silico molecular docking has been used increasingly to predict affinity and binding poses of interactions between proteins (receptors) and small molecules (ligands) (Shoichet, 2004). Several programs perform molecular docking; for example, AutoDock Vina is an open-source turnkey computational docking program based on a scoring function and gradient-optimization conformational search (Trott and Olson, 2010). It has been used to help identify novel PFAS interactions with transcription factors and cytochrome p450 proteins (Hvizdak et al., 2023; Li et al., 2024). Furthermore, the AlphaFold neural network used to predict protein structures (Jumper et al., 2021) has been improved to predict protein interactions with ligands and generate a binding pose, with AlphaFold 3’s (AF3) poses being significantly more accurate than Vina’s (Abramson et al., 2024). AF3 has been recently used to evaluate PFAS binding to various proteins. AF3 performed well but favored a PFAS headgroup–polar residue interaction. Thus, a complementary use of Vina and AF3 more accurately model PFAS interactions with proteins (Gong et al., 2025). While docking has been used to explore PFAS interactions with AHR (Ma et al., 2024) and CYPs (Hvizdak et al., 2023), it has not yet been used to investigate CYP1 family enzymes affinity for PFAS.

PFAS interactions with receptors have been extensively explored using in vitro assays, including proteins upstream and within the AHR pathway. Transactivation assays have tested PFAS with PPARs, PXR, FXR and AHR (Behr et al., 2020a; Lai, 2021; Houck et al., 2021). A study of 142 PFAS showed 9 activated AHR moderately, of which two (sodium perfluorodecane sulfonate, and 1-iodo,1H,1H,2H,2H-perfluoroheptane) activated AHR strongly (Houck et al., 2021). Furthermore, PFAS effects on enzymatic activity of CYPs has been tested in microsomes (Solan and Lavado, 2023), where PFOS and PFHxS were found to inhibit CYP1A2 (Solan and Lavado, 2023). Together, these findings suggest that PFAS have potential to interact with AHR receptors or CYP1 family proteins. CYP1A1, an enzyme metabolizing several xenobiotics, remains understudied in the context of PFAS exposures.

In this study, we hypothesized that some PFAS may interact with the AHR pathway, interfere with CYP1A1 activity, and exacerbate toxicity of AHR agonists. To test this, we performed in silico docking of selected PFAS and AHR pathway–related proteins. CYP1A1 interactions with PFAS were explored in vitro using microsomes and the ethoxyresorufin-o-deethylase (EROD) assay. We used larval zebrafish, a well-known biomedical model, to assess developmental toxicity and Cyp1a enzymatic activity using the EROD assay in vivo. EROD assay is a well-known and widely used method to assess AHR activation in zebrafish (Roy et al., 2020). Zebrafish larvae are sensitive to AHR activation and Cyp1a inhibition, which have induced developmental toxicity (Billiard et al., 2006; Jönsson et al., 2012; Roy et al., 2019). βNF was used as a proxy for AHR agonists to induce cyp1a expression, as it safer to dose than other AHR agonists (e.g. TCDD, B[a]P, PCB-126) and has a lower logK_ow (4.79) (Yun et al., 2015), in the context of aquatic toxicity testing. Previous studies have shown that B[a]P and βNF have relatively similar effects (Gesto et al., 2009), induce EROD fluorescence comparably in zebrafish (Boehler et al., 2018), and bind to similar residues of AHR (Diao et al., 2025). This study demonstrates a combination of in silico, in vitro, and in vivo approaches to generate hypotheses, reduce animal usage, and assess whether selected PFAS have potential to inhibit CYP1A1 enzymatic activity, thereby exacerbating AHR pathway–mediated toxicity.

2). Materials and Methods

2.1). Molecular Docking

Proteins relevant to AHR and pathways interacting with it were obtained from RSCB PDB (Supp. Table S1). PFAS ligands were obtained from PubChem (Supp. Table S2). Vina (version 1.1.2 (May 11, 2011)) (Trott and Olson, 2010) and AutoDock tools (version 4.2.6) were obtained from the Scripps Institute. PyMol 2.5.4 was obtained from Schrodinger, LLC. AF3 (Google DeepMind version 0.1.0) as available on the UNITY computing cluster (Massachusetts Green High Performance Computing Center) was used. Protein preparation was performed in accordance with Li et al. (2024). Ligands without 3D structures were converted using OpenBabel. Docking was performed using Vina using exhaustiveness 25 and energy range 4. Binding poses were visualized with PyMol. Protein sequences for AF3 input were obtained from UniProt (Table S1). Truncated sequences were used for AHR to increase confidence (Mamun et al., 2025). Ligands were specified with CCD IDs where possible and SMILES strings from PubChem otherwise (Supp. Table S2). Adenosine monophosphate (AMP) was used as an example molecule that does not bind to the proteins selected.

2.2). Chemicals

Chemicals were purchased as follows: PFOS, PFHxS, perfluorooctanoic acid (PFOA), 6:2 fluorotelomer alcohol (6:2 FTOH), βNF and D-glucose 6 phosphate (G6P) from Sigma Aldrich (St. Louis, MO, USA); 6:2 fluorotelomer sulfonate (6:2 FTS), perfluorohexanoic acid (PFHxA), and 7-ethoxyresorufin (7-ER) from Cayman Chemical (Ann Arbor, MI, USA). Nicotinamide adenine dinucleotide phosphate (NADP) monosodium salt was obtained from Millipore Corporation USA (St. Louis, MO, USA). CYPExpress CYP1A1 was obtained from Sigma Aldrich (St. Louis, MO, USA). All chemicals were dissolved in 100% dimethyl sulfoxide (DMSO) from Fisher Scientific (Fairlawn, NJ, USA). CYP1A1 microsomes with coenzymes were reconstituted to a concentration of 50 mg/mL in a potassium phosphate buffered saline solution (PBS, pH 7.4) and stored at −80°C. PFAS chemicals were stored at room temperature in 1.5 mL Eppendorf tubes, except 6:2 FTS and PFHxA, which were stored at −20°C. 7-ER and βNF chemicals were stored in amber glass vials at −20°C. NADP and G6P were reconstituted in PBS, pH 7.4; NADP was stored at −20°C in darkness and G6P at room temperature.

2.3). Chemical Exposures

In vitro, the 160 μM concentration of tested PFAS to assess inhibition of CYP1A1 was comparable with the highest value reported for half-maximal-inhibition concentration (IC50) in a previous study of CYPs and PFAS (Solan and Lavado, 2023). The concentration of 7-ER (0.1 μg/L) was below Michaelis’ constant as estimated by our prior experiment. Alpha-naphthoflavone (ANF) (0.1 μM) was used to inhibit CYP1A1 enzymatic activity. The CYPExpress kit (Oxford Biomed) was used following the manufacturer’s protocol with modifications previously performed by others (Fliszár-Nyúl et al., 2022; Tuet et al., 2022; Kaci et al., 2024). Eight reactions (6 PFAS, ANF, and 0.1% v/v DMSO), as well as a blank and a no-cofactors control were prepared in PBS, in duplicates, in dark-sided, 96-well, polystyrene plates. Each reaction mix contained 1 mg/mL CYP1A1 CYPExpress, 5.0 mM glucose 6 phosphate, and 2.0 mM NADP monosodium salt. Reaction mixes were shaken for 20 minutes at 37°C. Then, 7-ER was added to start the reaction. Fluorescence intensity measurements were taken immediately at 37°C. Separately, to determine PFOS IC50, six individual PFOS concentrations (4, 8, 20, 40, 80, and 160 μM) were chosen to fit a logarithmic curve. Eight reactions of six PFOS concentrations, ANF, and 0.1% DMSO were measured under the above-mentioned experimental conditions.

For in vivo experiments, concentrations of 16 μM PFOS and PFHxS were selected as they were sub-lethal and largely did not induce significant morphological deformities in this study (Supp. Fig. S1) and in Satbhai et al. (2025). βNF concentrations used in this study showed induction of Cyp1a-like activity (Billiard et al., 2006). The concentration of 7-ER was selected from a previous study (Roy et al., 2020). At 24 hpf, healthy embryos were exposed in eight groups as follows: exposure to one concentration of a PFAS (16 μM PFOS or 16 μM PFHxS), three βNF concentrations (3.67, 36.7, 183.6 nM), three co-exposures, consisting of 16 μM of the selected PFAS co-exposed with each βNF concentration, or a solvent control (DMSO, 0.02% v/v) (Supp. Fig. S2). For each treatment group, 10 embryos were placed in 10 mL dosing solution (0.3x Danieau’s medium, aerated overnight) in 20 mL borosilicate glass vials conditioned overnight with dosing solutions. A parallel experiment with 18 embryos in 18 mL of dosing solution was conducted for gene expression analysis. At 100 hpf, zebrafish larvae were rinsed with 0.3x Danieau’s medium thrice before collection.

2.4). Zebrafish Husbandry

Adult zebrafish (Danio rerio) were housed on a 14:10 hour light:dark cycle in a recirculating custom build aquatics system (Danio Labs, Boston, MA), maintained at 28°C, and fed twice daily with Gemma Micro 300 (Skretting, Westbrook, ME). All animals were treated humanely, and experiments were conducted following protocols approved by the University of Massachusetts Amherst Institutional Animal Care and Use Committee (Animal Welfare Assurance Number A3551–01). We used AB wildtype fish expressing transgenes to label the pancreatic islet (Tg(ins:GFP)) and liver (Tg(gut:GFP)). Embryos were collected 1 hour post-fertilization (hpf), washed, selected for developmental stage and health, and placed in 0.00003% methylene blue in 0.3x Danieau’s media (17 mM NaCl, 2 mM KCl, 0.12 mM MgSO₄, 1.8 mM Ca(NO₃)₂, 1.5 mM HEPES, pH 7.2). Embryos were maintained in an incubator at 28°C with a 14:10 hour light:dark cycle. At 3 hpf, embryos were washed thrice to remove methylene blue. At 24 hpf, embryos were assessed for health and were randomly distributed to experimental groups for exposures.

2.5). EROD, Morphology, and Gene Expression

In vitro CYP1A1 activity was measured using the EROD assay, by which CYP1A1 activity converts weakly fluorescent 7-ER to fluorescent resorufin. Resorufin concentration produced in vitro is approximated via fluorescence intensity. Measurements of excitation (530/590 nm) and gain 120 were taken at intervals of 1 minute for 15 minutes using a Cytation 3 (Biotek) plate reader.

In vivo, larvae at 100 hpf were anaesthetized using 0.02% w/v MS-222 solution, mounted in methylcellulose, and imaged with a customized Olympus upright microscope by Kramer Scientific, fitted with an Axiocam 503 camera (Carl Zeiss Inc.). Images of the resorufin product in the larval intestines produced by Cyp1a-like activity were captured using a dsRED filter. Full fish images were captured using brightfield settings. Liver and islet images were captured with a GFP filter. All images were converted to .tiff format using Zen software (ZEISS) and imported into FIJI open-source image analysis software for endpoint measurement (Version 2.16.0) (Schindelin et al., 2012). All images were blinded prior to analysis. EROD fluorescence intensity was measured using the brightest average pixel value (using a predetermined circular region of interest), corrected for background intensity and dark pixel value for the microscope. Each replicate’s EROD intensity was normalized to the average of its respective control and all replicates within a treatment group were merged for analysis.

Gene expression was assessed using RT-qPCR following the protocol detailed in Timme-Laragy et al. (2015). Primer sequences are made available (Supp. Table S3). The housekeeping gene b-actin (actb) was used for normalization. Its expression did not change across exposure groups.

2.6). Statistical analysis

Statistical analyses were performed in GraphPad Prism (Version 10.4.1). For the enzyme inhibition assay, linear regression was performed to obtain rate of fluorescence product formation, and a nonlinear regression (log[Inhibitor] vs response, variable slope, 4 parameters) was performed on the rates to estimate an IC50. For data from in vivo experiments, outliers were identified (Grubb’s test, α = 0.05) and removed from the statistical analysis. Additionally, if needed, complete data and statistical analysis including outliers are also available in Supp. Fig. S3. Data was assessed for normality (D’Agostino & Pearson test) and heteroscedasticity (Spearman’s test). For parametric data, a Two-Way Analysis of Variance (ANOVA) was performed with Tukey’s HSD post-hoc test. Otherwise, an ANOVA alternative was performed (Scherier-Ray-Hare test) using RStudio (version 2023.06.0+421). A Two-Way ANOVA was performed on ΔCt values Tukey’s HSD post-hoc test. All empirical experiments were independently replicated 3–4 times.

3). Results

3.1). In silico molecular docking

AF3 predicted structures of PFAS with CYP1A1, CYP1A2, and CYP1B1 possessed low (≤ 0.6) to moderate (≥ 0.6, ≤ 0.8) interaction predicted template modeling (iPTM) scores, indicating most to be possibly correct predictions and some failed predictions (Supp. Table S4). Predicted local distance difference test (pLDDT) scores were poor (≤ 80) (Supp. Table S4). For PFAS structures with AHR, AF3 iPTM scores were good (≥ 0.8), and pLDDT scores (≥ 80) indicated confident predictions. Vina predicted PFOS and PFHxS to make close interaction (3.0 Å, 2.1 Å) with the heme iron in CYP1A1 (Fig. 1A, B). Additionally, PFAS showed moderate affinities for AHR proteins (e.g., CYP1A1, CYP1A2) and proteins in cross-talking pathways (PXR) (Table 1, Supp. Table S5). The AutoDock generated best pose for PFOS, where the ω–1 carbon is closest to the heme iron and the headgroup close to the aspartic acid (ASP 313) within the binding pocket, aligned with the AF3 predicted structure of PFOS with CYP1A1 (Fig. 1A, Supp. Fig. S4A).

Fig. 1.

Structural representation of the highest predicted affinity binding pose for (a) PFOS with CYP1A1 (predicted affinity = −10.0 kcal/mol) and (b) PFHxS with CYP1A1 (predicted affinity = − 7.4 kcal/mol). For both panels, the left shows an overall structure and the right a magnified view of the binding site.

Table 1.

Affinities of PFOS and PFHxS for AHR-relevant proteins

Protein	Ligand	Affinity (kcal/mol)
AHR	B[a]P	−14.1
AHR	PFOS	−8.2
AHR	PFHxS	−7.5
CYP1A1	ANF	−14.7
CYP1A1	PFOS	−10.0
CYP1A1	PFHxS	−7.4
CYP1A2	ANF	−14.6
CYP1A2	PFOS	−8.8
CYP1A2	PFHxS	−8.6
PXR	Rifampicin	−14.6
PXR	PFOS	−8.5
PXR	PFHxS	−7.7

3.2). In vitro PFAS effects on CYP1A1 activity

Several PFAS representing sulfonates (PFOS, 6:2 FTS, PFHxS), carboxylates (PFOA, PFHxA), and an alcohol PFAS (6:2 FTOH) were assayed in vitro at 160 μM to test if they showed ≥ 50% inhibition of CYP1A1 enzymatic activity in 3–5 replicates. Of the PFAS tested, only PFOS showed ≥ 50% inhibition of CYP1A1 activity as compared to 7-ER alone (Fig 2A). Therefore, we tested 6 different concentrations of PFOS (4–160 μM), along with 7-ER (0.1 μg/mL) as a positive control, and ANF (1 μM) as a negative control. We estimated the IC50 as 35 μM (Fig 2B) under the conditions described in methods. Relative fluorescence of each tested PFOS concentration is shown in Fig. 2C.

Fig. 2.

Bar graphs and scatter plots showing PFAS inhibition of CYP1A1 deethylation of 7-ethoxyresorufin: (a) Relative inhibition of CYP1A1 by 160 μM of studied PFAS, (b) IC₅₀ of PFOS inhibition of CYP1A1, IC₅₀ = 34.91 μM (R² = 0.6404), and (c) relative inhibition of CYP1A1 by PFOS. All graphs represent the mean ± SEM for each group (n = 3–5).

3.3). In vivo effects of PFOS and PFHxS on Cyp1a-like activity and morphology

Larval zebrafish were exposed individually to PFOS, PFHxS, and βNF (a model AHR agonist) as well as co-exposed to either PFOS and βNF, or PFHxS and βNF. Pericardial area of the fish co-exposed to PFOS and 183.6 nM βNF group was significantly increased compared to 0.02% DMSO (p < 0.0001), as well as individually to 16 μM PFOS (p < 0.0001) and 183.6 nM βNF (p = 0.0076) (Fig 3A, B, D). These fish had higher incidences of deformities compared to 183.6 nM βNF individually, including pericardial edema, tail kink, yolk sac edema, and spinal deformity (Supp. Fig. S5D). In contrast, overall morphology of fish co-exposed to both PFHxS and βNF and individually exposed to βNF at same concentrations did not show significant differences particularly in fish length (Fig. 3A, C) pericardial area (Fig. 3A, E), or incidence of yolk sac edema, pericardial edema, tail kink, and spinal deformities (Supp. Fig. S5E). Individual exposures to 16 μM PFOS or 16 μM PFHxS did not result in significant differences in fish length or pericardial area. PFOS, but not PFHxS, showed lower incidences of swim bladder inflation (Supp. Fig. S5D, E). We observed reduced islet area in exposures to 16 μM PFOS or 16 μM PFHxS, but no significance (p ≥ 0.222, p = 0.811) (Supp. Fig. S5A, B). Fish liver length was unaltered across treatment groups (Supp. Fig. S5C).

Fig. 3.

Representative images and morphometric measurements of zebrafish exposed to a PFAS and βNF: (a) Representative images of whole zebrafish, (b) fish length of fish exposed to PFOS and βNF (n = 30–28) as well as (c) PFHxS and βNF (n = 29–27), and (d) pericardial area of fish exposed to PFOS and βNF (n = 30–27) as well as (e) PFHxS and βNF (n = 27–24). All graphs represent the mean ± SEM, and stars indicate significant differences as determined by a Two-Way ANOVA (b, c) or Scheirer-Ray-Hare test (d, e) with Tukey’s post–hoc test (p ≤ 0.05).

Individual exposures to 36.7 nM and 183.6 nM βNF significantly induced Cyp1a-like activity, as measured by 2.8–5.6 (184%–460%, p < 0.0001) fold increase in EROD fluorescence in the fish gut (Fig. 4 A, B, C). Fish co-exposed to 16 μM PFOS with 36.7 nM or 183.6 nM βNF showed significant decrease in EROD fluorescence (p < 0.0001) as compared to equivalent individual βNF concentrations (Fig. 4 B, C). In the 16 μM PFOS and 36.7 nM βNF co-exposure group, the EROD fluorescence was significantly higher (p < 0.0001) than in the 0.02% DMSO and 16 μM PFOS, individually. However, fish co-exposed to PFHxS and βNF group and individually exposed to βNF at comparable concentrations did not show any significant differences in EROD fluorescence (Fig. 4C, p ≥ 0.911).

Fig. 4.

Representative images and measurements of EROD fluorescence intensity in zebrafish exposed to a PFAS and βNF: (a) Representative images of EROD activity and (b) relative EROD fluorescence of fish exposed to PFOS and βNF (n = 29–24) as well as (c) PFHxS and βNF (n = 30–28). All graphs represent the mean ± SEM, and stars indicate significant differences as determined by a Scheirer-Ray-Hare test with Tukey’s post–hoc test (p ≤ 0.05).

3.4). PFOS effects on gene expression in vivo

We used RT-qPCR to investigate whether changes in cyp1a gene expression contributed to the decreased Cyp1a enzymatic activity (Fig 4). Compared to 0.02% DMSO, 183.6 nM βNF significantly upregulated cyp1a (by 15-fold, p = 0.0233); however, there was no observed significant upregulation compared to solvent control in the respective 16 μM PFOS co-exposure (2.8-fold increase, p = 0.806) (Fig. 5). Individual exposures to 16 μM PFOS and 0.02% DMSO did not show any difference in cyp1a expression. Abundance of ahr2 transcripts and studied AHR pathway-related genes (cyp7a1, nr0b2a, cyp3a65) remained unaltered across all treatment groups (Supp. Fig. S6).

Fig 5.

Relative gene expression of cyp1a shown normalized to β-actin as a housekeeping gene (mean ± SEM, n = 3–4 vials of 16–18 pooled larvae per exposure group, ANOVA with Tukey’s post-hoc test, stars represent p ≤ 0.05).

4). Discussion

In the present study, PFAS interactions with the AHR pathway were investigated using complimentary approaches: in silico molecular docking, in vitro enzyme activity, and in vivo zebrafish exposures. The integration of these approaches highlights the strength of the zebrafish model in differentiating between adaptive and toxic responses. The data provides evidence of interactions between CYP1A1/Cyp1a and PFOS as tested using in vitro, in vivo, and in silico methods. In vivo reduction in Cyp1a-like activity (as induced by 183.6 nM βNF) with PFOS exposures were concordant with morphological deformities and developmental toxicity. This effect was not seen with PFHxS co-exposures. PFOS inhibited CYP1A1 half-maximally at 35 μM in vitro and was predicted to have good affinity for CYP1A1 in silico, while PFHxS did not achieve half-maximal inhibition at 160 μM, nor was it predicted to have good affinity for CYP1A1. These findings indicate that PFOS can dysregulate the AHR pathway in zebrafish at sub-lethal concentrations.

PFAS have a wide range of modes of action, influenced by both chain length and moiety (Satbhai et al., 2022). Since empirical experimentation has identified some PFAS as agonists of AHR (Houck et al., 2021), and sulfonated but not carboxylated PFAS as CYP1A2 inhibitors (Solan and Lavado, 2023), we performed molecular docking to see if similar trends were observed in silico. While sulfonated PFAS (PFOS and 6:2 FTS) produced the highest predicted affinity of PFAS docked to CYP1A1 (Table 1), both PFOA and PFHxA were predicted to have affinities higher than PFHxS (Supp. Table S5), suggesting no evident trend for moiety. Some PFAS of different chain lengths had similar affinities; therefore, there was no trend in chain length vs affinity. We observed that only PFOS demonstrated good (≤ −10 kcal/mol) predicted affinity for CYP1A1 among all tested PFAS. There was no clear trend in the predicted affinity for AHR and PFAS. PFOSA is hypothesized to bind AHR (Ma et al., 2024) and it possessed the highest affinity (−8.9 kcal/mol) for AHR among PFAS tested. However, several PFAS also showed moderate predicted affinities about −8 kcal/mol. A known AHR agonist benzo[a]pyrene (B[a]P) possessed affinity of −14.1 kcal/mol. With the present molecular docking data, it is difficult to reach reliable conclusions about PFAS interactions with AHR, and further studies are needed to empirically determine these effects.

AF3 predicted structures for PFAS and CYP1A1 possessed moderate (0.53–0.71) iPTM scores, indicating potential confidence in several predicted PFAS interactions while pLDDT scores were poor (67.5–54.1), indicating low confidence in PFAS location within the predicted structures. Thus, only minimal interpretation is advised. Scores for ANF were 0.86 and 88.1 for iPTM and pLDDT, respectively. However, it is unclear whether the presence of CYP1A1 with ANF (4I8V) in the AlphaFold training data influenced this value. AF3 predicted higher confidence in PFAS structures with AHR, with iPTMs ranging in between 0.85–0.92 indicating likely correct predictions and pLDDT scores from 74.8–86.56, indicating decent to good confidence with PFAS position. No trend regarding chain length or moiety was observed in these scores (Supp. Table S5). AF3 structures for PFAS with both CYP1A1 and AHR aligned with binding poses generated by Vina. PFOS particularly was predicted by both tools to make close contact between its ω–1 carbon and the CYP1A1 heme iron (3.0 Å and 4.3 Å by AutoDock and AlphaFold) (Fig. 1, Supp. Fig. S4A). AutoDock also predicted an interaction between the headgroup of PFOS and the heme iron. These two poses have been indicative of inhibition of CYPs in previous studies (Hvizdak et al., 2023). We have observed similar poses with zebrafish Cyp1a, Cyp1b1, and Cyp1c1 predicted structures docked with PFOS (Supp. Fig. S7). AlphaFold and Autodock predicted PFAS would make an interaction with AHR where the headgroup was predicted to be in proximity to Ser 365 (AF3 and Vina) and Glu 383 (Vina) (Supp. Fig. S4C–F); these residues make interactions with select ligands which have partially charged regions (Dai et al., 2022; Diao et al., 2025).

Recent studies highlighted in vitro inhibition CYP enzymes by PFAS (Hvizdak et al., 2023; Solan and Lavado, 2023), yet CYP1A1 remained untested. The in vitro results showed inhibition of human CYP1A1 by PFOS among the C8 and C6 PFAS tested in this study. Previously, PFOS and PFHxS have inhibited CYP1A2 at comparable concentrations to those tested here (Solan and Lavado, 2023). Although similar in sequences, these CYPs have key differences in substrates and binding pocket geometries. Our in vitro findings for the six PFAS studied are in agreement with Solan and Lavado, 2023, where they showed that both C8 chain length and a sulfonate moiety were more potent with lower IC50 concentration with CYPs as compared to other PFAS.

We observed severe toxicological outcomes in co-exposures of PFOS and βNF not seen in individual exposures. In larval zebrafish, co-exposure to PFOS (16 μM) was found to reduce EROD activity by as much as 56% compared to individual exposure to βNF. This was concomitant with significant increases in pericardial area, and incidences of spinal deformities, craniofacial deformities, and yolk sac edema, characteristics of “blue sac” disease in fish (Spitsbergen et al., 1991). Notably, no significant change in morphology compared to control was seen in any of the concentrations tested individually, save fish length with βNF 183.6 nM in one experiment (p = 0.025), but not the other (p = 0.115). Previous work has demonstrated co-exposures of AHR agonists and CYP1A inhibitors can result in similar outcomes, such as co-exposures to ANF and βNF (Timme-Laragy et al., 2007), polychlorinated biphenyl-11 (PCB-11) and βNF (Roy et al., 2019) or B[a]P (Roy et al., 2020), and others (Wassenberg and Di Giulio, 2004; Billiard et al., 2006; Jayasundara et al., 2015).

AHR dysregulation during development can cause deformities (Lanham et al., 2014) especially of the heart (Ma et al., 2024). Furthermore, toxicants that interact with the AHR pathway potentially inhibit CYP1 family enzymatic activity, which uncouples redox reactions from monooxygenase activity and can release reactive oxygen species (De Matteis et al., 2012; Veith and Moorthy, 2018). While impairment of Cyp1a activity causes deformity, reduction of cyp1a expression can be similarly toxic. Morpholino knockdown of cyp1a with AHR agonist exposures caused deformities similar to some AHR agonist and CYP1A inhibitor co-exposures (Carney et al., 2004; Scott et al., 2011; Brown et al., 2015; Wincent et al., 2016; Endirlik et al., 2023). However, co-exposure toxicity may also be attributable to less metabolism of toxic AHR agonists, especially since AHR agonists that cannot be metabolized do not seem to have their toxicity exacerbated by Cyp1a inhibitors (e.g., PCB-11 and PCB-126 co-exposures) (Roy et al., 2020). Taken together, these findings show multiple ways in which the impairment of Cyp1a-like activity by PFOS could cause morphological deformities.

It was found PFOS affected the expression of cyp1a, suggesting that PFOS interacts with the AHR pathway upstream of Cyp1a enzymatic activity. Changes in cyp1a expression were also observed previously in βNF and PCB-11 co-exposure, but not with PCB-126 and PCB-11 co-exposure; thus, the changes in cyp1a expression observed here might not be seen with other ligands. The embryotoxicity and decrease in EROD fluorescence seen in co-exposures may be attributed to inhibition of Cyp1a by PFOS or changes in cyp1a expression and thereby reduced enzyme abundance—either phenomenon leading to AHR pathway dysregulation (Brown et al., 2015). Regardless, tested PFOS co-exposures dysregulate AHR and cause severe deformities during development. Therefore, mixtures of PFOS and AHR agonists (e.g., PAHs) should be tested to assess if their toxicities are underestimated.

Several pathways affected by PFAS crosstalk with AHR. PFAS interact with TLRs (Wang et al., 2024) and with NF-kB signaling (Han et al., 2018). NF-kB signaling allows AHR to bind the RelA and RelB subunit of NF-kB (Vogel et al., 2007; Vogel et al., 2014). PFAS may be PXR (Lai, 2021) or FXR (Wasel et al., 2022) agonists, but other studies have not seen their transactivation by PFAS in vitro (Behr et al., 2020a) so this interaction is unclear (Behr et al., 2020b). Activated PXR has been shown to bind activated AHR, and cyp1a expression (Cui et al., 2017). Activated FXR increases expression of nr0b2a, and the corresponding protein SHP can bind AHR (Klinge et al., 2001). HIF signaling is affected by PFOS exposure (Olufsen et al., 2014; Xue et al., 2025), potentially via MAPK (Lin et al., 2022). Activation of hypoxia signaling pathways inhibits AHR activation by competing for ARNT (also called HIF1 Beta) (Matson et al., 2008). PFAS are widely known to be PPAR agonists, and increase cellular ROS (Sant et al., 2017). High ROS conditions can prevent AHR/ARNT from binding DNA (Ireland et al., 1995; Xu et al., 1998). Expression of cyp1a can be induced by PPAR alpha activation via a putative response element on the promoter (Seree et al., 2004). Furthermore, AHR controls fgf21 expression which partially controls ppara expression (Girer et al., 2016; Girer et al., 2019). These interactions make attributing changes in cyp1a expression to a specific interaction difficult. No significant evidence of PFOS mediating AHR via FXR or PXR was seen in RT-qPCR findings (Fig. S5). Further investigation is needed to fully understand to what degree pathway crosstalk from PFAS interactions affect cyp1a expression.

Limitations:

We observed Cyp1a-like activity was decreased in vivo; however, we could not attribute this decrease to direct catalytic inhibition or indirect transcriptional downregulation; lower cyp1a exacerbated deformities in other studies (Brown et al., 2015). Furthermore, we did not determine if the other two zebrafish Cyp1 enzymes, Cyp1b1 and Cyp1c1, were inhibited, if they compensated for Cyp1a inhibition, or were affected in their expression by exposures. A more complete understanding of activity of individual Cyp1 enzymes would elucidate if Cyp1 enzymes maintain activity. Moreover, we did not determine whether Cyp inhibition increased ROS concentrations. We did not determine if Cyp inhibition and AHR dysregulation exacerbated toxicity by affecting osmoregulation in zebrafish, as observed previously (Wiegand et al., 2022). Some pathways which cross-talk with AHR and have been shown to be affected by PFAS exposure were not studied here, particularly the NF-kB and HIF pathways.

5). Conclusions

The present study highlights the importance of conducting mixture assessments of environmental toxicants across classes of chemicals. Understanding molecular interactions between PFAS and AHR could determine which PFAS are likely to exacerbate toxicity. This work also highlights the utility of the zebrafish embryo model to understand developmental toxicity of mixtures. A complimentary integration of multiple approaches may be best to address molecular and toxicological endpoints. Additional work is needed to understand the full effects of PFAS exposure on AHR-mediated toxicity.

• Complimentary approaches support PFOS interactions with AHR pathway

• PFOS inhibits CYP1A1 in vitro and has good predicted affinity for CYP1A1 in silico

• Reductions in zebrafish Cyp1a activity correlated with morphological deformity

• PFOS can dysregulate AHR; PFAS & PAH mixture toxicity may be underestimated