Toxicokinetic Profiling and Developmental Toxicity of Long-Chain Per- and Polyfluoroalkyl Substances in Zebrafish

Abstract

Perfluorooctanesulfonate (PFOS), perfluorodecanoic acid (PFDA), and perfluoroundecanoic acid (PFUnA) are three legacy long-chain per-/polyfluoroalkyl substances (PFAS). Although phased out of production, recent studies indicate substantial exposure can occur through consumption of contaminated freshwater fish, with minimal information known on the potential for developmental toxicity, especially for PFDA and PFUnA. In this study, zebrafish were used to characterize the toxicokinetics of these three PFAS as a model for developmental neurotoxicity, specifically behavior and dopamine (DA) dysregulation. Mass spectrometer quantification of water and tissue samples revealed that all three PFAS were rapidly absorbed during the early developmental window and bioconcentrated in tissue to a substantial degree. The visual motor response assay revealed altered behavioral patterns in the larvae, especially for PFDA and PFUnA. Enzyme-linked immunosorbent assay (ELISA) quantification of DA indicated these compounds induced elevations in total DA at nearly every exposure concentration for each compound down to the lowest test concentration used in this study at 0.004 ppb (ppb, μg/L) (equivalent to 4 ppb, US EPA drinking water regulation for PFOS). These findings confirm that PFOS, PFDA, and PFUnA exhibit a substantial capacity for bioaccumulation and demonstrate a need for further research into how developmental exposure impacts neurological health.

Graphical Abstract

INTRODUCTION

Per- and polyfluoroalkyl substances (PFAS) make up a class of persistent organic pollutants (POPs) that are incredibly widespread in the environment and in the tissues of organisms across the globe. PFAS are extremely resilient to degradation and exhibit a high capacity for bioaccumulation. Long-chain PFAS (6-carbon chain length or longer) are known to pose substantial health risks, and as such were voluntarily phased out of production in the early 2000s by all U.S. manufacturers, notably DuPont and 3 M.¹ Since these long-chain PFAS are no longer produced, much research on human health impacts has focused on the shorter-chain replacements still in production. Unfortunately, a recent 2023 study found that many legacy long-chain PFAS are still present to a substantial degree in the tissue of freshwater fish across the United States, particularly in fish from the Great Lakes.² These fish are regularly consumed by many Americans, and especially by certain immigrant angler communities. This poses a major human exposure vector to these legacy PFAS that had not been previously considered. For example, a prior study found that the blood serum of licensed anglers and Burmese immigrants in Western New York State contained roughly 2 and 6 times the U.S. population average of perfluorooctanesulfonate (PFOS, C8), respectively.³ In 2024, the U.S. EPA set the first enforceable maximum contaminant level (MCL) for PFOS and perfluorooctanoic acid (PFOA, 8C) at 4.0 ppt (ppt, ng/L) in drinking water. It was found in 2023 that eating one 8-oz fillet of freshwater fish at the median PFOS contamination level would increase a person’s serum PFOS level by 0.92 ng/mL, the equivalent exposure of spending one month drinking water contaminated at 48 ppt.² The three most abundant long-chain PFAS found in freshwater fish across the U.S. EPA’s Great Lakes and streams surveys were PFOS (74.2%), perfluoroundecanoic acid (PFUnA, C11) (9.6%), and perfluorodecanoic acid (PFDA, C10) (6.7%), and as such, these were the three compounds selected for this study. These three substances belong to the per-/polyfluoroalkyl acid (PFAA) subset of PFAS, defined as having an acidic headgroup. This structural similarity plays a role in their toxicokinetic and toxicodynamic behaviors, as they closely resemble the physical structure of fatty acids that exist naturally in our bodies. This allows them to bind with transporter and receptor proteins, in part enabling them to cross biological barriers and disrupt biochemical processes, causing harm and alterations.⁴ Various PFAS have been demonstrated to cross the placental barrier, blood-brain barrier (BBB), and blood-cerebrospinal fluid barrier in humans.^5-7

PFAS exposure is considered by the U.S. EPA to elicit reproductive effects, developmental delays and defects, increased risk of certain cancers, immune suppression and reduced vaccine response, hormone interference, and increased cholesterol and obesity.⁸ Long-chain PFAS have been shown to be capable of bypassing the BBB in a number of animals and accumulate in the brain, and as such, exposure is associated with a number of neurological effects.⁹ These impacts include neurodevelopmental delays and certain behavioral outcomes such as ADHD and overall executive function.^10-12 Research is limited on PFDA and PFUnA, but one known mechanism by which PFOS and PFOA act on the brain is by disrupting the dopaminergic system. In humans, PFOS has been shown to modulate the activity of dopamine (DA) receptors D1 and D2 (DRD1, DRD2), and PFOA exposure has been shown to detrimentally affect the maturation of DA neurons.^9,13

Zebrafish (Danio rerio) have been used in the research of developmental biology, particularly neural development, as well as vertebrate genetics since the 1950s, and have grown rapidly in popularity since the 1990s until the present.¹⁴ Zebrafish offer a number of advantages, such as small size, high fecundity, external fertilization, and externally visible organs.¹⁵ The zebrafish is now well-established and widely accepted as a reliable model, specifically in the study of toxicology. Additionally, zebrafish have a fully sequenced genome, which contains orthologs of 70% of all human genes, and a number of mutant strains exist for the study of specific diseases and genetic conditions.^16-18 Specifically, the catecholamine cascade is highly conserved between zebrafish and humans, which is pivotal to the dopaminergic component of this study.¹⁹ In this study, AB strain wild-type zebrafish were used because they are generally healthy and lacking in lethal mutations and are readily available in the U.S.²⁰ AB strain zebrafish are also among the most commonly used in all toxicity studies and, as such, are one of the more well-characterized strains, allowing ready comparisons among studies and more direct interpretation of results.²¹ The AB strain is also very genetically diverse, which helps to strengthen major findings in toxicity studies and is more representative of human genetic diversity.²²

The central hypothesis of this investigation is that developmental exposure to low concentrations of long-chain PFAS [i.e., down to 0.004 ppb (μg/L); equivalent to 4 ppt the U.S. EPA regulatory concentration for PFOS in drinking water) is associated with a detrimental neurotoxic effect in zebrafish, particularly in the dopaminergic system. Zebrafish were exposed during embryogenesis from 1 to 72 h postfertilization (hpf) to six concentrations each of PFOS, PFDA, and PFUnA and rinsed of the exposure solution. Some fish were then allowed to develop until 120 hpf. At each time point (i.e., 0, 72, and 120 hpf), samples of whole-body fish tissue and of exposure solution were collected and analyzed via mass spectrometry to quantify the content of each of the three PFAS. At 72 hpf, whole-body DA concentrations were measured, and at 120 hpf, a visual motor response assay assessed behavioral abnormalities to determine if changes in behavior were associated with alterations in DA regulation.

MATERIALS AND METHODS

Chemical Treatments for Zebrafish Assays.

PFOS, as a potassium salt (PFOS-K), (CAS# 2795-39-3, ≥98.0% purity, Sigma-Aldrich, St. Louis, MO), PFDA (CAS# 335-76-2, 98% purity, Sigma-Aldrich, St. Louis, MO), and PFUnA (CAS# 2058-94-8, 95% purity, Sigma-Aldrich, St. Louis, MO) were used to generate PFAS exposure solutions in this study. Near a neutral pH, the potassium salt of PFOS dissociates completely in aqueous solutions, producing primarily the sulfonate form and, as such, is a common form used in toxicity studies.²³ The chemicals were solubilized in embryo water (reverse osmosis water adjusted to ~7.3 pH and ~550 μS and vacuum filtered) at a targeted stock concentration of 1000 ppb (μg/L), and exposure solutions were diluted from stocks. PFAS stock solutions were neutralized (pH 7.0–7.5) by the addition of 5 M sodium hydroxide when necessary.

Zebrafish Exposure.

Embryos were collected and exposed within 1 hpf in 100 × 20 mm untreated polystyrene Petri dishes containing up to 50 embryos each and 20 mL of embryo water (negative control) or one of 5 concentrations of PFAS exposure solution. One biological replicate was either blocked as one dish of 50 embryos/larvae (behavioral analysis, DA ELISA analysis) or as two paired plates of 35 embryos/larvae each, pooled (toxicokinetic analysis) when >20 mL of exposure solution was required for the analysis of one replicate. Embryos/larvae were kept in incubators at 28 °C in the same room as adult fish. Protocols were approved by the Purdue University Animal Care and Use Committee (no. 1110000088), and all fish were treated humanely with regard to the prevention and alleviation of suffering. See Supplemental Methods for more details on zebrafish husbandry.

Larval Visual Motor Response Behavior Assessment.

Larval locomotor behavior in response to visual stimuli was assessed to determine if embryonic exposure to PFOS, PFDA, or PFUnA resulted in significant alterations from the negative control groups (Figure S1). A visual motor response (VMR) assay was performed with a white light routine using a Noldus DanioVision Observation Chamber (Noldus Information Technology, Wageningen, Netherlands). For each biological replicate, 50 embryos were submerged in 20 mL of 0 (negative control), 0.004, 0.04, 0.4, 4, or 40 ppb PFOS, PFDA, or PFUnA within 1 hpf. All embryos were rinsed at 72 hpf with embryo water to terminate the exposure period. From 72 to 120 hpf, fish were incubated at 28 °C in Petri dishes containing 20 mL of embryo water only. Behavior was assessed at 120 hpf. Within each replicate, 16 individual larvae were taken from each exposure group and placed into separate wells of a 96-well plate with 500 μL of embryo water each. The plate was placed into the DanioVision Observation Chamber, which was preheated and subsequently maintained at 28 °C by the Noldus temperature control unit. After allowing the larvae to acclimate to the dark for 10 min, fish were subjected to the white light routine, consisting of 5 periods of alternating dark and light for 10 min each (dark-light-dark-light-dark) for a total of 50 min.^24,25 During the dark phases, movement was observed using infrared light, which is not visible to zebrafish, and for the light phases, the chamber was illuminated from below with a 5000 lx light.²⁶ For both the dark and light phases, larvae motion was tracked with a Basler GenICam acA 1300–60 g camera recording at 25 fps. Tracking was smoothed to a minimum distance moved profile at >0.2 mm, and light phases were controlled via the EthoVision 12 software package. All assays were performed between 11:00 AM and 2:00 PM to minimize variations from the circadian rhythm. Total distance, velocity, and total time spent moving were calculated for each trial using EthoVision 12 software. Outlier analysis was performed with a Grubb’s test in GraphPad Prism 9. A repeated measures ANOVA was performed with blocking by light phase to determine significance using SAS 9.4 (α = 0.05). Significance was determined within each light phase for each exposure condition (i.e., by treatment group and for treatment group and phase interaction). Six to ten biological replicates were completed for each PFAS. Behavior data are reported with the F statistic expressed as F((df of variable), df (error)) = F value and the p-value. Data are presented as mean ± standard deviation.

Assessment of Dopamine Concentration in Larval Tissue.

Whole-body tissue DA concentration was assessed via ELISA to elucidate developmental deviations in the dopaminergic pathway (Figure S2). For each biological replicate, 50 embryos were submerged in 20 mL of 0 (negative control), 0.004, 0.04, 0.4, 4, or 40 ppb of PFOS, PFDA, or PFUnA within 1 hpf as described above. At 72 hpf, larvae were rinsed with embryo water and euthanized by storing at 4 °C for 2 h or −20 °C for 30 min following AVMA guidelines. From each exposure group, 35 embryos were collected in 1.5 mL microcentrifuge tubes, and excess water was removed. Larvae were suspended in 50 μL of 1× phosphate-buffered saline (PBS) and stored at −20 °C. Following two additional freeze–thaw cycles to degrade tissue and lyse cells, tissue samples were suspended in an additional 200 μL of 1× PBS and homogenized with a pestle. Homogenized samples were centrifuged at 5000 rcf for 5 min at 4 °C, and the supernatant was removed and stored at −80 °C. The day of the assay, samples were thawed and again centrifuged at 5000 rcf for 5 min at 4 °C in order to pellet any remaining particulates out of solution. For each sample, protein concentration was measured at 280 nm using a NanoDrop Microvolume Spectrophotometer, and subsamples of each sample were diluted accordingly with 1× PBS to yield 200 μL each, normalized to 3 mg/mL of protein. The assay was performed using a Cusabio fish DA ELISA kit (CSB-eq 027496FI; Cusabio, Houston, TX, USA) in accordance with the manufacturer’s protocol.²⁷ Each sample from 4 biological replicates, as well as each standard and blank, was run in triplicate (technical replicates). The optical density (OD) of each well was measured at 450 nm by using a microplate reader. The OD value of the blank wells was averaged and subtracted from each reading, and all other technical replicate values were then averaged for analysis. A standard curve was generated in GraphPad Prism 9 using the sigmoidal, four-parameter logistic (4-PL), “X is concentration” curve fit option, and DA concentration values calculated from the fit. Two ELISA plates were completed for each PFAS to total 8 biological replicates. Outlier analysis was performed with GraphPad Prism 9. An ANOVA was performed using SAS 9.4 (α = 0.05) with a post hoc least significant difference test when a significant ANOVA was observed. Data are presented as mean ± standard deviation and were calculated based on protein concentration and per larva.

Toxicokinetic Analysis of PFAS Exposure.

Main Procedure and Sample Collection.

Water and tissue concentrations were assessed at various points following the developmental exposure. The 72 and 120 hpf end points were completed in separate experiments (Figures S3 and S4). All collection tubes used in this process were prerinsed with high-performance liquid chromatography (HPLC)-grade methanol to reduce leaching of PFAS. During the 72 hpf phase, each replicate began by collecting two plates of 35 embryos for each of the six concentrations being tested (12 plates, 420 embryos total) and exposing them in 20 mL of 0 (negative control), 0.004, 0.04, 0.4, 4, or 40 ppb of PFOS, PFDA, or PFUnA within 1 hpf. At exposure initiation, 0.5 mL samples were collected of the 1000 ppb stock (prediluted 1:10 to avoid instrument overload), 0, 0.4, 4, and 40 ppb solutions, and 30 mL samples were collected of the 0.004 and 0.04 ppb solutions. At 72 hpf, embryos were rinsed 3 times with filtered embryo water, and the initial water removed from the plate was collected, pooling the water between the two plates for each concentration. These samples were centrifuged at 1800 rcf for 5 min to pellet out solids, then 0.5 mL of supernatant was collected for the 0, 0.4, 4, and 40 ppb samples, and 30 mL was collected for the 0.004 and 0.04 ppb samples. These samples were stored at −20 °C until the time of analysis. After rinsing between each set of two plates for each exposure concentration, a total of 50 larvae were collected in a 1.5 mL microcentrifuge tube. Larvae were euthanized by storing at −20 °C for 30 min. Excess water was removed, and tissue samples were stored at −80 °C. For the 120 hpf phase, an identical procedure was followed, except that at 72 hpf, water samples were collected in an identical manner, but tissue samples were not collected (Figure S4). Larvae continued development after rinsing of the exposure solution until 120 hpf, at which point water and tissue samples were again collected.

Water Sample Preparation—Direct Dilution.

Water samples were analyzed either by direct dilution or solid phase extraction (SPE), depending on PFAS concentration. For PFAS nominal concentrations >0.04 ppb, frozen samples were thawed on the day of dilution. Samples were spiked with 40 μL of PFAS internal standard master mix solution (IS) in MeOH at 125 ng/mL, and spike mass was recorded. Samples were diluted with 460 μL of LC-MS grade methanol, and the mass of methanol was recorded. All samples were vortexed at 3000 rpm for 5 min and then centrifuged at 13,300 rcf for 30 min. Supernatant was transferred into a 2 mL glass HPLC vial and sealed with caps containing silicone/polytetrafluoroethylene (PTFE) septa with the PTFE side oriented outward to avoid PFAS contamination. Samples were stored at 4 °C until the time of injection using liquid chromatography mass spectrometry (LC-MS). Quality control was addressed according to the U.S. EPA method 1633.²⁸ Following sample preparation, all stock solutions and water samples collected were analyzed at the environmental organic chemistry lab at Purdue University (West Lafayette, IN, USA) (see the Supporting Information in the Mass Spectrometry Method Details).

Water Sample Preparation—SPE.

Water samples below 0.4 ppb nominal concentration were subjected to the SPE procedure using a validated method.²⁹ On the day of analysis, samples were thawed and spiked with matching mass-labeled surrogates at 5 ng each and ammonium acetate for a final concentration of 2 mM. After adjusting the pH to ~7 with formic acid, samples were loaded onto a cleaned, conditioned, and equilibrated SPE cartridge (Oasis HLB 6 CC 200 mg Extraction Cartridges, Waters Corporation, Milford, MA, USA). After drying, the cartridges were eluted with methanol, and the extracts were gently evaporated with N2 and reconstituted in 1:1 v/v water:methanol to prepare for analysis using LC-MS. Quality control was addressed according to the U.S. EPA method 1633.²⁸ Samples were stored at 4 °C until the time of injection. Following sample preparation, all stock solutions and water samples collected were analyzed at the environmental organic chemistry lab at Purdue University (West Lafayette, IN, USA) (see Supplemental Mass Spectrometry Method Details).

Tissue Extraction.

Fish larvae samples were lyophilized for extraction following previously reported methods.³⁰ Briefly, 0.8 mL of 3:1 v/v tetrahydrofuran/nanopore water was added to freeze-dried samples in 1.5-mL polypropylene (PP) microcentrifuge tubes to which isotopically mass-labeled PFAS were added. Samples were extracted by vortexing (1500 rpm for 10 min) and sonicating (30 min), followed by centrifugation. The supernatant was transferred for evaporation under gentle N2 flow and reconstituted in a 1:1 v/v water:methanol solution to prepare for mass spectrometer (mass spec) analysis. After vortexing, sonicating, and centrifuging the reconstituted sample, the supernatant was transferred back into an HPLC vial for analysis by LC-MS. Quality control was addressed according to the U.S. EPA method 1633.²⁸ Samples were stored at 4 °C until the time of injection. Following sample preparation, all stock solutions and water samples collected were analyzed at the environmental organic chemistry lab at Purdue University (West Lafayette, IN, USA) (see the Supporting Information for Mass Spectrometry Method details).

Calculation of Bioconcentration Factor.

Bioconcentration factors (BCFs) were calculated using body burdens (ng/g) in tissue divided by the measured initial exposure concentration of each PFAS in the water using eq 1.

$$\begin{gathered} \text{Bioconcentration Factor(BCF)} \\ =\frac{\text{Body Burden in Tissue}\left(\text{ppb}\left(\frac{\text{ng}}{\mathrm{g}}\right)\right)}{\text{exposure Concentration in Water}\left(\text{ppb}\left(\frac{\text{ng}}{\text{mL}}\right)\right)} \end{gathered}$$ (1)

RESULTS

Behavioral Analysis.

PFOS.

VMR behavioral analysis of fish exposed to PFOS indicated a significant effect in all tested outcomes for the treatment group [distance moved [(F5, 833) = 3.50, p = 0.0039]; time spent moving [(F5, 836) = 2.18, p = 0.0447]; velocity [(F5, 833) = 3.51, p < 0.0038] and for the interaction between the treatment group and the phase (dark or light condition) [distance moved [(F20, 3332) = 3.13, p < 0.0001]; time spent moving [(F20, 3344) = 2.31, p = 0.0008]; velocity [(F20, 3332) = 3.13, p < 0.0001], but revealed limited effects on all three metrics at low concentrations (0.004–0.04 ppb) (Figure 1A-C). Specifically, hyperactivity was observed for the distance moved, time spent moving, and velocity in the 40 ppb treatment group during light phase 1 (L1) and light phase 2 (L2). Hypoactivity was observed for distance and velocity in the 0.4 ppb treatment group during dark phase 2 (D2).

Figure 1.

Visual motor response assay for zebrafish larvae at 120 hpf with an embryonic exposure (1–72 hpf) to PFOS, PFDA, or PFUnA. Graphs represent total distance moved, time spent moving, and velocity from left to right (D1: dark phase 1, L1: light phase 1, D2: dark phase 2, L2: light phase 2, D3: light phase 3). Error bars are SD, N = 10 (PFOS), N = 6 (PFDA), and N = 8 (PFUnA) with 16 subsamples per replicate to total 160, 96, or 128 fish per treatment, respectively. *p < 0.05 compared to 0 ppb treatment within each phase.

PFDA.

Similar to PFOS, VMR analysis showed a significant effect for the treatment group and for the interaction between the treatment group and phase for distance moved [treatment: [(F5, 522) = 23.25, p < 0.0001; treatment × phase: [(F20, 2088) = 3.04, p < 0.0001], time spent moving [treatment: [(F5, 541) = 2.87, p = 0.0143; treatment × phase: [(F20, 2164) = 2.39, p = 0.0005], and velocity [treatment: [(F5, 522) = 23.30, p < 0.0001; treatment × phase: [(F20, 2088) = 3.05, p < 0.0001]. PFDA-exposed fish showed hyperactivity for distance and velocity in the 40 ppb treatment group during all phases, in the 4 ppb treatment group during all phases except L2, and at all exposure concentrations during dark phase 1 (D1) (Figure 1D-F). Hyperactivity was also observed for distance moved, time spent moving, and velocity in the 40 ppb treatment group during L1 and L2. Aside from the light phase hyperactivity in the 40 ppb treatment group, no other effects were observed for the time spent moving.

PFUnA.

As with PFOS and PFDA, the VMR resulted in a significant effect for treatment group and for the interaction between treatment group and phase for distance moved [treatment: [(F5, 692) = 4.23, p = 0.0008; treatment × phase: [(F20, 2768) = 2.20, p = 0.0016], time spent moving [treatment: [(F5, 690) = 3.19, p = 0.0074; treatment × phase: [(F20, 2760) = 1.81, p = 0.0153], and velocity [treatment: [(F5, 692) = 4.23, p = 0.0008; treatment × phase: [(F20, 2768) = 2.20, p = 0.0016]. For fish exposed to PFUnA, VMR behavior analysis indicated hypoactivity for the distance moved and velocity at all concentrations except 40 ppb during all dark phases (Figure 1G-I). For time spent moving, hypoactivity was observed at various concentrations (0.004–4 ppb) during all dark phases, but never in the 40 ppb treatment group. Hyperactivity was observed for distance moved, time spent moving, and velocity in the 40 ppb treatment group during L1.

Dopamine Quantification.

For all three PFAS, elevated whole-body DA levels were observed at all exposure concentrations assessed, including the U.S. EPA’s new MCL for PFOS in drinking water (0.004 ppb; Figure 2; Figure S5). These elevations were statistically significant in all PFAS treatments, except for the 0.004 and 0.4 ppb PFDA exposure groups.

Figure 2.

Whole-body dopamine concentrations for PFOS (A), PFDA (B), and PFUnA (C) are expressed relative to the sample protein content. Error bars are SD, N = 8 of 35 pooled fish, *p < 0.05.

Toxicokinetic Analysis.

PFAS Body Burden.

Substantial bioconcentration was observed for all three compounds at all exposure levels (Figure 3). Only PFOS exhibited a marked degree of background contamination in the 0 ppb negative control group fish. For all three compounds, bioconcentration was rapid and substantial in the first 72 h during the static exposure period. This finding indicates that these long-chain PFAS are able to bypass the chorion prior to hatch, consistent with the ability to cross many biological barriers in mammals (e.g., BBB, placental barrier, blood-cerebrospinal fluid barrier, etc.). In no case was the compound substantially eliminated from the tissue between 72 and 120 hpf, following the rinsing of exposure solutions (see Supplemental Table S1 for tissue concentrations and Table S2 for mass accounting considerations).

Figure 3.

Body burden over time for PFOS (A), PFDA (B), and PFUnA (C). Error bars are SD. At 72 hpf, N = 4 (PFOS, PFUnA) or N = 3 (PFDA). At 120 hpf, N = 4 (PFOS, PFUnA) or N = 2 (PFDA). Replicates consisted of 50 pooled fish each.

BCFs varied considerably by PFAS, exposure concentration, and time point. BCF was, in most cases, higher at 120 hpf than at 72 hpf for all 3 compounds, indicating some amount of PFAS remained in the water following rinsing (Figure 4), and the PFAS were not eliminated from the fish’s bodies in the 48 h period following the rinse (Table 1). In addition, a general trend among all three PFAS was observed, with BCFs decreasing as treatment concentration increased. For all three PFAS, the highest BCF resulted from the 0.004 ppb treatment at 120 hpf, indicating a “diminishing return” on the concentration that can accumulate in tissue relative to water concentration, possibly indicating saturation behavior.

Figure 4.

Measured water concentration over time for PFOS (A), PFDA (B), and PFUnA (C) exposure solutions. Error bars are SD. At 0 and 72 h, N = 8 (PFOS, PFUnA) or N = 6 (PFDA). At 120 h, N = 3–4 (PFOS), N = 2–4 (PFDA), or N = 4 (PFUnA).

Table 1.

Calculated Bioconcentration Factors for Each Exposure Compound, Exposure Concentration, and Time Point

PFOS
nominal concentration (ppb)	0.004	0.004	0.04	0.04	0.4	0.4	4	4	40	40
time (h)	72	120	72	120	72	120	72	120	72	120
average BCF	290	3019	320	1853	224	1505	124	436	135	493
PFDA
nominal Concentration (ppb)	0.004	0.004	0.04	0.04	0.4	0.4	4	4	40	40
time (h)	72	120	72	120	72	120	72	120	72	120
average BCF	<LOD	7135	<LOD	661	129	661	205	278	339	438
PFUnA
nominal Concentration (ppb)	0.004	0.004	0.04	0.04	0.4	0.4	4	4	40	40
time (h)	72	120	72	120	72	120	72	120	72	120
average BCF	<LOD	2934	<LOD	160	205	173	119	200	221	330

Water PFAS Quantification.

PFAS concentrations in dosing solutions were lower than the theoretical target concentrations (Figure 4 and Table S3). Concentrations in water consistently decreased over time as PFAS were taken up by fish or adsorbed to the container. Rinsing of the exposure solution at 72 hpf was effective only at completely eliminating PFAS content in the water at 0.4 ppb for PFOS and at 0.004 ppb for PFDA and PFUnA (Figure 4). Otherwise, some amount of PFAS remained in the water in the Petri dishes at 120 hpf. It is important to note that this means the outcomes measured at the 120 hpf time point are representative of a continuous, low-level exposure, and the exposure condition was not fully terminated at 72 hpf. This phenomenon could be due to PFAS adsorption to the Petri dish prerinse and subsequent release into water postrinse when the concentration had dropped. This explanation would be consistent with the reduction in PFAS concentration observed in all exposure solutions relative to the target concentration. Consistent with our findings in the tissue analysis, only PFOS was present in the 0 ppb dosing solution at 0 hpf due to background contamination.

For all three PFAS, the measured concentration of the 1000 ppb stock solution, which was serially diluted to yield all exposure solutions, was lower than the target concentration, with PFUnA being the lowest at only 58.89% yield and PFDA being the highest at 77.55% yield (Table S3). Naturally, most subsequent dilutions were also then below their target concentrations. Generally, each subsequent dilution decreased in yield relative to the prior dilution, with the exception of PFOS and PFDA at 0.04 and 0.004 ppb (Table S3). In those cases, the trend reversed and diluted solutions exhibited greater yield than prior parent solutions, likely due to such low concentrations being highly sensitive to contamination. Average PFAS concentrations for all dosing solutions were well above the limit of quantification.

DISCUSSION

The results of this study demonstrated that developmental exposure to these PFAS induces behavioral alterations in zebrafish. These effects were more pronounced and consistent with exposure to PFDA or PFUnA than with PFOS, where effects even at the lowest exposure concentration (0.004 ppb nominal) were observed. This result is consistent with the commonly accepted finding that longer-chain PFAS are more harmful to health than shorter-chain PFAS.³¹ At the 40 ppb nominal exposure concentration, PFOS caused hyperactivity in the zebrafish at 120 hpf, which aligns with prior findings suggesting developmental PFOS exposure consistently results in hyperactivity in zebrafish exposed at higher concentrations (100 and 1000 ppb) and assessed at 144 hpf.³² PFUnA alone exhibited a U-shaped behavioral response curve, with low concentrations causing hypoactivity, which then trended back toward normal and (in one case) even hyperactivity at higher concentrations. This response pattern is not unprecedented, as a similar finding has been observed in the study of developmental PFBA exposure in zebrafish.³³ Additionally, a similar but inverted (inverted U-shape) biphasic behavioral response has been observed in zebrafish in response to ethanol exposure.³⁴ While this type of response is well-documented, it is not well understood. One possible explanation for this result could be due to biological compensatory mechanisms activating at lower exposures and being overcome by the toxic effect at higher exposures. Moreover, developmental exposure to PFDA and PFUnA elicited behavioral effects below the U.S. EPA’s MCL for PFOS and PFOA (0.004 ppb nominal). Our findings demonstrate that implementation of such limits may need to be considered for other long-chain PFAS that are potentially even more harmful. This result also implies the current limit for PFOS and PFOA could warrant further study and revision, especially considering the exposure window in this case was only 72 h compared to the lifetime of exposure intended with a drinking water MCL. Induction of a significant behavioral response is a primary indicator of neurotoxicity. To cause a behavioral effect, there must be some biological alteration occurring as a result of exposure, and it is important to characterize the nature of such an alteration, especially for the purpose of translating the findings to potential implications for human health. Since PFOS and PFOA are both thoroughly documented to affect the dopaminergic system across a variety of animal species, and notably in humans, DA was selected as a target for investigating the potential neurotoxicity.^13,35-37

A clear pattern of elevated DA levels across all exposures to all three PFAS was observed, and the elevation was significant at every exposure except for PFDA at nominal 0.004 and 0.4 ppb (0.002 and 0.187 ppb measured, respectively). This finding is consistent with a prior study reporting that developmental exposure to PFOS (32–3200 ppb) consistently resulted in increased DA in zebrafish assessed at 120 hpf.³⁸ The results also show that while there was a consistent elevation in DA levels across PFAS concentrations, this increase does not occur in a dose-dependent manner, but rather maintains a relatively flat increase beyond the 0.004 ppb nominal exposure concentration. Dose–response effects on DA content have not yet been properly investigated for any of these compounds in the present literature, with available studies using only 2–3 exposure concentrations. However, available data indicate that, at least for PFOS and PFOA, the effect on DA content does not follow a linear dose–response in zebrafish or northern leopard frogs.^37,38 Available data in the literature for typical DA content in zebrafish larvae is limited, but the available prior studies suggest the negative control group DA concentrations in the present study were near what should be expected, supporting the viability of the assay and the biological relevance of the results. One study using HPLC paired with electrochemical detection to measure DA reported the negative control group concentration at 120 hpf to be just under 30 ng/g (DA/protein),³⁹ which aligns almost exactly with the control group DA content at 72 hpf reported in the present study. While the dopaminergic system is likely not the only neurological pathway affected by these exposures, our findings indicate DA dysregulation is one of the mechanisms responsible for the observed behavioral effects. These results also indicate that even at exposures that do not induce significant behavioral abnormalities (i.e., low-dose PFOS), neurochemical alterations were still occurring, which is cause for concern. Research on the catecholamine cascade in zebrafish has found that the synthesis pathway for dopamine is wholly conserved between zebrafish and humans, with tyrosine hydroxylase (TH) converting precursor tyrosine into L-DOPA, which is subsequently converted by aromatic amino acid decarboxylase (AADC) to dopamine in both animals.¹⁹ Transport, storage, and receptor families and functions are also highly conserved between zebrafish and humans, so any potential impact on these systems in zebrafish will likely be reflected in humans, as well. In both species, dopamine is transported by packing into vesicles by monoamine transporter 2 and cleared from the synapse by dopamine transporter (DAT).¹⁹ The receptor families are also the same in both species, being categorized into D1-like (D1, D5) and D2-like (D2, D3, D4), with D1-like receptors functioning to stimulate adenylate cyclase and D2-like inhibiting it in both humans and zebrafish.⁴⁰ This is especially important because, as mentioned, DRD1 and DRD2 activity are already known to be modulated by PFOS exposure in humans.⁹ Downstream effects of dopamine dysregulation are also likely to be conserved between zebrafish and humans, as in both species, dopamine acts as a precursor for other catecholamine neurotransmitters.⁴⁰ Due to all of the aforementioned similarities in the catecholamine cascade between humans and zebrafish and given the data presented in this study, it is possible that developmental PFOS, PFDA, and PFUnA exposure could impact DA signaling in humans, even at low concentrations. Human studies on behavioral outcomes of PFAS exposure, such as ADHD and neurodevelopmental milestones, are largely inconclusive, often presenting opposite results.^11,12,41,42 Since it was shown in this study that neurological changes can be induced in the absence of behavioral effects, it may be important to examine neurotransmitters such as DA in conjunction with additional behavioral outcomes to come to a clear consensus, along with additional assessments on neuronal impacts.

The results from the whole-body tissue quantification of these PFAS in the zebrafish confirm that they can be absorbed through the biological barrier of the chorion, rapidly bioconcentrate, and are not efficiently eliminated. For all three PFAS, uptake was rapid during the first 72 h of development, bypassing the chorion in order to be absorbed by the fish, consistent with prior findings for developmental PFOS exposure in zebrafish.⁴³ In addition, PFAS were not eliminated to any substantial degree in the following 48 h. Body burden increased for all PFOS treatment groups from 72 to 120 hpf, indicating a steady-state was likely not achieved. In addition, while several of the lower treatment groups were not detected at 72 hpf for PFDA and PFUnA, measurable amounts of these PFAS were seen at 120 hpf, indicating that a steady state was not reached. This trend was measured in all treatment groups except for 0.4 ppb PFUnA. This observation is similar to that previously reported for PFOS.^44,45

BCFs were high in all exposure scenarios in this study except for 0.004 ppb nominal exposure to PFDA and PFUnA at 72 h, which was below the limit of detection. The highest BCF measured was 7135 in the 0.004 ppb PFDA nominal exposure at 120 hpf, while most were between 100 and 500. Prior studies found short-chain PFAS such as GenX (ammonium salt of hexafluoropropylene oxide dimer acid, C6) and PFBS (perfluorobutane sulfonate, C4) do not exhibit substantial bioconcentration, with 72-h BCFs of only 0.96–4.80 in zebrafish.⁴⁰ In contrast, prior studies on PFOS have shown tremendous bioconcentration in zebrafish, suggesting long-chain PFAS have a much higher capacity for bioconcentration than short-chain PFAS.^32,43 For example, when zebrafish were exposed to 2000 ppb PFOS, the 72-h BCF was approximately 208; based on calculations we performed using average zebrafish larvae body mass (see Supporting Information Estimation of PFOS 72-h BCF in Prior Study).⁴³ Moreover, chronic exposure to PFOS through 180 days of age in zebrafish had significantly lower BCFs for the higher concentration treatments (i.e., 20 and 100 ppb) compared to lower concentration treatments of 0.1 and 0.6 ppb.⁴⁴ These results are in line with the present study, indicating BCFs are dependent on exposure concentrations, with treatments below the 4 ppb exposure concentration having very high BCFs (in the thousands) and at higher exposure concentrations showing a “diminishing returns” trend, with all BCFs at 4 ppb or above being between 100 and 500. Overall, the findings of the present study confirm the bioaccumulative properties of PFOS, and extend this finding to PFDA and PFUnA, further supporting the idea that while short-chain PFAS do not substantially bioconcentrate, long-chain PFAS do.

Data are also limited in the current literature regarding the toxicokinetics of low exposure concentrations, but one prior study found zebrafish exposed at the time of fertilization to 100 or 1000 ppb of PFOS accumulated 21,600 and 213,500 ppb, respectively, at 144 hpf. At the highest exposure concentration in the present study, 40 ppb nominal (22–31 ppb measured), the larvae accumulated 11,026, 13,661, and 7444 ppb of PFOS, PFDA, and PFUnA, respectively, (average 10,710 ppb) at 120 hpf. While the exposure concentration was lower and the exposure period slightly shorter in the present study, these numbers do seem to be reasonably in line with the prior data for PFOS. These results demonstrate the potential for harm inherent in these compounds being so ubiquitous in the environment. First, because PFAS are capable of bypassing biological barriers, they can affect developing organisms during sensitive periods during which barriers, such as a placenta or chorion, would serve to protect them. Even if their concentration in a particular exposure vector, such as drinking water, is very low, these PFAS have the potential to rapidly accumulate to harmful levels in the bodies of exposed animals. Because elimination from the body is incredibly slow for most organisms, exposure events are essentially additive over the lifetime of an organism that is repeatedly exposed, and in longer-lived animals, body burdens may become quite high even from low-level exposure.

These findings bring us back to the apparent issue that initially prompted this investigation: substantial accumulation of long-chain PFAS in the tissue of fish, which are consumed by humans. As fish are particularly susceptible to PFAS bioaccumulation through both bioconcentration from living in contaminated water and biomagnification in high-level predators, fish that are living in water contaminated at safe drinking levels can accumulate PFAS levels in their tissue, which are hazardous to consume. This is especially true in large bodies of water, such as the U.S. Great Lakes, where predatory fish species have the potential to grow larger, live longer, and eat more contaminated prey over their lifetimes. In this study, fish exposed to 0.004 ppb nominal (measured at 0.0605, 0.0023, and 0.0008 ppb for PFOS, PFDA, and PFUnA, respectively), which is considered safe for lifetime drinking, accumulated 183, 16, and 2.45 ppb of PFOS, PFDA, and PFUnA, respectively, at 120 hpf; although actual PFOS water concentration was higher than expected, and PFDA and PFUnA water concentrations were lower than expected. Those fish exposed at 40 ppb nominal (measured at 22.34, 31.18, and 22.58 ppb for PFOS, PFDA, and PFUnA, respectively) all accumulated more than 7000 ppb of each compound in the same 120-h period. This finding also highlights why it was important in this study to examine neurotoxic effects of high-dose exposures, which far exceed what could be expected from the typical human exposure route of drinking water, as they may be comparable to the vector of eating contaminated fish. This exposure source is especially true among certain immigrant angler communities, such as Burmese immigrants in New York state, who may rely on fish consumption as their primary dietary intake.³ These communities have substantially elevated body burdens of long-chain PFAS, making further study into how this may be impacting their health warranted.³

For the quantification of these PFAS in dosing solutions, interesting patterns were observed. It is important to note that for all three compounds, their actual concentrations in the 1000 ppb stock solutions from which exposure solutions were derived were substantially lower than expected (i.e., 63% for PFOS, 78% for PFDA, 59% for PFUnA). Subsequently, nearly all exposure solutions derived from these stocks were lower in concentration than expected, and the yield tended to decrease with each subsequent dilution, except for at 0.04 and 0.004 ppb nominal for PFDA and PFOS. This reversal to the trend at 0.04 ppb nominal was likely due to both inherent contamination of the facility’s water and accidental contamination. Notably, the 0.004 ppb exposure solution for PFOS contained considerably more than the nominal concentration, likely indicating background PFOS contamination in our facility’s water was above 0.004 ppb. For all mass spec analyses, blank vials were prepared using ultrapure water, and results were adjusted to the blanks to account for any contamination that may occur during the processes of SPE, sample dilution, and injection on the mass spec. Therefore, any additional contamination not accounted for by the blank must have been the result of inherent contamination of our facility’s RO water (with which embryo water and solutions were prepared), leaching from plastics and containers (Petri dishes, pipet tips), or other accidental contamination (from skin, clothing, air, etc.).

It is important to bear in mind the measured exposure concentrations when interpreting and referencing the results of this study, as they were generally lower than nominal. This seems to be a common issue in the preparation of PFAS solutions for these studies. One study aimed at designing a proper methodology for preparing solutions of POPs found that for PFAAs (perfluoroalkyl acids, the class of PFAS to which PFOS, PFDA, and PFUnA belong), the measured concentrations in prepared solutions were only 30–40% of the nominal value.⁴⁶ There are a few potential reasons why this may occur, and in future studies with these compounds, steps are taken to avoid this issue. First, it is important to consider the materials and instruments used to prepare, transfer, and store these solutions. PFAS will adsorb heavily to the vessels in which they are stored, and the degree to which this occurs will depend on the type of PFAS and the material of the container. One study found up to 50% of PFOS can be lost from an aqueous solution due to adsorption to the storage container in just 48 h.⁴⁷ Polystyrene was found to adsorb the least from a mixture of PFAS, even less than glass, though the U.S EPA recommends using polypropylene (PP) and high-density polyethylene HDPE for sampling of PFAS.^47,48 Even when using polystyrene, there will still be a significant degree of adsorption, and it is advisable to appropriately compensate for the expected loss and to verify actual concentrations before performing experiments. It was also found in the same adsorption study that long-chain PFAS adsorb more than short-chain PFAS, consistent with our findings.⁴⁷ In the present study, polystyrene Petri dishes were used for the exposures; however, all stock and dosing solutions were prepared and stored in glass containers, which could explain some of the observed losses. Another source of loss in yield in PFAS solutions can occur as a result of the surfactant effect. Many PFAS, such as PFAAs, are powerful surfactants by nature, meaning they have a tendency to adsorb at the liquid–air interface in a solution, especially at concentrations below the critical micelle concentration CMC and will concentrate into aggregates such as perimicelles, which also adhere to the liquid–air interface, both of which are induced by agitation of a solution.^49-53 Therefore, if a PFAS solution is agitated by stirring or shaking, PFAS molecules will concentrate at the liquid–air interface, making the solutions nonuniform in distribution of PFAS molecules. If an aliquot is transferred from near the center of a container following agitation in order to generate a diluted solution, this aliquot will likely contain less than the expected concentration of PFAS. This property explains the trend of decreasing yields with decreased PFAS concentration in this study, as all exposure solutions were diluted serially from one another and mixed by stirring. An effective method to avoid this is to homogenize PFAS solutions by partially submerging the solution container in an ultrasonic bath filled with water and sonicating for 10 min. A final consideration for loss of yield when preparing PFAS solutions is the material of the stir bar, if used. Most commercially available stir bars are coated with Teflon (PTFE), which, while usually a good material for mixing chemical solutions due to its heat and chemical resistance, is itself a PFAS and exhibits a high degree of sorption with other aqueous PFAS, even being explored as a filtration medium for PFAS in water at large scales.^54,55 Therefore, it is preferable to use specialty Pyrex- or glass-coated stir bars if the use of a stir bar is necessary.

Looking at the change in concentrations over time, a reduction from 0 to 72 h was typically observed, indicating PFAS was leaving the solution through uptake by the zebrafish or adsorption to the Petri dish. In all instances, the total decrease in mass of PFAS in the water was substantially greater than the total mass of PFAS accumulated in the tissue of fish developing in that water. For example, at the 4 ppb exposure concentration for PFUnA, there was an average total decrease of 35.18 ng of PFUnA in the water in a given Petri dish, while the collective larvae within that dish accumulated on average only 0.70 ng of PFUnA in total (see Supplemental Table S2). Although polystyrene Petri dishes were used to minimize adsorption, substantial losses to adsorption occurred. For PFOS, the compound was only fully eliminated from the water at 120 hpf following the rinse at 72 hpf in the 0.4 ppb exposure groups. For PFDA and PFUnA, they were completely eliminated only at 120 hpf at the 0.004 ppb exposure level. It was expected these compounds would be mostly, if not fully, eliminated by the triple rinsing process, as was observed in our earlier study with GenX and PFBS following a similar exposure and analysis paradigm.⁴⁰ Although the recommended materials were used for the exposures, given the stronger adsorption properties of the longer-chain PFAS, most treatment groups contained measurable amounts at 120 hpf. Once the water concentration decreases to 0 ppb following rinsing, the PFAS molecules that were adsorbed to the Petri dish may have reentered the aqueous phase. Another contributing factor could be excretion of the compounds by the fish; however, this would be a smaller contributing factor since the total mass taken up by the fish in all cases was substantially less than the total mass in the water that would produce these 120-h concentrations. As such, the developing zebrafish had a continuous, low-level exposure from 1 to 120 hpf. Overall, these findings support the need for requiring analytical confirmation, especially for long-chain PFAS, in these types of studies. In addition, changes observed in behavioral outcomes and DA concentrations indicate studies are needed that mechanistically link these observations to neurotoxicity, given dietary exposure risk to these longer-chain PFAS via fish consumption.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c10194.

Methods for zebrafish husbandry and mass spectrometry, along with the experimental design and analysis time points for dopamine assessment, toxicokinetic analyses, dopamine content per larva, analytical measurements for water concentration and body burden, and estimation of BCF of PFOS from prior studies (PDF)