Background Per- and Polyfluoroalkyl Substances (PFAS) in Laboratory Fish Diet: Implications for Zebrafish Toxicological Studies

Abstract

Current attention is focused on determining the potential for per- and polyfluoroalkyl substances (PFAS) to adversely impact human health. Zebrafish are a popular biological model because they share early development pathways with humans. A dietary exposure paradigm is growing in popularity in the zebrafish model because the outcomes often translate to humans. To create a diet of known composition, it is crucial to understand background PFAS levels present in zebrafish diet. Background PFAS, if present, potentially confounds interpretation of toxicological data. To date, no studies document the PFAS background levels in laboratory fish diet and there is only limited information on some pet foods. The objective of this study was to develop and validate an analytical method for up to 50 target PFAS in high lipid and protein content laboratory fish diets and pet foods. Long-chain perfluoroalkyl carboxylic acids (C9-C13) and perfluorooctane sulfonate (PFOS) were quantified in 11 out of 16 laboratory fish diets and in three out of five pet fish foods. Foods for pet birds, lizards, and dogs were below the limit of detection for all PFAS. In two of the laboratory fish diets, PFOS concentrations were greater than 1.3 ng/g and the total PFAS for the three laboratory fish diets exceeded 1.0 ng/g. Hundreds of biomedical laboratories across the world utilize these commercial laboratory fish diets, and these results indicate that numerous zebrafish colonies may be inadvertently receiving significant dietary PFAS exposures. In light of this new information, it is critical to design PFAS studies with appropriate controls with measured background PFAS concentrations in the diet and to urge caution when interpreting the results.

Graphical Abstract

1. Introduction

Per- and polyfluoroalkyl substances (PFAS) are utilized for various applications such as electroplating (Chaplin, 2019), aqueous film forming foams (Backe et al., 2013; Barzen-Hanson et al., 2017), textiles (Heydebreck et al., 2016), and food packaging (Curtzwiler et al., 2021; Schaider et al., 2017). As a result, PFAS are detected ubiquitously in a number of environmental sources including landfill leachate (Hepburn et al., 2019), biosolid leachate (Gallen et al., 2016), industrial wastewater (Tavasoli et al., 2021), and at sites of aqueous film forming foam discharge. Widespread exposure of humans and wildlife through various routes, including diet, have led to concern over the potential health hazards of exposure (Sunderland et al., 2019). Investigations of PFAS toxicity have primarily focused on perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), with increasing numbers of toxicological studies published annually over the past two decades(Zeng et al., 2019). Understanding the toxicology of the large number of PFAS in commerce is challenging because we do not yet understand the relevant biological targets of these chemicals. Increasingly used for this purpose, the zebrafish model is a well-established human health and environmental toxicological model given its many intrinsic advantages, such as easy xenobiotic exposure, rapid growth rate, and comprehensive sequenced genome (Horzmann and Freeman, 2018).

During toxicological investigations, exposing of zebrafish to chemicals is achieved either by aqueous exposure, tissue microinjection, or by adding test substances (e.g., PFAS) to the food (fish diet). Data are typically interpreted with the assumption that the laboratory animal diet (fish diet) does not contain background PFAS (e.g., is blank for PFAS). Any PFAS mass in the background of laboratory fish diet would effectively increase both intended and unintended PFAS exposures, and has potentially the greatest effect on interpretation of results following low concentration exposures. In a previous study in rainbow trout, dietary PFAS exposures led to uptake and assimilation into a variety of tissues, including the liver, muscle, and blood, illustrating diverse distribution within the organism (Falk et al., 2015). Previous studies on background contaminants in laboratory fish diets focused on heavy metals (McKee et al., 2008; Nachman et al., 2012; Tye et al., 2018), hormones (Matsumoto et al., 2004), and other types of non-fluorinated, persistent organic contaminants (Berntssen et al., 2010). To the best of our knowledge, background PFAS concentrations in laboratory fish diets have not been investigated, so the implications for toxicological studies utilizing zebrafish is unknown.

Pet food is also a major product purchased by consumers for their pets, yet measurements of PFAS in pet foods has received little attention. Quantification of PFAS in fecal excretions from pet dogs and cats offers direct evidence that pets are exposed to PFAS (Ma et al., 2020). To date, analysis of pet foods has not included PFAS classes other than perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkyl sulfonates (PFSAs). For example, the analysis of 11 pet (cat & dog) foods for only PFCAs and PFSAs yielded C4, C6-C8 PFCAs and no detection of any PFSAs, including PFOS (Chinthakindi et al., 2021). Suominen et al. analyzed for PFCAs and PFSAs in the two farmed fish diet samples, but only reported the detection of PFOS and the C11 PFCA (PFUdA) with summed individual concentrations less than 5 ng/g (Suominen et al., 2011).

Animal foods are challenging to analyze given their high lipid and protein content, which requires extra cleanup steps prior to instrumental analysis. To the best of our knowledge, there are only two published analytical methods that report PFAS for fish diet and pet food (Chinthakindi et al., 2021; Suominen et al., 2011). Chinthakindi et al. (2021) utilized methanol and ethyl acetate for extraction of cat and dog food with a weak anion exchange (WAX) cleanup step. Method performance was documented by spiking native PFCA and PFSA standards and stable-isotope labelled surrogates into blank pet food, which gave recoveries ranging from 83.2 to 107% and a precision ranging from 5 to 14%. The method described by Suominen et al. (2011) consisted of buffered tetrabutylammonium hydrogen sulfate solutions for digestion, followed by liquid-liquid extraction with methyl tert-butyl ether. No accuracy and precision data for actual fish diet matrices were reported. Both Chinthakindi et al. (2021) and Suominen et al. (2011) gave limits of quantitation (LOQ) that were based on solvent-based calibration standards, and thus only represent instrumental LOQs. To date, the surrogate recovery of PFAS from complex fish and pet foods has not been determined using both extracted surrogates and internal standards, which are added to the final extract.

In this study, an analytical method was developed to quantify background PFAS in 16 laboratory fish diets and 10 commercial pet foods. The accuracy and precision of the whole method in fish diet matrix was determined along with the LOQs and limits of detection (LOD). Because PFAS are considered as proteinophilic.(Alesio et al., 2022) The protein content in the laboratory fish diets were then compared to total PFAS levels to determine if a correlation existed. No packaging was analyzed for this study. To the best of our knowledge, this is the first study to determine the PFAS background in commercial laboratory fish diets.

2. Materials and Methods

2.1. Materials and chemicals

All native and stable-isotope-labeled surrogate standards were purchased from Wellington Laboratories (Ontario, Canada). Details on standards can be found in the Supporting Information (Table S1). Native target PFAS standards included PFCAs (C4–C14, C16); PFSAs (C3–C10, C12); a chlorinated PFAS (Cl-PFOS); a cyclic sulfonate (PFEtCHxS); unsubstituted perfluoroalkyl sulfonamides (C4, C6, C8 FASAs); substituted sulfonamides (MeFOSA and EtFOSA); sulfonamido acetic acids (FOSAA, MeFOSAA, EtFOSAA); n:2 telomer sulfonates (C4, C6, C8, C10); saturated n:2 telomer acids (C6, C8, C10) and n:3 (C3, C5, C7); unsaturated n:2 telomer acids (C6, C8, C10); HFPO-DA, ADONA, F53-B (C9, C11); and diPAPs N:2 (C6, C8). Methanol (>99% purity, HPLC grade) and acetonitrile ((>99% purity, HPLC grade) were all purchased from Fisher Chemical (Hampton, New Hampshire); hexanes were purchased from Omnislov (Darmstadt, Germany). ENVI-carb SPE and WAX columns (500 mg wt. bed, volume 6 ml) were purchased from Supelco Inc. (Bellefonte, Pennsylvania). Coffee grinders were purchased locally.

2.2. Sample Collection and storage

Samples from all 16 laboratory fish diets were collected by Sinnhuber Aquatic Research Laboratory staff (Table 1). Ten pet foods were purchased locally including those for fish, birds, lizard, and dog. Table 1 contains additional information on the feed utilized in this study (e.g., food type, protein content, and life stage/pet type). Laboratory fish diets were all stored at 4 °C, while all pet foods were stored at room temperature.

Table 1.

Laboratory fish diet by life stage, mice food, and pet food by pet type.

	Sample number	Life stage (laboratory fish) or pet type	Protein content (%)
Laboratory Fish Diet	1	Complete life stage	59*
	2	Complete life stage	60*
	3	Adult	55*
	4	Complete life stage	50*
	5	Adult	46
	6	Complete life stage	59
	7	Complete life stage	60
	8	Subadult	50*
	9	Adult	43*
	10	Adult	41*
	11	Complete life stage	55
	12	All life stage	60
	13	All life stage	53.6
	14	All life stage	34
	15	Early stage	28
	16	Adult	47
Pet Food	17	Fish	45
	18	Fish	44
	19	Fish	52
	20	Fish	42
	21	Fish	45*
	22	Bird	14
	23	Bird	10
	24	Bird	14*
	25	Lizard	32
	26	dog	27*

^*indicates minimum protein content

2.3. Sample Preparation and Extraction

Ten gram portions of each laboratory fish diet and pet food were put into a grinder for homogenization. A 3 g subsample of ground diet/food was put into a 50 mL polypropylene (PP) tube, spiked with 0.9 ng of 30 stable-isotope surrogate standards (Table S1), and allowed to sit for 30 min. Next, 12 mL acetonitrile (can) was added, vortexed for 30 s, sonicated 10 min at room temperature, centrifuged for 10 min (4417 × g), and then the supernatant was decanted into a new 50 mL PP tube. The extraction was repeated one more time for total of 24 mL extract. Hexanes were added (12 mL) and vortexed 30 s and the hexanes layer (top) was removed and discarded. Thirty μL of ethylene glycol was added to prevent adsorption to container walls and loss of any semi-volatile PFASand the excecanACN was evaporated under a stream of nitrogen. The sample was then reconstituted in methanol (3 mL) and passed through an ENVI-carb SPE column (500 mg) that had been preconditioned with 6 mL of 5% glacial acetic acid in methanol (v/v). The ENVI-carb SPE eluent was collected in a 15 mL PP centrifuge tube, evaporated to 30 μL then reconstituted with 240 μL methanol, 150 μL water (1.2 g NaCl dissolved in 30 mL deionized water), and 30 μL of two additional stable-isotope labeled internal standards (M2PFOA and M8PFOS). The extract was centrifuged 5 mins (10000 × g) and the top 150 μL was transferred to an autosampler vial before injection.

2.4. Accuracy, Precision, Limits of Detection and Quantification

Experiments were performed to determine the accuracy, as determined by percent recovery, and precision, as determined by percent relative standard deviation (% RSD) of the whole method in a fish diet matrix. Fish Diet 1 (n=4) was selected for the spike and recovery experiments and because it is the diet used by the authors for toxicological studies. Background concentrations were determined and used for background subtraction when computing spike recoveries. Four replicates of 3 g Fish Diet 1 were spiked with 2.25 ng of 50 native PFAS standards and 0.9 ng of 30 stable-isotope surrogate standards and the mixture was processed through the procedure as described above. The final extract was spiked with 30 μL of two additional stable-isotope labeled internal standards, M2PFOA and M8PFOS. The target PFAS were ratioed to the extracted surrogates as a measure of relative recovery, while the surrogate recovery was determined as the area counts of surrogate standards M4PFOA and MPFOS divided by their respective internal standards, M2PFOA and M8PFOS, respectively. The ratios of surrogate to internal standards were then compared to those of calibration standards as a measure of surrogate recovery.

Because Fish Diet 1 had background PFAS, additional diets were analyzed to find a blank diet. Fish Diet 4 was used to determine the whole method LOD and LOQ. A total of eight replicates were spiked with native standards to achieve a range of concentrations from 0.0075, 0.015, 0.03, 0.075, 0.15, 0.30, 0.45, to 0.75 ng/g and extracted as described above. The LODs were calculated from regression and the LOQs were obtained by multiplying the LODs by 3.3 (Vial and Jardy, 1999). For comparison purposes, WAX was also performed; the details were given in the SI, but the WAX cleanup was not selected for use for routine analysis.

2.5. Instrumental Analysis by LC-QTOF and quality control (QC)

Extracts were analyzed using an Agilent 1260 (Santa Clara, CA) liquid chromatograph coupled with a SCIEX X500R QToF system (Framingham, MA). A 100 μL extract was injected onto an Agilent C18 guard column (4.6 mm × 12.5 mm × 5 μm; P.N 820950–925) and C18 analytical column (4.6 × 75 mm × 5 μm; P.N. 959933–902). Mobile phases are 20 mM Ammonium acetate with 3% Methanol in water (A) and Methanol (B). Mobile phase A started at 99% and was held for two min, then decreased to 45% at 2.1 min and then to 1.0% at 18.5 min. The initial conditions were restored 21.1 min and then held for 6.4 min. The concentration of native PFAS were determined by internal standard calibration curve established by 1/x weighted linear regression (7 points). Calibration curves were established ranging from 200 – 100,000 ng/L except for the FASAs where the calibration curve ranged from 200 – 50,000 ng/L. More information on target PFAS matching to surrogate standards can be found in Table S1. Two calibration standards (200 and 500 ng/L) were analyzed every 10 samples and were required to be within 70 – 130% of their nominal concentration. The LC-QToF was operated under negative electrospray ionization (ESI-) mode and data were acquired in data-independent acquisition mode (e.g., SWATH). Perfluorobutanoic acid and its mass labeled surrogate were analyzed in multiple reaction monitoring high resolution (MRM-HR) mode in order to reduce background. All the MS parameters can be found in the SI 2.5.

3. Results and Discu

3.1. Initial method optimization

Although a number of organic solvents are used for the extraction of PFAS from biological samples (Chinthakindi et al., 2021; Suominen et al., 2011), ACN was selected because proteins precipitate out of solution, while PFAS stay in supernatant (Flaherty et al., 2005). Becanse ACN is not suitable for ENVI-carb cleanup (Reiner et al., 2011), solvent exchange to methanol was required prior to ENVI-carb cleanup. In order to avoid repeated cycles of acetonitrile-methanol exchange, evaporation to dryness, and to eliminate the need for supervision during solvent evaporation, 30 μL ethylene glycol was added. Initial experiments employed 0.5 g of Fish Diet 1 and ENVI-carb SPE column (250 mg); however,no background PFAS was detected so the sample mass investigated was increased to 3 g diet and 500 mg ENVI-carb were used for final evaluation as described above. For this larger sample size, an additional cleanup was needed to improve chromatographic peak shape, so a single hexane extraction was added prior to ENVI-carb cleanup step to remove lipids.

3.2. Method performance

Thirty PFAS gave good recovery (100±30%), based on their surrogate standards, with recoveries ranging between 74 to 124% and with precision (% RSD) ranging from 2.9 to 28.3% (Table S2). The classes or individual PFAS that have good recovery include all PFCAs except PFPeA for which a peak for the native was observed but not its matched surrogate. Six out of nine PFSAs gave good recovery with exceptions including PFHpS, PFDS, and PFDoS (see Table S1 for acronyms). These analytes had higher than expected recovery that was attributed to higher area counts for the native PFAS and not their surrogates (Table S2) although the native PFAS were not observed in the unspiked Fish Diet 1. Other classes gave good recoveries and included the FASAs, fluorotelomer sulfonates (FTS) except for the 10:2 FTS, and some but not all sulfonamide derivatives (Table S2). Recoveries for PFPeA, EtFOSA, MeFOSSA, n:3 and n:2 FTCAs, n:2 UFTCAs, and phosphate classes could not be computed because one or more peaks for the target and/or surrogate were not detected. Loss of one or more of the target or surrogate peaks indicates matrix interferences, despite the cleanup steps employed. The poorly recovered PFAS tended to cluster around specific retention times (e.g., 9.5, 12.0 min; Table S2), which may indicate co-eluting interferences that lead to signal suppression for various molecular ions. Chinthakindi et al. 2021 reported good recovery of C4-C12 PFCAs, and for only even carbon-chain length PFSAs up to PFDS, while Suominen et al,. 2011 did not report recoveries.

The surrogate recovery of MFPOA and MPFOS, based on their respective internal standards, were lower and ranged from 10 to 62% and from 16 to 70%, respectively with precision 18 and 11%, respectively (Table S3). Surrogate recovery is a criterion for evaluating the extraction efficiency (Ahmadireskety et al., 2021), but it is not reported in previous fish diet studies. Thus, to our best knowledge, this is the first time surrogate recoveries are reported for fish food samples. The LODs for the 30 PFAS that gave recoveries between 70 and 130% ranged from 0.01 to 0.20 ng/g, while LOQs ranged from 0.03 to 0.66 ng/g (Table S2).

Previous reported instrumental LOQs (e.g., without matrix) ranged from 0.025 to 0.25 ng/g for 2 g samples (Chinthakindi et al., 2021) and from 0.04 – 0.09 ng/g for 0.5 g samples (Suominen et al., 2011). The instrumental LOQ (no matrix) for the present study is 0.03 ng/g; the higher overall method LOQs are attributed to the complex nature of the fish diet samples.

Additional efforts to increase recovery are described in the SI, including the use of WAX SPE as an alternative cleanup for methanol extracts. However, WAX SPE did not result in an increase in performance, but rather gave poor recovery for more PFAS. Therefore, WAX cleanup is not recommended.

3.3. Laboratory fish diets

Members of three classes of PFAS were quantified in 11 out of 16 laboratory fish diets, including PFCAs, PFSAs, and FASAs (Table 2). The comprehensive data set is shown in Table S5. Homologous series of long-chain PFCAs (C9-C13) were quantified in four laboratory fish diets (Nos. 1,2,8, and 10), while C7-C10 PFCAs were observed in Fish Diet 13 (Table 2). Both even- and odd-chain length PFCAs were detected, but the odd chain length PFCAs (e.g., PFNA, PFUdA and PFTrDA) gave both higher concentration and detection frequencies (Table 2). All PFCAs occurred as their linear isomer, indicating fluorotelomer origin; branched isomers were not detected (Benskin et al., 2010; Riddell et al., 2009). In the present study, only two PFSAs were detected, with PFHxS quantified at concentrations >LOQ in only laboratory Fish Diet 14 and the linear isomer of PFOS was quantified in seven out of 16 laboratory fish diets (Table 2; Table S5). Of the PFAS detected, PFOS was found at the highest levels with concentrations reaching 1.8 ng/g (Fish Diet 1) (Table 2). Although there are no published PFAS data for laboratory fish diets, Suominen reported less than 1 ng/g in one of two farmed fish diets. The C4 FASA (FBSA) was detected <LOQ in Fish Diet 1, while C8 FASA (FOSA) was quantified in Fish Diet 1 at 0.19 ng/g. To the best of our knowledge, this is the first report of any member of the FASA (perfluoroalkyl sulfonamide) class in any animal food.

Table 2.

PFAS concentrations in laboratory fish diets and pet fish foods with concentrations > LOQ. Data for all samples are presented in Table S5.

	Sample No.	PFHpA	PFOA	PFNA	PFDA	PFUdA	PFDoA	PFTrDA	PFHxS	PFOS	FOSA	Sum ng/g
Laboratory Fish Diet	1	ND	<LOQ	0.20	0.12	0.41	0.07	0.16	ND	1.8	0.19	3.0
	2	ND	<LOQ	0.08	0.06	0.22	<LOQ	0.13	<LOQ	0.40	<LOQ	0.89
	6	ND	ND	ND	ND	0.12	ND	0.23	ND	<LOQ	ND	0.35
	7	ND	ND	ND	ND	ND	ND	ND	ND	0.80	ND	0.8
	8	ND	ND	0.13	0.08	0.34	<LOQ	0.26	<LOQ	0.29	ND	1.1
	9	ND	ND	ND	<LOQ	0.12	ND	0.21	ND	<LOD	ND	0.33
	10	ND	<LOQ	0.08	0.09	0.18	0.06	0.11	<LOQ	0.34	ND	0.86
	12	ND	0.67	0.09	ND	ND	ND	ND	ND	1.3	ND	1.4
	13	0.28	0.34	0.58	0.06	<LOQ	ND	<LOQ	ND	ND	ND	0.64
	14	ND	ND	ND	ND	ND	ND	ND	0.87	<LOQ	ND	0.87
	15	ND	ND	ND	ND	ND	ND	ND	<LOQ	0.39	ND	0.39
Pet Food	17	ND	ND	ND	ND	ND	ND	ND	ND	0.27	ND	0.27
	18	ND	0.22	0.19	ND	0.17	ND	ND	ND	1.3	ND	1.66
	19	ND	ND	ND	ND	0.16	ND	ND	ND	0.32	ND	0.48

ND no target signal detected, < LOD peak is detected but concentration is less than LOD, <LOQ is less than the limit of quantification, NC not calculated because all values less than LOQ

3.4. Pet foods

Three PFCAs including C8 (PFOA), C9 (PFNA), and C11 (PFUdA) were quantified at concentrations up to 0.22 ng/g (Pet Fish Food 18) along with PFOS (1.3 ng/g) (Table S5). In addition, PFUdA also occurred in Pet Food 17 and with PFOS (1.3 ng/g) in Pet Food 19. Of the PFSAs, PFOS was the only homolog in three out of five pet fish foods (Table 2). None of the FASAs were detected in any pet fish foods. Pet foods for birds, lizard, and dog were below detection for all PFAS (Table 2, Table S5). The only other comparator data for pet foods is provided by Chinthakindi et al., 2021 who reported C4 and C6–8 PFCAs ranging from <LOQ-3.67 ng/g for 11 dog and cat foods with total PFCA concentration two times higher than the pet fish food (1.66 ng/g) in the present study. In contrast, Chinthakindi et al., 2021 did not detect any PFSAs. There are no comparator data for PFAS in foods for pet birds and lizards.

3.5. Implications for zebrafish toxicological studies

Understanding the extent to which there is background PFAS in laboratory fish diets is essential to enable appropriate interpretation of toxicological studies and to consider potential implications for zebrafish colony health. Laboratory Fish Diets 1 and 12 had at least one individual PFAS where the concentration exceeded > 1 ng/g (Table 2). Fish Food 1, 8, and 12 gave a summed PFAS concentration that exceeded 1 ng/g (Table 2). Only five of the 11 laboratory fish diets gave below detection or <LOQ for all target PFAS. Several studies have reported biomarkers of liver injury and altered liver weights and histopathology in rodents following dietary exposure to PFAS concentrations near 1 ng/g, as low as 2–20 ng/g, demonstrating the potential impact of background PFAS levels (Costello et al., 2022). Given increasing concern regarding health effects associated with PFAS exposure and a need to evaluate the toxicity of numerous chemicals in complex biological systems, a significant number of studies are being conducting using alternative models such as zebrafish.

Implications of background PFAS in laboratory zebrafish diets are wide-ranging and should be considered for a variety of study types. Future PFAS dietary exposure studies may be conducted at concentrations similar to the 1 ng/g background levels measured in the present study, as PFAS have been detected in food destined for human consumption at equivalent concentrations (Chen et al., 2018). Summed nominal exposure and background concentrations may lead to higher than expected exposures, particularly when chronic low concentration exposure is the aim of the study. For toxicological studies more broadly, toxic levels of contaminants (e.g., heavy metals, pesticides, and dioxins) in laboratory animal diets can cause altered phenotypes, gene expression changes, or chronic pathologies that may confound data interpretation, as reviewed for rodent diets and applicable for other model organisms (Mesnage et al., 2015; Pellizzon and Ricci, 2020). While PFAS have not yet been broadly considered in this context, the present study is an important contribution to our understanding of contamination in laboratory fish diets. Measured concentrations of PFOS at 1.8 ng/g and FOSA at 0.19 ng/g could be of particular concern as several studies have identified these compounds as more toxic than other PFAS, specifically in zebrafish, causing morphological effects, abnormal behavior, and delayed development in embryonic zebrafish (Dasgupta et al., 2020; Menger et al., 2020; Truong et al., 2022). Maternal transfer of bioaccumulative compounds could also result in high levels in embryos and next generations destined for use in studies (Sharpe et al., 2010). Chronic exposure to background levels of PFAS could lead to confounding chemical mixture effects during experiments investigating toxicity of other chemical classes, or even biomedical studies. Several studies have demonstrated a variety of mixture interactions within the PFAS chemical class, as well as synergistic interactions of mixtures between PFAS and other chemical classes, such as polychlorinated biphenyls (PCBs) and nanoparticles (Blanc et al., 2017; Du et al., 2016; Ojo et al., 2021). Consideration of unintended mixture interactions resulting from background contaminants in laboratory fish diets may be an important step for data interpretation.

Millions of zebrafish are fed commercial diets to maintain large research colonies and the diets are fed across generations. Owing to the bioaccumulative potential of some PFAS, levels of PFAS in zebrafish may be higher than measured in the diet, as demonstrated by Falk et al., in a dietary exposure of rainbow trout for PFHxS, PFOS, and PFNA (Falk et al., 2015). If the body burdens of PFAS increase, zebrafish colony health could be impacted. In addition to direct dietary exposure via consumption of diet containing background PFAS, it is also possible that zebrafish may be exposed aqueously to PFAS leached from the diet. Many zebrafish facilities utilize closed recirculating water systems; depending on the filtration system employed, PFAS, particularly shorter-chain compounds that are more hydrophilic, may remain in circulation and contribute to colony exposure (Barton et al., 2016; Li et al., 2020). Future studies should include determination of background PFAS in the diet and zebrafish tissues in facilities that rely on commercial diets prior to undertaking toxicological or biomedical studies.

3.6. Study Limitations

The origins of PFAS in the laboratory fish diet is unknown. Descriptions accompany the laboratory fish diets indicate that laboratory fish diet is based on fish protein sources. Because PFAS are considered proteinophilic (Alesio et al., 2022), linear regression was used to determine if there was a significant correlation between laboratory fish diet protein content (%) and total PFAS concentration (ng/g); however no correlation was found (Figure 1). Chinthakindi et al., 2021 reported PFAS in pet food packaging; thus, packaging cannot be eliminated as the source of the PFAS found in the laboratory fish diet.

Figure 1.

Protein content (%) and total PFAS concentration (ng/g).

The measurement of total fluorine was out of the scope of study. Analyses by LC-QToF were conducted only in negative mode for target PFAS; however, additional PFAS may be present such that the PFAS concentrations reported are conservative (e.g., minimum) estimates. Chinthakindi et al., 2021 detected the presence of PFAS precursors by the total oxidizable precursor (TOP) assay. After performing the TOP assay, their total PFCA concentration increased from 3.67 to 14.3 ng/g in pet food, which indicates up to a factor of 4 greater PFAS precursors in dog and cat foods. However, the TOP assay was not performed on the fish diets nor on the pet foods in this study. Limitations of the present study also include the fact that while 30 target PFAS were quantitively recovered (70–130%), 20 PFAS were not well recovered at < 70%. Surrogate recoveries for these 20 PFAS were also < 70% and indicates that for improved detection limits, further work is needed to optimize the analytical method to quantitatively recovery all 50 target PFAS.

4. Conclusions

Overall, in this study, an analytical method was developed and validated with good performance (100±30%) for 30 out of 50 native PFAS for laboratory fish diets and pet foods. More method development is required for the remaining 20 target PFAS that gave recoveries outside the 70 – 130% recovery range. The majority of the laboratory fish diets had background PFAS concentrations comprised of three classes, including PFCAs, PFSAs, and FASAs, while only PFCAs and PFSAs were detected in only pet fish food. Considering the increased evidence that chronic PFAS exposures are associated with a plethora of adverse biological effects, such as liver, kidney, and thyroid disease, lipid dysregulation, and immunotoxicity(Fenton et al., 2021), zebrafish diets should be selected such that background PFAS contribution to total PFAS concentrations are known when conducting chemical exposures in zebrafish for toxicological studies. Furthermore, future studies are needed to find the sources of the PFAS in each diet. The implications of background PFAS concentrations in laboratory fish diets is wide-ranging and important when considering potential effects on zebrafish colony health across generations, and when conducting and interpreting data for toxicological and biomedical studies.

Highlights.

• An analytical method was developed for 30 PFAS in laboratory fish diets and pet foods.

• Intermediate- and long-chain length (C9-C13) homologs dominated the PFCA class.

• Perfluorooctane sulfonate (PFOS) occurred at greatest concentration and frequency.

• Perfluoroalkyl sulfonamide (FOSA) was quantified in one laboratory fish diet.

• No PFAS were detected in foods for pet birds, lizards, and dogs.