BRAIN REGION DISTRIBUTION AND PATTERNS OF BIOACCUMULATIVE PERFLUOROALKYL CARBOXYLATES AND SULFONATES IN EAST GREENLAND POLAR BEARS (URSUS MARITIMUS)

Transcription

1 Environmental Toxicology and Chemistry, Vol. 32, No. 3, pp , 213 # 212 SETAC Printed in the USA DOI: 1.12/etc.217 BRAIN REGION DISTRIBUTION AND PATTERNS OF BIOACCUMULATIVE PERFLUOROALKYL CARBOXYLATES AND SULFONATES IN EAST GREENLAND POLAR BEARS (URSUS MARITIMUS) ALANA K. GREAVES,y ROBERT J. LETCHER,*y CHRISTIAN SONNE,z and RUNE DIETZz yecotoxicology and Wildlife Health Division, Science and Technology Branch, Environment Canada, and National Wildlife Research Centre and Department of Chemistry, Carleton University, Ottawa, Ontario, Canada zdepartment of Bioscience, Arctic Research Centre, Aarhus University, Roskilde, Denmark (Submitted 29 March 212; Returned for Revision 17 May 212; Accepted 16 October 212) Abstract The present study investigated the comparative accumulation of perfluoroalkyl acids (PFAAs) in eight brain regions of polar bears (Ursus maritimus, n ¼ 19) collected in 26 from Scoresby Sound, East Greenland. The PFAAs studied were perfluoroalkyl carboxylates (PFCAs, C 6 C 15 chain lengths) and sulfonates (C 4,C 6,C 8, and C 1 chain lengths) as well as selected precursors including perfluorooctane sulfonamide. On a wet-weight basis, blood brain barrier transport of PFAAs occurred for all brain regions, although inner regions of the brain closer to incoming blood flow (pons/medulla, thalamus, and hypothalamus) contained consistently higher PFAA concentrations compared to outer brain regions (cerebellum, striatum, and frontal, occipital, and temporal cortices). For pons/medulla, thalamus, and hypothalamus, the most concentrated PFAAs were perfluorooctane sulfonate (PFOS), ranging from 47 to 58 ng/g wet weight, and perfluorotridecanoic acid, ranging from 43 to 49 ng/g wet weight. However, PFOS and the longer-chain PFCAs (C 1 C 15 ) were significantly (p <.2) positively correlated with lipid content for all brain regions. Lipid-normalized PFOS and PFCA (C 1 C 15 ) concentrations were not significantly (p >.5) different among brain regions. The burden of the sum of PFCAs, perfluoroalkyl sulfonates, and perfluorooctane sulfonamide in the brain (average mass, 392 g) was estimated to be 46 mg. The present study demonstrates that both PFCAs and perfluoroalkyl sulfonates cross the blood brain barrier in polar bears and that wet-weight concentrations are brain region specific. Environ. Toxicol. Chem. 213;32: # 212 SETAC Keywords Perfluorinated compound Polyfluoroalkyl substance Polar bear East Greenland Brain region INTRODUCTION Perfluoroalkyl and polyfluoroalkyl substances (PFASs) have been used in a wide variety of commercial and industrial products over the last six decades including water- and oilrepellent coatings (e.g., for textiles, paper products, carpets, and food packaging), pharmaceuticals, and surfactants in cleaning products and fire-fighting foams [1]. Perfluoroalkyl carboxylates (PFCAs) and perfluoroalkyl sulfonates (PFSAs) are PFASs classified as perfluoroalkyl acids (PFAAs) that are of environmental relevance as they are highly resistant to chemical, thermal, and biological degradation. Fluorotelomer alcohols (FTOHs), fluorotelomer unsaturated carboxylic acids (FTUCAs), and perfluorosulfonamides (PFOSAs) are some major groups of PFCA or PFSA precursors that have also been detected in environmental compartments [2,3]. Over the last few decades, increasing levels of PFCA and PFSA exposure have been reported in various wildlife and fish species including those from the Arctic [2 4]. Several PFCAs and PFSAs, notably perfluorooctane sulfonate (PFOS), have been shown to be globally distributed in biota. Longer-chain (>C 6 ) PFSAs and (>C 8 ) PFCAs have been shown to generally bioaccumulate and biomagnify in aquatic food webs, resulting in top-predator species being most exposed [5]. The polar bear (Ursus maritimus) is the top predator in the Arctic marine ecosystem. Recently reported levels of various persistent organic pollutants have shown that polar bears from All Supplemental Data may be found in the online version of this article. * To whom correspondence may be addressed (robert.letcher@ec.gc.ca). Published online 27 December 212 in Wiley Online Library (wileyonlinelibrary.com). East Greenland are one of the most highly persistent organic pollutant contaminated subpopulations [2,3]. Spatial trends for bears from Alaska, Canada, and East Greenland and in livers collected in 27 and 28 indicated that the sum of all PFSAs (S-PFSA, mainly PFOS), the sum of all PFCAs (S-PFCA), and FOSA levels were highest in bears from East Greenland and southern Hudson Bay and lower but comparable among those from the Beaufort Sea, Gulf of Boothia, Lancaster/Jones Sound, Davis Strait, and Baffin Bay [6]. In general, PFOS concentrations as high as 4, ng/g wet weight have been reported in the livers of East Greenland polar bears collected in 26 [3,7]. Furthermore, from 1986 to 26, PFOS levels in East Greenland polar bear livers have increased exponentially, despite the voluntary phaseout of all C 8 compounds (e.g., PFOS, perfluorooctanoic acid [PFOA], perfluorooctane sulfonyl fluoride) by the 3M company in 21 to 22 [7]. In polar bears sampled from East Greenland in 26, PFOS levels [7] were comparable to or exceeded those of the most concentrated and lipophilic persistent organic pollutants reported in the adipose tissue of the same bears, namely, mean S-polychlorinated biphenyls (1,5 ng/g wet wt) and mean S-chlordanes (1,7 ng/g wet wt) [8]. Similarly, East Greenland polar bears sampled from 1991 to 21 had hepatic S-polychlorinated biphenyl and S-chlordane concentrations of 3, and 4,1 ng/g wet weight, respectively [3,9]. The liver is considered the major repository in the body for PFSAs, PFCAs, and possibly other PFAAs. It is believed that PFAAs are retained in the body by enterohepatic recirculation (e.g., PFOS) and by renal resorption via binding (e.g., PFOA) to organic anion transporter proteins [1,11]. This hypothesis has been supported by studies on free-ranging mammals, including polar bears, which reported no correlation of PFAA 713

2 714 Environ. Toxicol. Chem. 32, 213 A.K. Greaves et al. concentrations with extractable lipid levels in either liver or plasma samples [12 14]. The polar moiety of the PFSA and PFCA molecule preferentially binds with proteins over nonpolar molecules (e.g., free fatty acids), specifically fatty acid binding proteins, lipoproteins, and albumin, and is sequestered into protein-rich tissues such as liver and blood [15 17]. Very recently, Cassone et al. [18] reported that chicken egg embryos injected with perfluorohexane sulfonate (PFHxS) and perfluorohexanoic acid (PFHxA) had greater accumulation in proteinrich compartments (i.e., yolk sac and liver). Also, PFHxS and PFHxA were detected in the cerebral cortex of chicken embryos, suggesting that they are able to cross the blood brain barrier (BBB). Although concentrations of PFOS and other PFCAs and PFSAs in the liver of polar bears have been shown to be extremely high [3], no studies have investigated levels of any PFASs in other tissues or body compartments, including the brain or among brain regions [17,19 24]. In the brain, PFOS has been detected following laboratory animal exposure [25], demonstrating its ability to cross the BBB. Recent studies with rodents have found that neonatal exposure to PFOS or PFOA can cause disruption to the central nervous system, resulting in abnormal development of motor neurons and significant changes in gene expression [25,26]. Furthermore, in vitro PFOS and FOSA exposure in both differentiated and undifferentiated PC12 cells showed that both compounds were disruptive to DNA synthesis [27]. Given the high PFSA and PFCA exposures in polar bears and the high concern for possible effects in the brain, in the present study we examined the levels and patterns of a suite of PFCAs, PFSAs, and selected PFAS precursors in eight different brain regions of East Greenland polar bears. In addition, we estimated the whole-brain burden and evaluated distribution patterns, sex differences, correlations with lipid content, and correlations with age. Due to the selective nature of the BBB, we hypothesized that PFAA concentrations in the brain would be low and that small concentration differences between brain regions would be observed. MATERIALS AND METHODS Sample collection and age determination Brain tissue was collected from 19 East Greenland polar bears (13 males, six females) in the region of Ittoqqortoormiit/ Scoresby Sound (69 748N, 2 258W) between January and March 26 (Fig. 1). Samples were collected by Inuit hunters during their traditional regulated subsistence hunt on polar bears in East Greenland. Brain samples were dissected upon arrival in the laboratory after removal of the entire brain from the skull. Samples collected and analyzed included the pons and medulla (n ¼ 14), cerebellum (n ¼ 15), frontal cortex (n ¼ 16), occipital cortex (n ¼ 17), temporal cortex (n ¼ 15), striatum (n ¼ 11), thalamus (n ¼ 8), and hypothalamus (n ¼ 4) (Supplemental Data, Table S1, Fig. S1). The brain tissue that remained after dissection (n ¼ 15) was also collected and analyzed (Supplemental Data, Table S1). These samples were pooled with all other samples when investigating PFAS trends in the whole brain. Small internal residual amounts of blood remained in the brain capillaries. However, as this is the case for all sampled wildlife tissues, the minute contamination of residual blood is a nonissue, as is further discussed in the Results and Discussion section. Briefly, however, in a separate study that included analysis of the blood of these same bears, PFAS concentrations and patterns differed between blood and brain samples. Fig. 1. Sampling site in Scoresby Sound, East Greenland. Bear ages ranged from 3 to 19 years for males and from 4 to 15 years for females (Supplemental Data, Table S1). Age was determined by counting the annual growth layer in the cementum of the I 3 tooth after decalcification, thin layer sectioning (14 mm), and staining with toluidine blue, as previously described [8]. All samples were collected on-site less than 1 h postmortem and stored in polyethylene plastic bags. These polyethylene bags had been tested for PFASs, with none being detectable at the method limits of quantification (MLOQ; Table 1). All samples were kept frozen between the time of sampling and their arrival at the National Wildlife Research Centre (NWRC, Environment Canada). Samples were stored at Environment Canada s National Specimen Bank (NWRC) at 48C until analysis. Standards and chemicals All target compounds and internal standards (>98%) were obtained from Wellington Laboratories. The target compound standards consisted of perfluorosulfonates (PFBS, PFHxS, PFOS, PFDS), perfluorocarboxylates (PFHxA, PFHpA, PFOA, PFNA, PFDA, PFUnDA, PFDoDA, PFTrDA, PFTeDA), fluorotelomer alcohols (6:2, 8:2, 1:2 FTOH), fluorotelomer unsaturated acids (6:2, 8:2, 1:2 FTUCA), and perfluorosulfonamides (FOSA, Me-FOSA). A complete list of target compounds and their full chemical names and structures can be found in Table 1. Isotopically labeled standards consisted of 13 C-PFCAs (C 6,C 8 C 12 ), 18 O-PFHxS, 13 C-PFOS, 13 C-FTOHs (6:2, 8:2, 1:2), 13 C-FTUCAs (6:2, 8:2, 1:2), 13 C-FOSA, and 13 C-Me-FOSA (Supplemental Data, Table S2). Chemicals used for extraction were analytical-grade or higher, including formic acid (98 1%), from Sigma-Aldrich; ammonium hydroxide (28 3% w/v), from EMD Chemicals Canada; ammonium acetate, from Fisher Scientific; acetonitrile, from Caledon Laboratories; and diatomaceous earth, from VWR International. All solvents used were high-performance liquid chromatography (HPLC)-grade, including methanol, obtained from Caledon Laboratories, and Milli-Q water, treated on-site. PFAS extraction and analysis All PFAS analyses were performed at NWRC in the Letcher/ Organic Contaminants Research Laboratory (OCRL). The

3 Brain distribution of PFAAs in East Greenland polar bears Environ. Toxicol. Chem. 32, Table 1. A complete list of target compounds, including chemical structure and method detection limits, for analysis of per- and polyfluoroalkyl substances in 19 East Greenland polar bears Analyte Acronym Chain length Chemical formula Method limit of quantification (MLOQ, ng/g) Perfluorocarboxylates: PFCAs Perfluorohexanoic acid PFHxA C 6 CF 3 (CF 2 ) 4 CO 2 H.2 Perfluoroheptanoic acid PFHpA C 7 CF 3 (CF 2 ) 5 CO 2 H.2 Perfluorooctanoic acid PFOA C 8 CF 3 (CF 2 ) 6 CO 2 H.2 Perfluorononanoic acid PFNA C 9 CF 3 (CF 2 ) 7 CO 2 H.1 Perfluorodecanoic acid PFDA C 1 CF 3 (CF 2 ) 8 CO 2 H.1 Perfluoroundecanoic acid PFUnDA C 11 CF 3 (CF 2 ) 9 CO 2 H.1 Perfluorododecanoic acid PFDoDA C 12 CF 3 (CF 2 ) 1 CO 2 H.1 Perfluorotridecanoic acid PFTrDA C 13 CF 3 (CF 2 ) 11 CO 2 H.1 Perfluorotetradecanoic acid PFTeDA C 14 CF 3 (CF 2 ) 12 CO 2 H.2 Perfluoropentadecanoic acid PFPeDA C 15 CF 3 (CF 2 ) 13 CO 2 H.2 Perfluorosulfonates: PFSAs Perfluorobutane sulfonate PFBS C 4 CF 3 (CF 2 ) 3 SO 3 H.2 Perfluorohexane sulfonate PFHxS C 6 CF 3 (CF 2 ) 5 SO 3 H.1 Perfluorooctane sulfonate PFOS C 8 CF 3 (CF 2 ) 7 SO 3 H 3.1 Perfluorodecane sulfonate PFDS C 1 CF 3 (CF 2 ) 9 SO 3 H.2 Perfluorosulfonamides Perfluorooctane sulfonamide FOSA C 8 CF 3 (CF 2 ) 6 SO 2 NH 2.7 N-Methylperfluorooctane sulfonamide Me-FOSA C 8 CF 3 (CF 2 ) 7 SO 2 NH(CH 3 ).6 Fluorotelomer alcohols: FTOHs 1H,1H,2H,2H-perfluoro-octanol 6:2 FTOH C 8 CF 3 (CF 2 ) 5 (CH 2 ) 2 OH.8 1H,1H,2H,2H-perfluorodecanol 8:2 FTOH C 1 CF 3 (CF 2 ) 7 (CH 2 ) 2 OH.9 1H,1H,2H,2H-perfluorododecanol 1:2 FTOH C 12 CF 3 (CF 2 ) 9 (CH 2 ) 2 OH.5 Fluorotelomer unsaturated carboxylic acids: FTUCAs 2H-Perfluoro-2-octenoic acid 6:2 FTUCA C 8 CF 3 (CF 2 ) 4 CFCHCOOH.1 2H-Perfluoro-2-decenoic acid 8:2 FTUCA C 1 CF 3 (CF 2 ) 6 CFCHCOOH.1 2H-Perfluoro-2-dodecenoic acid 1:2 FTUCA C 12 CF 3 (CF 2 ) 8 CFCHCOOH.1 sample extraction and cleanup procedures have been previously described by Gebbink and Letcher [19]. Briefly, approximately 1 g of tissue was spiked with 1 ng of each isotopically labeled internal standard. Internal standards were used to calculate recoveries (Supplemental Data, Table S2). Internal standards consisted of an isotopically labeled version of all of the target compounds with the exception of the PFSAs (C 4,C 1 ) and PFCAs (C 7,C 13,C 14,C 15 ), for which labeled standards were not available. For the few compounds where no labeled standard was available, internal standards with the closest retention times were chosen as surrogates (Supplemental Data, Table S2): 13 C- PFHxA was the surrogate for PFBS and PFHpA, 13 C-PFUnDA was the surrogate for PFDS, and 13 C-PFDoDA was the surrogate for PFTrDA, PFTeDA, and PFPeDA. The internal standard spiked sample was homogenized with a 1 mm potassium hydroxide acetonitrile:water (8:2, v/v) solution to extract the PFASs. Cleanup of the samples was done by solid-phase extraction using Waters Oasis WAX cartridges, obtaining two fractions. The first fraction was eluted using 1 ml of methanol and contained all nonpolar compounds including FTOHs, PFOSAs, and lipids, while polar compounds stayed efficiently bound to the cartridge. Active carbon from Supelco (2 mg) was added to the first fraction to adsorb any matrixinterfering compounds (e.g., lipids) and subsequently filtered out using a centrifugal filter. The second fraction was eluted using 2 ml of 1% ammonium hydroxide in methanol and contained the acidic compounds including PFSAs, PFCAs, and FTUCAs. Full details regarding the solid-phase extraction can be found in the Supplemental Data. Analysis of the fractions was done using a Waters 2695 HPLC coupled with a Waters Quattro Ultima triple quadrupole mass spectrometer. The column used was an ACE C 18, 5 mm L 2.11 mm i.d., 3 mm particle size, from Canadian Life Science. The first fraction was ionized using atmospheric pressure photoionization with a krypton UV lamp. The injection volume was 2 ml. The source temperature and probe temperatures were kept at 15 and 258C, respectively. The mobilephase gradient was composed of HPLC-grade methanol and Milli-Q water, with a flow rate of.2 ml/min, with HPLC-grade toluene (5 ml/min) as a dopant. The second PFAS fraction used electrospray ionization in negative ion mode. The injection volume was 1 ml. The source and desolvation temperatures were kept at 12 and 358C, respectively. The mobile-phase gradient was composed of 2 mm ammonium acetate in HPLCgrade methanol and 2 mm ammonium acetate in Milli-Q water. For a complete list of monitored ion transitions, please refer to Supplemental Data, Table S2. Lipid extraction The lipid content in all samples was determined gravimetrically. The lipid extraction procedure has been described previously, although slight modifications were made, namely, replacing anhydrous sodium sulfate with diatomaceous earth as the extraction medium [28]. The tissue sample taken for PFAS analysis was not suitable to use directly for lipid extraction due to the nonvolatile nature of the solvent (1 mm potassium hydroxide in 8:2 [v/v] acetonitrile/water). Therefore, for lipid extraction approximately 1 g of a separate tissue sample was taken directly from the surrounding tissue of the original sample taken for PFAS analysis. This tissue sample was ground with approximately 8 g of diatomaceous earth. The lipids were then extracted with a Dionex ASE 2 accelerated solvent extractor using n-hexane:dicholoromethane (1:1 v/v). The extracted solvent-lipid solution was evaporated to 1 ml, and a portion of the solution was left to evaporate in an oven set to 18C overnight. The resulting mass of lipid was determined gravimetrically to determine the original lipid content in the tissue (Table 2).

4 716 Environ. Toxicol. Chem. 32, 213 A.K. Greaves et al. Table 2. Mean lipid contents and concentrations (ng/g wet wt standard error), including concentration ranges, of all detected per- and polyfluoroalkyl substances in all brain regions of 19 East Greenland polar bears collected in 26 Brain compartment Pons/medulla (n ¼ 14) Cerebellum (n ¼ 15) Frontal cortex (n ¼ 16) Occipital cortex (n ¼ 17) Temporal cortex (n ¼ 15) Striatum (n ¼ 11) Thalamus (n ¼ 8) Hypothalamus (n ¼ 4) Whole brain, average b (n ¼ 115) Lipid content (% standard error) a PFOA Mean n.d Range n.d..86 n.d..84 n.d..91 n.d..69 n.d..67 n.d n.d n.d. n.d PFNA Mean Range PFDA Mean Range PFUnDA Mean Range PFDoDA Mean Range PFTrDA Mean Range PFTeDA Mean Range PFPeDA Mean Range S-PFCA Mean Range PFHxS Mean Range n.d n.d n.d PFOS Mean Range PFDS Mean Range n.d n.d n.d n.d n.d n.d n.d. 1.6 n.d S-PFSA Mean Range FOSA Mean Range n.d n.d. 3.7 n.d. 3.3 n.d n.d n.d n.d. 5.5 a Lipid content ¼ (g lipid/g tissue) 1%. b Whole brain average ¼ the average concentration (ng/g wet wt) across all brain samples (n ¼ 115) for each detected PFAS. PFOA ¼ perfluorooctanoic acid; PFNA ¼ perfluorononanoic acid; PFDA ¼ perfluorodecanoic acid; PFUnDA ¼ perfluoroundecanoic acid; PFDoDA ¼ perfluorododecanoic acid; PFTrDA ¼ perfluorotridecanoic acid; PFTeDA ¼ perfluorotetradecanoic acid; PFPeDA ¼ perfluoropentadecanoic acid; S-PFCA ¼ PFOA þ PFNA þ PFDA þ PFUnDA þ PFDoDA þ PFTrDA þ PFTeDA þ PFPeDA; PFHxS ¼ perfluorohexane sulfonate; PFOS ¼ perfluorooctane sulfonate; PFDS ¼ perfluorodecane sulfonate; S-PFSA ¼ PFHxS þ PFOS þ PFDS; FOSA ¼ perfluorooctane sulfonamide. n.d. ¼ not detected.

5 Brain distribution of PFAAs in East Greenland polar bears Environ. Toxicol. Chem. 32, Quality assurance and quality control The MLOQ values were based on signal to noise ratios of 1 and ranged from.1 ng/g wet weight in all tissues for PFCAs C 9 to C 13 ;.2 ng/g wet weight for PFCAs C 6 to C 8,C 14, and C 15 ; and up to 3.1 ng/g wet weight for PFOS (Table 1). Analytes were identified based on their relative chromatographic retention times to authentic reference standards and their characteristic multiple reaction monitoring transitions (Supplemental Data, Table S2). A sample blank was run for every block of 1 samples. The sample blank underwent the exact same procedure as all brain samples, except no tissue was added to the tube. Traces of PFOA (mean concentration [ standard error] of.56 ng/g wet wt.12), PFNA (.8 ng/g wet wt.2), PFDA (.66 ng/g wet wt.19), and PFUnDA (.71 ng/g wet wt.19) were consistently (>7%) found in sample blanks. Therefore, for PFOA, PFNA, PFDA, and PFUnDA, background correction was done for every block of 1 samples using the background levels established by sample blanks. For all other PFASs, background levels were below the MLOQ; thus, no background correction was necessary. A control sample was also run every 1 samples to assess the precision of replicate sample analysis. The control sample consisted of an NWRC in-house standard reference material of pooled bird egg homogenate (collected in 23) from doublecrested cormorant (Phalacrocorax auritus) from the Great Lakes. For S-PFSAs (C 6, C 8, C 1 ) good reproducibility of quantitative analysis was obtained with a relative standard deviation (RSD) of 15% (n ¼ 13). Reproducibility was similarly high for S-PFCAs (C 8 C 15 ), with a RSD of 14%. The C 4 PFSA and the C 6 and C 7 PFCAs were not quantifiable in the standard reference material. The percentage of recoveries of all the internal standards of detected compounds ranged from % ( 13 C 8 -FOSA) to % ( 13 C 2 -PFUnDA) (Supplemental Data, Table S2). All concentrations were quantified by an internal standard/isotope dilution approach, where relative response factors were generated (additional detail provided in Supplemental Data). Thus, the quantification of all applicable PFASs in the samples was inherently recoverycorrected. Perfluoropentadecanoic acid (PFPeDA) was quantified using the PFTeDA calibration curve since a PFPeDA standard was not available. At the NWRC, the OCRL regularly participates in interlaboratory quality-assurance/quality-control exercises for PFAS analysis. Quality-assurance/quality-control interlaboratory exercises for the OCRL have included (among others) annual participation in the Northern Contaminants Program. For the most recent Northern Contaminants Program III phases 5 and 6 (211 and 212) exercises, and for matrices such as National Institute of Standards and Technology standard reference material 1947 (Lake Michigan fish tissue), the compliancy percentage for the OCRL was 1% for all PFSAs, PFCAs, and FOSA under study in the present study. Statistical analysis Statistical analysis was done using Statistica 8. (StatSoft, 28). For statistical analysis, all quantifiable PFAS concentrations were corrected for extractable lipid content in the samples (i.e., transformed from a wet wt concentration to a lipid wt concentration) and subsequently log 1 -transformed. Normality was tested using a Shapiro-Wilk s W test, with a ¼.5. Prior to lipid correction, PFAS concentrations were not normally distributed. Following lipid correction and log 1 transformation, the concentrations for the majority of PFSAs and PFCAs had a normal distribution. Statistical tests were not performed for PFASs where >5% of the samples were below the MLOQ. For all other compounds, where <5% of the samples were below the MLOQ, those samples below the MLOQ were assigned a lipid-normalized concentration of the MLOQ/3. The choice of MLOQ/3 was considered adequate after concluding that randomly generated values between and MLOQ showed no statistical difference. Correlations between PFAS wet-weight concentrations and lipid content were assessed using product-moment correlative matrices, with p.5 indicating statistical significance and a Pearson coefficient (r).5 indicating a strong correlation. Multiple linear regression analysis was performed to test for confounding factors such as sex and age, with a ¼.5. RESULTS AND DISCUSSION PFAA concentration relationships with extractable lipid content Chain lengths for PFCAs of C 8 to C 15 and for PFSAs of C 6,C 8, and C 1, as well as PFOSAs, were detected above the MLOQ in all brain compartments (Table 2). All other PFASs were either below the detection limit, detected with low frequency, or detected near the detection limit (Supplemental Data, Table S3). The lipid content in all samples ranged from to % (Table 2). To date, wildlife studies have shown no correlative relationship between any PFAA concentration and extractable lipid content in the liver (polar bears, mink [Mustela vison]) or plasma (bottlenose dolphins [Tursiops truncates]) [12 14]. However, to our knowledge, such relationships had yet to be investigated in the brain until the present study. Wet-weight concentrations of PFCAs (C 1 C 15 ) were all significantly ( p.2) positively correlated with lipid content throughout the brain, with Pearson coefficient r values ranging from.29 (PFDA) to.51 (PFTeDA; Supplemental Data, Table S4). Concentrations of PFOS were also weakly correlated with lipid content ( p ¼.13). No other PFSA was correlated with lipid content. Although strong and significant lipid content [PFAA] correlations existed within the whole brain (shown in Fig. 2), only a few compounds were correlated within an individual compartment. The PFCAs (C 13 C 15 ) were strongly correlated in the pons/medulla (r ¼.53.66), C 12 C 14 PFCAs were strongly correlated in the temporal cortex (r ¼.59.66), C 13 PFCA was strongly correlated in the frontal cortex (r ¼.6), and C 14 C 15 PFCAs were strongly correlated in the cerebellum (r ¼.61 and.6, respectively; Fig. 2). A complete list of r and p values can be found in Supplemental Data, Table S4. When the pons/medulla, temporal cortex, and cerebellum were removed from the whole-brain analysis, all lipid content [PFAA] correlations either disappeared, as was the case for PFOS, PFDA, PFDoDA, and PFPeDA, or became much weaker, as was the case for PFUnDA, PFTrDA, and PFTeDA. Pearson s productmoment correlation coefficient r values ranged from.24 to.29 (Supplemental Data, Table S4). The lipid extraction method used in the present study primarily extracts nonpolar compounds or compounds with low to nonpolar character, such as triglycerides and free fatty acids. However, there are many types of fatty acids in the brain, including lipids with larger polar heads, including phospholipids, gangliosides, and sphingolipids that were likely not isolated using the present method. Therefore, the observed relationships between PFAA concentrations and extractable

6 718 Environ. Toxicol. Chem. 32, 213 A.K. Greaves et al. Log1 Σ-PFCAs, (ng/g wet wt) Log1 PFTeDA, (ng/g wet wt) Whole brain Frontal cortex Pons/medulla Cerebellum Temporal cortex % Lipid, (g lipid/g tissue)x1% Pons/medulla Cerebellum Frontal cortex Occipital cortex Temporal Cortex Striatum Thalamus Hypothalamus Fig. 2. Correlative relationships between lipid content and perfluoroalkyl acid concentrations. (A) Correlation between extractable lipid content (%) and concentration (nanograms per gram wet wt) of the sum of all perfluoroalkyl carboxylates (S-PFCAs) across the whole brain (n ¼ 115, r ¼.47). (B) Correlation between lipid content (%) and concentration (nanograms per gram wet wt) of perfluorotetradecanoic acid (PFTeDA) for the frontal cortex (n ¼ 16, r ¼.6), cerebellum (n ¼ 15, r ¼.61), pons/medulla (n ¼ 14, r ¼.65), and temporal cortex (n ¼ 15, r ¼.6). lipid content are likely between longer-chain PFCAs (C 1 C 15 ) and nonpolar free fatty acids. The major fatty acids found in the brain are often between 12 and 26 carbons long, closely resembling the longer-chain PFCAs detected in the present study [29]. Château-Degat et al. [3] observed a correlation between PFOS and high-density lipoproteins in the blood of Inuit adults from Nunavik, Canada. The present study is the first to our knowledge that has linked PFAA concentrations (and particularly PFCA concentrations) with lipid content in a tissue other than blood. To examine differences in PFAS accumulation in different regions of the brain, all concentrations (nanograms per gram wet wt) were normalized for lipid content (e.g., nanograms per gram [ng/g] lipid wt) for subsequent statistical analysis. Sex differences and age correlations Sex differences were evaluated by brain region. The cerebellum showed higher concentrations ( p.48) of PFHxS and C 1 C 13 PFCAs in males compared to females, although all other brain regions showed no differences between the sexes. Due to the limited sex differences observed, sex was not considered to be a confounding factor of the relative PFCA, PFSA, and FOSA concentrations among brain regions. As such, all bears were treated identically, in one large cohort, irrespective of sex. Age trends were also evaluated by brain region. Increasing concentrations with age (p.47) were seen in both the cerebellum and occipital cortex for the C 1 to C 13 PFCAs. All other brain regions showed no correlation between PFCA, PFSA, or FOSA concentrations and age. Due to the limited age trends observed, age was not considered to be a confounding factor of the PFAS concentrations among brain regions. The PFCAs and PFSAs have a large range of half-lives, from hours in rats and rabbits to years in humans and other larger mammals [31,32]. Although the half-lives of bioaccumulative PFCAs and PFSAs in polar bears have not been reported, these limited age correlations in the cerebellum and occipital cortex indicate halflives in the order of years for longer-chain PFCAs (C 1 C 15 ), which suggests that they accumulate in the brain faster than they can be eliminated. PFCA, PFSA, and FOSA patterns and concentrations among brain regions Perfluorooctane sulfonamide (FOSA) is the most commonly detected precursor that is degraded to PFOS [33]. It was detected in 87% of the polar bear brain samples. In all brain compartments tested, a positive, statistically significant, and strong correlation (r ¼.7) between PFOS and FOSA existed for all samples for which both PFOS and FOSA were above the MLOQ. Thus, the mean PFOS:FOSA concentration ratio was highly conserved, with a value of 25 ( 1 standard error): 1 for all brain regions combined (Fig. 3). Correlations between PFOS and FOSA were also seen for every individual brain region with the exception of the hypothalamus (n ¼ 4), most likely due to its small sample set size. The region with the strongest PFOS:FOSA concentration correlation was the thalamus (r ¼.85), although differences in concentration ratios among compartments were not statistically significant (Supplemental Data, Table S5). These highly conserved PFOS:FOSA concentration ratios between brain regions indicate that the accumulation of PFOS and FOSA is consistent across the whole brain. Among brain regions, lipid-normalized S-PFCA concentrations accounted for 66% of S-PFAA concentrations, with the remaining 34% attributable to S-PFSAs. The dominant PFAA was PFOS ( ng/g lipid wt), accounting for 93% of the S-PFSA concentrations (Supplemental Data, Table S6). Levels of PFOS accounted for 31% of S-PFAA concentrations, followed by those of PFTrDA (27%), PFUnDA (14%), and PFPeDA (9%). In the brain longer-chain (C 11 C 15 ) PFCAs dominated shorter-chain (C 9 C 11 ) PFCAs. Residual amounts of blood remained in the brain after harvesting, which may have contaminated the samples. However, we have recently Log 1 [PFOS], (ng/g wet wt) Log 1 [FOSA], (ng/g wet wt) Fig. 3. Pearson s product-moment correlation analyses between perfluorooctane sulfonate (PFOS) and perfluorooctane sulfonamide (FOSA) across the entire brain. r ¼.7. Concentrations were log 1 -transformed and not lipid-corrected. The PFOS:FOSA ratio across all brain samples was 25 ( 1):1.

7 Brain distribution of PFAAs in East Greenland polar bears Environ. Toxicol. Chem. 32, completed a follow-up study that examined the blood of these same bears (Supplemental Data, Table S7). Concentrations and patterns of PFAS differed between blood and brain samples, which strongly suggests that blood contamination in the brain sample was limited. We found that in the blood of these bears the dominant PFASs were PFOS and C 9,C 11, and C 13 PFCAs and that in the brain the longer-chain C 11,C 13, and C 15 PFCAs dominated. For the mean concentrations of longer-chain PFCAs (C 1 C 15 ; Fig. 4a), after lipid normalization there were no significant differences among the eight brain regions ( p >.5; Fig. 4b). Brain compartments with higher extractable lipid content (and thus a more nonpolar chemical environment) may provide a suitable location for the accumulation of longer-chain PFCAs (containing a long hydrophobic tail). Lipids make up approximately 5% of the brain s dry weight, making it the second most lipid-rich tissue in the body, following adipose tissue [34]. Although it is believed that the brain is able to synthesize a few nonessential fatty acids, the majority of fatty acids needed for the brain s normal activity must cross the BBB. It is important to note that longer-chain PFCAs resemble saturated fatty acids, which are constantly being shuttled across the BBB to replenish lipids in the brain [34]. It has been hypothesized that although small- and mediumchain fatty acids (12 carbons and fewer) are capable of passive diffusion across the BBB, longer-chain fatty acids often require a mechanistic mode of transport to cross the BBB [34]. Fatty acids longer than 12 carbons are less soluble and less permeable to the lipid bilayer. As a result, passive diffusion is quite slow. In contrast, fatty acid transport proteins, fatty acid binding proteins, and fatty acid translocase proteins have been shown to transport fatty acids across the cell membrane and have similarly shown a high affinity for longer-chain fatty acids (1 carbons or longer), while showing very little affinity for shorter-chain lengths [34]. Here, we hypothesized that longerchain PFCAs (C 1 C 15 ), bound to albumin in the bloodstream, undergo a similar transport mechanism as saturated fatty acids at the BBB membrane, thus explaining the correlative relationship between longer-chain PFCA concentrations and extractable lipid content. It is interesting to note that the three brain regions with both the highest PFAA concentrations and the highest lipid contents (pons/medulla, thalamus, and hypothalamus) are also the brain regions that receive the freshest supply of blood (i.e., the most oxygen- and nutrient-rich, closest to the incoming internal carotid arteries and vertebral arteries [35]). Similarly, the brain regions with the lowest PFAA concentrations and lipid contents (frontal cortex and temporal cortex) are some of the brain regions that receive the least fresh supply of blood. It is possible that the correlation between lipid content and PFAA concentrations is a reflection of the nutrient levels in the blood. Brain regions receiving the freshest supply of blood will have a slightly higher exposure to both lipids and PFAAs. These brain regions will therefore shuttle more lipids and PFAAs across the A ng/g (wet wt) Pons/medulla Cerebellum Frontal cortex Occipital cortex Temporal cortex Striatum Thalamus Hypothalamus B ng/g lipid (lipid wt) PFOA PFNA PFDA PFUnDA PFDoDA PFTrDA PFTeDA PFPeDA PFOA PFNA PFDA PFUnDA PFDoDA PFTrDA PFTeDA PFPeDA Fig. 4. Concentrations of perfluoroalkyl carboxylates throughout the eight brain regions of East Greenland polar bears, with (A) wet-weight concentrations and (B) lipid-weight concentrations. Error bars represent standard error. Before lipid correction, statistically significant differences can be seen between compartments, especially for perfluoroundecanoic acid (PFUnDA) and perfluorotridecanoic acid (PFTrDA). These differences become statistically insignificant once corrected for lipid content in the tissues, indicating that perfluoroalkyl acid brain distribution occurs in a lipid-related fashion. PFOA ¼ perfluorooctanoic acid; PFNA ¼ perfluorononanoic acid; PFDA ¼ perfluorodecanoic acid; PFDoDA ¼ perfluorododecanoic acid; PFTeDA ¼ perfluorotetradecanoic acid; PFPeDA ¼ perfluoropentadecanoic acid.

8 72 Environ. Toxicol. Chem. 32, 213 A.K. Greaves et al. Table 3. Estimated perfluoroalkyl substances (PFAS) burden for the entire polar bear brain a based on the polar bear brain masses recorded (Supplemental Data, Table S1) and using brain anatomical data as described by Kamiya and Pirlot [36] Brain region Compartment analyzed b Percentile mass composition of brain (%) c S-PFCAs (mg) S-PFSAs (mg) PFOS (mg) FOSA (mg) S-PFASs (mg) d Neocortex Frontal cortex, occipital cortex, temporal cortex Cerebellum Cerebellum Diencephalon Thalamus, hypothalamus Rhinencephalon N/A Medulla oblongata Pons/medulla Striatum Striatum Mesencephalon N/A Hippocampus N/A Olfactory bulb N/A Septum N/A Whole brain a Based on the polar bear brain masses recorded (Supplemental Data, Table S1) and using brain anatomical data as described by Kamiya and Pirlot [36]. b N/A ¼ Brain region was not analyzed in the present study. Burden estimates are based on the average of the compartments analyzed. c Kamiya and Pirlot [36]. Percentile mass ¼ (mass compartment/total brain mass) 1%. d S-PFAS ¼ S-PFCA þ S-PFSA þ FOSA. S-PFCAs ¼ sum of all perfluoroalkyl carboxylates; S-PFSAs ¼ sum of all perfluoroalkyl sulfonates; PFOS ¼ perfluorooctane sulfonate; FOSA ¼ perfluorooctane sulfonamide; S-PFASs ¼ sum of all perfluoroalkyl substances. BBB than other brain regions, explaining the brain region specific accumulation patterns observed in this study. Estimate of measured PFAS burden in the entire brain To compare with other studies, we were able to calculate a reasonable estimate of the measured PFAS burden in the whole polar bear brain based on the eight brain regions under study. The whole-brain mass was recorded for 16 of the 19 bears (Supplemental Data, Table S1), with an average weight of 392 g. Kamiya and Pirlot [36] analyzed the brain anatomy of the sun bear (Ursus malayanus) and reported the distribution of brain compartments on a percentage of mass basis. Based on the compartments analyzed in the present study, concentrations could be directly estimated for 88.79% of the total brain weight. For the remaining 11.21% of the brain, which was not analyzed in the present study, the average concentrations throughout the brain were used to extrapolate burdens. The overall burden of PFCAs, PFSA, and FOSA in the polar bear brain was estimated to be 46 mg (Table 3). This is similar to the burden (28 mg) reported in the brain of harbor seals (Phoca vitulina) from the German Bight and higher than the burden (.16 mg) recently reported for the Great Lakes herring gull (Larus argentatus) brain [17,19]. The highest PFAS burden in the polar bear brain was presently found in the neocortex (24.1 mg) due to its large mass (59.47% of the total brain mass), followed by the cerebellum (6.5 mg) and the diencephalon (3.2 mg). Comparison with PFCAs and PFSAs in the brains of other studied mammals To compare the present brain PFCA and PFSA concentrations in the polar bear brain with data reported elsewhere in tissues of other mammals, comparison on a wet-weight basis is required. The dominant PFSA among all brain compartments was PFOS, representing 91% (thalamus) to 97% (hypothalamus) of the S-PFSA concentrations. The mean PFOS levels found in the brain regions of the present polar bears ranged from to ng/g wet weight (Table 2). These brain levels were comparable to those found in recent studies on the brain of harbor porpoises (24 ng/g wet wt) off the coast of Ukraine [21] and the brain of harbor seals from the German Bight (99 ng/g wet wt) [17]. In the brains of avian species (glaucous gulls, pelicans) [22,24], PFOS concentrations have been found to be lower (<5 and 3.5 ng/g wet wt, respectively) than those in the current polar bear brain samples (Fig. 5). The overall dominance of PFOS in the brains of the present study ( ng/g wet wt) is consistent with other reports for East Greenland polar bear liver studies. However, the PFOS concentrations in the brain are roughly 1-fold lower than previously reported liver concentrations (3, 4, ng/g wet wt) for bears collected in the same geographical location and/or the same year (26) [4,7,37,38]. Within all compartments, the major PFCAs were PFTrDA (C 13 ; ng/g wet wt), PFUnDA (C 11 ; ng/g wet wt), and PFPeDA (C 15 ) (1.4.5 ng/g wet weight). The major PFCA (C 13 ) in the brain has a longer chain than what has PFOS (ng/g wet wt) Polar bear (this study) (26) Harbor porpoise ( ) [21] Harbor seal (27) [17] Glaucous gull (24) [24] Brown pelican (24) [22] Fig. 5. Recently reported mean brain perfluorooctane sulfonate (PFOS) concentrations among mammalian and avian species (nanograms per gram wet wt).

9 Brain distribution of PFAAs in East Greenland polar bears Environ. Toxicol. Chem. 32, been reported for polar bear liver, where the major PFCA was PFNA (C 9 ) [37,38]. This is a more pronounced shift than what was found in harbor seals [17], where there was a slight shift from C 9 being dominant in the liver to C 1 being dominant in the brain. The shift toward longer-chain PFCAs in the brain may be an indication of different pharmacokinetics in the brain due to the transport across the BBB. The present study is one of a select few that has shown PFCA (PFTrDA) concentrations approaching levels of PFOS in any tissue of any wildlife. In the East Greenland polar bear liver PFOS has always been the dominant PFAA, with the next most abundant PFAA (PFNA) typically in the range of 1 to 2 times lower [7,37,38]. However, the dominance of PFOS relative to PFTrDA is statistically negligible ( p >.26) in the whole brain (PFOS vs PFTrDA ng/g wet wt) as well as in each of the eight individually analyzed brain regions. This leads to the suggestion that PFCAs may cross the BBB more efficiently than PFSAs, thus explaining the extreme dominance of PFOS over PFCAs in the liver and yet their similar levels in the brain. This hypothesis is supported by the study of Ahrens et al. [17], who observed a similar trend in harbor seals. In that study, long-chain PFCAs were proportionally more present in the brain than PFOS when compared to their levels in the liver: PFCA (C 11 C 14 ) levels in the brain were 28% of those found in the liver, whereas PFOS levels in the brain were only 1% of those found in the liver [17]. It appears that PFCAs enter the brain more easily than PFOS, which suggests that transport processes across the BBB have a preference for longer-chain PFCAs relative to the liver. This is consistent with the strong correlation that was found between long-chain PFCAs and extractable lipid content as previously discussed. Odd chain-length PFCAs were present at higher concentrations across all brain regions compared to even chain-length PFCAs of comparable chain length (i.e., PFUnDA > PFDA, PFTrDA > PFDoDA, PFPeDA > PFTeDA). This trend has previously been observed in the brain of red-throated divers (Gavia stellata) from the Baltic Sea [2] as well as in the liver of polar bears and other Arctic species [2,37]. This odd even relationship seems to be indicative of PFCA sources in the Arctic. It is known that, for example, FTOHs yield both even and odd chain-length PFCAs on abiotic degradation [39]. For example, 1:2 FTOH yields mainly PFDA and PFUnDA and 8:2 FTOH yields mainly PFOA and PFNA, with small amounts of shorter-chain PFCAs detected as well [39]. The odd even relationship seen in the brain, and previously in Arctic liver samples, may indicate that PFCA exposure in Arctic biota is in part due to volatile precursor compounds such as FTOHs that are transported to the Arctic and subsequently degraded to give odd and even chain-length PFCAs. It is unlikely that there are large amounts of biotic degradation of precursors (i.e., FTOHs, FTUCAs) occurring in polar bears due to the low detection frequency of FTOHs and FTUCAs in the present samples. Possible toxicological implications The present study showed that brain tissue with a high lipid content may be more susceptible to PFAA exposure, raising concern as to the potential toxicity effects of PFAA exposure in the brain. Recent studies on PFAAs in rodents have found that neonatal exposure to PFOS or PFOA can cause disruption to the central nervous system, resulting in abnormal development of motor neurons and significant changes in gene expression, including genes responsible for calcium signaling pathways, peroxisome proliferator activated receptor signaling, cell communication, and the cell cycle [25]. Other symptoms included deranged spontaneous behavior, hyperactivity that worsened with age, changes in exploratory behavior, and reduced muscle strength in males [26]. The toxicity of these compounds has been examined in multiple studies, although the mechanisms behind these toxic effects are still poorly understood. Furthermore, exposure concentrations in these studies far surpass concentrations found in free-ranging wildlife. For example, Wang et al. [25] administered PFOS both prenatally and neonatally in albino Wistar rats. They showed that at birth (day 1) the PFOS concentration in the brain of pups was mg/g wet weight and over 84 genes were significantly affected. By day 35, concentrations were still very high ( mg/g wet wt), but only 13 genes were significantly affected. In the present study, the mean PFOS concentration in the brain was much lower (only ng/g wet wt). As a result, caution should be exercised when directly comparing in-lab toxicity studies to free-ranging wildlife studies. Adverse effect levels have been determined for PFOS in the serum of cynomolgus monkeys [4]. The blood of the present polar bears has also been analyzed (Supplemental Data, Table S6), and with that, we may be able to more easily compare how close current concentrations are to adverse effect levels. Regardless, PFAS-induced neurotoxicity in polar bears is of potential concern due to the present finding of a brain-wide presence as well as recent studies showing increasing hepatic concentrations. CONCLUSIONS The present study is the first study, to our knowledge, that has shown a strong correlation between PFAA concentrations and extractable lipid content in wildlife tissue or body compartments. These correlative results support the hypothesis that longer-chain PFCAs (C 1 C 15 ), which closely resemble free fatty acids (in terms of shape, size, lipophilicities), are transported across the BBB using the same mechanisms as free fatty acids. Because of the assumption that PFASs associate primarily with proteins as opposed to lipids, very few studies have investigated the potential relationship between PFASs and lipids in the body. Luebker et al. [16] demonstrated weak binding of PFOS to liver fatty acid binding proteins. Future work should consider examining the potential interactions of multiple PFASs, especially longer-chain PFCAs (C 1 C 15 ) with fatty acid transport proteins, fatty acid binding proteins, and fatty acid translocase proteins at the BBB. Regardless, cerebral PFAS concentrations in polar bears should continue to be monitored since increased concentrations could be an early warning sign for potential neurotoxicity effects. SUPPLEMENTAL DATA Tables S1 S7. Fig. S1. (31 KB PDF). Acknowledgement We thank the East Greenland hunters for sampling the relevant tissues for this investigation as well as J. Brønlund, who coordinated and received the sampling. S. Joensen, L. Bruun, and T. Bechshøft assisted with the age determination. E.W. Born initiated East Greenland sampling and collaborated on several other parts of our work. P. Leifsson participated in part of the brain dissection. We also thank S. Chu (Organic Contaminants Research Laboratory/Letcher Group, National Wildlife Research Center) in the initial brain sample analysis. Funding for polar bear sampling was provided by the IPY Fuller #134 Program BearHealth by KVUG, DANCEA, and the Prince Albert Foundation (to R. Dietz). Likewise, funding was provided by a NSERC Discovery Grant, the Molson Foundation, and Environment Canada s Chemical Management Plan (to R.J. Letcher).