Document 4vwxZoqBzp07epYpDn4GpZ4BQ

9 PP AR226-1531 Environmental Toxicology and Chemistry, Vol. 22, No. 1, pp. 196-204, 2003 2003 SETAC Printed in the USA 0730-7268/03 $12.00 + .00 BIOCONCENTRATION AND TISSUE DISTRIBUTION OF PERFLUORINATED ACIDS IN RAINBOW TROUT (ONCORHYNCHUS MYKISS) Jo n a t h a n W .M a r t in , * ! Sc o tt A. Ma b u r y ,J Keith R. So l o m o n ,! and De r ek C.G. Mu ir University of Guelph, Department of Environmental Biology, Bovey Building, Guelph, Ontario N1G 2W1, Canada {University of Toronto, Department of Chemisrry, 80 St. George Street, Toronto, Ontario M5S 3H6, Canada Environment Canada, National Water Research Institute, 867 Lakerhore Road, Burlington, Ontario L7R 4A6 (Received 30 January 2002; Accepted 5 July 2002) RECEIVED OPPT NCIC 2003 OCT-9 AM 9:35 Abstract--Rainbow trout (Oncorhynchus mykiss) were exposed simultaneously to a homologous series of perfluoroalkyl carboxylates and sulfonates in a flow-through system to determine compound-specific tissue distribution and nioconcentration parameters for perfluorinated acids (PFAs). In general, PFAs accumulated to the greatest extent in blood > kidney > liver > gall bladder. Carboxylates and sulfonates with perfluoroalkyl chain lengths shorter than seven and six carbons, respectively, could not be detected in most tissues and were considered to have insignificant bioconcentration factors (BCFs). For detectable PFAs, carcass BCFs increased with increasing length of the perfluoroalkyl chain, ranging from 4.0 to 23,000, based on wet weight concentrations. Carboxylate carcass BCFs increased by a factor of eight for each additional carbon in the perfluoroalkyl chain between 8 and 12 carbons, but this relationship deviated from linearity for the longest PFA tested, possibly because of decreased gill permeability. In general, half-lives (3.9-28 d) and uptake rates (0.053-1,700 L/kg/d) also increased with increasing length of the perfluoroalkyl chain in all tissues. Sulfonates had greater BCFs, half-lives, and rates of uptake than the corresponding carboxylate of equal perfluoroalkyl chain length, indicating that hydrophobicity, as predicted by the critical micelle concentration, is not the only determinant of PFA bioaccumulation potential and that the acid function must be considered. Keywords-- Fi sh I^^ru^i^(3^^natedi^cM PerOuoroocnanesutfanate uurfacnants INTRODUCTION Perfluorinated acids (PFAs) are a class of anionic fluorinated surfactants characterized by a perfluoroalkyl chain and a sulfonate or carboxylate solubilizing group. Perfluorinated acids have been used increasingly over the past 20 years be cause of their temperature and chemical stability, lipophobicity, and effectiveness as surfactants at low concentrations [1]. The total production of fluorinated surfactants (anionics, cationics, and neutrals) was 200 t in 1979 [1], whereas in 2000, the total production of one PFA, perfluorooctane sulfonate (PFOS), was nearly 3,000 t [2]. Although this production vol ume represents only a small fraction of total surfactant man ufacturing (i.e., K0.01%) [2,3], PFOS recently emerged as a global contaminant after its detection in humans and wildlife from various geographic locations. By using liquid chromatography-tandem mass spectrom etry (LC-MS-MS) [4,5], human serum was found to contain ng/ml concentrations of PFOS, perfluorohexane sulfonate (PFHxS), perfluorooctanoate (PFOA), and perfluorooctane sulfonylamide [5]. Soon thereafter, wildlife samples collected in various global locations also were observed to contain PFOS [6,7], and organisms consuming fish, such as predatory birds and mink, contained greater concentrations of PFOS than their food sources [6,8]. The wide distribution of PFAs in the en vironment, despite low production volumes, is largely a result of recalcitrance to biotic and abiotic degradation mechanisms [9]. Furthermore, atmospheric transport of PFAs may be ac * To whom correspondence may be addressed (jmartin@chem.utoronto.ca). The current address of J.W. Martin is Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario M5S 3H6, Canada. commodated through recently detected transient volatile de rivatives [10], and direct atmospheric emission of perfluoroalkyl carboxylates (C3-C 14) occurs through thermolysis of polytetrafluoroethylene [11]. Because of concerns regarding their ubiquitous distribution in humans and wildlife, as well as their environmental persistence, the main manufacturer of PFAs currently is phasing out the production of long-chain perfluorinated acids [2]. Dietary exposure of fish to PFAs does not result in bio magnification [12]; however, the major uptake route for wa terborne xenobiotics in fish is directly across the gills [13], and direct uptake of chemicals from water (i.e., bioconcentra tion) is probably much more important than accumulation from food (i.e., biomagnification) [14]. Many hydrocarbon surfac tants are known to bioconcentrate in fish [3,15-19], and the European Economic Community has even adopted surface ac tivity as an indicator of a compound's bioconcentration po tential [20]. The predictably low Henry's law constant for PFAs indicates that they will accumulate in the aquatic environment [21,22], making fish a relevant test organism for PFA bioac cumulation testing. Under field conditions, fish collected downstream of a fire-fighting foam spill contained higher con centrations of several PFAs than fish collected upstream [23]. We describe here the results of a tissue distribution and bioconcentration study with rainbow trout (Oncorhynchus my kiss) exposed simultaneously to a suite of perfluoroalkyl carboxylates and sulfonates of varying fluorinated chain lengths in a flow-through aqueous exposure. We report compoundspecific bioaccumulation parameters determined by liquid chromatography-tandem mass spectrometry (LC-MS-MS), and discuss their relation to physicochemical properties and structure. 196 Bioconcentration and tissue distribution of PFAs MATERIALS AND METHODS Standards and reagents Standards of potassium perfluorobutane sulfonate (PFBS), potassium PFHxS (99.9%), and potassium PFOS (86.4%) were provided by the 3M Company (St. Paul, MN, USA). Standards of perfluoropentanoic acid (97%), perfluoroheptanoic acid (PFHpA, 99%), PFOA (98%), perfluorononanoic acid (PFNA, 97%), perfluorodecanoic acid (PFDA, 98%), perfluoroundecanoic acid (95%), perfluorododecanoic acid (PFDoA, 95%), and perfluorotetradecanoic acid (PFTA, 97%) were purchased from Sigma-Aldrich (Oakville, ON, Canada), and perfluorohexanoic acid (95%) was obtained from Oakwood Research Chemicals (West Columbia, SC, USA). Ammonium acetate (98%) and tetrabutylammonium hydrogensulfate were pur chased from Sigma-Aldrich, anhydrous sodium carbonate (99.8%) was purchased from J.T Baker (Phillipsburg, NJ, USA), and methyl-tert-butyl ether (99.5%) was purchased from EM Science (Gibbsburg, NJ, USA). Fish rearing Rainbow trout were purchased from Rainbow Springs (Thamesford, ON, Canada) and were allowed to acclimate to lab oratory conditions for two weeks before exposure. Carbonfiltered and dechlorinated water (with Na2SO3) was maintained at 12C, and a 12-h photoperiod was used. Fish were fed daily at a rate of 1.5% body weight per day, corrected for growth throughout the experiment. Trout feed was purchased from Martin Mills (Tavistock, ON, Canada). Stock perfluorinated acid solution preparation A stock solution containing all the test compounds was produced by first dissolving PFAs in a small amount of meth anol, which subsequently was dissolved in 30 L of reverseosmosis laboratory water. The resulting solution was contained in a polypropylene container and was stirred for 3 d before the beginning of the experiment. After allowing the solution to settle for several hours, some of the test compounds had not completely dissolved, based on the appearance of a white solid at the surface of the solution. This material was removed by filtering the entire solution through glass (GF/C) microfiber filters (Whatman, Kent, UK). The resulting solution was used for exposure and was constantly stirred while being delivered to exposure tanks. Bioconcentration exposure Juvenile rainbow trout (5-10 g) were exposed to a 1,000 fold dilution of the stock PFA solution in a flow-through ex posure design for 12 d, followed by 33 d of depuration in clean water. Two days before the initial exposure, fish were trans ferred to glass aquaria lined with polypropylene bags. Perfluorinated acids have been reported to bind to glass surfaces [5], and the plastic bags were used as a precaution to minimize equilibration time after introduction of the test compounds. Initial biomass loading was 8 g/L in the exposure tank, and 2.5 g/L in the control tank. Fish growth was monitored by weighing the total biomass every 2 to 3 d throughout the course of the experiment. Dilution water was gravity fed to each aquarium at 500 ml/ min, and a peristaltic pump delivered the stirred stock solution into the dilution stream of the treatment tank at 0.5 ml/min. At time 0 of the exposure, an initial volume of the stock solution (45 ml) was added to the exposure aquarium (45 L) to immediately achieve the desired exposure concentration. Environ. Toxicol. Chem. 22, 2003 197 Three fish from the exposure tank and one from the control tank were sampled at each predetermined interval during the uptake phase of the experiment (4.5, 9, 18, 36, 72, 144, and 288 h). At 288 h, the remaining fish were transferred to new aquaria receiving clean water at 3 L/min. Extra care was taken to reduce contamination of the depuration phase tanks by al lowing the fish to depurate for 5 min in a pail containing clean dilution water, and then transferring the fish by net to the depuration tanks. During the depuration phase, three fish from the treatment tank and one from the control tank were sampled at each time interval (4.5, 9, 18, 36, 72, 144, 288, 456, and 792 h). During the uptake phase, water samples (1 L) were collected below the surface in polypropylene bottles at 0.25, 4.5, 12, 18, 36, 72, 144, 197, 244, and 288 h. Water samples were also collected at 48 and 96 h of the depuration phase from both tanks to check for contamination. Sampled fish were anesthetized with MS-222, a blood sam ple (50-200 ^l) was drawn, and fish were subsequently killed by a blow to the head and cervical dislocation. An incision was made along the ventral surface from the anus to the gills, and the entire liver was removed for analysis. The gut, con sisting of esophagus, stomach, pyloric ceca, spleen, and in testines, was removed but not analyzed. The blood, liver, and carcass samples were analyzed separately for PFAs at each sampling time to determine the kinetics of uptake and depu ration. Tissue distribution exposure Four immature rainbow trout (30-48 g) were exposed in a separate tank under the same uptake conditions as bioconcen tration fish. On day 12, three fish were randomly sampled and anesthetized with MS-222. Blood samples were immediately drawn and the fish were euthanized by a blow to the head and cervical dislocation. Fish were subsequently dissected to sep arate and remove the spleen, heart, liver, gall bladder, gonads, gills (bones removed), adipose tissue (separated from pyloric ceca), gut (including esophagus, stomach, intestine, and py loric ceca), and kidney. A small sample of white muscle was also removed adjacent to the dorsal fin of each fish and sep arated from the skin. Analysis o f perfluorinated acids by liquid chromatography-tandem mass spectrometry Liver and blood samples were homogenized in 15-ml plastic (polypropylene copolymer) centrifuge tubes containing 3 ml of Na2CO3 (0.25 M), 1 ml of water, 1 ml of the ion-pairing agent tetrabutylammonium hydrogensulfate (0.5 M adjusted to pH 10) [5], and 100 ^l (25 ng) of the internal standard, PFNA. Carcass samples were first reduced to a fine powder by using a mortar and pestle with liquid nitrogen, and subsequently were homogenized in 50-ml plastic centrifuge tubes containing 10 to 20 ml of Na2CO3. An exact quantity (2-4 g) of the ho mogenate was then transferred to a separate centrifuge tube containing 1 ml of water, 1 ml of tetrabutylammonium hydro gensulfate, and 100 ^l of PFNA. The resulting homogenates were extracted with 5 ml of methyl-tert-butyl ether by shaking vigorously for 10 min, fol lowed by centrifugation to isolate the organic phase. The methyl-tert-butyl ether supernatant was collected in a separate plas tic tube, and this extraction process was repeated once more, combining the supernatants. The methyl-tert-butyl ether was blown to dryness under high-purity nitrogen gas, and the an alytes were taken up in 1 to 2 ml of 50:50 (v/v) water:methanol 198 Environ. Toxicol. Chem. 22, 2003 J.W. Martin et al. Table 1. Test compound acronym, structure, ion transition monitored, and mean aqueous exposure concentration 1 relative standard deviation (RSD) Test compound Acronym Structure Ion transition monitored by LC-MS-MSa Waterborne concentration (pgA) RSD (%) Perfluorocarboxylates Perfluoropentanoic acid Perfluorohexanoic acid Perfluoroheptanoic acid Perfluorooctanoic acid Perfluorodecanoic acid Perfluoroundecanoic acid Perfluorododecanoic acid Perfluorotetradecanoic acid Perfluorosulfonates Perfluorobutane sulfonic acid Perfluorohexane sulfonic acid Perfluorooctane sulfonic acid PFPA PFHxA PFHpA PFOA PFDA PFUnA PFDoA PFTA PFBS PFHxS PFOS CF3(CF2 )3CO2H CF3(CF2 )4 CO2H CF3(CF2 )5CO2H CF3(CF2 )6CO2H CF3(CF2 )8CO2H CF3(CF2 )9CO2H CF3(CF2 )10CO2H CF3(CF2 )12CO2H CF3(CF2 )3 SO3H CF3(CF2 )5 SO3H CF3(CF2 )7 SO3H aLC-MS-MS = liquid chromatography-tandem mass spectrometry. 263 219 313 269 363 319 413 369 513 469 563 519 613 569 713 669 299 99 399 99 499 99 1.7 11 1.7 10 1.6 12 1.5 13 0.71 24 0.48 26 0.20 30 0.014 30 1.4 10 1.4 11 0.35 26 by vortexing for 30 s. The solution was then filtered through 0.2-^m nylon filters into polypropylene vials for analysis. Direct analysis of uptake phase water samples was possible for all PFAs except for PFDoA and PFTA, which were below the limit of detection (~0.4 ^g/L). For direct analysis, 1-L samples were shaken vigorously and 1 ml was withdrawn from below the surface of each with a polypropylene syringe. These subsamples were then filtered through 0.2-^m nylon filters into polypropylene shell vials for analysis. These samples were analyzed without PFNA, and instead used external standards and quantitation by standard curve analysis. To determine PFDoA and PFTA uptake concentrations (36, 72, and 197 h), and to check for PFA contamination of depuration water, 25 ml of each water sample was removed and filtered (0.2-^m nylon) into a 50-ml plastic centrifuge tube containing 0.6 g of Na2CO3, 1 g of tetrabutylammonium hydrogensulfate, and 100 ^l of PFNA. Perfluorinated acids were extracted two times by using 20-ml aliquots of methyl-tert-butyl ether, centrifuging, combining the supernatants, and analyzing further in the same manner as tissue samples. Instrumental analysis was performed by LC-MS-MS by us ing previously described conditions [4,23]. Water and meth anol solvents (0.01 M ammonium acetate) were delivered at a total flow rate of 300 ^l/min by a Waters 600S controller (Milford, MA, USA), and samples (25 ^l) were injected with a Waters 717 plus autosampler. Chromatography was per formed on a Genesis C8 column (2.1 X 50 mm, Jones Chro matography, Lakewood, CO, USA). Initial mobile phase con ditions were 90:10 (v/v) water:methanol for 30 s, followed by a 10-min ramp to 0:100, a 3.5-min hold, and reverting to initial conditions at 14 min. The detector was a Quattro LC (Micro mass, Manchester, UK) equipped with an electrospray interface operating in negative ion mode. Data were acquired by MSMS by using a multiple reaction monitoring method that mon itored a single transition (parent daughter ion) for each compound (Table 1). Desolvation temperature was 350C, and the source block was maintained at 150C. Desolvation gas flow was between 600 and 700 L/h, and the capillary voltage always was 2.75 kV. Quantitation was performed relative to PFNA by using a standard curve constructed from known quantities of standards extracted from water in the same manner as tissue samples. All samples were blank subtracted before quantitation, and standard injections were made every six to nine samples to monitor sensitivity drift. Blank concentrations were absent for most PFAs; however, PFOA and PFOS were detected on oc casion below 10% of sample concentrations. Data analysis Fish weight (FW) was best predicted by the exponential growth model, FW = a exp(g t), where a is a constant, g is the growth rate, and t is the time. All tissue concentrations were corrected for growth dilution by determining the percent increase in FW at each sampling interval, relative to t = 0, by using the predicted exponential growth rate equation. The depuration rate constants (kd) were determined by linear re gression after fitting the growth corrected depuration concen trations (Cflshw) to the first-order decay model Cflshw = aexpp--kdt), where a is a constant. Depuration half-life was calculated by the formula ln(2)/kd. Uptake rate constants (ku) were determined by using iter ative nonlinear regression (Systat, Ver 9.0, Systat Software, Richmond, CA, USA), by fitting the growth-corrected tissue concentrations to the integrated form of the kinetic rate equa tion for constant aqueous exposure [14] CflShW= [(ku)(Cw)/(kd)]-[l --exp(--kdt)] (1) where Cwis the average exposure water concentration and kd is a fixed parameter. Bioconcentration factors (BCFs) were calculated by the quotient (ku/kd). RESULTS AND DISCUSSION Fish mortality, growth, and liver somatic index Only one fish (2%) died during the uptake phase, and no statistically significant difference was found in the rate of growth for exposed fish relative to controls (p = 0.41; Table 2). The mean initial fish mass (Table 2) was greater in this study relative to the dietary accumulation study [12], resulting in a lower rate of growth. No statistically significant difference was found between the liver somatic index of exposed and control fish (p = 0.854; Table 2). Perfluorinated acids cause hepatomegaly in rodents upon exposure via peroxisome pro liferation [24], but this effect was not apparent in these fish under this exposure scenario based on liver mass. Bioconcentration and tissue distribution of PFAs Environ. Toxicol. Chem. 22, 2003 199 Table 2 Uptake and depuration phase duration, growth rate constant, and the associated coefficient of determination, mortality, and liver somatic index (LSI) for exposed and control juvenile rainbow trout used in bioconcentration testing Exposure tank Control Uptake period (d) 12 12 Depuration period (d) 33 33 Mean initial fish mass (g) 7.3 7.9 Growth rate (10-3 g/d) (r2) 4.9 (0.50) 5.1 (0.43) Mortality (%) 2 0 LSIa (%) 1.1 0.03 1.1 0.07 aAverage of all fish sampled throughout the uptake and depuration period in each tank, for each experiment. Linear regression revealed no increase or decrease in LSI throughout. Water concentrations During the uptake period the average exposure water con centrations were between 0.014 and 1.7 ^g/L for individual PFAs (Table 2 and Fig. 1). The low exposure concentrations for longer PFAs were a result of their low water solubility, whereas more soluble PFAs were close to the nominal con centration (~2 ^g/L). Perfluorinated acid concentrations were relatively stable throughout the uptake phase after an initial decrease between 0.25 and 24 h (Fig. 1). This decrease was more pronounced for more hydrophobic compounds (i.e., PFOS > PFHxS; Fig. 1B), suggesting that rapid uptake by fish was influencing water concentrations to some extent de spite flow-through conditions. As biomass was removed through sampling, and as uptake rates decreased, the water concentrations stabilized or increased slightly at later sampling intervals, and approached the initial concentrations (i.e., 0.25 h). The mean water concentration was used as a constant es timate of Cwin Equation 1 for determination of ku. Perfluorinated acids could not be detected in depuration water above the limit of detection (~5 ng/L) in control or treated fish tanks. Tissue distribution In general, PFA concentrations were greatest in the blood > kidney > liver > gall bladder, and lowest in the gonads > adipose > muscle tissue (Fig. 2). Within the blood, the plasma contained between 94 and 99% of total PFAs, with only a minor fraction detectable in the cellular fraction. Perfluorinated acids also were detectable in the gills, suggesting that this was uptake depuration Blood uptake depuration Bood 10- 10 Tissue - e - PFOA --V-- PFDA - e -- PFUnA -0 -~ PFDoA --a-- PFTA Water PFOA --w - PFDA -- PFUnA PFDoA PFTA Liver Tissue PFHxS PFOS PFHxS Liver io Carcass Carcass 200 400 600 800 1000 1200 Time (h) 200 400 600 800 1000 1200 Fig. 1. Growth-corrected uptake and depuration phase concentrations for (A) perfluoroalkyl carboxylates and (B) perfluoroalkyl sulfonates in blood, liver, and carcass. Each tissue concentration representsthe mean ofthree different fish( 1 standarderror). Aqueousexposure concentrations during the uptake phase are shown in A and B for carboxylates and sulfonates, respectively. See Table 1 for definitions of test compounds. 200 Environ. Toxicol. Chem. 22, 2003 J.W. Martin et al. Fig. 2. Perfluorinated acid concentrations, 1 standard error, in various fish tissues after a 12-d aqueous exposure. See Table 1 for definitions of test compounds. the site of uptake, depuration, or both, as has been determined for other xenobiotics [13], including surfactants [15]. Recov ery of analytes from the heart and spleen was low (<10% based on internal standard response), thus preventing quanti tation of PFAs in these tissues. In almost all surfactant bio concentration tests for which tissue-specific data have been presented, the highest analyte (i.e., radiolabel) concentrations typically are found in the gall bladder, probably a result of metabolism in the liver [15]. The gall bladder concentrations determined herein may be lower than in previous surfactant studies [15] because the fish were fed daily, thus preventing concentration and accumulation of the bile [25]. Unlike lipophilic chlorinated organic pollutants, PFAs did not accumulate preferentially in adipose tissue. For example, dioxin congeners accumulated in the ceca and depot fat of rainbow trout to a greater extent than kidney, spleen, muscle, or skin tissue [26]. The tissue distribution of PFDA in male rats is similar to our results for rainbow trout, except that rat liver contained by far the greatest concentrations, followed by blood > kidney > heart > fat > testis > muscle [27]. This may be partially attributable to the lipophobic properties of PFAs; however, PFAs also have a high affinity for plasma albumin [28], and it has been hypothesized that PFAs may bind to hepatic proteins such as fatty acid-binding proteins [27], thus explaining the high PFA concentrations in blood and liver, respectively. The relatively high concentrations of PFAs in rainbow trout kidney may simply reflect the high perfusion of blood, and do not automatically indicate that uri nary elimination is a significant depuration route. For example, despite high kidney concentrations of PFDA in exposed rats, the primary mode of depuration was feces, with only minor excretion in urine [27]. The route by which humans are exposed to PFAs is not well characterized, but consumption of fish meat may well be a significant source of exposure. Assuming that muscle com prises 67% of a rainbow trout by weight [29], the meat could therefore contain 22, 61, and 81% of the total body burden of PFOS, PFHxS, and PFOA, respectively, based on the tissue concentrations (Fig. 2). Bioconcentration parameters Similar to the results of a PFA dietary accumulation study [12], only carboxylates with more than six perfluoroalkyl car bons, and sulfonates with more than four perfluoroalkyl car bons were detected in blood, liver, and carcass at all sampling times (Fig. 1). Shorter PFAs are expected to have insignificant bioconcentration potential, but their uptake or depuration ki netics could not be determined here, although the half-life of PFBS in liver of rainbow trout was 3.3 d in a dietary accu mulation study [12]. Recovery of most PFAs from liver and carcass is quantitative [12], and recovery from blood has al ready been demonstrated by Hansen et al. [5] for PFOS, PFHxS, and PFOA. We made no attempt to monitor for me tabolites of PFAs because the existing literature suggests that biotransformation will be insignificant or absent [27,30,31], and we assumed that depuration was entirely a function of elimination. Visual observation of depuration data indicated possible biphasic depuration for most PFAs in blood, liver, and carcass (Fig. 1); however, this could not be demonstrated statistically because of the small sample size. Linear regression Bioconcentration and tissue distribution of PFAs Environ. Toxicol. Chem. 22, 2003 201 Table 3. Rate of uptake (u), depuration (kd), half-life (d), steady-state bioconcentration factor (BCF; estimated by ku/kd), and the 12-d accumulation ratio (AR), calculated at the end of the uptake period. Error represents 1 standard error, and a number in parentheses represents the coefficient of determination (r2) for the corresponding regression analysis Test compound" ku (L/kg/d) kd X 10-3 (l/d) Half-life (d) BCF (L kg) 12-d AR Carcass PFOA PFDA PFUnA PFDoA PFTA PFOS PFHxS Blood PFOA PFDA PFUnA PFDoA PFTA PFOS PFHxS Liver PFOA PFDA PFUnA PFDoA PFTA PFOS PFHxS 0.53 0.041 (0.80) 29 1.0 (0.95) 120 4.8 (0.95) 700 29 (0.94) 580 26 (0.93) 53 1.3 (0.98) 0.62 0.021 (0.96) 4.1 160 500 1,700 1,400 240 5.3 0.31 (0.83) 4.6 (0.97) 18 (0.96) 65 (0.96) 120 (0.82) 8.4 (0.90) 0.48 (0.76) 1.4 0.13 (0.65) 74 5.3 (0.84) 260 20 (0.84) 910 79 (0.78) 960 98 (0.69) 260 17 (0.88) 5.8 0.31 (0.91) 130 17 (0.76) 62 8.2 (0.71) 46 6.5 (0.68) 38 5.5 (0.68) 24 5.5 (0.45) 48 6.5 (0.70) 65 6.2 (0.82) 150 53 (0.83) 58 7.2 (0.97) 46 5.3 (0.96) 41 4.3 (0.96) 48 5.28 (0.82) 57 6.7 (0.90) 70 6.5 (0.76) 180 13 (0.91) 65 8.9 (0.69) 55 7.7 (0.68) 50 6.7 (0.70) 31 5.5 (0.60) 50 7.4 (0.66) 58 7.0 (0.75) 5.2 0.67 11 1.5 15 2.2 18 2.6 28 6.4 15 2.0 11 1.0 4.5 1.6 12 1.5 15 1.8 17 1.7 15 1.6 12 1.4 10 0.94 3.9 0.28 11 1.5 13 1.7 14 1.9 22 3.7 14 2.0 12 1.5 4.0 450 2,700 18,000 23,000 1,100 9.6 0.60 62 400 2,700 5,300 150 0.99 27 2,700 11,000 40,000 30,000 4,300 76 9.7 350 1,400 4,500 4,200 570 9.7 8.0 1,100 4,900 18,000 30,000 5,400 100 0.59 180 770 2,900 6,000 860 13 3.2 350 1,400 6,600 8.500 690 7.6 25 1,900 5,500 18,000 20,000 3,100 59 12 1,100 3,800 11,000 8,700 2,900 54 aRefer to Table 1 for definitions of test compounds. analysis of all growth-corrected depuration data resulted in acceptable coefficients of determination (Table 3). For detected PFAs, depuration rate constants generally de creased with increasing length of the perfluoroalkyl chain for carboxylates and sulfonates in all tissues, ranging from 0.024 to 0.180/d, representing biological half-lives in the range of 3.9 to 28 d. For carboxylates, positive linear relationships were observed between half-life and perfluoroalkyl chain length for the carcass and liver, but not the blood (Fig. 3). In blood, the relationship deviated from linearity for the compound with the longest chain length, PFTA, which was eliminated more quick ly than expected from extrapolation of the observations from PFAs with shorter chain lengths. Perfluorotetradecanoic acid probably partitioned out of the blood into the carcass, rather than being eliminated, as has been shown for other anionic surfactants [18]. Based on the linear regression equations (Fig. 3), half-life increased by 2.8 to 3.8 d for each additional carbon in the perfluoroalkyl chain. Similar analysis could not be ap- Fig. 3. Half-life (ln 2/kd) and uptake rate (log ku) relationships with perfluoroalkyl chain length of carboxylates and sulfonates for (A) carcass, (B) liver, and (C) blood. Linear regression was applied to perfluoroalkyl carboxylate data, and the resulting equation and coefficient of determination (r2) are shown. The dashed portion of the regression line is an extrapolation and indicates that perfluorotetradecanoic acid was not included in the regression analysis. 202 Environ. Toxicol. Chem. 22, 2003 plied to the sulfonates because only two were detectable; how ever, in all tissues, the half-lives of PFHxS and PFOS were greater than those of carboxylates of equivalent perfluoroalkyl chain length (Fig. 3). The depuration half-lives determined in this study are slightly greater than were determined in a dietary PFA ex posure study [12]. Assuming that the mode of exposure is not important, we suggest that some of this variability may be attributable to the difference in fish size. For example, at the beginning of the depuration period in this study, the average trout mass was 8.4 g, compared with only 4.0 g in the dietary exposure study [12]. Hydrophobic organochlorine compounds previously have been demonstrated to have longer half-lives in larger fish [32,33], even when controlling for growth di lution. Body size also may help to explain the variability among half-lives reported for humans (1,428 d [2]), monkeys (180 d [34]), and male rats (89 d [34]). In comparison, the half-life in fish carcass was approximately 15 d for juvenile trout, but the possibility of elimination through the gills cannot be ruled out as an additional depuration mechanism for fish. The rate of depuration for PFAs is generally more rapid than for persistent organochlorine contaminants, including po lychlorinated biphenyls, toxaphene, hexachlorobenzene, mirex, and chlorinated alkanes [33,35]. However, the rate of dep uration for PFAs, with the exception of PFOA, is slower than for any previously investigated surfactant in fish [15], and this may be partially attributable to the lack of metabolism or bio transformation [27,30,31]. The relatively slow PFA depuration half-lives, combined with the observations of high blood, liver, and gall bladder concentrations, also support the theory that PFAs enter into enterohepatic recirculation in fish, the process whereby compounds are continuously recycled between the blood, liver, gall bladder, and intestines, where resorption oc curs via the portal vein. This is further supported by the ob servation of high PFA assimilation efficiencies [12], which indicate that intestinal resorption efficiency is similar to that of endogenous bile acids. Use of the relatively low assimilation efficiency of PFOA (59% [12]) as a measure of resorption from the gut during enterohepatic recirculation could explain why its half-life is shorter than for longer PFAs, which have a higher resorption capacity [12]. The divergence of sulfonate and carboxylate rates of depuration (Fig. 3) also could be attributable to the different efficiency of resorption from the gut during enterohepatic recirculation. For example, the as similation efficiency of PFHxS, which has only six perfluoroalkyl carbons, is greater than that for PFOA, which has seven perfluoroalkyl carbons [12]. Also, short PFAs possibly are eliminated in the urine to a greater extent than longer PFAs, as has been shown for PFHpA in rats [24]. Uptake rate constants varied by a factor of 3,000 for dif ferent PFAs, and increased with increasing perfluoroalkyl chain length (Fig. 3), ranging from 0.53 to 1,700 L/kg/d for all tissues (Table 3). A similar trend, showing increased uptake rates with increasing alkyl chain length, has been reported for a series of cationic monoalkyltrimethyl ammonium surfactants [36]. In general, these PFA uptake rates may be considered very rapid and, to our knowledge, the carcass kureported here for PFDoA exceeds any previously reported rate of uptake for anionic surfactants (i.e., 642 L/kg/d reported by Tolls et al. [3]). Given equal perfluoroalkyl chain lengths, sulfonates were taken up at a greater rate than carboxylates, always lying above the regression line for carboxylates (Fig. 3). For carboxylates, J.W. Martin et al. Fig. 4. Relationship between gill and blood concentrations, 1 stan dard error, in tissue distribution in fish. Each data point represents the mean of three fish, and is labeled with the perfluorinated acid and the length of the respective perfluoroalkyl chain in parentheses (e.g., C8 = eight carbons). See Table 1 for definitions of test compounds. uptake increased in a linear manner with perfluoroalkyl chain length until the length reached 13 carbons (i.e., PFTA). The rate of uptake for PFTA, presumably the most hydro phobic PFA tested, was less than expected based on extrap olation from shorter PFAs in liver, blood, and carcass (Fig. 3). Such a phenomenon is well characterized for hydrophobic chlorinated contaminants, and can result from exclusion based on limited diffusive mass transfer [37], molecular size [38], or from reduced bioavailability because of binding on dis solved or colloidal organic matter in the water column [39,40]. Reduced bioavailability because of association with dissolved and colloidal organic matter has been demonstrated for cationic surfactants [15], and anionic surfactants [41]. The octanolwater partition coefficient (KOW) for PFTA is unknown; how ever, it is feasible that the long perfluoroalkyl chain of PFTA causes the molecule to act similar to superhydrophobic neutral organochlorines (i.e., log KOW > 6), which interact with or ganic matter, resulting in an apparent reduction of the uptake rate [39]. Alternatively, based on the ratio of concentrations in the gills and blood plasma at day 12 ofthe tissue distribution study, the gills appear to be less permeable to PFTA than shorter PFAs (Fig. 4). Perfluorotetradecanoic acid was en riched at the gills relative to shorter perfluoroalkyl carboxylates and sulfonates, suggesting that gill permeation rates may limit uptake of PFTA, as has been shown for the high molecular weight cationic surfactants C16-alkylpyridinium and (C10)2-dialkyldimethylammonium [15]. A time course of the tissue dis tribution would allow for more conclusive arguments; how ever, gill permeation is known as the rate-limiting step for uptake of other anionic surfactants [18]. Steady-state carcass BCFs ranged from 4.0 to 23,000, and generally increased with the perfluoroalkyl chain length (Table 3 and Fig. 5). Similar trends, showing increased BCFs with increased alkyl chain length, have been reported for anionic linear alkylbenzene sulfonate surfactants [18] and some cat ionic surfactants [36]. For PFAs, blood and liver BCFs were 1 to 10 times greater than in the carcass (Table 3). Carboxylate BCFs were greatest in the blood, whereas sulfonate BCFs were greatest in the liver. Carcass BCFs closely approximate the whole-body BCF; however, removal of the liver and a portion of the blood from the intact fish resulted in a minor loss of Bioconcentration and tissue distribution of PFAs Environ. Toxicol. Chem. 22, 2003 203 Fig. 5. Carcass bioconcentration factor (BCF) relationship with perfluoroalkyl chain length. Linear regression was applied to perfluoroalkyl carboxylate data, and the resulting equation and coefficient of determination (r2) are shown. Perfluorotetradecanoic acid was not included in the regression analysis and the dashed portion of the regression line is an extrapolation of the regression equation. Fig. 6. Relationship determined by linear regression between the log critical micelle concentration (log CMC) and perfluoroalkyl chain length of potassium perfluoroalkyl carboxylate salts. The CMC for the potassium salt of perfluorooctane sulfonate (PFOS~K+) fits the relationship. All data have been taken from Kissa [42]. total PFAs. The resulting carcass contained between 85% (i.e., PFHxS) and 96% (i.e., PFTA) of total PFAs, depending on the test compound. Bioconcentration factors increased largely as a result of increasing uptake rate, whereas depuration rate varied only by a factor of eight for all tissues examined and had less relative influence on the observed BCF trends. Because steady state was not achieved for any PFA, a 12d accumulation ratio was calculated (Table 3) and compared to the estimated steady-state BCF. The 12-d accumulation ratio always was smaller than the BCF; however, for the PFAs with the shortest chain lengths, the 12-d accumulation ratio ap proached the estimated BCF to within 80% (i.e., PFOA). The tissue concentrations of PFAs with longer chain lengths were far from steady state at the end of the uptake period, reaching only 37% of the predicted steady-state value by day 12 (i.e., PFDoA and PFTA). The time to reach steady state is deter mined exclusively by the rate of depuration, and was predicted to exceed 10, 43, and 120 d for PFOA, PFOS, and PFTA, respectively [12]. The fact that steady state was not approached should not influence our estimates of the BCF calculated by the ratio of uptake and elimination rates. A previous compar ison of the kinetic and steady-state approaches to the study of surfactant bioaccumulation indicated that both methods yield similar results, and that the one-compartment, first-order model was suitable [15]. Perfluoroalkyl carboxylate carcass BCFs (i.e., log BCF) increased linearly with perfluoroalkyl chain length from PFOA to PFDoA by a factor of eight for each additional fluorinated carbon in the chain (Fig. 5). The trend deviated from linearity for PFTA as a result of the decreased rate of uptake (Fig. 3A). Given equal perfluoroalkyl chain length, sulfonates biocon centrated to a greater extent than carboxylates, largely as a result of higher uptake rates, but also because of lower dep uration rates, and did not fit the same linear relationship for carboxylates, always lying above the linear regression line (Fig. 5). Furthermore, PFHxS, which has six perfluoroalkyl carbons, had a carcass BCF of 9.6, whereas PFHpA, which also has six perfluoroalkyl carbons, could not be detected in most tissues despite a higher exposure concentration (Table 1). In a dietary exposure of fish to PFAs, perfluoroalkyl sul fonates also bioaccumulated to a greater extent than carbox- ylates of equivalent perfluoroalkyl chain length, and a similar linear trend was observed for carboxylate bioaccumulation fac tors [12]. Bioconcentration potential historically has been well pre dicted by the KOW, a measure of a substance's hydrophobicity. However, the KOW is a problematic parameter for surfactants because of their tendency to aggregate at the interface of a liquid-liquid system. Here we have used the perfluoroalkyl chain length as a measure of hydrophobicity; however, Tolls and Sijm [16] suggested that a more meaningful measure of the relative hydrophobicity for surfactants is the critical mi celle concentration (CMC), the concentration at which half of the molecules in solution are associated as micelles. Critical micelle concentration data are limited or nonexistent for many of the test compounds; however, the CMC of PFOS fits the relationship determined for several carboxylates of variable perfluoroalkyl lengths (Fig. 6) [42], indicating that the acid function has little influence on CMC. Analysis of our data indicates that bioaccumulation potential increases with de creasing CMC for carboxylates; however, the CMC alone does not explain the disproportionate accumulation of sulfonates relative to carboxylates. For PFAs, the acid function must be considered for prediction of the bioconcentration and bioac cumulation potential [12]. For perfluoroalkyl carboxylates, the slope of the log CMC data (Fig. 6) is remarkably similar, but opposite in sign, to the plot of log uptake rate in carcass (Fig. 3A), indicating that increased uptake for longer PFAs within a class may be a function of increasing hydrophobicity. The linear CMC-BCF relationships, described by Tolls and Sijm [16], had relatively low coefficients of determination, and it was noted that the CMC might not be the most suitable parameter for describing hydrophobicity of surfactants. The suggestion was made that interfacial tension or reversed-phase chromatographic retention time might be more suitable param eters. When using our chromatographic methods, PFOS and PFNA (i.e., both of which have eight perfluoroalkyl carbons and similar CMCs) could not be time-resolved on a reversedphase column, with PFNA eluting only 0.02 min after PFOS. Additional physical and chemical properties for this series of PFAs would be beneficial for future modeling efforts, partic ularly for the longer perfluoroalkyl carboxylates. 204 Environ. Toxicol. Chem. 22, 2003 Acknowledgement--T hisw orkw assuppoetedbyHealthCanadaand Environment Canada (Ottawa, ON) through the Toxic Substance Re search Initiative, and the Canadian Network of Toxicology Centres. We thank the 3M Company (St. Paul, MN, USA) for its generous donation of perfluoroalkyl-sulfonate chemical standards. REFERENCES 1. Fielding HC. 1979. Organofluorine surfactants and textile chem icals. In Banks RE, ed, Organofluorine Chemicals and Their Industrial Applications. Ellis Horwood, Chichester, UK, pp214234. 2. U.S. Environmental Protection Agency. 2001. Perfluorooctyl sul fonates; proposed significant new use rule. F ed Reg 65:62319 62333. 3. Tolls J, Haller M, Graaf ID, Thijssen MATC, Sijm DTHM. 1997. Bioconcentration of LAS: Experimental determination and ex trapolation to environmental mixtures. Environ Sci Technol 31: 3426-3431. 4. Moody CA, Kwan WC, Martin JM, Muir DCG, Mabury SA. 2001. Determination of perfluorinated surfactants in surface water samples by two independent analytical techniques: Liquid chromatographb/tannem mass spectrometry and 19F NMR. Anal Chem 73:2200-2206. 5. Hansen KJ, Clemen LA, Ellefson ME, Johnson HO. 2001. Com pound specific quantitative characterization of organic fluorochemicals in biological matrices. Environ Sci Technol 35:766 770. 6. Giesy JP, Kannan K. 2001. Global distribution of perfluorooctane sulfonate in wildlife. Environ Sci Technol 35:1339-1342. 7. Kannan K, Koistinen J, Beckman K, Evans T, Gorzelany JF, Han sen KJ, Jones PD, Helle E, Nyman M, Giesy JP. 2001. Accu mulation of perfluorooctane sulfonate in marine mammals. E n viron Sci Technol 35:1593-1598. 8. Kannan K, Franson JC, Bowerman WW, Hansen KJ, Jones PD, Giesy JP 2001. Perfluorooctane sulfonate in fish-eating birds in cluding bald eagles and albatrosses. Environ Sci Technol 35: 3065-3070. 9. Key BD, Howell RD, Criddle CS. 1998. Defluorination of or ganofluorine sulfur compounds by Pseudomonas sp. strain D2. Environ Sci Technol 32:2283-2287. 10. Martin JW, Muir DCG, Moody CA, Ellis DA, Kwan W, Solomon KR, Mabury SA. 2002. Collection of airborne fluorinated organ ics and analysis by gas chromatography/chemical ionization mass spectrometry. Anal Chem 74:584-590. 11. Ellis DA, Mabury SA, Martin JW, Muir DCG. 2001. Thermolysis of fluoropolymers as a potential source of halogenated organic acids in the environment. Nature 412:321-324. 12. Martin JW, Mabury SA, Solomon KR, Muir DCG. 2003. Dietary accumulation of perfluorinated acids in juvenile rainbow trout (Oncorhynchus mykiss). Environ Toxicol Chem 22:189-195. 13. Streit B. 1992. Bioaccumulation processes in ecosystems. Experientia 48:955-970. 14. Bruggeman WA, Martron LBJM, Kooiman D, Hutzinger O. 1981. Accumulation and elimination of di-, tri-, and tetrachlorobiphen yls by goldfish after dietary and aqueous exposure. Chemosphere 10:811-832. 15. Tolls J, Kloepper-Sams P, Sijm THM. 1994. Surfactant biocon centration--Acriticalreview. Chemosphere 29:693-717. 16. Tolls J, Sijm DTHM. 1995. A preliminary evaluation of the re lationship between bioconcentration and hydrophobic^ for sur factants. Environ Toxicol Chem 14:1675-1685. 17. Tolls J, Sijm DTHM. 1999. Bioconcentration and biotransfor mation of the nonionic surfactant octaethylene glycol monotridecy! ether14C-C13EO8. Environ Toxicol Chem 18:2689-2695. 18. Tolls J, Haller M, Seinen W, Sijm DTHM. 2000. LAS biocon centration: Tissue distribution and effect of hardness--Implica tions for processes. Environ Sci Technol 34:304-310. 19. Tolls J, Haller M, Labee E, Verweij M, Sijm DTHM. 2000. Ex perimental determination of bioconcentration of the nonionic sur factant alcohol ethoxylate. Environ Toxicol Chem 19:646-653. 20. European Economic Community. 1993. Council regulation 793/ 93 of 23 March on the evaluation and control of the risks of existing substances. O ff J E ur Commun L:84/81. 21. U.S. Environmental Protection Agency. 2000. Sulfonated per- J.W. Martin et al. fluorochemicals in the environment: Sources, dispersion, fate and effects. AR226-0620 . 3M Company, St. Paul, MN. 22. U.S. Environmental Protection Agency. 1999. Determination of the vapor pressure of PFOS using the spinning rotor gauge meth od. AR226-0048. Wildlife International, Easton, MD. 23. Moody CA, Martin JW, Kwan WC, Muir DCG, Mabury SA. 2001. Monitoring perfluorinated surfactants in biota and surface water samples following an accidental release of fire-fighting foam into Etobicoke Creek. Environ Sci Technol 36:545-551. 24. Kudo N, Bandai N, Suzuki E, Katakura M, Kawashima Y. 2000. Induction of perfluorinated fatty acids with different carbon chain length of peroxisomal B-oxidation in the liver of rats. Chem Biol Interact 124:119-132. 25. Eckert R, Randall D, Augustine G. 1996. Animal Physiology Mechanisms and Adaptations, 3rd ed. W.H. Freeman, New York, NY, USA. 26. Muir DCG, Yarechewski AL, Knoll A. 1986. Bioconcentration and disposition of 1,3,6,8-tetrachlorodibenzo-p -dioxin and octachlorodibenzo-p -dioxin byrainbow trout and fathead minnows. Environ Toxicol Chem 5:261-272. 27. Vanden Heuvel JP, Kuslikis BI, Van Rafelghem MJ, Peterson RE. 1991. Disposition of perfluoronecanoic acid in male and female rats. Toxicol Appl Pharmacol 107:450-459. 28. Guy WS, Taves DR, Brey WS Jr. 1976. Organic fluorocompounds in human plasma: Prevalence and characterization. In Filler R, ed, Biochemistry Involving Carbon-Fluorine Bonds. American Chemical Society, Chicago, IL, pp 117-134. 29. Barron MG, Tarr BD, Hayton Wl. 1987. Temperature-depern dence of cardiac output and regional blood flow in rainbow trout, Salmo gairdneri Richardson. J Fish Biol 31:735-744. 30. Banks RE, Smart BE, Tatlow JC. 1994. Organofluorine Chem istry Principles and Commercial Applications. Plenum, New York, NY, USA. 31. Goecke-Flora CM, Reo NV. 1996. Influence of carbon chain length on the hepatic effects of perfluorinated fatty acids. A 19F and 31P-NMR investigation. Chem Res Toxicol 9:689-695. 32. Sijm D, van der Linde A. 1995. Size-dependent bioconcentration kinetics of hydrophobic organic chemicals in fish based on dif fusive mass transfer and allometric relationships. Environ Sci Technol 29:2769-2777. 33. Fisk AT, Norstrom RJ, Cymbalisty CD, Muir DCG. 1998. Dietary accumulation and depuration of hydrophobic organochlorines: Bioaccumulation parameters and their relationship with the octanol/water partition coefficient. Environ Toxicol Chem 17:951 961. 34. U.S. Environmental Protection Agency. 2000. SIDS draft initial assessment report perfluorooctane sulfonic acid and its salts. AR226-0978. 3M Company, St. Paul, MN. 35. Fisk AT, Cymbalisty cD, Bergman A, Muir DCG. 1996. Dietary accumulation of C12- and C^-chlorinated alkanes by juvenile rainbow trout (Oncorhynchus mykiss). Environ Toxicol Chem 15: 1175-1782. 36. Versteeg DJ, Shorter SJ. 1992. Effect of organic carbon on the uptake and toxicity of quartenary ammonium compounds to the fathead minnow, Pimephales prom elas. Environ Toxicol Chem 11:571-580. 37. Gobas FAPC, Opperhuizen A, Hutzinger O. 1986. Bioconcentra tion of hydrophobic chemicals in fish: Relationship with mem brane permeation. Environ Toxicol Chem 5:637-646. 38. Opperhuizen A, van de Welde EW, Asyee GM, van der Steen JMD. 1987. Uptake and elimination by fish of polbnimethblsiloxanes (silicones) after dietary exposure and aqueous exposure. Toxicol Environ Chem 13:265-285. 39. Gobas FAPC, Clark KE, Shiu WY, Mackay D. 1989. Biocon centration of polbbrominaten benzenes and biphenyls and related superhydrophobic chemicals in fish: Role of bioavailability and elimination into the feces. Environ Toxicol Chem 8:231-245. 40. Verhaar HJM, De Jongh J, Hermens JLM. 1999. Modeling the bioconcentration of organic compounds by fish: A novel ap proach. Environ Sci Technol 33:4069-4072. 41. Traina SJ, McAvoy DC, Versteeg DJ. 1996. Association of linear alkylbenzenesulfonates with dissolved humic substances and its effect on bioavailabilty. Environ Sci Technol 30:1300-1309. 42. Kissa E. 1994. Fluorinated Surfactants: Synthesis, Properties, Applications. Marcel Dekker, New York, NY, USA.