In Vivo Passive Sampling of Nonpolar Contaminants in Brown Trout ( Salmo trutta )

In Vivo Passive Sampling of Nonpolar Contaminants in Brown Trout(Salmo trutta)Ian John Allan,* Kine Bæk, Thrond Oddvar Haugen, Kate Louise Hawley, Andreas Sven Høgfeldt,and Adam David Lillicrap

Oslo Centre for Interdisciplinary Environmental and Social Research, Norwegian Institute for Water Research, Gaustadalleen 21,NO-0349, Oslo, Norway

*S Supporting Information

ABSTRACT: Equilibrium passive sampling through in vivo implanta-tion can help circumvent complex extractions of biological tissues,provide more accurate information on chemical contaminant burdenbased on the fugacity of a chemical in an organism rather thanconventional normalization to lipid content, and improve theassessment of contaminant bioaccumulation potential. Here, weexplored the feasibility of in vivo implantation for the passive samplingof neutral hydrophobic contaminants through the insertion of a siliconetag into brown trout (Salmo trutta). Implanted fish from the upperreaches of the River Alna (Oslo, Norway) were relocated to a pollutedsection of the river for a 28 day caged exposure. “Whole fish” lipid-silicone distribution coefficients (Dlip‑sil) were calculated for chlorinatedcompounds measured in whole fish and in silicone tags of 13 fish. Dlip‑silranged from 13.6 to 40.0 g g−1 for polychlorinated biphenyl congeners 28−156 (CB28 and CB156), respectively, and are in closeagreement with literature in vitro lipid phase and tissue-based lipid-silicone partition coefficients. After dissection a further ofeight fish, muscle and liver samples were analyzed separately. Muscle-based Dlip‑sil values similar to the whole fish data wereobserved. However, lipid-normalized concentrations in the liver tended to be lower than in muscle for most compounds (by upto 50%). Values of whole fish Dlip‑sil for brominated diphenyl ethers determined for three fish were in the range of 8.6−51 g g−1

and in agreement with chlorinated substances. Finally, fugacity ratios calculated from equilibrium concentrations in fish-implanted and water-exposed silicone provided information on the bioaccumulation for chlorinated compounds as well as forsome polycyclic aromatic hydrocarbons. Equilibrium passive sampling through in vivo implantation can allow the comparison ofa chemical’s activity or fugacity in biotic as well as abiotic environmental compartments and at different trophic levels up tohumans.

■ INTRODUCTION

The assessment of the bioaccumulation of nonpolar nonionizedcontaminants such as polycyclic aromatic hydrocarbons(PAHs), polychlorinated biphenyls (PCBs) or polybrominateddiphenyl ethers (PBDEs) into aquatic organisms generallyrelies on solvent-based exhaustive extractions of contaminantsin either the whole organisms or specific tissues. Extraction andcleanup procedures for biological matrices can be complex andlengthy with extracts requiring extensive cleanup beforeanalysis. Further normalization of concentrations to the lipidcontent of the matrix being analyzed is undertaken assumingthat lipids are the main component responsible for a tissue ororganism’s fugacity capacity. In addition, the total lipid contentof the sample measured during sample preparation is only acrude descriptor of the actual lipid content and composition ofan organism or matrix.1 It does not distinguish between lipidtypes and may not necessarily be inclusive of all types of lipids.Possible differences in the capacity of different types of lipidssuch as membrane or storage lipids, or as a result of the lipidcomposition found in different species at different trophic levels

means that normalizing contaminant concentrations to the lipidcontent of the matrix being extracted is of limited value.2 Thisapproach does also not take into account other phases such asproteins that may exhibit significant sorption capacityparticularly relevant for organisms of low lipid content.3

These factors tend to weaken comparisons of contaminantconcentrations between different organisms and at differenttrophic levels.2

When a thermodynamic equilibrium is reached between twoenvironmental phases or compartments, all phases have equalfugacity, f or activity.4,5 Recently, the application of in vitropassive sampling with silicone polymer immersed in lipids6 orinserted into lipid-rich fish tissues7,8 showed that it was possibleto obtain an equilibrium between PCB concentrations in lipids

Received: April 24, 2013Revised: September 10, 2013Accepted: September 10, 2013Published: September 10, 2013

Article

pubs.acs.org/est

© 2013 American Chemical Society 11660 dx.doi.org/10.1021/es401810r | Environ. Sci. Technol. 2013, 47, 11660−11667

pubs.acs.org/est

and that in lipid-immersed silicone (or polydimethylsiloxane)with equal fugacity in the two media (eq 1):

=f flip sil (1)

with f lip and fsil, contaminant fugacities (expressed in Pa) in thelipid phase and in the silicone, respectively. Equation 1 can berewritten in terms of contaminant concentrations, Clip and Csil(mol m−3) and fugacity capacity of the two media, Zlip and Zsil(mol m−3 Pa−1), yielding lipid-silicone partition coefficients,Klip‑sil:

= = −C

C

Z

ZKlip

sil

lip

sillip sil

(2)

Values of Klip‑sil for olive, Tobis and seal oils representinglipids found at different trophic levels were in the range of 10−50 g g−1 for PCBs and other chlorinated pesticides.6 Theapplication of in vitro equilibrium sampling with polydime-thylsiloxane in intact fish tissues and Dugong blubber resultedin the measurement of lipid-silicone distribution coefficients,Klip‑sil in a similar range.8−10

Passive sampling methods have been used for two decadesfor the sampling of nonpolar nonionized and possiblybioaccumulative substances in water and sediments.11−13 Invivo passive sampling using solid phase microextraction fibresinserted into the muscle of living fish has been undertaken forthe measurement of some pharmaceutical compounds andpesticides.14,15 This technique has more recently been appliedto the measurement of contaminant bioaccumulation andelimination kinetics in fish with a reduced number of organismsneeded.16 When sampling (such as in these studies) isundertaken in the kinetic mode of uptake,16 tight controlover exposure conditions (e.g., exact physical location of theinserted fiber) and an understanding of factors influencingsampling rates are needed. In vivo implantation passivesampling has recently allowed the measurement of the fugacityof chlorinated and brominated persistent organic pollutants inhumans.17 At equilibrium, the fugacity of the chemical in thepolymeric material is equal to that in the organism, itselfcomposed of a range of tissues/phases of differing volumes, Vi(m3) and fugacity capacities Zi:

=∑

fMVZi i

sil (3)

with M, the total amount of chemical in the organism (mol).An advantage of this procedure is that it can allow themeasurement of contaminant fugacity in organisms with lowlipid content, with lipids of different types and where binding toother phases such as proteins becomes significant.The use of the same inert polymeric phase to estimate

contaminant fugacity in various abiotic and biotic matrices hasinherent advantages since this procedure can provideinformation on the relative “equilibrium” status between phasesor compartments independently of the matrix monitored.Fugacity ratios can be calculated to inform us on this“equilibrium” status.5 Passive sampling data can be related todata from organisms at low trophic levels and for which,contaminant concentrations in tissues are in equilibrium withthe freely dissolved concentrations in their environment.6,18 Inthese cases fugacity ratios close to unity may be expected.Contaminant concentrations in organisms at higher trophiclevels resulting from food ingestion may not necessarily bedirectly related to freely dissolved concentrations in their

environment or that of their prey(s). For these organisms,organism-water fugacity ratios above unity can occur and arerepresentative of biomagnification processes. Ecological factorssuch as home range and feeding habits, possible biotransforma-tion processes, and nonequilibrium conditions can affect thelevels of contaminants in these organisms.The aim of this pilot study was to evaluate the feasibility of in

vivo implantation passive sampling to measure the internalfugacity of nonpolar nonionized contaminants in brown trout(Salmo trutta) over 28 days. The hypothesis was thatcontaminants accumulating in the fish would subsequentlypartition into polydimethylsiloxane tags inserted in the bodycavity of brown trout. Data from the silicone tags forchlorinated and brominated compounds were evaluated againstfish tissue concentrations (on a lipid weight basis) and literaturevalues of lipid-silicone partition coefficients (Klip‑sil). Fugacityratios for chlorinated substances and for PAHs calculated fromsilicone exposed to the fish and in water were used to estimatethe degree of “equilibrium” established between a compoundsconcentration in fish and in water.

■ MATERIAL AND METHODS

Solvent and Standards. Ultrapure water was from anElgastat Maxima HPLC Deionization option 3 system. HPLC-grade dichloromethane and pentane were from Rathburn.Chromasolv-quality isopropanol was from Riedel-de-Haen andHPLC grade cyclohexane was from J.T. Baker. Standards forPAHs and their deuterated homologues from Chiron were ofanalytical-grade with purities of >99% for PAHs and >99.5% fordeuterated PAHs. Analytical-grade standards and surrogatestandards for organochlorinated compounds (OCs) andPBDEs were from LGC/Promochem. The full list of chemicalsunder study is given in Supporting Information (SI) Table S1.

Silicone Tag and Strip Preparation. Medical gradeSilastic A tubing (4.9 mm external diameter, 2.6 internaldiameter) purchased from Cole-Parmer/Tekmo was used toprepare 2.5 cm long tags. The silicone tubing was Soxhletextracted with ethyl acetate for 24 h prior to use. The siliconetags were then further cleaned in methanol before being kept inthe freezer at −20 °C until use. A total of 25 tags were preparedand their mass was on average 0.382 g (1.5% RSD). Tags werenot spiked with performance reference compounds.AlteSil silicone strips (2.5 cm wide, 60 cm long and 0.5 mm

thick) from Altec Products Ltd. were also Soxhlet extracted for24 h prior to use. Samplers were further soaked in methanolbefore spiking with performance reference compounds(deuterated PAHs). Spiking of PRC was performed accordingto procedures previously published using a methanol:watersolution.19,20 PRCs with concentrations between 1 and 6 μgsampler−1 (relative standard deviation <10%) used wereacenaphthene-d10, fluorene-d10, phenanthrene-d10, fluoran-thene-d10, chrysene-d12, and benzo[a]pyrene-d12. Once spiked,samplers were stored in clean metal containers at −20 °C untiluse.

Fieldwork and Field Site Description. This study wasconducted in 2012 in the River Alna in Oslo, a river originatingfrom a lake used for drinking water supply purposes. The uppersection of the river where fish were taken from is relativelyclean, whereas lower reaches (7−8 km downstream) where fishexposures were undertaken are polluted as a result of landfilland stormwater effluents and old industrial contamination withPCBs present in sediments.21

Environmental Science & Technology Article

dx.doi.org/10.1021/es401810r | Environ. Sci. Technol. 2013, 47, 11660−1166711661

Silicone strips were exposed in the River Alna close to thelocation of the fish caging. Triplicate samplers were exposed inthe river for 28 days. Field control samplers were used (duringdeployment and retrieval procedures) to evaluate possiblecontamination during handling and transport and to measurestarting PRC concentrations. Once retrieved from the river,silicone strips were kept frozen at −20 °C until extraction.Initial cleaning of the sampler surface took place in the riverduring retrieval.An application was approved by the Norwegian Animal

Research Authority (License ID 4156, 2012/54789). Electricfishing (Geomega FA3, 170−1500 V, 35−70 Hz, pulsed-DC)was used to obtain a sample of 21 brown trout (Salmo trutta)with minimum total length and weight of 13 cm and 24 g,respectively (Table 1, SI Figure S1). Fish were placed in a cagein the river until surgery. Prior to processing, fish wereindividually anesthetised in a benzocaine solution (1 mL L−1).Once sedated, the fish were weighed and measured for lengthand a small incision was made through the ventral body wallusing a scalpel blade. The silicone tag was then inserted into theperitoneal cavity (coleum) of the fish, between the ventral andanal fins. It was not necessary to close the incision with sutures.An example of placement of the tag in the fish cavity is shownon SI Figure S2. Blank silicone tag were manipulated in asimilar way to those inserted into the fish (manuallymanipulated for a few seconds and exposed to the air duringfish tagging). Passive integrated transponder (PIT) tags (8 ×1.4 mm, Oregon RFID) were injected into the muscle of thefish near the dorsal fin for identification purposes. Oncompletion of processing, fish were returned to an aeratedbucket filled with river water. Fish were then transported to theexposure site where they were placed in three cylindrical cages(80 cm long and 40 cm wide) for 28 days exposure. Fish didnot display any particular annoyance from tag implantation.After exposure all fish were sacrificed with a blow to the headand brought back to the laboratory where the silicone tags wereretrieved and fish were sampled.Silicone Tag and Strip, Fish, Liver, and Muscle Tissue

Extraction. Silicone tags and strips were extracted andanalyzed for PAHs and OCs. Biota samples were extractedand analyzed for OCs. Biota samples from three fish and theirrespective tags were also analyzed for PBDEs.The surface of water-exposed passive samplers was

thoroughly cleaned in the laboratory by rinsing with ultrapurewater and then dried using a clean paper tissue. Samplers weresoaked overnight in pentane (150 mL). Surrogate standards forPAHs (100 ng of d8-naphthalene, d10-biphenyl, d8-acenaph-thylene, d8-dibenzothiophene, d10-pyrene, d12-benz[a]-anthracene, and d12-perylene), and OCs (50 ng of PCB 30,53 and 204), were added. This step was repeated with freshpentane. Extracts were combined and reduced under a gentlestream of nitrogen to about one mL. Extracts were furtherreduced before PAH, PRC, and PCB analysis by GC/MS.

Silicone tags were removed from the fish, cleaned with atissue, and briefly rinsed/wiped with water and methanol toremove remaining fish tissue at the surface of the tags. Tagswere dried and stored at −20 °C until extraction. Surrogatestandards for PAHs, OCs (as above), and PBDEs (4.5 ng ofBDE 30, 119, and 9 ng of 181 and 13C-BDE 209) were addedand tags were soaked twice overnight with pentane (20 mL).Pentane extracts were reduced and the solvent changed tocyclohexane prior to analysis. Extracts were then furtherreduced prior to analysis. PAHs and OCs were analyzeddirectly, while extracts of tags from three fish were furthertreated with sulphuric acid and acetonitrile prior to PBDEsanalysis.Biota samples were homogenized with a PRO 250

homogenizer from Pro Scientific. For the small fish, the entirefish was homogenized while larger fish were gutted and muscleand liver were homogenized separately. All three types of biotasamples (whole fish, liver, and muscle, 5 g) were extracted usinga mixture of cyclohexane (15 mL) and isopropanol (15 mL).Samples were shaken for two hours on an orbital shaker at 200rpm. The solvent was then decanted and fresh solvent wasadded for a second extraction (shaking for another two hours).Solvent extracts were combined and water (10 mL) was addedto yield an organic cyclohexane phase. The organic phase wasseparated from the aqueous phase and the cyclohexane wasevaporated under a gentle stream of nitrogen to obtain the totalamount of fat extracted. The proportion of fat or lipid in thesamples was measured gravimetrically, bearing in mind that notall polar lipids may be extracted with this procedure. The lipidphase was dried, weighed, and transferred to smaller glasswarewhere cyclohexane and surrogate standards for OCs and BDEswere added (same standards and amounts as for siliconesamplers). Sulphuric acid (5 mL) was added to digest excess fatand the extracts were then evaporated under nitrogen andtransferred to GC-vial for analysis.

Passive Sampler and Biota Extract Analysis. Analysisfor PAHs (and PRCs), and OCs (9 PCB congener,pentachlorobenzene, hexachlorobenzene, and p,p′-DDE) wason a Agilent 7890A gas chromatograph (GC) linked to anAgilent 5975c inert XL EI/CI mass spectrometer (MS)operated in single ion monitoring mode (SIM) with electronimpact ionization (70 keV). Analysis for PBDEs was with aHewlett-Packard 6890Plus GC linked to a Hewlett-Packard5973 MS detector operated in negative chemical ionization(with methane) and SIM mode. A detailed description ofanalytical procedures can be found in the SI. Limits of detectionfor compounds of interest in silicone tags and strips and biotaare given in SI Table S1. The biota method is accredited and isbeing used in national monitoring programmes in Norway. Ourlaboratory uses spiked samples, certified reference materials andregularly participates in QASIMEME interlaboratory compar-isons. Recoveries of chlorinated and brominated compoundsfrom biota samples are consistently in the range of 80−120%.For passive samplers (tags and sheets), recoveries are between

Table 1. Characteristics of the 21 Caged and Tagged Fish of This Study

lipid content (%)c

n weight (g)a length (cm) weight loss (%)b fish liver muscle

whole fish 13 39 (26−286) 15.3 (13−30) 14 (3.8−22) 2.2 (0.7−4.4)liver/muscle 8 116 (24−188) 23 (13−27) 14 (0−20) 1.7 (1.3−4.3) 0.7 (0.4−0.7)

aweight of the whole fish before caging. bWeight loss of the fish during 28 day caging (%). cLipid content (weight weight basis) of the samplesmeasured during extraction and analysis of fish, liver and muscle samples.



90 and 110% for PAHs and chlorinated and brominatedcompounds.Passive Sampling Data Handling. The lipid content of

the fish used in this study were relatively low, therefore minimallipid absorption may be expected. Silicone tags were notweighed following exposure, so it is not possible to knowgravimetrically whether significant lipid absorption took place.Nonetheless contaminant concentrations in silicone tags werenot corrected for potential influence of absorbed lipids on lipid-silicone partition coefficients.6 Measurements in the fish(reported on a lipid weight basis) and in the silicone tagsenabled the estimation of contaminant distribution coefficients,Dlip‑sil (g g−1) between fish/fish tissues (on a lipid basis) andsilicone tags.For silicone strips exposed to the river water, a boundary

layer-controlled uptake model was used for estimating sampling

rates, Rs (from PRCs) and freely dissolved contaminantconcentrations in water, Cw from amounts accumulated insilicone strips.22 The unweighted nonlinear least-squaresmethod to estimate Rs from multiple performance referencecompounds23 was used to obtain values for the adjustableparameter βsil.

22 This parameter was then used to deduce Rs forchemicals with logKow > 3.5, according to Rs = βsil Ksw

−0.08, withKsw the sampler-water partition coefficient for AlteSilsilicone.22,24 More details are given in SI.The calculation of fugacity ratios through the comparison of

contaminant concentrations in silicone tags from in vivoexposure (through implantation) with water-exposed siliconesamplers may be a useful approach to assess contaminantbioaccumulation in an organism. Equilibrium between theconcentration of contaminants in water-exposed siliconesamplers and that in water is generally not reached for

Figure 1. Lipid-silicone distribution coefficients (Dlip‑sil, g g−1) for selected organochlorine compounds estimated from internal and in vivo silicone

exposure in 13 brown trout (Salmo trutta). Box boundaries indicate 25th and 75th percentiles, the horizontal line within the box is for the medianand whiskers indicate the 90th and 10th percentiles. Dots are for outliers.

Table 2. Lipid-Silicone Distribution Coefficients, Dlip‑sil (g g−1) Measured with In Vivo Implantation of Brown Trout (WholeFish, Liver and Muscle, This Study), and Reported (Literature) Lipid-Silicone Partition Coefficients, Klip‑sil for In Vitro LipidPhase-Based Batch Experiments and In Vitro Fish Muscle-Exposure Measurements

Dlip‑sil, this study (brown trout)b Klip‑sil, from literature

analyte logKowa whole fish liver muscle Genericc Norwegian Atlantic salmond Baltic Sea eeld Finnish lake eeld

PeCB 5.17 15.8 (1.6) 18.5 (5.6) 22.6 (3.8)HCB 5.50 13.9 (1.2) 12.2 (1.4) 13.1 (1.6) 13.5CB28 5.67 13.6 (2.1) 11.5 (2.7) 10.7 (2.1) 16.0 15.2 15.0 27.9CB52 5.84 13.8 (1.3) 8.2 (3.1) 9.7 (2.0) 16.0 18.0 14.2 31.4CB101 6.38 17.4 (1.8) 8.2 (1.0) 14.3 (3.1) 20.7 21.7 17.6 35.2CB118 6.74 22.8 (2.0) 12.1 (2.2) 18.5 (3.6) 26.0 26.7 20.4 38.2CB153 6.92 35.8 (4.9) 18.2 (2.7) 27.6 (5.7) 27.4 32.4 24.4 33.0CB105 6.65 27.2 (3.0) 25.7CB138 6.83 34.0 (5.3) 18.1 (3.5) 27.2 (5.9) 25.5 28.0 22.3 32.0CB156 7.18 40.0 (4.4) 32.6 (11.2) 33.0 (8.5) 32.3p,p′-DDE 5.69 23.5 (3.2) 9.6 (2.8) 14.3 (2.6) 19.8BDE28 5.98 4.6 (1.2)BDE47 6.60 27.2 (9.2)BDE100 7.22 38.7 (1.0)BDE99 7.22 53.5 (14)

aSee SI Table S13 for references for logKow values bMedian value for chlorinated compounds from 13 fish for the “whole fish”, and 8 fish for liverand muscle data. For PBDEs, “whole fish” median values are for three fish only. Median absolute deviations are given in brackets. cGeneric in vitrolipid phase-based Klip‑sil from based on the average of data for olive oil, Tobis oil and seal oil.6 dIn vitro tissue-based Klip‑sil for three different types offish.8



substances with logKow > 5−6 over commonly used monthlyexposures.20,25 For these substances, equilibrium concentra-tions in silicone that can be anticipated if samplers were left fora sufficiently long period of time are easily calculated fromestimated Cw and Ksw values.24 Despite differences in logKswvalues for the two silicones (from different suppliers for the tagsand sheets) generally not more than 0.1 log unit for PAHs andPCBs,24 data were corrected for these differences.

■ RESULTS AND DISCUSSIONWhole Fish−Silicone Distribution Coefficients for

Chlorinated Compounds. To assess the relevance of internalchemical fugacity of these substances absorbed in the browntrout and measured with implanted silicone tags, we comparedfish concentrations on a lipid weight basis with those in thesilicone tags (data in SI Tables S2 and S3). “Whole fish” lipid-silicone distribution coefficients, Dlip‑sil (g g−1) were measuredfor 13 fish and data are given in the boxplot of Figure 1. Dataare only given for those above limits of quantification. Dlip‑silvalues for PCBs and chlorinated organics are in the range 10 to50 g g−1. Median values for Dlip‑sil (n = 13) presented in Figure1 are given in Table 2 (together with median absolutedeviations). A very low variability in Dlip‑sil can be observed forsubstances such as HCB and CB28 up to CB118. Medianabsolute deviations (n = 13) were generally between 10 and20% of median values for all substances listed in Table 1. Thelowest median absolute deviations were for HCB and CB52.On average, the RSD was close to 20% for all substances. Theincrease in Dlip‑sil from CB28 to CB118 was observed to be, at amaximum, a factor of 2 for an increase in logKow of about onelog unit. More variability was seen for CB153 to CB156 (Figure1) and only minor increases in Dlip‑sil can be observed. CB180was not found in the silicone tags while PeCB was close tolimits of quantification. These values are in excellent agreementwith in vitro Klip‑sil from Jahnke et al. (2008)6 who measuredthe partitioning of organochlorinated compounds betweenlipids and polydimethylsiloxane (PDMS) immersed in threelipids of differing composition and origin. Since relativelysimilar partitioning to the three types of lipids (olive, Tobis,and seal oils) was observed by these researchers, we used theirgeneric (average) values for comparison. Values uncorrected forlipid uptake (Dlipid,pdms in Jahnke et al., 20086) were selectedand are given in Table 2 (but termed Klip‑sil here). For example,the median in vivo whole fish Dlip‑sil for brown trout for HCB isin close agreement with the generic in vitro Klip‑sil of 13.5 g g

−1.For the remaining chlorinated compounds of Table 2, thedeviation between median Dlip‑sil for brown trout and genericKlip‑sil

6 was on average below 20% (range of 6−34%). In vitroequilibrium passive sampling measurements of PCBs have alsobeen undertaken by direct insertion of the PDMS phase intothe fish tissue.8 Resulting in vitro Klip‑sil for Norwegian Atlanticsalmon, Baltic Sea eel and Finnish lake eel calculated fromJahnke et al. (2011)8 are also provided in Table 2. Values ofDlip‑sil for brown trout are in closest agreement with Klip‑sil forthe Norwegian Atlantic salmon and the Baltic Sea eel, withdeviations on average of 17 and 21%, respectively. For theFinnish lake eel, Klip‑sil for CB28 to CB118 were consistently50% higher than our results for brown trout. Differences wereminimal however for CB153 and C138.To our knowledge, lipid-silicone distribution coefficients for

PBDEs have not been reported in the literature. Here, theconcentration of polybrominated diphenyl ether congeners wasmeasured in tags and whole fish homogenates for three

randomly selected fish (SI Tables S4 and S5). Median values ofDlip‑sil for PBDEs are given in Table 2. The variability of in vivoDlip‑sil for BDE28 (median = 4.6 g g−1; range 3.4−18 g g−1) forthe three fish is higher than that found for PCBs with a similarhydrophobicity (Table 2). Median Dlip‑sil were 28 (18−38), 39(31−40) and 54 g g−1 (31−68) for BDE47, BDE100, andBDE99, respectively. Logarithms of lipid-silicone distributioncoefficients (logKlip‑sil) for PBDEs are in agreement with thosefor OCs when plotted as a function of logKow (SI Figure S3). Itis interesting to note that in vivo Dlip‑sil values are relativelyconstant for the range of hydrophobicity from logKow of 5.0 toover 6.0. A mild increase in lipid-silicone distributioncoefficients can be observed from logKow of 6 to 7.2.Our data was not corrected for possible amounts of lipid

absorbed by the silicone and comparison were made withuncorrected in vitro Klip‑sil from Jahnke et al. (2008).6 Thereason for this is that PDMS-absorbed lipids may notnecessarily have an influence on Klip‑sil.

26 Some uncertaintycan also result from potential difference in the capacity ofpolydimethylsiloxane or silicones from different sources toabsorb neutral hydrophobic compounds.24

Fish−Silicone Tag Equilibrium. An important question iswhether the silicone tags have reached some form ofequilibrium with surrounding tissues. Assuming a case wheremost of the resistance to mass transfer into the silicone tag is inthe polymer, approximate times needed for silicone tags toequilibrate would be in the range of 1−5 days for PCBsaccording to a simple version of mass transfer model (see SIFigure S4). This means that, for our in vivo measurements,times to equilibrium may be longer than that.The measurement of in vivo Dlip‑sil through implantation is in

excellent agreement with published data and tends to supportthe equilibrium nature of the sampling undertaken here. Theagreement between in vivo Dlip‑sil and literature values of invitro Klip‑sil from lipid-immersed silicone assays (e.g., forhexachlorobenzene) together with the lower variability ofwhole fish concentrations when normalized to the lipid content(not shown) tends to show that lipids were the maincomponent responsible for fugacity capacity in our browntrout. Correlation coefficients for lipid-normalized whole fish−silicone tag concentrations are consistently higher than thosebased on a wet weight basis (SI Table S6). This supports thenotion that concentrations in fish (lipid-normalized) and insilicone tags are likely to be related. While for lipid-rich tissues,“in tissue” exposures are expected to reach equilibrium within afew hours, equilibration times over seven days have beenobserved for less lipid-rich tissues.9 True equilibrium is difficultto define here since the fish were thinning (Table 1) duringexposure meaning that the system was dynamic.While the brown trout showed a relatively low overall lipid

content (2.2% wet weight basis) after exposure, the 14% bodyweight loss over the 28 days caging (Table 1) means that thelipid content and composition at the start of the experimentmay have been different due to lipid mobilization duringstarving.27 Starvation of seawater-reared brown trout for aperiod of eight weeks has been shown to result in a 10% bodyweight loss.28 In that study, losses of both lipids and proteinwere observed.28 Lipid losses may help remobilize contami-nants within the fish and this can affect internal equilibrium andthe time needed for tags to equilibrate with the fish. Theproportion of lipids in visceral tissues was shown to besignificantly higher than in muscle tissues or in the liver.28 Thismeans silicone tags implanted in the ventral part of the fish



were likely exposed to tissues with a higher lipid content(viscera and belly flap) than if exposed to the muscle or liver.This means that time to equilibrium may be shorter than thoseexpected for in vitro in tissue exposures to tissues of low lipidcontent. As opposed to silicone phases exposed to nonlivingmatrices, blood circulation and organism movements during invivo exposures are likely to play an important role in the timeneeded for an inserted reference phase such as silicone to cometo equilibrium with its surrounding tissues. Results fromintraperitoneal injection of PCB-spiked oil into Round Gobysuggested a rapid intertissue distribution of PCBs in these smallfish.29

Muscle- and Liver-Silicone Distribution Coefficients.Liver- and muscle-based Dlip‑sil (SI Table S7, S8 and S9) wereobtained for the 8 largest fish (median values for OCs given inTable 2). The range of Dlip‑sil measured for fish muscle overlapswell with values from “whole fish” measurements for mostsubstances under study. Ranges of values observed for muscletissues are wider than “whole fish” data. Liver-siliconedistribution coefficients are for most substances, lower thanthose for muscle tissues (Figure S5 in the SI). Mostpronounced differences are for CB101, CB118, CB153 andCB138. Concentrations in fish were the lowest and closest tolimits of detection for CB156 and this may be why data for thiscongener show high variability. For HCB, CB28 and CB52,Dlip‑sil values are in close agreement. For example, median Dlip‑silvalues for HCB are 13.9, 13.1, and 12.2 for “whole fish”, muscleand liver, respectively (Table 2). As shown on SI Figure S1, fishused for muscle and liver measurement were larger than thosefor “whole fish” analysis. The good agreement between in vivoDlip‑sil obtained for small and larger fish (muscle) indicate thatthe measurement is independent of the fish size.It is perhaps not surprising to observe the closest overlap of

Dlip‑sil between the whole fish data and the muscle data. Despitea higher lipid content of the liver samples (Table 1), theproportion of muscle tissues (and lipids) in the fish comparedwith that of the liver is much higher. Once inserted into thefish, the silicone tag was positioned in the peritoneal cavity,near the viscera and close to both muscle tissues and the liver.Despite aiming to implant all tags from the same incision point,these were not necessarily located in exactly the same positionin all fish and this may have affected the variability observedbetween paired fish-tag data. It may be that the lipidcomposition of the muscle tissues and of the liver differ,particularly after fasting. For example, saturated fatty acidmobilization upon starving appear to take place preferentially inthe viscera for rainbow trout.27 Different lipids may also exhibitdifferent capacities to absorb contaminants resulting indifferences in in vivo Dlip‑sil.

2 In our study the lipid content ofwhole fish, liver and muscle was measured during the matrixextraction for contaminant analysis. It is a relatively crudesurrogate of the actual lipid content and precludes furtherdetailed discussion of the reasons for differences in lipid-normalized contaminant concentrations in liver and muscletissues.Fish−Water Fugacity Ratios. Contaminant concentra-

tions in the silicone tags (assumed to be at equilibrium), ameasure of the fugacity of chemicals in the fish, can becompared with those found in the silicone sheets at equilibriumwith freely dissolved concentrations in water. Details of thecalculation of freely dissolved concentration and estimation ofequilibrium concentrations in silicone (and their associateduncertainties) are given in SI. The exposure time was not

sufficient to observe equilibrium between contaminantconcentrations in the silicone sheets and the freely dissolvedconcentration in water, particularly for analytes with logKow > 5(see SI Table S11). The concentration at equilibrium could,however, be calculated from the estimated freely dissolvedconcentrations in water and the sampler-water partitioncoefficients (Ksw). Estimated freely dissolved concentrationsfor selected PAHs and OCs are given in Table S11 in the SI.Concentrations of OCs tend to be similar to those measuredthe previous year at this site in the Alna while PAHconcentrations were substantially lower than in 2011.21 PAHconcentrations in silicone tags are given in SI Table S12.Fugacity ratio equivalents were calculated as the ratios of

equilibrium concentrations measured in the 13 fish-exposedsilicone tags (data not corrected for possible lipid content) overthose estimated for the silicone sheets exposed to the water.Results are shown in Figure 2 for PAHs and OCs. For these

lipophilic compounds that partition preferentially to the fishlipids, ratios close to one would tend to indicate that theconcentration in the fish is close to steady-state, or apparentequilibrium with the freely dissolved concentration in water.Ratios falling below 1 indicate that the fish is “under-concentrated” with respect to the water phase. This canoccur as a result of low accumulation rates (far fromequilibrium), biotransformation or dilution (e.g., throughgrowth). Ratios above one would represent cases ofbiomagnification or rapid weight loss. Spatiotemporal incon-sistencies in wild fish and passive sampler exposures could alsoresult in ratios deviating from one.30 Ratios for most PCBcongeners, HCB, PeCB and some PAHs such as anthracene,dibenzothiophene, phenanthrene, and fluorene were generallyclose to one. The excellent agreement of in vivo Dlip‑sil withliterature values of Klip‑sil and fish−water fugacity ratios close toone tend to indicate that the concentration of thesecontaminants in the fish is in apparent equilibrium with thatin the water. Fish−water fugacity ratios for acenaphthene andnaphthalene suggest that the fish are “over-concentrated” forthese two substances. It is likely that fish and water-exposedsilicone sheets are representative of different ambient freely

Figure 2. Fugacity ratios calculated as the ratios of equilibrium siliconeconcentrations measured in fish-exposed tags (n = 13 “whole fish”)and in water-exposed sheets. Data points represent individual tagconcentrations divided by the mean of triplicate water-exposed siliconesheet measurements.



dissolved concentrations in water for these compounds. Sincethe concentration of moderately hydrophobic substances suchas naphthalene in silicone sheet reaches equilibrium with that inwater very fast, data from silicone sheets is representative onlyof the last few days of exposure. Fugacity ratios for fluorantheneand pyrene are over an order of magnitude below one. This isnot surprising since fluoranthene and pyrene are efficientlymetabolized and have low half-lives in fish.31 Ratios for pyreneare close to an order of magnitude lower than those observedfor fluoranthene, and this is in agreement with the fishbiotransformation half-life of 0.556 and 2.57 days for pyreneand fluoranthene given by the BCFBAF v3.01 (EPI Suitev4.10).31 Of all PAHs detected and quantified in the siliconetags, fluoranthene and pyrene are the only ones whose fugacityratios appear to be affected by metabolism in the fish. The lesshydrophobic PAHs are generally close to the 1:1 relationship.Recent modeling32 showed that steady-state bioconcentrationfactors (BCFs) for substances with logKow > 5.0 are expected tobe significantly influenced by hepatic biotransformation. Incontrast these authors explained that hepatic biotransformationrates lower than rates of chemical elimination through fish gillsfor chemicals with logKow < 3 have negligible impact on BCFvalues. This means that hepatic transformation may have alimited impact on BCF values for these compounds.32 Themeasured fish−water ratio of equilibrium silicone concentrationfor pyrene of 0.003 is in agreement with pyrene fugacity ratiossummarized in Burkhard et al. (2011)33 generally well belowone for fish.Implications for Future Studies. In vivo equilibrium

passive sampling of neutral hydrophobic compounds can beused to measure the internal fugacity or activity in fish andother organisms.17 This measure of internal fugacity can becompared with the fugacity of the chemical in other (a)bioticcompartments measured with the same polymer (with the samefugacity capacity). Such fugacity ratios can provide relevantinformation on contaminant bioaccumulation potential withinan organism, accounting for all components contributing to theorganism’s contaminant absorption capacity (eq 3). Such amethod has the potential to be applied to organisms with lowerlipid content, although organism size and practical experimentaltime frame and time to equilibrium will remain crucial factorsthat require further investigations. When investigating organ-isms at different trophic levels across food webs, in vivosampling may need to be combined with in vitro tissue-basedmeasurements since not all organisms are amenable toimplantation. The PRC procedure may be developed in thefuture to assess equilibrium or steady-state conditions betweenthe organism and the polymeric phase. This way, tissue-siliconedistribution of “native”/naturally accumulated contaminantsand PRCs may be compared. The interpretation of results from“whole organism” ecotoxicological testing, BCF/BAF measure-ments, or caged organism exposures may benefit from ameasurement of contaminant activity/fugacity inside theorganism. This in vivo study confirms the validity of literatureK or Dlip‑sil values and supports their use for the calculation ofabiotic measures of lipid-normalized contaminant concentra-tions in biota based on passive sampling in water or sedimentsfor comparison with actual biota concentrations.

■ ASSOCIATED CONTENT*S Supporting InformationList of chemicals of interest in this study, further details andlimits of detection for the various analyses, fish length and

weights, fish tag information, raw data for OCs and PBDEsconcentrations in silicone tags and biota samples, modeling oftime to equilibrium for tags under polymer controlled-uptake,comparison of OC concentrations in fish liver and musclesamples, calculation of freely PAH and OC dissolvedconcentrations and their uncertainty, PAH concentrations intags and logKow values for OCs and PBDEs and their sources.This information is available free of charge via the Internet athttp://pubs.acs.org/.

■ AUTHOR INFORMATIONCorresponding Author*(I.J.A.) Phone: +47 22 18 5100; fax: +47 22 18 5200; e-mail:[email protected].

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSAlfhild Kringstad and Etienne Vermeirssen are thanked forsupport with the analytical work for this study and fordiscussions of the fish caging, respectively. We thank theNorwegian Research Council for funding this work throughNIVA’s basic research funding for innovation. We also thankfour anonymous reviewers for comments that helped improvethis manuscript.

■ REFERENCES(1) Randall, R. C.; Lee, H.; Ozretich, R. J. Evaluation of selected lipidmethods for normalizing pollutant bioaccumulation. Environ. Toxicol.Chem. 1991, 10 (11), 1431−1436.(2) van der Heijden, S. A.; Jonker, M. T. O. Intra- and interspeciesvariation in bioconcentration potential of polychlorinated biphenyls:Are all lipids equal? Environ. Sci. Technol. 2011, 45 (24), 10408−10414.(3) Endo, S.; Bauerfeind, J.; Goss, K. U. Partitioning of neutralorganic compounds to structural proteins. Environ. Sci. Technol. 2012,46 (22), 12697−12703.(4) Mackay, D. Correlation of bioconcentration factors. Environ. Sci.Technol. 1982, 16 (5), 274−278.(5) Mackay, D. Finding fugacity feasible. Environ. Sci. Technol. 1979,13 (10), 1218−1223.(6) Jahnke, A.; McLachlan, M. S.; Mayer, P. Equilibrium sampling:Partitioning of organochlorine compounds from lipids into poly-dimethylsiloxane. Chemosphere 2008, 73 (10), 1575−1581.(7) Ossiander, L.; Reichenberg, F.; McLachlan, M. S.; Mayer, P.Immersed solid phase microextraction to measure chemical activity oflipophilic organic contaminants in fatty tissue samples. Chemosphere2008, 71 (8), 1502−1510.(8) Jahnke, A.; Mayer, P.; Adolfsson-Erici, M.; McLachlan, M. S.Equilibrium sampling of environmental pollutants in fish: Comparisonwith lipid-normalized concentrations and homogenization effects onchemical activity. Environ. Toxicol. Chem. 2011, 30 (7), 1515−1521.(9) Jahnke, A.; Mayer, P.; Broman, D.; McLachlan, M. S. Possibilitiesand limitations of equilibrium sampling using polydimethylsiloxane infish tissue. Chemosphere 2009, 77 (6), 764−770.(10) Jin, L.; Gaus, C.; van Mourik, L.; Escher, B. I. Applicability ofpassive sampling to bioanalytical screening of bioaccumulativechemicals in marine wildlife. Environ. Sci. Technol. 2013, 47 (14),7982−7988.(11) Vrana, B.; Mills, G. A.; Allan, I. J.; Dominiak, E.; Svensson, K.;Knutsson, J.; Morrison, G.; Greenwood, R. Passive samplingtechniques for monitoring pollutants in water. TrAC, Trends Anal.Chem. 2005, 24 (10), 845−868.(12) Mayer, P.; Tolls, J.; Hermens, L.; Mackay, D. Equilibriumsampling devices. Environ. Sci. Technol. 2003, 37 (9), 184A−191A.



http://pubs.acs.org/

mailto:[email protected]

(13) Huckins, J. N.; Petty, J. D.; Booij, K. Monitors of OrganicChemicals in the Environment: Semipermeable Membrane Devices;Springer: New York, 2006.(14) Ouyang, G. F.; Oakes, K. D.; Bragg, L.; Wang, S.; Liu, H.; Cui, S.F.; Servos, M. R.; Dixon, D. G.; Pawliszyn, J. Sampling-rate calibrationfor rapid and nonlethal monitoring of organic contaminants in fishmuscle by solid-phase microextraction. Environ. Sci. Technol. 2011, 45(18), 7792−7798.(15) Zhou, S. N.; Oakes, K. D.; Servos, M. R.; Pawliszyn, J.Application of solid-phase microextraction for in vivo laboratory andfield sampling of pharmaceuticals in fish. Environ. Sci. Technol. 2008, 42(16), 6073−6079.(16) Adolfsson-Erici, M.; Akerman, G.; McLachlan, M. S. In-vivopassive sampling to measure elimination kinetics in bioaccumulationtests. Chemosphere 2012, 88 (1), 62−68.(17) Allan, I. J.; Bæk, K.; Kringstad, A.; Roald, H. E.; Thomas, K. V.Should silicone prostheses be considered for specimen banking? Apilot study into their use for human biomonitoring. Environ. Int. 2013,59, 462−468.(18) Maenpaa, K.; Leppanen, M. T.; Reichenberg, F.; Figueiredo, K.;Mayer, P. Equilibrium sampling of persistent and bioaccumulativecompounds in soil and sediment: Comparison of two approaches todetermine equilibrium partitioning concentrations in lipids. Environ.Sci. Technol. 2011, 45 (3), 1041−1047.(19) Booij, K.; Smedes, F.; van Weerlee, E. M. Spiking ofperformance reference compounds in low density polyethylene andsilicone passive water samplers. Chemosphere 2002, 46 (8), 1157−1161.(20) Allan, I. J.; Harman, C.; Kringstad, A.; Bratsberg, E. Effect ofsampler material on the uptake of PAHs into passive sampling devices.Chemosphere 2010, 79 (4), 470−475.(21) Allan, I. J.; Ranneklev, S. B. Occurrence of PAHs and PCBs inthe Alna River, Oslo (Norway). J. Environ. Monit. 2011, 13 (9), 2420−2426.(22) Rusina, T. P.; Smedes, F.; Koblizkova, M.; Klanova, J.Calibration of silicone rubber passive samplers: Experimental andmodeled relations between sampling rate and compound properties.Environ. Sci. Technol. 2010, 44 (1), 362−367.(23) Booij, K.; Smedes, F. An Improved Method for Estimating insitu sampling rates of nonpolar passive samplers. Environ. Sci. Technol.2010, 44 (17), 6789−6794.(24) Smedes, F.; Geertsma, R. W.; van der Zande, T.; Booij, K.Polymer-water partition coefficients of hydrophobic compounds forpassive sampling: Application of cosolvent models for validation.Environ. Sci. Technol. 2009, 43 (18), 7047−7054.(25) Allan, I. J.; Booij, K.; Paschke, A.; Vrana, B.; Mills, G. A.;Greenwood, R. Field performance of seven passive sampling devicesfor monitoring of hydrophobic substances. Environ. Sci. Technol. 2009,43 (14), 5383−5390.(26) Jahnke, A.; Mayer, P. Do complex matrices modify the sorptiveproperties of polydimethylsiloxane (PDMS) for non-polar organicchemicals? J. Chromatog., A 2010, 1217 (29), 4765−4770.(27) Jezierska, B.; Harzel, J. R.; Gerking, S. D. Lipid mobilizationduring starvation in the rainbow trout, Salmo gairdneri Richardson,with attention to fatty acids. Journal of Fish Biology 1982, 21, 681−692.(28) Regost, C.; Arzel, J.; Cardinal, M.; Laroche, M.; Kaushik, S. J.Fat deposition and flesh quality in seawater reared, triploid browntrout (Salmo trutta) as affected by dietary fat levels and starvation.Aquaculture 2001, 193 (3−4), 325−345.(29) O’Neil, J. A.; Johnson, T. B.; Drouillard, K. G. Validation ofrapid assimilation of PCBs following IP dosing in the round goby(Neogobius melanostomus). Bull. Environ. Contam. Toxicol. 2013, 91(2), 135−140.(30) Selck, H.; Drouillard, K.; Eisenreich, K.; Koelmans, A. A.;Palmqvist, A.; Ruus, A.; Salvito, D.; Schultz, I.; Stewart, R.; Weisbrod,A.; van den Brink, N. W.; van den Heuvel-Greve, M. Explainingdifferences between bioaccumulation measurements in laboratory andfield data through the use of a probabilistic modeling approach. Integr.Environ. Assess. Manage. 2012, 8 (1), 42−63.

(31) Arnot, J. A.; Mackay, D.; Parkerton, T. F.; Bonnell, M. Adatabase of fish biotransformation rates for organic chemicals. Environ.Toxicol. Chem. 2008, 27 (11), 2263−2270.(32) Nichols, J. W.; Huggett, D. B.; Arnot, J. A.; Fitzsimmons, P. N.;Cowan-Ellsberry, C. E. Towards improved models for predictingbioconcentration of well-metabolized compounds by rainbow troutusing measured rates of in vitro intrinsic clearance. Environ. Toxicol.Chem. 2013, 32 (7), 1611−1622.(33) Burkhard, L. P.; Arnot, J. A.; Embry, M. R.; Farley, K J; Hoke, R.A.; Kitano, M.; Leslie, H. A.; Lotufo, G. R.; Parkerton, T. F.;Sappington, K. G.; Tomy, G. T.; Woodburn, K. B. Comparinglaboratory and field measured bioaccumulation endpoints. Integr.Environ. Assess. Manage. 2011, 8 (1), 17−31.



Documents

In Vivo Passive Sampling of Nonpolar Contaminants in Brown Trout ( Salmo trutta )