Download pdf - Long-term effects of a binary mixture of perfluorooctane sulfonate (PFOS) and bisphenol A (BPA) in zebrafish (Danio rerio)

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Aquatic Toxicology 118– 119 (2012) 116– 129

Contents lists available at SciVerse ScienceDirect

Aquatic Toxicology

jou rn al h om epa ge: www.elsev ier .com/ locate /aquatox

Long-term effects of a binary mixture of perfluorooctane sulfonate (PFOS) andbisphenol A (BPA) in zebrafish (Danio rerio)

Su. Keitera,∗, L. Baumanna, H. Färberb, H. Holbechc, D. Skutlarekb, M. Engwalld, T. Braunbecka

a Aquatic Ecology and Toxicology Group, Centre for Organismal Studies, University of Heidelberg, Im Neuenheimer Feld 504, D-69120 Heidelberg, Germanyb Institute for Hygiene and Public Health, University of Bonn, Sigmund-Freudstr. 25, D-53127 Bonn, Germanyc Institute of Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmarkd Man-Technology-Environment Research Centre (MTM), Department of Natural Science, University of Örebro, Fakultetsgatan 1, S-701 12 Örebro, Sweden

a r t i c l e i n f o

Article history:Received 21 January 2012Received in revised form 2 April 2012Accepted 3 April 2012

Keywords:Perfluorooctane sulfonateBisphenol ALong-termMixtureVitellogeninZebrafish

a b s t r a c t

Previous in vitro studies have reported the potential of perfluorooctane sulfonate (PFOS) to increasethe toxicity of other compounds. Given the complex nature of mixtures of environmental pollutants inaquatic systems together with the persistent and bioaccumulative properties of PFOS, this study aimedat evaluating the long-term effects and toxicity-increasing behavior of PFOS in vivo using the zebrafish(Danio rerio). Fish were maintained in flow-through conditions and exposed to single and binary mix-tures of PFOS and the endocrine disruptor bisphenol A (BPA) at nominal concentrations of 0.6, 100 and300 �g/L and 10, 200 and 400 �g/L, respectively. F1 and F2 generations were evaluated from 0 to 180days post-fertilization (dpf) and F3 generation was evaluated from 0 to 14 dpf. Survival was documentedin all generations, whereas growth, fecundity, fertilization rate, histological alterations (in liver, thyroidand gonads) and vitellogenin (Vtg) induction in males were evaluated for F1 and F2 generations. Datafor growth were collected at 30, 90 and 180 dpf and data for histological evaluations and Vtg induc-tion were analyzed at 90 and 180 dpf. No significant effects on survival were seen in the F1 generationin any treatment following 180 d exposure; however, in the F2 generation, 300 �g/L PFOS both aloneand in combination with BPA (10, 200 and 400 �g/L) induced 100% mortality within 14 dpf. PFOS (0.6and 300 �g/L) did not increase the Vtg-inducing potential of BPA (10, 200 and 400 �g/L) in a binarymixture. In contrast, binary mixtures with 300 �g/L PFOS suppressed the Vtg levels in F1 males at 90dpf when compared to single BPA exposures. Whereas the lowest tested PFOS concentration (0.6 �g/L)showed an estrogenic potential in terms of significant Vtg induction, Vtg levels were generally found todecrease with increasing PFOS-exposure in both F1 and F2 generations. In F1 generation, BPA-exposurewas found to increase Vtg levels in a concentration-dependent manner. Histological analyses of F1 and F2fish revealed hepatocellular vacuolization, predominantly in males, following PFOS-exposure both aloneand in combination with BPA. Hepatotoxicity by PFOS might explain the suppressed Vtg response seenin PFOS-exposed F1 and F2 males. PFOS-exposed fish also showed granulomas, mainly in the liver. Givenprevious reports of the immunosuppressive potential of PFOS, the granulomas could be a consequenceof a PFOS-induced reduction of the immune response potential. In conclusion, the hypothesis that thepresence of PFOS increases the endocrine potential of BPA could not be confirmed in zebrafish. Adverseeffects on liver structure and survival were only seen at concentrations well above ecologically relevantconcentrations; however, the decline in survival rates following PFOS-exposure seen over generationsagain documents the importance of long-term studies for the investigation of persistent environmentalpollutants.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

The frequent use of perfluorinated chemicals (PFCs) in indus-trial applications and domestic products has led to a continuous

∗ Corresponding author.E-mail address: [email protected] (Su. Keiter).

detection of PFCs in a wide range of environmental matrices on aglobal basis including aquatic systems (Giesy and Kannan, 2001;Taniyasu et al., 2003; Stock et al., 2007; Yeung et al., 2009) Perfluo-rooctane sulfonate (PFOS; C8F17SO3

−) has frequently been reportedas one of the most commonly detected PFCs in biotic and abi-otic samples (Martin et al., 2004; Ahrens et al., 2009; Naile et al.,2010; Li et al., 2011; Thompson et al., 2011). Due to the persistent,bioaccumulative and toxic (PBT) properties of PFOS (OECD, 2002), a

0166-445X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.aquatox.2012.04.003


Su. Keiter et al. / Aquatic Toxicology 118– 119 (2012) 116– 129 117

restriction for commercialization and use was implemented withinthe EU in 2008 (EU, 2006). PFOS was also added to the list of prioritysubstances to be controlled within the European Water Frame-work Directive (EU, 2008). However, due to a continued productionof PFOS and its precursors in non-EU member countries (UNIDO,2009), combined with the long-range transport potential of PFCs(Young et al., 2007; Dreyer et al., 2009), sustained global emissionof PFOS is likely to occur. The worldwide detection of PFOS in drink-ing water including Europe (Skutlarek et al., 2006; Ericson et al.,2009; Quinete et al., 2009; McLaughlin et al., 2011) further doc-umented the need to fully understand the toxic potentials of thiscompound.

Given the aquatic occurrence and persistent properties of PFOS,multi-generation studies of fish assessing ecologically relevantconcentrations are warranted (Oakes et al., 2005). To date, themajority of PFOS-related studies with fish have focused on acuteeffects, leaving chronic and reproductive effects largely unex-plored. Ankley et al. (2005) performed a partial life-cycle test withthe fathead minnow (Pimephales promelas) and observed deviat-ing sex steroid concentrations and histopathological alterationsin ovaries following exposure to 300 �g/L PFOS. The discoveryof such sublethal effects holds great ecological importance. Yet,the continued lack of studies over multiple generations withaquatic vertebrates limits a comprehensive assessment of pos-sible risks of PFOS. Notable concentrations of PFOS in eggs offield-sampled fish have suggested maternal effects (Kannan et al.,2005), thus further emphasizing the relevance of long-term stud-ies with PFOS to explore its potential to induce population-relevanteffects.

Another issue that needs to be addressed is the fact that contam-inants in aquatic systems frequently occur in complex mixtures.Given previous indications of the toxicity increasing behaviourof PFOS in binary mixtures (Hu et al., 2003; Jernbro et al., 2007;Liu et al., 2009), long-term assessments of combinations of PFOSwith other pollutants appears a promising strategy to shed morelight on the complex toxicology of PFOS. To our knowledge, nostudy has previously assessed the toxic potential of PFOS in fishin vivo on a long-term basis in combination with an environmentalchemical.

In the present study, PFOS was investigated in binary mix-tures with the well known endocrine disrupting chemical (EDC)bisphenol A (BPA), chosen due to its extensive applications andnumerous indications of its endocrine disrupting effects in fish(Kang et al., 2002; Lahnsteiner et al., 2005; Mandich et al., 2007;Hatef et al., 2012). The effects of waterborne PFOS and BPA werestudied in zebrafish (Danio rerio) over two full generations inorder to test the hypothesis that PFOS increases the bioavailabil-ity and thus the estrogenic potential by BPA in a binary mixture.A further aim was to assess whether chronic exposure to PFOSinduces effects over generations, i.e. effects that cannot be fore-seen with short-term studies. The long-term study was designedto allow a simultaneous evaluation of PFOS and BPA, both singlyand in combination, at nominal exposure concentrations rangingfrom 0.6–300 �g/L and 10–400 �g/L, respectively. The range ofconcentrations tested were chosen in light of measured concen-trations in aquatic systems (Yang et al., 2005; Rostkowski et al.,2006; Skutlarek et al., 2006), along with previously tested con-centrations of PFOS and BPA in fish (Ankley et al., 2005; Oakeset al., 2005; Staples et al., 2011). Toxicological endpoints wereselected at different levels of biological organization and includedVtg induction, histological alterations in liver, thyroid and gonads,somatic growth (total length and body weight) and reproductivesuccess (egg production and fertilization rate). Data were collectedat 30, 90 and 180 d post-fertilization (dpf) in F1 and F2 gen-erations. Post-hatch survival was documented in F1, F2 and F3generations.

2. Materials and methods

2.1. Chemicals and preparation of stock solutions

Heptadecafluorooctanesulfonic acid potassium salt (PFOS; CASno. 2795-39-3; ≥98% purity) and 4-ethylaminobenzoate (benzo-cain) were purchased from Sigma–Aldrich (Schnelldorf, Germany).Bisphenol A (BPA; CAS no. 80-05-7; ≥99% purity) and pro-tease inhibitor cocktail (P8340) was obtained from Sigma–Aldrich(Taufkirchen, Germany). PFOS and BPA were both delivered to thetest vessels without use of a carrier solvent. A first stock solutionof PFOS (300 mg/L) was prepared by dissolving 1.5 g PFOS in 5 Lof deionized water with over-night magnetic stirring. The solutionafter the first dilution step, hereafter named second stock solution(0.016, 2.6 and 7.8 mg/L), was freshly prepared four times a weekby diluting the first stock solution (300 mg/L) with deionized water.Nominal concentrations of PFOS in the test vessels were 0.6, 100and 300 �g/L. Stock solutions of BPA (500 mg/L) were freshly pre-pared twice a week by adding 500 mg in 1 L of deionized water.Complete solubilisation was achieved by alkalinization and rigor-ous over-night stirring as previously described by Pickford et al.(2003). The pH of the first BPA stock solution ranged from 10.97 to11.55 throughout the test. The second stock solution of BPA (0.26,5.2 and 10.4 mg/L) was freshly prepared four times a week by dilu-tion of the first stock solution (500 mg/L) using deionized water.Nominal concentrations of BPA in the test vessels were 10.0, 200and 400 �g/L. For the co-exposure treatments, all tested BPA con-centrations (10, 200 and 400 �g/L) were combined with the lowestand the highest nominal concentration of PFOS (0.6 and 300 �g/L),respectively.

2.2. Fish maintenance

The test was initiated with fertilized eggs of zebrafish (Daniorerio; Westaquarium strain) obtained from non-exposed adultsreared in the laboratory of Aquatic Ecology and Toxicology, Hei-delberg University, Germany. Throughout the study, the fish weremaintained in a light-isolated room with an artificial 14/10 hlight/dark period. Adult fish were fed twice daily with freshlyhatched Artemia nauplii (Aquafauna Bio-Marine Inc., Hawthorne,CA) complemented with TetraMin flake food (Tetra, Germany).Larvae were initially fed twice daily with liquid starter food(Hobby liquizell, Gelsdorf, Germany) followed by Sera micronpowder food (Sera, Heinsberg, Germany) and freshly hatchedArtemia nauplii. The diets were analyzed for PFOS contents priorto the start of the study (data not shown). Tap water and deion-ized water were mixed until conductivity (600–750 �S), hardness(276 ± 17.8) and pH (8.0–8.2) were stably balanced. The water mixwas supplied from an aerated reservoir and used to culture allembryos and fish. The final test water was routinely character-ized for pH (8.25–8.75), total hardness (167–356 mg/L), ammonium(<5 mg/L) and nitrite (<1 mg/L). Temperature (26.0 ± 1.0 ◦C) and dis-solved oxygen (6.45–10.97 mg/L) was checked weekly. In order toensure high water quality, food remains and debris were removeddaily and vessel surfaces were gently scraped once a week withthe exception of sensitive periods during early larvae develop-ment.

2.3. Experimental design

During the course of the study, fish were continuously exposedto PFOS and BPA, either singly or in combination. Each treat-ment group was replicated twice holding a starting number of 80fish per replicate (160 individuals per treatment). At 2–4 h post-fertilization (hpf), eggs were transferred to glass dishes (12 cmdiameter, 8 cm high) and exposed to the different treatments at


118 Su. Keiter et al. / Aquatic Toxicology 118– 119 (2012) 116– 129

26 ± 1 ◦C under semi-static conditions (complete renewal of solu-tions after 24 h) until 48 hpf, when they were transferred torespective test vessel. Whole-glass tanks (18 cm × 40 cm × 40 cm),adjusted for a 10 L working volume were utilized as test vessels.The test vessels were placed on top of serial-connected heatingmattresses ensuring constant heating conditions. A flow-throughsystem with a three-fold water exchange per day was appliedthroughout the study in order to provide adequate supply of freshtest solution. External aeration by pressurized air was installed foreach test vessel. Test solutions were daily refilled into light-isolated10 L glass bottles located above the test vessels. Each test solu-tion was constantly held in motion by magnetic stirring. Peristalticpumps (M312; Gilson, Villiers-le-Bel, France) were used for a con-tinuous delivery of test solution (50 ml/h) from each glass bottleto paired test vessels serving as replicates A and B for each treat-ment group. For each test vessel, a water flow rate of 1.25 L/h wasadjusted by means of rotameters (Rota Yokogawa, Wehr, Germany).A detailed outline of the flow-through system has been describedby Knörr and Braunbeck (2002). Flow rates of water supply andtest solutions were controlled daily. Outflow water passed throughactive carbon filters (Prantner, Reutlingen, Germany) before releaseto the sewer. Each replicate of the F1 and F2 generations wassub-sampled at 30 and 90 d post-fertilization (dpf). At 30 dpf,the number of fish in each replicate test vessel was reduced by35 individuals for measurements of length and weight. Sampledfish were anaesthetized and euthanized in a saturated solutionof 4-ethylaminobenzoate (benzocain). At 90 dpf, each replicatewas further reduced by 25 individuals for measurement of length,weight and vitellogenin (Vtg) as well as for histological evaluationof liver, thyroid and gonads. After sacrifice, length and weight weredocumented, and fish were transversely trimmed behind the analfin separating the tail for Vtg measurements and the rest of the bodyfor histology. Tails were instantly frozen in liquid nitrogen, and therest of the body was placed in histology cassettes and submergedin cooled Davidson’s fixative (Romeis, 1989) for a minimum of 24 hbefore histological processing. For each replicate test vessel, a totalof 10 males and 10 females were retained for reproduction experi-ments and breeding of the F2 and F3 generations. After terminationof the breeding experiments (approximately at 180 dpf), remain-ing adults were sampled following the exact procedure as describedabove for sub-sampling at 90 dpf. Post-hatching survival was doc-umented for the F3 generation at 14 dpf, when the experiment wasterminated without subsequent sampling.

2.4. Fertilization and egg production in F1 and F2 adults

Breeding experiments for evaluation of fecundity and fertil-ization rate were performed with F1 and F2 adults starting atapproximately 4 months of age. Breeding trials for each treat-ment group were repeated six to seven times (F1) and nine toten times (F2) with a minimum of one week of recovery inbetween to avoid stress related bias. At the day before spawning,five individuals of either sex from each test vessel were ran-domly selected and transferred to breeding tanks prior to theonset of darkness. The spawning facility was constructed of sixbreeding tanks (15 cm × 16 cm × 25 cm) which were held togetherunder un-exposed semi-static conditions with constant air supply(7.37–8.10 mg/L) and heating (25.0 ± 1.0 ◦C). Each breeding tankwas equipped with green colored nylon netting serving as a breed-ing stimulant. The bottom of the breeding tanks was covered by astainless steel grid (mesh size 1.25 mm) to allow the eggs to passthrough into separate spawning trays and thus to avoid cannibalismby parental fish. About 20–30 min after the onset of light, spawningtrays were removed and the eggs were collected and any furtherdebris was removed. Eggs were counted and visually inspectedunder a stereo microscope and transferred to petri dishes (18 cm

diameter) containing freshly prepared artificial water accordingto ISO (1996) (maximum 100 eggs/200 ml water). The eggs wereincubated at 26.0 ± 1.0 ◦C over night, after which coagulated andfertilized eggs were counted.

2.5. Chemical analysis

Water samples for chemical analysis of PFOS concentrationswere collected directly from the test vessels on a monthly basisand stored at −80 ◦C prior to chemical analysis. PFOS concen-trations were determined using solid phase extraction (SPE) andliquid chromatography-tandem mass spectrometry (LC–MS/MS;Skutlarek et al., 2006). BPA concentrations were not analyticallyverified. In a previous study, no degradation of BPA was detected innatural waters within 1 day (Dorn et al., 1987). As the flow-throughsystem in the present study was maintained to provide three tankvolume changes per day, the rate of fresh test solution was consid-ered enough to compensate for any bacterial degradation occurringin the test vessels.

2.6. Measurement of vitellogenin (Vtg) in tail homogenate

Deep frozen tails of sampled zebrafish males and 10 × the tis-sue wet weight of ice-cold homogenate buffer (50 mM Tris–HCL,pH 7.4, 1% protease inhibitor cocktail) were added to 2 ml Eppen-dorf tubes containing one stainless steel bead each (5 mm diameter;Qiagen, Hilden, Germany). The tissue homogenization was per-formed with a tissue lyser II (30 s, 15 Hz, Qiagen). The homogenatewas centrifuged for 30 min at 24,652 × g at 4 ◦C and the super-natant for Vtg measurements was collected and stored in aliquotsof 50 �l at −20 ◦C until further analysis. Vtg concentrations weremeasured by a method based on a direct non-competitive sand-wich ELISA previously described by Holbech et al. (2001) with thefollowing modifications as described by Morthorst et al. (2010):Dextran-HRP conjugated antibodies were replaced by a two-stepprocess in order to enhance sensitivity of the assay. First, 150 �Lof biotin-conjugated antibody was added to each well, and theplate was incubated on a horizontal shaker (100 vibrations/min)for 1 h at room temperature. After rinsing five times with washingbuffer (PBS, 0.1% Tween-20, 0.1% BSA) 150 �L of streptavidin HRP-conjugated antibody was added to each well, and the plate wasincubated on a horizontal shaker for 1 h at room temperature.

2.7. Histology

For histological examination of liver, thyroid and gonads, tis-sues (central body portions without tail) fixed in Davidson’s fixativewere processed in a Leica TP 1020 Tissue Processor (Leica Microsys-tems, Wetzlar, Germany) and embedded in Histoplast S (Serva,Heidelberg, Germany). Four micron sections were mounted on glassslides, stained with periodic acid-Schiff (PAS) staining (Romeis,1989) and examined using a light microscope (Leitz Aristoplan). Forfurther details on embedding and staining procedures, see Schmidtand Braunbeck (2011). Gonadal staging and severity grading wassemi-quantitatively assessed according to the criteria outlined byBraunbeck et al. (2010). Briefly, testes and ovaries were stagedbased on the abundance of specific gametogenic cell types by useof a staging system ranging from 0 (undeveloped) to 4 (spent) or 5(post-ovalutory) for males and females, respectively. Severity grad-ing was based on the degree to which a histomorphological changewas present in a tissue section and was employed according toa system ranging from grade 1 (minimal) to grade 4 (severe). Inorder to allow a comparison between treatments and generationsregarding severity and maturation, the maturity index developedand described by Baumann (2008) was adopted. Briefly, each stageof maturity was accorded to a fixed value which increases with



Fig. 1. Measured water concentrations of PFOS throughout the 330 d exposure trialof the F1, F2 and F3 generation zebrafish. Nominal concentrations of PFOS were 0(�; n = 1), 0.6 (�; n = 3), 100 (©; n = 1) and 300 �g/L (�; n = 4), where n representsthe amount of sampled test vessels within each exposure group at each samplingtime. Concentrations of PFOS were measured using LC–MS/MS.

increasing maturity of the fish (maturity stage 0 corresponds tovalue 1; stage 1 corresponds to value 2; etc.). The values of eachreplicate aquaria were summed up, divided by the number offemale and male fish, respectively and the mean value of eachtreatment group was calculated. The same principle was appliedfor the severity index. Sections investigated for the presence ofMycobacterium spp. were stained according to the Ziehl–Neelsentechnique.

2.8. Statistical analysis of data

In case assumptions of normality and equal variances heldtrue, differences in length, weight, survival and Vtg concentra-tions were evaluated with one-way ANOVA followed by the posthoc Holm–Sidak method. In case assumptions of normality andequal variance failed, the non-parametric Kruskal–Wallis ANOVAon ranks test was used followed by multiple comparisons versusthe control group or single exposure treatments (Dunn’s method).To compare groups (in a set less than three), the Student’s t-testor the Mann–Whitney rank sum test (when no normality or equalvariance) was applied. Differences were considered significant atthree different levels (*, p < 0.05; **, p < 0.01, ***, p < 0.001) relativeto controls or to the single exposure treatments. Analyses wereperformed with SigmaStat® Statistical Software version 3.5 (Systat-Jandel Scientific, Erkrath, Germany).

3. Results

3.1. Chemical analysis

Mean measured concentrations of PFOS throughout the exper-imental period are shown in Fig. 1. Monthly mean concentrations(standard deviation; total number of sampled test vessels) ofPFOS in controls, 0.6, 100 and 300 �g/L treatment groups were0.073 (0.080; n = 10), 0.734 (0.131; n = 26), 106.9 (16.26; n = 9)and 267.6 (44.99; n = 27) �g/l, respectively. The measured PFOSconcentrations in the control vessels already at day 1 and later onare believed to reflect the background concentration in the watersupply to the test facility. The peak in the control group at exposureday 142 was due to a handling error causing a temporarily higherPFOS concentration (0.29 �g/L) compared with the backgroundlevel. At the subsequent sampling time (exposure day 183), the

measurements of PFOS (0.076 �g/L) indicate a return to the rangeof the background level.

3.2. Survival in the F1, F2 and F3 generation

Survival for the F1, F2 and F3 generations is summarized inFig. 2. No significant mortality or malformations were observedin the F1 generation over the course of the 180-day exposure. Inthe subsequent F2 generation, post-hatch malformations such asbody flexure followed by 100% mortality was documented withina period of 14 dpf in the highest PFOS (300 �g/L) treatment, withor without BPA. Given this, all treatment groups with the highestPFOS exposure (300 �g/L) had to be terminated and were not fur-ther investigated in subsequent F2 and F3 generation. During thesame time period, an unexplained drop in survival (survival rate37.5 ± 17.7%) was observed in the 200 �g/L BPA exposure group(Fig. 2). Furthermore, a decrease in fish density shortly after swim-up was observed in one of the two replicates of PFOS 100 �g/L(survival rate 5%). In the F3 generation, no post-hatch malforma-tions were documented throughout the examined 14 dpf period.Although not significant, the lowest survival rate (41.8 ± 29.2%) inthe F3 generation was observed in the PFOS 100 �g/L treatmentgroup, whereas survival rates for the remaining treatments rangedbetween 77.5 and 91.6%.

3.3. Growth in the F1 and F2 generation

Growth data (total length and body weight) for male and femaleF1 and F2 zebrafish at 30-, 90- and 180 dpf are summarized inFigs. 3 and 4, respectively. Males and females were pooled at 30dpf in F1 and F2 generations.

3.3.1. PFOSF1 generation. At 90 dpf, length and weight in male zebrafish

were significantly reduced in all PFOS exposures (0.6, 100 and300 �g/L), if compared to the controls (Fig. 3b, d). In adult malesat 180 dpf, length was significantly suppressed at the higher PFOSexposure groups (100 and 300 �g/L; Fig. 3b). In adult females at180 dpf, length and weight was significantly reduced in all PFOSexposure groups (0.6, 100 and 300 �g/L; Fig. 3a and c).

F2 generation. At 90 dpf, length and weight in male (Fig. 4b andd) and female zebrafish (Fig. 4a and c) were significantly reduced inboth PFOS exposure groups (0.6 and 100 g/L) (weight not significantin males in the 100 �g/L exposure group). At 180 dpf, length andweight in adult males and females exposed to PFOS 100 �g/L weresignificantly lower than in controls.

3.3.2. BPAF1 generation. After exposure to 200 �g/L BPA, reduction of

length and weight was evident for both males and females at 90dpf (length not significant in females; Fig. 3a–d). However, thisgrowth inhibition was most likely due to a counting error at thefirst time of sampling (30 dpf), resulting in a slightly higher fishdensity throughout a two month period in one replicate (61 versusapprox. 44 individuals in the other treatment groups).

F2 generation. At 90 dpf, all BPA treatment groups (10, 200 and400 �g/L) showed reduced length and weight in both males andfemales compared to the controls (Fig. 4a–d). In adult males andfemales at 180 dpf, length and weight at the highest BPA concen-tration (400 �g/L) were significantly lower than in controls.

3.3.3. PFOS and BPA mixtureF1 generation. At 90 and 180 dpf, the highest tested PFOS concen-

tration (300 �g/L) in combination with BPA (10, 200 and 400 �g/L)significantly reduced length and weight in males (Fig. 3b and d). Asignificant reduction of length and weight was noted for females at



Fig. 2. Survival of the F1 (A), F2 (B) and F3 (C) generation zebrafish at 180 (A andB) and 14 (C) dpf after exposure to PFOS, BPA or a binary mixture of PFOS and BPA.Data are given as average ± S.D. of two replicate aquaria. +Due to a low survival rate(5%) in one of the two replicates of the PFOS 100 �g/L treatment group in the F2generation (B), only the second replicate is illustrated in the figure.

180 dpf following co-exposure to PFOS (0.6 and 300 �g/L) and BPA(10, 200 and 400 �g/L). Length changes were not significant in thePFOS 300 �g/L + BPA 10 �g/L exposure group (Fig. 3a).

F2 generation. At 90 dpf, length and weight in both males andfemales in all binary mixture groups (0.6 �g/L PFOS + 10, 200 and400 �g/L BPA) were significantly lower than in controls (Fig. 4a–d).At 180 dpf, length and weight were significantly reduced in malesand females exposed to a binary mixture with 0.6 �g/L PFOS and thetwo highest tested BPA concentrations (200 and 400 �g/L); weightchanges in males not significant in the PFOS 0.6 �g/L + BPA 200 �g/Lexposure group.

3.4. Vitellogenin (Vtg) induction in males of the F1 and F2generations

Vtg concentrations (outliers excluded) measured in F1 and F2male zebrafish at 90 and 180 dpf are summarized in Fig. 5. Any Vtgvalue greater than 1.5 times the spread outside the closest hinge ofthe boxplot of each treatment was considered an outlier.

3.4.1. PFOSF1 generation. At 90 dpf, males exposed to 0.6 �g/L PFOS showed

a significantly induced Vtg synthesis if compared to controls(Fig. 5a). In the presence of outliers (11,448 and 13,723 ng/g Vtg incontrols and 66,253 ng/g Vtg in the PFOS 0.6 �g/L exposure group)this difference was not great enough to be statistically significant.Vtg levels generally tended to decrease in a concentration-dependent manner; however, Vtg synthesis in F1 males exposedto 100 and 300 �g/L PFOS was never significantly different fromcontrols (Fig. 5a and b).

F2 generation. Comparable with the corresponding F1 males, F2males exposed to 0.6 �g/L PFOS for 90 d displayed a statisticallysignificant Vtg induction over controls (Fig. 5c). Due to the low Vtgconcentrations in adult control males, both PFOS exposure groups(0.6 and 100 �g/L) were statistically higher at 180 dpf (Fig. 5d).

3.4.2. BPAF1 generation. At 90 dpf, the Vtg synthesis was significantly

induced in males exposed to the highest tested BPA concentration(400 �g/L) (Fig. 5a). Overall, Vtg levels in the BPA treatment groupstended to increase in a concentration-dependent manner at bothsampling times (Fig. 5a and b).

F2 generation. Exposure to 10 �g/L and 400 �g/L BPA signifi-cantly induced the Vtg levels in male zebrafish over controls at90 dpf (Fig. 5c). In presence of outlier value (985.2 ng/g Vtg), theVtg induction in the BPA 200 �g/L group was also significantlyhigher compared to controls. Due to the low Vtg concentrations inthe adult control males, all BPA exposure groups were statisticallyhigher at 180 dpf (Fig. 5d).

3.4.3. PFOS and BPA mixtureF1 generation. At 90 dpf, males in the lowest PFOS (0.6 �g/L)

binary mixture groups displayed a BPA-dependent increase in Vtgsynthesis; however, no statistical significance was found comparedto controls (Fig. 5a). At 90 dpf, Vtg levels in males co-exposedto 0.6 �g/L PFOS and 10 �g/L BPA were significantly lower, ifcompared with the single exposures of the single compounds.In contrast to the lowest PFOS (0.6 �g/L) binary mixture groups,Vtg levels in the highest PFOS (300 �g/L) binary mixture groupsdecreased in a concentration-dependent manner with a significantreduction in the highest binary mixture group (PFOS 300 �g/L + BPA400 �g/L), if compared with 400 �g/L BPA alone. As seen in malesat 90 dpf, adults at 180 dpf co-exposed to 0.6 �g/L PFOS and 10,200 and 400 �g/L BPA displayed a BPA concentration-dependentincrease in Vtg synthesis with the highest binary mixture group(PFOS 0.6 �g/L + BPA 400 �g/L) being significantly higher compared



Fig. 3. Length and weight of female (A and C) and male (B and D) zebrafish in the F1 generation exposed to PFOS, BPA or a binary mixture of PFOS and BPA after 30, 90 and180 dpf. Length and weight data at 30 dpf consist of males and females pooled together. Data are given as average ± S.D. of two replicate aquaria. Significant differencesfrom negative controls are shown (*, p < 0.05, **, p < 0.01, ***, p < 0.001). Statistic significant differences were detected with one-way ANOVA (post hoc Holm–Sidak method)or Kruskal–Wallis ANOVA on ranks (post hoc Dunn’s method).

with the single exposures of both compounds as well as com-pared to controls (Fig. 5b). In presence of outlier in the controlgroup (10,976 ng/g Vtg) no significant difference between PFOS0.6 �g/L + BPA 400 �g/L and controls was found. At 180 dpf, adultmales in all highest PFOS (300 �g/L) binary exposure groups dis-played significantly lower Vtg levels compared to controls. Inpresence of outliers in the two highest binary exposure groupswith PFOS 300 �g/L (123,000 ng/g and 39,864 ng/g respectively) nosignificance was found.

F2 generation. At 90 dpf, Vtg levels in males co-exposed to0.6 �g/L PFOS and 10, 200 and 400 �g/L BPA were significantlyhigher than in controls (Fig. 5c). A significant drop in Vtg synthe-sis was seen with a mixture of 0.6 �g/L PFOS and 400 �g/L BPA, ifcompared with isolated exposure of BPA (400 �g/L). Given the lowVtg concentrations in the adult control males, all binary exposuregroups at 180 dpf were statistically elevated (Fig. 5d). Similarly tothe corresponding F1 adults at 180 dpf, adult F2 males exposed to

the lowest PFOS (0.6 �g/L) binary exposure groups showed a slightBPA-dependent induction of Vtg synthesis with significantly higherVtg levels in males exposed to the highest binary mixture (PFOS0.6 �g/L + BPA 400 �g/L), if compared with the single exposure ofBPA (400 �g/L).

3.5. Histological alterations

3.5.1. GonadsTestis maturation stages (Braunbeck et al., 2010) for F1 and F2

male zebrafish did never differ from controls in any exposure group,nor was there a difference between generations except for 200 �g/LBPA where adult F2 males showed slightly advanced maturation, ifcompared to adult F1 males (details not shown). No obvious dif-ference in number or size of Leydig cells was recorded for anytreatment group in either generation. In the F1 generation at 90dpf, one case of moderate testis-ova (perinucleolar oocytes) was



Fig. 4. Length and weight of female (A and C) and male (B and D) zebrafish in the F2 generation exposed to PFOS, BPA or a binary mixture of PFOS and BPA after 30, 90 and180 dpf. Length and weight data at 30 dpf consist of males and females pooled together. Data are given as average ± S.D. of two replicate aquaria. Significant differencesfrom negative controls are shown (*, p < 0.05, **, p < 0.01, ***, p < 0.001). +Only one replicate aquaria was sampled. Statistic significant differences were detected with one-wayANOVA (post hoc Holm–Sidak method) or Kruskal–Wallis ANOVA on ranks (post hoc Dunn’s method).

observed in the 10 �g/L BPA exposure group (1 out of 12 males).Likewise, after 90 d exposure in the F2 generation, a minimal caseof testis-ova (perinucleolar oocytes) was observed in the controlgroup (1 out of 13 males). No case of testis-ova was recorded foradult males at 180 dpf in any generation. Ovarian maturation stages(Braunbeck et al., 2010) in F1 and F2 females did not differ fromcontrols in any treatment group both at 90 and 180 dpf. With theexception of 90 d old F2 females exposed to the highest BPA con-centration (400 �g/L; both alone and in combination with PFOS),all F2 females displayed a transient delay in ovarian maturation,if compared with F1 females (details not shown). At 180 dpf noobvious difference in ovarian maturation stages between the twogenerations was seen.

3.5.2. ThyroidNo deviations from controls were recorded in any treatment

regarding size, structure, distribution and number of thyroid folli-cles.

3.5.3. LiverVacuolization of the liver (Fig. 6) was the major histological

alteration found exclusively in PFOS-exposed fish in F1 and F2generations, both alone and in combination with BPA. At 90 dpf,hepatocellular vacuolization in F1 males was recorded for the upperinvestigated PFOS range (100 �g/L and 300 �g/L) both in singleand binary exposures. At 180 dpf, adult F1 males displayed a sim-ilar chemical-related vacuolization, although with less individuals



Fig. 5. Measured vitellogenin (Vtg) concentrations in tail homogenate of male zebrafish in F1 and F2 generation at 90 (A and C) and 180 (B and D) dpf after exposure to PFOS,BPA or a binary mixture of PFOS and BPA. Vtg concentrations are presented as box plots with median line (line within box), 25th and 75th percentiles (lower and upperboundary of box) and 10th and 90th percentiles (lower and upper whiskers). Significant comparisons versus negative control are indicated with *, p < 0.05, **, p < 0.01 and ***,p < 0.001. Significant comparisons versus BPA-treated group and PFOS-treated group are indicated with a and b, respectively. Statistic significant differences were detectedwith one-way ANOVA (post hoc Holm–Sidak method) or Kruskal–Wallis ANOVA on ranks (post hoc Dunn’s method). The number of measured male fish in each treatmentgroup (pooled data from all tanks in each treatment) is indicated in parentheses in the bottom of the figure.

affected (Fig. 7a). Vacuolization was also detected in F2 males after90 d of exposure to 100 �g/L PFOS, whereas no such findings weremade in adult F2 males (Fig. 7b). Vacuolization was documented inboth sexes; however, compared with males, far fewer females wereaffected. Female F1 fish at 90 dpf displayed vacuolization followingexposure to all binary mixtures with the highest PFOS concentra-tion (300 �g/L PFOS in combination with 10, 200 and 400 �g/LBPA; Fig. 7a). No hepatic alterations were found in adult femalesat 180 dpf. In F2 females, no case of hepatocellular vacuolizationwas observed at any time.

3.5.4. Granulomatous inflammationDistinct granulomas (Fig. 8) were frequently observed in fish

exposed to 100 and 300 �g/L PFOS both in single and binary

exposure groups. Granulomas were mainly located in the liver,but were also documented in other visceral organs such as pan-creas and kidney as well as in the reproductive organs of bothsexes. Fig. 9a and b shows the mean histological index of granu-lomas in the liver of males and females in F1 and F2 generations.The occurrence of granulomas was clearly gender-specific withmale fish being more susceptible than female fish in both gener-ations. At 90 dpf, F1 males exposed to PFOS 300 �g/L, alone andin combination with all BPA concentrations, displayed granulo-mas in liver, pancreas and testis with the order of severity being:liver > pancreas > testis. With an overall higher grade of severity,an identical pattern was documented for adult F1 males with theexception that granulomas in liver and pancreas were noted alsowith 100 �g/L PFOS. For F2 males, kidney-located granulomas in the



Fig. 6. Light micrographs showing the liver structure in control (A) and PFOS-exposed (B) males of zebrafish in the F1 generation at 90 d post-fertilization. Hepatocellularvacuolization (arrows) is observed in PFOS-exposed fish (B). Identical vacuolization was also detected in binary mixtures with BPA (micrographs not shown). Sections of4 �m thickness stained with periodic-acid Schiff (PAS) and Mayer‘s hematoxylin.

Fig. 7. Hepatocellular vacuolization in male and female zebrafish in F1 (A) and F2 (B) generation following exposure to PFOS, BPA or to a binary mixture of PFOS and BPA.Data are given as average ± S.D. of two replicate aquaria. +Only one replicate aquarium was sampled.

Fig. 8. Light micrographs showing granuloma structures (arrows) located in the liver of PFOS-exposed zebrafish males of the F1 generation at 90 d post-fertilization (A).High magnification of a granuloma structure (B). Identical granulomas were also detected in binary mixtures with BPA (micrographs not shown). Section of 4 �m thicknessstained with periodic-acid Schiff (PAS) and Mayer‘s hematoxylin. L, liver; I, intestine and S, spleen.



Fig. 9. Granulomas located in the liver of male and female zebrafish in F1 (A) and F2 (B) generation following exposure to PFOS, BPA or to a binary mixture of PFOS and BPA.Data are given as average ± S.D. of two replicate aquaria. + Only one replicate aquarium was sampled.

100 �g/L PFOS treatment were observed already at 90 dpf (1 out of9 males). One case of kidney granuloma was also documented inthe lowest PFOS (0.6 �g/L) binary exposure group (0.6 �g/L PFOSin combination with 200 �g/L BPA; 1 out of 13 males). In adultF2 males, the severity grade of granulomas was increased and fol-lowed a similar pattern of affected organs as in the F1 generation(liver > pancreas ≈ testis). At 90 dpf, F1 females in the highest PFOS(300 �g/L) binary exposures displayed a minimal case of granulomain the liver, pancreas and ovary with the order of severity being:liver > pancreas > ovary. In adult F1 females, a higher grade of sever-ity was noted with granulomas primarily located in the ovary andthe liver (ovary > liver). A minimal and moderate case of granulomawas furthermore detected in the ovary of two control females (2 outof 11 females). No granulomas were observed in F2 females at 90dpf in any treatment; however, at 180 dpf, a case of minimal granu-loma was detected in the ovary of one F2 female (1 out of 7 females)following exposure to PFOS 100 �g/L.

3.6. Egg production and Fertilization rate in the F1 and F2generation

Fecundity (total amounts of eggs spawned per female) in F1and F2 generations is shown in Fig. 10 (mean values ± SD of tworeplicate aquaria). Mean (±SD) of eggs spawned per control female(190 ± 12) in F1 generation represented the highest reproductiveoutput among all treatment groups. Co-exposure of 0.6 �g/L PFOSand 400 �g/L BPA significantly reduced fecundity compared to sin-gle exposures of the two chemicals as well as compared to controls(Fig. 10a). The scatter plots (Fig. 10b and c) serves to illustratethe varying amount of eggs spawned per F1 and F2 female overthe breeding study, thus explaining the high standard deviationwhich was generally seen in all treatment groups including the con-trol groups. Fertilization rate of the F1 offspring ranged between58% and 79% for all treatment groups with no significant differ-ence compared to controls (72 ± 2; Fig. 11). Fertilization rate ofthe F2 offspring ranged between 63 and 80% for all treatmentgroups including the control. A significantly higher fertilization ratecompared to controls was observed in the 0.6 �g/L PFOS exposuregroup. No statistical significance was found compared with the fer-tilization rate of the F1 offspring (59–79%). The amount of fertilizedeggs transferred from the F1 groups to the F2 generation variedbetween 160 and 170 for each group (80–85 fertilized eggs per

replicate). The same amounts of fertilized eggs were transferredfrom F2 to start the F3 generation with the exception of 400 �g/LBPA with 145 eggs.

4. Discussion

The present study was designed to determine long-term effectsand synergizing behaviour of PFOS in binary mixtures with BPA inzebrafish (Danio rerio). Both chemicals were investigated in isolatedand combined exposure scenarios over two full generations withfocus on Vtg concentrations, histological alterations and reproduc-tive effects. Whereas PFOS did not increase the endocrine potentialof BPA; PFOS-exposure resulted in hepatocellular vacuolization andreduced survival for the F1 offspring.

A number of studies have investigated the acute and chronic tox-icity of PFOS and BPA towards a wide range of aquatic organisms,confirming a lethal potency at doses well above those normallyreported in environmental samples (OECD, 2002; Pickford et al.,2003; Ankley et al., 2004; Du et al., 2009; Mihaich et al., 2009; Hanand Fang, 2010). In the present study, PFOS and BPA were evalu-ated at nominal concentrations of 0.6–300 �g/L and 10–400 �g/L,respectively. Following exposure to the maximum PFOS concen-tration (300 �g/L), malformations such as body flexure followed by100% mortality was observed within 14 dpf in F2 generation. Iden-tical observations in all highest PFOS (300 �g/L) binary mixtures,indicate that the effects seen were PFOS-related. These results arein agreement with two recent studies, where malformations fol-lowed by 100% mortality after 96 h post-hatch (Du et al., 2009) and7 dpf (Wang et al., 2011), were reported for embryos and larvaederived from maternal exposure to 250 �g/L PFOS. Decreased off-spring survival following maternal PFOS exposure has also beenreported for other test organisms, e.g., swordtail fish (Xiphophorushelleri; Han and Fang, 2010), Northern bobwhite quail (Colinus vir-ginianus; Newsted et al., 2007), mouse (Mus musculus; Lau et al.,2003) and rat (Rattus norvegicus; Lau et al., 2003), thus furthersupporting the results obtained in the present study. In contrast,Ankley et al. (2005) found no significant effects on offspring survival(≤300 �g/L PFOS) in a partial life-cycle study with the fathead min-now (Pimephales promelas). However, the survival rate followingexposure to 300 �g/L PFOS represented the lowest among all treat-ments. In addition to inter-species differences in sensitivity, thediverging rates in offspring survival reported in the literature could



Fig. 10. Total number of eggs spawned by F1 (white bars) and F2 (black bars) generation females exposed to PFOS, BPA or a binary mixture of PFOS and BPA (A). Data aregiven as average ± S.D. of two replicate aquaria. Comparisons significant at <0.05 are indicated with * (versus negative control), a (versus BPA-treated group) and b (versusPFOS-treated group). Statistic significant differences were detected with one-way ANOVA (post hoc Holm–Sidak method). Due to technical problems in F2 generation controlgroup, no statistical analyses were carried out. The two lower scatter plots illustrate the variation in egg production per female within each replicate aquarium in the F1 (B)and F2 (C) generation (open circles: replicate 1; filled circles: replicate 2).

possibly be related to the duration of the parental exposure; 180 din this present study, 70 and 42 d in the study by Du et al. (2009) andHan and Fang (2010), respectively, and finally, 21 d in the study of(Ankley et al., 2005). The present study also revealed unexplainedmortality in the F2 generation after exposure to 200 �g/L BPA. How-ever, since no significant mortality was detected in the lowest orhighest BPA exposure groups (10 and 400 �g/L, respectively), norin any of the lowest PFOS (0.6 �g/L) binary exposures, the observedmortality is not believed to reflect a BPA-derived toxicity.

BPA- and PFOS-exposure have been associated with reducedmean body weights and body lengths in different organisms such ascrustaceans (Lemos et al., 2010), fish (Sohoni et al., 2001; Han andFang, 2010; Wang et al., 2011), amphibians (Ankley et al., 2004) andmammals (Seacat et al., 2002). In the present study, suppressedgrowth was seen in both F1 and F2 adults following 180 d PFOSexposure. Exposure to BPA had no consistent effects on growth in

the F1 generation; however, in adult F2 fish at 180 dpf, the high-est tested BPA concentration of 400 �g/L significantly decreasedgrowth in both males and females.

The liver is well known to be one of the target organs follow-ing PFOS exposure (Hagenaars et al., 2008; Ivan et al., 2008; Cuiet al., 2009). In the present study, histological evaluation of the liverrevealed vacuolization in fish exposed to PFOS (100 and 300 �g/L),both alone and in combination with BPA. For both F1 and F2 gen-eration males, the prevalence of hepatocellular vacuolization wasfound to be more severe at 90 dpf than at 180 dpf, suggesting anadaptive response over time. This is in contrast to the findings by Duet al. (2009) showing that the severity of vacuolization in 250 �g/LPFOS-exposed zebrafish males was unchanged after 30 d in cleanwater. In accordance with findings by Du et al. (2009), effects weremore pronounced in males in terms of severity and amount of indi-viduals affected, thus, pointing towards a gender-specific toxicity.



Fig. 11. Fertilization rate of eggs spawned by females of F1 (white bars) and F2 (blackbars) generation exposed to PFOS, BPA or a binary mixture of PFOS and BPA. Data aregiven as average ± S.D. of two replicate aquaria. Comparisons significant at <0.05 areindicated with * (versus negative control) and a (versus BPA-treated group). Statisticsignificant differences were detected with one-way ANOVA (post hoc Holm–Sidakmethod) or Kruskal–Wallis ANOVA on ranks (post hoc Dunn’s method)..

Another histopathological finding with a dominating prevalencein PFOS-exposed males was an increased occurrence of distinctspherical granulomas. Granulomas were observed both in isolatedPFOS-exposures as well as in binary mixtures, thus, strongly indi-cating a PFOS-related effect. In a study by Novotny et al. (2010),identical granulomas were detected in the liver of the freshwaterfish Aphyosemion gardneri and diagnosed as generalized mycobac-teriosis, a chronic and progressive bacterial disease commonly seenin wild and cultured fish (Chinabut, 1999). Although we wereunable to trace Mycobacterium spp. in lesions, previous studies havedemonstrated that an occurrence of a mycobacterial infection isoften not well correlated with the presence of visualized Mycobac-terium spp. (Watral and Kent, 2007). Mycobacteriosis is consideredto be precipitated by stress (Gauthier and Rhodes, 2009). Sincechemical stress is generally supposed to act immunosuppressive(Prosser et al., 2011), it may be speculated that the granulomasseen in PFOS-exposed fish occurred as a result of a suppressedimmune system. Since PFOS has previously been reported to exertimmunotoxic effects (Peden-Adams et al., 2008), an immunosup-pressive action of PFOS could help to explain the observations in thepresent study. Further support is provided by the study of Jacobsonet al. (2010) showing a correlation between PFOS-exposure and anincreased incidence of a parasitic infection in amphipods (Mono-poreia affinis).

Vtg levels were significantly elevated in F1 males after 90d exposure to the highest BPA concentration tested (400 �g/L).Induced Vtg synthesis in fish following BPA-exposure has pre-viously been reported for fathead minnows with significant Vtginductions in males already at 160 �g/L (Sohoni et al., 2001). Themeasured Vtg concentrations of approximately 1 and 2 �g/g inF1 control males at 90 and 180 dpf are rather high; however,Vtg levels in the same order of magnitude have previously beenreported for unexposed males of zebrafish (Holbech et al., 2001;Christianson-Heiska et al., 2008) and rainbow trout (Oncorhynchusmykiss; Copeland et al., 1986). No significant increase in Vtg con-centration was seen for F1 adults; however; overall, Vtg levels inthe BPA exposure groups increased in a concentration-dependentmanner at both sampling times. As in the F1 generation, at 90 dpf,Vtg levels in F2 males were significantly elevated in the highest

BPA 400 �g/L exposure group. Given the unexplained low Vtg con-centrations in adult control males in the F2 generation (180 dpf;approximately 11 ng/g), all BPA concentrations tested (10, 200 and400 �g/L) were significantly higher.

PFOS and other PFCs such as PFOA have previously been demon-strated to have an estrogenic potential in fish both in vivo (Oakeset al., 2005; Wei et al., 2007; Du et al., 2009; Benninghoff et al.,2011) and in vitro (Liu et al., 2007). In the present study, the lowesttested PFOS concentration of 0.6 �g/L was shown to significantlyincrease Vtg levels in F1 and F2 males. In line with our findings, Duet al. (2009) reported a significant up-regulation of Vtg mRNA inzebrafish males exposed to PFOS in a similar concentration range(10 �g/L). Surprisingly, we discovered that in contrast to the Vtgresponse induced by 0.6 �g/L PFOS, the estrogenic potential tendedto decrease with an increasing PFOS-exposure. Similar observa-tions were made in a study with fathead minnow (Ankley et al.,2005), where lower PFOS concentrations were shown to stimulatesteroidogenesis whereas higher concentrations had a suppressingeffect. In the present study, binary mixture exposures with thehighest PFOS (300 �g/L) concentration were observed to signifi-cantly suppress Vtg levels in F1 males at 90 dpf if compared withsingle exposures of BPA. In adult F1 males at 180 dpf, the addition ofPFOS 0.6 �g/L to the highest BPA (400 �g/L) exposure significantlyinduced Vtg synthesis compared with the two chemicals alone aswell as compared to controls. The induction was higher than wouldhave been expected based on an additive toxicity assumption; how-ever, since this trend was not seen at the other sampling times inF1 and F2 generations, this observation is not believed to reflect atoxicity-increasing effect by PFOS.

Previous studies have reported suppressed Vtg levels inpathogen-infected (Hecker and Karbe, 2005; Burki et al., 2010),thus offering a possible explanation to the observations in thepresent study. As already discussed, the prevalence of granulomasin liver of fish exposed to 300 �g/L PFOS could be indicative of abacterial infection following suppression of the immune system.The mechanisms underlying reduced Vtg inductions in diseasedfish are unknown; however, given the energy requiring processof Vtg synthesis, it is believed that this metabolic cost is saved infavor for immunological defense against the disease (Rushbrooket al., 2007; Burki et al., 2010). However, since observations in thepresent study indicated an increased severity of granulomas withage, the stronger Vtg suppressive effect of PFOS in younger F1 malescompared with F1 adults contradicts this theory. An additionalexplanation could be related to the hepatotoxic effects seen in alltreatments with the highest PFOS concentration (300 �g/L) as indi-cated by vacuolization of hepatocytes. As previously mentioned, thegrade of vacuolization was found to be more severe at 90 dpf thanat 180 dpf, thus, paralleling the Vtg-suppressive response seen byPFOS. It appears reasonable that an overall decreased fitness of theliver would lower the Vtg synthesizing capacity of the hepatocytes,thus potentially helping to explain the decreased estrogenic poten-tial of PFOS at higher concentrations. In any case, in agreement withVan der Ven et al. (2003), the present study further highlights theimportance of histopathology as a tool when evaluating endocrinedisruptors in zebrafish.

Reproductive success is considered to be one of the most ecolog-ically relevant endpoints in fish life-cycle exposures (Arcand-Hoyand Benson, 1998). In the F1 generation, spawning in the lowestPFOS (0.6 �g/L) binary exposures decreased in a BPA concentration-dependent manner with a significantly lower fecundity in thehighest binary mixture (PFOS 0.6 �g/L + BPA 400 �g/L), both com-pared with the single exposures of the two chemicals and comparedwith controls. However, no further significance was found in thebreeding study with the F1 generation fish. The low fecundity doc-umented in the F2 control group is believed to reflect stress inducedby technical problems temporarily experienced within the test



facility; thus, no further statistical or comparative analyzes werecarried out for female fecundity in the F2 generation. The variationin the total amount of eggs spawned per female seen in both gen-erations is commonly seen in zebrafish and has previously beenreported in the literature (Brion et al., 2004; Christianson-Heiskaet al., 2008). One could ask if the present test design had sufficientpower to detect the effects of PFOS and BPA: The numbers of eggstransferred from F1 to F2 and from F2 to F3 were between 160 and170 (with one outlier of 145 eggs in the F3 400 �g/L BPA group)per treatment. These numbers are higher than the numbers of eggsused to start the OECD TG 234 (2011), where comprehensive poweranalysis and statistics have been included in the validation of thetest. In TG 234, the groups are started with 120 fertilized eggs andthe endpoints also include Vtg and sex determination. Thus, thepresent design should indeed be valid in relation to sample size.

5. Conclusions

The present study documents that the tested concentrationsof PFOS did not increase the Vtg inducing potential of BPA whencombined in a binary mixture. In contrast, binary mixtures withthe highest tested PFOS concentration (300 �g/L) tended to sup-press the Vtg induction in F1 males at 90 dpf when comparedwith the single exposures of BPA. Whereas the lowest tested PFOSconcentration (0.6 �g/L) showed estrogenic potential in terms ofVtg induction, Vtg levels were generally found to decrease withincreasing PFOS exposure. PFOS-induced hepatotoxicity at higherconcentrations may be a possible explanation for the Vtg suppres-sion observed with increasing PFOS concentrations. Vtg levels inthe F2 generation generally followed a similar pattern as previouslyseen in the F1 generation after exposure to BPA and PFOS. Survivalin the F2 generation was severely reduced at the highest testedPFOS exposure (300 �g/L) within 14 dpf. Since adverse effects onhepatotoxicity and survival were only observed at concentrationsof BPA and PFOS well above ecologically relevant concentrations,these results suggest a low environmental risk of a combined expo-sure to PFOS and BPA. However, while this study was limited to abinary mixture of chemicals, more complex mixtures of pollutantsin the environment and additional stressors might well interferewith the outcome.

Acknowledgements

The authors would like to thank Dr. Susanne Knörr and ErikLeist for their assistance during the experimental work. The firstauthor was supported by the Landesgraduiertenförderung (LGFG),Heidelberg University, Ministry of Science, Research and the Artsof Baden-Württemberg.

References

Ahrens, L., Siebert, U., Ebinghaus, R., 2009. Temporal trends of polyfluoroalkyl com-pounds in harbor seals (Phoca vitulina) from the German Bight, 1999–2008.Chemosphere 76, 151–158.

Ankley, G.T., Kuehl, D.W., Kahl, M.D., Jensen, K.M., Butterworth, B.C., Nichols, J.W.,2004. Partial life-cycle toxicity and bioconcentration modeling of perfluo-rooctanesulfonate in the northern leopard frog (Rana pipiens). EnvironmentalToxicology and Chemistry 23, 2745–2755.

Ankley, G.T., Kuehl, D.W., Kahl, M.D., Jensen, K.M., Linnum, A., Leino, R.L., Villeneuve,D.A., 2005. Reproductive and developmental toxicity and bioconcentration ofperfluorooctanesulfonate in a partial life-cycle test with the fathead minnow(Pimephales promelas). Environmental Toxicology and Chemistry 24, 2316–2324.

Arcand-Hoy, L.D., Benson, W.H., 1998. Fish reproduction: an ecologically relevantindicator of endocrine disruption. Environmental Toxicology and Chemistry 17,49–57.

Baumann, L., 2008. Effects of endocrine disruptors in zebrafish (Danio rerio) asrevealed with the fish sexual development test. Diploma Thesis, Faculty of Bio-sciences, University of Heidelberg.

Benninghoff, A.D., Bisson, W.H., Koch, D.C., Ehresman, D.J., Kolluri, S.K., Williams, D.E.,2011. Estrogen-like activity of perfluoroalkyl acids in vivo and interaction with

human and rainbow trout estrogen receptors in vitro. Toxicological Sciences 120,42–58.

Braunbeck, T., Johnson, R., Wolf, J., 2010. Guidance document on the diagnosis ofendocrine-related histopathology in fish gonads. Series on Testing and Assess-ment No. 123, 114 pp.

Brion, F., Tyler, C.R., Palazzi, X., Laillet, B., Porcher, J.M., Garric, J., Flammarion, P., 2004.Impacts of 17 beta-estradiol including environmentally relevant concentrations,on reproduction after exposure during embryo-larval-, juvenile- and adult-lifestages in zebrafish (Danio rerio). Aquatic Toxicology 68, 193–217.

Burki, R., Wahli, T., Segner, H., Burkhardt-Holm, P., Rexroad, C.E., Afanasyev, S.,Antikainen, M., Krasnov, A., 2010. Pathogenic infection attenuates the responseof the estrogenic biomarker, vitellogenin, in rainbow trout. Comparative Bio-chemistry and Physiology A: Molecular and Integrative Physiology 157, S23-S23.

Chinabut, S., 1999. Mycobacteriosis and Nocardiosis. In: Woo, P.T.K., Bruno, D.W.(Eds.), Fish Diseases and Disorders, vol. 3, Viral, Bacterial and Fungal Infections.CAB International, New York, NY.

Christianson-Heiska, I.L., Haavisto, T., Paranko, J., Bergelin, E., Isomaa, B., 2008. Effectsof the wood extractives dehydroabietic acid and betulinol on reproductive phys-iology of zebrafish (Danio rerio)—a two-generation study. Aquatic Toxicology 86,388–396.

Copeland, P.A., Sumpter, J.P., Walker, T.K., Croft, M., 1986. Vitellogenin levels in maleand female rainbow trout (Salmo gairdneri Richardson) at various stages of thereproductive cycle. Comparative Biochemistry and Physiology B: Biochemistry& Molecular Biology 83, 487–493.

Cui, L., Zhou, Q.-f., Liao, C.-y., Fu, J.-j., Jiang, G.-b., 2009. Studies on the toxicologicaleffects of PFOA and PFOS on rats using histological observation and chemicalanalysis. Arch Environ Contamin Toxicol 56, 338–349.

Dorn, P.B., Chou, C.S., Gentempo, J.J., 1987. Degradation of bisphenol A in naturalwaters. Chemosphere 16, 1501–1507.

Dreyer, A., Weinberg, I., Temme, C., Ebinghaus, R., 2009. Polyfluorinated compoundsin the atmosphere of the Atlantic and Southern oceans: evidence for a globaldistribution. Environmental Science & Technology 43, 6507–6514.

Du, Y., Shi, X., Liu, C., Yu, K., Zhou, B., 2009. Chronic effects of water-borne PFOSexposure on growth, survival and hepatotoxicity in zebrafish: a partial life-cycletest. Chemosphere 74, 723–729.

Ericson, I., Domingo, J.L., Nadal, M., Bigas, E., Llebaria, X., van Bavel, B., Lindstrom,G., 2009. Levels of perfluorinated chemicals in municipal drinking water fromCatalonia, Spain: Public health implications. Archives of Environment Contam-ination and Toxicology 57, 631–638.

EU, 2006. Directive 2006/122/EC of the European Parliament and of the Council of12 December 2006 amending for the 30th time Council Directive 76/769/EECon the approximation of the laws, regulations and administrative provisions ofthe Member States relating to restrictions on the marketing and use of certaindangerous substances and preparations (perfluorooctane sulfonates). OfficialJournal L 372, 32–34.

EU, 2008. Directive 2008/105/EC of the European Parliament and of the Council of16 December 2008 on environmental quality standards in the field of waterpolicy, amending and subsequently repealing Council Directives 82/176/EEC,83/513/EEC, 84/156/EEC, 84/491/EEC, 86/280/EEC and amending Directive2000/60/EC of the European Parliament and of the Council. Official Journal L348, 84–97.

Gauthier, D.T., Rhodes, M.W., 2009. Mycobacteriosis in fishes: a review. VeterinaryJournal 180, 33–47.

Giesy, J.P., Kannan, K., 2001. Global distribution of perfluorooctane sulfonate inwildlife. Environmental Science and Technology 35, 1339–1342.

Hagenaars, A., Knapen, D., Meyer, I.J., van der Ven, K., Hoff, P., De Coen, W., 2008.Toxicity evaluation of perfluorooctane sulfonate (PFOS) in the liver of commoncarp (Cyprinus carpio). Aquatic Toxicology 88, 155–163.

Han, J., Fang, Z., 2010. Estrogenic effects reproductive impairment and develop-mental toxicity in ovoviparous swordtail fish (Xiphophorus helleri) exposed toperfluorooctane sulfonate (PFOS). Aquatic Toxicology 99, 281–290.

Hatef, A., Alavi, S.M.H., Abdulfatah, A., Fontaine, P., Rodina, M., Linhart, O., 2012.Adverse effects of bisphenol A on reproductive physiology in male goldfishat environmentally relevant concentrations. Ecotoxicology and EnvironmentSafety 76, 56–62.

Hecker, M., Karbe, L., 2005. Parasitism in fish—an endocrine modulator of ecologicalrelevance? Aquatic Toxicology 72, 195–207.

Holbech, H., Andersen, L., Petersen, G.I., Korsgaard, B., Pedersen, K.L., Bjerregaard, P.,2001. Development of an ELISA for vitellogenin in whole body homogenate ofzebrafish (Danio rerio). Comparative Biochemistry and Physiology C: Toxicology& Pharmacology 130, 119–131.

Hu, W.Y., Jones, P.D., DeCoen, W., King, L., Fraker, P., Newsted, J., Giesy, J.P., 2003.Alterations in cell membrane properties caused by perfluorinated compounds.Comparative Biochemistry and Physiology C: Toxicology & Pharmacology 135,77–88.

ISO, 1996. Water quality—determination of the acute lethal toxicity of sub-stances to a freshwater fish (Brachydanio rerio Hamilton-Buchanan (Teleostei,Cyprinidae)). Part 1: Static method. ISO 7346, 1.

Ivan, C., Hierlihy, S.L., Virginia, L., Peter, P., AndrÃ(e, N., Sheryl, T., Michael, B., Keith,T., Genevieve, B., 2008. Altered fatty acid homeostasis and related toxicologicsequelae in rats exposed to dietary potassium perfluorooctanesulfonate (PFOS).Journal of Toxicology and Environmental Health A: Current Issue 71, 1526–1541.

Jacobson, T., Holmstrom, K., Yang, G.D., Ford, A.T., Berger, U., Sundelin, B., 2010.Perfluorooctane sulfonate accumulation and parasite infestation in a field pop-ulation of the amphipod Monoporeia affinis after microcosm exposure. AquaticToxicology 98, 99–106.



Jernbro, S., Rocha, P.S., Keiter, S., Skutlarek, D., Faerber, H., Jones, P.D., Giesy, J.P.,Hollert, H., Engwall, M., 2007. Perfluorooctane sulfonate increases the geno-toxicity of cyclophosphamide in the micronucleus assay with V79 cells—furtherproof of alterations in cell membrane properties caused by PFOS. EnvironmentalScience and Pollution Research 14, 85–87.

Kang, I.J., Yokota, H., Oshima, Y., Tsuruda, Y., Oe, T., Imada, N., Tadokoro, H., Honjo,T., 2002. Effects of bisphenol A on the reproduction of Japanese medaka (Oryziaslatipes). Environmental Toxicology and Chemistry 21, 2394–2400.

Kannan, K., Tao, L., Sinclair, E., Pastva, S.D., Jude, D.J., Giesy, J.P., 2005. Perfluorinatedcompounds in aquatic organisms at various trophic levels in a Great Lakes foodchain. Archives of Environment Contamination and Toxicology 48, 559–566.

Knörr, S., Braunbeck, T., 2002. Decline in reproductive success sex reversal, anddevelopmental alterations in Japanese medaka (Oryzias latipes) after continuousexposure to octylphenol. Ecotoxicology and Environment Safety 51, 187–196.

Lahnsteiner, F., Berger, B., Kletzl, M., Weismann, T., 2005. Effect of bisphenol A onmaturation and quality of semen and eggs in the brown trout, Salmo trutta f.fario. Aquatic Toxicology 75, 213–224.

Lau, C., Thibodeaux, J.R., Hanson, R.G., Rogers, J.M., Grey, B.E., Stanton, M.E., Buten-hoff, J.L., Stevenson, L.A., 2003. Exposure to perfluorooctane sulfonate duringpregnancy in rat and mouse. II. Postnatal evaluation. Toxicological Science 74,382–392.

Lemos, M.F.L., van Gestel, C.A.M., Soares, A.M.V.M., 2010. Developmental toxicity ofendocrine disrupters bisphenol A and vinclozolin in a terrestrial isopod. Archivesof Environment Contamination and Toxicology 59, 274–281.

Li, F., Sun, H., Hao, Z., He, N., Zhao, L., Zhang, T., Sun, T., 2011. Perfluorinated com-pounds in Haihe River and Dagu Drainage Canal in Tianjin, China. Chemosphere84, 265–271.

Liu, C.S., Du, Y.B., Zhou, B.S., 2007. Evaluation of estrogenic activities and mechanismof action of perfluorinated chemicals determined by vitellogenin induction inprimary cultured tilapia hepatocytes. Aquatic Toxicology 85, 267–277.

Liu, W., Zhang, Y.-B., Quan, X., Jin, Y.-H., Chen, S., 2009. Effect of perfluorooc-tane sulfonate on toxicity and cell uptake of other compounds with differenthydrophobicity in green alga. Chemosphere 75, 405–409.

Mandich, A., Bottero, S., Benfenati, E., Cevasco, A., Erratico, C., Maggioni, S., Mas-sari, A., Pedemonte, F., Vigano, L., 2007. In vivo exposure of carp to gradedconcentrations of bisphenol A. General and Comparative Endocrinology 153,15–24.

Martin, J.W., Whittle, D.M., Muir, D.C.G., Mabury, S.A., 2004. Perfluoroalkyl contam-inants in a food web from Lake Ontario. Environmental Science and Technology38, 5379–5385.

McLaughlin, C.L., Blake, S., Hall, T., Harman, M., Kanda, R., Foster, J., Rumsby, P.C.,2011. Perfluorooctane sulphonate in raw and drinking water sources in theUnited Kingdom. Water Environment Journal 25, 13–21.

Mihaich, E.M., Friederich, U., Caspers, N., Hall, A.T., Klecka, G.M., Dimond, S.S., Staples,C.A., Ortego, L.S., Hentges, S.G., 2009. Acute and chronic toxicity testing of bisphe-nol A with aquatic invertebrates and plants. Ecotoxicology and EnvironmentSafety 72, 1392–1399.

Morthorst, J.E., Holbech, H., Bjerregaard, P., 2010. Trenbolone causes irreversiblemasculinization of zebrafish at environmentally relevant concentrations.Aquatic Toxicology 98, 336–343.

Naile, J.E., Khim, J.S., Wang, T., Chen, C., Luo, W., Kwon, B.O., Park, J., Koh, C.H., Jones,P.D., Lu, Y., Giesy, J.P., 2010. Perfluorinated compounds in water sediment, soiland biota from estuarine and coastal areas of Korea. Environmental Pollution158, 1237–1244.

Newsted, J.L., Coady, K.K., Beach, S.A., Butenhoff, J.L., Gallagher, S., Giesy, J.P., 2007.Effects of perfluorooctane sulfonate on mallard and northern bobwhite quailexposed chronically via the diet. Environmental Toxicology and Pharmacology23, 1–9.

Novotny, L., Halouzka, R., Matlova, L., Vavra, O., Bartosova, L., Slany, M., Pavlik, I.,2010. Morphology and distribution of granulomatous inflammation in fresh-water ornamental fish infected with mycobacteria. Journal of Fish Diseases 33,947–955.

Oakes, K.D., Sibley, P.K., Martin, J.W., MacLean, D.D., Solomon, K.R., Mabury, S.A., VanDer Kraak, G.J., 2005. Short-term exposures of fish to perfluorooctane sulfonate:Acute effects on fatty acyl-CoA oxidase activity, oxidative stress, and circulatingsex steroids. Environmental Toxicology and Chemistry 24, 1172–1181.

OECD, 2002. Hazard Assessment of Perfluorooctane Sulfonate (PFOS) and its Salt.ENV/JM/RD(2002)17/FINAL.

OECD, 2011. OECD Guideline No. 234. Fish Sexual Development Test. OECD Publish-ing.

Peden-Adams, M.M., Keller, J.M., EuDaly, J.G., Berger, J., Gilkeson, G.S., Keil, D.E., 2008.Suppression of humoral immunity in mice following exposure to perfluorooc-tane sulfonate. Toxicological Sciences 104, 144–154.

Pickford, D.B., Hetheridge, M.J., Caunter, J.E., Hall, A.T., Hutchinson, T.H., 2003.Assessing chronic toxicity of bisphenol A to larvae of the African clawedfrog (Xenopus laevis) in a flow-through exposure system. Chemosphere 53,223–235.

Prosser, C.M.P.C.M., Unger, M.A., Vogelbein, W.K., 2011. Multistressor interactionsin the zebrafish (Danio rerio) Concurrent phenanthrene exposure and Mycobac-terium marinum infection. Aquatic Toxicology 102, 177–185.

Quinete, N., Wu, Q., Zhang, T., Yun, S.H., Moreira, I., Kannan, K., 2009. Specific profilesof perfluorinated compounds in surface and drinking waters and accumula-tion in mussels, fish, and dolphins from southeastern Brazil. Chemosphere 77,863–869.

Romeis, B., 1989. Mikroskopische Technik. Urban und Schwarzenberg.Rostkowski, P., Yamashita, N., So, I.M.K., Taniyasu, S., Lam, P.K.S., Falandysz, J., Lee,

K.T., Kim, S.K., Khim, J.S., Im, S.H., Newsted, J.L., Jones, P.D., Kannan, K., Giesy,J.P., 2006. Perfluorinated compounds in streams of the Shihwa industrial zoneand Lake Shihwa, South Korea. Environmental Toxicology and Chemistry 25,2374–2380.

Rushbrook, B.J., Katsiadaki, I., Barber, I., 2007. Spiggin levels are reduced in malesticklebacks infected with Schistocephalus solidus. Journal of Fish Biology 71,298–303.

Schmidt, F., Braunbeck, T., 2011. Alterations along the hypothalamic-pituitary-thyroid axis of the zebrafish (Danio rerio) after exposure to propylthiouracil.J of Thyroid Research 2011, 1–17.

Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T., Butenhoff, J.L.,2002. Subchronic toxicity studies on perfluorooctanesulfonate potassium saltin Cynomolgus monkeys. Toxicological Sciences 68, 249–264.

Skutlarek, D., Exner, M., Färber, H., 2006. Perfluorinated surfactants in sur-face and drinking waters. Environmental Science and Pollution Research 13,299–307.

Sohoni, P., Tyler, C.R., Hurd, K., Caunter, J., Hetheridge, M., Williams, T., Woods, C.,Evans, M., Toy, R., Gargas, M., Sumpter, J.P., 2001. Reproductive effects of long-term exposure to bisphenol A in the fathead minnow (Pimephales promelas).Environmental Science & Technology 35, 2917–2925.

Staples, C.A., Hall, A.T., Friederich, U., Caspers, N., Klecka, G.M., 2011. Earlylife-stage and multigeneration toxicity study with bisphenol A and fatheadminnows (Pimephales promelas). Ecotoxicology and Environment Safety 74,1548–1557.

Stock, N.L., Furdui, V.I., Muir, D.C.G., Mabury, S.A., 2007. Perfluoroalkyl contaminantsin the Canadian arctic: evidence of atmospheric transport and local contamina-tion. Environmental Science & Technology 41, 3529–3536.

Taniyasu, S., Kannan, K., Horii, Y., Hanari, N., Yamashita, N., 2003. A survey of per-fluorooctane sulfonate and related perfluorinated organic compounds in waterfish, birds, and humans from Japan. Environmental Science and Technology 37,2634–2639.

Thompson, J., Eaglesham, G., Mueller, J., 2011. Concentrations of PFOS, PFOA andother perfluorinated alkyl acids in Australian drinking water. Chemosphere 83,1320–1325.

UNIDO, 2009. Perfluorooctane sulfonate (PFOS) production and use: Past and currentevidence.

Van der Ven, L.T.M., Wester, P.W., Vos, J.G., 2003. Histopathology as a tool for theevaluation of endocrine disruption in zebrafish (Danio rerio). EnvironmentalToxicology and Chemistry 22, 908–913.

Wang, M.Y., Chen, J.F., Lin, K.F., Chen, Y.H., Hu, W., Tanguay, R.L., Huang, C.J., Dong,Q.X., 2011. Chronic zebrafish PFOS exposure alters sex ratio and maternalrelated effects in F1 offspring. Environmental Toxicology and Chemistry 30,2073–2080.

Watral, V., Kent, M.L., 2007. Pathogenesis of Mycobacterium spp. in zebrafish (Daniorerio) from research facilities. Comparative Biochemistry and Physiology C: Tox-icology & Pharmacology 145, 55–60.

Wei, Y.H., Dai, J.Y., Liu, M., Wang, J.S., Xu, M.Q., Zha, J.M., Wang, Z.J., 2007. Estrogen-like properties of perfluorooctanoic acid as revealed by expressing hepaticestrogen-responsive genes in rare minnows (Gobiocypris rarus). EnvironmentalToxicology and Chemistry 26, 2440–2447.

Yang, F.X., Xu, Y., Pfister, G., Henkelmann, B., Schramm, K.W., 2005. Nonylphenol,bisphenol-A and DDTs in Lake Donghu, China. Fresenius Environmental Bulletin14, 173–180.

Yeung, L.W.Y., Yamashita, N., Taniyasu, S., Lam, P.K.S., Sinha, R.K., Borole, D.V., Kan-nan, K., 2009. A survey of perfluorinated compounds in surface water and biotaincluding dolphins from the Ganges River and in other waterbodies in India.Chemosphere 76, 55–62.

Young, C.J., Furdui, V.I., Franklin, J., Koerner, R.M., Muir, D.C.G., Mabury, S.A., 2007.Perfluorinated acids in arctic snow: new evidence for atmospheric formation.Environmental Science and Technology 41, 3455–3461.