Transcriptional effects on glutathione S-transferases in first feeding Atlantic cod (Gadus morhua) larvae exposed to crude oil

Chemosphere 79 (2010) 905–913

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

Transcriptional effects on glutathione S-transferases in first feeding Atlantic cod (Gadus morhua) larvae exposed to crude oil P.A. Olsvik a,*, T. Nordtug b, D. Altin c, K.K. Lie a, I. Overrein d, B.H. Hansen b a

National Institute of Nutrition and Seafood Research, Nordnesboder 1-2, N-5005 Bergen, Norway SINTEF, Materials and Chemistry, Marine Environmental Technology, N-7465 Trondheim, Norway c Biotrix, N-7022 Trondheim, Norway d SINTEF, Fisheries and Aquaculture, N-7465 Trondheim, Norway b

a r t i c l e

i n f o

Article history: Received 15 January 2010 Received in revised form 3 March 2010 Accepted 14 March 2010 Available online 3 April 2010 Keywords: Atlantic cod Exposure Oil dispersion Water-soluble fraction of oil Glutathione S-transferase Transcription

a b s t r a c t Polycyclic aromatic hydrocarbons (PAHs) and other oil compounds are known to induce stress and impact health of marine organisms. Water-soluble fractions of oil contain components known to induce glutathione S-transferases (GSTs), one of the major classes of phase II detoxifying enzymes present in essentially all eukaryotic organisms. In this study, the transcriptional responses of six GSTs (GST pi, GST mu, GST omega, GST theta, GSY zeta and GST kappa) were examined in early larvae of Atlantic cod Gadus morhua exposed to five concentrations of dispersed oil (containing oil droplets and water-soluble fraction) and water-soluble fractions (WSF) of oil. When Atlantic cod larvae were exposed to WSF P PAH/L for 4 days), expression of GSTM3 and GSTO1 was significantly (containing 1.31 ± 0.31 lg increased, whereas no differences in GST expression were observed in larvae exposed to a corresponding P 50% lower amount of dispersed oil (containing 0.36 ± 0.10 lg PAH/L for 4 days). The study suggest that although the oil clearly had severe negative effects on the larvae (i.e. concentration-dependent lethality and growth reduction), only minor effects on GST transcription could be observed using RNA obtained from pooled whole-larvae homogenates. This result indicates that the expression of these important detoxification enzymes is only moderately inducible at such an early developmental stage either reflecting low tolerance of cod larvae to dispersed oil or alternatively that using whole-larvae homogenates may have masked tissue-specific mRNA induction. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction The glutathione S-transferases (GSTs) (GSTs: EC 2.5.1.18) represent a family of phase II enzymes that provide cellular protection against the toxic effects of a number of environmental toxicants. Mechanisms of detoxification by GSTs involve catalytic substrate conjugation and oxidant reduction with reduced glutathione (GSH). Advances in genomics have facilitated the identification and classification of multiple forms of fish GST, as has been true for other phase II drug-metabolizing enzymes (Blanchette et al., 2007; Schlenk et al., 2008). The GST super-family includes cytosolic and mitochondrial GSTs. Based on amino acid sequence similarities, seven classes of cytosolic GSTs are recognized in mammalian organisms, designated alpha, mu, pi, sigma, theta, omega, and zeta (Hayes et al., 2005). The kappa class GSTs are quite distinct from cytosolic GSTs, and they represent the mitochondrial GST isoenzymes. Even though GSTs and uridine diphosphate glucuronosyltransferases (UGTs), together with the sulfotransferases * Corresponding author. E-mail address: [email protected] (P.A. Olsvik). 0045-6535/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2010.03.026

(SULTs), make up the main phase II detoxifying enzymes, involved in the biotransformation of many POPs, they are also considered to be important antioxidants, involved in antioxidant responsive element (ARE)-regulated gene expression and the protection of cells against reactive oxygen species. Relatively little is known about GST transcription in fish exposed to various toxicants (Schlenk et al., 2008). Typically, only modest alteration of overall GST activity has been reported under most conditions (2-fold or less) (Henson et al., 2001; Henson and Gallagher, 2004). For example, Hasselberg et al. (2004) reported a modest 1.3-fold reduced GST activity in first-spawning male Atlantic cod exposed to 0.02 ppm alkylphenol. Four classes of GSTs have previously been characterized in salmonids: pi, mu, theta and alpha (Donham et al., 2005). Analysis of our Atlantic cod EST data (as of August 2009) suggests that these four isoforms also are present in this North-Atlantic species (unpublished data). Based upon studies from the plaice (Pleuronectes platessa), a unique fish GST class, Rho, has been designated (Schlenk et al., 2008). Trute et al. (2007) characterized GSTs in coho salmon (Oncorhynchus kisutch) and described two major isoforms in liver, the pi and thetaclass GSTs, but noted that they might have a limited capacity to

906

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

conjugate substrates of various toxicants and endogenous compounds associated with cellular oxidative stress. We recently showed that intraperitoneal injection of 50 mg kg1 body mass bnaphthoflavone in Atlantic salmon (Salmo salar) resulted in an 3.6-fold induction of GST pi in the liver, while in comparison CYP1A was 121 fold induced (Olsvik et al., 2007). In juvenile Atlantic cod exposed to environmental relevant levels of nonylphenol (30 lg L1), the transcriptional levels of liver GST pi increased 4.6-fold after 3 weeks of exposure (Olsvik et al., 2009). These results clearly show that the pi class GST is inducible in Atlantic cod exposed to alkylphenols, one of the major components in water-soluble fractions of crude oil. Oil exploration and production in the Atlantic Ocean move northwards towards more sensitive areas like the Lofoten/Vesterålen area in Norway and the Arctic regions. Because of this there is a fundamental need to understand what potential effects oil may have on different sentinel species, especially fish larvae which are expected to be particularly vulnerable to dispersed oil (Elskus et al., 2005). For larvae of the Atlantic cod (Gadus morhua), one of the ecological most important North Atlantic fish species, very little is known about the levels to which dispersed oil causes narcosis, growth effects and death. A low tolerance to exposure might be a result of low ability to metabolize oil components. The aim of this study was therefore to study effect limits for transcriptional induction of a set of GSTs, survival and growth in Atlantic cod larvae exposed to environmentally realistic weathered oil with (dispersed oil) and without oil droplets (water-soluble fraction of oil [WSF]). Cod larvae were exposed for 4 days during the first feed period (9–13 days post hatching (dph)) and sampled at 17 dph. Chemical composition of oil compounds in the exposure media was analyzed and larvae survival and growth were measured at the end of the experiment. The transcriptional levels of six GSTs were quantified in whole-larvae homogenates. These genes were encoding GST pi, GST mu, GST omega, GST zeta, GST theta and GST kappa. Genes encoding several additional relevant proteins were also attempted measured, including cytochrome P450 1A (CYP1A), CYP3C, sulfotransferase (SULT) and uridine diphosphate glucuronosyltransferase (UGT).

2. Materials and methods 2.1. Materials and experimental set up Fresh crude oil was artificially weathered by heating to 200 °C. The resulting 200 °C + residue was dispersed into filtered seawater (5 lm) through a series of nozzles yielding a constant flow of dispersion with a homogenous droplet size. The principle of the exposure system (Fig. 1) is based on generating two dilution series, one with dispersion and a second with the corresponding water-soluble fraction (WSF) isolated from the dispersions. For each dilution step the dispersion is diluted with seawater to a third of the previous concentration. To distinguish between toxic stress caused by oil droplets and WSF, each concentration in the dispersion dilution series was filtered to a parallel exposure unit containing only the WSF. The filter unit consisted of fine glass wool on top of a Watman GFC glass filter. Retention of oil droplets was verified by chemical analysis and particle counting. The exposure containers consisted of 5 L borosilicate glass bottles with their bottoms removed. Exposure solution and clean seawater (controls) were added in the lower part of the exposure container through Teflon tubing (bore 1 mm). Water was drained from the surface through a 300 lm plankton mesh. The temperature was controlled by submerging the units into a water bath. The flow through in all exposure units was kept constant at

10 mL ± 0.5 mL min1. A peristaltic pump (Watson-Marlow) equipped with MarphreneÒ tubing was used to drive the dispersion through the glass filters. Dispersions were added by passive flow through the inlet Teflon tubes as resistance and adjusted by changing the height of the inlet water column. Three parallel exposure units were used in order to achieve biological replicates for every exposure concentration. The principle of the exposure units is shown in Fig. 1. Fertilized cod eggs (Gadus morhua) from Marine Harvest Cod (Norway) were transported at an age of 42 day degrees and disinfected upon arrival at the laboratory (glutaraldehyde: 400 ppm for 6 min). The eggs were incubated at 7 °C with a water exchange rate of seven times per day until 4 dph, when the larvae were transferred to the exposure units at 7.5 °C. The temperature was gradually increased and reached 12 °C at 15 dph. Each experimental unit was stocked with 240 larvae at 4 dph and dead larvae were replaced after 1 day in the experimental units. The larvae were given micro algae (Isochrysis galbana) to a density of 1–2 mg carbon L1 and rotifers (12 000 rotifers L1) three times a day from 3 dph until the last sampling at 17 dph. The rotifers ‘Cayman’, belonging to Brachionus plicatilis sensu lato group (Gomez et al., 2002), were grown on rotifer diet (Reed mariculture, US; 0.8 mL million1 rotifers day1) and Multigain (Danafeed, DK; 0.12 g million1 rotifers1 day1). Cod larvae were exposed to a gradient of different concentrations of dispersed oil or the corresponding water soluble oil fraction (as described) from 9 dph until 13 dph, and the experiment was ended at 17 dph after a recovery period in clean sea water when larvae was sampled for gene expression analyses. The effect of the exposure on the cod larvae was compared with identical control units containing non-exposed cod larvae. In order to verify the oil droplet size distribution, water samples were collected from the exposure chambers on day one and three of the exposure period and analyzed on a particle analyzer (Coulter Counter Multisizer 3; Beckman). The samples contained both rotifers and oil droplets. 2.2. Larvae survival and dry weight Dry weight was determined at the end of the experiment at day 17 (n = 36 larvae per treatment). Larvae were anesthetized with metacaine (FinquelÒ, Argent Laboratories, Redmond, USA) and rinsed for a few seconds in fresh water before transferred to pre weighted tin capsules. Capsules were dried for minimum 48 h at 60 °C before weight determination. Survival was estimated based on the number of larvae transferred to the units, counts of dead larvae removed during the experiment, and the survival larvae corrected for larval sampling. 2.3. Image capture and fluorescence microscopy The feeding activity of the larvae was assessed as gut filling visualized by the autofluorescence of algal chlorophyll a and degradation products. Prior to examination, a number of larvae were randomly chosen from the exposure units at 14 dph and irreversibly anesthetized with metacaine. Examination of the anesthetized larvae was performed on an inverted microscope (Nikon TE2000, Nikon Corp., Tokyo, Japan) with a 10 planapochromatic objective (Nikon Corp., Tokyo, Japan). Fluorescence in the gut was induced by illuminating the specimen with a 120 W mercury arc lamp (xcite 120, EXFO Corp., Quebec, Canada) passing through a B-2A filtercube (Nikon Corp., Tokyo, Japan) in the microscope. Images were captured with a Peltier cooled CCD camera (DS 5Mc, Nikon Corp., Tokyo, Japan) controlled from a computer running NIS Elements F (Nikon Corp., Tokyo, Japan).

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

907

Fig. 1. (A) Principle design of the experimental setup. Three parallel exposure setups were used in order to achieve biological replicates for every exposure concentration. Note that the dilution gradient is reversed relative to normal practice with the highest concentration furthest from the inlet. This is done in order to maintain a high flow rate of the dispersion through the inlet tube, thus minimizing settling of oil droplets within the tube. (B) The figure shows an overview of the experiment (feeding, exposure period and when the cod larvae were sampled for gene expression analyses, growth measurement and survival was observed). X is the time when the yolk sack was consumed.

2.4. Larvae sampling and RNA extraction At the end of the experiment, the whole cod larvae were immediately rinsed with distilled water and blotting paper and thoroughly homogenized with RNase-free disposable pellet pestles specially designed to match 1.5 mL microtubes (Kontes, New Jersey) in 0.8 mL Trizol reagent (Invitrogen, Carlsbad, CA, USA) on ice, flash-frozen in liquid nitrogen, and stored at 80 °C before RNA isolation. Total RNA was extracted from the batches of pooled larvae using Trizol reagent according to the manufacturer’s instructions and stored in 50 lL RNase-free MilliQ H2O. Genomic DNA was eliminated from the samples by DNase treatment according to the manufacturer’s description (Ambion, Austin, TX, USA). The RNA was then stored at 80 °C before further processing. The quality of the RNA was assessed with the NanoDrop ND1000 UV–Vis Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). The RNA 6000 Nano LabChipÒ kit (Agilent Technologies, Palo Alto, CA, USA) was used to evaluate the integrity of the RNA. The RNA integrity number (RIN) of the 27 samples was 7.33 ± 0.12 (mean ± SEM). 2.5. Chemical analyzes Water samples for chemical analysis (approx. 900 mL each) were collected 1 and 3 days into the exposure period from all exposures and from control groups and acidified with diluted hydrochloric acid. Acidified water samples were extracted with dichloromethane (DCM), dried over Na2SO4 and concentrated to 1 mL. Analysis for semi-volatile compounds, including phenols, naphthalenes and 3–5 ring polycyclic aromatic hydrocarbons (PAH) was performed by gas chromatography–mass spectrometry (GC–MS) operated in selected ion monitoring (SIM) mode. The system comprised a HP6890 N gas chromatograph fitted with a Hewlett–Packard HP7683B Series auto-sampler and a HP5975B quadrupole mass selective detector. The column was a Phenomenex ZB-5MS fused silica capillary column (30 m  0.25 mm id  0.25 mm film thickness). The carrier gas was helium at a con-

stant flow of 1.0 mL min1. A 1.0 mL sample was injected into a 310 °C splitless injector. The oven temperature was programmed from 40 °C for 1 min, then to 315 °C at 6 °C min1 and held for 15 min. Data and chromatograms were monitored and recorded using MSD ChemStation (version D.03.00.611) software. The quadrupole mass spectrometer ion source temperature was 230 °C. Determination of the total extractable organic compounds (TEOC) was performed on DCM extracts by gas chromatography with flame ionization detection (GC-FID). The system comprised a HP6890 gas chromatograph fitted with a Hewlett–Packard HP7683B Series auto-sampler. The column was a HP-5 fused silica capillary column (30 m  0.32 mm id  0.25 mm film thickness). The carrier gas was helium at a constant flow of 1.0 mL min1. A 1.0 mL sample was injected into a 310 °C splitless injector. The oven temperature was programmed from 40 °C for 1 min, then to 315 °C at 6 °C min1 and held for 15 min. Unless otherwise noted, concentrations are given as the mean value of the analyses from day 1 and day 3 of the exposure period. 2.6. Quantitative real-time RT-PCR PCR primer sequences used for the quantification of the transcriptional levels of the target genes GSTP, GSTM, GSTO, GSTZ, GSTT, GSTK, as well as the reference genes b-actin (ACTB), elongation factor 1 alpha (EF1A) and ubiquitin (UBI), are shown in Table 1. Also included are the PCR primer sequences used to assess the transcriptional levels of four additional genes, i.e. CYP1A, CYP3C, SULT2, and UGT. The transcriptional levels of three of these genes were too low in the examined larvae samples to be accurately quantified. The primer pairs amplify PCR products between 69 and 145 base pairs (bp) long. qPCR assays were designed using Primer Express 2.0 software (Applied Biosystems, Foster City, CA, USA) to select appropriate primer sequences from sequenced cod genes. Due to the lack of genomic information, exon–exon borders were not considered when designing the PCR primers. A two-step real-time RT-PCR protocol was developed to measure the mRNA levels of the 10 target genes in liver tissue of Atlantic cod. The RT reactions were run in duplicate on 96-well reaction

908

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

Table 1 Accession numbers, PCR primers, amplicon sizes and PCR efficiencies of the RT-qPCR assays used in the current experiment. Gene

Accession no.

Forward primer (50 –30 )

Reverse primer (50 –30 )

Amplicon size (bp)

PCR efficiency

GST mu GST pi GST theta GST omega GST zeta GST kappa CYP1A CYP3C UGT SULT ACTB EF1A Ubiquitin

>Contig6399 EX730032 >Contig6168 >Contig6402 >Contig293 >Contig6905 EX725014 EX725058 EX722276 EX724532 EX739174 EX722124 EX735613

CCGGTTGACGGTGTAGATTCA GTCCCCCTGCTGCCATTC ATCACCCTGCATATACGAAAACG CATGCGCTCAGTCCACTTCTT AACCTCGCTGGATGAAATGC CCAGCCCAGGAGGTTTGTTT CCTTGACCTCTCGGAGAAAGAC CCCTATGCCTACATGCCCTTT GTAAAAATAATGAAGTGGCTACCTCAAA TGGCCAACTACTCATCATTTGAAA CACAGCCGAGCGTGAGATT CGGTATCCTCAAGCCCAACA GGCCGCAAAGATGCAGAT

TCGAGGCTCTGGAGAAGATTTC CCTCCATACACCGCCACCTA GCCAAAACCTTCCAGGACAA TGGCTGGTGACTCAATCACAA CCAGGTGCGGATGATATGTG CCGTACACCTGGCTTGCATT CGCCCCGCTAGCTATAGACA TCCAGAGGAACAACGGTGTCT GAACGCCGTGGCAGATG CTGGGTTACGGTGAAGTGGTTT ACGAGCTAGAAGCGGTTTGC GTCAGAGACTCGTGGTGCATCT CTGGGCTCGACCTCAAGAGT

119 126 123 125 122 127 145 143 115 101 95 93 69

1.97 1.98 1.81 2.00 1.96 1.99 nd nd 2.08 nd 2.12 2.06 1.94

nd = Not detected.

plates with the GeneAmp PCR 9700 machine (Applied Biosystems, Foster City, CA, USA) using TaqMan Reverse Transcription Reagent containing Multiscribe Reverse Transcriptase (50 U lL1) (Applied Biosystems, Foster City, CA, USA). Twofold serial dilutions of total RNA were made for efficiency calculations. Six serial dilutions (1000–31 ng) in triplicates were analyzed in separate sample wells. Total RNA input was 500 ng in each reaction for all genes. No template control (ntc) and RT-control were run for quality assessment. RT-controls were not performed for every individual sample, but were run for each assay or gene. Reverse transcription was performed at 48 °C for 60 min by using oligo dT primers (2.5 lM) for all genes in 30 lL total volume. The final concentration of the other chemicals in each RT reaction was: MgCl2 (5.5 mM), dNTP (500 mM of each), 10 TaqMan RT buffer (1), RNase inhibitor (0.4 U lL1) and Multiscribe reverse transcriptase (1.67 U lL1) (Applied Biosystems). Diluted cDNA (1.0 lL cDNA from each RT reaction) was transferred to a new 96-well reaction plate and the qPCR run in 10 lL reactions on the LightCycler 480 Real-Time PCR System (Roche Applied Sciences, Basel, Switzerland). Real-time PCR was performed by using SYBR Green Master Mix (LightCycler 480 SYBR Green master mix kit, Roche Applied Sciences,), which contains FastStart DNA polymerase and gene-specific primers (500 nM). PCR was achieved with a 5 min activation and denaturizing step at 95 °C, followed by 45 cycles of a 15 s denaturing step at 95 °C, a 60 s annealing step and a 30 s synthesis step at 72 °C. Target gene mean normalized expression (MNE) was determined using a normalization factor calculated by the geNorm software based on the three selected reference genes, i.e. ACTB, EF1AB and ARP (Vandesompele et al., 2002). geNorm determines the individual stability of a gene within a pool of genes, and calculates the stability according to the similarity of their expression profile by pair-wise comparison, using the geometric mean as a normalizing factor. The gene with the highest M, i.e. the least stabile gene, is then excluded in a stepwise fashion until the most stabile genes are determined. Here a normalizing factor based on all three examined reference genes was used to calculate the MNE.

2.7. Statistics The GraphPad Prism 4.0 software (GraphPad Software, Inc., San Diego, CA, USA) was used for the statistical analyses of the gene expression data. Since each sample for transcriptomic analysis consisted of a pool of many larvae, a parametric t-test was used to compare the transcriptional levels of the examined genes between the control and the group exposed to the highest oil concentrations with three replicate groups, i.e. D2 oil dispersion or W3 for WSF.

Based on previous experience, gene expression data most often follow a Gaussian distribution, even if individual variation often can be large due to health status, gender and the fact that most fish species are highly outbred. Spearman rank correlation was used for non-parametric correlation analysis. Equality of variance was tested with Bartletts test. An alpha level of 0.05 was considered significant. 3. Results 3.1. Chemical analyzes Table 2 shows a summary of the chemical data determined from GC–MS SIM analysis of the dispersion (D1–5) and WSF (W1–5) solutions. Concentrations are given as average of three replicates sampled and analyzed at two times (at day 1 and 3) during the exposure period. As in a previous experiment (Hansen et al., 2009) oil droplet distribution ranged from approximately 4 to 14 lm with a peak at about the mean oil droplet size of 8 lm and a half width of the droplet fraction of 10 lm. 3.2. Cod larvae survival All larvae exposed to the highest concentration of dispersed oil (D5) and almost all exposed to the second highest concentration of dispersed oil (D4) died during the exposure experiment, leaving larvae for RNA extraction only from the three lowest concentrations of dispersed oil (D1–D3). With only one remaining larvae after dry weight sampling in one of the D3 (oil dispersion) replicates, we were only able to obtain gene expression data from two groups (n = 2, N = 16) exposed to this concentration and also for the D1 (oil dispersion) group due to a technical error. For the WSF exposure, seven individual larvae (1 + 6 + 0) remained in the highest exposure groups (W5), and 17 individuals (7 + 10 + 0) in the groups exposed to the second highest concentration (W4). Fig. 2A and B shows survival (% survival relative to 9 dph) of the larvae in the chambers after 4 days exposure to dispersed oil (D) or WSF (W) followed by 4 days recovery in clean seawater. Fig. 2C and D shows the weight of the surviving larvae (as lg dry weight) at termination of the experiment at 17 dph. 3.3. Gene expression As of August 2009, we were able to obtain cod mRNA sequences encoding five of the seven separate families of cytosolic GST enzymes (GST class mu, omega, pi, theta and zeta) and one of the two membrane families (GST class kappa). Thus, transcriptional

909

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

Table 2 Summary of chemical data determined by GC–MS in SIM mode for the dispersed (D1–D5) and WSF (W1–W5) exposure media. Concentrations are given as average and standard error of three replicates (one from each of the replicate chambers) sampled and analyzed at two times during the exposure experiment. Sample

Sum all compounds mass (lg/L)

P Decalins mass (lg/L)

P Naphthalenes mass (lg/L)

P

Phenantrenes mass (lg/L)

P Dibenzothiophenes mass (lg/L)

P

PAH 2 + rings mass (lg/L)

P Phenols mass (lg/L)

D1 D2 D3 D4 D5

0.91 ± 0.16 1.56 ± 0.30 9.99 ± 1.32 32.98 ± 4.43 102.01 ± 9.28

0.00 0.01 ± 0.01 0.56 ± 0.14 2.78 ± 0.33 8.85 ± 0.78

0.33 ± 0.09 0.88 ± 0.18 5.95 ± 0.67 20.76 ± 2.33 64.39 ± 5.00

0.05 ± 0.01 0.12 ± 0.03 1.01 ± 0.11 3.08 ± 0.39 8.75 ± 1.06

0.01 ± 0.01 0.05 ± 0.01 0.27 ± 0.03 0.74 ± 0.08 1.99 ± 0.16

0.14 ± 0.08 0.36 ± 0.10 2.89 ± 0.55 8.48 ± 1.98 25.68 ± 5.67

0.42 ± 0.11 0.28 ± 0.06 0.42 ± 0.10 0.33 ± 0.08 1.11 ± 0.42

W1 W2 W3 W4 W5

0.96 ± 0.18 1.21 ± 0.25 5.80 ± 0.59 17.02 ± 1.42 39.33 ± 2.09

0.00 0.00 0.04 ± 0.01 0.10 ± 0.03 0.09 ± 0.02

0.27 ± 0.06 0.52 ± 0.18 3.94 ± 0.37 13.31 ± 0.93 33.39 ± 1.59

0.03 ± 0.01 0.07 ± 0.02 0.40 ± 0.04 0.88 ± 0.06 1.32 ± 0.05

0.03 ± 0.01 0.02 ± 0.02 0.13 ± 0.01 0.22 ± 0.01 0.32 ± 0.02

0.17 ± 0.06 0.27 ± 0.10 1.31 ± 0.31 2.71 ± 0.54 4.15 ± 0.57

0.50 ± 0.09 0.40 ± 0.07 0.39 ± 0.10 0.46 ± 0.09 0.41 ± 0.05

P P P Decalins include decalin and C1–C4-alkylated homologues (C0–C4). Naphthalenes include naphthalene and C1–C4-akylated homologues (C0–C4). Phenantrenes P include Phenatrene, anthracene and their C1–C4-alkylated homologues (C0–C4). Dibenzothiophenes include dibenzothiophenes and C1–C4-alkylated homologues (C0–C4). P PAH 2 + rings include benzothionphenes (C0–C4), acenaphthylene, acenaphthene, dibenzofuran, fluorenes (C0–C3), phenanthrenes (C0–C4), anthracenes (C0–C4), dibenzothiophenes (C0–C4), fluoranthenes (C0–C3), pyrenes (C0–C3), benz(a)anthracene, chrysenes (C0–C4), benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(e)pyrene, benzo(a)pyrene, perylene, indeno(1,2,3-c,d)pyrene, dibenz(a,h)anthracene and benzo(g,h,i)perylene.

Fig. 2. Survival (% of total) of Atlantic cod larvae in the chambers after exposure to dispersed oil (A) and WSF (B) at day 17 dph. Weight of larvae, as an indicator of growth, measured after the end of the experiment, is given as dry weight for the individual groups exposed to dispersed oil (C) and WSF (D). D1–5 = dispersed oil solutions. W1– 5 = WSF of oil solution.

data were not obtained for GST class alpha, sigma (cytosolic) and microsomal GST. In addition, transcriptional data were attempted obtained for four other potential marker genes for oil exposure, i.e. CYP1A, CYP3C, SULT2 and UGT. No transcriptional data were obtained for CYP1A, CYP3C and SULT2 in the examined cod larvae due to their very low expression levels, even though we previously have been able to quantify the transcriptional levels of several of these genes in cod larvae of similar age. Table 3 shows the best annotation of the six GST class genes (BLASTX), e-score and func-

tion according to the GeneCard database (of mammalian homologs). Due to low RNA yield, expression data was obtained only from larvae exposed to the three lowest concentrations of dispersion and WSF (see above). Statistical analysis was therefore performed between the control and the D2 group (n = 3, N = 33), and between the control and W3 group (n = 3, N = 56); i.e. between the control and group exposed to the highest concentration of either dispersed oil or WSF of oil of which we were able to obtain data from three

910

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

Table 3 GST gene function. Protein function is obtained from the GeneCards database. Gene

Best annotation

E-score BLASTX

Function (GeneCards database) The mu class of enzymes functions in the detoxification of electrophilic compounds, including carcinogens, therapeutic drugs, environmental toxins and products of oxidative stress, by conjugation with glutathione This GST family member is a polymorphic gene encoding active, functionally different GSTP1 variant proteins that are thought to function in xenobiotic metabolism and play a role in susceptibility to cancer, and other diseases Glutathione S-transferase (GST) theta 1 (GSTT1) is a member of a super-family of proteins that catalyze the conjugation of reduced glutathione to a variety of electrophilic and hydrophobic compounds. Acts on 1,2-epoxy-3-(4nitrophenoxy)propane, phenethylisothiocyanate 4nitrobenzyl chloride and 4-nitrophenethyl bromide. Displays glutathione peroxidase activity with cumene hydroperoxide This gene encodes a member of the theta class glutathione Stransferase-like (GSTTL) protein family. In mouse, the encoded protein acts as a small stress response protein, likely involved in cellular redox homeostasis. Exhibits glutathionedependent thiol transferase and dehydroascorbate reductase activities This gene is a member of the glutathione S-transferase (GSTs) super-family and plays a significant role in the catabolism of phenylalanine and tyrosine. Bifunctional enzyme showing minimal glutathione-conjugating activity with ethacrynic acid and 7-chloro-4-nitrobenz-2-oxa-1,3-diazole and maleylacetoacetate isomerase activity. Has also low glutathione peroxidase activity with T-butyl and cumene hydroperoxides. Is able to catalyze the glutathione dependent oxygenation of dichloroacetic acid to glyoxylic acid This gene encodes a member of the kappa class of the glutathione transferase super-family of enzymes that function in cellular detoxification. The encoded protein is localized to the peroxisome and catalyzes the conjugation of glutathione to a wide range of hydrophobic substates facilitating the removal of these compounds from cells. Significant glutathione-conjugating activity is found only with the model substrate, 1-chloro-2,4-dinitrobenzene (CDNB)

GST mu

GSTM3

Glutathione S-transferase mu 3 [Anoplopoma fimbria]

Expect = 1e87

GST pi

GSTP1

Glutathione S-transferase pi [Danio rerio]

Expect = 4e78

GST theta

GSTT1

Glutathione S-transferase theta-1 [Salmo salar]

Expect = 1e94

GST omega

GSTO1

Glutathione S-transferase omega [Takifugu obscurus]

Expect = 9e95

GST zeta

GSTZ1

Gluthathione S-transferase zeta [Takifugu obscurus]

Expect = 1e73

GST kappa

GSTK1

Glutathione S-transferase kappa 1 [Anoplopoma fimbria]

Expect = 4e102

replicate tanks. Only two of the six GSTs were significantly upregulated in exposed larvae (Fig. 3). Transcripts encoding GSTM1 (Fig. 3A) and GSTO1 (Fig. 3B) were significantly upregulated in larvae exposed to the W3 solution compared to the control (t-test, P = 0.007 and P = 0.03, respectively). The control group consisted of 184 larvae pooled from six replicate tanks, whereas the W3 group consisted of 56 larvae pooled from three replicate tanks. None of the other quantified genes showed any transcriptional response to the exposure, although GSTT1 showed a non-significant trend towards downregulation in larvae exposed to oil dispersion (Fig. 3C). Spearman rank correlation was used to search for possible co-regulation among the six examined GST isoforms. The transcriptional levels of GSTM3 and GSTO1 were highly correlated (Spearman rank correlation, r2 = 0.93). Significant correlations (P < 0.05) were also observed between GSTM3 and GSTZ1 (r2 = 0.51), GSTM3 and GSTK1 (r2 = 0.52), GSTO1 and GSTK1 (r2 = 0.53), GSTO1 and GSTZ1 (r2 = 0.62), GSTZ1 and GSTT1 (r2 = 0.64) and between GSTZ1 and GSTK1 (r2 = 0.48). 3.4. Microscopy In order to further assess the impact of the oil exposure, a selected number of larvae were subjected to microscopic gut content evaluations 1 day after end of exposure (14 dph). Representative images of cod larvae exposed to WSF (W4), dispersed oil (D4) and controls are shown in Fig. 4. Absence of gut content in larvae exposed to D4 (oil dispersion) suggests that they ceased feeding.

Microscopic images of cod larvae were taken without (left) and with (right) fluorescence detection. Fig. 4A shows cod larvae from the control group, Fig 4B shows cod larvae from the W4 group and Fig. 4C shows cod larvae from the D4 group. WSF-exposed cod larvae continued feeding even at high concentrations of WSF, although visual inspections indicated lower feeding compared to controls. In the fluorescence images food (rotifers) are visible inside the intestine in the control cod larvae and the cod larvae exposed to WSF, however, in the dispersed oil-exposed cod larvae, the intestine contains no food. An oil droplet is, however, visible in the intestine (arrow). 4. Discussion Experiments conducted in our laboratory have previously shown a dose-dependent increase in GST gene expression in the copepod Calanus finmarchicus exposed to naphthalene (Hansen et al., 2008), and more recently after exposure to dispersed oil and water-soluble fractions (WSF) of North Sea oil using an identical system as used in the current experiment (Hansen et al., 2009). Hence, the experimental setup using a realistically weathered crude oil should represent an excellent exposure system for investigating potential expression of detoxification enzymes in cod larvae. In the current examination effects of oil dispersion and WSF of oil were studied at environmentally realistic levels. Only two of the six examined GST genes, GSTM3 and GSTO1, displayed significant increased expression in Atlantic cod larvae exposed to WSF (W3)

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

Fig. 3. Significantly increased expression of GST mu (A) and GST omega (B) in Atlantic cod larvae exposed to WSF (W3) compared to the control (Control W). GST theta (C). Control D: n = 5, D1: n = 2, D2: n = 3, D3: n = 2, Control W: n = 6, W1: n = 3, W2: n = 3, W3: n = 3. Number of larvae in the groups is shown above each column.

of North Sea oil compared to the control larvae. Based on the current knowledge of the biological function of the GST enzymes these genes are encoding, the results suggest that exposure to WSF have mediated both a detoxifying reaction (GSTM3 – phase II conjugation or reduced glutathione with electrophilic compounds) and a

911

stress response reaction (GSTO1 – oxidative stress) in the cod larvae. None of the other examined genes displayed any altered expression in the oil-exposed cod larvae. The survival data clearly show that the larvae were severely affected by the treatment. At the highest concentration of both dispersed oil and WSF of oil (D5/W5), high mortalities were observed (Fig. 2). All cod larvae were dead after the exposure in the highest dispersion concentration (D5), and at the highest WSF concentration only 9.1% had survived. Also at the second highest concentrations (D4/W4), survival was low, with 6.0% and 14.8% for dispersed oil and WSF, respectively. Also the weight of the larvae was affected by exposure to WSF and dispersed oil. The data strongly suggest a reduction of growth when comparing the weight of the exposed larvae compared to controls (Fig. 2). GSTs are widely distributed in nature, and most eukaryotic species contain multigene families, many of which are expressed in many cell types (Hayes et al., 2005). GSTs are most abundant in the liver, comprising about 2–4% of total cytosolic enzymes (Schlenk et al., 2008). Proteins of the alpha, mu, pi and theta-like families have been identified in many fish species, based on immunological cross-reactivity with family-specific antisera (Blanchette et al., 2007; Schlenk et al., 2008). The current examination shows that Atlantic cod possesses most of the GST isoforms previously described in fishes. Konishi et al. (2005) have shown that the pi-class homolog is the predominant GST isoform in gadoids. In contrast, our results suggest that the six examined GST isoforms were relatively homogenously expressed in whole-larvae cod homogenates (based on raw Ct evaluations). Kim et al. (2010) recently described the expression profiles of seven GST genes (GSTA, GSTT, GSTM, GSTK, GSTO, GSTZ and GSTMA (microsomal GST)) in Cd-exposed pufferfish (Takifugu obscurus). At basal levels, liver expression was highest for GSTM, followed by GSTT, GSTMA and GSTZ, with 5- to 10-fold higher expression than in other tissues (examined tissues: liver, skin, muscle, intestine, kidney, gonad, gills, eye and brain). Their results suggest that GSTM, GSTT, GSTMA and GSTZ are primarily associated with detoxification in liver. Contrarily, GSTO was evenly expressed between the different tissues. It has previously been shown that tissue distribution of GSTs in fish are linked to differential susceptibility to antioxidant damage or the presence of detoxifying enzymes (Doi et al., 2004; Lee et al., 2006). In mammals, the GST mu class of enzymes functions in the detoxification of electrophilic compounds, including carcinogens, therapeutic drugs, environmental toxins and products of oxidative stress, by conjugation with glutathione, whereas the GST omega class of enzymes encode members of the theta class glutathione S-transferase-like (GSTTL) protein family, acting as small stress response protein, likely involved in cellular redox homeostasis and exhibits glutathionedependent thiol transferase and dehydroascorbate reductase activities (GeneCards database). According to the Comparative Toxicogenomics Database (CTD) (CTD, 2009), GSTM respond to several components in oil, such as polycyclic aromatic hydrocarbons (PAHs) and benzene, in addition to other chemicals and endogenous glutathione. GSTO, on the other hand, respond to chemicals like arsenic, estradiol, ethinyl estradiol, genistein and others. Our results therefore suggest that cod larvae GSTM is involved in the metabolism of some of the heavy components in the oil, such as PAHs, most likely predominantly by glutathione conjugation in the liver, while other components in the oil probably triggered the increased transcription of GSTO1. These components may have been chemicals acting via the antioxidant response element (ARE) or via the estrogen receptor, as suggested by annotated data curated by the CTD. Since we were using RNA from whole-larvae homogenates, it is uncertain whether the increased GSTO1 expression was liver-specific. The increased transcription of GSTM3 in cod larvae exposed to WSF is likely to be induced as a result of exhaustion of cytosolic glutathione (GSH)

912

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

Fig. 4. Microscopic images of Atlantic cod larvae taken without (left) and with (right) fluorescence detection. (A) Cod larvae from the control group. (B) Cod larvae from the W4 group exposed to water-soluble fractions of oil. (C) Cod larvae from the D4 group exposed to dispersed oil. In the fluorescence images food (rotifers) are visible inside the intestine in the control cod larvae and the cod larvae exposed to WSF, however, in the dispersed oil-exposed cod larvae the intestine contains no food. An oil droplet is, however, visible in the intestine (arrow). Pictures are taken at 14 dph.

following oil exposure. In liver of pufferfish exposed to Cd, GSTM showed the strongest up-regulation amongst the up-regulated GSTs (Kim et al. 2010), in line with our findings. It remains unknown as to why the other four GST genes examined in the present study did not show any increased transcription in the exposed cod larvae. One possible explanation is the duration of the exposure experiment; increased expression of GST genes in fish often occur 6–24 h after the initialization of exposure (Bouraoui et al., 2008; Kim et al. 2010). In pufferfish exposed to Cd, GSTM and GSTO, as well as GSTZ, reached a transcriptional peak after 24 h of exposure to 5 mg Cd L1 for 96 h (Kim et al. 2010). Exposure duration of 4 days, as used in our experiment, may therefore have boosted the transcription initially, but mediated a return toward the baseline levels after 4 days of exposure. Although major cellular detoxifying systems like phase II conjugation reaction often are similar between species, varying responses are to be expected across different phyla. Further, fish species do not always respond similarly regarding expression of GSTs, therefore there is a need for careful evaluation of responses across species. In addition, the biological control of GST enzymes is complex depending on sex, age, tissue, and species, and also varies between individuals (Hayes and Pulford, 1995; Chiou et al., 1997). Thus, detailed evaluation on the

function of each GST member in Atlantic cod larvae requires further research. In contrast to cod larvae exposed to WSF (W3), GSTM3 and GSTO1 transcription was not significantly induced in larvae exposed to dispersed oil (D2). One possible explanation for this finding, apart from the fact that the dispersed oil solution was more diluted (Table 2), may be the observed difference in gut content between the two groups of larvae. The microscopic evaluation of gut content of a selected number of larvae exposed to the second highest concentrations of dispersion (D4) and WSF (W4) suggest that the WSF group was feeding more efficiently (representative larvae are shown in Fig. 4). WSF-exposed cod larvae continued feeding even at high concentrations of WSF, although visual inspections indicated lower feeding compared to controls. The reduction in feeding seems to be an adequate explanation for reduced weight compared to the control larvae, and perhaps also an explanation for no induced stress gene expression in the dispersion-exposed larvae (compared to WSF-exposed larvae). In case the cod larvae exposed to the oil dispersion in general were eating less of the supplied rotifers than the larvae in the corresponding WSF concentration, the developmental status of the liver might have been affected as less developed livers may have reduced ability to re-

P.A. Olsvik et al. / Chemosphere 79 (2010) 905–913

spond to toxicants. Thus, the combination of waterborne and the more efficient foodborne exposure with rotifers contaminated by oil once they were added to the exposure water may have enhanced the uptake of toxicants in the WSF larvae, possible depleting the defense faster and triggering new synthesis of detoxifying and protecting enzymes in these larvae. If reduced tolerance for dispersed oil and a reduction in growth is linked in this way, a marine oil spill in a cod spawning area may have devastating effects on a larvae population. Certainly, these aspects need further investigation. Biologists often pool RNA samples extracted from different subjects in order to obtain enough material for downstream experiments and for cost-efficiency (Kendziorski et al., 2003). Peng et al., (2003) showed that appropriate RNA pooling can provide power and improve efficiency for gene expression experiments with a modest increase in total number of subjects. On the other hand, pooling RNA from multiple whole-larvae may dilute tissuespecific responses. As mentioned above, GST enzymes are ubiquitously distributed and may comprise 2–4% of total cytosolic proteins in liver cells (Schlenk et al., 2008). Atlantic cod larvae at age 17 dph old have a distinct liver, although the metabolizing capacity of the liver during early development in Atlantic cod remains unknown. Using RNA extracted from pooled whole-larvae Atlantic cod homogenates we have previously observed increased mRNA expression of glutathione peroxidase isoforms and heat shock proteins in stressed individuals (unpublished results). RTqPCR can therefore be applied to assess responses in fish larvae exposed to toxicants using RNA obtained from pooled whole-larvae homogenates. Fish express CYP1A very early in embryonic development (Binder and Stegeman, 1984; Mattingly and Toscano, 2001). In killifish (Fundulus heteroclitus) embryos, CYP1A activity is detectable from 3 dph (Binder and Stegeman, 1984). Tissue-specific examinations indicate that the cardiovascular system express CYP1A in early larvae (Guiney et al., 1997). It is therefore very likely that the examined Atlantic cod larvae have a well-developed capacity to metabolize toxic compounds in the oil. Surprisingly, we were not able to detect any quantifiable levels of CYP1A transcripts in the examined cod larvae. In conclusion, only genes encoding class GST mu and GST omega showed a significant increased expression in Atlantic cod larvae exposed to WSF for 4 days. The current examination suggest that the mRNA expression of these important detoxification enzymes is only moderately inducible at early developmental stages indicating a low detoxification capacity and potentially a low tolerance of cod larvae to oil; alternatively that using whole-larvae homogenates may have masked tissue-specific mRNA induction. Acknowledgement The authors wish to thank Synnøve Winterthun and Hui-shan Tung for excellent technical help, and Elisabeth Holen, all NIFES, for project administration. We will also thank Werner Johansen for technical skills during trial, and Marte Schei and Merethe Selnes for dry weight analysis. This project was financed by the Research Council of Norway (RCN) Projects 173534/I30 and 184716/S40. References Binder, R.L., Stegeman, J.J., 1984. Microsomal electron-transport and xenobiotic monooxygenase activities during the embryonic period of development in the killifish, Fundulus heteroclitus. Toxicol. Appl. Pharmacol. 73, 432–443. Blanchette, B., Feng, X., Singh, B.R., 2007. Marine glutathione S-transferases. Mar. Biotechnol. 9, 513–542. Bouraoui, Z., Banni, M., Ghedira, J., Clerandeau, C., Guerbej, H., Narbonne, J.F., Boussetta, H., 2008. Acute effects of cadmium on liver phase I and phase II

913

enzymes and metallothionein accumulation on sea bream Sparus aurata. Fish Physiol. Biochem. 34, 201–207. Chiou, H.Y., Hsueh, Y.M., Hsieh, L.L., Hsu, L.I., Hsu, Y.H., Hsieh, F.I., Wei, M.L., Chen, H.C., Yang, H.T., Leu, L.C., Chu, T.H., Chen-Wu, C., Yang, M.H., Chen, C.J., 1997. Arsenic methylation capacity, body retention, and null genotypes of glutathione S-transferase M1 and T1 among current arsenic-exposed residents in Taiwan. Mutation Res. 386, 197–207. CTD, 2009. Comparative Toxicogenomics Database. . Doi, A.M., Pham, R.T., Hughes, E.M., Barber, D.S., Gallagher, E.P., 2004. Molecular cloning and characterization of a glutathione S-transferase from largemouth bass (Micropterus salmoides) liver that is involved in the detoxification of 4hydroxynonenal. Biochem. Pharmacol. 67, 2129–2139. Donham, R.T., Morin, D., Jewell, W.T., Lame, M.W., Segall, H.J., Tjeerdema, R.S., 2005. Characterization of cytosolic glutathione S-transferases in juvenile Chinook salmon (Oncorhynchus tshawytscha). Aquat. Toxicol. 73, 221–229. Elskus, A.A., Collier, T.K., Monosson, E., 2005. Interaction between lipids and persistent organic pollutants in fish. In: Mommsen, T.P., Moon, T.W. (Eds.), Environmental Toxicology. Elsevier, Amsterdam, pp. 152–191. GeneCards. . Gomez, A., Serra, M., Caravalho, G.R., Lunt, D.H., 2002. Speciation in ancient cryptic species complexes: evidence from the molecular phylogeny of Brachionus plicatilis (Rotifera). Evolution 56 (7), 1431–1444. Guiney, P.D., Smolowitz, R.M., Peterson, R.E., Stegeman, J.J., 1997. Correlation of 2,3,7,8-tetrachlorodibenzo-p-dioxin induction of cytochrome P4501A in vascular endothelium with toxicity in early life stages of lake trout. Toxicol. Appl. Pharmacol. 143, 256–273. Hansen, B.H., Altin, D., Vang, S.H., Nordtug, T., Olsen, A.J., 2008. Effects of naphthalene on gene transcription in Calanus finmarchicus (Crustacea: Copepoda). Aquat. Toxicol. 86, 157–165. Hansen, B.H., Nordtug, T., Altin, D., Booth, A., Hessen, K.M., Olsen, A.J., 2009. Gene expression of GST and CYP330A1 in lipid-rich and lipid-poor female Calanus finmarchicus (Copepoda: Crustacea) exposed to dispersed oil. J. Toxicol. Environ. Health – Part A 72, 131–139. Hasselberg, L., Meier, S., Svardal, A., 2004. Effects of alkylphenols on redox status in first spawning Atlantic cod (Gadus morhua). Aquat. Toxicol. 69, 95–105. Hayes, J.D., Pulford, D.J., 1995. The glutathione S-transferase supergene family: regulation of GST and the contribution of the isoenzymes to cancer chemoprotection and drug resistance. Crit. Rev. Biochem. Mol. Biol. 30, 445– 600. Hayes, J.D., Flanagan, J.U., Jowsey, I.R., 2005. Glutathione transferases. Ann. Rev. Pharmacol. Toxicol. 45, 51–88. Henson, K.L., Gallagher, E.P., 2004. Glutathione S-transferase expression in pollution-associated hepatic lesions of brown bullheads (Ameiurus nebulosus) from the Cuyahoga River, Cleveland, Ohio. Toxicol. Sci. 80, 26–33. Henson, K.L., Stauffer, G., Gallagher, E.P., 2001. Induction of glutathione Stransferase activity and protein expression in brown bullhead (Ameiurus nebulosus) liver by ethoxyquin. Toxicol. Sci. 62, 54–60. Kendziorski, C.M., Zhang, Y., Lan, H., Attie, A.D., 2003. The efficiency of pooling mRNA in microarray experiments. Biostatistics 4, 465–477. Kim, J.H., Dahms, H.U., Rhee, J.S., Lee, Y.M., Lee, J., Han, K.N., Lee, J.S., 2010. Expression profiles of seven glutathione S-transferase (GST) genes in cadmiumexposed river pufferfish (Takifugu obscurus). Comp. Biochem. Physiol. C – Toxicol. Pharmacol. 151 (1), 99–106. Konishi, T., Kato, K., Araki, T., Shiraki, K., Takagi, M., Tamaru, Y., 2005. Molecular cloning and characterization of alpha-class glutathione S-transferase genes from the hepatopancreas of red sea bream, Pagrus major. Comp. Biochem. Physiol. C – Toxicol. Pharmacol. 140, 309–320. Lee, Y.M., Seo, J.S., Jung, S.O., Kim, I.C., Lee, J.S., 2006. Molecular cloning and characterization of theta-class glutathione S-transferase (GST-T) from the hermaphroditic fish Rivulus marmoratus and biochemical comparisons with alpha-class glutathione S-transferase (GST-A). Biochem. Biophys. Res. Commun. 346, 1053–1061. Mattingly, C.J., Toscano, W.A., 2001. Posttranscriptional silencing of cytochrome P4501A1 (CYP1A1) during zebrafish (Danio rerio) development. Dev. Dynam. 222, 645–654. Olsvik, P.A., Lie, K.K., Sæle, Ø., Sanden, M., 2007. Spatial transcription of CYP1A in fish liver. BMC Physiol. 7, 12. Olsvik, P.A., Lie, K.K., Sturve, J., Hasselberg, L., Andersen, O.K., 2009. Transcriptional effects of nonylphenol, bisphenol A and PBDE-47 in liver of juvenile Atlantic cod (Gadus morhua). Chemosphere 75, 360–367. Peng, X.J., Wood, C.L., Blalock, E.M., Chen, K.C., Landfield, P.W., Stromberg, A.J., 2003. Statistical implications of pooling RNA samples for microarray experiments. BMC Bioinform. 4, 26. Schlenk, D., Celander, M., Gallagher, E.P., George, S., James, M., Kullman, S.W., Hurk, P.V.D., Willett, K., 2008. Biotransformation in fishes. In: Guiulio, R.T., Hinton, D.E. (Eds.), The Toxicology of Fishes. CRC Press, Boca Raton, pp. 153–234. Trute, M., Gallis, B., Doneanu, C., Shaffer, S., Goodlett, D., Gallagher, E., 2007. Characterization of hepatic glutathione S-transferases in coho salmon (Oncorhynchus kisutch). Aquat. Toxicol. 81, 126–136. Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A., Speleman, F., 2002. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 3 (RESEARCH0034).