Environ. Sci. Technol. 2009, 43, 6341–6348
Hepatic Transcriptomic and Metabolomic Responses in the Stickleback (Gasterosteus aculeatus) Exposed to Environmentally Relevant Concentrations of Dibenzanthracene T I M D . W I L L I A M S , * ,† H U I F E N G W U , † EDUARDA M. SANTOS,‡ JON BALL,‡ IOANNA KATSIADAKI,§ MARGARET M. BROWN,| PAUL BAKER,| FERNANDO ORTEGA,† FRANCESCO FALCIANI,⊥ JOHN A. CRAFT,| CHARLES R. TYLER,‡ JAMES K. CHIPMAN,† AND MARK R. VIANT† School of Biosciences, The University of Birmingham, Birmingham, B15 2TT, U.K., School of Biosciences, University of Exeter, Exeter, Devon, EX4 4QJ, U.K., Centre for Environment, Fisheries and Aquaculture Science, Cefas Weymouth Laboratory, Weymouth, Dorset DT4 8UB, U.K., School of Life Sciences, Glasgow Caledonian University, Glasgow, G4 0BA, U.K., and School of Immunity and Infection, The University of Birmingham, Birmingham, B15 2TT, U.K.
Received March 23, 2009. Revised manuscript received June 9, 2009. Accepted July 6, 2009.
A three-spined stickleback (Gasterosteus aculeatus) cDNA array and one-dimensional 1H nuclear magnetic resonance (NMR) spectroscopy-based metabolomics approach, together with individual biomarkers, were employed to investigate the responses of male sticklebacks to polycyclic aromatic hydrocarbon exposure. Fish were exposed to 1,2:5,6-dibenzanthracene (DbA) at concentrations between 0.01 and 50 µg per liter dissolved in the ambient water for four days, and hepatic transcript and metabolite profiles were determined in comparison with those of solvent-exposed controls. Induction of gene expression was apparent for cytochrome P450 1A (CYP1A) and CYP2family monooxygenases and these responses were strongly correlated with DbA exposure concentrations (for CYP1A r > 0.996). Expression of suites of genes related to bile acid biosynthesis, steroid metabolism, and endocrine function were also affected, as demonstrated by gene ontology analyses. Expression changes in selected genes were confirmed by realtime PCR. Metabolomics highlighted notable changes in concentrations of taurine, malonate, glutamate, and alanine. Thesestatisticallysignificantresponsestoenvironmentallyrelevant concentrations of DbA at the transcriptomic and metabolomic * Corresponding author tel: +44 121 414 3393, +44 7796 321 103; fax: +44 121 414 5925; e-mail:
[email protected]. † School of Biosciences, The University of Birmingham. ‡ School of Biosciences, University of Exeter. § Centre for Environment, Fisheries and Aquaculture Science. | School of Life Sciences, Glasgow Caledonian University. ⊥ School of Immunity and Infection, The University of Birmingham. 10.1021/es9008689 CCC: $40.75
Published on Web 07/17/2009
2009 American Chemical Society
levels provided sensitive markers characteristic of environmentally relevant low-level DbA exposure. Metabolic pathways were identified where both gene expression and metabolite concentrations were altered in response to DbA.
Introduction Dibenzanthracene (DbA) is a polycyclic aromatic hydrocarbon (PAH) known to act via the aryl-hydrocarbon receptor (1) and to induce cytochrome P450 1A (CYP1A)-dependent ethoxyresorufin-O-deethylase (EROD) activity in fish (2). It has been shown that P450s metabolically activate DbA (3) and DbA is classified as a carcinogen in mammals (4). Some PAHs are also known to affect estrogen-responsive systems (5). PAHs may be found in the water column at polluted sites (6), but they are mainly associated with sediment. In the UK, PAHs are present at concentrations up to 3.8 mg/kg (in the Mersey, Liverpool, UK) (7), and historically in estuaries at concentrations in excess of 10 mg/kg (8). “Omic” technologies are increasingly being employed to investigate responses of organisms to environmental pollutants, both to identify early warning biomarkers of exposure and effect and to elucidate the mechanisms and pathways by which such pollutants manifest their toxicities (9-11). A number of transcriptomic investigations have addressed the effects of PAHs on fish, including benzo(a)pyrene, pyrene, and beta-naphthoflavone treatments of trout (Oncorhynchus mykiss) (12-15), 3-methylcholanthrene treatment of flounder (Platichthys flesus) (16), and a pilot experiment in which stickleback were treated with DbA (17). To our knowledge there has been no previous metabolomics study of the effects of PAHs on fish, or studies integrating metabolomics and transcriptomics data in any aquatic organism. The responses of stickleback to DbA exposure were characterized using transcriptomics (cDNA microarray with 14,496 probes) (18) and metabolomics (one-dimensional 1H nuclear magnetic resonance (NMR) spectroscopy), together with a number of biomarkers and organism-level end points (plasma concentrations of vitellogenin, amounts of spiggin in the kidney, and morphometric parameters). The three-spined stickleback (Gasterosteus aculeatus) was chosen as a model study species because of its wide distribution across Northern Europe, America, and Asia in both freshwater and marine environments, wide use as a model organism to evaluate the risks to fish posed by environmental pollutants (reviewed by Katsiadaki et al. (19)), and the availability of a genome sequence (http://www.ensembl.org/Gasterosteus_aculeatus/ Info/Index).
Experimental Section Fish Treatments and Water Chemistry. All chemicals were obtained from Sigma-Aldrich (Dorset, UK) unless stated otherwise. Sticklebacks were collected from a pristine site and maintained in our laboratory as previously described (18). Males were allocated randomly into groups of 12 fish and exposed, in duplicate 40 L tanks, to sublethal doses of 0.01, 0.1, 1, 10, and 50 µg/L DbA in 0.005% (v/v) acetone solvent, solvent control, and water control in an aerated flowthrough system where the water replacement was 1 L/h. Water was sampled at day 0, 2, and 4, filtered (0.45 µm) and 0.5 L was spiked with 100 µL of Aristar grade nitric acid, preserved with dichloromethane, and assayed at Cefas Burnham Laboratory (Essex, UK), according to the method described in Kelly et al. (20). Fish were anesthetized in benzocaine and sacrificed humanely, according to UK Home Office regulations, and the wet weight and fork length were VOL. 43, NO. 16, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6341
determined. Blood was collected from the caudal sinus using heparinized capillary tubes, and the plasma was separated and used for vitellogenin analysis (n ) 18). Livers and kidneys were immediately dissected, weighed, and frozen. Additional nonexposed female sticklebacks from the same population were used to generate a reference pool of hepatic RNA. Morphometric Parameters and Biomarkers. Hepatosomatic, gonadosomatic, and nephrosomatic indices (HSI, GSI, and KSI) and condition factor were calculated as described previously (18) to evaluate the general health and reproductive status of the fish. Vitellogenin protein concentration in plasma and spiggin protein in kidney tissue were determined using ELISA methods (21, 22). Statistical significance was determined using Student’s t test with a P-value cutoff of 0.05. cDNA Microarrays. The stickleback “S2” cDNA microarray was used for gene expression profiling as previously described (18). Briefly, liver tissue from individual sticklebacks was homogenized (23) and aliquots were taken for both metabolomics and transcriptomics. Liver homogenates (ca. 10 mg wet weight) were used to prepare total RNA (Qiagen, Crawley, UK) that were reverse-transcribed to cDNA and labeled with Cy5-dCTP or Cy3-dCTP fluorophores (Amersham, Amersham, UK). Pooled or individual samples were hybridized to the microarray against a common reference pool synthesized from livers of untreated females. Data were captured using an Axon 4000B scanner and Genepix 3 software (Molecular Devices, Wokingham, UK). Two microarray experiments were performed, with samples randomized in each experiment. In the first (Individual) experiment samples from 6 individual fish per exposure were hybridized against a common reference sample using one array per individual. The treatment groups analyzed were solvent control, 0.01, 0.1, and 10 µg/L DbA. In the second (Pooled) experiment, an aliquot was taken from each individual liver homogenate (8 fish per concentration from solvent control, 0.01, 1, 10, and 50 µg/L DbA exposures) and pooled for each exposure group. Labeled cDNA was hybridized against a common reference sample, derived from nonexposed females, using one array for each pooled sample. Microarray data were submitted to ArrayExpress at EMBLEBI (http://www.ebi.ac.uk/microarray-as/ae/) using maxdLoad2 software (24). The individual experiment has been assigned the accession number E-MAXD-42, and the pooled experiment has been assigned accession number E-MAXD43. Data from the experiment performed on individual fish were quantile normalized using the affy package (http:// www.bioconductor.org) under the R programming environment and analyzed by SAM (25) with a false discovery rate (FDR) cutoff of 0.1. The pooled experiment analysis was carried out to identify genes whose expression correlated with DbA over a wider concentration range. Data from the pooled samples were analyzed using Genespring (Agilent, Stockport, UK), Lowess normalized, and filtered to remove low intensity variable spots. Relationships between the expression of genes and the doses of DbA were determined using standard correlation. Blast2GO (26, 27) was used to compare differential representation of GO terms by Fishers Exact Test using the Gossip package (28) among the DbA responsive genes. Real-Time PCR. Real-time PCR was conducted for validation of microarray data, with cDNA from the pooled stickleback tissue samples. This used an ABI7500 FAST system with Detection Software v1.3.1 (Applied Biosystems, Foster City, CA) with SYBR Mastermix (Invitrogen, Paisley, UK) and primers for the G. aculeatus genes cytochrome P450 1A, cytochrome c oxidase subunit IV isoform 2, prostaglandin D2 synthase, and hydroxymethylglutaryl CoA synthase. Three reference genes were employed (18S rRNA, beta-tubulin, and 6342
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 16, 2009
ribosomal protein L8), and were not found to significantly differ in expression over the experiment (ANOVA, P > 0.05). Primers were synthesized by Biomers.net (Ulm, Germany); sequences are shown in Table S1 in the Supporting Information. Dissociation curves for each primer pair showed single products, and PCR efficiencies were greater than 93% in all cases. The delta-delta Ct method of relative quantification (29) was employed, and the mean fold changes versus the solvent control relative to the mean of three reference genes was used for correlation analysis. 1D 1H NMR Metabolomics. For metabolomics, liver homogenate aliquots from individuals above were further extracted using methanol/chloroform/water (2:2:1.8 final volumes) (30, 23). One-dimensional 1H NMR spectroscopy was performed upon the hydrophilic fraction as previously described (18). Briefly, NMR spectra were measured at 500.11 MHz using an Avance DRX-500 spectrometer and cryogenic probe (Bruker, Coventry, UK), with 200 transients collected into 32k data points. NMR data sets were zero-filled to 64k points, exponential line-broadenings of 0.5 Hz were applied before Fourier transformation, and spectra were phase and baseline corrected, then calibrated (TMSP, 0.0 ppm) using TopSpin software (version 1.3; Bruker). The subsequent processing and statistical analyses of the NMR data have been described in detail in a previous study (18). Briefly, taurocholic acid, an abundant bile acid with highly variable concentration in the liver extracts, was subtracted from each spectrum using Chenomx NMR metabolomics software (version 4.6; Chenomx, Edmonton, Alberta). Next, residual water was removed, each spectrum was segmented into 0.005 ppm bins, and the total area of each binned spectrum was normalized to unity so as to facilitate comparison among the samples. SAM software (25) was used to find significant metabolic differences among exposure groups with appropriate false discovery rate (FDR) cutoffs. Peaks that changed significantly (at FDR < 0.01) were subsequently identified using Chenomx software and associated spectral libraries. Bins that showed significantly different peak intensities across exposures (at FDR < 0.1) were isolated, subjected to the generalized log transformation (31), meancentered, and principal component analysis (PCA) was conducted. ANOVA was used to test the significance of peak intensity changes of specific metabolites.
Results and Discussion Measured concentrations of DbA in water declined over the course of the experiment (from a mean of 95% at day 0 to 42% at day 2 and 32% at day 4) in comparison to the nominal concentrations. Many factors may have contributed to progressively falling water concentrations, and they likely included adsorption onto the tank surface or organic materials within the tanks, uptake by the fish, and/or metabolic transformation by microorganisms within the tanks. DbA exposure did not affect condition factor, GSI, KSI, or concentrations of vitellogenin serum protein or kidney spiggin protein. The HSI of fish treated with 10 µg/L DbA was statistically significantly higher than that of the solvent control group (P ) 0.0489). Gene Expression. All genes identified by SAM analysis as changing significantly in the experiment that analyzed responses in individuals are shown in Table S2; together with correlations to DbA concentration in the experiment with pooled samples and tentative identifications of unannotated ESTs (derived by comparison with the stickleback genome sequence). The transcripts also showing 1.5-fold or more change in expression in at least one of the treatments (compared to the solvent control) and annotated to known proteins are presented in Table 1. In addition, 29 further transcripts passed the above criteria but were not securely annotated ( Table S2). The PCA scores plot obtained for
TABLE 1. Identified Genes Showing Significant (SAM, FDR
