Differential gene expression as a toxicant-sensitive endpoint in zebrafish embryos and larvae

Aquatic Toxicology 81 (2007) 355–364

Differential gene expression as a toxicant-sensitive endpoint in zebrafish embryos and larvae Doris Voelker a,∗,1 , Christoph Vess a,1 , Michaela Tillmann b , Roland Nagel b , Georg W. Otto c , Robert Geisler c , Kristin Schirmer a , Stefan Scholz a a

Helmholtz Centre for Environmental Research – UFZ, Department of Cell Toxicology, Permoserstrasse 15, 04318 Leipzig, Germany b Technische Universit¨ at Dresden, Institute for Hydrobiology, Mommsenstrasse 13, 01062 Dresden, Germany c Max-Planck-Institute for Developmental Biology, Department 3 – Genetics, Spemannstrasse 35, 72076 Tuebingen, Germany Received 24 November 2006; received in revised form 12 December 2006; accepted 14 December 2006

Abstract The zebrafish (Danio rerio) embryo toxicity test (DarT) is under consideration as an alternative to the acute fish toxicity test. Microscopically visible developmental disorders or death are the endpoints used to report on toxicity in DarT. These endpoints are easily observed. They, however, rarely reveal mechanisms leading to a toxic effect and are relatively insensitive compared to chronic toxic effects. We hypothesized that, by using gene expression profiles as an additional endpoint, it may be possible to increase the sensitivity and predictive value of DarT. Therefore, as a proof of principle, we exposed zebrafish embryos to the reference compound 3,4-dichloroaniline (3,4-DCA) and analyzed gene expression patterns with a 14k oligonucleotide array. Important stress response genes not included in the microarray were additionally quantified by reverse transcriptase polymerase chain reaction. Six genes involved in biotransformation (cyp1a, ahr2), stress response (nfe212, maft, hmox1) and cell cycle control (fzr1) were significantly regulated. With the exception of fzr1, these genes proved to be differentially expressed in post hatch life stages as well. The identified genes point toward an aryl hydrocarbon receptor-mediated response. Differential gene expression in embryos exposed for 48 h was observed at 3,4-DCA concentrations as low as 0.78 ␮M, which is more than 10-fold below the concentrations that elicited visible toxic effects. Upon exposure for 5 days, differential expression was detected at concentrations as low as 0.22 ␮M of 3,4-DCA, which was close to the lowest observed effect concentration (0.11 ␮M) in the 30-day early life stage test. This study therefore indicates that gene expression analysis in DarT is able to reveal mechanistic information and may also be exploited for the development of replacement methods for chronic fish tests. © 2007 Elsevier B.V. All rights reserved. Keywords: Zebrafish; Embryo test; Early life stage test; Chronic toxicity; Microarray

1. Introduction Acute and chronic fish tests are commonly applied animal tests to investigate the toxic properties of chemicals and are performed as part of many regulations for chemicals and pesticides (Commission of the European Communities, 1996). For acute fish tests, an alternative test system using zebrafish embryos has recently been developed (Nagel, 2002). It was found that toxicity in zebrafish embryos exposed for 48 h, using survival and developmental disorders as endpoints, correlated well with the acute 48–96 h fish test. This so-called DarT (Danio rerio embryo toxicity test) is considered a pain-free in vitro test and ∗ 1

Corresponding author. Tel.: +49 341 235 2928; fax: +49 341 235 2434. E-mail address: [email protected] (D. Voelker). These authors contributed equally to the work.

0166-445X/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2006.12.013

is therefore accepted as a replacement for animal experiments. The German waste water dues law (‘Wasserabgabengesetz’) has already been revised to include the testing of zebrafish embryos as a replacement for the acute fish test. Despite the good correlation of the DarT with acute fish toxicity, the endpoints employed provide low sensitivity compared to effects detected during chronic exposure, such as analyzed in the juvenile growth test (OECD guideline 215), early life stage test (OECD guideline 210) or life cycle test (e.g. OECD guideline 416). Furthermore, it generally does not provide any information on the mode of action leading to toxic effects. These shortcomings may be overcome if gene expression is used as an endpoint. Since chemicals manifest their effects starting from molecular interactions, it can be postulated that toxicogenomic approaches are likely to identify highly sensitive, mechanism-based markers that may also have the potential to indicate long term detrimental

356

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

effects (Corvi, 2002). Such markers can be related to either adaptive processes or are the cause for toxic effects (Girardot et al., 2004). They have been shown in mammals to indicate harmful impacts of chemicals in cases where classical toxicological endpoints showed no obvious adverse effects (Heinloth et al., 2004). In order to explore whether DarT can be extended to a ‘Gene-DarT’ (gene expression Danio rerio embryo test) with the potential to elucidate mechanisms and to predict chronic effects by analysis of gene expression, we applied a 14k zebrafish oligonucleotide microarray to identify genes that were altered in zebrafish embryos exposed to 3,4-dichloroaniline (3,4-DCA). Genes that were not included in the microarray but were potential candidates for alterations by chemical exposure were additionally analyzed by quantitative RT-PCR. Those genes that showed a statistically significant differential expression in embryos were also analyzed in post hatched stages for the validation of stage specific sensitivity. 3,4-DCA was used as model compound since it is the reference compound for the standardized DarT protocol (DIN 38415-6, 2001). Its chronic toxicity has been analysed in a life cycle test and in early life stages of zebrafish (Nagel et al., 1991; Sch¨afers et al., 1993). Only little is known on the potential molecular effects of 3,4-DCA, but an induction of heat shock protein 70 (HSP70) has been observed in human cells (Ait-Aissa et al., 2000). Quantitative structure activity analysis has so far indicated that the toxicity of 3,4-DCA can be described by the polar narcosis mode of action (Arnold et al., 1990). Thus, it may resemble many other toxic chemicals that are anticipated to act as narcotics or so-called base-line toxicants (Escher and Schwarzenbach, 2002). 2. Materials and methods 2.1. Danio rerio embryo test (DarT) The zebrafish wildtype strain WiK was cultured at 26 ± 1 ◦ C at a 14:10 h light:dark cycle. Fish were fed daily, once with Artemia spec. ad libitum and twice with commercial flake food (Tetra, Melle, Germany). Collection of eggs and exposure of embryos were performed according to established protocols (Nagel, 1998; Schulte and Nagel, 1994). The test chemical 3,4-dichloroaniline (3,4-DCA, Fluka Chemie AG, Buchs, Germany; purity 99%) was added to the test medium as different volumes of a stock solution (308.5 ␮M 3,4-DCA) in water. Exposures were performed in 10 ml test medium (2 mM CaCl2 , 0.5 mM MgSO4 , 0.75 mM NaHCO3 , 0.08 mM KCl) in 70 mm glass Petri dishes. They were performed in triplicates starting on 3 different days. Chemical analysis of 3,4-DCA was performed at the onset and after 48 h of exposure using GC–MS (gas chromatography–mass spectrometry). Measured concentrations reflected nominal concentrations within a range of 97–110% for all concentrations except for the highest concentration (12.4 ␮M). At 12.4 ␮M 3,4-DCA a decrease of the concentration to 50% of the nominal concentration was observed in repeated experiments after 48 h of exposure. For the determination of toxicity, 30 embryos each were exposed to 0.78, 1.6, 3.1, 6.2 and 12.4 ␮M 3,4-DCA. Control embryos were incubated in test medium only. After 48 h of expo-

sure, lethality and developmental disorders were recorded as toxic endpoints. Lethality was identified by coagulation of the embryo, missing heart beat, the failure to develop somites, and a non-detached tail (Nagel, 2002). Sublethal effects were indicated by deformations of the embryo. Embryonic staging was performed according to Kimmel et al. (1995). For the determination of gene expression by means of microarrays, 150 48 hpf (hours post fertilization) embryos were exposed (25 embryos per dish) to 3.1 and 12.4 ␮M 3,4-DCA. The concentrations corresponded to the LOEC (lowest observed effect concentration) for survival and sublethal effects (12.4 ␮M) and a concentration of four-fold below the LOEC (3.1 ␮M) in the DarT. For confirmation of regulated genes by means of RT-PCR, 30 embryos were exposed to either 3.1, 6.2 or 12.4 ␮M of 3,4-DCA. The LOEC for survival was chosen for exposure as the highest possible concentration with minimum effects on developmental disorders. Thus, at the end of the exposure, embryos were microscopically observed and dead embryos were removed. Likewise, embryos with visible developmental abnormalities, such as shortened tail or reduced head size were excluded from the analysis to avoid ambiguities due to differing developmental stages and severity of damage. In the end, 100 intact embryos per concentration were pooled for the microarray analyses in 2 ml reaction tubes, shock frozen in liquid nitrogen and stored at −80 ◦ C for subsequent RNA isolation. RT-PCR (see below) experiments were carried out with different RNA samples using 30 embryos per treatment. In order to confirm if genes regulated by 3,4-DCA were also regulated by a well-known aryl hydrocarbon receptor agonist (Stegeman and Kloepper-Sams, 1987), one set of experiments was carried out using ␤-naphthoflavone (␤-NF, Sigma–Aldrich, Seelze, Germany). ␤-NF was added as 1000-fold concentrated stock solution in DMSO to reach final concentrations of 0.01, 0.1 and 1 ␮M. The concentration of DMSO in all treatments including controls was 0.1%. Experiments were carried out in the same way as for 3,4-DCA with subsequent RNA isolation for RT-PCR. 2.2. Early life stage test (ELST) The ELST was performed according to the OECD guideline 210 (OECD 210, 1992). Embryos (210 per concentration) were exposed to 0.11, 0.22, 0.46, 0.93 and 1.85 ␮M 3,4-DCA. The range of concentrations was selected based on previous experiments, which identified a LOEC of 1.2 ␮M (Nagel et al., 1991). This previous study also showed that mean measured concentrations in a flow-through system were within a range of 83–117% of the nominal range. This was confirmed in the current study by sampling exposure tanks of 0.46–3.1 ␮M nominal concentrations at day 14 (GC/MS analysis done by DVGW Technologiezentrum Wasser, Karlsruhe, Germany, measured concentrations 41–103% of nominal concentrations). 3,4-DCA was added as aqueous stock solution. For each of the five exposure concentrations, two replicates were carried out. Four replicates were used for the control. Exposure started at 2 hpf and was performed in a flow-through system by continuous dosing. Embryos and larvae were kept in 600 ml glass beakers for

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

14 days with a flow rate equivalent to 11 test chamber volumes per 24 h, and were then transferred to 5 l glass aquaria (flow rate equivalent to three test chamber volumes per 24 h). Temperature was adjusted to 26 ± 1 ◦ C and photoperiod set to 12:12 h light/dark throughout the entire experiment. Fish were fed from 6 dpf (days post fertilization) to 9 dpf with Paramecium ad libitum. Feeding was continued with Artemia spec. ad libitum once and commercial flake food (Tetra, Melle, Germany) twice per day. Dissolved oxygen, pH and total hardness were measured weekly; temperature was monitored daily in at least seven of the 14 vessels. At 5 and 30 dpf, survival rate, length and weight were recorded. The percentage of larvae or juvenile fish with reduced pigmentation, edema and body deformation was recorded as sublethal endpoint at days 5, 14 and day 30. At day 5 no sublethal endpoints were observed. At day 14, pigmentation and deformation were the most sensitive sublethal endpoints and resulted in a NOEC (no observed effect concentration) and LOEC (lowest observed effect concentration) of 0.11 and 0.22 ␮M, respectively. At day 30, both endpoints were already significantly affected at the lowest tested concentration of 0.11 ␮M. For RNA isolation, 65 larvae and 3 fish per replicate were shock-frozen in liquid nitrogen and stored at −80 ◦ C. RNA samples from the ELST were used, by means of RT-PCR, to verify whether genes identified as being regulated by 3,4-DCA in the DarT were regulated in later life stages of the fish as well. 2.3. RNA extraction Total RNA was extracted from 100 (for microarray analysis) or 30 (for RT-PCR) homogenized zebrafish embryos originating from DarT using Trizol Reagent (Invitrogen, Karlsruhe, Germany) according to the manufacturer’s instructions. Likewise, total RNA was extracted from a pool of 65 larvae or 3 juvenile fish originating from the ELST. Genomic DNA contamination was removed by treatment with DNAse I (Roche, Grenzach, Germany) for 15 min at 25 ◦ C. Total RNA was subsequently purified by phenol–chloroform extraction using phase lock tubes (Eppendorf, Hamburg, Germany).

357

slides were dried by centrifugation. Microarray hybridizations were performed for all replicates of samples treated with 3.1 and 12.4 ␮M 3,4-DCA. For every single experiment, dye-swaps were performed, i.e. hybridization of each experimental sample to the microarray slides was carried out twice to obtain a pair of cyanine 3/5 and cyanine 5/3 labeled cDNAs of control and 3,4-DCA treated embryos (‘dye swap’). The hybridized slides were scanned at 10 ␮m resolution using the ScanArray Express Scanner (Perkin-Elmer, USA) at 532 nm (Cy3) and 635 nm (Cy5). The laser power of the scanner was adjusted between 80 and 100% and PMT (photo multiplier value) was preset between 60 and 80, depending on the amount of background fluorescence and the saturation of signals. Microarray data were analyzed using the TM4 software package (http://www.tm4.org/, Saeed et al., 2003). TIGR Spotfinder 3.0.0 beta was used to identify spots on the array and to assess the quality of spots for downstream analyses. TIGR MIDAS 2.19 was then used to adjust the data by applying LOWESS normalization (Quackenbush, 2002). The control channel was taken as the reference for all transformations. Correlation of normalized intensities between hybridizations was weak for low intensity spots (≤0.67 and ≤0.29 for the lower 30% quantile of the 3.1 and 12.4 ␮M 3,4-DCA experiments, respectively). Therefore, these data were removed prior to subsequent analysis, which resulted in an average correlation of 0.83 (3.1 ␮M 3,4-DCA) and 0.73 (12.4 ␮M 3,4-DCA). Genes with significantly altered expression patterns were identified by one-class SAM (significance analysis of microarrays; Tusher et al., 2001). SAM is based on permutations of the data set as well as a user-defined threshold, which defines the number of expected falsely positive regulated genes. This threshold was chosen to identify significant genes with the lowest possible false discovery rate (FDR), which amounted to 12.8% for the experiments described therein. The data discussed in this publication have been deposited in NCBIs Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE6228. 2.5. Reverse transcription polymerase chain reaction (RT-PCR)

2.4. Microarray analysis Microarrays with a 14k zebrafish oligonucleotide set (MWG Biotech AG, Ebersbach, Germany) were obtained from the Leiden Genome Technology Centre (University of Leiden, NL, for details see Meijer et al., 2005). Aminoallyl-labeled cDNA was synthesized from 10 ␮g of DNAse I treated and purified total RNA as described in Yu et al. (2002). Cy3- and Cy5-labeled cDNA of control and treated embryos were combined and resuspended in 32 ␮l hybridization buffer (50% formamide, 5× SSC, 0.1% SDS and 0.1 mg/ml salmon sperm DNA). The samples were denatured by heating to 95 ◦ C for 3 min. A volume of 30 ␮l of the reaction was hybridized to the 14k zebrafish oligonucleotide slides. The slides were covered by cover slips, inserted into hybridization chambers and incubated for 14 h at 42 ◦ C in a water bath. Washing was performed in 0.1% SDS, 2× SSC for 10 min, followed by three additional washes for 10 min in 0.2× SSC, 0.1× SSC and 0.05× SSC at room temperature. The

For both conventional, as well as real time RT-PCR, cDNA was synthesized from 2 ␮g of total RNA using the RevAidTM First Strand cDNA Synthesis Kit (Fermentas, St. Leon-Rot, Germany) according to the manufacturer’s instructions. Primers (for sequences and accession numbers see supplementary information) were designed using the computer program Primer3 (freely available at http://frodo.wi.mit.edu/cgi-bin/ primer3/primer3 www.cgi). For conventional (qualitative) RT-PCR, target genes and the housekeeping genes bactin or cyclophilin A were amplified from 1 ␮l of cDNA using 1 unit of Taq Polymerase (Promega, Mannheim, Germany), 50 mM Tris–HCl, pH 9.0, 1.5 mM MgCl2 , 15 mM (NH4 )2 SO4 , 0.1% (v/v) Triton-X 100, 0.2 mM dNTP mix and 0.4 ␮M of each primer in a 50 ␮l reaction volume. The number of cycles was adjusted for each gene to result in amplification products below the saturation of the reaction. PCR-fragments were analyzed by agarose gel elec-

358

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

trophoresis and ethidium bromide staining and changes in gene expression levels were assessed by visual inspection of band densities. The image analysis software ImageJ (Version 1.33u, available free at http://rsb.info.nih.gov/ij/) was used for semiquantitative analysis of gene expression in embryos exposed to ␤-naphthoflavone. Relative expression levels were calculated by comparison with agarose gel band intensities of cyclophilin A as a housekeeping gene. Real time (quantitative) RT-PCR was carried out in a reaction volume of 25 ␮l. One reaction contained 1 ␮l of cDNA, 1× qPCR buffer (Taq PCR Core Kit, Qiagen, Hilden, Germany), 3.5 ␮M MgCl2 , 0.2 mM dNTP mix (Taq PCR Core Kit, Qiagen, Hilden, Germany), 0.02 ␮M FITC (fluorescein isothiocyanate, Bio-Rad, Munich, Germany), 0.05 ␮l SYBR green I (100× dilution in DMSO; Sigma, Steinheim, Germany), 0.5 units Taq DNA polymerase (Taq PCR Core Kit, Qiagen) and 0.1 ␮M of each primer. The amplification was carried out in 96-well plates (Peqlab, Erlangen, Germany) in the iCycler Real-Time PCR Detection System (Bio-Rad). Gene expression relative to the housekeeping gene bactin was calculated according to Muller et al. using Q-Gene software (Muller et al., 2002). The expression of a target gene was investigated by RT-PCR (real-time and semi-quantitative) from at least three independent embryo tests. Differential expression showed a similar concentration-dependent trend in each replicated experiment but the magnitude of mRNA expression levels was found to fluctuate. Therefore, a data standardisation that is common in multivariate statistics was applied (Afifi and Clark, 1996). Gene expression data from different experiments were divided by the mean that was calculated from all concentrations of each of these experiments. The resulting mean of these standardized data sets is equal to one. The advantage of this technique over other approaches – such as the representation of the data as percentage of the expression in controls – is that variation of control data is visible. 2.6. Statistics Normal distribution was confirmed by the Shapiro–WilksW-test (p < 0.05) based on the residues of log-transformed

data using the software Statistica 7.0 (StatSoft Europe GmbH, Hamburg, Germany). Statistically significant difference of logtransformed data was evaluated with parametric ANOVA and Dunnett’s-test (p < 0.05) using the software GraphPad InStat 3.0 (GraphPad software, San Diego, CA, USA). Due to correspondence of data with normal distribution of greater than 50% (R/S-test), NOEC and LOEC of the DarT and ELST were determined by the parametric Dunnett’s test (p < 0.05, ToxRat software, ToxRat Solutions GmbH, Alsdorf, Germany). 3. Results and discussion 3.1. Identification of 3,4-DCA-sensitive genes in zebrafish embryos Seven genes (five induced and two repressed) were identified by significance analysis of microarrays (SAM) upon exposures of embryos for 48 h to the lowest observed effect concentration (LOEC) of 12.4 ␮M 3,4-DCA (Table 1). The identified genes are generally known to be involved in detoxification (cyp1a, ahr2), cell proliferation (fzr1) or basic cell function (ferritin middle subunits, proteasome component c7-i, alpha-2-macroglobulin). The greatest increase in abundance due to 3,4-DCA compared to the control was recorded for cyp1a with a six-fold induction. Repression, as recorded for fzr1 and alpha-2-macrogobulin, amounted to a 30–40% decline. No significantly altered genes were identified by SAM in embryos exposed for 48 h to 3.1 ␮M 3,4-DCA (data not shown). Three of the seven genes identified from the microarrays were confirmed by quantitative RT-PCR in an independent set of samples to be differentially expressed in a concentration-dependent manner (Fig. 1A–C): cyp1a, ahr2 and fzr1 were detected to be regulated even at concentrations below those that elicit toxic effects. The genes cyp1a and ahr2 were found to be significantly induced at ≥0.78 and 6.2 ␮M 3,4-DCA, respectively (Fig. 1 A and B). The gene fzr1 showed a significantly reduced expression at ≥3.1 ␮M 3,4-DCA (Fig. 1C). The genes identified by microarray analysis and confirmed in independent samples by quantitative RT-PCR appear robust in their expression but their number seems small. At least three

Table 1 List of genes that were identified as differentially expressed by microarray analysis in zebrafish exposed for 12.4 ␮M 3,4-dichloroaniline for 48 h Gene description

NCBI accession number

SAM score

Fold change (fluorescence intensity ratios)

Induced cytochrome P450 1A (cyp1a) ferritin, middle subunita aryl hydrocarbon receptor 2 (ahr2) proteasome component c7-i ferritin, middle subunita

AF057713 BQ783379 NM 131264 BI710610 BG892155

2.01 1.78 1.57 1.33 1.30

6.37 1.94 1.68 1.55 1.53

Repressed fizzy-related protein (fzr1) alpha-2-macroglobulin

NM 131098 BI326783

−1.39 −1.39

0.68 0.62

Microarrays were performed in three replicates plus dye-swaps. Significantly differentially expressed genes were identified by SAM (significance analysis of microarrays) from co-hybridizations of control vs. treated cDNA of zebrafish embryos at a false discovery rate of 12.8%. The fold change is calculated from the fluorescence intensity values of control vs. 3,4-dichloroaniline exposed embryos. a BQ783379 and BG892155 are distinct genes with 40% nucleotide sequence identity. Both sequences show greatest homology to the human ferritin middle subunit.

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

359

Fig. 1. Differential gene expression of cyp1a (A), ahr2 (B), fzr1 (C), nfe212 (D), maft (E) and hmox1 (F) in zebrafish embryos exposed from 2 h post fertilization for 48 h to 3,4-dichloroaniline. Gene expression was analyzed with quantitative RT-PCR from at least three different experiments (see Section 2 for details). For normalization the gene for cytoplasmatic actin was used. Asterisks indicate statistically significant difference compared to control samples (p < 0.05).

proposals can be advanced to explain the low number of identified genes. Firstly, regulated genes were deduced from whole body RNA. Thus, locally differentially expressed genes may be masked by the average body expression. The ‘Gene-DarT’ (gene expression Danio rerio embryo test), however, does not allow for spatial analysis of gene expression. Secondly, differentially expressed genes were identified from embryos not exhibiting developmental disorders. This was purposely done in order to avoid the identification of stage-specific genes due to developmental delays by 3,4-DCA. Further, we were interested in genes that are regulated at concentrations prior to visible effects. As well, genes regulated due to the initial insult may no longer be responsive at the visible effect concentrations because of an expression-overwriting toxicity. Finally, the number of identified genes might be small due to a limited number of chemical target genes on the 14k microarray. The 14k oligonucleotide used here is based on a rather early stage of the zebrafish genome sequencing. It was established in October 2002. A large set of gene sequence data that contributed to the preparation of the 14k microarray was generated from ESTs (expressed sequence tags) obtained from a variety of untreated

tissues or embryonic stages. Indeed, collation of the genes represented on the oligonucleotide array indicated that potential candidate genes were not represented on the array. Among them are, for example, recently described stress response genes related to the mitogen-activated protein kinase (MAPK) or antioxidant defence pathway (Kobayashi et al., 2002; Martin et al., 2004; Motohashi and Yamamoto, 2004; Pratt et al., 2002; Takagi et al., 2004). Sequences of these genes were available from public databases. We therefore decided to screen a total of 20 potential candidate genes that were not included in the 14k microarray for differential expression in embryos exposed to 3,4-DCA by RT-PCR. A list of all investigated genes is given as supplementary information in table S1. Of these genes, three were found to be differentially expressed, which was confirmed by quantitative RT-PCR. These genes were nfe212 (NF-E2 related factor 2), maft (v-maf musculoaponeurotic fibrosarcoma oncogene homolog) and hmox1 (phase2 stress response enzyme heme oxigenase 1) (Fig. 1D–F). Primer sequences of these genes can be found in Table 2. Statistically significant induction of nfe212, maft and hmox1 was demonstrated at an exposure concentration of 6.2 ␮M 3,4-

360

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

Table 2 List of primers for significantly differentially expressed genes in zebrafish embryos exposed to 3,4-dichloroaniline and the housekeeping genes cytoplasmatic bactin and cyclophilin A Gene name or homology

NCBI accession number

Primer sequence (5 –3 )

bactin (cytoplasmatic)

AF057040

GCCAACAGAGAGAAGATGACACAGa CAGGAAGGAAGGCTGGAAGAGa TAGTCATTCCAGAAGCGTTTACC TACAGAGACACCCTGGCTTACAT

cyclophilin A

NM 212758

GACTTCACAAACCACAATGGAA CCAAAGCCCTCTACTTTCTTGA

ahr2 (aryl hydrocarbon receptor 2)

NM 131264

TACAGACTTCCAAAGAGGCACAC TGACATCCCAGATACCAGATTG

cyp1a (cytochrome P450 1A)

AF057713

CGTAATCTGCGGGATCTGTTa TTCTCATCGGACACTTGCAGa ATTCATCCTTCCTTCCCTTCAC ACCTTCTCGCCTTCCAACTTAT

fzr1 (fizzy related protein 1)

AW173921

TTTTGGAACACTCTCACAGCAC ATTGCCAGATAAAGCACCCTAT

hmox1 (heme oxygenase 1)

NM 199678

GCTTCTGCTGTGCTCTCTATACG CTCTCAGTCTCTGTGCATATCG

maft (v-maf musculoaponeurotic fibrosarcoma oncogene homologs)

AB167543

GAGGGACCAACAACATTTCAG ATTCAGACCAATCACAGCACAA

nfe212 (NF-E2 p45-related factor)

AB081314

GTTGTCCCTAGATGCAAGTCC TCTTCAGCTTGTCTTTGGTGAA

a

Primers exclusively used for qualitative RT-PCR.

DCA. Transcription of nfe212 was elevated 2.2-fold compared to the transcription of control embryos. Corresponding values for maft and hmox1 were 6.4- and 4-fold increased, respectively. At the highest exposure concentration tested (12.4 ␮M), gene expression of nfe212 and hmox1 were not statistically significant different from controls. It is possible that, as stated above, toxic effects interfere with the induction of these genes at this concentration, which was the lowest to reduce survival and induce developmental disorders of embryos (LOEC).

3.2. Regulation of 3,4-DCA sensitive genes at different life stages and comparative sensitivity The six genes identified as being differentially expressed in 48 h old embryos were found to be regulated in post-hatch life stages as well. This was assessed by qualitative RT-PCR with visual inspection of the bands (Fig. 2). Thus, an increase in transcription was observed in 5 dpf larvae (post hatch sac fry stage) for cyp1a, ahr2, nfe212, maft and hmox1 at concentra-

Fig. 2. Analysis of gene expression of cyp1a, ahr2, fzr1, nfe212, maft and hmox1 in sac fry stages (5 days post fertilization, A) and juvenile stages (30 days post fertilization, B) of zebrafish. Zebrafish embryos were exposed from 2 h post fertilization for 5 or 30 days, respectively, to different concentrations of 3,4-dichloroaniline. Gene expression was analyzed by RT-PCR. Amplification of bactin served as a control for a constitutively expressed gene.

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

tions between 0.22 and 0.93 ␮M DCA (Fig. 2A). Differential expression of these genes was also detected in 30 day old juveniles exposed to 0.46 and 0.93 ␮M 3,4-DCA (Fig. 2B). For fzr1, a slightly reduced gene expression was detected only in 30 dpf fish exposed to 0.93 ␮M 3,4-DCA (Fig. 2B). The derivation of acute and chronic toxicity along with gene expression allowed to compare the sensitivity of the endpoints with respect to 3,4-DCA concentrations. Significant gene regulation in embryos, i.e. in the DarT, was found at 3,4DCA concentrations 2–16-fold below the LOEC (12.4 ␮M). The highest sensitivity was observed for cyp1a expression (0.78 ␮M). With respect to sublethal chronic effects in an ELST (LOEC = 0.11 ␮M) with 30 day old juveniles, gene expression in the DarT was 7.1–58-fold less sensitive. In contrast, in 5 dpf post hatch sac fry stages, i.e. in a 3-day elongated DarT, differential gene expression was observed in the same concentration range (≥0.22 ␮M) as chronic toxicity in the ELST. The similar trend for gene expression changes in embryos, post-hatch sac fry and juvenile stages indicates that the same cellular and molecular signaling pathways for the response to chemicals or stress are already established in embryonic stages. The lower sensitivity of gene expression in pre-hatched embryos is intriguing but cannot yet be explained. It may be due to a lower sensitivity of these early life stages in general. However, the lower sensitivity may also be explained by the

361

bioavailability (reduced uptake due to the chorion or reduced bioaccumulation) and the duration of exposure. This may also explain the more prominent induction of cyp1a at 30 dpf if compared to 5 dpf (Fig. 2B). Furthermore, sensitivity could be biased by the gene set that was used in the analysis. Upon completion of the zebrafish genome project, the number of potential target genes that can be analyzed may increase and lead to the identification of more genes with higher sensitivity. 3.3. Mechanism of 3,4-DCA-effects deduced from regulated genes The genes that were identified by the combination of microarray and RT-PCR provide novel insights into the molecular mode of action of 3,4-DCA, which is classified as a polar narcotic. Cyp1a is a member of the cytochrome P450 isoenzyme family. Its transcription is controlled by the Ah-receptor (AHR), a nuclear receptor, which can bind polyaromatic and other substances (Bock, 1994; Denison and Heath-Pagliuso, 1998; Waller and McKinney, 1995). The receptor–ligand complex associates with a specific promoter region, so-called dioxin responsive elements (DRE), and induces transcription of downstream genes. Like cyp1a, the expression of ahr is also regulated via DREs (Safe, 1995). The stress response genes induced by 3,4-DCA in zebrafish embryos are involved in the induction of genes encoding phase

Fig. 3. Illustration of signaling pathways that might be involved in the induction of genes in zebrafish embryos exposed to 3,4-dichloroaniline: cyp1a and ahr2 transcription are enhanced via binding of 3,4-dichloroaniline to AHR2. Subsequently, the AHR2–ligand complex is supposed to act as inducer of several transcriptional activators (e.g. NFE212 and MAFT) of phase 2 biotransformation enzymes as well as antioxidant enzymes (e.g. hmox1). It is known that these enzymes are also mediated via the MAP-kinase stress response (modified from Denison and Nagy, 2003; Miao et al., 2005; Rushmore and Kong, 2002; ARNT, aryl hydrocarbon receptor nuclear translocator; HSP90, heat shock protein 90; RE, regulatory element).

362

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

2 detoxification enzymes and antioxidant proteins. The p45 NFE2 related factor2 (nfe212) is one of the major inducers of phase 2 detoxification enzymes (Motohashi and Yamamoto, 2004) and is controlled, via several transcriptional cofactors, by MAPKs (mitogen activated protein kinases, Shen et al., 2004). A transcriptional regulation of nfe212 via activation of the Ah-receptor was also demonstrated (Miao et al., 2005). Furthermore, heat shock proteins are induced via the NFE212 pathway (Kwak et al., 2003), which may explain the induction of heat shock protein hmox1 (Martin et al., 2004). By catalyzing the conversion to biliverdin and bilirubin, HMOX1 contributes to the antioxidant response of the cell elicited upon oxidative stress by reactive oxygen species. Translocation of NFE212 to the nuclei upon exposure to xenobiotics or oxidative stress leads to the dimerization with socalled small MAF proteins (Motohashi and Yamamoto, 2004), of which the induction of gene expression was also observed in this study. NFE212/MAF-protein dimers are mainly targeting the antioxidant (ARE) and electrophile response elements (EpRE) of genes encoding for different phase 2 detoxification enzymes or enzymes of antioxidant functions. They enhance their expression (Alam et al., 1999). Furthermore, the promoter region of a maf gene exhibits a response element where NFE212/MAF heterodimers are able to bind (Katsuoka et al., 2005). This indicates an auto regulatory feedback pathway, which transcriptionally regulates small maf genes. In summary, five of the six genes differentially expressed in zebrafish embryos exposed to 3,4-DCA are highly indicative of a regulatory network which is based mainly on the Ah-receptor and/or the activation of MAP-kinases as nodal points (Fig. 3). The induction of Ah-receptor-/MAP-kinaseregulated genes indicates a general defence response, which is likely to be activated by other chemicals as well. The genes cyp1a, nfe212 and hmox1 were also altered by the model AHR agonist and cyp1a-inducer ␤-naphthoflavone (Fig. 4). These results support the notion that the induction of these genes is primarily mediated via activation of the AHR pathway. The fizzy related protein 1 (fzr1) does not appear to be directly linked to the above described defence response. In contrast to cyp1a, nfe212 and hmox1, its expression was not reduced by exposure of embryos to ␤-naphthoflavone and thus did not indicate a regulation linked to the AHR (Fig. 4). FZR1 is one of the key regulators of a large multi-subunit complex, the anaphasepromoting complex or cyclosome (APC/C). This complex is catalyzing the ubiquitination of cyclin B. A controlled degradation of cell cycle regulators like cyclin B1 is necessary to maintain the sequence of the different cell cycle phases. The proteolysis of mitotic cyclins A and B is essential for the cell’s exit of mitosis (Inbal et al., 1999; Raff et al., 2002). 3,4-DCA might influence cell cycle control by repressing the transcriptional activity of FZR1. Decrease of fzr1 transcription mediated by 3,4-DCA could result in a block of mitosis, a crucial step particularly in embryonic development. Functional analysis like gene knock-down or overexpression in exposed embryos would be an important next step to unravel the potential implications of fzr1 down-regulation.

Fig. 4. Gene expression of cyp1a (A), nfe212 (B), hmox1 (C) and fzr1 (D) in zebrafish embryos exposed from 2 h post fertilization for 48 h to the Ah-receptor activator ␤-naphthoflavone. Gene expression was analyzed with RT-PCR and densitiometric gel electrophoresis from three different experiments. For normalization, the expression of the gene cyclophilin A was used. Asterisks indicate statistically significant differences compared to control samples (p < 0.05).

3.4. Perspectives for the development of a gene expression Danio rerio embryo test (‘Gene-DarT’) It has been postulated that the zebrafish embryo is a highly suitable model to indicate the impact of toxicants by gene expression profiling (Hill et al., 2005; Pichler et al., 2003). However, so far only few toxicology-related studies report the utilisation of microarrays in zebrafish embryos: Hoyt et al. (2003) used a targeted cDNA array to analyze the effect of 4-nonylphenol after exposure of 48 hpf zebrafish embryos. Nine robustly regulated genes were identified, although their expression was not confirmed in an independent set of experiments by, for example, RT-PCR. Further, Ton et al. (2003) analyzed the effects of hypoxia by a cDNA array. Another related example is the identification of genes involved in the sonic hedgehog pathway by exposure of embryos to cyclopamine (Xu et al., 2006). Thus, this is one of the first examples for a screen of gene expression changes in whole zebrafish embryos exposed to a toxic chemical using a large-scale microarray analysis. We showed that analysis

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

of gene expression was able to indicate additional mechanistic information, e.g. the involvement of Ah-receptor-regulated pathways as a likely adaptive response, which cannot be revealed by the classical endpoints in the zebrafish (Danio rerio) embryo test (DarT). Our data – shown for the reference compound 3,4dichloroaniline – indicate that analysis of gene expression in embryos exhibits a higher sensitivity (up to 16-fold) than classical toxic endpoints such as survival and developmental disorders in the embryo test. If compared to chronic toxicity, gene expression analysis in embryos was less sensitive. Sensitivity might be increased however by extension of the exposure to post-hatched embryonic stages. In order to deploy the ‘Gene-DarT’ as predictive model, it will be necessary to identify and include other sensitive genes and to extend the analysis to other compounds including different chemical structures and mode of actions. Robustly regulated genes, such as the ones identified here, could be studied by functional gene manipulation to reveal their role in toxicity. In this way, the expression of the marker genes in the Gene-DarT could be used to predict chronic effects. Acknowledgements We thank C. Petzold and P. Popp (Helmholtz Centre for Environmental Research - UFZ, Department Analytical Chemistry) for analysis of 3,4-DCA in the embryo tests. Karen Duis (ECT Floersheim, Germany) is acknowledged for critically reading the manuscript. We also thank the Genome Technology Centre of the University of Leiden for providing the spotted oligonucleotide microarrays. This study was supported by a grant of the Ministry for Education and Research BMBF (PTJ-BIO/0313016 and PTJ-BIO/0313017). G.W.O. and R.G. are supported by a European Commission 6th Framework Program grant (contract LSHG-CT-2003-503496, ZF-MODELS). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.aquatox.2006.12.013. References Afifi, A.A., Clark, V., 1996. Computer-aided Multivariate Analysis. Chapman & Hall, London. Ait-Aissa, S., Porcher, J.-M., Arrigo, A.-P., Lambre, C., 2000. Activation of the hsp70 promoter by environmental inorganic and organic chemicals: relationships with cytotoxicity and lipophilicity. Toxicology 145 (2–3), 147–157. Alam, J., Stewart, D., Touchard, C., Boinapally, S., Choi, A.M., Cook, J.L., 1999. Nrf2, a Cap’n’Collar transcription factor, regulates induction of the heme oxygenase-1 gene. J. Biol. Chem. 274 (37), 26071–26078. Arnold, L.M., Lin, D.T., Schultz, T.W., 1990. QSAR for methyl- and/or chloro-substituted anilines and the polar narcosis mechanism of toxicity. Chemosphere 21 (1–2), 183–191. Bock, K.W., 1994. Aryl hydrocarbon or dioxin receptor: biologic and toxic responses. Rev. Physiol. Biochem. Pharmacol. 125, 1–42. Commission of the European Communities, 1996. Technical guidance document in support of commission directive 93/67/EEC on risk assessment for new notified substances and commission regulation (EC) No. 1488/94 on risk assessment for existing substances. Commission of the European Communities.

363

Corvi, R., 2002. Genomics: an in vitro toxicology point of view. ATLA-Altern. Lab. Anim. 30, 129–131. Denison, M.S., Heath-Pagliuso, S., 1998. The Ah receptor: a regulator of the biochemical and toxicological actions of structurally diverse chemicals. Bull. Environ. Contamin. Toxicol. 61 (5), 557–568. Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and endogenous chemicals. Annu. Rev. Pharmacol. Toxicol. 43, 309–334. DIN 38415-6, 2001. Deutsche Einheitsverfahren zur Wasser-, Abwasser- und Schlammuntersuchung - Suborganismische Testverfahren (Gruppe T) Teil 6: Giftigkeit gegen¨uber Fischen; Bestimmung der nicht akut giftigen Wirkung von Abwasser auf die Entwicklung von Fischeiern u¨ ber Verd¨unnungsstufen (T 6). Deutsches Institut f¨ur Normung e.V., Ausgabe 2001:9. Escher, B., Schwarzenbach, P., 2002. Mechanistic studies on baseline toxicity and uncoupling of organic compounds as a basis for modeling effective membrane concentrations in aquatic organisms. Aquat. Sci., 64. Girardot, F., Monnier, V., Tricoire, H., 2004. Genome wide analysis of common and specific stress responses in adult Drosophila melanogaster. BMC Genom. 5 (1), 74. Heinloth, A.N., Irwin, R.D., Boorman, G.A., Nettesheim, P., Fannin, R.D., Sieber, S.O., Snell, M.L., Tucker, C.J., Li, L., Travlos, G.S., Vansant, G., Blackshear, P.E., Tennant, R.W., Cunningham, M.L., Paules, R.S., 2004. Gene expression profiling of rat livers reveals indicators of potential adverse effects. Toxicol. Sci. 80 (1), 193–202. Hill, A.J., Teraoka, H., Heideman, W., Peterson, R.E., 2005. Zebrafish as a model vertebrate for investigating chemical toxicity. Toxicol. Sci. 86 (1), 6–19. Hoyt, P.R., Doktycz, M.J., Beattie, K.L., Greeley Jr., M.S., 2003. DNA microarrays detect 4-nonylphenol-induced alterations in gene expression during zebrafish early development. Ecotoxicology 12 (6), 469–474. Inbal, N., Listovsky, T., Brandeis, M., 1999. The mammalian fizzy and fizzyrelated genes are regulated at the transcriptional and post-transcriptional levels. FEBS Lett. 463 (3), 350–354. Katsuoka, F., Motohashi, H., Engel, J.D., Yamamoto, M., 2005. Nrf2 transcriptionally activates the mafG gene through an antioxidant response element. J. Biol. Chem. 280 (6), 4483–4490. Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., 1995. Stages of embryonic development of the zebrafish. Dev. Dynam. 203 (3), 253–310. Kobayashi, M., Itoh, K., Suzuki, T., Osanai, H., Nishikawa, K., Katoh, Y., Takagi, Y., Yamamoto, M., 2002. Identification of the interactive interface and phylogenic conservation of the Nrf2–Keap1 system. Genes Cells 7 (8), 807–820. Kwak, M.K., Wakabayashi, N., Itoh, K., Motohashi, H., Yamamoto, M., Kensler, T.W., 2003. Modulation of gene expression by cancer chemopreventive dithiolethiones through the Keap1-Nrf2 pathway. Identification of novel gene clusters for cell survival. J. Biol. Chem. 278 (10), 8135–8145. Martin, D., Rojo, A.I., Salinas, M., Diaz, R., Gallardo, G., Alam, J., De Galarreta, C.M., Cuadrado, A., 2004. Regulation of heme oxygenase-1 expression through the phosphatidylinositol 3-kinase/Akt pathway and the Nrf2 transcription factor in response to the antioxidant phytochemical carnosol. J. Biol. Chem. 279 (10), 8919–8929. Meijer, A.H., Verbeek, F.J., Salas-Vidal, E., Corredor-Adamez, M., Bussman, J., van der Sar, A.M., Otto, G.W., Geisler, R., Spaink, H.P., 2005. Transcriptome profiling of adult zebrafish at the late stage of chronic tuberculosis due to Mycobacterium marinum infection. Mol. Immunol. 42 (10), 1185– 1203. Miao, W., Hu, L., Scrivens, P.J., Batist, G., 2005. Transcriptional regulation of NRF2 expression by the AHR-XRE signaling pathway: direct cross-talk between phase I and II drug-metabolizing enzymes. J. Biol. Chem. 280 (21), 20340–20348. Motohashi, H., Yamamoto, M., 2004. Nrf2–Keap1 defines a physiologically important stress response mechanism. Trends Mol. Med. 10 (11), 549–557. Muller, P.Y., Janovjak, H., Miserez, A.R., Dobbie, Z., 2002. Processing of gene expression data generated by quantitative real-time RT-PCR. BioTechniques 32 (6), 2–7. Nagel, R., 1998. Der vollst¨andige Life-Cycle Test mit Fischen - Verfahrensvorschlag. In: Heger, W., Jung, S., Martin, S., Raennefahrt, I.,

364

D. Voelker et al. / Aquatic Toxicology 81 (2007) 355–364

Schiecke, U., Schmitz, S., Teichmann, H., Peter, H. (Eds.), UBA - Texteband 58/98. Umweltbundesamt, Berlin. Nagel, R., 2002. DarT: the embryo test with the Zebrafish Danio rerio—a general model in ecotoxicology and toxicology. Altex 19 (Suppl. 1), 38–48. Nagel, R., Bresch, H., Caspers, N., Hansen, P.D., Markert, M., Munk, R., Scholz, N., ter Hofte, B.B., 1991. Effect of 3,4-dichloroaniline on the early life stages of the zebrafish (Brachydanio rerio): results of a comparative laboratory study. Ecotoxicol. Environ. Safe. 21 (2), 157–164. OECD 210, 1992. OECD guideline for testing of chemicals. Test no. 210: fish, early life stage toxicity test. Pichler, F.B., Laurenson, S., Williams, L.C., Dodd, A., Copp, B.R., Love, D.R., 2003. Chemical discovery and global gene expression analysis in zebrafish. Nat. Biotechnol. 21 (8), 879–883. Pratt, S.J., Drejer, A., Foott, H., Barut, B., Brownlie, A., Postlethwait, J., Kato, Y., Yamamoto, M., Zon, L.I., 2002. Isolation and characterization of zebrafish NFE2. Physiol. Genom. 11 (2), 91–98. Quackenbush, J., 2002. Microarray data normalization and transformation. Nat. Genet. 32 (Suppl.), 496–501. Raff, J.W., Jeffers, K., Huang, J.Y., 2002. The roles of Fzy/Cdc20 and Fzr/Cdh1 in regulating the destruction of cyclin B in space and time. J. Cell. Biol. 157 (7), 1139–1149. Rushmore, T.H., Kong, A.N., 2002. Pharmacogenomics, regulation and signaling pathways of phase I and II drug metabolizing enzymes. Curr. Drug Metab. 3 (5), 481–490. Saeed, A.I., Sharov, V., White, J., Li, J., Liang, W., Bhagabati, N., Braisted, J., Klapa, M., Currier, T., Thiagarajan, M., Sturn, A., Snuffin, M., Rezantsev, A., Popov, D., Ryltsov, A., Kostukovich, E., Borisovsky, I., Liu, Z., Vinsavich, A., Trush, V., Quackenbush, J., 2003. TM4: a free, open-source system for microarray data management and analysis. Biotechniques 34 (2), 374–378. Safe, S.H., 1995. Modulation of gene expression and endocrine response pathways by 2,3,7,8-tetrachlorodibenzo-p-dioxin and related compounds. Pharmacol. Ther. 67 (2), 247–281. Sch¨afers, C., Oertel, D., Nagel, R., 1993. Effects of 3,4-dichloroaniline on fish populations with differing strategies of reproduction. In: Braunbeck, W.H.T.,

Segner, H. (Eds.), Ecotoxicology and Ecophysiology. VCH, Weinheim, pp. 133–146. Schulte, C., Nagel, R., 1994. Testing acute toxicity in the embryo of zebrafish, Brachydanio rerio, as an alternative of the acute fish test: preliminary results. ATLA-Altern. Lab. Anim. 22, 12–19. Shen, G., Hebbar, V., Nair, S., Xu, C., Li, W., Lin, W., Keum, Y.S., Han, J., Gallo, M.A., Kong, A.N., 2004. Regulation of Nrf2 transactivation domain activity. The differential effects of mitogen-activated protein kinase cascades and synergistic stimulatory effect of Raf and CREB-binding protein. J. Biol. Chem. 279 (22), 23052–23060. Stegeman, J.J., Kloepper-Sams, P.J., 1987. Cytochrome P-450 isozymes and monooxygenase activity in aquatic animals. Environ. Health Perspect. 71, 87–95. Takagi, Y., Kobayashi, M., Li, L., Suzuki, T., Nishikawa, K., Yamamoto, M., 2004. MafT, a new member of the small Maf protein family in zebrafish. Biochem. Biophys. Res. Commun. 320 (1), 62–69. Ton, C., Stamatiou, D., Liew, C.C., 2003. Gene expression profile of zebrafish exposed to hypoxia during development. Physiol. Genom. 13 (2), 97–106. Tusher, V.G., Tibshirani, R., Chu, G., 2001. Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. U.S.A. 98 (9), 5116–5121. Waller, C.L., McKinney, J.D., 1995. Three-dimensional quantitative structureactivity relationships of dioxins and dioxin-like compounds: model validation and Ah receptor characterization. Chem. Res. Toxicol. 8 (6), 847–858. Xu, J., Srinivas, B.P., Tay, S.Y., Mak, A., Yu, X., Lee, S.G., Yang, H., Govindarajan, K.R., Leong, B., Bourque, G., Mathavan, S., Roy, S., 2006. Genomewide expression profiling in the zebrafish embryo identifies target genes regulated by hedgehog signaling during vertebrate development. Genetics 174 (2), 735–752. Yu, J., Othman, M.I., Farjo, R., Zareparsi, S., MacNee, S.P., Yoshida, S., Swaroop, A., 2002. Evaluation and optimization of procedures for target labeling and hybridization of cDNA microarrays. Mol. Vis. 8, 130–137.