ToxSci Advance Access originally published online on August 17, 2006
Toxicological Sciences 2006 94(1):71-82; doi:10.1093/toxsci/kfl080
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gene Expression Profiles in Fathead Minnow Exposed to 2,4-DNT: Correlation with Toxicity in Mammals
* Department of Nutritional Sciences and Toxicology, Morgan Hall and Berkeley Institute of the Environment, University of California, Berkeley, California 94720
US Army Corps of Engineer Research and Development Center, Environmental Laboratory, Vicksburg, Mississippi 39180
Division of Systems Toxicology, National Center for Toxicological Research, Jefferson, Arkansas 72079
Eon Corporation, Davis, California 95616
1To whom correspondence should be addressed. Fax: (510) 642-0535. E-mail: wintz{at}berkeley.edu.
Received August 14, 2006; accepted August 15, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Toxicogenomics, the genome-wide analysis of gene expression to study the effect of toxicants, has great potential for use in environmental toxicology. Applied to standard test organisms, it has possible applications in aquatic toxicology as a sensitive monitoring tool to detect the presence of contaminants while providing information on the mechanisms of action of these pollutants. We describe the use of a complementary DNA (cDNA) microarray of the fathead minnow (Pimephales promelas) a standard sentinel organism in aquatic toxicology, to better understand the mechanisms of toxicity of 2,4-dinitrotoluene (2,4-DNT) which is released in the environment through military and industrial use. We have constructed a fathead minnow microarray containing 5000 randomly picked anonymous cDNAs from a whole fish cDNA library. Expression profiles were analyzed in fish exposed to 2,4-DNT for 10 days at three concentrations (11, 22, and 44µM, respectively) below the measured median lethal concentration (58µM). Sequence analysis of cDNAs corresponding to differentially expressed genes affected by exposure revealed that lipid metabolism and oxygen transport genes were prominently affected in a dose-specific manner. We measured liver lipids and demonstrate that lipid metabolism is indeed perturbed following exposure. These observations correlate well with available toxicological data on 2,4-DNT. We present possible modes of action of 2,4-DNT toxicity and suggest that fathead minnow cDNA microarrays can be useful to identify mechanisms of toxicity in fish and as a predictive tool for toxicity in mammals.
Key Words: dinitrotoluene; peroxisome proliferators; methemoglobinemia; microarrays; ecotoxicogenomics.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Within the last decade, the development of complementary DNA (cDNA) microarrays has made genome-wide analysis of gene expression possible and has changed the way toxicological problems are investigated (Nuwaysir et al., 1999
). The data generated by microarray-based toxicity assays have provided valuable insight into the mechanism of toxicity of many xenobiotics and has been used extensively in drug discovery and assessment programs (Pennie et al., 2004). More recently, the use of genomics and gene expression profiling has begun to expand to the field of environmental toxicology (Lettieri, 2006). Changes in gene expression in standard sentinel organisms can potentially be used both as a monitoring tool to identify environmental pollutants in unknown samples based on signature expression profiles as well as to provide information on the mechanism of action and the safety of chemicals released in the environment (Itakura et al., 2005; Lattier et al., 2001, 2002; McClain et al., 2003; Miracle et al., 2003; Volz et al., 2005, 2006) and their potential impact on the ecosystem (Hinton et al., 2005).
While expression profiling and genomic technologies hold great potential in environmental toxicology, challenges remain before these tools will become part of the standard environmental toxicity battery of tests (Snape et al., 2004). Expression profiling on environmental samples is complicated because organisms in their natural habitat can be exposed to a variety of natural and man-made stressors. Careful analysis under controlled experimental conditions will be necessary to establish a clear correlation between particular stresses and gene expression profiles. In addition, the link between gene expression and existing toxicity assays and other outcome measures for individual organisms and populations must be established before extending these approaches to field studies (Miracle et al., 2003; Moore, 2002). In the light of the importance of linking gene expression with outcome measures for ecotoxicogenomics, we decided to utilize the fathead minnow (Pimephales promelas) for our studies. The fathead minnow is the accepted standard vertebrate aquatic toxicity organism used in acute and chronic toxicity assays. For decades, this organism has been utilized in toxicity studies to develop water quality criteria and understand the effects of pollutants on aquatic ecosystems. As a result of these ongoing efforts, well-established, rapid, and straightforward procedures now exist. These include standardized culture methods validated for exposure studies as well as accepted outcome assessment procedures. There is an extensive database of exposure outcomes, which makes the fathead minnow an ideal vertebrate model system for environmental toxicogenomics. The explosives 2,4-DNT and its isomer 2,6-dinitrotoluene are utilized in a variety of manufacturing processes (e.g., production of dyes, munitions, and gelatinizing and plasticizing agents). 2,4-DNT's production and use in military training activities has resulted in its release to the aquatic environment through various surface water pathways (Simini et al., 1995). The aquatic toxicology of the DNT compounds is poorly understood and this lack of knowledge contributes to uncertainty when assessing the potential risks involved in exposure of aquatic species to these compounds in their environment.
The purpose of this study is to begin to characterize the effects of 2,4-DNT and determine the mechanisms of toxicity in the freshwater fish species, fathead minnow, through the use of expression profiling and to assess the utility of this method in environmental toxicology.
The results presented in this study suggest that 2,4-DNT, in good correlation of known symptoms in mammals, affects oxygen transport and lipid metabolism in liver. The utility of expression profiling in environmental toxicology is discussed in view of the data presented.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Chemicals.
The 2,4-DNT (> 98% pure) was purchased from Chem Service (West Chester, PA), and acetone used to prepare the exposure stock solutions was purchased from Fisher Scientific (Fair Lawn, NJ). The high-performance liquid chromatography–grade acetonitrile and CaCl2 were purchased from J.T. Baker Chemical Co. (Phillipsburg, NJ).
Experimental organisms.
Fathead minnows (P. promelas) were obtained from Aquatic Research Organisms (Hampton, NH). Organisms were shipped overnight and acclimated to the laboratory conditions for at least 24 h prior to use to ensure the health and quality of the fish. Acclimation was conducted using dechlorinated tap water (DTW) generated by passing municipal tap water through a high-efficiency carbon filtration system.
Toxicity assessment.
The 10-day toxicity experiment was conducted using modifications, described here, of methods outlined by U.S. EPA (2002). Toxicity experiments were conducted with adult fathead minnow (age = 3–6 months, 2.75 ± 0.84 g) held for 2–3 days in 40-l glass aquaria until experiment initiation. Toxicological endpoints were survival and weight of fish at the end of the exposure period. Exposure treatments were 11, 22, 44, and 88 µmol/l (2, 4, 8, and 16 mg/l) of 2,4-DNT. Exposure solutions were prepared in acetone and spiked into DTW using a constant concentration of acetone of 1.0 ml/l. Control treatments were a DTW treatment and four acetone-only treatments (1.0 ml/l). Single exposure chambers (10 gallon aquaria) containing 25 l of exposure water and eight replicate fish were used for all treatments for a chamber loading of 0.88 g of fish/l. Exposures were conducted at 23 ± 1°C with a 16:8-h light:dark photoperiod under gold fluorescent light bulbs (
> 500 nm) to prevent photodegradation of the nitroaromatic compounds by ultraviolet (UV) radiation. Eighty percent of the exposure solutions was exchanged every 24 h by siphoning with minimum handling stress to the organisms. Fish were fed approximately 0.25 g of ground Tetramin flaked fish food purchased from Tetra Sales (Blacksburg, VA) daily before and every 48 h during the experiments prior to water renewal (U.S. EPA, 2002). Water quality parameter (pH, dissolved oxygen, and temperature) measurements and exposure solution 1-ml samples were taken daily before and after water renewals for chemical analysis. Ammonia and hardness were measured at experiment initiation and termination. All water quality parameters were within acceptable limits (pH = 6–9, dissolved oxygen 
4 mg/l, total ammonia 
5 mg/l, hardness = 60–160 mg/l as CaCO3 and temperature 23 ± 1°C) (U.S. EPA, 2002). At experiment termination, the surviving fish from each of the exposure treatments were removed from the beaker using a large dipnet, rinsed with DTW, blotted dry on absorbent paper to remove excess water and sacrificed by pithing. Liver tissues excised from each fish and remaining fish tissues were weighed into preweighed beadbeater vials and snap frozen by liquid N for holding at –80°C until microarray and chemical analysis. Median lethal concentrations (LC50) values and associated 95% confidence limits were determined using the trimmed Spearman-Karber Method using ToxCalc 5.0 (Tidepool Scientific, McKinleyville, CA).
Chemical analysis.
Samples of exposure water stored at 4°C were analyzed for 2,4-DNT using a modified version of U.S. EPA Method 8330 (1998). Chemical analysis of water samples was conducted using an Agilent 1100 high-performance liquid chromatography (HPLC) (Palo Alto, CA) equipped with a diode-array detector. The column used was a SUPELCO RP-Amide C-16 with a sample injection volume of 100 µl and flow rate of 1 ml/min. Solvent ratios were 45% water and 55% methanol, and UV absorbance was measured at 230 nm. Laboratory reporting limits for the analysis of water samples using this method is 0.1 mg/l. The whole fish carcass samples were homogenized over liquid nitrogen by mortar and pestle and a 0.2-g aliquot was extracted with 500 µl of HPLC grade acetonitrile in the presence of 100 mg of 1-mm glass beads (Biospec Products, Bartlesville, OK). Liver tissue samples were extracted with 300 µl of HPLC grade acetonitrile in the presence of 100 mg of 1-mm glass beads (Biospec Products). Both tissue sample types were homogenized on a mini-beadbeater (Biospec Products) for 3- x 60-s intervals at a speed setting of 4200 oscillations/min and immediately placed on ice to cool between intervals. Samples were sonicated for 1 h (Branson 3200, Branson Ultrasonics Corporation, Danbury, CT) at 18°C in a waterbath (Neslab RTE-111, Neslab Instruments, Inc. Newington, NH), followed by centrifugation for 10 min at 7500 x g (10,000 rpm) and 4°C. Supernatant (200 µl) was removed into a syringe (Nalge Nunc International, Rochester, NY; Norm-Ject [5 ml] Tuttlingen, Germany) with a small 0.45-µm PTFE filter attached to the syringe by luer-lock. Cold 1% CaCl2 (200 µl) was added to syringe and sample was filtered into eppendorf tubes and transferred with glass Pasteur pipette into amber HPLC vials. Samples were stored at 4°C and in dark conditions until analysis by HPLC using a modified version of U.S. EPA Method 8330 (1998) described above. The method detection limit for the acetonitrile 2,4-DNT tissue extraction technique determined by spiking untreated fathead minnow samples with 2,4-DNT is 2.46 µg/g in wet weight tissue.
RNA extraction.
Whole fish were crushed in liquid nitrogen using a mortar and pestle followed by homogenization using a tissue homogenizer in 2 ml of RNAwiz (Ambion, TX). RNAs were extracted from the homogenate using the manufacturer's protocol (http://ambion.com). Liver RNAs were extracted from 50 mg of liver tissue using the RNAeasy Mini Kit (Qiagen, Valencia, CA).
cDNA library construction and amplification.
A fish cDNA library was constructed in lambda TriplEx2 using the SMART cDNA kit (Clonetech, Mountain View, CA) using a pool of RNA extracted from unexposed fathead minnow eggs, and fish of different ages (less than 5, 5–10, 10–60, 60–150, and more than 150 days old). The library was amplified, plated, and 5000 randomly selected plaques were grown in 96-well plates to produce phage supernatant. Phage supernatant was used as template to amplify the cDNA inserts using two custom made oligonucleotides encompassing the SfiA cloning sites (SfiA: GGCCATTACGGCCGGG and SfiB: CGAGAGGCGGCCGACAT). PCR reactions were carried out in 96-well plates in a 50 µl volume using an amplification program consisting of 3 min at 94°C, followed by 30 cycles of 30 s at 94°C, 1 min at 65°C, 2 min at 72°C. All PCR reactions were cleaned by ethanol precipitation, dissolved in 50 µl of water, and printed in duplicate on poly-lysine coated glass slides.
Microarray hybridization and analysis.
cDNAs were labeled using the 3-DNA labeling kit (Array 900) from Genisphere, Inc (Hatfield, PA) and hybridized according to the manufacturer's suggested protocol (http://genisphere.com). Scanning of slides and quantification of hybridization signals were performed using an ArrayWoRx Biochip Reader (Applied Precision, Issaquah, WA) and GenePix software version 3.01 (Molecular Devices, Sunnyvale, CA), respectively. To identify genes with significant change in expression, microarray data were analyzed using an algorithm developed in-house and implemented in S-plus language (R version of the software is available in Louginov et al., 2004). Technical replicates were normalized to remove possible nonlinearity, and checked for homogeneity using box plots. To avoid between-slide normalization we applied an approach based on sequential single-slide data analysis and utilized the 
-outlier-generating model and outlier regions approach (Loguinov et al., 2004) to identify differentially expressed cDNAs. An average false positive cutoff of 1 was applied to identify candidates for differential gene expression. The software takes microarray data from single-slide experiments as an input and generates summary tables with the log2 ratio of normalized signal intensities of control versus exposed samples for each of candidate gene.
Cloning of the fathead minnow cDNA for peroxisome proliferator activator receptors alpha and gamma.
The fathead minnow peroxisome proliferator activator receptors alpha and gamma (PPAR and PPAR
) were PCR amplified from total liver cDNA using degenerate oligonucleotides and cloned into pGEM-T vector (Promega, Madison WI). Degenerate oligonucleotides, GTTCAYGCNTGYGARGGVTGYAARGG and CTRRAIKCIGTYRAICAITGICTYG-TRCG, were derived from a conserved zinc finger motif (ZnF-C4) and the ligand-binding domain (HOLI), respectively. Primers were designed based on a consensus sequences derived from the Danio rerio (XM_680825
[GenBank]
.1), mouse (NM_011144
[GenBank]
), and Fugu rubres (DQ157766
[GenBank]
) PPAR sequences. Total liver cDNA was used as a template in a PCR reaction consisting of 40 cycles of 30 s at 94°C, 1.5 min annealing at 45°C, and 2 min extension at 72°C. The fragments generated were cloned in the pDrive vector, 20 clones randomly selected and then sequenced.
Real-time PCR.
Two to five micrograms of DNAse I–treated total RNA was used to synthesize cDNA using oligo-dT and SuperscriptIII reverse transcriptase (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. cDNA was diluted to 50 µl with water and used as template for quantitative PCR (q-PCR). PCR reactions (95°C for 2 min, 40 cycles of 95°C for 15 s, 60°C for 1 min) were carried out in 96-well plates (ABI Optical Reaction Plates, Applied Biosystems, Foster City, CA), using Sybr Green PCR Master Mix (Applied Biosystems) and amplification was monitored using an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Analysis of the data was done using ABI's SDS 2.0 software. Forward and reverse primers used for q-PCR were designed using Oligoperfect Designer (invitrogen.com): 18S ribosomal RNA (rRNA) (AAACGGCTACCACATCCAAG and TTACAGGGCCTCGAAAGAGA); fatty acid–binding protein (ACCGTCACCAACTCCTTCAC and ATCGCTCCGTGTTACAGACC); 14-kDa apolipoprotein (ATTACCTTCGGCCCTACGTT and GGACAAGCAGCAGTGTTTGA); ApoAI (CCTGTTCATTCAAAACTGCCTA and CAAAAAGCAGCTTAACAAAGGA); transferrin (AAGCTGGTGACGTTGCTTT and TTTCTTTGCCCATTCTGGTC); hemoglobin alpha chain (GAGGAAGCATGGCAAGGTTA and TGTGTGCCAAGATCCTGAAG); PPAR (CGGCTCTGACCACCTCTAAG and CTGGAGGGCTGAATAAG); PPAR (CGGTCTCCACTCAGGATGAT and AACCCTTCTGCGAGATGATG); HNF-4 (ACATCATCCTGCTGGGAAAC and AGGTCTCACCAGCTCTTCCA); and acyl conenzyme A (CoA) dehydrogenase (GCAGGTGATCCCTCTGTAGC and GGATCAGTAACGCGGAACAT). A t-test was performed, using Microsoft Excel spreadsheet tools to assess the significance (p value) of the difference between samples and controls.
MS sample preparation.
Fish liver were removed from –80°C and allowed to thaw at room temperature and then transferred into a 1.5-ml polypropylene microcentrifuge tube and weighed. Samples were spiked with deuterium-labeled (H2 = 31) phospholipids at 1 µg/mg tissue. Twenty microliters of water followed by 20 µl of cold acetonitrile per milligram liver tissue were added. Samples were then centrifuged at 10,000 x g for 10 min. The water and acetonitrile were decanted and discarded. Samples were extracted with 50 µl/mg of 2:1 chloroform:methanol and centrifuged for 10 min at 10,000 x g. Ten microliters of each sample was diluted to 40 µl of methanol in a liquid chromatography/mass spectometry (LC/MS) sample vial.
HPLC–Electrospray ionization-tandem mass spectrometry.
The LC flow was provided with a capillary HPLC system from LC Packings/Dionex (Amsterdam, Netherlands) and comprised of an UltiMate quaternary pump, and a Famos autosampler. Ten-microliter injections were used. For positive ionization, the carrier solvent was 100% methanol with 0.1% formic acid. For negative ionization, the carrier solvent was 100% methanol. Electrospray ionization (ESI) was done in both positive and negative modes with a Waters Micromass Quattro Ultima triple quadrupole mass spectrometer (Manchester, U.K.). A precursor scan of 184 in the positive mode was used for the analysis of phosphatidylcholine (PC) and sphingomyelin (Beger et al., 2006; Taguchi et al., 2005). For the analysis of phosphatidylethanolamine (PE), a neutral loss of 141 in the positive mode was used. In the negative ESI mode, phosphatidylserine (PS) was analyzed by a neutral loss scan of 87. The dwell time for all analytes was set at 1 s. The collision energy was set at 29 V for all analytes. The capillary voltage was set at 3.49 kV, the cone voltage was set at 48 V, and hexapoles 1 and 2 were set at 18.1 and 0 V, respectively. The source and desolvation temperatures were 120 and 250°C. The cone gas flow rate was 114 l/h, and the desolvation gas was set at 241 l/h. Argon was used as the collision gas and set at a pressure of 2.64 mTorr. Statistica version 7.0 (Statsoft, Tulsa, OK) was used to calculate t-tests for independent samples by phospholipids levels at control and 4 and 8 mg. The t-tests were calculated using the Brown and Forsythe homogeneity of variances.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Construction of a FHM cDNA Microarray
A challenge of utilizing standard environmental toxicology organisms in genomic studies is the limited or nonexisting genomic or expression sequence information. At the time this study was initiated, very limited sequence information was available for the fathead minnow. We therefore decided to utilize anonymous cDNA microarrays as a cost-effective strategy to develop gene expression information for the fathead minnow. In this approach, we identified candidate differentially expressed genes by sequencing of a limited number of clones after hybridization studies rather than sequencing of all clones before hand. The primary fathead minnow cDNA library contained approximately 2 x 106 clones. The inserts of 192 randomly selected cDNA clones were amplified by PCR and analyzed by fractionation on agarose gel. Each clone was scored for the size of its insert and for the presence of multiple bands indicative of the presence of multiple inserts in a single vector and/or the presence of multiple clones in a single preparation. This analysis revealed that 25% of the clones amplified two or more fragments by PCR or did not yield a PCR product. The fragments size range in size between 0.2 and more than 3 kb. Five thousand clones were amplified and used to print the array without discrimination of size or insert quality. It should be emphasized that the anonymous array design, using a nonnormalized library, results in some degree of redundancy in the amplified clones and that the same transcript may be represented by multiple clones as shown in Table 1.
|
Toxicology of 2,4-DNT in Fathead Minnow
No significant mortality was observed before the concentration of 16 mg/l of 2,4-DNT at which survival was greatly affected (Fig. 1A). This allowed us to determine the LC50 to be
58µM. Measurement of liver 2,4-DNT indicated an accumulation in the liver of three to five times the concentration in the water (Fig. 1B). However, nonliver tissues (whole fish without the liver) accumulated 2,4-DNT to higher levels. The concentration factor varied between 9 and 15 times the concentration in the water. We observed significant variation between individual fish in the levels of 2,4-DNT measured in the liver and in the nonliver tissues. The standard error varied from 23 to 77% in biological samples and from 7 to 16% in the water samples, suggesting that there is variability in the way individual fish respond to exposure to 2,4-DNT.
|
Identification of 19 cDNAs Responding to Exposure to 2,4-DNT
Effects of exposure to waterborne 2,4-DNT were investigated in the fathead minnow using static renewal exposure at three different concentrations: 11.22, 22.44, and 44.88µM (2, 4, and 8 mg/l, respectively), the lowest concentration representing one-fifth of the LC50 in our experimental conditions. 2,4-DNT is known to cause liver damage in exposed animals suggesting that this organ is the primary target for 2,4-DNT toxicity (Tchounwou et al., 2003
). In addition, we have observed an accumulation of 2,4-DNT in the liver of exposed fish (Fig. 1A). Therefore, we have focused our analysis on gene expression in the liver. For each exposure, four fish were harvested, their liver dissected, and RNA extracted. Equal quantities of RNA from each liver of fish within the same exposure were pooled and used for microarray hybridizations. Each exposed RNA pool was compared to the same pool of control fish RNA in a reference hybridization design. We recognize that there is a debate in the microarray field over the use of pooling versus individual sample hybridization (Allison et al., 2006; Jolly et al., 2005; Zhang and Gant, 2005) as well as the use of reference versus loop designs (Churchill, 2002; Vinciotti et al., 2005). As our primary goal was to identify differential expression rather than determine intraindividual variation, we decided to use a pooling approach which can reduce variability between arrays and enables increase in sample size without requiring large number of arrays. This approach may have a reduced sensitivity (Allison, 2006, #76); however, a comparison of microarray hybridizations performed with a pool of RNA from two fish, revealed the same candidate genes as with the pool of four fish (data not shown). Similarly, we chose a common reference design given the limited number of comparisons and to ensure comparability of the data sets. Four replicate hybridizations that included two dye swaps were performed for each exposure. Each hybridization was analyzed independently using an in-house algorithm implemented in S-plus to identify candidate genes for regulation in response to 2,4-DNT exposure. The lists of candidate cDNAs generated for each technical replicate were compared and only the cDNAs that were identified in at least three out of the four technical replicates were considered for further analysis. We then narrowed the list further to 42 cDNAs for which both duplicated spots on the array agreed. We sequenced these clones, and the 19 different genes that could be identified by comparisons to databases using BLASTN or BLASTX are presented in Table 1. As shown, the same clone or different cDNAs representing the same gene were identified for some differentially expressed gene which is reflective of the redundancy present in the anonymous cDNA microarray but also provides additional confidence in the validity of the differential expression observed. Five of the 19 genes that we identified as differentially expressed were represented by three or more clones which if this is a representative sample would indicate about 25% redundancy and indicate about 75% unique or 2800 unique clones in the library. The expression profiles we have identified could be observed across three independent pools of fish that were treated with three different concentrations of 2,4-DNT, yet to increase the statistical significance of the results, the experiments should be replicated several times and possibly in different labs.
2,4-DNT Affects Respiration and Lipid Metabolism Genes in a Dose Dependent Manner
A significant proportion of the genes affected by 2,4-DNT are related to respiration and lipid metabolism (Fig. 2). The lipid metabolism genes identified code for apolipoproteins, namely, a 14-kDa fish-specific apolipoprotein (Kondo et al., 2001), ApoAI and ApoB (Genbank accession DQ676854), and the liver fatty acid–binding protein (L-FABP, Genbank accession DQ676856). The respiration-related genes code for the alpha chain of hemoglobin (Genbank accession DQ785103), transferrin (Tf, Genbank accession DQ676851), and cytochrome oxidase subunit I, respectively. While the hemoglobin alpha chain gene is upregulated in response to 2,4-DNT exposure, the lipid metabolism and Tf genes are all downregulated (Fig. 3). We also observed a dose-dependent response for the apolipoproteins, Tf, and hemoglobin genes with a decrease in the amplitude of the response at higher doses. Similarly, the increase in expression of hemoglobin is less important at the higher dose (Fig. 3). We also identified two mitochondrial sequences that appear to be downregulated at the 2 mg/l treatment but not at higher concentrations. This observation could be the result of decreased transcription of the mitochondrial genome or a decrease in the mitochondrial genome copy number via mitochondrial transcription factor A (Garstka et al., 1994; Larsson et al., 1998). Interestingly, downregulation of the mitochondrial genes occurs at the lower concentrations of 2,4-DNT (2 mg/l) while upregulation of hemoglobin genes occurs at higher concentrations of 2,4-DNT (4 and 8 mg/l), suggesting that a compensatory response to oxygen deprivation involving hemoglobin genes could take place at higher concentrations of 2,4-DNT.
|
|
Besides respiration or lipid metabolism genes, genes involved in fast muscle function (parvalbumin 3 [Genbank accession DQ676855] and troponin I [Genbank accession DQ785101]), in aminoacid biosynthesis/catabolism (branched-chain alpha-ketoacid dehydrogenase kinase and glutamate synthase), gene expression (7.6-kDa polypeptide of RNA polymerase II (DQ785102) and ribosomal protein L10a), apoptosis (Fas associated factor 1), and glycolysis glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were found to be affected, mostly downregulated, by exposure to 2,4-DNT. A number of cDNAs with only short stretches of similarity with known proteins (Na/Ca exchange protein, tissue inhibitor of metalloproteinase 2, PI3 kinase) were also identified, as well as unidentified cDNAs with no significant matches to known genes (Table 1).
Verification of Expression Profiles Using Real-Time q-PCR
The same RNA pools used for microarray hybridizations were used to confirm the expression pattern of selected genes (hemoglobin, Tf, 14-kDa ApoA1, and L-FABP) by real-time q-PCR. The 18S rRNA was used to normalize the amount of RNA between samples. At least two different pairs of primers were used for each gene and each sample was analyzed in triplicate. The results, which are consistent with the microarray data, including the dose dependency of the response, are presented in Figure 4.
|
2,4-DNT Downregulates PPAR
, PPAR, and Short-Branched Chain Acyl CoA Dehydrogenase cDNA but not HFN4Our findings of altered gene expression of lipid metabolism and altered liver lipid profiles led us to assess the role for PPAR
in the perturbation of lipid metabolism in response to 2,4-DNT. PPAR is a key regulator of liver lipid metabolism that acts through transcriptional activation of lipid metabolism genes such as the fatty acid oxidation genes (Peters et al., 1997). PPAR activation is transduced via the interaction between L-FABP and fatty acids or peroxisome proliferator activators such as Wy-14,643 and ciprofibrate. We cloned the fathead PPAR and PPAR cDNAs using degenerate oligonucleotides based on two conserved domains of the PPAR protein (the zinc finger ZnF-2 and a ligand-binding domain HOLI). These oligonucleotides allowed us to amplify a major DNA fragment of the expected size of 0.9 kb. Shotgun cloning of the PCR fragments led us to identify the fathead minnow cDNAs for PPAR (genbank accession DQ676847) and PPAR (DQ676853) as well as for transcription factor of the hepatocyte nuclear factor family (HNF), HNF-4 (DQ676846), and a cDNA potentially coding for an enzyme involved in short chain fatty acid oxidation, acyl CoA dehydrogenase (DQ676848). Sequence alignment of PPAR and HNF-4 (data not shown) suggests that HNF-4 possesses a ZnF-2 zinc finger similar to the one in PPAR.
Real-time PCR revealed that the expression of PPAR and acyl CoA dehydrogenase is decreased by a factor of 5 (p value < 0.0001) and 2 (p value < 0.01), respectively, in the 2,4-DNT-exposed fish. Since PPAR is responsible for regulating lipid metabolism, a decrease in expression of this magnitude is likely to affect lipid metabolism and may contribute to the observed phenotype. PPAR is only moderately downregulated (less than two-fold decrease) and HNF-4 is not affected in a significant manner (p value > 0.01) (Fig. 5).
|
Phospholipids are Increased in the Liver of 2,4-DNT Exposed Fish
In order to determine if our observation of altered lipid gene expression in the exposed fish resulted in changes in lipid metabolism, we measured phospholipid levels in the exposed and control fish livers and total lipid in nonliver tissues. Analysis of phospholipids in exposed fish livers showed a significant (p value < 0.05) increase in (18:1, 20:4) PC, PE content, and possibly PS content at 8 mg/l of 2,4-DNT with a p value of 0.12 (Fig. 6). In addition, the increase in PE appears to be dose dependent (Fig. 6). This is in agreement with our observation that livers of fish exposed to high doses of 2,4-DNT became yellow which can be an indication of high fat content. Measurements of total lipid in nonliver tissues did not show a significant change in total lipid content in these tissues suggesting that increased lipid content is confined to the liver.
|
Promoter Analysis
In order to provide further insight into the biologic response of the fish to 2,4-DNT, we assessed whether the coexpression of the genes identified in this study could be mediated via common regulatory motifs present in the promoters of the genes.
Since the sequence of the fathead minnow genome is not yet available we have used the zebra fish (D. rerio) genomic sequence to identify potential regulatory motifs that could be involved in the transcriptional responses. We used D. rerio messenger RNA (mRNA) sequences corresponding to the genes identified as affected by 2,4-DNT in the fathead minnow to search the zebra fish genome for promoter regions using the ENSEMBL genome browser (ensembl.org). One thousand five hundred nucleotides immediately upstream of the initiation codon of genes were selected and searched for known transcription factor–binding sites (Transfac database) using the search software MATCH on the Biobase website (http://www.gene-regulation.com/pub/programs.html). We did not identify PPAR-binding motifs (peroxisome proliferator response element [PPRE]) (Lawrence et al., 2001), suggesting that the differential expression is not directly mediated via PPAR. We did identify motifs for HNF, a family of transcription factors that play a pivotal role in regulating liver gene expression, in the promoters of several of these genes (Table 2). We found motifs for HNF-1, HNF-3, HNF-4, FOXD3 (HNF-3/Fkh homolog 2), and HFH-3 (HNF-3/Fkh homolog 3) upstream of the lipid metabolism and respiration-related genes (Table 2). The 14-kDa apolipoprotein gene, L-FABP, ApoAI, and hemoglobin have HNF-3 or HNF-3-related binding sites. Similarly, PPAR has potential binding sites for HNF-1 and FOX3D. A potential HNF-4-binding site was identified upstream of the Tf gene. Coregulation of these genes in response to 2,4-DNT could be mitigated via members of the HNF family of transcription factors.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
We have used cDNA microarrays of the sentinel organism fathead minnow to begin to unravel the molecular mechanisms of the toxic response to 2,4-DNT, a chemical used in the fabrication of ammunitions, explosives, and flexible polyurethane foams. 2,4-DNT represents a significant health threat since more than 13,000 pounds are released in the environment according to a 1998 survey and it has been found to contaminate ground waters (ATSDR, 1998
). The fathead minnow is a standard organism used in aquatic toxicology to assess toxicity of chemicals and water quality by monitoring endpoints such as death (acute toxicity) and growth and reproduction (chronic toxicity). In this work we use gene expression profiling to address issues in environmental toxicology.
Correlation of Expression Profiles in Fish with Known Symptoms in Mammals
The symptoms of 2,4-DNT intoxication have been well characterized in laboratory animals (Dacre et al., 1985; Tchounwou, 2003) and to some extent in humans through the study of exposed workers (Faust, 1995; Tchounwou, 2003). Methemoglobinemia, and its associated effects (anemia, reticulocytosis and increased number of Heinz bodies), is a major symptom of intoxication by 2,4-DNT (ATSDR, 1998). 2,4-DNT belongs to the nitrobenzene family of compounds that are known to cause methemoglobinemia as a primary effect (Beauchamp et al., 1982). In our study we have identified three genes that can be correlated to the effect of 2,4-DNT on oxygen transport in the blood, namely hemoglobin, Tf, and cytochrome oxidase subunit I genes. In agreement with this, we have observed that fish exposed to 2,4-DNT tend to swim close to the water surface (data not shown), which may represent a behavioral response to functional hypoxia in these fish. While nitrobenzene compounds are known to cause methemoglobinemia the mechanism by which it is caused is not known. It is likely that nitrite ions, potential byproducts of 2,4-DNT catabolism, are responsible for the oxidation of iron leading to methemoglobinemia (Huey and Beitinger, 1985).
2,4-DNT Affects Lipid Metabolism in Fish
Expression profiles suggested that 2,4-DNT affects lipid metabolism and our analysis of phospholipid content provided support for this hypothesis. Mass spectrometry analysis of phospholipids demonstrated that PC, PE, and PS are accumulating in the liver of exposed fish suggesting that downregulation of lipid metabolism could be the consequence of accumulation of phospholipids in the liver. Lipid metabolism, and in particular fatty acid oxidation, is regulated via PPAR (Peters et al., 1997) a nuclear factor that acts as a transcriptional regulator in conjunction with retinoic acid receptors (Gearing et al., 1993; Kliewer et al., 1992). PPAR is activated by fatty acids and by various chemicals known as peroxisome proliferators or hypolipidemic drugs to increase ß-oxidation of fatty acids and to downregulate genes involved in lipid biosynthesis. PPAR ligands are shuttled across the cell to the nucleus by the fatty acids–binding protein (Wolfrum et al., 2001). One possible interpretation of our observations was that 2,4-DNT, or a product of its metabolism, could act as a peroxisome proliferator activator and activate PPAR. Apoplipoprotein genes downregulated in exposed fish liver are also downregulated in mouse liver treated with the hypolipidemic drug Wy-14,643 (Hamadeh et al., 2002) and in rats treated with ciprofibrate (Yadetie et al., 2003). Our data also match the downregulation of short chain acyl CoA dehydrogenase which was shown to be reduced at the protein level by Wy-14,643 in mouse (Chu et al., 2004). However, L-FABP is upregulated in mouse treated with Wy-14.643 and in rat treated with ciprofibrate but downregulated in our experiments, suggesting that 2,4-DNT may not act by activating PPAR but rather could be affecting its expression, as evidenced by our q-PCR data. It should be noted that species-specific differences in the regulation of genes via the PPRE have been reported (Lawrence et al., 2001). In particular, human acyl CoA oxidase is not induced by Wy-14,643 while it is in rodents (Lambe et al., 1999). Hypolipidemic drugs or peroxisome proliferators are a diverse class of chemicals that induce peroxisome proliferation that is linked to lipid metabolism and to liver carcinogenesis in rats (Reddy, 2004). This correlates well with the liver carcinogenicity of 2,4-DNT in rats (Leonard et al., 1987) suggesting that, as for peroxisome proliferators, 2,4-DNT could be a nongenotoxic carcinogen in rats. The effect of 2,4-DNT on liver lipid metabolism is not specific to 2,4-DNT but is likely to be a general response to exposure to toxicants or to certain classes of toxicants. Indeed, lipid metabolism was shown to be affected in fish exposed to perfluorooctanoic acid and pefluorooctane sulfonate (Oakes et al., 2004, 2005).
Our observed effect on lipid metabolism gene expression is in agreement with reported effects of 2,4-DNT exposure in mammals, including demyelination of the peripheral nervous system (Tchounwou et al., 2003) and liver damage when exposed for prolonged period of times (Dacre et al., 1985). Downregulation of apolipoproteins can be correlated to demyelination since active expression of ApoA1 has been linked to myelination of peripheral nerves in chicken while apolipoprotein E has been associated with the response to nerve injury and axon regeneration in mammals (Gillen et al., 1995; Kim et al., 2001; LeBlanc and Poduslo, 1990; Saher et al., 2005; Vance et al., 2000). ApoAI is also a target for downregulation by peroxisome proliferators in rats (Hamadeh et al., 2002). However, we could not detect significant changes in total lipids within nonliver tissues of fish treated with 2,4-DNT (data not shown). Longer periods of exposure may be required to significantly affect whole body lipid.
The Tf gene was downregulated in response to 2,4-DNT in our experiments. Tf plays a role in iron transport in the serum and as a growth factor in a variety of cells. In addition, Tf has been shown to be part of the hypolipidemic drug pathway in mammals via PPAR which downregulates its expression by displacing HNF-4, the transcription factor controlling Tf gene expression (Hertz et al., 1996). Tf is also downregulated in rat hepatocytes treated with the PPAR agonist ciprofibrate. Our analysis of promoter elements revealed a potential HNF-4-binding element upstream of the zebra fish Tf gene suggesting that the same mechanism of regulation may exist in fish. Our results indicate that expression of HNF-4 is not affected by 2,4-DNT, therefore transcriptional regulation of this gene does not play a role in response to 2,4-DNT. However, we did observe a significant downregulation of PPAR. Downregulation of the Tf gene could also be a consequence of methemoglobinemia-induced hypoxia since hypoxia is known to decrease Tf expression in fish (Gracey et al., 2001).
Our data revealed an inverse dose-response particularly for Tf and -hemoglobin which show a decrease in the response at higher doses of 2,4-DNT. While the data at hand is not sufficient to clearly explain this response, one can postulate that at higher doses of 2,4-DNT, other compensatory mechanisms could modulate the expression of these genes.
Application of Expression Profiling to Environmental Toxicology
Expression profiling can give important clues on the mode of action of toxicants. Despite the fact that we only screened a relatively small number of cDNAs we could identify a pattern in the gene expression profiles suggesting that 2,4-DNT interferes with oxygen transport and with lipid metabolism. In our study, we utilized a pooling strategy in which the RNA of four fish was combined in equal amounts and used for the hybridization studies with multiple technical replicates. While this approach can reveal a consensus response and reduce the effects of minor variability, it is also can be sensitive to large effects from a single sample or a few samples. While individual sample hybridizations could reveal such discrepancies, this approach can require significantly more hybridizations. Although the pooling strategy may be more cost effective, it will be important to be aware of these potential confounding factors in the interpretation of expression profiling data from studies using a pooling approach. Any potential biomarker identified will require further validation to determine its robustness in application to environmental studies. Our results clearly show that expression profiling is a more sensitive indicator of adverse phenotypic effects than tests based on common biometric endpoints used in environmental toxicology such as survival. No changes in 10-day survival are noted until > 8 mg/l of 2,4-DNT. By expression profiling we noted significant changes in gene expression at 2 mg/l, which led us to suspect an effect on lipid metabolism. We demonstrated abnormal accumulation of lipids at 4 mg/l of 2,4-DNT and by 8 mg/l an abnormal fatty liver was clearly apparent by visual inspection. At these concentrations despite the clear adverse phenotypic effects, no toxicity was evident by standard survival assays. The abnormal lipid profile is indicative of compromised energy metabolism in these fish. We therefore suggest that gene expression coupled to adverse phenotypic outcomes may be more sensitive and appropriate measures of the environmental effects of toxicants. The ultimate concern of environmental toxicology is how toxicants are affecting populations and their organization by looking at how they affect survival and reproduction. We would suggest that other endpoints, such as the molecular endpoints considered in this work, may provide important monitoring tools for assessing the effects of environmental contaminants. While it is difficult to extrapolate these effects to a field situation, we would suggest that 2,4-DNT exposure at the levels which lead to abnormal lipid levels would reasonably be expected to have significant impacts on natural populations including altered energy budgets of exposed fish resulting in substandard food utilization and allocation of resources for reproduction. Our data are in good agreement with the toxicological data and known symptoms of intoxication in mammals, suggesting that fathead minnow expression data will be useful to extrapolate to mammals and humans. Similarly, this work demonstrates the utility of expression profiling to identify phenotypic biomarkers of exposure. This work provides only a first glimpse at the mechanism of toxicity of 2,4-DNT. Use of more extensive cDNA microarrays and analysis of gene expression in different organs combined with more sensitive physiological endpoints will allow us to further understand the mode of action of 2,4-DNT.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
ACKNOWLEDGMENTS |
|---|
This research was supported by a University of California discovery grant (Vulpe/lsi-bio01-10093) and by the Army Environmental Quality Program of the US Army Corps of Engineers. Permission was granted by the Chief of Engineers to publish this information. Special thanks to the Centre National de la Recherche Scientifique for its continuing support and to Christina Sasaki, Erika Wong, and Lynn Escalon for technical help. The views presented in this article do not necessarily reflect those of the U.S. Food and Drug Administration.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAREFERENCES |
|---|
Allison DB, Cui X, Page GP, Sabripour M. (2006) Microarray data analysis: from disarray to consolidation and consensus. Nat. rev. 7:55–65.
ATSDR. (1998) Toxicological Profile for 2,4- and 2,6-Dinitrotoluene(U.S. Department of Health and Human Services, Public Health Service, Atlanta, GA).
Beauchamp RO Jr,, Irons RD, Rickert DE, Couch DB, Hamm TE Jr. (1982) A critical review of the literature on nitrobenzene toxicity. Crit. Rev. Toxicol. 11:33–84.[Medline]
Beger RD, Schnackenberg LK, Li D, Dragan Y. (2006) Metabonomic models of human pancreatic cancer using 1D proton NMR spectra of organic plasma extracts. Metabolomics (in press).
Chu R, Lim H, Brumfield L, Liu H, Herring C, Ulintz P, Reddy JK, Davison M. (2004) Protein profiling of mouse livers with peroxisome proliferator-activated receptor alpha activation. Mol. Cell Biol. 24:6288–6297.
Churchill GA. (2002) Fundamentals of experimental design for cDNA microarrays. Nature genetics 32:Suppl., 490–495.
Dacre JC, Ellis H, Hong C, Lee C, Glennon J. (1985) Subchronic and chronic toxicity studies of 2,4-dinitrotoluene. Part I. Beagle dogs. J. Am. Coll. Toxicol. 4:233–242.
Faust RA. (1995) Toxicity Summary for 2,6-Dinitrotoluene(Oak Ridge National Laboratory, Oak Ridge, TN) pp. 1–11.
Garstka HL, Facke M, Escribano JR, Wiesner RJ. (1994) Stoichiometry of mitochondrial transcripts and regulation of gene expression by mitochondrial transcription factor A. Biochem. Biophys. Res. Commun. 200:619–626.[CrossRef][ISI][Medline]
Gearing KL, Gottlicher M, Teboul M, Widmark E, Gustafsson JA. (1993) Interaction of the peroxisome-proliferator-activated receptor and retinoid X receptor. Proc. Natl. Acad. Sci. U.S.A. 90:1440–1444.
Gillen C, Gleichmann M, Spreyer P, Muller HW. (1995) Differentially expressed genes after peripheral nerve injury. J. Neurosci. Res. 42:159–171.[CrossRef][ISI][Medline]
Gracey AY, Troll JV, Somero GN. (2001) Hypoxia-induced gene expression profiling in the euryoxic fish Gillichthys mirabilis. Proc. Natl. Acad. Sci. U.S.A. 98:1993–1998.
Hamadeh HK, Bushel PR, Jayadev S, Martin K, DiSorbo O, Sieber S, Bennett L, Tennant R, Stoll R, Barrett JC, et al. (2002) Gene expression analysis reveals chemical-specific profiles. Toxicol. Sci. 67:219–231.
Hertz R, Seckbach M, Zakin MM, Bar-Tana J. (1996) Transcriptional suppression of the transferrin gene by hypolipidemic peroxisome proliferators. J. Biol. Chem. 271:218–224.
Hinton D, Kullman S, Hardman R, Volz D, Chen P, Carney M, Bencic D. (2005) Resolving mechanisms of toxicity while pursuing ecotoxicological relevance? . Mar. Pollut. Bull. 51:635–648.[CrossRef][ISI][Medline]
Huey DW and Beitinger TL. (1985) Hematological responses of larval Rana catesbiana to sublethal nitrite exposures. Bull. Environ. Contam. Toxicol. 25:574–577.[CrossRef]
Itakura T, Ogino Y, Mahata SC, El-Kady MA, Aoki JY, Kato H, Kaminishi Y. (2005) Estrogen-responsive element (ERE)-like motifs affect the 3-methylcholanthrene induction of eel CYP1A gene. Environ. Sci. 12:65–70.[Medline]
Jolly RA, Goldstein KM, Wei T, Gao H, Chen P, Huang S, Colet JM, Ryan TP, Thomas CE, Estrem ST. (2005) Pooling samples within microarray studies: a comparative analysis of rat liver transcription response to prototypical toxicants. Physiological genomics 22:346–55.
Kim DS, Lee SJ, Park SY, Yoo HJ, Kim SH, Kim KJ, Cho HJ. (2001) Differentially expressed genes in rat dorsal root ganglia following peripheral nerve injury. Neuroreport 12:3401–3405.[CrossRef][ISI][Medline]
Kliewer SA, Umesono K, Noonan DJ, Heyman RA, Evans RM. (1992) Convergence of 9-cis retinoic acid and peroxisome proliferator signalling pathways through heterodimer formation of their receptors. Nature 358:771–774.[CrossRef][Medline]
Kondo H, Kawazoe I, Nakaya M, Kikuchi K, Aida K, Watabe S. (2001) The novel sequences of major plasma apolipoproteins in the eel Anguilla japonica. Biochim. Biophys. Acta 1531:132–142.[Medline]
Lambe KG, Woodyatt NJ, Macdonald N, Chevalier S, Roberts RA. (1999) Species differences in sequence and activity of the peroxisome proliferator response element (PPRE) within the acyl CoA oxidase gene promoter. Toxicol. Lett. 110:119–127.[CrossRef][ISI][Medline]
Larsson NG, Wang J, Wilhelmsson H, Oldfors A, Rustin P, Lewandoski M, Barsh GS, Clayton DA. (1998) Mitochondrial transcription factor A is necessary for mtDNA maintenance and embryogenesis in mice. Nat. Genet. 18:231–236.[CrossRef][ISI][Medline]
Lattier DL, Gordon DA, Burks DJ, Toth GP. (2001) Vitellogenin gene transcription: A relative quantitative exposure indicator of environmental estrogens. Environ. Toxicol. Chem. 20:1979–1985.[CrossRef][ISI][Medline]
Lattier DL, Reddy TV, Gordon DA, Lazorchak JM, Smith ME, Williams DE, Wiechman B, Flick RW, Miracle AL, Toth GP. (2002) 17alpha-Ethynylestradiol-induced vitellogenin gene transcription quantified in livers of adult males, larvae, and gills of fathead minnows (Pimephales promelas). Environ. Toxicol. Chem. 21:2385–2393.[CrossRef][ISI][Medline]
Lawrence JW, Li Y, Chen S, DeLuca JG, Berger JP, Umbenhauer DR, Moller DE, Zhou G. (2001) Differential gene regulation in human versus rodent hepatocytes by peroxisome proliferator-activated receptor (PPAR) alpha. PPAR alpha fails to induce peroxisome proliferation-associated genes in human cells independently of the level of receptor expression. J. Biol. Chem. 276:31521–31527.
LeBlanc AC and Poduslo JF. (1990) Regulation of apolipoprotein E gene expression after injury of the rat sciatic nerve. J. Neurosci. Res. 25:162–171.[CrossRef][ISI][Medline]
Leonard TB, Graichen ME, Popp JA. (1987) Dinitrotoluene isomer-specific hepatocarcinogenesis in F344 rats. J. Natl. Cancer Inst. 79:1313–1319.[ISI][Medline]
Lettieri T. (2006) Recent applications of DNA microarray technology to toxicology and ecotoxicology. Environ. Health Perspect. 114:4–9.[ISI][Medline]
Loguinov AV, Mian SI, Vulpe CD. (2004) Exploratory differential gene expression analysis in microarray experiments with no or limited replication. Genome Biol. 5:R18.[CrossRef][Medline]
McClain JS, Oris JT, Burton GA Jr, Lattier D. (2003) Laboratory and field validation of multiple molecular biomarkers of contaminant exposure in rainbow trout (Oncorhynchus mykiss). Environ. Toxicol. Chem. 22:361–370.[CrossRef][ISI][Medline]
Miracle AL, Toth GP, Lattier DL. (2003) The path from molecular indicators of exposure to describing dynamic biological systems in an aquatic organism: Microarrays and the fathead minnow. Ecotoxicology 12:457–462.[CrossRef][ISI][Medline]
Moore MN. (2002) Biocomplexity: the postgenome challenge in ecotoxicology. Aquat. Toxicol. 59:1–15.[CrossRef][ISI][Medline]
Nuwaysir EF, Bittner M, Trent J, Barrett JC, Afshari CA. (1999) Microarrays and toxicology: The advent of toxicogenomics. Mol. Carcinog. 24:153–159.[CrossRef][ISI][Medline]
Oakes KD, Sibley PK, Martin JW, MacLean DD, Solomon KR, Mabury SA, Van Der Kraak GJ. (2005) Short-term exposures of fish to perfluorooctane sulfonate: Acute effects on fatty acyl-CoA oxidase activity, oxidative stress, and circulating sex steroids. Environ. Toxicol. Chem. SETAC 24:1172–1181.[CrossRef]
Oakes KD, Sibley PK, Solomon KR, Mabury SA, Van der Kraak GJ. (2004) Impact of perfluorooctanoic acid on fathead minnow (Pimephales promelas) fatty acyl-CoA oxidase activity, circulating steroids, and reproduction in outdoor microcosms. Environ. Toxicol. Chem. SETAC 23:1912–1919.[CrossRef]
Pennie W, Pettit SD, Lord PG. (2004) Toxicogenomics in risk assessment: An overview of an HESI collaborative research program. Environ. Health Perspect. 112:417–419.[ISI][Medline]
Peters JM, Hennuyer N, Staels B, Fruchart J-C, Fievet C, Gonzalez FJ, Auwerx J. (1997) Alterations in lipoprotein metabolism in peroxisome proliferator-activated receptor alpha-deficient mice. J. Biol. Chem. 272:27307–27312.
Reddy JK. (2004) Peroxisome proliferators and peroxisome proliferator-activated receptor alpha: Biotic and xenobiotic sensing. Am. J. Pathol. 164:2305–2321.
Saher G, Brugger B, Lappe-Siefke C, Mobius W, Tozawa R, Wehr MC, Wieland F, Ishibashi S, Nave KA. (2005) High cholesterol level is essential for myelin membrane growth. Nat. Neurosci. 8:468–475.[ISI][Medline]
Simini M, Wentsel RS, Checkai RT, Phillips CT, Chester NA, Major MA, Amos JC. (1995) Evaluation of soil toxicity at Joliet Army Ammunition Plant. Environ. Toxicol. Chem. 14:623–630.
Snape JR, Maund SJ, Pickford DB, Hutchinson TH. (2004) Ecotoxicogenomics: the challenge of integrating genomics into aquatic and terrestrial ecotoxicology. Aquat. Toxicol. 67:143–154.