Taking Microarrays to the Field: Differential Hepatic ... - ACS Publications

Dec 12, 2011 - proved effective in the detection of polychlorinated biphenyls,1 organotins,2 and dieldrin,3 while polar organic chemical integrative ...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/est

Taking Microarrays to the Field: Differential Hepatic Gene Expression of Caged Fathead Minnows from Nebraska Watersheds Marlo K. Sellin Jeffries,*,†,⊥,# Alvine C. Mehinto,*,‡,⊥ Barbara J. Carter,§,∇ Nancy D. Denslow,‡ and Alan S. Kolok†,∥ †

Department of Environmental, Agricultural and Occupational Health, University of Nebraska - Medical Center, Omaha, Nebraska 68198, United States ‡ Department of Physiological Sciences and Center for Environmental and Human Toxicology, Gainesville, Florida 32611-0885, United States § EcoArray, Inc., Gainesville, Florida 32608, United States ∥ Department of Biology, University of Nebraska at Omaha, Omaha, Nebraska 68182-0040, United States S Supporting Information *

ABSTRACT: This study aimed to evaluate the utility of microarrays as a biomonitoring tool in field studies. A 15,000-oligonucleotide microarray was used to measure the hepatic gene expression of fathead minnows (Pimephales promelas) caged in four Nebraska, USA watersheds - the Niobrara and Dismal Rivers (low-impact agricultural sites) and the Platte and Elkhorn Rivers (highimpact agricultural sites). Gene expression profiles were site specific and fish from the low- and high-impact sites aggregated into distinct groups. Over 1500 genes were differentially regulated between fish from the low- and high-impact sites. Many gene expression differences (1218) were also noted when the Platte and Elkhorn minnows were compared to one another and Platte fish experienced a higher degree of transcript alterations than Elkhorn fish. These findings indicate that there are differences between the low-impact and highimpact sites, as well as between the two high-impact sites. Historical water quality data support these results as only trace levels of agrichemicals have been detected at the low-impact sites, while substantial levels of agrichemicals have been reported at the high-impact sites with agrichemical loads at the Platte generally exceeding those at the Elkhorn. Overall, this study demonstrates that microarrays can be utilized to discriminate sites with different contaminant loads from one another.



INTRODUCTION As concern regarding the presence of emerging contaminants has grown, several biomonitoring tools have been developed to assess both their presence and biological effects in field settings. For example, semipermeable membrane devices (SPMDs) have proved effective in the detection of polychlorinated biphenyls,1 organotins,2 and dieldrin,3 while polar organic chemical integrative samplers (POCISs) have been successfully used to detect organic compounds such as pharmaceuticals,4 steroid hormones,5 and certain pesticides.6 Molecular techniques such as enzyme-linked immunosorbent assay and quantitative realtime polymerase chain reaction (qPCR) have been employed to evaluate the effects of chemical exposures on targeted proteins and genes and to identify potential biomarkers of exposure.7−9 While these techniques have been valuable in assessing the occurrence and biological effects of contaminants in natural environments, they are often limited by their specificity for certain classes of contaminants. For example, vitellogenin induction in male fish has been established as a © 2011 American Chemical Society

reliable biomarker of environmental exposure to estrogenic compounds.10−12 However, it does not necessarily have predictive value for other classes of contaminants. Because surface waters generally contain complex mixtures of contaminants, more comprehensive biomonitoring tools capable of not only assessing the occurrence of contaminants in the environment but also linking the presence of those contaminants to biological outcomes are needed. Transcriptomics technologies, such as microarrays, have been identified as tools that may be particularly useful in fulfilling these needs.13 Microarrays have been used to determine the gene expression profiles of fish exposed to environmental pollutants including methylmercury,14 pesticides,15,16 and endocrine-disrupting compounds (EDCs).17−19 The objective Received: Revised: Accepted: Published: 1877

November 2, 2011 December 7, 2011 December 12, 2011 December 12, 2011 dx.doi.org/10.1021/es2039097 | Environ. Sci. Technol. 2012, 46, 1877−1885

Environmental Science & Technology

Article

Table 1. General Characteristics and Contaminant Loads of the Four Selected Field Sites watershed area (km2)a stream flow (m3/s)b specific conductance (μs/cm)b dissolved oxygen (mg/L)b suspended sediment (mg/L)b temperature (°C)c pHc Nutrients in water (mg/L)b ammonia nitrate phosphorus chloride Pesticides in water (μg/L)b acetochlor alachlor atrazine simazine metolachlor Pesticides in POCIS (ng)c acetochlor alachlor atrazine deethylatrazine deisopropylatrazine dimethenamid propazine simazine metolachlor

Dismal

Niobrara

Elkhorn

Platte

2,500 6.40 18.2 (2.7) 8.0 (0.2)

34,913 24.4 218 8.9 44.0 20.5 (3.5) 8.2 (0.3)

18,135 64.0 613 8.5 518.8 21.8 (2.0) 8.3 (0.2)

222,740 379 606 8.9 936.9 24.7 (4.9) 8.5 (0.3)

-

0.024 0.37 0.15 -

0.040 0.77 0.65 10.0

0.079 1.50 0.84 39.3

-

-

0.28 0.011 1.20 0.013 0.38

0.70 0.066 2.75 0.018 1.25

ND ND 11 (1) ND ND ND ND ND ND

ND ND 13 (1) ND ND ND ND ND ND

98 (29) ND 967 (218) 200 (42) 105 (22) ND 19 (3) ND 379 (51)

194 (32) 12 (1) 1983 (339) 356 (68) 184 (32) 17 (3) 33 (5) 11 (2) 775 (127)

a

Watershed areas in the Dismal, Niobrara, Elkhorn, and Platte were reported by Mast and Turk,21 Soenksen et al.,22 Kolok et al.,20 and Eschner et al.,23 respectively. bData collected by the United States Geological Survey (USGS)24 and obtained at http://waterdata.usgs.gov/ne/nwis/qwdata?. Reported values reflect averages of data collected during May and June from 2005 to 2009, except for stream flows which represent averages in June from 2005 to 2009. cData reported in Sellin et al.;6average (±standard error). Temperatures were obtained every 90 s by loggers placed alongside minnows from the current study. pH was determined on the first and last deployment days. Pesticide data reported reflect amounts of pesticides in polar organic chemical integrative samplers (POCIS) deployed alongside minnows from this study, ND = below detectable limits (10 ng for pesticides).

making both watersheds ideal reference sites. Descriptions of these sites, as well as water quality data, are shown in Table 1. Field Study. Adult fathead minnows were caged at each of the aforementioned sites during the summer of 2007 as described in Sellin et al.6 At each site, 36 sexually mature fish (21 males and 15 females) were divided equally into three cages and deployed for seven days. Three polar organic chemical integrative samplers (POCIS, Environmental Sampling Technologies, Inc., St. Joseph, MO) and a temperature logger (Onset, Bourne, MA) were also placed at each site.6 The 7-day deployments were initiated within a three-day period, from June 4−6, with retrievals occurring from June 11−13. Following the 7-day deployments, fish were euthanized with tricaine methanesulfonate (MS-222) and body mass was determined. Livers and gonads were dissected from each fish, weighed, and flash frozen in liquid nitrogen. Hepatosomatic (HSI) and gonadosomatic (GSI) indices were calculated by dividing the mass of the tissues into the body mass of the fish and then multiplying by 100. Female livers and a subset of male livers were utilized for targeted gene expression analysis via qPCR and the results of that analysis are reported in Sellin et al.6 Male livers used in this study were sent to EcoArray, Inc. (Gainesville, FL) where they were stored at −80 °C until further analysis.

of this study was to assess the utility of microarray analysis as a biomonitoring tool to discriminate field sites with different levels or types of contaminants from one another. To accomplish this, the hepatic gene expression profiles of fathead minnows (Pimephales promelas) caged at four sites in Nebraska were examined. Two of the selected sites are located in lowimpact agricultural watersheds, while the other sites are located in high-impact agricultural watersheds. It was hypothesized that differences in the gene expression profiles of caged minnows would correlate with differences in water quality, specifically contaminant load, between the sites.



MATERIALS AND METHODS Site Descriptions. This study was carried out at four sites in Nebraska - the Elkhorn, Platte, Dismal, and Niobrara Rivers (Supporting Information). The Elkhorn and Platte Rivers drain watersheds characterized by high densities of row crop agriculture (primarily corn and soybean rotation) and beef cattle feedlots.20 As such, surface waters of these watersheds are expected to contain agricultural pesticides and veterinary pharmaceuticals. The Dismal and Niobrara river watersheds are located in the Sandhills region of Nebraska where agricultural activity is limited to low-density cattle grazing 1878

dx.doi.org/10.1021/es2039097 | Environ. Sci. Technol. 2012, 46, 1877−1885

Environmental Science & Technology

Article

Statistical Analysis. Differences in body mass and organ indices among groups were determined by one-way ANOVA (StatView 5.0, SAS Institute) followed by Newman-Kuels multiple comparison tests (p < 0.05). Microarray Analysis. Gene expression profiling was conducted at EcoArray, Inc. (Gainesville, FL). Total RNA was isolated from male livers (n = 3−5 per site) using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA) according to manufacturer’s protocol and quantified on a NanoDrop ND1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). RNA integrity was determined using a BioAnalyzer 2100 (Agilent, Palo Alto, CA). The average RNA integrity number (RIN) was 7.6. Fathead minnow 15,000-oligonucleotide arrays were designed by EcoArray and manufactured by Agilent (Palo Alto, CA). A one-color labeling method was utilized, and microarray hybridizations were performed according to the manufacturer’s instructions. Briefly, RNA samples were used for the synthesis/ amplification of cyanine 3 (Cy3) labeled cRNA using Agilent’s Low RNA Input Fluorescent Linear Amplification Kit. The resulting cRNA products were purified with RNeasy purification columns (Qiagen). The amplification yield and Cy3 incorporation were evaluated using the NanoDrop spectrophotometer, and samples with a specific activity >8 pmol Cy3/μL were used on the arrays. Fragmentation of the cRNA was performed using Agilent’s Gene Expression Hybridization Kit. A final volume of 50 μL per sample was hybridized to the microarrays for 17 h at 65 °C. The slides were washed and scanned using an Agilent DNA microarray scanner, and raw images were processed using Agilent Feature Extraction Software v9.5. Microarray data from this study were deposited into the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE33436). Bioinformatics. Microarray data were imported into JMP Genomics v4.1. Raw intensity data were subjected to a log2 transformation followed by Loess normalization. Gene probes differentially regulated between the sites were identified by oneway ANOVA (p < 0.05; fold change >1.5). Functional enrichment analysis was performed using the Gene Ontology (GO) annotations provided by EcoArray Inc. Differentially regulated genes that were over-represented in the biological process GO category were determined by Fisher Exact Test (p < 0.05, no post hoc correction). The most significant genes (p < 0.01; fold change >1.5) were subjected to a hierarchical clustering analysis using Cluster 3.025 and visualized in Java TreeView.26

Figure 1. Comparison of hepatic gene regulation using Dismal as a control site. Significant genes (ANOVA, p < 0.05) were arranged based on the fold expression for the Niobrara vs Dismal site and kept in that order for the other site comparisons. In cases where gene expression changes were not significant at a given site, they were graphed as no change.

(GO:0001558) were over-represented in Niobrara fish relative to Dismal fish. High- versus Low-Impact Sites. Over 1500 transcripts were differentially expressed between fish from the low-impact and high-impact sites (Figure 2, Supporting Information). A subset of these altered transcripts is shown in Table 2. The largest changes in gene expression were observed between fish from the Platte and Dismal Rivers with fish deployed at the Platte exhibiting an up-regulation of 891 transcripts and a downregulation of 861 transcripts relative to Dismal fish. The main biological processes altered in the Platte fish relative to those caged at the low-impact sites included DNA repair (GO:0006281), response to DNA damage stimulus (GO:0006974), and ion transport (GO:0006811) (Table 3). In relation to minnows from the low-impact sites, those deployed at the Elkhorn River experienced alterations in the expression of several genes related to lipid, cholesterol, and steroid biosynthesis (e.g., lanosterol synthase, mevalonate decarboxylase, and 7-dehydrocholestrol reductase), and the majority of these altered transcripts were down-regulated. The main biological processes disrupted in Elkhorn River minnows



RESULTS Body Mass and Organ Indices. The average body mass, HSI, and GSI of the fish retrieved from each site ranged from 1.3 to 1.5 g, 1.4 to 1.7, and 1.0 to 1.8, respectively. There were no significant differences in body mass or organ indices between the groups (ANOVA, p > 0.26 in each case). Differential Hepatic Gene Expression of Male Minnows. Low-Impact Sites. A comparison of gene expression profiles between minnows deployed at the low-impact sites revealed that 529 transcripts (269 increased and 260 reduced) were differentially expressed in the fish caged at the Niobrara relative to the Dismal (Figure 1, see Supporting Information for a full list of altered transcripts). Genes involved in protein biosynthesis (GO:0006412) and regulation of cell growth 1879

dx.doi.org/10.1021/es2039097 | Environ. Sci. Technol. 2012, 46, 1877−1885

Environmental Science & Technology

Article

minnows from the low-impact sites. There were 620 and 584 transcripts commonly altered in fish from both high-impact sites relative to the Dismal and Niobrara sites, respectively (Figure 2, Supporting Information). The expression of a number of genes involved in DNA repair and response to DNA damage stimulus (e.g., MutS homologue 2, apex nuclease and ligase I) were down-regulated in minnows caged at both the Platte and Elkhorn Rivers (Supporting Information). Fish caged in these two watersheds also exhibited differential expression of several genes encoding cytochrome P450 (CYP) enzymes. Specifically, the expression of CYP2C19 and CYP51A1 among Elkhorn fish and CYP1A1 among Platte fish was significantly down-regulated relative to those deployed at the low-impact sites. In addition, fish deployed at each of the high-impact sites exhibited a down-regulation of CYP39A1 relative to fish from both of the low-impact sites and an upregulation of CYP3A4 relative to Niobrara fish. When compared to fish caged at the Dismal River, fish from both high-impact sites experienced alterations in G-protein coupled receptor protein signaling pathway (GO:0007186), chromatin assembly or disassembly (GO:00006333), and chromatin modification (GO:0016568). However, no other biological processes were found to be consistently altered in fish caged at both high-impact sites relative to those caged at the low-impact sites (Table 3). High-Impact Sites. Significant differences in the expression of 1218 transcripts were observed between males caged at the Platte and Elkhorn Rivers (Figure 1, Supporting Information). Several of these discriminative genes, including tumor protein P53 binding protein, DNA methyltransferase 3A, and activating

Figure 2. Venn diagrams of down-regulated and up-regulated hepatic genes in male fathead minnows caged at Elkhorn and Platte Rivers against the Dismal (a) and Niobrara Rivers (b).

included lipid catabolism (GO:0016042) and steroid biosynthesis (GO:0006694) (Table 3). Similarities in gene expression patterns were observed when minnows from the high-impact sites were compared to

Table 2. Subset of Transcripts Differentially Regulated among Fish Caged at the Two High-Impact Sites Relative to Those Caged at the Low-Impact Sitesa probe ID Up-Regulated EA_Pp_52006a EA_Pp_60180a EA_Pp_58540a EA_Pp_68092a EA_Pp_60408a EA_Pp_61451a EA_Pp_69420a

gene name potassium voltage gated channel, Shabrelated subfamily, member 1 haptoglobin similar to neurotoxin/C59/Ly-6-like protein cyclin-dependent kinase inhibitor 1A CD209 antigen RAS-like estrogen-regulated growth inhibitor activating transcription factor 3

EA_Pp_56320a EA_Pp_55594a Down-Regulated EA_Pp_52419a

complement component 7 solute carrier family 26, member 5

EA_Pp_70715a EA_Pp_50338a EA_Pp_51250a

deiodinase, iodothyronine, type II lanosterol synthase cytochrome P450, family 2, subfamily C, polypeptide 29 DNA methyltransferase 3A tumor protein P53 binding protein cytochrome P450 39A phosphoserine phosphatase retinol dehydrogenase 8 like

EA_Pp_64706a EA_Pp_60472a EA_Pp_53730a EA_Pp_68162a EA_Pp_52740a

fatty acid desaturase 2

Elkhorn vs Dismal

biological process

Platte vs Dismal

Elkhorn vs Niobrara

Platte vs Niobrara

potassium ion transport

4.9

7.0

proteolysis

14.5 23.7

6.9 31.4

cell cycle arrest endocytosis protein transport and small GTPase mediated signal transduction regulation of transcription, DNAdependent induction of apoptosis sulfate transport

15.8 251.3 2.9

3.5

8.0 11178.3 3.1

3.8

14.0

15.2

4.1

4.4

5.5 5.5

12.9 7.9

4.1 14.4

9.9 20.5

lipid and fatty acid biosynthesis, fatty acid desaturation

4.0

5.2

12.5 lipid and steroid biosynthesis

12.6 15.6

transcription positive regulation of transcription bile acid and cholesterol catabolism amino acid biosynthesis estrogen biosynthesis

2.2 2.4 5.1 7.9 51.5

12.1 17.5 20.3

2.5 3.4 5.9 12.4 30.3

2.9 2.3 5.2 7.5 20.3

3.3 3.2 5.9 11.8 12.0

a

Selected genes include those involved in the altered biological processes identified by functional enrichment analysis or that were found to be highly differentially regulated. 1880

dx.doi.org/10.1021/es2039097 | Environ. Sci. Technol. 2012, 46, 1877−1885

Environmental Science & Technology

Article

Table 3. Functional Enrichment Analysis Using Gene Ontology (GO) Biological Processesa p-value/% DR biological process

Elkhorn vs Dismal

DNA-dependent DNA replication (GO:0006261) DNA repair (GO:0006281) response to DNA damage stimulus (GO:0006974) ion transport (GO:0006811) chromatin assembly or disassembly (GO:0006333) G-protein coupled receptor protein signaling pathway (GO:0007186) microtubule-based movement (GO:0007018) cholesterol biosynthesis (GO:0006695) lipid catabolism (GO:0016042) lipid biosynthesis (GO:0008610) steroid biosynthesis (GO:0006694) regulation of GTPase activity (GO:0043087)

0.06/0.47 0.01/1.17 0.01/0.39 0.03/0.70 0.02/0.47 0.02/0.39 0.01/0.55

Platte vs Dismal