Article pubs.acs.org/est
Using Transcriptomic Tools to Evaluate Biological Effects Across Effluent Gradients at a Diverse Set of Study Sites in Minnesota, USA Jason P. Berninger,*,† Dalma Martinović-Weigelt,‡ Natàlia Garcia-Reyero,§ Lynn Escalon,⊥ Edward J. Perkins,⊥ Gerald T. Ankley,¶ and Daniel L. Villeneuve¶ †
National Research Council, U.S. Environmental Protection Agency, 6201 Congdon Blvd., Duluth, Minnesota 55804, United States University of St. Thomas, St. Paul, Minnesota 55105, United States § Institute of Genomics, Biocomputing, and Biotechnology, Mississippi State University, Starkville, Mississippi 39762, United States ⊥ U.S. Army Engineer Research and Development Center, 3909 Halls Ferry Rd., Vicksburg, Mississippi 39180, United States ¶ Mid-Continent Ecology Division, U.S. Environmental Protection Agency, 6201 Congdon Blvd., Duluth, Minnesota 55804, United States ‡
S Supporting Information *
ABSTRACT: The aim of this study was to explore the utility of “omics” approaches in monitoring aquatic environments where complex, often unknown stressors make chemical-specific risk assessment untenable. We examined changes in the fathead minnow (Pimephales promelas) ovarian transcriptome following 4-day exposures conducted at three sites in Minnesota (MN, USA). Within each site, fish were exposed to water from three locations along a spatial gradient relative to a wastewater treatment plant (WWTP) discharge. After exposure, site-specific impacts on gene expression in ovaries were assessed. Using an intragradient point of comparison, biological responses specifically associated with the WWTP effluent were identified using functional enrichment analyses. Fish exposed to water from locations downstream of the effluent discharges exhibited many transcriptomic responses in common with those exposed to the effluent, indicating that effects of the discharge do not fully dissipate downstream. Functional analyses showed a range of biological pathways impacted through effluent exposure at all three sites. Several of those impacted pathways at each site could be linked to potential adverse reproductive outcomes associated with the hypothalamic−pituitary−gonadal (HPG) axis in female fathead minnows, specifically signaling pathways associated with oocyte meiosis, TGF-beta signaling, gonadotropin-releasing hormone (GnRH) and epidermal growth factor receptor family (ErbB), and gene sets associated with cyclin B-1 and metalloproteinase. The utility of this approach comes from the ability to identify biological responses to pollutant exposure, particularly those that can be tied to adverse outcomes at the population level and those that identify molecular targets for future studies.
■
INTRODUCTION Many environmental surveillance and monitoring efforts are hampered by complex chemical mixtures, the presence of unknown contaminants, and nonchemical, environmental stressors. Biological response-based approaches provide a means to integrate all potential impacts and are increasingly used as a means to complement and address the limitations of chemical-focused environmental assessments.1 Historically, these methods evaluate whole organism responses based on quantifiable adverse responses (mortality, growth, and reproduction). However, these direct assessments are not necessarily ideal, due to the likelihood that, if effects are observed, the ecosystem may already be significantly impacted. An alternate approach is utilizing more sensitive molecular and biochemical responses to environmental exposure to provide a predictive early warning for potential hazards, possibly allowing actions to be taken before organismal level impacts occur. The utility of predictive end points are most fully realized when they © 2014 American Chemical Society
can be credibly linked to probable adverse outcomes at the individual or population level. This linking of molecular alterations to changes at higher levels of biological complexity has been termed an adverse outcome pathway (AOP)2 and provides a framework for understanding the implications of a molecular event in a risk assessment context. Using the AOP framework in the context of environmental monitoring and predictive assessment does have limitations. For example, in terms of selecting appropriate molecular end points, it requires a priori assumptions about the type of effects likely to be present, a potential difficulty for assessments of areas with complex mixtures and unknown chemicals. One potential bridge for that limitation is the use of “omic” tools. Received: Revised: Accepted: Published: 2404
September 10, 2013 January 14, 2014 January 16, 2014 January 16, 2014 dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
Environmental Science & Technology
Article
the gradient and site-specific differences that might exist between laboratory control and/or remote reference site waters. Finally, the study was repeated at three different sites in MN to evaluate the utility of the approach across different chemical, land use, and population profiles. The goal of this research was to evaluate the following hypotheses: (1) transcriptomic profiles downstream of a WWTP discharge would be distinct from those upstream and exhibit some responses in common with those initiated by exposure to the effluent itself; (2) the impact of the point source (WWTP effluent) on the fish transcriptome could be discriminated from the influences of other site-related variables; and (3) using supervised and unsupervised approaches, possible adverse outcomes downstream of the effluent discharge could be inferred from functional analyses of the transcriptomic results, thus supporting the use of transcriptomics as a tool in AOP-linked predictive effects-based monitoring and screening efforts. An additional goal of this study was to compare trends and similarities of the inferred biological responses to the effluents, among the three associated studies.
Techniques such as transcriptomics, proteomics, and metabolomics provide a means to examine tissue-specific impacts on a large diversity of biological pathways. 3,4 Thus, omics approaches may be well suited for broad-based surveillance/ monitoring where complex mixtures and unknown stressors may be present. Through the application of various functional analyses (e.g., based on gene ontologies and other functional annotations) and AOP knowledge, omics can support both supervised approaches, targeting existing AOPs, identifying responses previously established as functionally significant within the pathway, and unsupervised approaches, potentially identifying alternate pathways of prospective concern, through which it may be possible to draw some inferences/develop hypotheses to aid in targeting subsequent monitoring.1 Utilizing a combination of supervised and unsupervised analyses allows omics data to meet the need to link molecular/biochemical responses to adverse outcomes important to environmental regulations while more comprehensively exploring the biological impacts of complex mixtures, multiple stressors, and unknown chemicals. Although the use of omics approaches in field-based monitoring and assessments remains quite an emerging area, some pioneering work has been done. For example, Williams et al.5 and Falciani et al.6 used a transcriptomic approach with wild fish to identify likely chemical initiators and changes in gene profiles, which matched well with established chemical monitoring results, particularly for historical pollutants (organics and metals). Wild fish, while perhaps the best integrator of possible impacts at the population level, are not ideal for site-specific environmental monitoring and surveillance, as they may be transient site residents or not present at all in the site of interest. To avoid that problem, a handful of studies have used laboratory reared fish in the application of transcriptomics for monitoring in complex effluents.7−11 In some experiments,7,8 effluent was brought into the lab to reduce external stressors (e.g., low dissolved oxygen (DO), limited food, temperature fluctuation) and fish exposed to standard laboratory water are used as the point of comparison. Other studies9−11 utilized in situ cages to expose fish to effluents. Changes in the transcriptome of exposed fish were evaluated relative to a reference site. For example, GarciaReyero et al.9 and Ing et al.10 both utilized a site upstream of an effluent discharge as the reference site, to provide a comparison that more closely represented ambient conditions. The current study (and its two associated studies12,13) sought to expand on previous efforts by coupling many of the strongest elements of previous work, with extensive information about the site’s chemical and physical nature,14 and grounding the approach in established, sound ecotoxicological practice. We utilized laboratory reared fathead minnows (Pimephales promelas) to examine changes in the ovarian transcriptome following exposures to complex effluents. Analysis of ovarian tissue provided the concurrent studies,12,13 which utilize liver, a complementary approach using an alternative tissue that specifically reflects reproductive processes/pathways. To reduce external stressors, fish were exposed in field-based units15 where temperature, DO, and feeding could be controlled. Within each site, fish were exposed at three locations along a spatial gradient relative to a wastewater treatment plant (WWTP) discharge. In an effort to discern the specific impact(s) of the WWTP discharges, the upstream site within each gradient was used as reference location. This intragradient comparison serves to reduce the potential influence of other point sources outside of
■
METHODS Site Selection. The methods for site selection and experimental design are described in detail in MartinovićWeigelt et al.12 Briefly, three study sites, differing in land use, population size, agricultural and urban influence, and geography were selected within the state of Minnesota: Ely (Shagawa Lake; site E), Hutchinson (Crow River; site H), and Rochester (Zumbro River; site R). The study sites represent a diversity of land-use, WWTP size, receiving waters (lakes versus rivers), water quality, and effluent chemical composition (Table S1, Supporting Information).14 Within each of these sites, three sampling locations were selected representing a gradient with regard to a specific WWTP discharge (Figure S1, Supporting Information): an upstream location representing the basal condition of the receiving waters prior to the influx of effluent; an effluent location, using final, finished effluent prior to mixing with receiving water and representing the most extreme potential impact; and a downstream location representing the conditions after the final effluent had mixed with the receiving water (Figure S2a−c, Supporting Information). Additional site descriptions are presented in the Supporting Information (SM1). Site demographics, land use, treatment systems, water quality, select chemicals, and targeted biological activity (i.e., primarily estrogenic activity) were evaluated one year prior to the present study (Table S1, Supporting Information).14 Fish Exposures. At the specific locations (upstream, effluent, downstream), sexually mature female fathead minnows (SM2) were exposed to location specific water in mini-mobile environmental monitoring units (MMU)15 for a period of 4 d, at each of three sites (R, H, E) during the summer and fall of 2010 (SM3). The MMUs allowed for consistency in aeration, temperature, and feeding. In the case of the riverine R and H sites, the MMU was supplied with a continuous flow of location-specific water. Due to logistical constraints, exposures at the lake site (E) were conducted under static renewal conditions within the MMUs. However, the cumulative number of daily water exchanges was equivalent to that in the flowthrough studies. After exposure, fish were anesthetized (buffered tricaine methane sulfonate [MS-222]) and necropsied at a facility near the site, and specific tissues were excised, flash frozen in liquid nitrogen, and stored at −80 °C. Exposure 2405
dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
Environmental Science & Technology
Article
Figure 1. Evaluation of gene expression across a wastewater effluent gradient at three geographically separated sites within Minnesota, US. The total number of differentially expressed genes was determined by statistical analysis within sites. DEGs were grouped on the basis of gene response profiles relative to upstream at each site. The three groups consisted of unique gene responses at the downstream (DRP) and effluent (ERP) locations and genes with the same response, relative to upstream grouped within the common response profile (CRP). A small percentage of genes overlap within the DRP and ERP groups and are shown in a separate column. An example gene map of site R is shown; for clarity, genes not significantly different from upstream were not displayed.
group variances) and a critical p-value of 0.01, with no multiple test correction. The decision to not use multiple testing corrections follows that of previous field studies8,11,12 to allow for a broader characterization of potential biological impacts. In order to focus on the effects of wastewater on the downstream site, the upstream location was regarded as the basal or reference environmental condition for each site. While the upstream location is neither pristine nor a perfect control/ reference site, it represents the site-specific environmental conditions of the water body prior to the input of effluent within each site gradient. To group the data for further evaluation, significant differences relative to the upstream location (TMeV v4.8) were determined, as the comparisons most germane to our study objectives. Specifically, this comparison provided a means to evaluate the influence of effluent on the downstream location. Additionally, the use of a site-specific reference was intended to help control for the many ambient environmental conditions that vary markedly among sites but only minimally among locations within a site. For the purpose of visualization and functional analysis, DEGs were arranged on the basis of the fold change (log2) in gene expression (up or down) relative to the upstream reference for each gene (Figure 1). The DEGs were grouped on the basis of three response profiles relative to the upstream reference: CRP, DRP, and ERP. The common response profile (CRP) category consists of the group of genes differentially expressed at both the downstream and effluent locations, relative to upstream, and whose fold changes were in the same direction. The downstream response profile (DRP) contains genes unique to the downstream location, where downstream was significantly different from upstream, but effluent was not or was significantly different but responded with an opposite response relative to downstream. The effluent response profile (ERP) category consists of genes unique to the effluent location, where effluent was significantly different from
methods for this study were identical to those described by Martinović-Weigelt et al.12 Transcriptomics. For the microarray work, methods follow those described in Villeneuve et al.16 Total RNA was isolated from ovary tissue of fathead minnow using RNeasy kits (Qiagen, Valencia, CA). RNA quality was evaluated using an Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA), and RNA was quantified spectrophotometrically (Nanodrop, Wilmington, DE). Total RNA samples were stored at −80 °C until used for microarray analyses. RNA from 53 female fathead minnows (3 sites × 3 locations; n = 6 per location, except site E-downstream where n = 5) were analyzed using a fathead minnow 15 000 feature microarray (GEO Platform Accession GPL9248) purchased from Agilent Technologies (Palo Alto, CA). One μg of total RNA from each sample was used in complementary RNA synthesis, complementary RNA labeling, amplification, and hybridizations as recommended in the manufacturer’s kits and protocols (Quick Amp labeling kit; Agilent). Microarrays were hybridized following the manufacturer’s guidelines for One-Color Microarray-Based Gene Expression Analysis (version 5.7; Agilent). Microarrays were scanned with a SureScan High-Resolution Microarray scanner G2505 C (Agilent). Data were extracted from microarray images using Feature Extraction Software (Agilent). Text versions of the raw data values have been deposited to the Gene Expression Omnibus (GEO: http://www.ncbi.nlm.nih.gov/geo/; Accession # GSE53204). Microarray data were normalized using Fastlo17 implemented in R (http://www.r-project.org/). Data were normalized as a single group, including all sites. The determination of significantly differentially expressed genes (DEGs) was based on ANOVA followed by pairwise t tests (p < 0.01 for both) for each intrasite comparison (upstream vs downstream, upstream vs effluent, downstream vs effluent) implemented in MultiExperiment Viewer (TMeV) v4.8 (www. tm4.org)18 using a Welch approximation (assuming unequal 2406
dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
Environmental Science & Technology
Article
Figure 2. Principal component analysis of differentially expressed genes in fathead minnow ovaries at three sites (R, E, H) across an established effluent gradient. Each site has three study locations: upstream (○), effluent (▲), and downstream (■). Bars represent standard deviation across each principal component (PC). All locations within sites are significantly different (ANOVA p < 0.05), with the exception of upstream and downstream on PC2 for sites R and E and effluent and downstream on PC1 for site H.
■
RESULTS AND DISCUSSION Differences in Physical and Chemical Site Characteristics. Chemical and water quality parameters were measured one year prior to this fish analysis but still provide a qualitative assessment of the potential influence of effluent on the downstream ecosystem. The flow characteristics of the two riverine sites, H and R, show that effluent contributes substantially to downstream flow (Table S1: Water Quality Measurements, Supporting Information). A number of the measured water quality parameters exhibit this effluent influence on downstream sites, as evidenced by their change from upstream. Changes were observed in downstream conductivity, pH, turbidity, chloride, sulfate, ammonia/ nitrate/nitrite, phosphorus, sodium, and suspended sediment that reflect the influx of effluent into the system, particularly at the R and H sites (Table S1, Supporting Information). The number of anthropogenic chemicals detected at the upstream locations was generally indicative of land-use and number of upstream WWTPs at the R and H sites.12,14 The number of chemical pollutants in the effluent is likely related to the efficiency and treatment technology differences between larger and smaller WWTPs. Analysis of downstream locations showed more chemicals than those measured at upstream but lower than in the effluent, likely due to dilution, degradation, or adsorption. These changes suggest that there is potential for effluent to cause effects on downstream biological communities despite dilution with receiving water. Differentially Expressed Genes, Gene Response Profiles, and Principle Component Analysis. The E and H sites had a comparable number of DEGs (400−500), while analysis of the R site resulted in an order of magnitude more DEGs (∼4900; Figure 1). The PCA of differentially expressed genes at the three sites showed separation (with some overlap) of the three locations within each site, with the majority of the variation (∼60%) and the significant differences in PC scores occurring within the first two principle components (PC1 and PC2; Figure 2). For sites R and E (Figure 2R,E), PCA results were similar showing significant separation of all sites along PC1 (X-axis) which captures the greatest amount of the variability in the data (i.e., 43% at R site; 42% at E site). Effluent also separated from UP and DS along PC2 (Y-axis) suggesting some distinct characteristics in the expression profile in the effluent-exposed fish, compared to those exposed to river or lake waters. At site H, (Figure 2H) both effluent and downstream separated from upstream along PC1 (X-axis, capturing 30% of the variation) suggesting that responses common to the effluent and downstream sites drove the loadings along PC1. However, there was still clear separation
upstream, but downstream was not or was significantly different but the response was opposite relative to effluent. Principal components analysis (PCA; with median centering) was used to examine the relationship among DEGs at each location within a site (MS Excel, Multibase2013 extension; http://www.numericaldynamics.com). The groups of genes within the gene response profiles were also evaluated using PCA to determine if differences among response profiles were significant and would warrant further analysis. Differences in average principal component scores along each principal component were evaluated using ANOVA with Bonferroni post hoc analysis (Sigma Plot v. 12, Systat Software Inc.). Functional Annotation. To evaluate the biological implications of these three response profiles (DRP, ERP, CRP) for each site (R, H, E) in an unsupervised manner, DAVID (Database for Annotation, Visualization and Integrated Discovery v6.719,20) functional enrichment analyses were used to identify over-represented biological pathways among the set of DEGs (SM4) that could be mapped to KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway orthologs in DAVID. For the individual pathways, enrichment was considered significant if p < 0.05, FDR (false discovery rate) < 0.25, and at least two DEGs were associated with the pathway. A less stringent FDR ( E). Transcriptomic approaches were able to detect a HPG-relevant biological signature at all sites. While not likely to be necessary for all contaminated sites, these studies show the promise of omics-based approaches to help with molecular target prioritization and hypotheses development, particularly when coupled with appropriate experimental design for sites with complex mixtures of unknown chemicals. This intrasite comparison approach provides a logical, experimentally sound approach to limiting the ambient biological variability across a monitoring site by focusing on the point source (in this case effluent) rather than trying to discern differences between laboratory controls and field-based responses indicative of general ambient conditions but not necessarily related to point source pollution. While this approach allows for the observation of changes moving across the effluent gradient, it cannot necessarily account for environmentally degraded conditions that occur across the entire system. An external laboratory control may be useful in this aspect, but this is also limited by an inability to differentiate between site-specific (e.g., confounding environmental factors) and anthropogenic influences (e.g., chemical composition of effluent). This type of analysis provides a simple/reasonable and reproducible study design, which has been demonstrated to be applicable across different types of environments (i.e., lakes or rivers, land use difference, upstream anthropogenic influences).
three distantly located sites with different levels of background contamination. The study design, using the upstream location site as a point of comparison rather than a separate lab control, provided a means to reduce variability in the gene expression data caused by ambient conditions common to all locations within a site. This method evaluation does make the assumption that there is a substantial degree of similarity in water conditions at all three locations along the gradient and that, by using an intragradient point of comparison (upstream in this design), variability will be reduced and biological responses due to anthropogenic influences are therefore more evident. Using response categories, the amount of differential and common response among DEGs can be observed. The most obvious is the common response profile, where similarity in response between effluent and downstream-exposed fish suggests the presence of effluent in the system is initiating common molecular responses. Similarly, the influence of upstream waters on downstream responses is observed in the nonoverlapping portion of the effluent response profile (both upstream and downstream are significantly different from effluent but not from each other). The response profile approach allows analysis efforts to be focused on the genes that are likely reflecting changes initiated by exposure to the point source. Even in highly polluted systems, the ability to discern relative biological contributions associated with each source is critically important to monitoring and assessment efforts. In addition to identifying effluent-specific responses, this methodology allows those responses to be tracked as the effluent moves downstream. While each of the three sites were unique in terms of ambient conditions, the transcriptomic data show that effluent exerts a considerable influence upon the biological responses of organisms in the immediate downstream area. Additionally, this analysis provided a means to identify common trends in biological responses to effluent, particularly with the similarity in gene expression response profiles. In the DAVID−KEGG pathway identification, in particular, the influence of effluent was clearly observable downstream. Supervised and Unsupervised Transcriptomics Approaches in Environmental Monitoring. Ecotoxicology research focuses on improving environmental protection often through supporting risk assessment and regulatory decisionmaking. Our transcriptomic analysis has been focused on establishing linkages to impacts on reproductive end points in that these effects are critically important in risk assessment and regulatory frameworks.29 Through the focused lens of an AOP, transcription effects can be connected to population level through increasing levels of biological complexity and ultimately to population level.1,2 In this study, the unsupervised pathway analysis and the supervised gene set analysis identified responses that have been previously linked to reproductive AOPs as a result of exposure to effluent at all three study sites. The supervised approach, enriched gene set analysis, was consistent with responses found as a result of endocrine-active chemical exposures in the laboratory.21 While the same chemicals previously identified as inducing expression of the gene sets may not have been observed or measured in the field (e.g., 17β trenbolone),14 several chemicals were present that are known to initiate the same molecular responses along the androgen receptor activation AOP.2 Utilizing the unsupervised approach, the genes and gene pathways were identified at each site that enables the selection and prioritization of targeted biological end points for future
■
ASSOCIATED CONTENT
S Supporting Information *
Descriptions of physical and chemical differences among sites and locations, further explanations of some methods, and spreadsheets with in depth descriptions and identifications of differentially expressed genes, gene expression profiles, and 2410
dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
Environmental Science & Technology
Article
impacts of a wastewater effluent on the fathead minnow. Aquat. Toxicol. 2011, 101 (17), 38−48. (8) Vidal-Dorsch, D. E.; Colli-Dula, R. C.; Bay, S. M.; Greenstein, D. J.; Wiborg, L.; Petschauer, D.; Denslow, N. D. Gene expression of fathead minnows (Pimephales promelas) exposed to two types of treated municipal wastewater effluents. Environ. Sci. Technol. 2013, 47 (19), 11268−11277. (9) Ings, J. S.; Servos, M. R; Vijayan, M. M. Hepatic transcriptomics and protein expression in rainbow trout exposed to municipal wastewater effluent. Environ. Sci. Technol. 2011, 45 (6), 2368−2376. (10) Garcia-Reyero, N.; Adelman, I. R.; Martinović, D.; Liu, L.; Denslow, N. D. Site-specific impacts on gene expression and behavior in fathead minnows (Pimephales promelas) exposed in situ to streams adjacent to sewage treatment plants. BMC Bioinf. 2009, 10 (S11), S11. (11) Sellin-Jeffries, M. K.; Mehinto, A. C.; Carter, B. J.; Denslow, N. D.; Kolok, A. S. Taking microarrays to the field: Differential hepatic gene expression of caged fathead minnows from Nebraska watersheds. Environ. Sci. Technol. 2012, 46 (3), 1877−1885. (12) Martinović-Weigelt, D.; Mehinto, A. C.; Ankley, G. T.; Denslow, N. D.; Barber, L. B.; Lee, K. E.; King, R. J.; Schoenfuss, H. L.; Schroeder, A. L.; Villeneuve, D. L. Transcriptomic effects-based monitoring for endocrine active chemicals: Assessing relative contribution of treated wastewater to downstream pollution. Environ. Sci. Technol. 2013, DOI: 10.1021/es404027n. (13) Skelton, D. M.; Ekman, D. R.; Martinović-Weigelt, D.; Ankley, G. T.; Villeneuve, D. L.; Teng, Q.; Collette, T. W. Metabolomics for in situ environmental monitoring of surface waters impacted by contaminants from both point and nonpoint sources. Environ. Sci. Technol. 2013, DOI: 10.1021/es404021f. (14) Lee, K. E.; Langer, S. K.; Barber, L. B.; Writer, J. H.; Ferrey, M. L.; Schoenfuss, H. L.; Furlong, E. T.; Foreman, W. T.; Gray, J. L.; ReVello, R. C.; Martinović, D.; Woodruff, O. P.; Keefe, S. H.; Brown, G. K.; Taylor, H. E.; Ferrer, I.; Thurman, E. M. Endocrine active chemicals, pharmaceuticals, and other chemicals of concern in surface water, wastewater-treatment plant effluent, and bed sediment, and biological characteristics in selected streams, Minnesota-design, methods, and data, 2009. U.S. Geological Survey Data Series 575; USGS: Reston, VA, 2011; 54 p., with appendixes. (15) Kolok, A. S.; Miller, J. T.; Schoenfuss, H. L. The mini mobile environmental monitoring unit: A novel bio-assessment tool. J. Environ. Monit. 2012, 14 (1), 202−208. (16) Villeneuve, D. L.; Garcia-Reyero, N.; Escalon, B. L.; Jensen, K. M.; Cavallin, J. E.; Makynen, E. A.; Durhan, E. J.; Kahl, M. D.; Thomas, L. M.; Perkins, E. J.; Ankley, G. T. Ecotoxicogenomics to support ecological risk assessment: A case study with bisphenol A in fish. Environ. Sci. Technol. 2012, 46 (1), 51−59. (17) Ballman, K. V.; Grill, D. E.; Oberg, A. L.; Therneau, T. M. Faster cyclic loess: Normalizing RNA arrays via linear models. Bioinformatics 2004, 20 (16), 2778−2786. (18) 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. TM4: A free, open-source system for microarray data management and analysis. Biotechniques 2003, 34 (2), 374−378. (19) Huang, D. W.; Sherman, B. T.; Lempicki, R. A. Systematic and integrative analysis of large gene lists using DAVID Bioinformatics Resources. Nature Protoc. 2009, 4 (1), 44−57. (20) Huang, D. W.; Sherman, B. T.; Lempicki, R. A. Bioinformatics enrichment tools: Paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009, 37 (1), 1−13. (21) Villeneuve, D. L.; Garcia-Reyero, N.; Martinović-Weigelt, D.; Li, Z.; Watanabe, K. H.; Orlando, E. F.; LaLone, C. A.; Edwards, S. W.; Burgoon, L. D.; Denslow, N. D.; Perkins, E. J.; Ankley, G. T. A graphical systems model and tissue-specific functional gene sets to aid transcriptomic analysis of chemical impacts on the female teleost reproductive axis. Mutat. Res. 2012, 746 (2), 151−162. (22) Viarengo, A.; Lowe, D.; Bolognesi, C.; Fabbri, E.; Koehler, A. The use of biomarkers in biomonitoring: A 2-tier approach assessing
KEGG pathways identified via DAVID. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]; phone: 218-529-5027; fax: 218-529-5003. Notes
Mention of trade names or commercial products does not constitute endorsement or recommendation for use. The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the US Environmental Protection Agency. This work was partly funded by the US Army Environmental Quality Research Program (including BAA 11-4838). Permission for publishing this information has been granted by the Chief of Engineers. The authors thank Kathy Lee (USGS), Heiko Schoenfuss (St. Cloud State University), Leah Wehmas, and their co-workers for their efforts in collecting samples that were used in this study and Anthony Schroeder for helpful comments on the manuscript.
■
REFERENCES
(1) Ekman, D. R.; Ankley, G. T.; Blazer, V. S.; Collette, T. W.; Garcia-Reyero, N.; Iwanowicz, L. R.; Jorgenson, Z. G.; Lee, K. E.; Mazik, P. M.; Miller, D. H.; Perkins, E. J.; Smith, E. T.; Tiege, J. E.; Villeneuve, D. L. Biological effects-based tools for monitoring impacted surface waters in the Great Lakes: A multi-agency program in support of the GLRI. Environ. Pract. 2013, 15 (4), 409−426. (2) Ankley, G. T.; Bennett, R. S.; Erickson, R. J.; Hoff, D. J.; Hornung, M. W.; Johnson, R. D.; Mount, D. R.; Nichols, J. W.; Russom, C. L.; Schmieder, P. K.; Serrano, J. A.; Tietge, J. E.; Villeneuve, D. L. Adverse outcome pathways: A conceptual framework to support ecotoxicology research and risk assessment. Environ. Toxicol. Chem. 2010, 29 (3), 730−741. (3) Ankley, G. T.; Daston, G. P.; Degitz, S. J.; Denslow, N. D.; Hoke, R. A.; Kennedy, S. W.; Miracle, A. L.; Perkins, E. J.; Snape, J.; Tillitt, D. E.; Tyler, C. R.; Versteeg, D. Toxicogenomics in regulatory ecotoxicology. Environ. Sci. Technol. 2006, 40 (13), 4055−4065. (4) Van Aggelen, G.; Ankley, G. T.; Baldwin, W. S.; Bearden, D. W.; Benson, W. H.; Chipman, J. K.; Collette, T. W.; Craft, J. A.; Denslow, N. D.; Embry, M. R.; Falciani, F.; George, S. G.; Helbing, C. C.; Hoekstra, P. F.; Iguchi, T.; Kagami, Y.; Katsiadaki, I.; Kille, P.; Liu, L.; Lord, P. G.; McIntyre, T.; O’Neill, A.; Osachoff, H.; Perkins, E. J.; Santos, E. M.; Skirrow, R. C.; Snape, J. R.; Tyler, C. R.; Versteeg, D.; Viant, M. A.; Volz, D. C.; Williams, T. D.; Yu, L. Integrating omic technologies into aquatic ecological risk assessment and environmental monitoring: Hurdles, achievements, and future outlook. Environ. Health Perspect. 2010, 118 (1), 1−5. (5) Williams, T. D.; Turan, N.; Diab, A. M.; Wu, H.; Mackenzie, C.; Bartie, K. L.; Hrydziuszko, O.; Lyons, B. P.; Stentiford, G. D.; Herbert, J. M.; Abraham, J. K.; Katsiadaki, I.; Leaver, M. J.; Taggart, J. B.; George, S. G.; Viant, M. R.; Chipman, K. J.; Falciani, F. Towards a system level understanding of non-model organisms sampled from the environment: a network biology approach. PLoS Comput. Biol. 2011, 7 (8), No. e1002126. (6) Falciani, F.; Diab, A. M.; Sabine, V.; Williams, T. D.; Ortega, F.; George, S. G.; Chipman, J. K. Hepatic transcriptomic profiles of European flounder (Platichthys f lesus) from field sites and computational approaches to predict site from stress gene responses following exposure to model toxicants. Aquat. Toxicol. 2008, 90 (2), 92−101. (7) Garcia-Reyero, N.; Lavelle, C. M.; Escalon, B. L.; Martinović, D.; Kroll, K. J.; Sorensen, P. W.; Denslow, N. D. Behavioral and genomic 2411
dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
Environmental Science & Technology
Article
the level of pollutant-induced stress syndrome in sentinel organisms. Comp. Biochem. Physiol., Part C 2007, 146 (3), 281−300. (23) Ankley, G. T.; Bencic, D. C.; Cavallin, J. E.; Jensen, K. M; Kahl, M. D.; Makynen, E. A.; Martinović, D.; Mueller, N. D.; Wehmas, L. C.; Villeneuve, D. L. Dynamic nature of alterations in the endocrine system of fathead minnows exposed to the fungicide prochloraz. Toxicol. Sci. 2009, 112 (2), 344−353. (24) Ekman, D. R.; Villeneuve, D. L.; Teng, Q.; Ralston-Hooper, K. J.; Martinović-Weigelt, D.; Kahl, M. D.; Jensen, K. M.; Durhan, E. J.; Makynen, E. A.; Ankley, G. T.; Collette, T. W. Use of gene expression, biochemical and metabolite profiles to enhance exposure and effects assessment of the model androgen 17β-trenbolone in fish. Environ. Toxicol. Chem. 2011, 30 (2), 319−329. (25) Villeneuve, D. L.; Mueller, N. D.; Martinović, D.; Makynen, E. A.; Kahl, M. D.; Jensen, K. M.; Durhan, E. J.; Cavallin, J. E.; Bencic, D.; Ankley, G. T. Direct effects, compensation, and recovery in female fathead minnows exposed to a model aromatase inhibitor. Environ. Health Perspect. 2009, 117 (4), 624−631. (26) Wang, R. L.; Bencic, D.; Villeneuve, D. L.; Ankley, G. T.; Lazorchak, J.; Edwards, S. A transcriptomics-based biological framework for studying mechanisms of endocrine disruption in small fish species. Aquat. Toxicol. 2010, 98, 230−244. (27) Ekman, D. R.; Hartig, P. C.; Cardon, M.; Skelton, D. M.; Teng, Q.; Durhan, E. J.; Jensen, K. M.; Kahl, M. D.; Villeneuve, D. L.; Gray, L. E., Jr.; Collette, T. W.; Ankley, G. T. Metabolite profiling and a transcriptional activation assay provide direct evidence of androgen receptor antagonism by bisphenol a in fish. Environ. Sci. Technol. 2012, 46 (17), 9673−9680. (28) Xu, J.; Huang, W.; Chengrong, Z.; Luo, D.; Li, S.; Zhu, Z.; Hu, W. Defining global gene expression changes of the hypothalamicpituitary-gonadal axis in female sGnRHAntisense transgenic common carp (Cyprinus carpio). PLoS One 2011, 6 (6), No. e21057. (29) Ankley, G. T.; Breen, B. D.; Collette, T. W.; Conolly, R.; Denslow, N. D.; Edwards, S.; Ekman, D. R.; Jensen, K. M.; Lazorchak, J.; Martinović, D.; Miller, D. H.; Perkins, E. J.; Orlando, E. F.; GarciaReyero, N.; Villeneuve, D. L.; Wang, R. L.; Watanabe, K.. Endocrine disrupting chemicals in fish: Developing exposure indicators and predictive models of effects based on mechanism of action. Aquat. Toxicol. 2009, 92 (3), 168−178.
2412
dx.doi.org/10.1021/es4040254 | Environ. Sci. Technol. 2014, 48, 2404−2412
