Citation
Transcriptome profiling in th fathead minnow (Pimephales promelas) following exposure to complex chemical mixtures

Material Information

Title:
Transcriptome profiling in th fathead minnow (Pimephales promelas) following exposure to complex chemical mixtures a landscape based approach
Added title page title:
Landscape based approach
Creator:
Bertolatus, David ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (84 pages) : ;

Subjects

Subjects / Keywords:
Minnows ( lcsh )
Fishes -- Effect of human beings on ( lcsh )
Fishes -- Effect of human beings on ( fast )
Minnows ( fast )
Shenandoah River (Va. and W.V.) ( lcsh )
United States -- Shenandoah River ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Anthrohropogenic chemicals of emerging concern [CECs] are commonly detected in surface waters and several are known to cause adverse effects in aquatic vertebrates. Although there is a large body of research documenting the effects of exposure to single chemicals in laboratory settings, less is known about the effects caused by the complex mixtures of chemicals that occur in aquatic ecosystems. To characterize these effects, we exposed adult fathead minnows (Pimephales promelas) to water from four different locations within the Shenandoah River watershed (VA, USA) using flow-through mobile laboratories. The exposure locations were chosen to capture unique and representative landuse in surrounding watersheds, including agricultural, urban, mixed-use, and forested. In addition to biological sampling, water samples were taken every 7 days during the fish exposure and analyzed for 460 chemical constituents. Each location had a unique chemical profile that was generally consistent with landuse in the surrounding watershed. Whole-organism and molecular responses also differed between the locations. Fish exposed at both agricultural and WWTP impacted sites had a reduced number of nuptial tubercles. At the agricultural site, survivorship was significantly reduced. Genome-wide transcription profiles were measured to investigate the molecular underpinnings of these higher-level changes and to gain an unbiased observation of the physiological state of animals following exposure. Differentially expressed genes and pathways were identified using ANOVA, gene set, and sub-network enrichment analyses. Transcript biomarkers of endocrine disruption, including er1, er2, ar, vtg1, and vtg3, showed little to no differential expression in exposed fish, suggesting these organisms did not experience estrogenic endocrine disruption. Pathways and sub-networks related to immune function, cholesterol synthesis, and metabolism were affected by exposure at various sites. These data provide insightful hypotheses regarding the specific effects of exposure to different types of complex mixtures and demonstrate the value of our complex mixture/landscape research approach.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by David Bertolatus.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
985114355 ( OCLC )
ocn985114355
Classification:
LD1193.L45 2016m B47 ( lcc )

Downloads

This item has the following downloads:


Full Text
TRANSCRIPTOME PROFILING IN THE FATHEAD MINNOW
(Pimephalespromelas) FOLLOWING EXPOSURE TO COMPLEX CHEMICAL MIXTURES: A LANDSCAPE BASED APPROACH
by
DAVID BERTOLATUS Bachelor of Arts, University of Iowa, 2007 Bachelor of Science, Metropolitan State University of Denver, 2013
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Integrative Biology
2016


2016
DAVID BERTOLATUS ALL RIGHTS RESERVED
11


This thesis for the Master of Science degree by David Bertolatus Has been approved for the Department of Integrative Biology by
Alan M. Vajda, Chair Larry B. Barber Chris Miller John Swallow
July 30, 2016


Bertolatus, David (MS, Integrative Biology)
Transcriptome profiling in the fathead minnow {Pimephalespromelas) following exposure to complex chemical mixtures: A landscape based approach
Thesis directed by Assistant Professor Alan M. Vajda
ABSTRACT
Anthropogenic chemicals of emerging concern [CECs] are commonly detected in surface waters and several are known to cause adverse effects in aquatic vertebrates. Although there is a large body of research documenting the effects of exposure to single chemicals in laboratory settings, less is known about the effects caused by the complex mixtures of chemicals that occur in aquatic ecosystems. To characterize these effects, we exposed adult fathead minnows (.Pimephales promelas) to water from four different locations within the Shenandoah River watershed (VA, USA) using flow-through mobile laboratories. The exposure locations were chosen to capture unique and representative landuse in surrounding watersheds, including agricultural, urban, mixed-use, and forested In addition to biological sampling, water samples were taken every 7 days during the fish exposure and analyzed for 460 chemical constituents. Each location had a unique chemical profile that was generally consistent with landuse in the surrounding watershed. Whole-organism and molecular responses also differed between the locations. Fish exposed at both agricultural and WWTP impacted sites had a reduced number of nuptial tubercles. At the agricultural site, survivorship was significantly reduced. Genome-wide transcription profiles were measured to investigate the molecular underpinnings of these higher-level changes and to gain an unbiased observation of the physiological state of animals following exposure. Differentially expressed genes and pathways were identified
IV


using ANOVA, gene set, and sub-network enrichment analyses. Transcript biomarkers of endocrine disruption, including erl, er2, ar, vtgl, and vtg3, showed little to no differential expression in exposed fish, suggesting these organisms did not experience estrogenic endocrine disruption. Pathways and sub-networks related to immune function, cholesterol synthesis, and metabolism were affected by exposure at various sites. These data provide insightful hypotheses regarding the specific effects of exposure to different types of complex mixtures and demonstrate the value of our complex mixture/landscape research approach.
The form and content of this abstract are approved. I recommend its publication.
Approved: AlanM. Vajda
v


AKNOWLEDGMENTS
This work was funded by the U.S. Geological Survey's Contaminant Biology and Toxic Substances Hydrology programs.
This thesis would not have been possible without the collaboration, guidance, and support of a wide network of individuals too numerous to name individually. I would like to specifically thank Dr. Alan Vajda for starting me on my path as a researcher and continuing to support my academic development, Dr. Larry Barber of the U.S. Geological survey for his trust and support of me as a graduate researcher, Dr. Chris Martyniuk of the University of Florida for his large contribution of time and resources to my training. I would also like to thank my committee members Dr. Chris Miller and Dr. John Swallow for providing guidance and feedback on this thesis. Chemical laboratory analyses were preformed by collaborators at the following U.S Geological Survey branches: National Research Program Boulder, CO; National Water Quality Laboratory Denver, CO; Kansas Water Science Center, Lawrence, KS; and California Water Science Center, Sacramento, CA. The framework of this project and field logistics were organized by Dr. Larry Barber and Dr. Alan Vajda. The research activities described here have been reviewed and approved by the University of Colorado Institutional Animal Care and Use Committee, Protocol #92514(05)1E.
vi


TABLE OF CONTENTS
Chapter
1. A LANDSCAPE FRAMEWORK FOR ASSESSING THE BIOLOGICAL IMPACTS
OF CHEMICALS OF EMERGING CONCERN...................................... 1
1.1 INTRODUCTION 1
1.1.1 Chemicals of emerging concern..................................1
1.1.2 Endocrine disruption...........................................1
1.1.3 Mixture effects................................................2
1.1.4 The landscape framework to address complex mixture effects.....3
1.2 RELEVANT BACKGROUND 4
1.2.1 Our chemical society...........................................4
1.2.2 Emerging concern over impacts..................................5
1.2.3 The Endocrine disruptor hypothesis.............................6
1.2.4 Complex mixture models.........................................7
1.2.5 Regulatory and conceptual frameworks..........................10
1.2.6 Transcriptomics to study exposure to CECs.....................12
2. MOBILE LABORATORY EXPOSURE AND TRANSCRIPTOME PROFILING... 16
2.1 INTRODUCTION 16
2.1.1 The Shenandoah Valley.........................................16
2.1.2 Research objectives...........................................17
2.2 Methods 19
2.2.1 Site descriptions.............................................19
vii


2.2.2 Mobile laboratory exposure and sampling..............................21
2.2.3 Water sampling & chemical profiling...................................22
2.2.4 RNA extraction & microarray analysis..................................22
2.2.5 Statistical analysis & bioinformatics.................................24
2.3 RESULTS AND DISSCUSSION 26
2.3.1 Chemical profiles.....................................................26
2.3.2 Morphological endpoints & Vitellogenin................................28
2.3.3 Differentially expressed genes........................................29
2.3.4 Differentially expressed pathways in the liver of fathead minnows.....33
2.3.5 Hierarchical clustering...............................................37
2.4 CONCLUSIONS 38
References 64
viii


LIST OF TABLES
Table
1. Genes selected a priori for analysis based on known responses to 57
environmental contaminants
2. Chemicals detected in the Shenandoah Valley, August-Sept. 2014 58
3. Number of differentially expressed genes observed at each 61
exposure site, FDR corrected
4. Highly differentially regulated pathways identified by Gene-Set 62
Enrichment Analysis at each of the three locations
5. Highly up- and down-regulated sub-networks of genes as 63
identified by sub-network enrichment analysis
IX


41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
LIST OF FIGURES
Map of the Shenandoah River watershed including locations of mobile laboratory deployment
Nutrient loadings from four locations in the Shenandoah Valley Pesticides detected in the Shenandoah Valley
Halogenated disinfection byproducts detected at site in the Shenandoah Valley
Hierarchical clustering of water samples for each location based on binary presence/absence data
Survivorship of fish exposed for 21-days
Gonadosomatic index of fish exposed to waters from the Shenandoah watershed
Number of nuptial tubercles for fish exposed to waters from the Shenandoah watershed
Relative concentration of plasma vitellogenin in fish exposed to waters from the Shenandoah watershed
Venn diagram of differentially expressed genes
Expression of transcripts related to reproduction in the liver
Expression of transcripts involved in the aryl hydrocarbon receptor pathway
Schematic of the mevalonate pathway
A schematic diagram of the IL6/STAT signal transduction pathway
A schematic diagram of the classical complement pathway
Heatmap and hierarchical clustering of differentially expressed genes


ABBREVIATIONS
AhR Aryl hydrocarbon receptor
AOP Adverse outcome pathway
AR Androgen receptor
CEC Chemical of emerging concern
CERC Columbia Environmental Research Center
EDA Effects directed analysis
DDT Di chi orodipheny ltri chi oroethane
DEG Differentially expressed gene
DBP Disinfection byproduct
EDC Endocrine disrupting chemical
EE2 17a-ethinylestradiol
ER Estrogen receptor
FHM Fathead minnow
GSEA Gene set enrichment analysis
NRWD North River wastewater downstream
NRWU North River wastewater upstream
PCFH Passage Creek Fish Hatchery
PCB Polychlorinated biphenyl
SFSR South Fork Shenandoah River
SNEA Sub-network enrichment analysis
TIE Toxicity identification evaluations
WWTP Wastewater treatment plant


1. A landscape framework for assessing the biological impacts of chemicals of emerging concern
1.1 INTRODUCTION
1.1.1 Chemicals of emerging concern
Over the past decades, a wide variety of anthropogenic chemicals have been identified in surface waters (Focazio et al., 2008; Kolpin et al., 2002). These unregulated and unmonitored chemicals are referred to as contaminants of emerging concern [CECs] and include pharmaceuticals (Daughton, 2001), pesticides (Kolpin et al., 2013), exogenous hormones (Lange et al., 2002; Ternes et al., 1999), and other industrial chemicals (Blackburn and Waldock, 1995; Fromme et al., 2002; Suja et al., 2009).
Several emerging contaminants have been shown to cause adverse biological outcomes in aquatic organisms at concentrations frequently found in the environment. Adverse outcomes associated with CECs include abnormal growth and development (Hayes et al., 2010; Jobling et al., 1998), altered behavior, increased mortality, and reduced fecundity (Kidd et al., 2007; Schwindt et al., 2014). These effects can lead to reduced fitness in individuals and ultimately to population declines or extinction (Kidd et al., 2007).
1.1.2 Endocrine disruption
Endocrine disruption is a common mechanism through which many CECs affect organisms. Virtually all aspects of vertebrate physiology are modulated by endogenous hormones that function at very low concentrations (Norris and Carr, 2006). Due to these low effective concentrations, endocrine signaling is also susceptible to disruption from low concentrations of exogenous molecules, known as endocrine disrupting chemicals
1


[EDCs] (Crain and Guillette, 2000). The effects of endocrine disruption are often sub-lethal and can result in adverse outcomes not monitored by traditional screening programs. Thus, it is important to design CEC studies to explicitly test for endocrine disruption and other subtle adverse effects.
1.1.3 Mixture effects
A significant body of research documents the effects of exposure to environmentally relevant concentrations of CECs in laboratory settings (reviewed in Paskova et ah, 2011; Santos et ah, 2010). However, wildlife encounter CECs as components of complex mixtures that vary in time and space. The effects of exposure to these mixtures may be different than the effects of exposure to a single constituent. Additive or antagonistic effects of exposure to CECs have been demonstrated and synergistic effects from complex mixtures are possible (Altenburger et al., 2012; Kortenkamp, 2007). Adding to this complexity, mixture constituents have multiple origins including point and non-point sources (Daughton, 2001; Neumann et al., 2002). Point sources include effluents from municipal wastewater treatment plants [WWTPs] and industrial operations. Non-point sources include runoff from agricultural fields and urban areas. A better understanding of the biological consequences propagated by complex mixtures of CECs and the pathways through which they enter the environment is needed for effective regulation and management.
2


1.1.4 The landscape framework to address complex mixture effects
To address this environmental complexity, we have developed a landscape-based framework for linking complex chemical mixtures to adverse outcomes across multiple levels of biological organization. At the core of this framework is the hypothesis that the specific chemical mixtures present in a given body of water are associated with the landscape patterning of the watershed. We reason that chemicals are used in specific applications and therefore, enter the environment through non-random pathways that correspond to these applications. By characterizing the landscape of a watershed, it might be possible to probabilistically predict the chemicals present in downstream waters. We can extend this association a step further to the biological effects of exposure. Physiological responses in organisms will also occur in non-random ways based on the composition of the contaminant mixture in their environment. Given a baseline understanding of the biological effects of exposure to various contaminant mixtures, it should be possible to predict biological outcomes from a contamination profile. Thus, our overarching hypothesis is that landscape patterning, watershed contamination, and biological effects are related and predictive models can be developed using these three variables. However, a knowledge gap exists in understanding the biological effects caused by exposure to complex environmental mixtures.
Transcriptome profiling is a promising tool for characterizing the biological consequences of exposure to CECs (Ankley and Daston, 2006; Villeneuve et al., 2012). Transcriptome profiling provides information on a wide range of physiological process that are affected by exposure to chemicals. Bioinformatic tools are available to analyze transcription changes in thousands of genes and to identify cellular pathways and
3


biological processes that are affected (Mehinto et al., 2012; Subramanian et al., 2005). These data are useful for three purposes: 1) examining patterns in biological responses between exposure sites, thus serving as an initial exploration of the landscape framework, 2) comparing an observed transcriptomic response to existing datasets involving simple exposures and well-defined adverse outcome pathways and, 3) generating hypotheses regarding the mechanisms of action involved in exposure to contaminant mixtures.
1.2 RELEVANT BACKGROUND
1.2.1 Our chemical society
The large-scale production of chemicals can be traced back to the industrial revolution in Europe. Indeed, the ability to synthesize large volumes of specific molecules was integral in the development of a modern, technological society. Following World War II there was an exponential increase in the number and quantity of synthetic chemicals used in virtually all aspects of society. Currently, the Chemical Abstract Service registry contains over 100 million unique chemicals that have been discovered or synthesized. Of these, it is uncertain how many are currently used in commercial applications but estimates range from about 25,000 to 84,000 in the U.S (IOM, 2014). In 2005, the EPA tracked the manufacture or import of over 13 million tons of chemical products and this number excludes fuels, pesticides, pharmaceuticals, and food products (U.S. EPA, 2009; Wilson and Schwarzman, 2009). Given the volume and ubiquity of chemicals in use, ecosystem contamination resulting in human and wildlife exposures is unavoidable.
4


It should be stated that while industrial chemical synthesis is a recent and human-driven phenomenon, organisms have been exposed to exogenous chemicals over evolutionary time scales. For example, plants produce endocrine-active phytochemicals that are capable of interfering with vertebrate reproduction (Vajda and Norris, 2006). Organisms have evolved mechanisms for maintaining homeostasis in the face of a changing chemical environment including xenobiotic receptors capable of inducing expression of metabolizing enzymes (Hahn, 2002). Thus, the risk posed to organisms by anthropogenic chemicals may be less a product of the novelty of these molecules and more a product of the concentration and eco-evolutionary context in which exposure occurs.
1.2.2 Emerging concern over impacts
Concern over the health effects in humans and wildlife from exposure to these chemicals greatly increased following the 1962 publication of Rachel Carsons Silent Spring, in which ecosystem degradation and human health impacts are attributed to the increasing use of pesticides, especially dichlorodiphenyltrichloroethane [DDT], Silent Spring is often credited with catalyzing the modern environmental movement that began around this time. Whatever the cause, growing public concern about chemical contamination led to the formation of the Environmental Protection Agency in 1970 and soon after passage of the Clean Water Act in 1972 that gave the EPA authority to regulate point sources of contamination to surface waters. Around this time several commonly used chemicals, including DDT and polychlorinated biphenyls [PCBs] were banned in the U.S. after it was discovered that they were environmentally persistent, capable of
5


bioaccumulation and magnification, and posed significant toxicity risk to humans and the environment. Leading up to and following the ban of DDT and PCBs it was shown that these chemicals are capable of activating the estrogen receptor [ER] in various vertebrate tissues and interfering with the normal physiological processes mediated by estrogen signaling, providing a mechanistic explanation for the observed adverse reproductive and developmental outcomes in wildlife (Bitman et al., 1968; Gaido et al., 1997).
Since the mid 20th century, it has been known that certain synthetic chemicals can mimic the effects of endogenous estrogens, often with negative biological outcomes. Diethylstilbestrol [DES] was first synthesized in 1938 by Sir Charles Dodd as an estrogen mimic to study hormone action (Gilbert and Epel, 2009). It was widely prescribed as a therapeutic drug until is was linked with uterine cancer and abnormal reproductive morphology in women exposed in utero. Dodd also discovered that bisphenol A [BPA] is estrogenic in 1936, before it was widely incorporated into plastic production (Gilbert and Epel, 2009). BPA exposure has now been linked to many adverse effects including abnormal reproductive morphology and low sperm count in humans and wildlife (vom Saal and Hughes, 2005).
1.2.3 The Endocrine disruptor hypothesis
Toward the end of the 20th century, increasing evidence of reproductive abnormalities in human and wildlife populations led to the creation of the endocrine disruptor hypothesis, which posits that anthropogenic chemicals are negatively affecting organisms through interaction with endogenous hormone signaling. In 1991, a group of scientists gathered at the Wingspread Conference where they coined the term endocrine disruptor
6


and released a consensus statement saying that many chemicals in the environment are capable of interfering with normal endocrine function and that exposure to these chemicals is likely responsible for adverse outcomes in wildlife and humans (reviewed in Colborn and Clement, 1992). Endocrine disruptors were defined as any substance that binds to or blocks binding of a hormone to a receptor, alters hormone synthesis or metabolism, or affects the transport or excretion of an endogenous hormone. In the two decades since the Wingspread Conference an enormous body of research has been produced documenting the endocrine disrupting effects of many chemicals. Much of this research focuses on the disruption of the hypothalamic-pituitary-gonadal [HPG] axis through modulation of estrogen and androgen signaling. It is now known that 17a-ethinylestradiol [EE2], PCBs, alkylphenolic compounds, phthalates, organochlorine pesticides, and bisphenols can interact directly with vertebrate estrogen receptors (Kuiper et al., 1998, 1997; Lemaire et al., 2004; Nakai et al., 1999). Other chemicals are known to indirectly affect endocrine signaling through modulation of hormone synthesis (Cheshenko et al., 2008). Additional research has shown disruption of other hormone networks including glucocorticoid, thyroid, and progestogen signaling (Kumar et al., 2015; Suzuki et al., 2015; Tabb and Blumberg, 2006).
1.2.4 Complex mixture models
It has long been recognized that mixtures of chemicals may have different effects on biological organisms than individual constituents (Berenbaum, 1989). These effects are generally classified as additive, synergistic, or antagonistic. At a basic level, additive effects occur when the response to a mixture is predicted by the sum of the effects
7


expected from individual components. Synergism occurs when the effect of the mixture is greater than expected from the sum of its individual components. Lastly, antagonism occurs when the observed effect of a mixture is less than the sum of expected effects of individual components. The type of mixture effect observed depends on the components of the mixture and their modes of action. Any two mixture constituents can affect a biological response through direct chemical-chemical interactions, through interaction with the same bio-molecule (e.g. two receptor agonists), or interaction with different biomolecules involved in the same process.
Two general reference models have been developed to evaluate mixture effects: the dose or concentration addition model, and the independent action model (Kortenkamp and Altenburger, 1998). The concentration addition model was developed by Loewe and Muischneck for chemicals with similar modes of action. It predicts effects based on an additivity principle. As pointed out elsewhere, additive effects are not simply the sum of individual effects for all mixture constituents at a given concentration, as this would not account for the non-linear dose-response curves observed for most chemicals (Berenbaum, 1989; Kortenkamp and Altenburger, 1999, 1998). Rather, the concentration addition model is based on iso-effective doses or the concentrations at which two chemicals produce the same effect (Berenbaum, 1989). In this way chemicals in mixtures can be thought of as dilutions of each other. Experimental observations can be compared to model predictions to evaluate whether mixtures are producing synergistic, additive, or antagonistic effects. Predictions based on the concentration addition model have proven accurate for several classes of CECs based on experimental observations.
8


The response addition or independent action concept provides another reference model for predicting mixture effects. This model is based on the concept of statistical independence and predicts mixture effects based on the product of probabilities for individual constituents to cause effects (Kortenkamp and Altenburger, 1998; Martin et al., 2009). The independent action model is used for mixtures where the constituents have different mechanisms of action (Spurgeon et al., 2010) and has also been experimentally validated in certain contexts.
Although these models have been shown to generate accurate predictions in certain instances, their use in the study of complex environmental mixtures is limited for several reasons. First and foremost, both models require knowledge of the effects caused by individual chemicals and for many CECs this data is simply not available. The time and cost of individual exposure represents a major logistical challenge given the thousands of chemicals that may be present in the environment. These approaches also require assumptions about the mechanism of action for mixture constituents to choose the appropriate model. Even if dose response data are available for a chemical, its mechanism might be unknown. Additionally, complex mixtures will inevitably contain sub-mixtures that act on the same biomolecule and sub-mixtures that act of different biomolecules involved in the same biological process and no models have been successfully developed for this scenario. Finally, neither of these models account for chemical-chemical interactions. The limitations of predictive modeling have led to calls for different approaches for the study of complex environmental mixtures in recent years. Several recent papers have advocated using omics based technologies to study complex
9


mixture effects at the systems level (Garcia-Reyero and Perkins, 2011; Spurgeon et al., 2010).
1.2.5 Regulatory and conceptual frameworks
Several experimental protocols have been developed in an attempt to overcome the disadvantages of purely mathematical modeling approaches, especially in regulatory and risk-assessment contexts. Following the establishment of the Clean Water Act, the EPA developed the toxicity identification evaluation [TIE] process to enable the identification of specific toxicants from contaminant mixtures (US EPA, 1991). The TIE protocol follows three phases; In phase I, organisms are exposed to whole mixtures to access overall toxicity. Next, whole mixtures are manipulated to remove classes of chemicals and organisms are again exposed to manipulated waters. In this way the range of potential chemicals can be narrowed down to specific classes such as divalent cation metals, nonionic organics, or volatile chemicals (Burgess et al., 2013). The goal of phase II is to identify specific toxicants from within the chemical class identified in phase I, generally by measuring concentrations of candidate chemicals. Phase III is a confirmation step where organisms are exposed to a candidate chemical to test whether that chemical alone is capable of producing the toxicity observed from the mixture.
Effects directed analysis [EDA] is a similar approach proposed for identifying causal toxic constituents from complex mixtures (Brack, 2003; Burgess et al., 2013). EDA also begins with a toxicity test of whole mixtures, but may employ targeted in vivo or molecular assays in lieu of whole organism tests. The second step of EDA involves manipulations to narrow the range of offending chemicals, however, EDA uses more
10


precise chemical fractionation techniques to achieve a narrowed scope of potential chemicals (Brack, 2003). Fractionation is followed by chemical analysis, toxicity testing, and potentially subsequent fractionation. Using fractionation makes it possible to get a much narrower and more precise list of candidate chemicals, whereas the manipulations of TIE are only capable of identifying broader classes. This specificity comes at the expense of ecological relevance, as the fractionation processes can dramatically alter the bioavailability of mixture constituents, thus the observed effects caused by fractions and sub-fractions may not be relevant to the original ecosystem (Burgess et al., 2013).
Both TIE and EDA have been used to successfully identify individual toxicants responsible for adverse outcomes in wild populations (Amato et al., 1992; Brack et al., 1999; Thomas et al., 2001). However, there are limitations to these approaches for characterizing the effects of complex mixtures. Namely, these methods are largely designed to identify single chemicals or classes of chemical that have high acute toxicity. As such, they are useful in heavily polluted sites where acute toxicity is high, but less so for the larger portions of habitat where chronic exposure to CECs occurs and adverse effects can be subtler (Reineke et al., 2002). Furthermore, the fundamental concept behind both TIE and EDA involves reducing the complexity of mixtures. Although this reductionist approach is certainly useful in certain contexts, it ignores mixture effects that originate from chemicals in different classes and non-specific modes of action (Brack, 2003). Incorporating omics data into these approaches increase their applicability to landscape scale investigations of ecotoxicology by enabling measurement of fast acting biomarkers for a wider range of effects, although this adds time and cost to an already resource intensive protocol.
11


To help identify these responsive biomarkers, capitalize on emerging omics data, and facilitate risk-assessment Ankley et al. proposed the adverse outcome pathway [AOP] framework (Ankley et al., 2010). The AOP framework seeks to mechanistically link molecular, cellular and organismal data from chemical exposures to an ecologically meaningful endpoint, the adverse outcome. This approach increases the relevance of commonly and easily measured endpoints and enables the development of new, cost-effective and sensitive biomarkers. An AOP begins with a molecular initiating event where a biomolecule, such as a receptor or enzyme, interacts with a xenobiotic chemical. This interaction triggers a cascade of events that cross levels of biological organization that eventually results in an adverse outcome at the level of the organism, population or ecosystem. The critical steps and thresholds along the pathway can then serve as biomarkers for the outcome itself, enabling for efficient and cost-effective monitoring. This framework has been widely adopted by the ecotoxicology community and moderate to well defined AOPs exist for several classes of contaminants (Perkins et al., 2015). In complex mixture studies, transcriptome profiles can be compared to known AOPs to provide clues regarding the chemical initiators and mechanisms of action propagated by the mixture.
1.2.6 Transcriptomics to study exposure to CECs
Transcriptomic studies, using microarrays and RNA sequencing, have been widely employed to study the effects of exposure to CECs. However, the majority of these studies focus on single chemicals in laboratory settings. A handful of studies have examined the transcriptomic effects of exposure to complex mixtures in the fathead
12


minnow [FHM], Of these, most focus on estrogenic EDCs associated with point source municipal WWTP effluent.
In the first published study using microarrays to measure effects of complex mixture exposure, Garcia-Reyero et al. compared the transcriptome profiles from testis of FHM exposed in wire-cages above and below a WWTP in Minnesota (2008). They identified 119 genes unregulated and 128 genes down regulated in the downstream fish relative to the upstream fish. A hierarchical clustering analysis using DEGs grouped upstream and downstream fish in separate clusters. This pioneering study demonstrated that fish transcriptomes can be altered in specific ways following exposure to complex mixtures. However, the biological interpretation of these changes is limited by the lack of a control or outgroup, e.g. it is not possible to resolve whether genes were downregulated in the downstream cohort or upregulated in the upstream cohort. Clustering based on DEGs between upstream and downstream samples all but guaranteed that individual fish will cluster according to their exposure location, limiting interpretation of this analysis. (Garcia-Reyero et al., 2008).
Subsequent investigations of these WWTP-impacted streams incorporated three additional sites (Garcia-Reyero et al., 2010). The three additional sites included two downstream of WWTPs and one in an agricultural area. Hierarchical clustering of male fish liver and gonad transcriptomes showed that expression patterns were correlated with exposure site, however there was no clustering of sites upstream or downstream of WWTPs, thus it is unclear if transcriptome profiles are responding to differences in chemical contamination or other environmental variables. The above studies do not
13


provide any chemical characterization of the exposure water, limiting the conclusions that can be made in terms of direct CEC effects. (Garcia-Reyero et al., 2010)
Laboratory studies show that exposure of male FHM to WWTP effluent, EE2, or 17f )-trenbolone [TB] alters the transcriptome of the liver and testis (Garcia-Reyero et al., 2010). The transcription profile from fish exposed to WWTP effluent was more similar to that of fish exposed to EE2 than to TB, but ultimately WWTP effluent produced a unique transcription signature. The authors employed a functional enrichment analysis to identify biological processes that were statistically overrepresented among DEGs. Several of the significantly enriched processes in effluent exposed fish are known to be affected by CECs, including hypoxia inducible factor [HIF] signaling, oxidative stress response, and aryl-hydrocarbon receptor [AhR] signaling (Simmons et al., 2009). Again, no data on the chemical composition of the effluent was provided. (Garcia-Reyero et al., 2010)
Another series of related studies measured genome-wide transcription levels in FHM ovaries (Martinovic-Weigelt et al., 2014) and male livers (Beminger et al., 2014) from fish exposed to pure effluent from three different WWTPs, as well as waters above and below the effluent discharge. Two of the locations used flow-through mobile laboratory exposures, while a third site used static renewal exposures. Transcriptome profiles of individuals generally clustered based on exposure site, and the transcriptomes of upstream and downstream exposed fish were more similar to each other than to effluent exposed fish. This suggests that factors other than effluent composition play a role in driving transcriptome patterns in fish below WWTP effluent. However, 43.4% of all DEGs overlapped between effluent and downstream fish, but not in upstream fish, suggesting that effluent exposure does affect transcription to some degree in downstream
14


fish. These studies also measured concentrations of 71 organic contaminants, including steroid hormones, pharmaceuticals, and alkylphenols. Hierarchical clustering of chemical profiles was similar to that of transcriptome profiles in that effluent sites were the most dissimilar and upstream and downstream sites clustered closely together (Martinovic-Weigelt et al., 2014). This is consistent with the hypothesis that exposure to CECs alters transcriptomes in fish.
Taken together, these studies demonstrate the value of transcriptome profiling for complex mixture studies. Clustering of transcriptome profiles largely corresponds with exposure site in all of these cases, demonstrating that the genome-wide profile is responding in non-random ways to site-specific variables. Explicitly accounting for environmental variables in future studies will help identify the subset of transcriptional variation that is due to chemical exposure. Mobile laboratory exposures can remove some of these confounding environmental variables, such as temperature and dissolved oxygen levels. Another point to make is that transcriptomic studies of complex mixtures can greatly benefit from chemical analysis of exposure water, because this allows for direct comparison between chemical and transcriptome profiles, which is necessary for generating hypotheses about specific drivers of transcriptome variation.
15


2. Mobile laboratory exposure and transcriptome profiling
2.1 INTRODUCTION
2.1.1 The Shenandoah Valley
The Potomac River watershed represents a microcosm for investigating our landscape-based framework. The watershed contains a diversity of landuse types, including urban, suburban, agricultural and forested areas. Widespread fish kills involving several species of centrarchid fishes occurred in the South Branch of the Potomac River in 2002 and 2005 (Blazer et al., 2010). In Virginia, major fish kills were observed on the North Fork of the Shenandoah River in 2004 and the South Fork of the Shenandoah in 2005. Lower-level mortality events involving young-of-year fish have occurred frequently from 2002 until at least 2010 (Blazer et al., 2010). Several studies have investigated the conditions associated with these fish kills. An increased frequency of skin lesions was observed in fish from areas with mortality events. Culturing of microorganisms from lesions found a variety of opportunistic bacterial and fungal pathogens. Because these pathogens are also frequently found on healthy fish and because no pathogen was found on all fish or at all sampling locations, it was hypothesized that immunosuppression was contributing to the occurrence of lesions, rather than a single disease epidemic (Blazer et al., 2010). Interestingly, histopathological analysis found a high frequency of intersex fish, i.e. genetically male fish with immature oocytes in the testes (Blazer et al., 2010, 2007). High levels of intersex in populations of fish is generally associated with exposure to estrogenic EDCs. Estrogenic EDCs are frequently found in municipal WWTP effluent and agricultural run-off. An investigation into the association between landscape variables and estrogenic EDCs found that density
16


of animal feeding operations and the proportion of agriculture were the best predictors of estrogenicity in downstream waters, although these variables only accounted for a fraction of total variance (Young et al., 2014). Interestingly, there was no difference in the prevalence of intersex or vitellogenin levels in males between fish caught upstream and downstream of WWTPs (Iwanowicz et al., 2009). Taken as a whole, these data suggest that multiple causes contribute to the observed mortality events with CECs serving as a potential modulator of fish health.
2.1.2 Research objectives
To characterize the effects of exposure to complex environmental mixtures and investigate linkages between landscape variables and contamination, we developed an integrated site assessment protocol. This protocol combines multiple landscape, chemical, and biological analyses at each sampling location. The measurements span multiple levels of biological organization including gross anatomical endpoints, tissue histology, and protein and transcript abundance. For this thesis, I will focus on on the analysis of genome-wide transcription data in support of the following objectives.
Objective 1: Measure genome-wide transcription levels in FHM exposed to different environmental source waters and identify differentially expressed genes.
Transcriptome profiles were generated for individuals at each exposure location using 60,000 feature FHM microarrays. Individual genes that are affect by exposure can be identified by comparing exposed fish to a control or reference cohort. This first level of analysis is useful for identifying transcripts that show the highest response in relative fold
17


change, and to generate hypotheses regarding organismal implications and candidate chemical drivers.
Objective 2: Determine the response of genes relevant to detoxification and endocrine disruption. In addition to a whole-genome analysis for DEGs, I employed a hypothesis-driven approach to identify non-random changes in transcription for select genes with known sensitivity to CEC exposure. Such genes include hormone receptors, xenobiotic receptors, and downstream targets of receptor response elements [Table 1],
This approach allows for pairwise comparisons between all groups, which would be intractable for all 60,000 microarray features. These data will be useful for evaluating the impacts of commonly studied classes of CECs and legacy contaminants at these locations.
Objective 3: Identify biological processes and molecular pathways differentially affected by specific exposures. Statistical tests have been developed for identifying groups of genes that are disproportionately affected in an omics data set. Identifying molecular pathways or processes that are altered by exposure provides a greater biological context for interpreting transcriptome data. Pathway analysis is also helpful in identifying subtle changes in gene expression that may not be significant in a DEG analysis due to small effect size, yet may have important biological implications. For example, if an individual transcript is up-regulated by 50% or less, it is unlikely to meet the significance threshold of a hypothesis test, especially when corrections for multiple comparisons are used. However, if that gene and several related genes are all upregulated by 50%, the actual biological effect on an organism could be substantial.
18


Objective 4: Use hierarchical clustering to explore patterns in transcription and chemical profiles at the different locations. If our landscape hypothesis is correct, then changes in gene transcription and chemical profiles are expected to correlate with exposure site. Hierarchical clustering can be used to measure correlations between individuals based on many response variables. Significant clustering of transcription profiles based on exposure site will indicate a similarity in the transcription response to exposure. Clustering of chemical profiles by site will indicate that each location had a relatively consistent and unique chemical fingerprint. This clustering would represent a proof of concept for the landscape framework.
2.2 Methods
2.2.1 Site descriptions
Four locations within the greater Shenandoah River watershed were selected to investigate linkages between landscape patterning, chemical contamination, and biological impacts. The locations were chosen to capture unique and representative landuse characteristics in the surrounding watersheds. The Shenandoah River watershed has a diversity of land uses (agricultural, municipal, domestic, industrial, forested, and recreational) in a relatively small geographic area (1372 mi2). Logistical requirements such as road access and electricity prevented a random sampling design. The Shenandoah River watershed is composed of two forks, the North Fork and the South Fork, that flow northward until converging near Front Royal, VA [Fig 1], The North River is a tributary
19


to the South Fork and Passage Creek is a tributary to the North Fork. Landscape characteristics are as follows:
North River Wastewater Upstream [NRWU]. Two sites were selected on the North River that occurred upstream and downstream of municipal waste water treatment plant [WWTP], The watershed of the upper North River site is dominated by intensive agriculture including row-crop farming and confined animal feeding operations. Thus, the NRWU site was designated as the agricultural landuse site.
North River Wastewater Downstream [NRWD]. The Harrisonburg Rockingham Regional Sewer Authority operates a WWTP near Mount Crawford, VA. This plant processes the municipal wastewater of approx. 100,000 residents as well as several industrial operations. The average daily discharge is 13-14 million GPD. The water intake for this downstream site was placed in the North River directly below the outfall of the plant to capture a strong chemical signature from municipal wastewater.
South Fork Shenandoah River State Park [SFSR]. The third site was located at the Andy Guest Shenandoah River State Park, near Bentonville, VA. This is the largest watershed selected and it contains a variety of landuse types, including forested areas, agriculture, and small municipalities. This location was designated the mixed use site.
Passage Creek Fish Hatchery [PCFH]. Passage Creek is a small tributary to the North Fork of the Shenandoah River. Its watershed is largely forested within the George Washington National Forest. The watershed also contains some rural
20


development as well as dispersed agriculture. This location was designated the reference site, as it was expected to show the smallest anthropogenic chemical impact.
2.2.2 Mobile laboratory exposure and sampling
Mobile laboratories were deployed to the four locations described above. Water from each site was pumped from the main channel of each river into a head tank on top of each laboratory. From the head tank, water was fed into one of two splitter tanks, where it was heated to 25 1 C and then flowed into individual 5 gallon glass aquaria. Water in aquaria was aerated with diffusion bubblers, and the photoperiod was maintained at 14hr light, lOhr dark. Flow rates to individual aquaria were maintained at -200 mL/min to provide replacement of 95% of the water volume approximately every 4 hours. Materials that contact exposure waters were limited to stainless steel, glass, or Teflon tubing.
Reproductively recrudescent 12-month-old male FHM were provided by the Columbia Environmental Research Center [(CERC); Columbia, MO], Fish were randomly assigned to exposure sites and to aquaria within sites. Five individuals were placed in each aquarium. Exposure began on August 13th, 2014 and continued for 21 days until September 2nd, 2014. Fish were fed laboratory fish chow provided by CERC.
Fish were sampled on days 0, 7 and 21. The initial control cohort was sampled on day 0, without being exposed to environmental waters. Twenty male fish were sampled on day 0 and 10 male fish were sampled from each site on both day 7 and day 21. Fish were anesthetized with tricaine methanesulfonate (MS-222). Gross anatomical characteristics were measured including mass, maximum total length, nuptial tubercle number and
21


prominence, liver mass, gonad mass, and dorsal fat pad prominence (Jensen et al., 2001). Condition factor was calculated as (mass/length)*100. Gonadal somatic index [GSI] was calculated as (gonad mass/total mass)* 100. Blood was collected from the caudal vein in heparinized capillary tubes. Plasma was isolated via centrifugation and stored at '80C until it was analyzed for vitellogenin protein by enzyme-linked immunosorbent assay (ELISA) using an anti-FHM kit (Biosense; Bergen, Norway). Livers were dissected, weighed, and divided into three fragments. Individual tissue fragments were preserved in 10% neutral-buffered formalin, RNA/ater (ThermoFisher Scientific), or snap frozen on dry ice.
2.2.3 Water sampling & chemical profiling
To characterize the chemical profile of the exposure waters, grab samples were collected every 7 days during the fish exposures. Grab samples were collected inside the mobile laboratories from water flowing out of the head tank. Multiple preservation methods were used from each sampling including acidification and filtering. Samples were placed on ice and shipped to 5 different analytical laboratories for analysis. A total of 460 chemical constituents, including parent compounds and metabolites, were measured for each sample using 17 different analytical methods. Chemical analytes included 109 pharmaceuticals, 48 trace elements, 29 halogenated disinfection byproducts, 43 hormones and phytoestrogens and 135 pesticides.
2.2.4 RNA extraction & microarray analysis
For the 7-day exposure cohort, total RNA was extracted from liver tissues preserved
22


in RNAlater using Qiagen RNeasy plus mini kits according to the manufactures instructions. RNA purity and concentration were measured on a NanoDrop ND-1000 UV-Vis Spectrophotometer. All samples had an A260/A280 ratio greater than 2.0. RNA quality was measured with an Agilent 2100 Bioanalyzer. Average RNA integrity numbers [RINs] were 9.48 0.25 (mean SD). All RIN values were above 8.9.
A sum-sample of 31 RNA extracts (N = 7 or 8 /site) were prepared for microarray hybridization using Agilent One-color Low Input Quick Amp Labeling kits according to the manufacturers instructions (Agilent Technologies v6.9.1, Santa Clara, CA). Briefly, samples were spiked with Agilent One-Color Spike-Mix, diluted by a factor of 10,000. Ploy-T, T7 primers were annealed to samples by incubating at 65 C for 15 minutes. cDNA was synthesized from sample mRNA using AffinityScript reverse transcriptase. Samples were incubated at 40C for 2 hours immediately followed by 70C for 15 minutes. Cyanine 3-labeled cRNA was synthesized from the cDNA templates by in vitro transcription using T7 RNA polymerase in the presence of cyanine-3-cytosine triphosphate. Samples were incubated at 40C for 2 hours. Amplified and labeled cRNA was then purified using a modified Qiagen RNeasy protocol. Following purification, samples were tested for purity, nucleic acid concentration, and Cyanine-3 concentration on a NanoDrop ND-1000 UV-Vis Spectrophotometer. All samples had a specific activity (nmol cyanine-3/ug RNA) greater than 7.
The FHM 8 x 60k oligonucleotide microarray slides (GPL15775) were manufactured by Agilent Technologies. Each array contained 61,718 individual probes. These probes contained 22,010 unique gene annotations. Six-hundred nanograms of labeled RNA was used form each sample for hybridization. Arrays were allowed to hybridize at 65C for
23


17 hours, then washed according to the manufacturers instructions. Microarrays were scanned at 5 pm with the Agilent G2505 C Microarray Scanner and Agilent Feature Extraction Software vlO. 1.1.1 was used to extract raw signal intensities from microarray images. All microarray data were within acceptable quality control parameters. Raw signal intensities were imported into JMP Genomics v6.0 (SAS Institute Inc., Cary, NC) quantile normalized, and log2 transformed. The intensity of spike-in and dark spot Agilent controls was used to determine the lower limit of detection for this assay. It was determined that 2.7 was the optimal lower bound and all values less than 2.7 were assigned this value.
2.2.5 Statistical analysis & bioinformatics
A one-way analysis of variance was used to identify differentially expressed genes [DEGs] in each treatment group. A false discovery rate of 5% was used to correct for multiple testing. Fish from NRWU, NRWD and SFSR were compared to the reference PCFH. Probes with p < 0.05 following FDR correction were considered differentially expressed. ANOVA was performed in JMP Genomics (V6.0).
A subset of genes was selected a priori for focused analysis based on their relevance to environmental contamination studies. Genes selected for focused analysis were categorized as reproduction related, or xenobiotic responsive [Table 1], Signal intensities for all probes assigned to a given gene were aggregated and mean intensity was calculated. A bootstrapped resampling of the means was used to generate an empirical F-statistic distribution based on 5000 iterations. The observed F-statistic was
24


compared to the empirical distribution to calculate a p-value. For genes with p-values < 0.05, pairwise comparisons were conducted using Tukeys HSD.
Gene set enrichment analysis [GSEA] and sub-network enrichment analysis [SNEA] were performed in Pathway Studio 9.0 (Ariadne, Rockville, MD) using the ResNet 10.0 database (Nikitin et al., 2003). A total of 37,169 microarray probes (60%) were mapped to mammalian homologs. GSEA was performed on gene sets in the categories cell signaling, receptor signaling, cell process pathways, and metabolic pathways. Each category contains groups of genes known to be involved in a molecular pathway. The GSEA algorithm uses a modified Kolmogorov-Smirnov test to identify pathways that are significantly up- or down-regulated as a whole (Subramanian et al., 2005). SNEA was conducted using cellular processes as the seeds for sub-network construction. The SNEA algorithm then creates de novo sub networks based on database information linking genes to a seed cellular process and identifies sub networks that are up- or down-regulated in experimental groups (Mehinto et al., 2012).
A complete-linkage hierarchical clustering analysis using Euclidean distance was performed to identify patterns in expression profiles. For clustering, the dataset was reduced to probes that had been identified as differentially expressed in one or more treatment groups. Probes that were annotated to the same gene symbol were averaged. An additional cluster analysis was performed on the chemical profiles using binary presence/absence data and Jaccard distance. Hierarchical clustering and bootstrapped ANOVA were performed in R (R Core Team, 2016).
25


2.3 RESULTS AND DISSCUSSION
2.3.1 Chemical profiles
A total of 149 of 460 chemical analytes were detected at one or more sites. This includes 47 trace elements, 20 neutral or acidic organic contaminants, 9 hormones, 11 pesticides, 10 halogenated disinfection byproducts, 27 pharmaceuticals, and 3 antibiotics [Table 2], A total of 70 chemicals were detected at PCFH, 77 at NRWU, 91 SFSR, and 116 at NRWD. The majority of CECs detected were found at concentrations in the low ng/L range. The chemical data is generally in line with our characterization of the landuse in surrounding watersheds, with a few notable exceptions. The fewest chemicals were detected at the PCFH reference site. Of 69 total detects at PCFH, excluding nutrients, 61 were found at all 4 sites and the majority of these were trace elements, phytoestrogens and hormones. Several CECs were detected at this site including the pesticides fipronil, metalaxyl and simazine, as well as the pharmaceuticals metformin and diphenhydramine. PCFH also had the lowest concentration of nutrients, including nitrate, nitrite, ammonia and phosphorous [Fig. 2], These findings are consistent with a largely forested watershed with limited agriculture and residences.
At the two North River locations, surprisingly, more pesticides were found at NRWD than at NRWU, with 8 and 6 detections, respectively [Fig. 3], Four pesticides were detected at SFSR. Atrazine, metolachlor, and simazine occurred at similar concentrations between NRWU and NRWD, whereas concentrations of imidacloprid and azoxystrobin were higher at NRWD. Boscalid, diuron, and fluxapyroxade were detected only at NRWD. Based on the highly agricultural landuse patterns in the NRWU watershed, we expected to find more pesticides at NRWU than at the other locations. It is possible that
26


some of the pesticide load detected at NRWD originated from upstream and not WWTP effluent, however this cannot explain the detection of chemicals at NRWD that were not present at NRWU. In addition, this phenomenon is unlikely to account for the total concentration of chemicals detected at both sites because upstream waters were heavily diluted with WWTP effluent, and thus concentrations measured at NRWD should be lower than NRWU if the only source of pesticide was from upstream. Therefore, it is reasonable to conclude that the WWTP effluent per se is the source of much of the pesticide load observed at NRWD. Previous studies have also detected numerous pesticides in WWTP effluent in Europe (Kock-Schulmeyer et al., 2013; Kuster et al., 2008). The HRRSA-WWTP influent includes wastewater from an industrial facility that processes agricultural products. This facility is a likely source of some of the pesticides observed at NRWD. In addition, pesticides may enter the WWTP influent through urban run-off from lawns and gardens or from human waste (Hill et al., 1995).
NRWD had the most pharmaceutical detections at 22, while SFSR had 17, and NRWU had 6. This is consistent with our expectation that WWTP effluent would contain the highest number of pharmaceuticals. The high number of detections at SFSR indicates that this location may be impacted by upstream WWTP effluents which is consistent with the mixed-use character of the watershed.
Of 29 halogenated disinfection byproducts [DBPs] tested for, 10 were detected at NRWD [Fig 4], No DBPs were found at the other sites, with the exception of 1 detection of trichloromethane at PCFH. This indicates that NRWD was indeed highly impact by the nearby WWTP effluent. The fact that no DBPs were detected at NRWU confirms that this location was outside the upstream mixing zone of the WWTP effluent and therefore,
27


contaminants detected at this location originated from upstream and not from the nearby but downstream effluent.
Hierarchical clustering of chemical profiles based on binary presence/absence data showed samples from each site clustering together [Fig. 5], demonstrating that each location had a unique chemical profile that was driven more by the specific location rather than by sampling time. Additional sampling of other watersheds with similar landuse profiles is needed to evaluate the degree to which contaminant profiles can be predicted from landscape variables. NRWD samples were the most unique relative to the other locations. This is likely due to the presence of numerous DBPs detected only at this site. The reference PCFH profile was unique compared to NRWU and SFSR, which clustered together based on chemical composition. This indicates there is more similarity in contaminant profile between our agricultural and mixed-use watersheds than between the reference watershed and either of these two. Overall, the chemical data provides proof of concept for our landscape framework. The watersheds we selected based on landuse each had unique contaminant profiles that were often consistent across our 4-week sampling period. There were several deviations from our predictions, such as the high prevalence of pesticides at NRWD. This may reflect the integrated nature of agricultural and industrial systems, which will present a unique challenge for linking landuse to river contamination.
2.3.2 Morphological endpoints & Vitellogenin
Survivorship was decreased at the NRWU location, where 52% of individuals died during the course of the 21 exposure (ANOVA, p > 0.05) [Fig. 6], Survivorship was high
28


(>90%) at all other sites. For the 7-day cohort there were no differences in GSI, nuptial tubercle abundance or plasma VTG (p > 0.05) [Fig. 7 & 8], GSI and nuptial tubercle abundance were significantly reduced at NRWU and NRWD. The magnitude of reduction was greater at NRWD for both GSI and nuptial tubercles. This reduction in GSI and secondary sex characteristics is consistent with previous studies of WWTP effluent (Barber et al., 2012; Vajda et al., 2011). Plasma VTG was significantly elevated at PCFH [Fig. 9], although the magnitude of this change was relatively small compared to other studies of endocrine disruption (see 2.3.3 below). Decreases in GSI and nuptial tubercles are often associated with exposure to EDCs, in many cases estrogenic substances (Vajda et al., 2011). However, the lack of a corresponding VTG induction and low concentrations of measured estrogens at these locations makes an estrogenic mechanism of action unlikely. Reduction in nuptial tubercle abundance can also be caused by antiandrogens (Panter et al., 2004).
2.3.3 Differentially expressed genes
Individual genes were identified as up- or downregulated using PCFH as the reference. The number of DEGs was similar between all sites. Fish exposed to waters at NRWU displayed 202 DEGs, whereas fish exposed at NRWD displayed 297 DEGs, and fish exposed at SFSR had 219 DEGs [Table 3], At each site, the largest number of DEGs were only identified as differentially expressed in that location, suggesting that each exposure site had a unique transcriptional response to exposure [Fig. 10], The NRWU and NRWD sites had the most DEGs in common at 67. The high level of overlap between DEGs at the two North River Sites likely reflects the environmental similarity
29


between these locations. This overlap could be caused by a response to CECs or other environmental variables that were similar between the two sites.
Given the number of DEGs at each site it is not possible or meaningful to discuss all of the individual genes that were affected by exposure. Therefore, I will limit reporting and discussion to the genes with the largest fold-change relative to the reference and genes that have known responses to CECs and thus were hypothesized to change a priori.
At the mixed-use SFSR site, the three most up-regulated genes in terms of fold-change were D site of albumin promoter (albumin D-box) binding protein b [dbpb], period 1 [perl], and nuclear receptor subfamily 1 group D member 1 [nr Id I], The three most down-regulated genes were: nuclear factor, interleukin 3-regulated, member 5 [nfil3-5]; protein arginine methyltransferase 6 [prmt6\; and ankyrin repeat domain 10a [ankrdlOa], At the agricultural NRWU site the most up-regulated genes were potassium voltage-gated channel subfamily E member 4 \kcne4\ EGL nine homolog 3 [egln3\, and nuclear receptor subfamily 1 group D member 1 [nrldl\. The most down-regulated genes were myelocytomatosis oncogene homolog [mycn\, transmembrane protein 80 [tmem80\ and DNA cross-link repair 1A [dclrela\. At the WWTP impacted NRWD site the most up-regulated genes were heat shock protein B8 [hspb8], potassium voltage-gated channel subfamily E member 4 [kcne4], and solute carrier family 16, member 3 [slcl6a3\. The most down-regulated genes were hydroxysteroid dehydroxenase 4 \hsdl7b4\
Cytochrome P450 Family 26 Subfamily B Polypeptide 1 [cyp26bl\ and alanine-glyoxylate aminotransferase b [agxt\.
It is interesting to note that, among these highly regulated DEGs, there is overlap between locations. All three of the highest up-regulated genes at SFSR {dbpb, perl,
30


nrldl) were also identified as differentially expressed at NRWU and NRWD. This overlap could represent a similar response to the CECs that were present at all three locations. Alternatively, this might be a consequence of similar nutrient levels at these sites or other environmental factors that are shared at these locations but not at PCFH.
Activating transcription factor 3 \atf3\ was highly up-regulated at NRWD. A if3 is a stress-responsive gene that is up-regulated following a variety of physical and chemical stressors (Hai et al., 1999). Its immediate role in the stress response is not fully known.
Of the chemicals detected at NRWD, acetaminophen (Stamper et al., 2015), carbamazepine (Schulpen et al., 2015), metformin (Limonciel et al., 2014), and nicotine (Malpass et al., 2014) have been shown to up-regulate the expression of atf3. Nicotine was detected at all 4 sites and concentrations were lower at NRWD compared to the other 3 locations. Therefore, nicotine is unlikely to be responsible for the up-regulation of this transcript which was only detected at NRWD. Carbamazepine is an anti-convulsant pharmaceutical prescribed to patients with epilepsy and is commonly detected in WWTP effluents. Carbamazepine significantly increased the expression of atf3 after 1 and 7 days in human stem cells (Schulpen et al., 2015). Metformin is a pharmaceutical prescribed to patients with type-2 diabetes. It was detected at all sites, but in the highest concentration at NRWD. Atf3 was up-regulated in primary rat hepatocytes after exposure to metformin (Limonciel et al., 2014). Acetaminophen is an over-the-counter analgesic and antipyretic medication which up-regulated the transcription of atf3 in-vitro (Stamper et al., 2015). Acetaminophen was only detected at NRWD. Interestingly, ATF3 can bind the androgen receptor [AR] and inhibit androgen signaling, thus it is possible that the increase in at/3
31


observed in NRWD fish may be involved in the decease of nuptial tubercles seen in fish exposed for 21-days.
Heat shock protein a8 [hspa8, a.k.a. hcs70] was also highly induced at NRWD.
Hspa8 is generally considered to be a constitutively expressed member of the heat-shock family, unlike the highly stress inducible hspala and hspalb (Chen et al., 2006; Liu et al., 2012). Of the chemicals detected at NRWD, acetaminophen has been shown to up-regulate the expression of hspa8 in human liver slices (Elferink et al., 2011).
Among the transcripts related to reproduction, only vtgl displayed differential expression following exposure [Fig. 11], Vtg was upregulated 2.4 fold at PCFH compared to SFSR (p= 0.0492) and this is consistent with the VTG protein data.. The fold change in vtg expression is relatively small compared to previous studies of estrogenic endocrine disruption at WWTPs (Bahamonde et al., 2015; Garcia-Reyero et al., 2010). It is not clear whether this finding represents vitellogenin induction at the PCFH site or is part of a normal background fluctuation in the levels of vtg transcript in male FHM. Following careful examination of these genes as a whole, there is little evidence of reproductive endocrine disruption at these sites. This finding is especially surprising for NRWD which was heavily impacted by WWTP effluent, which often contains EDCs (Barber et al.,
2011; Vajda et al., 2011). These results are consistent with a yeast estrogen screen of water samples which found no estrogenicity above the detection limit of 0.025 ng/L E2 (Iwanowicz, pers. comm.). Furthermore, this finding is consistent with the observed concentrations of known EDCs from water samples at these sites. The transcript and protein VTG data indicate the reduction in nuptial tubercles at NRWU and NRWD was
32


not caused by estrogenic EDCs, but may be attributable to non-estrogenic endocrine disruption.
Among transcripts involved in aryl-hydrocarbon receptor [AhR] signaling, cytochrome p450 lal \cyplal] was upregulated 2.8 fold atNRWD compared to the reference PCFH (p < 0.001) [Fig. 12], No significant differences were observed between sites in other AhR pathway genes including ahrl, ahr2, arntl, arnt2, and cyplbl. Cyplal, also known as aryl hydrocarbon hydroxylase, is an enzyme involved in the catabolism of certain xenobiotic organic molecules. Its expression is induced by the AhR/ARNT complex following binding between AhR and a ligand. Tetrachlodibenzodioxin [TCDD] is the most well known and commonly studied AhR ligand, although a number of other emerging and legacy contaminants are known to be AhR agonists (Denison and Nagy, 2003). Of the chemicals detected only atNRWD, the pesticide diuron is known to increase the expression of cyplal in mammals (Ihlaseh et al., 2011). Other chemicals detected at NRWD and other sites are also known to increase the expression of cyplal including 4-nonylphenol (Lee et al., 2005), and carbamazepine (Oscarson et al., 2006).
2.3.4 Differentially expressed pathways in the liver offathead minnows
The GSEA analysis identified multiple differentially expressed pathways at each exposure site. There were 54 affected pathways at SFSR, and 58 affected pathways at both NRWD andNRWU [Table 4], SNEA analysis identified 130, 138, and 113 subnetworks that were significantly affected by exposure at NRWD, NRWU, and SFSR, respectively [Table 5], Altered pathways and sub-networks in general included carbohydrate, glucose, and sterol metabolism, as well as pathways related to immune
33


function. Below, the discussion focuses on those related to metabolism and the immune
system.
At both NRWD and NRWU, the two most down-regulated pathways were biosynthesis of cholesterol and the mevalonate pathway, a precursor to cholesterol synthesis [Fig. 13], This finding was corroborated by SNEA, which identified the mevalonate pathway and sterol biosynthesis as down-regulated sub-networks at both of these locations. These same gene sets were up-regulated at SFSR, although with a smaller median fold change compared to that observed at NRWU and NRWD. This finding at NRWD is in contrast to a previous study that identified significant up-regulation of a cholesterol biosynthesis gene-set in the livers of FHM exposed to several WWTP effluents and downstream waters, although in this study, a different enrichment analysis and microarray were used (Martinovic-Weigelt et al., 2014). This discrepancy could be a result of different contaminant exposures, as Martinovic-Weigelt et al. also found evidence of estrogenic endocrine disruption in effluent and effluent imapcted sites, which was not detected in the present study (2014). Garcia-Reyero and colleagues also identified enrichment of the gene ontology categorey cholesterol biosynthetic process in livers of FHM exposed to waters downstream of a WWTP, althought the direction of this change was unclear (2009). A study investigating the transcriptiomic response of fish exposed to river water from highly agricultural areas found down-regulation of the gene ontology category cholesterol biosysnthesis (Sellin Jeffries et al., 2012), suggesting that agricultural chemicals could be involved in the suppression of these pathways and subnetworks at NRWD and NRWU. Pathways related to carbohydrate metabolism were significantly up-regulated at NRWU. Glycogen and glucose metabolism were among the
34


highest up-regulated gene-sets [Table 4], This was corroborated by SNEA which identified glycogen metabolism, glycogenesis, carbohydrate utilization, glucose import and gluconeogenesis as significantly up-regulated. It is becoming clear, through this study and others, that organismal metabolic processes are comprimised by WWTP effluent and there is a significant transcriptional response involved.
Pathways involved in immune function and pathogen resistance were down-regulated at NRWU, including Interleukin-6/STAT [IL6/STAT] signaling (median fold change[fc] = -1.41, NES = -1.59, p = 0.03) [Fig. 14] as well as classical (median fc = -1.40, NES = -1.85 p < 0.001) [Fig. 15] alternative (median fc = -1.40, NES = -1.95, p = 0.004) and lectin-induced complement pathways (median fc = -1.4, NES = -1.98, p < 0.001). IL6/STAT is a signal transduction pathway whereby interlukin-6, a pleiotropic cytokine, binds with a receptor complex, leading to the phosphorylation and dimerization of STAT transcription factors which then regulate the expression of responsive genes (Aaronson and Horvath, 2002). In mammals this pathway mediates several immune related processes including proliferation of B- and T-cells, the acute-phase response and inflammation (Heinrich et al., 1998) and similar functions have been observed in fish (Uribe et al., 2011). The complement system is part of the innate immune response and mediates the recognition and elimination of invading microbes or damaged host cells (Zipfel and Skerka, 2009). In fish, the complement system is thought to play a larger role in immune defense compared to adaptive immune responses. There are three mechanisms that activate the complement cascade, the classical pathway, the alternative pathway, and the lectin-induced pathway. All three of these pathways were down-regulated in NRWU fish. Interestingly, as part of the acute-phase response in mammals, IL6 activates the
35


expression of several complement proteins, including C3, C4, and C9 (Gabay and Kushner, 1999; Steel and Whitehead, 1994). Thus, the down-regulation of IL6/STAT signaling may be mechanistically related to the suppression of complement pathways in the livers of FHM exposed atNRWU. This effect on immunosuppression is an interesting candidate pathway potentially contributing to the increased mortality among NRWU exposed fish.
Several chemicals detected atNRWU are known to down-regulate IL-6 expression and/or secretion based on mammal in vitro assays (Medjakovic et al., 2010). These include the isoflavones biochanin A, formononetin, diadzein and the diadzein metabolite equol, which were all detected either only at or in the highest concentration at NRWU. Biochanin A decreased expression of IL-6 in osetoblastic murine cells following an H202 inflammatory challenge (Lee and Choi, 2005). Biochanin A and equol decreased the secretion of IL-6 from LPS stimulated macrophages (Mueller and Jungbauer, 2008). Diadzein decreased the secretion and transcription of IL-6 in a dose dependent manner in LPS stimulated RAW264.7 cells (Choi et al., 2012). Therefore, these isoflavones serve as candidate causal agents for the observed decrease of the IL-6/STAT pathway atNRWU. Immune processes can also be inhibited by generalized stress responses. Unfortunately, less data are available regarding the interactions between chemicals and whole pathways than there are for individual genes, so this discussion is limited to chemical mediators of IL-6 itself.
The most prevalent source of isoflavones is legumes, especially soy and red clover (Medjakovic et al., 2010). Formononetin and biochanin A are found predominantly in red clover while genistein and diadzein are more abundant in soy (Clarke et al., 2008). Soy is
36


frequently grown in the agricultural NRWU watershed. In 2012, 9,847 acres of soybeans were harvested in Rockingham County, VA, were NRWU is located (USDA NASS, 2012). Red clover is frequently used as a cover crop on agricultural fields, but as such no official statistics are kept at the county level. The fact these chemicals were found predominantly at NRWU suggests a non-point agricultural origin as opposed to a point source WWTP origin.
At the mixed-use SFSR site, pathways related to immune function were the most strongly up-regulated. Several interleukin/STAT pathways in the common y-chain family were increased, including IL2/STAT, IL7/STAT, IL9/STAT, IL15/STAT and IL21/SdTAT. These pathways share the receptor component IL2G also known as the common y-chain. The initiating cytokines and distinguishing receptor components were not on the microarray, thus it is not possible to resolve whether one specific pathway was increased relative to the others or if all these pathways were up-regulated simultaneously. The common y-chain pathways are generally involved in growth and proliferation of immune cells, such as T-, B-, and NK cells (Akdis et al., 2011).
2.3.5 Hierarchical clustering
Hierarchical clustering of differentially expressed genes was used to explore the overall response to exposure and identify patterns between sites [Fig. 16], The first level of clustering separated all the NRWD fish and 3 NRWU fish from the PCFH fish, the SFSR fish and the remaining NRWU fish. All the NRWD fish clustered together with the exception of 1 that clustered with the 3 NRWU fish. All the PCFH fish clustered together, which is perhaps not surprising given that DEGs were identified by comparison to this
37


reference site. The remaining NRWU and SFSR fish divided into two clusters based on transcriptomic response, each containing individuals from both sites. NRWU showed the most variable response to exposure with individuals clustering amongst NRWD and SFSR fish. The NRWD fish had a largely consistent response that showed the largest distance from the majority of the other DEG profiles. The NRWD contaminant profiles also had the largest difference from other sites, thus there is a similar pattern between total contaminant profile and overall transcriptomic response at NRWD. This pattern is also observed to a degree between NRWU and SFSR, where contaminant and transcription profiles from these sites cluster together, although there are differences within cluster patterning. Overall, there are similarities in the clustering of contaminant profiles and the clustering of transcription profiles, suggesting that the unique mixture of CECs in exposure water may be driving part of the total transcriptional response in fish, although this cannot be quantitatively established at this time. Improved systems-level analysis techniques will be needed to formally evaluate linkages between contaminant mixtures and transcriptional response.
2.4 CONCLUSIONS
This study serves as a proof-of-concept for our landscape-based framework. Each site had a unique contaminant profile and exposed fish displayed unique biological responses. Many of the contaminants detected matched expectations based on landuse, such as a high preponderance of pharmaceuticals and disinfection byproducts at the WWTP impacted NRWD, high levels of pesticides at the agricultural NRWU and a moderate level of pesticides and pharmaceuticals at the mixed use SFSR. Deviations from
38


expectations such as the high preponderance of pesticides at NRWD highlight the complexity of chemical occurrence and transport in a highly industrialized society. Additional variables beyond basic landuse should be included in future studies investigating the linkage between landscapes, river contamination and biological effects. Additional data are needed to quantitatively evaluate the power of landscape variables for predicting contamination and associated biological effects.
The use of mobile laboratories greatly reduces the number of confounding variables in environmental mixture studies. Effects in exposed fish can be attributed to differences in water composition, however this does include factors other than chemical contaminants. Differences in the base chemistry of water sources and microbial communities may contribute to observed effects. Measuring or controlling for these variables in future studies will increase our ability to attribute effects to contaminant mixtures per se.
This study further validates the use of transcriptome profiling in investigations of chemical mixtures. Microarray analysis revealed a diverse and unique transcriptional response at each exposure site, demonstrating the opportunity for transcriptome profiling to characterize a range of responses without being limited to a narrow range of biological endpoints established a priori. The genes and pathways identified as differentially expressed in this study provide valuable mechanistic information regarding the effects of exposure to environmental mixtures. For example, transcript data were useful for evaluating the underlying causes of the observed decrease in nuptial tubercles at NRWU and NRWD. This evaluation was facilitated by the well defined AOP for estrogenic initiating events. For adverse outcomes with less defined AOPs, such as the mortality at
39


NRWU, transcriptome profiling provided hypotheses for causal mechanisms. These hypotheses will be useful for characterizing new AOPs. Additionally, the large amount of data available from single-chemical laboratory exposures makes it possible to identify potential chemical initiators. The hypotheses generated regarding systems-level changes can now be evaluated with specific assays. Although the use of transcript and other omics techniques in field settings is increasing, it is still a relatively new area of research. Additional methodologies are needed to more effectively determine quantitative linkages between changes in transcriptome profiles and the constituents of a complex mixture. Development of these techniques will be facilitated by more studies that capture as much chemical and environmental data as possible. These rich datasets can help pave the way to a more predictive ecotoxicological science.
40



Explanation
1 | Watersheds Counties Shenandoah National Park 7 A X if. u ^ / / > i / IMDMICX .' /f O'j
Shenandoah Rivers y Winchester e ^ .
Shenandoah Toxics Sites 2014 /v4 I J
i
PCFH



& *
WWa£.

SFSR
-' i .
Fredericksburg
7>* NRWD
-*. /
?r
Chariot esville
'
_____________________________________________________________________
0 9 18 Miles
IiWiWi"1
0 10 20 Kilomelers
Figure 1. Map of the Shenandoah River watershed including locations of mobile
laboratory deployment.
41


Nutrients in the Shenandoah watershed, August-September 2014
Total P Phosphate Ammonia Nitrite
008 010 008 010 008 010 008 010
900 900 900 900
004 004 . 004 004
^
oce oce oce m m oce
t o -*--r
mg/ 0 00 00 0 a m 00 0 a 00 0
C o TO NRWD NRWU PCFH SFSR Nitrate NRWD NRWU PCFH SFSR Total N NRWD NRWU PCFH SFSR Nitrate + Nitrite NRWD NRWU PCFH SFSR Organic N

(D q o m c - - § -
O O . -J- lO . f 0 lO t
3 -
~
s - s - s - q .
s - s - s -
? - ? - s - A

s - s - s - " t
r
§ - § - § - § -
NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR
Figure 2. Nutrient loadings from four locations in the Shenandoah Valley. The reference PCFH locations had the lowest total nutrient load. The agricultural NRWU and the WWTP impacted NRWD had the highest nutrient loads. Bars represent the mean and points are measurements from individual samples (N = 4/site).
42


Pesticides detected in the Shenandoah watershed, August-September 2014 Atrazine Azoxystrobin Boscalid Fipronil
0 s - 0 0
o
8 - + O 8 - 8 -
a - 8 - 8 -
8 -
/ LO
8 -
8 - § - 8 - 8 -
8 -
o - c: o - A M m o - M o - O *
NRWD NRWU PCFH SFSR Fluxapyroxade NRWD NRWU PCFH SFSR Iprodione NRWD NRWU PCFH SFSR Metal axyl NRWD NRWU PCFH SFSR Metolachlor
2 8 - 8 - 8 -
Q) 01
o s ~ 8 - 8 - 8 -
TO £ - LO
C *
O c .
o > - V- LO - LO - LO -
o - m mm o - m mm mm o - m mm o - m
NRWD NRWU PCFH SFSR Simazine NRWD NRWU PCFH SFSR Imidacloprid NRWD NRWU PCFH SFSR Diuron NRWD NRWU PCFH SFSR
8 - 8 - 8 -
s - s - s -
8 - ? 8 - 8 -
8 - £ - 8 -
0 0 0

- LO -
o - "S" o - -TT- .. ___ o - mm
NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR
Figure 3. A total of 11 pesticides were detected at one or more of the locations. PCFH had the lowest number of pesticide detections. More pesticides were detected at the WWTP impacted NRWD than the agricultural NRWU. Bars represent the mean and points are measurements from individual samples (N = 4/site).
43


Disinfection by-products detected in the Shenandoah watershed, August-Setember, 2014 Bromodichloromethane Dibromochloromethane Dichloroacetonitrile Bromochloroacetonitrile
8 - 8 - 8 - 8 -

8 - 8 - 8 - 8 -

i - i - - i - i -

8 - 1 - 1 - 1 -
o - o - O - m_ m o -
NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR
Trichloromethane Tribromom ethane Trichloroacetaldehyde 1,1,1 -trichloro-2-propanone
8 - 8 - 8 - 8 -

c § - 8 - 8 - 8 -
5 3 - 8 - 8 - 8 -
O
15 g . 8 - 8 - 8 -
c (D 8 - o o o
O C *
o m m m m
NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR
Dichloroiodomethane Dibromoacetonitrile
O O

8 - 8 -
8 - 8 - m
NRWD NRWU PCFH SFSR NRWD NRWU PCFH SFSR
Figure 4. Halogenated disinfection byproducts detected at site in the Shenandoah Valley. DBPs were detected predominately at NRWD, with the exception of one detection of trichloromethane at PCFH. Bars represent the mean and points are measurements from individual samples (N = 4/site).
44


Figure 5. Hierarchical clustering of water samples for each location based on binary presence/absence data. The contaminant profiles from each location cluster together, indicating there is more similarity within locations than between different locations. NRWD had the most unique profile. Agricultural = NRWU, WWTP = NRWD, Mixed Use = SFSR, and Reference = PCFH.
45


1.2
O 0.6
<
1/1 0.8
c
1
0.4
0.2
0 Hiiiiiiiiiiiiiiiiiiiii
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Day
Figure 6. Survivorship at PCFH, NRWD, and SFSR was > 90%, but was significantly reduced to 48% at NRWD by exposure Day 21 (ANOVA, p < 0.05).
46


Gonadosomatic Index
5i
3-
0
2-
1-

o i i i i
TJ U TJ TJ
X 3 Q X
U- > > W
O ^ ^ Li.
Q. 9; ^ W
M N N N
T3 "D "D "D
X 3 Q X
U- > > W
O jy jy LL
Q. ~ ~
Figure 7. Gonadosomatic index was decreased at NRWU and NRWD for 21-day exposed fish compared to the initial controls (ANOVA, p < 0.05, N = 10/factor). The middle bar is the median. Upper and lower boundaries of the box are the 3rd and 1st quartiles, respectively and whiskers are the maximum and minimum values. Asterisks indicate significant difference from initial control [IC] at p < 0.05.
47


Figure 8. The number of nuptial tubercles was reduced at NRWU and NRWD in 21-day exposed fish compared to initial controls (ANOVA, p < 0.05, N = 10/factor). Nuptial tubercles are a secondary sex characteristic that are modulated by androgen signaling. Error bars = SEM. The middle bar is the median. Tipper and lower boundaries of the box are the 3rd and 1st quartiles, respectively and whiskers are the maximum and minimum values. Asterisks indicate significant difference from initial control [IC] at p < 0.05.
48


Figure 9. Vitellogenin protein was statistically increased at PCFH compared to the initial control cohort (ANOVA, p > 0.05, N = 10/factor). The magnitude of this change is relatively small compared to other studies of estrogenic WWTP effluents. Bars represent the mean and points are measurements from individual samples. Asterisks indicate significant difference from initial control [IC] at p < 0.05.
49


NRWD NRWU
SFSR
Figure 10. Venn diagram of differentially expressed genes with annotations and the overlap between sites. For each site the largest groups of DEGs is unique to that exposure location (Oliveros, 2015). Genes were considered differentially expressed at p < 0.05 following false discovery rate correction.
50


Expression of reproduction related genes
Estrogen receptor 1 Estrogen receptor 2 Androgen receptor
Figure 11. Expression of transcripts related to reproduction in the liver. There was a higher level of vitellogenin 1 transcript in fish exposed at the reference PCRH site (fc = 2.4, p = 0.0496, N = 31). Overall these data show little evidence of estrogenic endocrine disruption at these sites. The middle bar is the median. Upper and lower boundaries of the box are the 3rd and 1st quartiles, respectively. Whiskers are the maximum and minimum values that are not outliers and circles represent outliers. Different letters denote significant differences at p < .05 following Tukeys HSD correction.
51


AhR pathway genes
Aryl hydrocarbon receptor 1b1
Aryl hydrocarbon receptor 2
AhR nuclear translocator 2
PCFH NRWU NRWD SFSR
Figure 12. Genes involved in the aryl hydrocarbon receptor pathway. Transcription of cytochrome p450 lal was significantly up-regulated at NRWD (fc = 2.8, p < 0.001, N = 7 or 8/site).). None of the other genes had altered transcription following exposure. The middle bar is the median. Upper and lower boundaries of the box are the 3rd and 1st quartiles, respectively. Whiskers are the maximum and minimum values that are not outliers and circles represent outliers. Different letters denote significant differences at p < .05 following Tukeys HSD correction.
52


Figure 13. Schematic of the mevalonate pathway (Pathway Studio 9.0, Ariadne, Rockville, MD). This pathway is responsible for synthesizing mevalonate, a precursor of cholesterol. The mevalonate pathway was identified as down-regulated at both NRWU and NRWD and up-regulated at SFSR. Here, only the data from NRWU is represented. Green = decreased expression, red = increased expression.
53


Figure 14. A schematic diagram of the IL6/STAT signal transduction pathway (Pathway Studio 9.0, Ariadne, Rockville, MD). Green indicates transcripts that were down-regulated at NRWU and red indicates genes that were up-regulated.
54


Figure 15. A schematic diagram of the classical complement pathway (Pathway Studio
9.0, Ariadne, Rockville, MD). Green indicates transcripts that were down-regulated at NRWU and red indicates genes that were up-regulated.
55


Figure 16. Heatmap and hierarchical clustering of differentially expressed genes. Red indicates increased transcription and green indicates decreased transcription, relative to the mean expression for each gene. Agricultural = NRWU, WWTP = NRWD, Mixed Use = SFSR, and Reference = PCFH.
56


Table 1. Genes selected a priori for analysis based on known responses to environmental contaminants.
Gene name______________________Gene symbol
Reproduction related
Estrogen receptor 1 erl
Estrogen receptor 2 er2
Androgen receptor ar
Vitellogenin 1 vtgl
Vitellogenin 3 vtg2
Xenobiotic responsive
Aryl-hydrocarbon receptor 1 ahrl
Aryl-hydrocarbon receptor 2 ahr2
AhR nuclear translocator 1 arntl
AhR nuclear translocator 2 arnt2
Cytochrome p450 1A1 cyplal
Cytochrome p450 1B1 cyplbl
57


Table 2. Summary of chemicals detected in the Shenandoah Valley, August-Sept. 2014
___________________________NRWD NRWU__________PCFH_____SFSR
Base chemistry
National Research Program, Boulder, CO
Temp 24.58 24.70 24.63 25.35
Dissolved oxygen (%) 92.75 98.15 92.13 98.08
Specific conductivity 569.00 352.63 147.03 303.95
pH 8.13 8.13 8.20 8.49
Nitrate 12.52 10.51 1.39 3.56
Dissolved organic content 3.44 1.93 2.83 2.15
uv254 0.05 0.03 0.06 0.04
SUVA 0.02 0.02 0.02 0.02
Nutrients
USGS National Water Quality Program, Denver, CO
Total P X X X X
Phosphate X X X
Total N X X X X
Ammonia X X X
Nitrite X X X
Nitrate & Nitrite X X X
Nitrate X X X
Organinc N X X X X
Trace elements
USGS National Reasearch Program, Boulder, CO; inductively-coupled plasma/mass spectrometry and inductively-coupled plasma/atomic emission spectrometry________
Al X X X X
As X X X X
B X X X X
Ba X X X X
Be Bi X X X X X X
Ca X X X X
Cd X X X X
Ce X X X X
Co X X X X
Cr X X X
Cs X X X X
Cu X X X X
Dy X X X X
Er X X X X
Eu X X X X
Fe X X X X
Ga X X X X
Gd X X X X
Ho X X X X
K X X X X
La X X X X
Li X X X
Lu X X X X
Mg X X X X
Mn X X X X
Mo X X X X
Na X X X X
Nd X X X X
Ni X X X X
P X X X X
Pb X X X X
Pr X X X X
Rb X X X X
S X X X X
Sb X X X X
Se X X X X
Si02 X X X X
Sm X X X X
Sr X X X X
Tb X X X X
Th X X X X
Tl X X X X
Tm X X X X
U X X X X
V X X X X
Y X X X X
Yb X X X X
Zn X X X X
Zr X X X X
58


Table 2. Contd
NRWD NRWU______PCFH_____SFSR
Disinfection by-products
USGS California Water Science Center, Sacramento, CA; solid phase extraction with gas chromatography/mass spectrometry
Trichloromethane X X
Bromodichloromethane X
Dibromochloromethane X
Tribromomethane X
Dichloroiodomethane X
Dichloroacetonitrile X
Bromochloroacetonitrile X
Dibromoacetonitrile X
T richloroacetaldehyde X
1,1,1-trichloro-2-propanone X
Pesticides
USGS California Water Science Center, Sacramento, CA; solid phase extraction with
liquid or qas chromatoqraphy/mass spectrometry
Atrazine X X X
Azoxystrobin X X
Boscalid X
Fipronil X
Fluxapyroxade X X
Metalaxyl X X
Metolachlor X X X
Simazine X X X
Imidacloprid X X
Diuron X
CLLE
USGS National Research Program, Boulder, CO; continuous liquid-liquid extraction
with qas chromatoqraohv/tandem mass spectrometry
1,3-dichlorobenzene X
1,4-dichlorobenzene X
1,2-dichlorobenzene X
2,6-di-tert-butyl-1,4-benzoquinone X X
5-methyl-1 H-benzotriazole X
N,N-diethyl-m-toluamide X X
Desethylatrazine X X X
4-nonylphenol X X
Atrazine X X X
Caffeine X
Galaxolide X
Tonalide X
Diphenhydramine X X X X
Triclosan X X
Carbamazapine X X
Coprostanol X
Cholesterol X X X X
59


Table 2 Contd
________________________NRWD NRWU_____PCFH
Pharmacuticals
USGS National Water Quality Laboratory, Denver, CO; direct-aqueous-injection liquid chromatoqraphy/tandem mass spectrometry_____________________________
Atrazine X X
Acetaminophen X
Bupropion X
Caffeine X X X
Carbamazepine X X
Cotinine X X X
Sulfamethoxazole
Thiabendazole X X
Lidocaine
Meprobamate X
Phenytoin X
Temazepam X
Triamterene
Fluconazole X
Acyclovir X
Metformin X X X
Nicotine X X X
Diazepam X
Methocarbamol X
Atenolol X
Fexofenadine X
Methyl-1 h-benzotriazole X X
Tramadol
Metoprolol X
Sitagliptin X
Venlafaxine
Glyburide X
SFSR
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Antibiotics
USGS Kansas Water Science Center, Lawrence, KS; liquid chromatography/tandem mass spectrometry__________________________________________________________
Azithromycin X
Carbamazepine X X
Sulfamethoxazole X X
Hormones & Phytoestrogens
USGS National Research Program, Boulder, CO; solid-phase extraction, derivatization
and qas chromatoqraphv/tandem mass spectrometrv
Equol X X X
Estrone X X X X
4-androstene-3,17-dione X
Equlin X X
Estriol X X
Coprostanol X X X X
Coumesterol X X X X
Cholesterol X X X X
Dihydrocholesterol X X X X
USGS OGR Laboratorv, Lawrence, KS; liquid chromatoqraphv/tandem mass
Daidzein X
Formononentin X X X X
Equol X X X
Biochanin A X
60


Table 3. Number of differentially expressed genes observed at each exposure site, FDR corrected.
Site Abbreviation Site description__________________Total PEGS Up-regulated Down-regulated
NRWU North River upstream of WWTP 202 118 84
NRWD North River downstream of WWTP 297 232 65
SFSR South Fork Shenandoah River at Shenandoah River State Park 219 112 107
61


Table 4. Highly differentially regulated pathways identified by gene-set enrichment
analysis.
Location Gene-set Name
# of Measured Entities Normalized ES Median change p-value
NRWD
NRWU
SFSR
Heme oxidation
Irinotecan metabolism
Capecitabine and Fluorafur metabolism
Ascorbate biosyntesis
Pentose-phosphate shunt
EphrinB-> JUN signaling
Vitamin A (retinol) metabolism and visual cycle
Urea cycle and arginine metabolism
EctodysplasinR -> AP-1 signaling
NGFR -> AP-1/CEBPB/CREB/ELK-SRF/TP53 signaling
Extracellular Matrix Turnover
Selencompound biosynthesis
Tight Junction Assembly (Occludin)
tRNATranscription and Processing
Actin Cytoskeleton Assembly
Exocytosis
Mevalonate pathway Biosynthesis of cholesterol
Glycogen metabolism Nicotinate and nicotinamide metabolism EphrinB-> JUN signaling Phenylalanine and Tyrosine metabolism Glucose metabolism
Respiratory chain and oxidative phosphorylation
Glutathione metabolism
Glyoxylate and glycerate metabolism
Copy of 'TLR -> AP-1 signaling'-TB
Protein Nuclear Import and Export
Coagulation Cascade
L-sugars oxidation
Classical Complement Pathway
Lectin-induced Complement Pathway
Alternative Complement Pathway
IL6R -> STAT signaling
IL6ST -> STAT5B signaling
Biosynthesis of cholesterol
Mevalonate pathway
IL21R -> STAT signaling
IL9R -> STAT signaling
IL7R -> STAT signaling
IL15R -> STAT signaling
IL2R -> STAT signaling
IL10R -> STAT signaling
VasopressinR2 -> STAT signaling
Ketonogenesis
EphrinB-> JUN signaling
Single-Strand Base Excision DNA Repair
Tryptophan metabolism
Glutathione metabolism
Direct DNA Repair
Fatty acid oxidation
Selencompound biosynthesis
Alternative Complement Pathway
Intermediate Filament Polymerization
tRNA Transcription and Processing
Galactose metabolism
21 2.03 1.67 <0.001
21 1.89 1.52 <0.001
22 1.95 1.48 <0.001
33 1.83 1.41 0.003
17 1.80 1.40 0.004
15 1.59 1.32 0.026
40 1.50 1.28 0.023
33 1.99 1.28 <0.001
16 1.71 -1.09 0.014
35 1.48 -1.11 0.040
89 -1.35 -1.14 0.038
70 -1.43 -1.15 0.021
241 -1.65 -1.18 <0.001
40 -1.58 -1.21 0.022
39 -1.64 -1.21 0.010
31 -1.66 -1.24 0.005
16 -2.20 -1.61 <0.001
31 -2.73 -1.66 <0.001
29 1.92 1.36 <0.001
21 2.00 1.36 <0.001
15 1.76 1.33 0.015
35 1.72 1.25 0.004
55 2.06 1.23 <0.001
66 1.46 1.18 0.032
54 1.52 1.18 0.014
21 1.77 1.18 0.014
15 1.65 1.12 <0.001
54 -1.75 -1.27 0.005
23 -2.06 -1.28 <0.001
5 -1.58 -1.29 0.024
21 -1.85 -1.40 <0.001
15 -1.99 -1.40 <0.001
14 -1.95 -1.40 0.005
8 -1.59 -1.41 0.038
7 -1.66 -1.67 0.008
31 -3.17 -1.92 <0.001
16 -2.39 -1.99 <0.001
6 1.74 1.67 <0.001
6 1.79 1.67 <0.001
7 1.66 1.60 0.013
8 1.51 1.60 0.032
7 1.68 1.60 0.039
6 1.53 1.60 0.046
7 1.54 1.59 0.031
9 1.86 1.49 0.004
15 1.91 1.41 0.004
98 -1.42 -1.10 0.035
105 -1.51 -1.11 0.014
54 -1.59 -1.12 0.021
103 -1.51 -1.12 <0.001
38 -1.82 -1.14 0.004
70 -1.74 -1.15 <0.001
14 -1.48 -1.16 0.039
33 -1.49 -1.20 0.038
41 -1.82 -1.23 0.004
14 -1.76 -1.37 0.004
62


Table 5. Highly differentially regulated sub-networks of genes identified by sub-network
enrichment analysis.
Location Cell process seed name
# of Measured Neighbors Median change p-value
NRWD pyrimidine salvage 6 2.28 0.002
response to acid 6 2.21 0.022
response to heavy metal 7 2.07 0.006
renal vasodilatation 8 2.02 0.001
response to pheromone 5 1.92 0.017
bile acid conjugation 7 1.73 0.028
endometrium differentiation 16 1.60 <0.001
meiotic spindle assembly 11 1.58 0.032
ethanol oxidation 6 1.53 0.006
spinal cord development 10 -1.30 0.012
proteolysis 13 -1.30 0.007
viral genome replication 7 -1.32 0.013
lactotrope development 5 -1.35 0.013
mevalonate pathway 24 -1.46 0.007
sterol transport 17 -1.46 0.025
blastocyst growth 5 -1.52 0.037
Schwann cell formation 15 -1.53 0.016
spleen development 5 -1.55 0.024
respiratory mechanics 5 -1.56 0.026
NRWU alcohol metabolism 6 1.58 0.013
gonad morphogenesis 5 1.50 0.034
gall bladder relaxation 6 1.48 0.024
amino acid biosynthesis 7 1.47 0.040
fatty acid elongation 15 1.43 0.030
lipoprotein catabolism 10 1.38 0.019
peritoneal macrophage function 5 1.32 0.006
actomyosin based movement 19 1.31 0.046
uterine angiogenesis 7 1.30 0.008
response to cadmium ion 13 -1.30 0.034
biliary flow 37 -1.30 0.008
sterol biosynthesis 29 -1.33 <0.001
lactotrope development 5 -1.36 0.015
secondary metabolism 9 -1.37 0.042
mevalonate pathway 24 -1.39 0.004
thymocyte adhesion 15 -1.39 0.015
membrane protein ectodomain proteolysis 7 -1.45 0.044
long-chain fatty acid transport 9 -1.46 0.008
myometrium growth 5 -1.87 0.017
gastrointestinal secretions 5 -2.32 0.043
SFSR heme metabolism 10 1.89 0.004
pyrimidine salvage 6 1.83 0.028
alcohol metabolism 6 1.56 0.009
endometrium differentiation 16 1.52 <0.001
peptide transport 14 1.51 0.013
peritoneal macrophage function 5 1.50 0.003
nucleosome mobilization 6 1.46 0.044
monocyte activity 6 1.42 0.021
uterine angiogenesis 7 1.39 0.003
mitochondrial amplification 9 -1.30 0.041
digestive process 7 -1.30 0.032
long-chain fatty acid transport 9 -1.32 0.048
spinal cord development 10 -1.35 0.003
eosinophil function 16 -1.35 0.018
desmosome assembly 23 -1.36 0.045
response to stimulus 7 -1.37 0.005
neutrophil homeostases 5 -1.37 0.045
pre-miRNA processing 8 -1.42 0.046
mature B cell differentiation 6 -1.55 0.006
killing virus-infected cells 9 -1.58 0.035
63


References
Aaronson, D.S., Horvath, C.M., 2002. A Road map for those who dont know JAK-STAT. Science. 296, 1653-1655. doi:10.1126/science,1071545
Akdis, M., Burgler, S., Crameri, R., Eiwegger, T., Fujita, H., Gomez, E., Klunker, S., Meyer, N., OMahony, L., Palomares, O., Rhyner, C., Quaked, N., Schaffartzik, A., Van De Veen, W., Zeller, S., Zimmermann, M., Akdis, C.A., 2011. Interleukins, from 1 to 37, and interferon-y: Receptors, functions, and roles in diseases. J. Allergy Clin. Immunol. 127, 701-721. doi: 10.1016/j jaci.2010.11.050
Altenburger, R., Scholz, S., Schmitt-Jansen, M., Busch, W., Escher, B.I., 2012. Mixture toxicity revisited from a toxicogenomic perspective. Environ. Sci. Technol. 46, 2508-2522. doi:10.1021/es2038036
Amato, J.R., Lukasewycz, M.T., Robert, E.D., Mount, D.I., Durhan, E.J., Ankley, G.T., 1992. An example of the identification of diazinon as a primary toxicant in an effluent. Environ. Toxicol. Chem. 11, 209-216. doi: 10.1002/etc.5620110210
Ankley, G., Daston, G., 2006. Toxicogenomics in regulatory ecotoxicology. Environ. Sci. Technol. 40, 4055-4065.
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., Serrrano, J.A., Tietge,
J.E., Villeneuve, D.L., 2010. Adverse outcome pathways: A conceptual framework to support ecotoxicology research and risk assessment. Environ. Toxicol. Chem. 29, 730-741. doi: 10.1002/etc.34
Bahamonde, P.A., McMaster, M.E., Servos, M.R., Martyniuk, C.J., Munkittrick, K.R., 2015. Molecular pathways associated with the intersex condition in rainbow darter (Etheostoma caeruleum) following exposures to municipal wastewater in the Grand River basin, ON, Canada. PartB. Aquat. Toxicol. 159, 302-316. doi: 10.1016/j.aquatox.2014.11.022
Barber, L.B., Brown, G.K., Nettesheim, T.G., Murphy, E.W., Bartell, S.E., Schoenfuss, H.L., 2011. Effects of biologically-active chemical mixtures on fish in a wastewater-impacted urban stream. Sci. Total Environ. 409, 4720-8. doi: 10.1016/j.scitotenv.2011.06.039
Barber, L.B., Vajda, A.M., Douville, C., Norris, D.O., Writer, J.H., 2012. Fish endocrine disruption responses to a major wastewater treatment facility upgrade. Environ. Sci. Technol. 46, 2121-31. doi:10.1021/es202880e
Berenbaum, M.C., 1989. What is synergy? Pharmacol. Rev. 41, 93-141. doi :10.1007/s 13398-014-0173-7.2
Beminger, J.P., Martinovic-Weigelt, D., Garcia-Reyero, N., Escalon, L., Perkins, E.J., Ankley, G.T., Villeneuve, D.L., 2014. Using transcriptomic tools to evaluate biological effects across effluent gradients at a diverse set of study sites in Minnesota, USA. Environ. Sci. Technol. 48, 2404-2412. doi:10.1021/es4040254
64


Bitman, J., Cecil, H.C., Harris, S.J., Fries, G.F., 1968. Estrogenic activity of o, p-DDT in the mammalian uterus and avian oviduct. Science. 162, 371-372. doi:10.1126/science. 162.3851.371
Blackburn, M.A., Waldock, M.J., 1995. Concentrations of alkylphenols in rivers and estuaries in England and Wales. Water Res. 29, 1623-1629. doi: 10.1016/0043-1354(94)00340-0
Blazer, V.S., Iwanowicz, L.R., Iwanowicz, D.D., Smith, D.R., Young, J.A., Hedrick, J.D., Foster, S.W., Reeser, S.J., 2007. Intersex (testicular oocytes) in smallmouth bass from the Potomac River and selected nearby drainages. J. Aquat. Anim. Health 19, 242-53. doi:10.1577/H07-031.1
Blazer, V.S., Iwanowicz, L.R., Starliper, C.E., Iwanowicz, D.D., Barbash, P., Hedrick, J.D., Reeser, S.J., Mullican, J.E., Zaugg, S.D., Burkhardt, M.R., Kelble, J., 2010. Mortality of centrarchid fishes in the potomac drainage: Survey results and overview of potential contributing factors. J. Aquat. Anim. Health 22, 190-218. doi :10.1577/H10-002.1
Brack, W., 2003. Effect-directed analysis: A promising tool for the identification of organic toxicants in complex mixtures? Anal. Bioanal. Chem. 377, 397-407. doi: 10.1007/s00216-003 -213 9-z
Brack, W., Altenburger, R., Ensenbach, U., Moder, M., Segner, H., Schuurmann, G.,
1999. Bioassay-directed identification of organic toxicants in river sediment in the industrial region of bitterfeld (Germany)-A contribution to hazard assessment. Arch. Environ. Contam. Toxicol. 37, 164-74.
Burgess, R.M., Ho, K.T., Brack, W., Lamoree, M., 2013. Effects-directed analysis (EDA) and toxicity identification evaluation (TIE): Complementary but different approaches for diagnosing causes of environmental toxicity. Environ. Toxicol.
Chem. 32, 1935-1945. doi:10.1002/etc.2299
Chen, H.-S., Jia, J., Su, H.-F., Lin, H.-D., Chen, J.-W., Lin, S.-J., Yang, J.-Y., Lai, H.-C., Mestril, R., Wang, P.H., 2006. Downregulation of the constitutively expressed Hsc70 in diabetic myocardium is mediated by insulin deficiency. J. Endocrinol. 190, 433-40. doi:10.1677/joe. 1.06692
Cheshenko, K., Pakdel, F., Segner, H., Kah, O., Eggen, R.I.L., 2008. Interference of endocrine disrupting chemicals with aromatase CYP19 expression or activity, and consequences for reproduction of teleost fish. Gen. Comp. Endocrinol. 155, 31-62. doi:10.1016/j.ygcen.2007.03.005
Choi, E.-Y., Jin, J.-Y., Lee, J.-Y., Choi, J.-I., Choi, I.S., Kim, S.-J., 2012. Antiinflammatory effects and the underlying mechanisms of action of daidzein in murine macrophages stimulated with Prevotella intermedia lipopolysaccharide. J.
Periodontal Res. 47, 204-11. doi: 10.1111/j. 1600-0765.2011.01422.x
Clarke, D.B., Bailey, V., Lloyd, A.S., 2008. Determination of phytoestrogens in dietary supplements by LC-MS/MS. Food Addit. Contam. Part A. Chem. Anal. Control. Expo. Risk Assess. 25, 534-47. doi: 10.1080/02652030701658340
65


Colborn, T., Clement, C. (Eds.), 1992. Chemically-induced alterations in sexual and functional development: The wildlife/human connection. Princeton Scientific Publishing.
Crain, A.C., Guillette, L. (Eds.), 2000. Environmental endocrine disrupters: An evolutionary perspective. CRC Press.
Daughton, C.G., 2001. Pharmaceuticals and personal care products in the environment: overarching issues and overview., in: Daughton, C.G., Jones-Lepp, T. (Eds.), Pharmaceuticals and Personal Care Products in the Environment: Scientific and Regulatory Issues. American Chemical Society Symposium Series 791, pp. 2-38.
Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon receptor by
structurally diverse exogenous and endogenous chemicals. Annu Rev Pharmacol Toxicol 43, 309-334. doi:10.1146/annurev.pharmtox.43.100901.135828
Elferink, M.G.L., Olinga, P., van Leeuwen, E.M., Bauerschmidt, S., Polman, I,
Schoonen, W.G., Heisterkamp, S.H., Groothuis, G.M.M., 2011. Gene expression analysis of precision-cut human liver slices indicates stable expression of ADME-Tox related genes. Toxicol. Appl. Pharmacol. 253, 57-69. doi:10.1016/j.taap.2011.03.010
Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United StatesII) untreated drinking water sources. Sci. Total Environ. 402, 201-16. doi: 10.1016/j. scitotenv .2008.02.021
Fromme, H., Kiichler, T., Otto, T., Pilz, K., Muller, J., Wenzel, A., 2002. Occurrence of phthalates and bisphenol A and F in the environment. Water Res. 36, 1429-1438. doi: 10.1016/S0043 -13 54(01 )003 67-0
Gabay, C., Kushner, I., 1999. Acute-phase proteins and other systemic responses to
inflammation. N. Engl. J. Med. 340, 448-454. doi: 10.1056/NEJM199902113400607
Gaido, K.W., Leonard, L.S., Lovell, S., Gould, J.C., Baba'i, D., Portier, C.J., McDonnell, D.P., 1997. Evaluation of chemicals with endocrine modulating activity in a yeast-based steroid hormone receptor gene transcription assay. Toxicol. Appl. Pharmacol. 143, 205-212. doi: 10.1006/taap. 1996.8069
Garcia-Reyero, N., Adelman, I., Liu, L., Denslow, N., 2008. Gene expression profiles of fathead minnows exposed to surface waters above and below a sewage treatment plant in Minnesota. Mar. Environ. Res. 66, 134-136. doi:10.1016/j.marenvres.2008.02.073
Garcia-Reyero, N., Lavelle, C.M., Escalon, B.L., Martinovic, D., Kroll, K.J., Sorensen, P.W., Denslow, N.D., 2010. Behavioral and genomic impacts of a wastewater effluent on the fathead minnow. Aquat. Toxicol. 101, 38-48. doi:10.1016/j.aquatox.2010.08.014
Garcia-Reyero, N., Perkins, E.J., 2011. Systems biology: Leading the revolution in ecotoxicology. Environ. Toxicol. Chem. 30, 265-273. doi:10.1002/etc.401
66


Gilbert, S.F., Epel, D., 2009. Endocrine Disruptors, in: Ecological Developmental Biology: Integrating Epigenetics, Medicine, and Evolution, pp. 197-244.
Gilbert, S.F., Epel, D., 2009. Ecological Developmental Biology: Integrating Epigenetics, Medicine, and Evolution 82, 231-232. doi: 10.1126/science. 1175598
Hahn, M.E., 2002. Aryl hydrocarbon receptors: diversity and evolution. Chem. Biol. Interact. 141, 131-160. doi:10.1016/S0009-2797(02)00070-4
Hai, T., Wolfgang, C.D., Marsee, D.K., Allen, A.E., Sivaprasad, U., 1999. ATF3 and stress responses. Gene Expr. 7, 321-35.
Hayes, T.B., Khoury, V., Narayan, A., Nazir, M., Park, A., Brown, T., Adame, L., Chan, E., Buchholz, D., Stueve, T., Gallipeau, S., 2010. Atrazine induces complete feminization and chemical castration in male African clawed frogs (Xenopus laevis). Proc. Natl. Acad. Sci. U. S. A. 107, 4612-4617. doi: 10.1073/pnas.0909519107
Heinrich, P.C., Behrmann, I., Muller-Newen, G., Schaper, F., Graeve, L., 1998. Interleukin-6-type cytokine signalling through the gpl30/Jak/STAT pathway. Biochem. J. 334, 297-314. doi:10.1042/bj3340297
Hill, R.H., Head, S.L., Baker, S., Gregg, M., Shealy, D.B., Bailey, S.L., Williams, C.C., Sampson, E.J., Needham, L.L., 1995. Pesticide residues in urine of adults living in the United States: reference range concentrations. Environ. Res. doi :10.1006/enrs. 1995.1071
Ihlaseh, S.M., Bailey, K.A., Hester, S.D., Jones, C., Ren, H., Cardoso, A.P.F., Oliveira, M.L.C.S., Wolf, D.C., de Camargo, J.L. V, 2011. Transcriptional profile of diuron-induced toxicity on the urinary bladder of male wistar rats to inform mode of action. Toxicol. Sci. 122, 330-338. doi:10.1093/toxsci/kfrl08
IOM, (Institue of Medicine), 2014. Institue of Medicine, in: Identifying and Reducing Environmental Health Risks of Chemicals in Our Society: Workshop Summary. The National Academies Press, Washington, DC.
Iwanowicz, L.R., Blazer, V.S., Guy, C.P., Pinkney, A.E., Mullican, J.E., Alvarez, D.A., 2009. Reproductive health of bass in the Potomac, U.S.A., drainage: part 1. Exploring the effects of proximity to wastewater treatment plant discharge. Environ. Toxicol. Chem. 28, 1072-1083. doi:10.1897/08-433.1
Jensen, K.M., Korte, J.J., Kahl, M.D., Pasha, M.S., Ankley, G.T., 2001. Aspects of basic reproductive biology and endocrinology in the fathead minnow (Pimephales promelas). Comp. Biochem. Physiol. Part C Toxicol. Pharmacol. 128, 127-141. doi: 10.1016/S1532-0456(00)00185-X
Jobling, S., Nolan, M., Tyler, C.R., Brighty, G., Sumpter, J.P., 1998. Widespread sexual disruption in wild fish. Environ. Sci. Technol. 32, 2498-2506. doi: 10.102 l/es9710870
Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E., Lazorchak, J.M., Flick, R.W., 2007. Collapse of a fish population after exposure to a synthetic estrogen. Proc. Natl. Acad. Sci. U. S. A. 104, 8897-8901.
67


doi:10.1073/pnas.0609568104
Kock-Schulmeyer, M., Villagrasa, M., Lopez de Alda, M., Cespedes-Sanchez, R.,
Ventura, F., Barcelo, D., 2013. Occurrence and behavior of pesticides in wastewater treatment plants and their environmental impact. Sci. Total Environ. 458-460, 466-76. doi:10.1016/j.scitotenv.2013.04.010
Kolpin, D.W., Blazer, V.S., Gray, J.L., Focazio, M.J., Young, J.A., Alvarez, D.A., Iwanowicz, L.R., Foreman, W.T., Furlong, E.T., Speiran, G.K., Zaugg, S.D., Hubbard, L.E., Meyer, M.T., Sandstrom, M.W., Barber, L.B., 2013. Chemical contaminants in water and sediment near fish nesting sites in the Potomac River basin: Determining potential exposures to smallmouth bass (Micropterus dolomieu). Sci. Total Environ. 443, 700-716. doi:10.1016/j.scitotenv.2012.09.063
Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in El .S. streams, 1999 2000: A national reconnaissance. 36, 1202-1211. doi: 10.1021/esOl 1055j
Kortenkamp, A., 2007. Ten years of mixing cocktails: A review of combination effects of endocrine-disrupting chemicals. Environ. Health Perspect. 115, 98-105. doi:10.1289/ehp.9357
Kortenkamp, A., Altenburger, R., 1999. Approaches to assessing combination effects of oestrogenic environmental pollutants. Sci. Total Environ. 233, 131-140. doi: 10.1016/S0048-9697(99)00228-4
Kortenkamp, A., Altenburger, R., 1998. Synergisms with mixtures of xenoestrogens: A reevaluation using the method of isoboles. Sci. Total Environ. 221, 59-73. doi: 10.1016/S0048-9697(98)00261 -7
Kuiper, G.G., Carlsson, B., Grandien, K., Enmark, E., Haggblad, J., Nilsson, S.,
Gustafsson, J.A., 1997. Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors alpha and beta. Endocrinology 138, 863-70. doi:10.1210/endo. 138.3.4979
Kuiper, G.G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998. Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta. Endocrinology 139, 4252-63. doi:10.1210/endo. 139.10.6216
Kumar, V., Johnson, A.C., Trubiroha, A., Tumova, J., Ihara, M., Grabic, R., Kloas, W., Tanaka, H., Kroupova, H.K., 2015. The challenge presented by progestins in ecotoxicological research: a critical review. Environ. Sci. Technol. 49, 2625-38. doi:10.1021/es5051343
Kuster, M., Lopez de Alda, M.J., Hernando, M.D., Petrovic, M., Martin-Alonso, J., Barcelo, D., 2008. Analysis and occurrence of pharmaceuticals, estrogens, progestogens and polar pesticides in sewage treatment plant effluents, river water and drinking water in the Llobregat river basin (Barcelona, Spain). J. Hydrol. 358, 112-123. doi: 10.1016/j.jhydrol.2008.05.030
68


Lange, I.G., Daxenberger, A., Schiffer, B., Witters, H., Ibarreta, D., Meyer, H.H.., 2002. Sex hormones originating from different livestock production systems: fate and potential disrupting activity in the environment. Anal. Chim. Acta 473, 27-37. doi: 10.1016/S0003 -2670(02)00748-1
Lee, K.-H., Choi, E.-M., 2005. Biochanin A stimulates osteoblastic differentiation and inhibits hydrogen peroxide-induced production of inflammatory mediators in MC3T3-E1 cells. Biol. Pharm. Bull. 28, 1948-1953. doi:10.1248/bpb.28.1948
Lee, Y.-M., Williams, T.D., Jung, S.-O., Lee, J.-S., 2005. cDNA cloning and expression of a cytochrome P450 1A (CYP1 A) gene from the hermaphroditic fish Rivulus marmoratus. Mar. Pollut. Bull. 51, 769-75. doi:10.1016/j.marpolbul.2005.06.015
Lemaire, G., Terouanne, B., Mauvais, P., Michel, S., Rahmani, R., 2004. Effect of
organochlorine pesticides on human androgen receptor activation in vitro. Toxicol. Appl. Pharmacol. 196, 235-46. doi:10.1016/j.taap.2003.12.011
Limonciel, A., Moenks, K., Stanzel, S., Truisi, G.L., Parmentier, C., Aschauer, L.,
Wilmes, A., Richert, L., Hewitt, P., Mueller, S.O., Lukas, A., Kopp-Schneider, A., Leonard, M.O., Jennings, P., 2014. Transcriptomics hit the target: Monitoring of ligand-activated and stress response pathways for chemical testing. Toxicol. Vitr. 30, 7-18. doi: 10.1016/j.tiv.2014.12.011
Liu, T., Daniels, C.K., Cao, S., 2012. Comprehensive review on the HSC70 functions, interactions with related molecules and involvement in clinical diseases and therapeutic potential. Pharmacol. Ther. 136, 354-74. doi:10.1016/j.pharmthera.2012.08.014
Malpass, G.E., Arimilli, S., Prasad, G.L., Howlett, A.C., 2014. Regulation of gene expression by tobacco product preparations in cultured human dermal fibroblasts. Toxicol. Appl. Pharmacol. 279, 211-219. doi:10.1016/j.taap.2014.06.001
Martin, H.L., Svendsen, C., Lister, L.J., Gomez-Eyles, J.L., Spurgeon, D.J., 2009. Measurement and modeling of the toxicity of binary mixtures in the nematode Caenorhabditis elegans A test of independent action. Environ. Toxicol. Chem. 28, 97-104. doi: 10.1897/07-215.1
Martinovic-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., 2014. Transcriptomic effects-based monitoring for endocrine active chemicals: Assessing relative contribution of treated wastewater to downstream pollution. Environ. Sci. Technol. 48, 2385-2394. doi:10.1021/es404027n
Medjakovic, S., Mueller, M., Jungbauer, A., 2010. Potential health-modulating effects of isoflavones and metabolites via activation of PPAR and AhR. Nutrients 2, 241-279. doi: 10.3 3 90/nu203 0241
Mehinto, A.C., Martyniuk, C.J., Spade, D.J., Denslow, N.D., 2012. Applications of next-generation sequencing in fish ecotoxicogenomics. Front. Genet. 3, 1-10. doi: 10.3389/fgene.2012.00062
69


Mueller, M., Jungbauer, A., 2008. Red clover extract: a putative source for simultaneous treatment of menopausal disorders and the metabolic syndrome. Menopause 15,
1120-1131. doi: 10.1097/gme.ObO 13e31817062ce
Nakai, M., Tabira, Y., Asai, D., Yakabe, Y., Shimyozu, T., Noguchi, M., Takatsuki, M., Shimohigashi, Y., 1999. Binding characteristics of dialkyl phthalates for the estrogen receptor. Biochem. Biophys. Res. Commun. 254, 311-4. doi: 10.1006/bbrc. 1998.9928
Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W., Liess, M., 2002. The significance of entry routes as point and non-point sources of pesticides in small streams. Water Res. 36, 835-842. doi:10.1016/S0043-1354(01)00310-4
Nikitin, A., Egorov, S., Daraselia, N., Mazo, I., 2003. Pathway studiothe analysis and navigation of molecular networks. Bioinformatics 19, 2155-2157. doi: 10.1093/bioinformatics/btg290
Norris, D.O., Carr, J.A., 2006. Introduction to Endocrinology, in: Norris, D.O., Carr, J.A. (Eds.), Endocrine Disruption: Biological Bases for Health Effects in Wildlife and Humans. Oxford University Press.
Oliveros, J.C., 2015. Venny. An interactive tool for comparing lists with Venns diagrams, http://bioinfogp.cnb.csic.es/tools/venny/index.html.
Oscarson, M., Zanger, U.M., Rifki, O.F., Klein, K., Eichelbaum, M., Meyer, U.A., 2006. Transcriptional profiling of genes induced in the livers of patients treated with carbamazepine. Clin. Pharmacol. Ther. 80, 440-456. doi:10.1016/j.clpt.2006.08.013
Panter, G.H., Hutchinson, T.H., Hurd, K.S., Sherren, A., Stanley, R.D., Tyler, C.R., 2004. Successful detection of (anti-)androgenic and aromatase inhibitors in pre-spawning adult fathead minnows (Pimephales promelas) using easily measured endpoints of sexual development. Aquat. Toxicol. 70, 11-21. doi:10.1016/j.aquatox.2004.06.007
Paskova, V., Hilscherova, K., Blaha, L., 2011. Teratogenicity and embryotoxicity in aquatic organisms after pesticide exposure and the role of oxidative stress. Rev. Environ. Contam. Toxicol. 211, 25-61. doi:10.1007/978-1-4419-8011-3 2
Perkins, E.J., Antczak, P., Burgoon, L., Falciani, F., Garcia-Reyero, N., Gutsell, S., Hodges, G., Kienzler, A., Knapen, D., McBride, M., Willett, C., 2015. Adverse outcome pathways for regulatory applications: Examination of four case studies with different degrees of completeness and scientific confidence. Toxicol. Sci. 148, 14-25. doi:10.1093/toxsci/kfvl81
R Core Team, 2016. R: A language and environment for statistical computing. R
Foundation for Statistical Computing. Vienna, Austria. https://www.R-project.org.
Reineke, N., Bester, K., Hiihnerfuss, H., Jastorff, B., Weigel, S., 2002. Bioassay-directed chemical analysis of River Elbe surface water including large volume extractions and high performance fractionation. Chemosphere 47, 717-723. doi:10.1016/S0045-6535(01)00340-X
70


Santos, L.H.M.L.M., Araujo, A., Fachini, A., Pena, A., Delerue-Matos, C., Montenegro, M.C.B.S.M., 2010. Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. J. Hazard. Mater. 175, 45-95. doi:10.1016/j.jhazmat.2009.10.100
Schulpen, S.H.W., Pennings, J.L.A., Piersma, A.H., 2015. Gene expression regulation and pathway analysis after valproic acid and carbamazepine exposure in a human embryonic stem cell-based neurodevelopmental toxicity assay. Toxicol. Sci. 146, 311-320. doi: 10.1093/toxsci/kfv094
Schwindt, A.R., Winkelman, D.L., Keteles, K., Murphy, M., Vajda, A.M., 2014. An environmental oestrogen disrupts fish population dynamics through direct and transgenerational effects on survival and fecundity. J. Appl. Ecol. 51, 582-591. doi: 10.1111/1365-2664.12237
Sellin Jeffries, M.K., Mehinto, A.C., Carter, B.J., Denslow, N.D., Kolok, A.S., 2012. Taking microarrays to the field: Differential hepatic gene expression of caged fathead minnows from Nebraska watersheds. Environ. Sci. Technol. 46, 1877-1885. doi:10.1021/es2039097
Simmons, S.O., Fan, C.-Y., Ramabhadran, R., 2009. Cellular stress response pathway
system as a sentinel ensemble in toxicological screening. Toxicol. Sci. Ill, 202-225. doi: 10.1093/toxsci/kfp 140
Spurgeon, D.J., Jones, O. a. H., Dome, J.-L.C.M., Svendsen, C., Swain, S., Sturzenbaum, S.R., 2010. Systems toxicology approaches for understanding the joint effects of environmental chemical mixtures. Sci. Total Environ. 408, 3725-3734. doi:10.1016/j.scitotenv.2010.02.038
Stamper, B.D., Garcia, M.L., Nguyen, D.Q., Beyer, R.P., Bammler, T.K., Farin, F.M., Kavanagh, T.J., Nelson, S.D., 2015. p53 contributes to differentiating gene expression following exposure to acetaminophen and its less hepatotoxic regioisomer both In vitro and in vivo. Gene Regul. Syst. Bio. 9, 1-14. doi: 10.4137/GRSB.S253 88
Steel, D.M., Whitehead, A.S., 1994. The major acute phase reactants: C-reactive protein, serum amyloid P component and serum amyloid A protein. Immunol. Today 15, 81-88. doi: 10.1016/0167-5699(94)90138-4
Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., Mesirov, J.P., 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102, 15545-50. doi: 10.1073/pnas.0506580102
Suja, F., Pramanik, B.K., Zain, S.M., 2009. Contamination, bioaccumulation and toxic effects of perfluorinated chemicals (PFCs) in the water environment: a review paper. Water Sci. Technol. 60, 1533. doi:10.2166/wst.2009.504
71


Suzuki, G., Sato, K., Isobe, T., Takigami, H., Brouwer, A., Nakayama, K., 2015.
Detection of glucocorticoid receptor agonists in effluents from sewage treatment plants in Japan. Sci. Total Environ. 527-528, 328-34. doi:10.1016/j.scitotenv.2015.05.008
Tabb, M.M., Blumberg, B., 2006. New modes of action for endocrine-disrupting chemicals. Mol. Endocrinol. 20, 475-82. doi: 10.1210/me.2004-0513
Temes, T.., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.-D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants I. Investigations in Germany, Canada and Brazil. Sci. Total Environ. 225, 81-90. doi: 10.1016/S0048-9697(98)00334-9
Thomas, K. V., Hurst, M.R., Matthiessen, P., Waldock, M.J., 2001. Characterization of estrogenic compounds in water samples collected from United Kingdom estuaries. Environ. Toxicol. Chem. 20, 2165-2170. doi:10.1002/etc.5620201005
U.S. EPA, 2009. Inventory Update Reporting: Non-confidential 2006 IUR records by chemical, including manufacturing, processing and use information. Washington, DC.
U.S. EPA, 1991. Methods for aquatic toxicity identification evaluations: Phase I toxicity characterization proceadures. Duluth, MN.
Uribe, C., Folch, H., Enriquez, R., Moran, G., 2011. Innate and adaptive immunity in teleost fish: a review. Vet. Med. (Praha). 56, 486-503.
USDA NASS, 2012. Quick stats [WWW Document], United States Dep. Agric. Natl. Agric. Stat. Serv. URL https://www.nass.usda.gov/Quick_Stats/index.php (accessed 4.24.16).
Vajda, A.M., Barber, L.B., Gray, J.L., Lopez, E.M., Bolden, A.M., Schoenfuss, H.L., Norris, D.O., 2011. Demasculinization of male fish by wastewater treatment plant effluent. Aquat. Toxicol. 103, 213-21. doi:10.1016/j.aquatox.2011.02.007
Vajda, A.M., Norris, D.O., 2006. Endocrine-active phytochemicals: Environmental signaling context and mechanisms, in: Norris, D.O., Carr, J.A. (Eds.), Endocrine Disruption: The Biological Basis for Health Effects in Humans and Wildlife. Oxford University Press, pp. 390-423.
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., 2012. Ecotoxicogenomics to support ecological risk assessment: A case study with bisphenol A in fish. Environ. Sci. Technol. 46, 51-59. doi:10.1021/es201150a
vom Saal, F.S., Hughes, C., 2005. An extensive new literature concerning low-dose effects of bisphenol A shows the need for a new risk assessment. Environ. Health Perspect. 113, 926-933. doi:10.1289/ehp.7713
Wilson, M.P., Schwarzman, M.R., 2009. Toward a new U.S. chemicals policy: Rebuilding the foundation to advance new science, green chemistry, and environmental health. Environ. Health Perspect. doi: 10.1289/ehp.0800404
72


Young, J., Iwanowicz, L., Sperry, A., Blazer, V., 2014. A landscape-based
reconnaissance survey of estrogenic activity in streams of the upper Potomac, upper James, and Shenandoah Rivers, USA. Environ. Monit. Assess. 186, 5531-5545. doi: 10.1007/s 10661-014-3 801-y
Zipfel, P.F., Skerka, C., 2009. Complement regulators and inhibitory proteins. Nat. Rev. Immunol. 9, 729-740. doi:10.1038/nri2620
73


Full Text

PAGE 1

TRANSCRIPTOME PROFILING IN THE FATHEAD MINNOW ( Pimephales promelas) FOLLOWING EXPOSURE TO COMPLEX CHEMICAL MIXTURES: A LANDSCAPE BASED APPROACH by DAVID BERTOLATUS Bachelor of Arts, University of Iowa, 2007 Bachelor of Science, Metropolitan State Univer sity of Denver, 2013 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Ma s ter of Science Integrative Biology 2016

PAGE 2

ii 2016 D AVID BERTOLATUS ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Science degree by David Bertolatus Has been approved for the Department of Integrative Biology by Alan M. Vajda, Chair Larry B. Barber Chris Miller John Swallow July 3 0 2016

PAGE 4

iv B ertolatus, David ( MS Integrative Biology) Transcriptome profiling in the fathead minnow ( Pimephales promelas) following exposure to complex chemical mixtures: A landscape based approach Thesis directed by Assistant Profe ssor Alan M. Vajda ABSTRACT Anthr opogenic chemicals of emerging concern [CECs] are commonly detected in surface waters and several are known to cause adverse effects in aquatic vertebrates Although there is a large body of research documenting the effects of exposure to single chemicals in laboratory settings less is known about the effects caused by the complex mixtures of chemicals that occur in aquatic ecosystems To characterize these effects, we exposed adult fathead minnows ( Pimephales promelas ) to water from four different locatio ns within the Shenandoah River watershed (VA, USA) using flow through mobile laboratories. The exposure locations were chosen to capture unique and representative landuse in surrounding watersheds including agricultural, urban mixed use and forested. In addition to biological sampling, water samples were taken every 7 days during the fish exposure and analyzed for 460 chemical constituents. Each location had a unique chemical profile that was generally consistent with landuse in the surrounding watershed Whole organism and molecular responses also differed between the locations. Fish exposed at both agricultural and WWTP impacted sites had a reduced number of nuptial tubercles. At the agricultural site, survivorship was significantly reduced. Genome wide transcription profiles were measured to investigate the molecular underpinnings of these higher level changes and to gain an unbiased observation of the physiological state of animals following exposure. Differentially expressed genes and pathways were id entified

PAGE 5

v using ANOVA, g ene s et, and s ub n etwork e nrichment a nalyses. Transcript biomarkers of endocrine disruption, including er1, er2, ar, vtg1, and vtg3, showed little to no differential expression in exposed fish, suggesting these organisms did not expe rience estrogenic endocrine disruption. Pathways and sub networks related to immune function, cholesterol synthesis, and metabolism were affected by exposure at various sites. These data provide insightful hypotheses regarding the specific effects of expos ure to different types of complex mixtures and demonstrate the value of our complex mixture/landscape research approach. The form and content of this abstract are approved. I recommend its publication. Approved: Alan M. Vajda

PAGE 6

vi AK NOWLEDGMENTS This work w as funded by the U.S. Geological Survey's Contaminant Biology and Toxic Substances Hydrology programs This thesis would not have been possible without the collaboration, guidance, and support of a wide network of individuals too numerous to name individ ually I would like to specifically thank Dr. Alan Vajda for starting me on my path as a researcher and continuing to support my academic development Dr. Larry Barber of the U.S. Geological survey for his trust and support of me as a graduate researcher, Dr. Chris Martyniuk of the University of Florida for his large contribution of time and resources to my train ing I would also like to thank my committee members Dr. Chris Miller and Dr. John Swallow for providing guidance and feedback on th is thesis. Chemical laboratory analyses were preformed by collaborators at the following U.S Geological Survey branches: National Research Program Boulder, CO; National Water Quality Laboratory Denver, CO; Kansas Water Science Center Lawrence KS; and Cal ifornia Water Science Center Sacramento, CA The framework of this project and field logistics were organized by Dr. Larry Barber and Dr. Alan Vajda. The research activities described here have been reviewed and approved by the University of Colorado Inst itutional Animal Care and Use Committee, Protocol #92514(05)1E.

PAGE 7

vii TABLE OF CONTENTS Chapter 1. A LANDSCAPE FRAME WORK FOR ASSESSING T HE BIOLOGICAL IMPACTS OF CHEMICALS OF EMER GING CONCERN ................................ ............................... 1 1.1 INTRODUCTION 1 1.1.1 Chemicals of emerging concern ................................ ................................ ............. 1 1.1.2 Endocrine disruption ................................ ................................ .............................. 1 1.1.3 Mixture effects ................................ ................................ ................................ ....... 2 1.1.4 The landscape framework to addre ss complex mixture effects ............................. 3 1.2 RELEVANT BACKGROUND 4 1.2.1 Our chemical society ................................ ................................ .............................. 4 1.2.2 Emerging concern over impacts ................................ ................................ ............. 5 1.2.3 The Endocrine disrup tor hypothesis ................................ ................................ ...... 6 1.2.4 Complex mixture models ................................ ................................ ....................... 7 1.2.5 Regulatory and conceptual frameworks ................................ ............................... 10 1.2.6 Transcriptomics to study exposure to CECs ................................ ........................ 12 2. MOBILE LABORATORY EXPOSURE AND TRANSCR IPTOME PROFILING ... 16 2.1 INTRODUCTION 16 2.1.1 The Shenandoah Valley ................................ ................................ ....................... 16 2.1.2 Research objectives ................................ ................................ .............................. 17 2.2 Methods 19 2.2.1 Site descriptions ................................ ................................ ................................ ... 19

PAGE 8

viii 2.2.2 Mobile laboratory exposure and sampling ................................ ........................... 21 2.2.3 Water sampling & chemical profiling ................................ ................................ .. 22 2.2.4 RNA extraction & micr oarray analysis ................................ ................................ 22 2.2.5 Statistical analysis & bioinformatics ................................ ................................ ... 24 2.3 RESULTS AND DISSCUSSION 26 2.3.1 Chemical profiles ................................ ................................ ................................ 26 2.3.2 Morphological endpoint s & Vitellogenin ................................ ............................ 28 2.3.3 Differentially expressed genes ................................ ................................ ............. 29 2.3.4 Differentially expressed pathways in the liver of fathead minnows .................... 33 2.3.5 Hierarchical clustering ................................ ................................ ......................... 37 2.4 CONCLUSIONS 38 Re ferences 64

PAGE 9

ix LIST OF T ABLES Table 1 Genes selected a priori for analysis b ased on known responses to 57 environmental contaminants 2. Chemicals detected in the Shenandoah Valley, August Sept. 2014 58 3 Number of differentially expressed genes observed at each 61 exposure site, FDR corrected 4 Highly differentially regu lated pat hways identified by Gene Set 62 Enrichment Analysis at each of the three locations 5 Highly up and down regulated sub networks of genes as 63 identified by s ub n etwork e nrichment a nalysis

PAGE 10

x LIST OF F IGURES Figure 1. Map of the Shenandoah R iver watershed including locations of 41 mobile laboratory deployment 2. Nutrient loadings from four locations in the Shenandoah Valley 42 3. Pesticides detected in the Shenandoah Valley 43 4. Halogenated disinfection byproducts detected at site in the 44 Shenandoah Valley 5. Hierarchical clustering of water samples for each location based 45 on binary presence/absence data 6. Survivorship of fish exposed for 21 days 46 7. Gonadosomatic index of fish exposed to waters from the 47 Shenan doah watershed 8. Number of nuptial tubercles for fish exposed to waters from the 48 Shenandoah watershed 9. Relative concentration of plasma vitellogenin in fish exposed to 49 waters from the Shenandoah watershed 10. Venn diagram of differentia lly expressed genes 50 11. Expression of transcripts related to reproduction in the liver 51 12. Expression of transcripts involved in the aryl hydrocarbon 52 receptor pathway 13. Schematic of the mevalonate pathway 53 14. A schematic diagram of the IL6/STAT signal transduction 54 pathway 15. A schematic diagram of the classical complement pathway 55 16. Heatmap and hierarchical clustering of differentially 56 expressed genes

PAGE 11

xi ABBREVIATIONS AhR Aryl hydrocarbon receptor AOP Adverse outc ome pathway AR Androgen rec e ptor CEC Chemical of emerging concern CERC Columbia Environmental Research Center EDA Effects directed analysis DDT Dichlorodiphenyltrichloroethane DEG Differentially expressed gene DBP Disinfection byproduct EDC Endocri ne disrupting chemical EE2 17 ethinylestradiol ER Estrogen receptor FHM Fathead minnow GSEA Gene set enrichment analysis NRWD North River wastewater downstream NRWU North River wastewater upstream PCFH Passage Creek Fish Hatchery PCB Polychlorina ted biphenyl SFSR South Fork Shenandoah River SNEA Sub network enrichment analysis TIE Toxicity identification evaluations WWTP Wastewater treatment plant

PAGE 12

1 1. A landscape framework for assessing the biological impacts of chemicals of emerging concern 1.1 INTRODUCTION 1.1.1 Chemicals of emerging concern Over the past decades, a wide variety of anthropogenic chemicals have been identified in surface waters ( Focazio et al., 2008; Kolpin et al., 2002) These unregulated and unmonitored chemicals are referred to as contaminants of emerging concern [CECs] and include pharmaceuticals ( Daughton, 2001) pesticides ( Kolpin et al., 2013) exogenous hormones (Lange et al., 2002; Ternes et al., 1999) and other industrial chemicals (Blackburn and Waldock, 1995; Fromme et al., 2002; Suja et al., 2009) Several emerging contaminants have been shown to cause adverse bio logical outcome s in aquatic organisms at concentrations frequently found in the environment A dverse outcomes associated with CECs include abnormal growth and dev elopment ( Hayes et al., 2010; Jobling et al., 1998) altered behavior, increased mortality, and reduced fecundit y ( Kidd et al., 2007 ; Schwindt et al., 2014) These effects can lead to reduced fitness in individuals and ultimately to popul ation declines or extinction ( Kidd et al., 2007 ) 1.1.2 Endocrine disruption Endocrine disruption is a common mechanism t hrough which many CECs affect organisms. Virtually all aspects of vertebrate physiology are modulated by endogenous hormones that function at very low concentrations (Norris and Carr, 2006) Due to these low effective concentrations, endocrine signaling is also susceptible to disruption from low concentrations of exogenous molecules know n as endocrine disrupting chemicals

PAGE 13

2 [EDCs] (Crain and Guillette, 2000) The effects of endocrine disruption are often sub lethal and can result in adverse out comes not monitored by traditional screening programs Thus, it is important to design CEC studies to explicitly test for endocrine disruption and other subtle adverse effects 1.1.3 Mixture effects A significant body of research documents the effects of exposure to environmentally relevant concentrations of CECs in laboratory settings (reviewed in Pa!kov‡ et al., 2011; Santos et al., 2010) However, wildlife encounter CECs as components of complex mixtures that vary in time and space. The effects of ex posure to these mixtures may be different than the effects of exposure to a single constituent. Additive or antagonistic effects of exposure to CECs have been demonstrated and synergistic effec ts from complex mixtures are possible (Altenburger et al., 2012; Kortenkamp, 2007) Adding to this complexity, mixture constituents have multiple origins including point and non point source s (Daughton, 2001; Neumann et al., 2002) Point source s include effluents from municipal wastewater treatment plants [WWTPs] and industrial operations. Non point sources include runoff from a gricultural fields and urban areas. A better understanding of the biological consequences propagated by complex mixture s of CECs and the pathways through which the y enter the environment is needed for effective regulation and management.

PAGE 14

3 1.1.4 The landscape framework to address complex mixture effects To address this environmental complexity, we have developed a landscap e based framework for linking complex chemical mixtures to adverse outcomes across multiple levels of biological organization. At the core of this framework is the hypoth esis that the specific chemical mixture s present in a given body o f water are associat ed with the la ndscape patterning of the water shed. We reason that c hemicals are used in specific applications and therefore enter the environment through non random pathways that correspond to these applications. By characterizing the landscape of a water shed it might be possible to probabilistically predict the chemicals present in downstream waters. We can extend this association a step further to the biological effect s of exposure Physiological responses in organisms will also occur in non random ways based on the composition of the contaminant mixture in their environment Given a baseline understanding of the biological effects of exposure to various contaminant mixtures, it should be possible to pre dict biological outcomes from a contamination profi le Thus, our overarching hypothesis is that landscape patterning, watershed contamination, and biological effects are related and predictive models can be developed using these three variables However, a knowledge gap exists in understanding the biologic al effects caused by exposure to complex environmental mixtures. Transcript ome profiling is a promising tool for characterizing the biological consequences of exposure to CECs (Ankley and Daston, 2006; Villeneuve et al., 2012) T ranscript ome profil ing provide s information on a wide range of ph ysiological process that are affe cted by exposure to chemicals Bioinformatic tools are available to analyze transcription changes in thousands of genes and to identify cellular pathways and

PAGE 15

4 biological processes that are affected (Mehinto et al., 2012; Subramanian et al., 2005) These data are useful for three purposes : 1) examining patterns in biological responses between exposure site s, thus serving as an initial exploration of the landscape framework, 2) comparing an observed transcriptomic response to existing datasets involving simple exposures and well defined adverse outcome pathways and 3 ) generating hypothes e s regarding the mechanisms of action involved in expo sure to contaminant mixtures. 1.2 RELEVANT BACKGROUND 1.2.1 Our chemical society The large scale production of chemicals can be traced back to the industrial revolution in Europe. Indeed, t he ability to synthesize large volumes of specific molecules was integral in the development of a modern, technological society. Following World War II there was an exponential increase in the number and quantity of synthetic chemicals used in virtually all aspects of society. Currently, the Chemical Abst ract Service registry contains ov er 100 million unique chemicals that have been discovered or synthesized Of these, it is uncertain how many are currently used in commercial applications but estimates range from about 25,000 to 84,000 in the U.S (IOM, 2014) In 2005, the EPA tracked the manufacture or import of over 13 mi llion tons of chemical products and this number excludes fuels, pesticides, pharmaceuticals, and food products (U.S. EPA, 2009; Wilson and Schwarzman, 2009) Given the volume and ubiquity of chemicals in use, ecosystem contamination resulting in human and wildlife exposures is unavoidable.

PAGE 16

5 It should be stated that while industrial chem ical synthesis is a recent and human driven phenomenon, organisms have been exposed to exogenous chemica ls over evolutionary time scales For example, plants produce endocrine active phytochemicals that are capable of interfering with vertebrate reproducti on (Vajda and Norris, 2006) Organisms have evolved mechanisms for maintaining homeostasis in the face of a chan ging chemical environment including xenobiotic receptors capable of inducing expression of metabolizing enzymes (Hahn, 2002) Thus, the risk posed to organisms by anthropogenic chemicals may be less a product of the novelty of these molecules and more a product of the concentration and eco evolutionar y context in which exposure occurs. 1.2.2 Emerging concern over impacts Concern over the health effects in humans and wildlife from exposure to these chemicals greatly increased following the 1962 publication of Rachel C arson's Silent Spring, in which ecosystem degradation and human health impacts are attributed to the increasing use of pesticides, especially dichlorodiphenyltrichloroethane [DDT]. Silent Spring is often credited with catalyzing the modern environmental mo vement that began around this time. Whatever the cause, g rowing public concern about chemical contamination led to the formation of the Environmental Protection Agency in 1970 and soon after passage of the Clean Water Act in 1972 that gave the EPA authorit y to regulate point sources of contamination to surface waters. Around this time several commonly used chemicals including DDT and polychlorinated biphenyls [PC Bs] were banned in the U.S. after it was discovered that they were environmentally persistent, cap able of

PAGE 17

6 bioaccumulation and magnification, and posed significant toxicity risk to humans and the environment. Leading up to and following the ban of DDT and PCBs it was shown that these chemicals are capable of activating the estrogen receptor [ ER ] in v arious vertebrate tissues and interfering with the normal physiological processes mediated by estrogen signaling providing a mechanistic explanation for the observed adverse reproductive and developmental outcome s in wildlife (Bitman et al., 1968; Gaido et al., 1997) Since the mid 20 th century it has been known that c ertain synthetic chemicals can mimic the effects of endogenous est rogens often with negative biological outcomes Diethylstilbestrol [DES] was first synthesized in 1938 by Sir Charles Dodd as an estrogen mimic to study hormone action (Gilbert and Epel, 2009) It was widely prescribed as a therapeutic drug until is was linked with uterine cancer and abnormal reproductive morphology in women exposed in utero. Dodd also discovered that bisphenol A [BPA] is estroge nic in 1936, before it was widely incorporated into plastic production (Gilbert and Epel, 2009) BPA exposure has now been linked to many adverse effects including a bnorm al reproductive morphology and low sperm count in humans and wildlife (vom Saal and Hughes, 2005) 1.2. 3 The Endocrine disruptor hypothesis Toward the end of the 20 th century, increasing evidence of reproductive abnormalities in human and wildlife populations led to the creation of the endocrine disruptor hypothesis, which posits that anthropogenic chemicals are negatively affecting organisms through interaction with endogenous hormone signaling. In 1991 a group of scientists gathered at the Wingspread Confere nce where they coined the term "endocrine disruptor"

PAGE 18

7 and released a consensus statement saying that many chemicals in the environment are capable of interfering with normal endocrine function and that exposure to these chemicals is likely responsible for a dverse outcomes in wildlife and humans (reviewed in Colborn and Clement, 1992) Endocrine disruptors were defined as any substance that binds to or blocks binding of a hormone to a receptor, alters h ormone synthesis or metabolism, or aff ects the transport or excretion of an endogenous hormone. In the two decades since the Wingspread Conference an enormous body of research has been produced documenting the endocrine disrupting effects of many chemicals Much of this research focuses on the disruption of the hypothalamic pituitary g onadal [HPG] axis through modulation of estrogen and androgen signaling. It is now known that 17 eth i nylestradiol [EE2], PCBs, alkylphenolic compounds, phthalates, organochlo rine pesticides, and bisphenols can interact directly with vertebrate estrogen receptors (Kuiper et al., 1998, 1997; Lemaire et al., 2004; Nakai et al., 1999) Other chemicals are known to indirectly affect endocrine signaling thro ugh modulation of hormone synthesis (Cheshenko et al., 2008) Additional research has shown disruption of other hormone networks including glucoc orticoid, thyroid, and progestoge n signaling (Kumar et al., 2015; Suzuki et al., 2015; Tabb and Blumber g, 2006) 1.2.4 Complex mixture models It has long been recognized that mixtures of chemicals may have different effects on biological organisms than individual constituents (Berenbaum, 1989) These effects are generally classified as additive synergistic, or antagonistic. At a basic level, a dditive effects occur when the response to a mix ture is predicted by the sum of the effects

PAGE 19

8 expected from individual components. Synergis m occur s when the effect of the mixture is greater than expected from the sum of its individual components. Lastly antagonism occurs when the observed effect of a mix ture is less than the sum of expected effects of individual components. The type of mixture effect observed depends on the components of the mixture and their modes of action. Any two m ixture constituents can affec t a biological response thro ugh direct che mical chemical interactions, through interaction with the same bio molecule (e.g. two receptor agonists), or interaction with different biomolecules involved in the same process. Two general reference models have been developed to evaluate mixture effec ts : the dose or concentration addition model, and the independent action model (Kortenkamp and Altenburger, 1998) The concentration addition model was developed by Loewe and Muischneck for chemicals with similar modes of action. It predicts effects based on an additivity principle. As pointed out elsewhere, addit ive effects are no t simply the sum of individual effects for all mixture constituents at a given concentration, as this would not account for the non linear dose response curves observed for most chemicals (Berenbaum, 1989; Kortenkamp and Altenburger, 1999, 1998) Rather, the concentration addition model is based on iso effective doses or the concentrations at which two chemicals produ ce the same effect (Berenbaum, 1989) In this way chemicals in mixtures can be thought of as dilutions of each other. Experimental observations can be compared to model predictions to evaluate whether mixtures are producing synergistic, additive or antagonistic effects. Predictions based on the concentration addition model have proven accurat e for several classes of CECs based on experimental observations.

PAGE 20

9 The response addition or independent action concept provides another reference model for predicting mixture effects. This model is based on the concept of statistical independence and pred icts mixture effects based on the product of probabilities for individual constituent s to cause effects (Kortenkamp and Altenburger, 1998; Martin et al., 2009) The independent action model is used for mixtures where the constituents have different mechanisms of action (Spurgeon et al., 2010) and has also been experimentally validated in certain contexts. Although these models have been shown to generate accurate predictions in certain instances, their use in the study of complex environmental mix tures is limited for several reasons. First and foremost, both models require knowledge of the effects caused by individual chemicals and for many CECs this data is simply not available. The time and cost of individual exposure represents a major logistica l challenge given the thousands of chemicals that may be present in the environment. These approaches also require assumptions about the mechanism of action for mixture constituents to cho o se the appropriate model. Even if dose response data are av ailable for a chemical, its mec h an ism might be unknown. Additionally complex mixtures will inevitably contain sub mixtures that act on the same biomolecule and sub mixtures that act of different biomolecules involved in the same biological process and no models h ave been successfully developed for this scenario. Finally, neither of these models account for chemical chemical interactions. The limitations of predictive modeling have led to calls for different approaches for the study of complex environmental mixture s in recent years. Several recent papers have advocated using omics' based technologies to study complex

PAGE 21

10 mixture effects at the systems level (Garcia Reyero and Perkins, 2011; S purgeon et al., 2010) 1.2.5 Regulatory and conceptual frameworks Several experimental protocols have been developed in an attempt to overcome the disadvantages of purely mathematical modeling approaches, especially in regulatory and risk assessment con texts. Following the establishment of the Clean Water Act the EPA developed the toxicity identification evaluation [TIE] process to enable the identification of specific toxicants from contaminant mixtures (US EPA, 1991) The TIE protocol follows three phases; In phase I, organisms are exposed to whole mixtures to access overall toxicity Next, whole mixtures are manipulated to remove cl asses of chemicals and organisms are again exposed to manipulated waters. In this way the range of potential chemicals can be narrowed down to specific classes such as divalent cation metals, nonionic organics, or volatile chemicals (Burgess et al., 2013) The goal of phase II is to identify specific toxicants from within the chemical class identified in phase I, generally by measuring concentrations of candidate chemicals Phase III is a confirmation step where organisms are exposed to a candidate chemical to test whether that chemical alone is capable of producing the toxicity observed from the mixture. Effects directed analysis [EDA] is a similar approach proposed for identifying causal toxic constituents from complex mixtures (Brack, 2003; Burgess et al., 2013) EDA also begins with a toxicity test of whole mixtures, but may employ target ed in vivo or molecular assays in lieu of whole organism tests. The second step of EDA involves manipulations to narrow the range of offending chemicals, however, EDA uses more

PAGE 22

11 precise chemical fracti onation techniques to achieve a narrowed scope of potential chemicals (Brack, 2003) Fractionation is followed by chemical analysis, toxicity test ing, and potentially subsequent fractionation. Using fractionation makes it possible to get a much narrower and more precise list of candidate chemicals whereas the manipulations of TIE are only capable of identifying broader classes. This specificity comes at the e xpense of ecological relevance, as the fractionation processes can dramatically alter the bioavai labil ity of mixture constituents, thus the observed effects caused by fractions and sub fractions may not be relevant to the original ecosystem (Burgess et al., 2013) Both TIE and EDA have been used to successfully identify individual toxicants responsibl e for adverse outcomes in wild populations (Amato et al., 1992; Brack et al., 1999; Thomas et a l., 2001) However, there are limitations to these approaches for characterizing the effects of complex mixtures. Namely, these methods are largely designed to identify single chemicals or classes of chemical that have high acute toxicity. As such, they a re useful in heavily polluted sites where acute toxicity is high, but less so for the larger portions of habitat where chronic exposure to CECs occurs and adverse effects can be subtler (Reineke et al., 2002) Furthermore, the fundamental concept behind both TIE and EDA involves reducing the complexity of mixtures. Although this reductionist approach is certainly useful in certain conte xts, it ignores mixture effects that originate from chemicals in different classes and non specific modes of action (Brack, 2003) Incorporating omics' data into these approaches increase their applicability to landscape scale investigations of ecotoxicology by enabling measurement of fast acting biomarkers for a wide r range of effects although this add s time and cost to an already resource intensive protocol

PAGE 23

12 To help identify these responsive biomarkers capitalize on emerging omics data, and facilitate risk assessment Ankley et al. proposed the adverse outcome pathway [AOP] framework (Ankley et al., 2010) The AOP framework seeks to mechanistically link molecular, cellular and organismal data from chemical exposures to a n ecologically meaningful endpoint, the adverse outcome. T his approach increas es the relevance of commonly and easily measured endpoints and enab les the development of new cost effective and sensitive biomarkers An AOP be gins with a molecul ar initiating event where a biomolecule, such as a receptor or enzyme, interacts with a xenobiotic chemical. This interaction triggers a cascade of events that cross levels of biological organization that eventually results in an adverse outcome at the lev el of the organism, population or ecosystem The critical steps and thresholds along the pathway can then serve as biomarkers for the outcome itself enabling for efficient and cost effective monitoring This framework has been widely adopted by the ecotox icology community and moderate to well defined AOPs exist for several classes of contaminant s (Perkins et al., 2015) In complex mixture studies, transcriptome profiles can be compare d to known AOPs to provide clues regarding the chemical initiators and mechanisms of action propagated by the mixture. 1.2. 6 Transcriptomics to study exposure to CECs Transcriptomic studies, using microarrays and RNA sequencing, have been widely employed to study the effects of exposure to CECs However, the majority of these studies focus on single chemicals in laboratory settings. A handful of studies have examined the transcriptomic effects of exposure to complex mixtures in the fathead

PAGE 24

13 minnow [FHM] O f these, most focus on estrogenic EDCs associated with point source municipal WWTP effluent In the first published study using microarrays to measure effects of complex mixture exposure, Garcia Reyero et al. compared the transcriptome profiles from testi s of FHM exposed in wire cages above and below a WWTP in Minnesota (2008) They identified 119 genes unregulated and 128 genes down regulated in the downstream fish relative to the upstream fish. A hierarchical clustering analysis using DEGs grouped upstre am and downstream fish in separate clusters. This pioneering study demonstrated that fish transcriptomes can be altered in specific ways following exposure to complex mixtures. However, the biological interpretation of these changes is limited by the lack of a control or outgroup, e.g. it is not possible to resolve whether genes were downregulated in the downstream cohort or upregulated in the upstream cohort. C lustering based on DEGs between upstream and downstream samples all but guarantee d that individua l fish will cluster according to their exposure location, limiting interpretation of this analysis. (Garcia Reyero et al., 2008) Subsequent investigations of these WWTP impacted streams incorporated three additional sites ( Garcia Reyero et al., 2010) The three additional si tes included two downstream of WWTP s and one in an agricultural area Hierarchical clustering of male fish liver and gonad transcriptomes showed that expression patterns w ere correlated with exposure site, however there was no clustering of sites upstream or downstream of WWTP s thus it is uncle ar if transcriptome profiles are responding to differences in chemical contamination or other environmental variables. The above studies do not

PAGE 25

14 provide any chemical characterization of the exposure water, limiting the conclusion s that can be made in terms of direct CEC effects. (Garcia Reyero et al., 2010) Laboratory studies show that e xposure of male FHM to WWTP effluent EE2 or 17 trenbo lone [ TB ] alters the transcriptome of the liver and testis (Garcia Reyero et al., 2010) The transcription profile from fish exposed to WWTP effluent was more similar to that of fish exposed to EE2 than to TB, but ultimately WWTP effluent produced a unique transcription signature. The authors employed a function al enrich ment analysis to identify biological process es that were statistically overrepresented among DEGs. Several of the significantly enriched processes in effluent exposed fish are known to be affected by CECs, including hypoxia inducible factor [HIF] signaling oxidative stress response, and aryl hydrocarbon receptor [AhR] signaling (Simmons et al., 2009) Again, no data on the chemical composition of the effluent was provided. (Garcia Reyero et al., 2010) Another series of related studies measured gen ome wide transcription levels in FHM ovaries (Martinovi! Weigelt et al., 2014) and male livers (Berninger et al., 2014) from fish exposed to pure effluent from three different WWTPs, as well as waters above and below the effluent discharge Two of the locations used flow through mobile laboratory exposures, while a third sit e used static renewal exposures. Transcriptome profiles of individuals generally clustered based on exposure site and the transcriptomes of upstream and downstream exposed fish were more similar to each other than to effluent exposed fish. This suggests t hat factors other than effluent composition play a role in driving transcript ome patterns in fish below WWTP effluent. However, 43.4% of all DEGs overlapped between effluent and downstream fish, but not in upstream fish, suggesting that effluent exposure d oes affect transcription to some degree in downstream

PAGE 26

15 fish These studies also measured concent rations of 71 organic contaminants, including steroid hormones, pharmaceuticals and alkylphenols. Hierarchical clustering of chemical profiles was similar to th at of transcriptome profiles in that effluent sites were the most dissimilar and upstream and downstream sites clustered closely together (Martinovi! Weigelt et al., 2014) This is consistent with the hypothesis that exposure to CECs alters transcriptomes in fish. Taken together, these studies demonstrate the value of transcriptome profiling for complex mixture studies. Clustering of trans cript ome profiles largely corresponds with exposure site in all of these cases, demonstrating that the genome wide profile is responding in non random ways to site specific variables. Explicitly accounting for environmental variables in future studies will help identify the subset of transcriptional variation that is due to chemical exposure. Mobile laboratory exposures can remove some of these confounding environmental variables such as temperature and dissolved oxygen levels. Another point to make is tha t transcriptomic studies of c omplex mixture s can greatly benefit from chemical analysis of exposure water because this allows for direct comparison between chemical and transcriptome profiles which is necessary for generating hypotheses about specific dr ivers of transcriptome variation

PAGE 27

16 2. Mobile laboratory exposure and transcriptome profiling 2.1 INTRODUCTION 2.1.1 The Shenandoah Valley The Potomac River watershed represents a microcosm for investigating ou r landscape based framework The watershed contains a diversity of landuse types, including urban, suburban, agricultural and forested areas. Widespread fish kills involving several species of centrarchid fishes occu r red in the South B ranch of the Potomac River in 2002 and 2005 (Blazer et al., 2010) In Virginia, major fish kills wer e observed on the North Fork of the Shenandoah River in 2004 and the South Fork of the Shenandoah in 2005 Lower level mortality events involving young of year fish have occurred frequently from 2002 until at least 2010 (Blazer et al., 2010) Several studies have investigated the conditions associated with these fish kills. An increased frequency of skin lesions was observed in fish from areas with mortality events. Culturing of microorganisms from lesions found a variety of opportunistic bacterial and fungal pathogens. Because these pathogen s are also frequently found on healthy fish and because no pathogen was found on all fish or at all sampling locations, it was hypothesized that immunosuppression was contributing to the occurrence of lesions, rather than a single disease epidemic (Blazer et al., 2010) Interestingly, h istopa thological analysis found a high frequency of intersex fish, i.e. genetically male fish with immature oocytes in the testes (Blazer et al., 2010, 2007) High levels of intersex in populations of fish is generally associat ed with exposure to estrogenic EDCs. Estrogenic EDCs are frequently found in municipal WWTP effluent and agricultural run off. An investigation into the association between landscape variables and estrogenic EDCs found that density

PAGE 28

17 of animal feeding operat ions and the proportion of agriculture were the best predictors of estrogenicity in downstream waters, although these variables only accounted for a fraction of total variance (Young et al., 2014) Interestingly, there was no difference in the prevalence of intersex or vitellogenin levels in males between fish caught upstream and downstream of WWTPs (Iwanowicz et al., 2009) Taken as a whole, these data suggest that multiple causes contribute to the observed mortality events with CECs serving as a potential modulator of fish health. 2.1.2 Research objectives To characterize the effects of exposure to complex environmental mixtures and investigate linkages between landscape variables and contamination, we developed an integrated site assessment protocol. This protocol combines multiple landscape, chemical, and biological analyses at each sampling location. The measurements span multiple level s of biological organization including gross anatomical endpoints, tissue histology, and protein and transcript abundance. For this thesis I will focus on on the analysis of genome wide transcription data in support of the following objectives Ob jective 1: Measure genome wide transcription levels in FHM exposed to different environmental source waters and identify differentially expressed genes Transcriptome profiles were generated for individuals at each exposure location using 60,000 feature F HM microarrays. Individual genes that are affect by exposure can be identified by comparing exposed fish to a control or reference cohort. This first level of analysis is useful for identifying transcripts that show the highest response in relative fold

PAGE 29

18 ch ange, and to generate hypotheses regarding organismal implications and candidate chemical drivers. Objective 2: Determine the response of genes relevant to detoxification and endocrine disruption. In addition to a whole genome analysis for DEGs I employ ed a hypothesis driven approach to identify non rand om changes in transcription for select genes with known sensitivity to CEC exposure. Such genes include hormone receptors, xenobiotic receptors, and downstream targets of receptor response elements [Table 1 ] This approach allows for pairwise comparisons between all groups, which would be intractable for all 60,000 microarray features. These data will be useful for evaluating the impacts of commonly studied classes of CECs and legacy contaminants at these locations. Objective 3: Identify biological processes and molecular pathways differentially affected by specific exposures Statistical test s have been developed for identifying groups of genes that are disproportionately affected in an omics' data set. I dentifying molecular pathways or processes that are altered by exposure provides a greater biological context for interpreting transcriptome data. Pathway analysis is also helpful in identifying subtle changes in gene expression that may not be significan t in a DEG analysis due to small effect size, yet may have important biological implications. For example, if a n individual transcript is up regulated by 50% or less, it is unlikely to meet the significance threshold of a hypothesis test, especially when c orrections for multiple comparisons are used. However, if that gene and several related genes are all upregulated by 5 0%, the actual biological effect on an organism could be substantial.

PAGE 30

19 Objective 4: Use hierarchical clustering to explore patterns in tr anscription and chemical profiles at the different locations If our landscape hypothesis is correct, then changes in gene transcription and chemical profiles are expected to correlate with exposure site. Hierarchical clustering can be used to measure corr elations between individuals based on many res ponse variables. Significant clustering of transcription profiles based on exposure site will indicate a similarity in the transcription response to exposure Clustering of chemical profiles by site will indica te that each location ha d a relatively consistent and unique chemical fingerprint. This clustering would represent a proof of concept' for the landscape framework 2.2 Methods 2.2.1 Site descriptions Four locations withi n the greater Shenandoah River w atershed were selected to investigate linkages between landscape patterning, chemical contamination, and biological impacts. The locations were chosen to capture unique and representative landuse characteristics in the surrounding watersheds. The Shenandoa h River w atershed has a diversity of land uses (agricultural, municipal, domestic, industrial, forested, and recreational) in a relatively small geographic area (1372 mi 2 ). Logistical requirements such as road access and electricity prevented a random samp ling design. The Shenand oah River watershed is composed of two forks, the North For k and the South Fork, that flow northward until converging near Front Royal, VA [Fig 1] The North River is a tributary

PAGE 31

20 to the South Fork and Passage Creek is a tributary to the North Fork. Landscape characteristics are as follows: North River Wastewater Upstream [NRWU]. Two sites were selected on the North River that occurred upstream and downstream of municipal waste water treatment plant [WWTP]. The watershed of the upper North River site is dominated by intensive agriculture including row crop farming and confined animal feeding operations. Thus, the NRWU site was designated as the "agricultural" landuse site. North River Wastewater Downstream [NRWD]. The Harrisonburg Rock ingham Regional Sewer Authority operates a WWTP near Mount Crawford, VA. This plant processes the municipal wastewater of approx. 100 ,000 residents as well as several industrial operations. The average daily discharge is 13 14 million GPD. The water intake for this downstream site was placed in the North River directly below the outfall of the plant to capture a strong chemical signature from municipal wastewater. South Fork Shenandoah River State Park [ SFSR ]. The third site was located at the Andy Guest S henandoah River State Park, near Bentonville, VA. This is the largest watershed selected and it contains a variety of landuse types, including forested areas, agriculture, and small municipalities This location was designated the "mixed use" site. Passag e Creek Fish Hatchery [ PCFH ]. Passage Creek is a small tributary to the North Fork of the Shenandoah River. Its watershed is largely forested within the George Washington National Forest. The watershed also contains some rural

PAGE 32

21 development as well as disper sed agriculture. This location was designated the "reference" site, as it was expected to show the smallest anthrop o genic chemical impact. 2.2.2 Mobile laboratory exposure and sampling Mo bile laboratories were deployed to the four locations described ab ove. Water from each site was pumped from the main channel of each river into a head tank on top of each laboratory. From the head tank, water was fed into one of two splitter tanks w h ere it was heated to 25 1 ¡ C and then flo wed into individual 5 gallon glass aquaria. Water in aquaria was aerated with diffusion bubblers and the photoperiod was maintained at 14hr light, 10hr dark. Flow rates to individual aquaria were maintained at ~200 mL/min to provide replacement of 95% of the water volume approximatel y every 4 hours. Materials that contact exposure waters were limited to stainless steel, glass, or Teflon # tubing. Reproductively recrudescent 12 month old male FHM were provided by the Columbia Environmental Research Center [( CERC ); Columbia, MO ] Fish we re randomly assigned to exposure sites and to aquaria within sites. Five individ uals were placed in each aquarium Exposure began on August 13 th 2014 and continued for 21 days until September 2 nd 2014. Fish were fed laboratory fish chow provided by CERC. Fish were sampled on days 0, 7 and 21. The initial control cohort was sampled on day 0, without being exposed to environmental waters. Twenty male fish were sampled on day 0 and 10 male fish were sampled from each site on both day 7 and day 21. F ish were anesthetized with tricaine methanesulfonate (MS 222). Gross anatomical characteristics were measured including mass maximum total length, nuptial tubercle number and

PAGE 33

22 prominence, liver mass, gonad mass, and dorsal fat pad prominence (Jensen et al., 2001) Condition factor was calculated as (mass/length)*100. Gonadal somatic index [GSI] was calculate d as (gonad mass/total mass)*100. Blood w as collected from the caudal vei n in hep arinized capillary tubes Plasma was isolated via centrifugation and stored at 80 ¡ C until it was analyzed for vitellogenin protein by enzyme linked immunosorbent assay (ELIS A) using an anti FHM kit (Biosense; Bergen, Norway). Livers were dissected, weighed, and divided into three fragments Individual tissue fragments were preserved in 10% neutral buffered formalin, RNA later # (ThermoFisher Scienti fic), or snap frozen on dry i ce. 2.2.3 Water sampling & chemical profiling To characterize the chemical profile of the exposure waters, grab samples were collected every 7 days during the fish exposures. Grab samples were collected inside the mobile laboratories from water flowing out of the head tank. Multiple preservation methods were used from each sampling including acidification and filtering. Samples were placed on ice and shipped to 5 different analytical laboratories for analy sis. A total of 460 chemical constituents inclu ding parent compounds and metabolites, were measured for each sample using 17 different analytical methods. Chemical analytes included 109 pharmaceuticals 48 trace elements, 29 halogenated disinfection byproducts, 43 hormones and phytoestrogens and 135 pe sticides. 2.2. 4 RNA extraction & microarray analysis For the 7 day exposure cohort, t otal RNA was extracted from liver tissues preserved

PAGE 34

23 in RNA later # using Qiagen RNeasy plus mini kits according to the manufacture s instructions. RNA purity and concentrat ion were measured on a NanoDrop ND 1000 UV Vis Spectrophotometer. All samples had an A260/A280 ratio greater tha n 2.0. RNA quality was measured with a n Agi lent 2100 Bioanalyzer Average RNA integrity numbers [RINs] were 9.48 0.25 (mean SD). All R IN val ues were above 8.9. A sum sample of 31 RNA extracts (N = 7 or 8 /site) were prepared for microarray hybridization using Agilent One color Low Input Quick Amp Labeling kits according to the manufacturers instructions (Agilent Technologies v 6.9.1 Santa Clar a, CA ) Briefly, samples were spiked with Agilent One Color Spike Mix, diluted by a factor of 10,000. Ploy T, T7 primers were annealed to samples by incubating at 65¡ C for 15 minutes. cDNA was synthesized from sample mRNA using AffinityScript reverse tran scriptase. Samples were incubated at 40¡C for 2 hours immediately followed by 70¡C for 15 minutes. Cyanine 3 lab e led cRNA was synthesized from the cDNA templates by in vitro transcription using T7 RNA polymerase in the presence of cyanine 3 cytosine tri ph osphate. Samples were incubated at 40¡C for 2 hours. Amplified and labeled cRNA was then purified using a modified Qiagen RNeasy protocol. Following purification, samples were tested for purity, nucleic acid concentration, and Cyanine 3 concentration on a NanoDrop ND 1000 UV Vis Spectrophotometer. All samples had a specific activity (nmol cyanine 3/ g RNA) greater than 7. The FHM 8 x 60k oligonucleotide microarray slides ( GPL15775) were manufactured by Agilent Technologies. Each array contained 61,718 indiv idual probes. These probes contained 22,010 unique gene annotations Six hundred nano gram s of labeled RNA was used form each sample for hybridization Arrays were allowed to hybridize at 65 ¡ C for

PAGE 35

24 17 hours, then washed according to the manufacturers instruc tions. Microarrays were scanned at 5 m with the Agilent G2505 C Microarray Scanner and Agilent Feature Extraction Software v10.1.1.1 was used to extract raw signal intensities from microarray images. All microarray data were within acceptable quality control par ameters. Raw signal intensities were imported into JMP # Genomics v6.0 (SAS Institute Inc. Cary, NC) quantile normalized, and log 2 transformed The intensity of spike in and "dark spot" Agilent controls was used to determine the lower limit of detection fo r this assay. It was determined that 2.7 was the optimal lower bound and all values less than 2.7 were assigned this value 2.2. 5 Statistical analysis & bioinformatics A one way analysis of variance was used to identify differentially expressed genes [DEG s ] in each treatment group. A false discovery rate of 5% was used to correct for multiple testing. Fish from NRWU, NRWD and SFSR were compared to the reference PCFH. Probes with p < 0.05 following FDR correction were considered differentially expressed. AN OVA was p er formed in JMP Genomics (V 6.0 ). A subset of genes w as selected a priori for focused analysis based on their relevance to environmental contamination studies. Genes selected for focused analysis were categorized as "reproduction related," or "xeno biotic responsive [Table 1 ]. Signal intensities for all probes assigned to a given gene were aggregated and mean intensity was calculated. A bootstrapped resampling of the means was used to generate an empirical F statistic distribution based on 5 000 ite rations. The observed F statistic was

PAGE 36

25 compared to the empirical distribution to calculate a p value. For genes with p values < 0.05, pairwise comparisons were conducted using Tukey's HSD. Gene s et e nrichment a nalysis [GSEA] and s ub n etwork e nrichment a naly sis [SNEA] were p er formed in Pathway Studio 9.0 (Ariadne, Rockville, MD ) using the ResNet 10 .0 database (Nikitin et al., 2003) A total of 37,169 microarray probes (60%) were mapped to mammalian homologs. GSEA was p er formed on gene sets in the categories cell signaling, receptor si gnaling, cell process pathways, and metabolic pathways. Each category contains groups of genes known to be involved in a molecular pathway. The GSEA algorithm uses a modified Kolmogorov Smirnov test to identify pathways that are significantly up or down r egulated as a whole (Subramanian et al., 2005) SNEA was conducted using cellular processes as the seeds for su b network construction The SNEA algorithm then creates de novo sub networks based on database information linking genes to a seed' cellular process and identifies sub networks that are up or down regulated in experimental groups (Mehinto et al., 2012) A com plete linkage hierarchical clustering analysis using Euclidean distance was p er formed to identify patterns in expression profiles. For clustering, the dataset was reduced to probes that had been identified as differentially expressed in one or more treatme nt group s Probes that were annotated to the same gene symbol were averaged. An additional cluste r analysis was p er formed on the chemical profiles using binary presence/absence data and Jaccard distance Hierarchical clustering and bootstrapped ANOVA were p er formed in R (R Core Team, 2016)

PAGE 37

26 2.3 RESULTS AND DISSCUSSION 2.3.1 Chemical profiles A total of 149 of 460 chemical analytes were detected at one or more sites. This includes 47 trace elements, 20 neutral or acidic organic contaminants, 9 hormone s, 11 pesticides, 10 halogenated disinfection byproducts, 27 pharmaceuticals, and 3 antibiotics [Table 2] A total of 70 chemicals were detected at PCFH, 77 at NRWU, 91 SFSR, and 116 at NRWD. The majority of CECs detected were found at concentrations in t he low ng/L range. The chemical data is generally in line with our characterization of the landuse in surrounding watersheds, with a few notable exceptions. The fewest chemicals were detected at the PCFH reference site. Of 69 total detects at PCFH excludi ng nutrients, 61 were found at all 4 sites and the majority of these were trace elements, phytoestrogens and hormones. Several CECs were detected at this site including the pesticides fipronil, metalaxyl and simazine, as well as the pharmaceuticals metform in and diphenhydramine. PCFH also had the lowest concentration of nutrients, including nitrate, nitrite, ammonia and phosphorous [Fig 2 ]. These findings are consistent with a largely forested watershed with limited agriculture and residences. At the two North River locations, surprisingly, more pesticides were found at NRWD than at NRWU, with 8 and 6 detections, respectively [Fig. 3 ]. Four pesticides were detected at SFSR. Atrazine, metolachlor, and simazine occurred at similar concentrations between NRWU and NRWD whereas concentrations of imidacloprid and azoxystrobin were higher at NRWD. Boscalid, diuron, and fluxapyroxade were detected only at NRWD. Based on the highly agricultural landuse patterns in the NRWU watershed we expected to find more pestic ides at NRWU than at the other locations. It is possible that

PAGE 38

27 some of the pesticide load detected at NRWD originated from upstream and not WWTP effluent, however this cannot explain the detection of chemicals at NRWD that were not present at NRWU. In addit ion, this phenomenon is unlikely to account for the total concentration of chemicals detected at both sites because upstream waters were heavily diluted with WWTP effluent, and thus concentrations measured at NRWD should be lower than NRWU if the only sour ce of pesticide was from upstream. Therefore, it is reasonable to conclude that the WWTP effluent per se is the source of much of the pesticide load observed at NRWD. Previous studies have also detected numerous pesticides in WWTP effluent in Europe (Kšck Schulmeyer et al., 2013; Kuster et al., 2008) The HRRSA WWTP influent includes wastewater from an industrial facility that processes agricultural products. This facility is a likely source of some of the pesticides observed at NRWD. In addition, pesticides may enter the WWTP influent through urban run off from lawns and gardens or from human waste (Hill et al., 1995) NRWD had the most pharm aceutical detections at 22, while SFSR had 17, and NRWU had 6. This is consistent with our expectation that WWTP effluent would contain the highest number of pharmaceuticals. The high number of detections at SFSR indicates that this location may be impacte d by upstream WWTP effluents which is consistent with the mixed use character of the watershed. Of 29 halogenated disinfection byproducts [DBPs] tested for 10 were detected at NRWD [Fig 4 ]. No DBPs were found at the other sites, with the exception of 1 d etection of trichloromethane at PCFH. This indicates that NRWD was indeed highly impact by the nearby WWTP effluent. The fact that no DBPs were detected at NRWU confirms that this location was outside the upstream mixing zone of the WWTP efflu ent and there fore,

PAGE 39

28 contaminants detected at this location originated from up stream and not from the nearby but downstream effluent. Hierarchical clustering of chemical profiles based on binary presence/absence data showed sample s from each site clustering together [Fi g 5 ] d emonstrating that each location had a unique chemical profile that was driven more by the specific location rather than by sampling time. Additional sampling of other watersheds with similar landuse profiles is needed to evaluate the degree to whic h contaminant profiles can be predicted from landscape variables. NRWD samples were the most unique relative to the other locations. This is likely due to the presence of numerous DBPs detected only at this site. The reference PCFH profile was unique comp ared to NRWU and SFSR, which clustered together based on chemical composition This indicates there is more similarity in contaminant profile between our agricultural and mixed use watersheds than between the reference watershed and either of these two. Ov erall, the chemical data provides proof of concept for our landscape framework. The watersheds we selected based on landuse each had unique contaminant profiles that wer e often consistent across our 4 week sampling period. There were several deviations fro m ou r predictions, such as the high prevalence of pesticides at NRWD. This may reflect the integrated nature of agricultural and industrial systems, which will present a unique challenge for linking landuse to river contamination. 2.3.2 Morphological endp oints & Vitellogenin Survivorship was decreased at the NRWU location, where 52% of individuals died during the course of the 21 exposure (ANOVA, p > 0.05) [Fi g. 6 ] Survivorship was high

PAGE 40

29 ( $ 90 %) at all other sites. For the 7 day cohort there were no diffe rences in GSI, nuptial tubercle abundance or plasma VTG (p > 0.05) [Fig 7 & 8 ]. GSI and nuptial tubercle abundance were signifi cantly reduced at NRWU and NRWD The magnitude of reduction was greater at NRWD for both GSI and nuptial tubercles. This reducti on in GSI and secondary sex characteristics is consistent with previous studies of WWTP effluent (Barber et al., 2012; Vajda et al., 2011) Plasma VTG was significantly elevated at PCFH [Fig. 9] although the magnitude of this change was relatively small compared to other studies of endocrine disruption (see 2.3.3 below ) Dec r eases in GSI and nuptia l tubercles are often associated with exposure to EDCs, in ma n y cases estrogenic substances (Vajda et a l., 2011) However, the lack of a corresponding VTG induction and low concentrations of measured estrogens at these locations makes an estrogenic mechanism of action unlikely Reduction in nuptial tubercle abundance can also be caused by anti androgens (Panter et al., 2004) 2.3.3 Differentially expressed gene s Individual genes were identified a s up or downregulated using PCFH as the reference The number of D EGs was similar between all sites. Fish exposed to waters at NRWU displayed 202 DEGs, whe reas fish exposed at NRWD displ a yed 297 DEGs, and fish exposed at SFSR had 219 DEGs [Table 3 ]. At each site, the largest number of DEGs were only identified as differe ntially expressed in that location, suggesting that each exposure site had a unique transcriptional response to exposure [Fig 10 ] The NRWU and NRWD site s had the most DEGs in common at 67. The high level of overlap between DEGs at the two North River Sit es likely reflects the environmental similarity

PAGE 41

30 between these locations. This overlap could be caused by a response to CECs or other environmental variables that were similar between the two sites. Given the number of DEGs at each site it is not possible or meaningful to discuss all of the individual genes that were affect ed by exposure. Therefore, I will limit reporting and discussion to the gen e s with the largest fold change relative to the reference and genes that have known response s to CECs and thus w ere hypothesized to change a priori. At the mixed use SFSR site the three most up regulated genes in terms of fold change were D site of albumin promoter (albumin D box) binding protein b [ dbpb], period 1 [ per1 ], an d nuclear receptor subfamily1group D m ember 1 [ nr1d1 ]. The three most down regulated genes were: nuclear factor, inter leu kin 3 regulated, member 5 [ nfil3 5 ]; protein arginine methyltransferase 6 [ prmt6 ]; and ankyrin repeat domain 10a [ ankrd10a ]. At the agricultural NRWU site the most up regula ted genes were potassium voltage gated channel subfamily E member 4 [ kcne4 ], EGL nine homolog 3 [ egln3 ], and nuclear receptor subfamily1group D member 1 [ nr1d1 ]. The most down regulated genes were myelocytomatosis oncogene homolog [ mycn ], transmembrane pro tein 80 [ tmem80 ], and DNA cross link repair 1A [ dclre1a ]. At the WWTP impacted NRWD site the most up regulated genes were heat shock protein B8 [ hspb8 ], potassium voltage gated channel subfamily E member 4 [ kcne4 ], and solute carrier family 16, member 3 [ s lc16a3 ]. The most down regulated genes were hydroxysteroid dehydroxenase 4 [ hsd17b4 ], Cytochrome P450 Family 26 Subfamily B Polypeptide 1 [ cyp26b1 ], and alanine glyoxylate aminotransferase b [ agxt ]. It is interesting to note that among these highly regul ated DEGs there is overlap between locations. All three of the high est up regulated genes at SFSR ( db p b per1,

PAGE 42

31 nr1d1 ) were also identified as differentially expressed at NRWU and NRWD. This overlap could represent a similar response to the CECs that were present at all three locations. Alternatively, this might be a consequence of similar nutrient levels at these sites or other environmental factors that are shared at these locati ons but not at PC FH. Activating transcription factor 3 [ atf3 ] was highly up regulated at NRWD. Atf3 is a stress responsive gene that is up regulated following a variety of physical and chemical stressors (Hai et al., 1999) Its immediate role in the stress response is not fully known. Of the chemicals detected at NRWD acetaminophen (Stamper et al., 2015) carbamazepine (Schulpen et al., 2015) metformin (Limonciel et al., 2014) and nic otine (Malpass et al., 2014) have been shown to up regulate the expression of atf3. Nicotine was detect ed at all 4 sites and concentrations were lower at NRWD compared to the other 3 locations. T herefore nicotine is unlikely to be responsible for the up regulation of this transcript which was only detected at NRWD. Carbamazepine is an anti convulsant pharm aceutical prescribed to patients with epilepsy and is commonly detected in WWTP effluents. C arbamazepine significantly increased the expression of atf3 after 1 and 7 days in human stem cells (Schulpen et al., 2015) Metformin is a pharmaceutical prescribed to patients with type 2 diabetes It was detected at all sites, but in the highest concentration at NRWD. Atf3 was up regulated in primary rat hepatocytes after exposure to metformin (Limonciel et al., 2014) Acetaminophen is an over the counter analgesic and antipyretic medication which up regulated the transcription of atf3 in vitro (Stamper et al., 2015) Acetaminophen was only detected at NRWD. Interestingly, ATF3 can bind the androgen receptor [AR] and inhibit androgen signaling, thus it is possible that the increase in atf3

PAGE 43

32 observed in NRWD fish may be involved in the decease of nuptial tubercles seen in fish exposed for 21 days. Heat shock protein a8 [ hspa8 a.k.a. hcs70 ] was also highly induced at NRWD. Hspa8 is generally considered to be a constitutively expressed member of the heat sho ck family, unlike the highly stress inducible hspa1a and hspa1b (Chen et al., 2006; Liu et al., 2012) Of the chemicals detected at NRWD, acetaminophen has been shown to up regulate the expression of hspa8 in human liver slices (Elferink et al., 2011) Among the transcripts related to reproduction, only vtg 1 displayed differential expression following exposure [Fig. 11 ] Vtg was upregulated 2.4 fold at PCFH compared to SFSR (p= 0.04 92) and this is consistent with the VTG protein data. The fold change in vtg expression is relatively small compared to previous studies of estrogenic endocrine disruption at WWTPs (Bahamonde et al., 2015; Garcia Reyero et al., 2010) It is not clear w he ther thi s finding represents vitellogen in inducti on at the PCFH site or is part of a normal background fluctuation in the levels of vtg transcript in male FHM Following careful examination of these genes as a whole there is little evidence of reproductive endocrine disruption at these sites. This findi ng is especially surprising for NRWD which was heavily impacted by WWTP effluent which often contains EDCs (Barber et al., 2011; Vajda et al., 2011) These results are consistent with a yeast estrogen screen of water samples which found no estrogenic ity above the detection limit of 0.025 ng/L E2 ( Iwanowicz, pers. comm. ) Furthermore, this finding is consistent with the observed concentrations of known EDCs from water samples at these sites. The transcript and protein VTG data indicate the reduction in nuptial tubercles at NR WU and NRWD was

PAGE 44

33 not caused by estrogenic EDCs but may be attributable to non estrogenic endocrine disruption. Among transcripts involved in aryl hydrocarbon receptor [AhR] signaling, cytochrome p450 1a 1 [ cyp1a1 ] was upregulated 2. 8 fold at NRWD compared to the reference PCFH (p < 0.001) [Fig 12 ] No significant differences were observed between sites in other AhR pathway genes including ahr1, ahr 2, arnt1, arnt2, and cyp1b1 Cyp1a1 also known as aryl hydrocarbon hydroxylase, is an enzyme involved in the catabolism of certain xenobiotic organic molecules. Its expression is induced by the AhR/ARNT complex following binding between AhR and a ligand. Tetrachlodibenzodioxin [TCDD] is the most well known and commonly studied AhR ligand, although a number of other emerging and legacy contaminants are known to be AhR agonists (Denison and Nagy, 2003) Of the chemicals detected only at NRWD, the pesticid e d iuron is known to increase the expression of cyp1a1 in mammals (Ihlaseh et al., 2011) Other chemicals detected at NRWD and other sites are also known to increase the expression of cyp1a1 including 4 nonylphenol (Lee et al., 2005) and carbamazepine (Oscarson et al., 2006) 2.3.4 Differentially expressed p athways in the liver of fathead minnows The GSEA analysis identified multiple differentially expressed pathways at each expos ure site. There were 54 affected pathways at SFSR and 58 affected pathways at both NRWD and NRWU [Table 4 ] SNEA analysis identified 130, 138, and 113 sub networks that were significantly affected by exposure at NRWD, NRWU, and SFS R, respectively [Table 5 ] Altered pathways and sub networks in general included carbohydrate, glucose, and sterol metabolism, as well as pathways related to immune

PAGE 45

34 function. Below, the discussion focuses on those related to metabolism and the immune system. At both NRWD and NRWU the two most down regulated pathways were biosynthesis of cholesterol and the mevalonate pathway a precursor to cholesterol synthesis [Fig 13 ] This finding was corroborated by SNEA, which identified the mevalonate pat hway and sterol biosynthesis as do wn regulated sub networks at both of these locations These same gene sets were up regulated at SFSR although with a smaller median fold change compared to that observed at NRWU and NRWD This finding at NRWD is in contrast to a previous study that identi fied significant up regulation of a cholesterol biosynthesis gene set in the livers of FHM exposed to several WWTP effluents and downstream waters a lthough in this study a different enrichment analysis and microarray w ere used (Martinovi! Weigelt et al., 2014) This discrepancy could be a result of different contaminant exposures as Martinovi Weigelt et al. also found evidence of estrogenic endocrine disruption in effluent and effluent imapcted sites which was not detected i n the present study (2014). Garcia Reyero and colleagues also identified enrichment of the gene ontology categorey "cholesterol biosynthetic process" in livers of FHM exposed to waters downstream of a WWTP althought t he direction of this change was unclea r (2009) A study investigating the transcriptiomic response of fish exposed to river water from highly agricultural areas found d own regulation of the gene ontology category cholesterol biosysnthesis (Sellin Jeffries et al., 2012) suggesting that agricul tural chemicals could be involved in the suppression of these pathways and sub networks at NRWD and NRWU. Pathways related to carbohydrate metabolism were significantly up regulated at NRWU. Glycogen and glucose metabolism were among the

PAGE 46

35 highest up regulat ed gene sets [Table 4 ]. This was corroborated by SNEA which identified glycogen metabolism, glycogenesis, carbohydrate utilization, glucose import and gluconeogenesis as significantly up regulated. It is becoming clear, through this study and others, that organismal metabolic processes are comprimised by WWTP effluent and there is a significant transcriptional response involved. Pathways invo lved in immune function and pathogen resistance were down regulated at NRWU, including Interleukin 6/ STAT [IL6/STAT] signaling (median fold change[fc] = 1.4 1 NES = 1.59, p = 0.03) [Fig. 14 ] as well as classical (median fc = 1.40, NES = 1.85 p < 0.001) [Fig. 15 ] alternative (median fc = 1.40, NES = 1.95, p = 0.004) and lectin induced complement pathway s (median f c = 1.4, NES = 1.98, p < 0.001). IL6/ STAT i s a signal transduction pathway whereby interlukin 6 a pleiotropic cytokine, binds with a receptor complex, leading to the phosphorylation and dimerization of STAT transcription factors which then regulate the expression of responsive genes (Aaronson and Horvath, 2002) In mammals t his pathway mediates several immune related processes including proliferation of B and T cells, the acute phase response and infla mmation (Heinrich et al., 1998) and similar functions have been observed in fish (Uribe et al., 2011) The complement system is part of the innate immune response and mediates the recognition and elimination of invading microbes or damaged host cells (Zipfel and Skerka, 2009) In fish, the complement syste m is thought to play a larger rol e in immune defense compared to adaptive immune responses. There are three mechanisms that activate the complement cascade, the classical pathway, the alternative pathway, and the lectin induced pathway. All three of these pathways were down regulated in NRWU fish. Interestingly, as part of the acute phase response in mammals, IL6 activates the

PAGE 47

36 expression of several complement proteins, including C3, C4, and C9 (Gabay and Kushner, 1999; Steel a nd Whitehead, 1994) Thus, the down regulation of IL6/STAT signaling may be mechanistically related to the suppression of complement pathways in the livers of FHM exposed at NRWU This effect on immunosuppression is a n interesting candidate pathway potent ially contributing to the increased mortality among NRWU exposed fish. Several chemicals detected at NRWU are known to down regulate IL 6 expression and/or secretion based on mammal in vitro assays (Medja kovic et al., 2010) The se include the isoflavones bio chanin A formononetin diadzein and the diadzein metabolite equol which were all detected either only at or in the highest concentration at NRWU. Biochanin A decreased expression of IL 6 in osetoblas tic murine cells following an H 2 0 2 inflammatory challenge (Lee and Choi, 2005) Biochanin A and equol decreased the secretion of IL 6 from LPS stimulated macrophages (Mueller and Jungbauer, 2008) Diadzein decreased the secretion and transcription of IL 6 in a dose dependent manner in LPS stimulated RAW264.7 cells (Choi et al., 2012) Therefore, these isoflavones serve as candidate causal agents for the observed decrease of the IL 6/STAT pathway at NRWU. Immune processes can also be inhibited by generalized stress responses. Unfortunately, less data are available regarding the interactions between chemicals and whole pathways than there are for individual genes, so this discussion is limited to chemical mediators of IL 6 itself. The most prevalent source of isoflavones is legumes, especially soy and red clover (Medjakovic et al., 2010) Formononetin and biochanin A are found predominantly in red clover while geni stein and diadzein are more abundant in soy (Clarke et al., 2008) Soy is

PAGE 48

37 frequently grown in the agric ultural NRWU watershed. In 2012, 9,847 acres of soybeans were harvested in Rockingham Count y, VA, were NRWU is located (USDA NASS, 2012) Red clover is frequently used as a cover crop on agricultural fields, but as such no official statistics are kept at the county level. The fact these chemicals were found predomin antly at NRWU suggests a non point agricultural origin as opposed to a point source WWTP origin. At the mixed use SFSR site, pathways related to immune function were the most strongly up regulated. Several interleukin/STAT pathways in the common % chain f amily were increased, including IL2/STAT, IL7/STAT, IL9/STAT, IL15/STAT and IL21/S d TAT. These pathways share the receptor component IL2G also known as the common % chain. The initiating cytokines and distinguishing receptor components were not on the micro array, thus it is not possible to resolve w he ther one specific pathway was increased relative to the others or if all these pathways were up regulated simultaneously The common % chain pathways are generally involved in growth and proliferation of immune cells, such as T B and NK cells (Akdis et al., 2011) 2.3.5 Hierarchical clustering Hierarchical clustering of differentially expressed genes was used to explore the overall response to exposure and identify patterns between sites [Fig 16 ]. The first level of cluster ing separated all the NRWD fish and 3 NRWU fish from the PCFH fish, the SFSR fish and the remaining NRWU fish. All the NRWD fish clustered together with the exception of 1 that clustered with the 3 NRWU fish. All the PCFH fish clustered together, which is perhaps not surprising given that DEGs were identified by comparison to this

PAGE 49

38 reference site. The remaining NRWU and SFSR fish divided into two clusters based on transcriptomic response, each containing individuals from both sites. NRWU showed the most variable response to exposure with individ uals clustering amongst NRWD and SFSR fish. The NRWD fish had a largely consistent response that showed the largest distance from the majority of the other DEG profiles. The NRWD contaminant profiles also had the largest difference from other sites, thus t here is a similar pattern between total contaminant profile and overall transcriptomic response at NRWD. This pattern is also observed to a degree between NRWU and SFSR, where contaminant and transcription profiles from these sites cluster together, althou gh there are differences within cluster patterning. Overall there are similarities in the clustering of contaminant profiles and the clustering of transcription profiles, suggesting that the unique mixture of CECs in exposure water may be driving part of the total transcriptional response i n fish, although this cannot be quantitatively established at this time. Improved systems level analysis techniques will be needed to formally evaluate linkages between contaminant mixtures and transcriptional response. 2.4 CONCLUSIONS This study serves as a proof of concept for our landscape based framework. E ach site had a unique contaminant profile an d exposed fish displayed unique biological responses Many of the contaminants detected matched expectations based on landuse, such as a high preponderance of pharmaceuticals and disinfection byproducts at the WWTP impacted NRWD high levels of pesticides at the agricultural NRWU and a moderate level of pesticides and pharmaceuticals at the mixed use SFSR. Deviations f rom

PAGE 50

39 expectations such as the high preponderance of pesticides at NRWD highlight the complexity of chemical occurrence and transport in a highly industrialized society. Additional variables beyond basic landuse should be included in future studies investiga ting the linkage between landscapes, river contamination and biological effects. Additional data are needed to quantitatively evaluate the power of landscape variables for predicting contamination and associated biological effects The use of mobile labor atories greatly reduces the number of confounding variables in environmental mixture studies. Effects in exposed fish can be attributed to differences in water composition, however this does include factors other than chemical contaminants Differences in the base chemistry of water sources and microbial communities may contribute to observed effects. Measuring or controlling for these variables in future studies will increase our ability to attribute effects to contaminant mixtures per se. This study furth er validates the use of transcriptome profiling in investigations of chemical mixtures. Microarray analysis revealed a diverse and unique transcriptional response at each exposure site, demonstrating the opportunity for transcriptome profiling to character ize a range of responses without being limited to a narrow range of biological endpoints established a priori The genes and pathways identified as differentially expressed in this study provide valuable mechanistic information regarding the effects of exp osure to environmental mixtures. For example, transcript data were useful for evaluating the underlying causes of the observed decrease in nuptial tubercles at NRWU and NRWD. This evaluation was facilitated by the well defined AOP for estrogenic initiating events. For adverse outcomes with less defined AOPs, such as the mortality at

PAGE 51

40 NRWU, transcriptome profiling provided hypotheses for causal mechanisms. These hypotheses will be useful for characterizing new AOPs Additionally, t he large amount of data avai lable from single chemical laboratory exposures makes it possible to identify potential chemical initiators. The hypotheses generated regarding systems level changes can now be evaluated with specific ass ays. Although the use of transcript and other omics techniques in field settings is increasing, it is still a relatively new area of research. Additional method ologies are needed to more effectively determine quantitative linkages between changes in transcriptome profiles and the constituents of a complex mixture. Development of these techniques will be facilitated by more studies that capture as much chemical and environmental data as possible These rich datasets can help pave the way to a more predictive eco t oxicolo gi cal science.

PAGE 52

41 Figure 1. Map of th e Shenandoah River watershed including locations of mobile laboratory deployment. !"#$ %&'( %&') *#*&

PAGE 53

42 Figure 2 Nutrient loadings from four locations in the Shenandoah Valley. The reference PCFH locations had the lowest total nutrient load. The agricultural NRWU and the WWTP impacted NRWD had the highest nutrient loads. Bars represent the mean and points are measurements from individual samples (N = 4/site) NRWD NRWU PCFH SFSR 0.00 0.02 0.04 0.06 0.08 0.10 Total P NRWD NRWU PCFH SFSR 0.00 0.02 0.04 0.06 0.08 0.10 Phosphate NRWD NRWU PCFH SFSR 0.00 0.02 0.04 0.06 0.08 0.10 Ammonia NRWD NRWU PCFH SFSR 0.00 0.02 0.04 0.06 0.08 0.10 Nitrite NRWD NRWU PCFH SFSR 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Nitrate NRWD NRWU PCFH SFSR 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Total N NRWD NRWU PCFH SFSR 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Nitrate + Nitrite NRWD NRWU PCFH SFSR 0.0 0.1 0.2 0.3 0.4 0.5 Organic N Nutrients in the Shenandoah watershed, August September 2014 Concentration (mg/L)

PAGE 54

43 Figure 3 A total of 11 pesticides were detected at one or more of the locations. PCFH had the lowest number of pesticide detections. More pesticides were detected at the WWTP impacted NRWD than the agricultural NRWU. Bars represent the mean and points are measurements from individual samples (N = 4/site). NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Atrazine NRWD NRWU PCFH SFSR 0 20 40 60 80 100 120 140 Azoxystrobin NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Boscalid NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Fipronil NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Fluxapyroxade NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Iprodione NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Metalaxyl NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Metolachlor NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Simazine NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Imidacloprid NRWD NRWU PCFH SFSR 0 5 10 15 20 25 30 Diuron Pesticides detected in the Shenandoah watershed, August September 2014 Concentration (ng/L)

PAGE 55

44 Figure 4. Halogenated disinfection byproducts detected a t site in the Shenandoah Valley. DBPs were detected predominately at NRWD, with the exception of one detection of trichloromethane at PCFH. Bars represent the mean and points are measurements from individual samples (N = 4/site) NRWD NRWU PCFH SFSR 0 200 400 600 800 Bromodichloromethane NRWD NRWU PCFH SFSR 0 200 400 600 800 Dibromochloromethane NRWD NRWU PCFH SFSR 0 200 400 600 800 Dichloroacetonitrile NRWD NRWU PCFH SFSR 0 200 400 600 800 Bromochloroacetonitrile NRWD NRWU PCFH SFSR 0 50 100 150 200 250 300 350 Trichloromethane NRWD NRWU PCFH SFSR 0 50 100 150 200 250 300 350 Tribromomethane NRWD NRWU PCFH SFSR 0 50 100 150 200 250 300 350 Trichloroacetaldehyde NRWD NRWU PCFH SFSR 0 50 100 150 200 250 300 350 1,1,1 trichloro 2 propanone NRWD NRWU PCFH SFSR 0 20 40 60 80 100 Dichloroiodomethane NRWD NRWU PCFH SFSR 0 20 40 60 80 100 Dibromoacetonitrile Disinfection by products detected in the Shenandoah watershed, August Setember, 2014 Concentration (ng/L)

PAGE 56

45 Figure 5 Hierarchical clustering of water samples for each location based on binary presence/absence dat a The contaminant profiles from each location cluster together, indicating there is more similarity within locations than between different locations. NRWD had the most uni que profile. Agricultural = NRWU, WWTP = NRWD, Mixed Use = SFSR, and Reference = PCFH. 0.0 0.1 0.2 0.3 0.4 0.5 Hierarchical clustering of chemical profiles (presence/absence) WWTP 1 WWTP 2 WWTP 3 WWTP 4 Reference 4 Reference 2 Reference 1 Reference 3 Agricultural 2 Agricultural 4 Agricultural 1 Agricultural 3 Mixed Use 4 Mixed Use 3 Mixed Use 1 Mixed Use 2

PAGE 57

46 Figure 6 Survivorship at PCFH, NRWD, and S FSR was 90%, but was significantly reduced to 48% at NRWD by exposure Day 21 (ANOVA, p < 0.05). !"# !"$ !"% !"& '"# # ( $ ) % & + '! '' '# '( '$ ') '% '* '& '+ #! #' ,-./ 0123 0124 5.51 567898:7;<9= 4>?

PAGE 58

47 Figure 7 Gonadosomat ic index was decreased at NRWU and NRWD for 21 day exposed fish compared to the initial controls (ANOVA, p < 0.05 N = 10/factor ) The middle bar is the median. Upper and lower boundaries of the box are the 3 rd and 1 st quartiles, respectively and whiskers are the maximum and minimum values. Asterisk s indicate significant difference from initial control [IC] at p < 0.05. IC PCFH d7 NRWU d7 NRWD d7 SFSR d7 PCFH d21 NRWU d21 NRWD d21 SFSR d21 0 1 2 3 4 5 GSI Gonadosomatic Index

PAGE 59

48 Figure 8 The number of nuptial tubercles was reduced at NRWU and NRWD in 21 day exposed fish compared to initial controls (ANOVA, p < 0.05 N = 10/factor ) Nuptial tubercles are a secondary sex characteristic that are modulated by androgen signaling. Error bars = SEM. The middle bar is the median. Upper and lower boundaries of the box are the 3 rd and 1 st quartiles, respectively and whis kers are the maximum and minimum values. Asterisk s indicate significant difference from initial control [IC] at p < 0.05. IC PCFH D7 NRWU D7 NRWD D7 SFSR D7 PCFH D21 NRWU D21 NRWD D21 SFSR D21 0 10 20 30 40 Number of Tubercles Nuptial Tubercles

PAGE 60

49 Figure 9 Vitellogenin protein was statistically increased at PC F H compared to the initial control cohort (ANOVA, p > 0.05 N = 1 0/factor ) The magnitude of this change is relatively small compared to other studies of estrogenic WWTP effluents. Bars represent the mean and points are measurements from individual samples. Asterisks indicate significant difference from initial control [IC] at p < 0.05. IC PCFH d7 NRWU d7 NRWD d7 SFSR d7 PCFH d21 NRWU d21 NRWD d21 SFSR d21 -1 0 1 2 3 Plasma Vitellogenin Log vtg ng mL -1

PAGE 61

50 Figure 10 Venn diagram of differentially expressed genes with annotations and the overlap between sites. For each site the largest groups of DEGs is unique to that exposure location (Oliveros, 2015) Genes were considered differentially exp ressed at p < 0.05 following fals e discovery rate correction.

PAGE 62

51 Figure 11 Expression of transcripts related to reproduction in the liver. There was a higher level of vitellogenin 1 transcript in fish exposed at the reference PCR H site (fc = 2 .4, p = 0. 0 496 N = 31 ) Overall these data show little evidence of estrogenic endocrine disruption at these sites. The middle bar is the median. Upper and lower boundaries of the box are the 3 rd and 1 st quartiles, respectively. Whiskers are the maximum and minimum values that are not outliers and c ircles represent outliers. Different letters denote significant differences at p < .05 follow ing Tukey's HSD correction PCFH NRWU NRWD SFSR 8.0 8.5 9.0 9.5 10.0 10.5 11.0 Estrogen receptor 1 p = 0.9834 PCFH NRWU NRWD SFSR 7.0 7.5 8.0 8.5 9.0 9.5 10.0 Estrogen receptor 2 p = 0.3806 PCFH NRWU NRWD SFSR 10.0 10.5 11.0 11.5 12.0 12.5 13.0 Androgen receptor p = 0.8374 PCFH NRWU NRWD SFSR 3 4 5 6 7 8 Vitellogenin 1 p = 0.0496 A AB AB B PCFH NRWU NRWD SFSR 2.5 3.0 3.5 4.0 4.5 Vitellogenin 3 p = 0.4226 Expression of reproduction related genes Signal intensity (Log2)

PAGE 63

52 Figure 12 Genes involved in the aryl hydrocarbon receptor pathway. Transcription of cytochrome p450 1a1 was significantly up reg ulated at NRWD (fc = 2.8, p < 0.001 N = 7 or 8/site). ) None of the other genes had altered transcription following exposure. The middle bar is the median. Upper and lower boundaries of the box are the 3 rd and 1 st quartiles, respectively. Whiskers are the maximum and minimum values that are not outliers and circles represent outliers. Different letters denote significant differences at p < .05 follow ing Tukey's HSD correction PCFH NRWU NRWD SFSR 2 3 4 5 6 7 8 Aryl hydrocarbon receptor 1b1 p = 0.431 PCFH NRWU NRWD SFSR 5 6 7 8 9 Aryl hydrocarbon receptor 2 p = 0.526 PCFH NRWU NRWD SFSR 4.0 4.5 5.0 5.5 6.0 6.5 AhR nuclear translocator p = 0.805 PCFH NRWU NRWD SFSR 7 8 9 10 11 AhR nuclear translocator 2 p = 0.342 PCFH NRWU NRWD SFSR 4 5 6 7 8 Cytochrome p450 1a1 p < 0.001 A AB B A PCFH NRWU NRWD SFSR 2.5 3.0 3.5 4.0 4.5 Cytochrome p450 1b1 p = 0.419 AhR pathway genes Signal intensity (Log2)

PAGE 64

53 Figure 13 Schematic of the mevalonate pathway (Pathway Studio 9.0, Ariadne Rockville, MD) This pathway is responsible for synthesizing mevalonate, a precursor of cholesterol. The mevalonate pathway was identified as down regulated at both NRWU and NRWD and up regulated at SFSR. Here, only the data from NRWU is represented. Gre en = decreased expression, red = increased expression.

PAGE 65

54 Figure 14 A schematic diagram of the IL6/STAT signal transduction pathway (Pathway Studio 9.0, Ariadne, Rockville, MD) Green indicates transcripts that were down regulated at NRWU and red indic ates genes that were up regulated.

PAGE 66

55 Figure 15 A schematic diagram of the classical complement pathway (Pathway Studio 9.0, Ariadne, Rockville, MD) Green indicates transcripts that were down regulated at NRWU and red indicat es genes that were up regula ted.

PAGE 67

56 Figure 16 Heatmap and hierarchical clustering of differentially expressed genes. Red indicates increased transcription and green indicates decreased transcription, relative to the mean expression for each gene. Agricultural = NRWU, WWTP = NRWD, Mixed Use = SFSR, and Reference = PCFH. Agricultural.3 WWTP.8 Agricultural.7 Agricultural.4 WWTP.4 WWTP.5 WWTP.1 WWTP.7 WWTP.2 WWTP.3 WWTP.6 Reference.2 Reference.1 Reference.5 Reference.3 Reference.6 Reference.4 Reference.8 Reference.7 Agricultural.1 Agricultural.2 Mixed.Use.2 Mixed.Use.5 Mixed.Use.4 Mixed.Use.3 Mixed.Use.6 Mixed.Use.1 Mixed.Use.8 Agricultural.5 Mixed.Use.7 Agricultural.6

PAGE 68

57 Table 1 Genes selected a priori for analysis based on known responses to environmental contaminants. Gene name Gene symbol Reproduction related Estrogen receptor 1 er1 Estrogen receptor 2 er2 Androgen re ceptor ar Vitellogenin 1 vtg1 Vitellogenin 3 vtg3 Xenobiotic responsive Aryl hydrocarbon receptor 1 ahr1 Aryl hydrocarbon receptor 2 ahr2 AhR nuclear translocator 1 arnt1 AhR nuclear translocator 2 arnt2 Cytochrome p450 1A1 cyp1a1 Cytochrom e p450 1B1 cyp1b1

PAGE 69

58 Table 2. Summary of chemicals detected in the Shenandoah Valley, August Sept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`-M99> 9.,HKL:-,KL@>MB=>8B=JHK8<,;@_H:J.;,=>.;M9-M`MK:-8H>,-8998:B>9.,HKL:-,KL@> %*V*>!MK8:BM;>#MK,L>aJM;8K@>&L:ALM-]>$,B<,L]>'S !MK8:BM;>",9,MLHQ>&L:ALM-]>T:J;=,L]>'S

PAGE 70

59 Table 2. Cont'd !"#$ !"#% &'() *(*" !"#"$%&'(")$*+,-./)01'(# +,-./01,1234/563 7 7 8,1219-./01,1234/563: 7 $-;,121./01,1234/563: 7 +,-;,121234/563 7 $-./01,1-191234/563: 7 $-./01,15.3416-4,-03: 7 8,121./01,15.3416-4,-03: 7 $-;,1215.3416-4,-03: 7 +,-./01,15.345093/<93 7 =>=>=?4,-./01,1?@?A,1A56163: 7 2&#("'"0&# B4,5C-63: 7 7 7 BC1DI?9-./01,1;36C363 7 =>J?9-./01,1;36C363 7 =>@?9-./01,1;36C363 7 @>K?9-?43,4?;F4<0?=>J?;36C1LF-6163 7 7 M?234/<0?=)?;36C14,-5C103 7 !>!?9-34/<0?2?410F52-93 7 7 $3E34/<054,5C-63 7 7 7 J?616<0A/3610 7 7 B4,5C-63 7 7 7 '5NN3-63 7 O505D10-93 7 +1650-93 7 $-A/36/<9,52-63 7 7 7 7 +,-.01E56 7 7 '5,;525C5A-63 7 7 '1A,1E45610 7 '/103E43,10 7 7 7 7 %*O*:'50-N1,6-5:#543,:*.-36.3:'3643,>:*5.,523641>:'BP:E10-9:A/5E3:3D4,5.4-16:Q-4/: R5E:./,12541R,5A/:*5.,523641>:'BP:E10-9:A/5E3:3D4,5.4-16:Q-4/: 0-LF-9:1,:R5E:./,12541R,5A/:81F093,>:'TP:.164-6F1FE:0-LF-9?0-LF-9:3D4,5.4-16: Q-4/:R5E:./,12541R,5A/
PAGE 71

60 Table 2 Cont'd !"#$ !"#% &'() *(*" !"#$%#&'()&#*+, +,-./012 3 3 3 +42,.50167821 3 9:7-67061 3 '.;;2012 3 3 3 3 '.-<.5./27012 3 3 3 '6,01012 3 3 3 3 *:=;.52,86>./6=2 3 ?80.<21@./6=2 3 3 3 A0@64.012 3 B27-6<.5.,2 3 3 &821C,601 3 ?25./27.5 3 ?-0.5,2-212 3 (=:461./6=2 3 3 +4C4=6D03 B2,;6-501 3 3 3 3 !046,012 3 3 3 3 $0./27.5 3 B2,864.-<.56= 3 3 +,216=6= 3 (2>6;21.@012 3 3 B2,8C=EF8E<21/6,-0./6=2 3 3 3 ?-.5.@6= 3 B2,67-6=6= 3 *0,.G=07,01 3 3 H21=.;.>012 3 I=C<:-0@2 3 -.()/)0()&+, +/0,8-65C401 3 '.-<.5./27012 3 3 *:=;.52,86>./6=2 3 3 10$%0.2+,3,!"4(02+($052.+ JK:6=L 3 3 3 JM,-612L 3 3 3 3 NE.1@-6M,212EOPFQE@0612 3 JK:=01 3 3 JM,-06= 3 3 '67-6M,.16=L 3 3 3 3 '6:52M,2-6= 3 3 3 3 '86=2M,2-6= 3 3 3 3 $08C@-6486=2M,2-6=L 3 3 3 3 $.0@/201 3 (6-5616121,01 3 3 3 3 JK:6= 3 3 3 90648.101L+L 3 %*I*LRI"LA.<6-.,6-CPLA.S-2142PLT*UL=0K:0@L48-65.,6G-.78CV,.1@25L5.MML %*I*L!.,061.=L#.,2-LW:.=0,CLA.<6-.,6-CPL$21D2-PL'RUL@0-24,E.K:26:ME01X24,061L =0K:0@L48-65.,6G-.78CV,.1@25L5.MMLM724,-652,-C %*I*LT.1M.ML#.,2-L*402142L'21,2-PLA.S-2142PLT*UL=0K:0@L48-65.,6G-.78CV,.1@25L 5.MMLM724,-652,-C %*I*L!.,061.=L"2M2.-48L&-6G-.5PL96:=@2-PL'RULM6=0@E78.M2L2>,-.4,061PL@2-0D.,0/.,061L .1@LG.ML48-65.,6G-.78CV,.1@25L5.MMLM724,-652,-C

PAGE 72

61 Table 3 Number of differentially expressed genes observed at each exposure site FDR corrected Site Abbreviation Site description Total DEGS Up-regulated Down-regulated NRWU North River upstream of WWTP 202 118 84 NRWD North River downstream of WWTP 297 232 65 SFSR South Fork Shenandoah River at Shenandoah River State Park 219 112 107

PAGE 73

62 Table 4 Highly differentially regulated pathways identified by gene set enrichment analysis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`)6)%"',R++).K6S FM *CDGQ *CDBC EDECE 5A"#S%"+&+ FC *CDGG *CDBQ EDEEN 1)<$6"'$%),;$%9a$S CG *BDBE *CDGC IEDEEC U&"+S'%9)+&+,"0,#9"6)+%)3"6 FC *BDHF *CDGG IEDEEC -=>X (6S#":)',.)%$K"6&+. BM CDMB CDFG IEDEEC -&#"%&'$%),$'4,'&#"%&'$.&4),.)%$K"6&+. BC BDEE CDFG IEDEEC 5;93&'U,*V,,WX-,+&:'$6&': CN CDHG CDFF EDECN T9)'S6$6$'&'),$'4,^S3"+&'),.)%$K"6&+. FN CDHB CDBN EDEEQ (62#"+),.)%$K"6&+. NN BDEG CDBF IEDEEC =)+;&3$%"3S,#9$&',$'4,"A&4$%&<),;9"+;9"3S6$%&"' GG CDQG CDCL EDEFB (62%$%9&"'),.)%$K"6&+. NQ CDNB CDCL EDECQ (6S"AS6$%),$'4,:6S#)3$%),.)%$K"6&+. BC CDHH CDCL EDECQ O";S,"0,b^!=,*V,RT*C,+&:'$6&':b*^U CN CDGN CDCB IEDEEC T3"%)&',-2#6)$3,J.;"3%,$'4,5A;"3% NQ *CDHN *CDBH EDEEN O"$:26$%&"',O$+#$4) BF *BDEG *CDBL IEDEEC !*+2:$3+,"A&4$%&"' N *CDNL *CDBM EDEBQ O6$++&#$6,O".;6).)'%,T$%9a$S BC *CDLN *CDQE IEDEEC !)#%&'*&'42#)4,O".;6).)'%,T$%9a$S CN *CDMM *CDQE IEDEEC R6%)3'$%&<),O".;6).)'%,T$%9a$S CQ *CDMN *CDQE EDEEN J!G=,*V,8^R^,+&:'$6&': L *CDNM *CDQC EDEFL J!G8^,*V,8^R^NU,+&:'$6&': H *CDGG *CDGH EDEEL U&"+S'%9)+&+,"0,#9"6)+%)3"6 FC *FDCH *CDMB IEDEEC 1)<$6"'$%),;$%9a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a$S CQ *CDQL *CDCG EDEFM J'%)3.)4&$%),P&6$.)'%,T"6S.)3&7$%&"' FF *CDQM *CDBE EDEFL %=-R,^3$'+#3&;%&"',$'4,T3"#)++&': QC *CDLB *CDBF EDEEQ ($6$#%"+),.)%$K"6&+. CQ *CDHG *CDFH EDEEQ

PAGE 74

63 Table 5. Highly differentially re gulated sub networks of genes identified b y sub network enrichment analysis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

PAGE 75

64 References Aaronson, D.S., Horvath, C.M., 2002. A Road map for those who don't know JAK STAT. Science. 296, 1653 1655. doi:10.1126/science.1071545 Akdis, M., Burgler, S., Crameri, R., Eiwegger, T., Fujita, H., Gomez, E., Klunker, S., Meyer, N., O'Mahony, L., Paloma res, O., Rhyner, C., Quaked, N., Schaffartzik, A., Van De Veen, W., Zeller, S., Zimmermann, M., Akdis, C.A., 2011. Interleukins, from 1 to 37, and interferon ": Receptors, functions, and roles in diseases. J. Allergy Clin. Immunol. 127, 701 721. doi:10.101 6/j.jaci.2010.11.050 Altenburger, R., Scholz, S., Schmitt Jansen, M., Busch, W., Escher, B.I., 2012. Mixture toxicity revisited from a toxicogenomic perspective. Environ. Sci. Technol. 46, 2508 2522. doi:10.1021/es2038036 Amato, J.R., Lukasewycz, M.T., Rob ert, E.D., Mount, D.I., Durhan, E.J., Ankley, G.T., 1992. An example of the identification of diazinon as a primary toxicant in an effluent. Environ. Toxicol. Chem. 11, 209 216. doi:10.1002/etc.5620110210 Ankley, G., Daston, G., 2006. Toxicogenomics in reg ulatory ecotoxicology. Environ. Sci. Technol. 40, 4055 4065. 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., Serrrano, J.A., Tietge, J.E., Villeneuve, D.L., 2 010. Adverse outcome pathways: A conceptual framework to support ecotoxicology research and risk assessment. Environ. Toxicol. Chem. 29, 730 741. doi:10.1002/etc.34 Bahamonde, P.A., McMaster, M.E., Servos, M.R., Martyniuk, C.J., Munkittrick, K.R., 2015. Mo lecular pathways associated with the intersex condition in rainbow darter ( Etheostoma caeruleum ) following exposures to municipal wastewater in the Grand River basin, ON, Canada. Part B. Aquat. Toxicol. 159, 302 316. doi:10.1016/j.aquatox.2014.11.022 Barbe r, L.B., Brown, G.K., Nettesheim, T.G., Murphy, E.W., Bartell, S.E., Schoenfuss, H.L., 2011. Effects of biologically active chemical mixtures on fish in a wastewater impacted urban stream. Sci. Total Environ. 409, 4720 8. doi:10.1016/j.scitotenv.2011.06.03 9 Barber, L.B., Vajda, A.M., Douville, C., Norris, D.O., Writer, J.H., 2012. Fish endocrine disruption responses to a major wastewater treatment facility upgrade. Environ. Sci. Technol. 46, 2121 31. doi:10.1021/es202880e Berenbaum, M.C., 1989. What is syne rgy? Pharmacol. Rev. 41, 93 141. doi:10.1007/s13398 014 0173 7.2 Berninger, J.P., Martinovi! Weigelt, D., Garcia Reyero, N., Escalon, L., Perkins, E.J., Ankley, G.T., Villeneuve, D.L., 2014. Using transcriptomic tools to evaluate biological effects across effluent gradients at a diverse set of study sites in Minnesota, USA. Environ. Sci. Technol. 48, 2404 2412. doi:10.1021/es4040254

PAGE 76

65 Bitman, J., Cecil, H.C., Harris, S.J., Fries, G.F., 1968. Estrogenic activity of o, p' DDT in the mammalian uterus and avian o viduct. Science. 162, 371 372. doi:10.1126/science.162.3851.371 Blackburn, M.A., Waldock, M.J., 1995. Concentrations of alkylphenols in rivers and estuaries in England and Wales. Water Res. 29, 1623 1629. doi:10.1016/0043 1354(94)00340 D Blazer, V.S., Iwan owicz, L.R., Iwanowicz, D.D., Smith, D.R., Young, J.A., Hedrick, J.D., Foster, S.W., Reeser, S.J., 2007. Intersex (testicular oocytes) in smallmouth bass from the Potomac River and selected nearby drainages. J. Aquat. Anim. Health 19, 242 53. doi:10.1577/H 07 031.1 Blazer, V.S., Iwanowicz, L.R., Starliper, C.E., Iwanowicz, D.D., Barbash, P., Hedrick, J.D., Reeser, S.J., Mullican, J.E., Zaugg, S.D., Burkhardt, M.R., Kelble, J., 2010. Mortality of centrarchid fishes in the potomac drainage: Survey results and overview of potential contributing factors. J. Aquat. Anim. Health 22, 190 218. doi:10.1577/H10 002.1 Brack, W., 2003. Effect directed analysis: A promising tool for the identification of organic toxicants in complex mixtures? Anal. Bioanal. Chem. 377, 397 407. doi:10.1007/s00216 003 2139 z Brack, W., Altenburger, R., Ensenbach, U., Moder, M., Segner, H., Schuurmann, G., 1999. Bioassay directed identification of organic toxicants in river sediment in the industrial region of bitterfeld (Germany) A contribut ion to hazard assessment. Arch. Environ. Contam. Toxicol. 37, 164 74. Burgess, R.M., Ho, K.T., Brack, W., Lamoree, M., 2013. Effects directed analysis (EDA) and toxicity identification evaluation (TIE): Complementary but different approaches for diagnosing causes of environmental toxicity. Environ. Toxicol. Chem. 32, 1935 1945. doi:10.1002/etc.2299 Chen, H. S., Jia, J., Su, H. F., Lin, H. D., Chen, J. W., Lin, S. J., Yang, J. Y., Lai, H. C., Mestril, R., Wang, P.H., 2006. Downregulation of the constitutivel y expressed Hsc70 in diabetic myocardium is mediated by insulin deficiency. J. Endocrinol. 190, 433 40. doi:10.1677/joe.1.06692 Cheshenko, K., Pakdel, F., Segner, H., Kah, O., Eggen, R.I.L., 2008. Interference of endocrine disrupting chemicals with aromata se CYP19 expression or activity, and consequences for reproduction of teleost fish. Gen. Comp. Endocrinol. 155, 31 62. doi:10.1016/j.ygcen.2007.03.005 Choi, E. Y., Jin, J. Y., Lee, J. Y., Choi, J. I., Choi, I.S., Kim, S. J., 2012. Anti inflammatory effects and the underlying mechanisms of action of daidzein in murine macrophages stimulated with Prevotella intermedia lipopolysaccharide. J. Periodontal Res. 47, 204 11. doi:10.1111/j.1600 0765.2011.01422.x Clarke, D.B., Bailey, V., Lloyd, A.S., 2008. Determina tion of phytoestrogens in dietary supplements by LC MS/MS. Food Addit. Contam. Part A. Chem. Anal. Control. Expo. Risk Assess. 25, 534 47. doi:10.1080/02652030701658340

PAGE 77

66 Colborn, T., Clement, C. (Eds.), 1992. Chemically induced alterations in sexual and fun ctional development: The wildlife/human connection. Princeton Scientific Publishing. Crain, A.C., Guillette, L. (Eds.), 2000. Environmental endocrine disrupters: An evolutionary perspective. CRC Press. Daughton, C.G., 2001. Pharmaceuticals and personal car e products in the environment: overarching issues and overview., in: Daughton, C.G., Jones Lepp, T. (Eds.), Pharmaceuticals and Personal Care Products in the Environment: Scientific and Regulatory Issues. American Chemical Society Symposium Series 791, pp. 2 38. Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and endogenous chemicals. Annu Rev Pharmacol Toxicol 43, 309 334. doi:10.1146/annurev.pharmtox.43.100901.135828 Elferink, M.G.L., Olinga, P., van Leeuwen, E.M., Bauerschmidt, S., Polman, J., Schoonen, W.G., Heisterkamp, S.H., Groothuis, G.M.M., 2011. Gene expression analysis of precision cut human liver slices indicates stable expression of ADME Tox related genes. Toxicol. Appl. Pharmacol. 2 53, 57 69. doi:10.1016/j.taap.2011.03.010 Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States -II) untreated drinking water sources. Sci. Total Environ. 402, 201 16. doi:10.1016/j.scitotenv.2008.02.021 Fromme, H., KŸchler, T., Otto, T., Pilz, K., MŸller, J., Wenzel, A., 2002. Occurrence of phthalates and bisphenol A and F in the envir onment. Water Res. 36, 1429 1438. doi:10.1016/S0043 1354(01)00367 0 Gabay, C., Kushner, I., 1999. Acute phase proteins and other systemic responses to inflammation. N. Engl. J. Med. 340, 448 454. doi:10.1056/NEJM199902113400607 Gaido, K.W., Leonard, L.S., L ove l l S ., G oul d, J .C., Ba ba # D ., P ort i e r, C.J ., M c D onne l l D.P., 1997. Evaluation of chemicals with endocrine modulating activity in a yeast based steroid hormone receptor gene transcription assay. Toxicol. Appl. Pharmacol. 143, 205 212. doi:10.1006/ta ap.1996.8069 Garcia Reyero, N., Adelman, I., Liu, L., Denslow, N., 2008. Gene expression profiles of fathead minnows exposed to surface waters above and below a sewage treatment plant in Minnesota. Mar. Environ. Res. 66, 134 136. doi:10.1016/j.marenvres.20 08.02.073 Garcia Reyero, N., Lavelle, C.M., Escalon, B.L., Martinovi!, D., Kroll, K.J., Sorensen, P.W., Denslow, N.D., 2010. Behavioral and genomic impacts of a wastewater effluent on the fathead minnow. Aquat. Toxicol. 101, 38 48. doi:10.1016/j.aquatox.2010.08.014 Garcia Reyero, N., Perkins, E.J., 2011. Systems biology: Leading the revolution in ecotoxicology. Environ. Toxicol. Chem. 30, 265 273. doi:10.1002/etc.401

PAGE 78

67 Gilbert, S.F., Epel, D., 2009. Endocrine Disruptors, in: Ecological Developmental Biology: Integratin g Epigenetics, Medicine, and Evolution. pp. 197 244. Gilbert, S.F., Epel, D., 2009. Ecological Developmental Biology: Integrating Epigenetics, Medicine, and Evolution 82, 231 232. doi:10.1126/science.1175598 Hahn, M.E., 2002. Aryl hydrocarbon receptors: di versity and evolution. Chem. Biol. Interact. 141, 131 160. doi:10.1016/S0009 2797(02)00070 4 Hai, T., Wolfgang, C.D., Marsee, D.K., Allen, A.E., Sivaprasad, U., 1999. ATF3 and stress responses. Gene Expr. 7, 321 35. Hayes, T.B., Khoury, V., Narayan, A., Na zir, M., Park, A., Brown, T., Adame, L., Chan, E., Buchholz, D., Stueve, T., Gallipeau, S., 2010. Atrazine induces complete feminization and chemical castration in male African clawed frogs ( Xenopus laevis ). Proc. Natl. Acad. Sci. U. S. A. 107, 4612 4617. doi:10.1073/pnas.0909519107 Heinrich, P.C., Behrmann, I., MŸller Newen, G., Schaper, F., Graeve, L., 1998. Interleukin 6 type cytokine signalling through the gp130/Jak/STAT pathway. Biochem. J. 334, 297 314. doi:10.1042/bj3340297 Hill, R.H., Head, S.L., Ba ker, S., Gregg, M., Shealy, D.B., Bailey, S.L., Williams, C.C., Sampson, E.J., Needham, L.L., 1995. Pesticide residues in urine of adults living in the United States: reference range concentrations. Environ. Res. doi:10.1006/enrs.1995.1071 Ihlaseh, S.M., B ailey, K.A., Hester, S.D., Jones, C., Ren, H., Cardoso, A.P.F., Oliveira, M.L.C.S., Wolf, D.C., de Camargo, J.L. V, 2011. Transcriptional profile of diuron induced toxicity on the urinary bladder of male wistar rats to inform mode of action. Toxicol. Sci. 122, 330 338. doi:10.1093/toxsci/kfr108 IOM, (Institue of Medicine), 2014. Institue of Medicine, in: Identifying and Reducing Environmental Health Risks of Chemicals in Our Society: Workshop Summary. The National Academies Press, Washington, DC. Iwanowicz, L.R., Blazer, V.S., Guy, C.P., Pinkney, A.E., Mullican, J.E., Alvarez, D.A., 2009. Reproductive health of bass in the Potomac, U.S.A., drainage: part 1. Exploring the effects of proximity to wastewater treatment plant discharge. Environ. Toxicol. Chem. 28 1072 1083. doi:10.1897/08 433.1 Jensen, K.M., Korte, J.J., Kahl, M.D., Pasha, M.S., Ankley, G.T., 2001. Aspects of basic reproductive biology and endocrinology in the fathead minnow (Pimephales promelas). Comp. Biochem. Physiol. Part C Toxicol. Pharmacol 128, 127 141. doi:10.1016/S1532 0456(00)00185 X Jobling, S., Nolan, M., Tyler, C.R., Brighty, G., Sumpter, J.P., 1998. Widespread sexual disruption in wild fish. Environ. Sci. Technol. 32, 2498 2506. doi:10.1021/es9710870 Kidd, K.A., Blanchfield, P.J., M ills, K.H., Palace, V.P., Evans, R.E., Lazorchak, J.M., Flick, R.W., 2007. Collapse of a fish population after exposure to a synthetic estrogen. Proc. Natl. Acad. Sci. U. S. A. 104, 8897 8901.

PAGE 79

68 doi:10.1073/pnas.0609568104 Kšck Schulmeyer, M., Villagrasa, M. L—pez de Alda, M., CŽspedes S‡nchez, R., Ventura, F., Barcel—, D., 2013. Occurrence and behavior of pesticides in wastewater treatment plants and their environmental impact. Sci. Total Environ. 458 460, 466 76. doi:10.1016/j.scitotenv.2013.04.010 Kolpin, D.W., Blazer, V.S., Gray, J.L., Focazio, M.J., Young, J.A., Alvarez, D.A., Iwanowicz, L.R., Foreman, W.T., Furlong, E.T., Speiran, G.K., Zaugg, S.D., Hubbard, L.E., Meyer, M.T., Sandstrom, M.W., Barber, L.B., 2013. Chemical contaminants in water and sedim ent near fish nesting sites in the Potomac River basin: Determining potential exposures to smallmouth bass ( Micropterus dolomieu ). Sci. Total Environ. 443, 700 716. doi:10.1016/j.scitotenv.2012.09.063 Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M. Zaugg, S.D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U .S. streams, 1999 2000: A national reconnaissance. 36, 1202 1211. doi:10.1021/es011055j Kortenkamp, A., 2007. Ten years of mixing c ocktails: A review of combination effects of endocrine disrupting chemicals. Environ. Health Perspect. 115, 98 105. doi:10.1289/ehp.9357 Kortenkamp, A., Altenburger, R., 1999. Approaches to assessing combination effects of oestrogenic environmental polluta nts. Sci. Total Environ. 233, 131 140. doi:10.1016/S0048 9697(99)00228 4 Kortenkamp, A., Altenburger, R., 1998. Synergisms with mixtures of xenoestrogens: A reevaluation using the method of isoboles. Sci. Total Environ. 221, 59 73. doi:10.1016/S0048 9697(9 8)00261 7 Kuiper, G.G., Carlsson, B., Grandien, K., Enmark, E., HŠggblad, J., Nilsson, S., Gustafsson, J.A., 1997. Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors alpha and beta. Endocrinology 138, 863 70. doi:10.1210/endo.138.3.4979 Kuiper, G.G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998. Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta. Endocrinol ogy 139, 4252 63. doi:10.1210/endo.139.10.6216 Kumar, V., Johnson, A.C., Trubiroha, A., Tumov‡, J., Ihara, M., Grabic, R., Kloas, W., Tanaka, H., Kroupov‡, H.K., 2015. The challenge presented by progestins in ecotoxicological research: a critical review. E nviron. Sci. Technol. 49, 2625 38. doi:10.1021/es5051343 Kuster, M., L—pez de Alda, M.J., Hernando, M.D., Petrovic, M., Mart’n Alonso, J., Barcel—, D., 2008. Analysis and occurrence of pharmaceuticals, estrogens, progestogens and polar pesticides in sewage treatment plant effluents, river water and drinking water in the Llobregat river basin (Barcelona, Spain). J. Hydrol. 358, 112 123. doi:10.1016/j.jhydrol.2008.05.030

PAGE 80

69 Lange, I.G., Daxenberger, A., Schiffer, B., Witters, H., Ibarreta, D., Meyer, H.H.., 2002 Sex hormones originating from different livestock production systems: fate and potential disrupting activity in the environment. Anal. Chim. Acta 473, 27 37. doi:10.1016/S0003 2670(02)00748 1 Lee, K. H., Choi, E. M., 2005. Biochanin A stimulates osteobla stic differentiation and inhibits hydrogen peroxide induced production of inflammatory mediators in MC3T3 E1 cells. Biol. Pharm. Bull. 28, 1948 1953. doi:10.1248/bpb.28.1948 Lee, Y. M., Williams, T.D., Jung, S. O., Lee, J. S., 2005. cDNA cloning and expres sion of a cytochrome P450 1A (CYP1A) gene from the hermaphroditic fish Rivulus marmoratus. Mar. Pollut. Bull. 51, 769 75. doi:10.1016/j.marpolbul.2005.06.015 Lemaire, G., Terouanne, B., Mauvais, P., Michel, S., Rahmani, R., 2004. Effect of organochlorine p esticides on human androgen receptor activation in vitro. Toxicol. Appl. Pharmacol. 196, 235 46. doi:10.1016/j.taap.2003.12.011 Limonciel, A., Moenks, K., Stanzel, S., Truisi, G.L., Parmentier, C., Aschauer, L., Wilmes, A., Richert, L., Hewitt, P., Mueller S.O., Lukas, A., Kopp Schneider, A., Leonard, M.O., Jennings, P., 2014. Transcriptomics hit the target: Monitoring of ligand activated and stress response pathways for chemical testing. Toxicol. Vitr. 30, 7 18. doi:10.1016/j.tiv.2014.12.011 Liu, T., Dani els, C.K., Cao, S., 2012. Comprehensive review on the HSC70 functions, interactions with related molecules and involvement in clinical diseases and therapeutic potential. Pharmacol. Ther. 136, 354 74. doi:10.1016/j.pharmthera.2012.08.014 Malpass, G.E., Ari milli, S., Prasad, G.L., Howlett, A.C., 2014. Regulation of gene expression by tobacco product preparations in cultured human dermal fibroblasts. Toxicol. Appl. Pharmacol. 279, 211 219. doi:10.1016/j.taap.2014.06.001 Martin, H.L., Svendsen, C., Lister, L.J ., Gomez Eyles, J.L., Spurgeon, D.J., 2009. Measurement and modeling of the toxicity of binary mixtures in the nematode Caenorhabditis elegans -A test of independent action. Environ. Toxicol. Chem. 28, 97 104. doi:10.1897/07 215.1 Martinovi! 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., 2014. Transcriptomic effects based monitoring for endocrine active chemicals: Assessing relative contribution of treated wastewater to downstream pollution. Environ. Sci. Technol. 48, 2385 2394. doi:10.1021/es404027n Medjakovic, S., Mueller, M., Jungbauer, A., 2010. Potential health modulating effects of isoflavones and metabolites via activation of PPAR and AhR. Nutrients 2 241 279. doi:10.3390/nu2030241 Mehinto, A.C., Martyniuk, C.J., Spade, D.J., Denslow, N.D., 2012. Applications of next generation sequencing in fish ecotoxicogenomics. Front. Genet. 3, 1 10. doi:10.3389/fgene.2012.00062

PAGE 81

70 Mueller, M., Jungbauer, A., 2008. Red clover extract: a putative source for simultaneous treatment of menopausal disorders and the metabolic syndrome. Menopause 15, 1120 1131. doi:10.1097/gme.0b013e31817062ce Nakai, M., Tabira, Y., Asai, D., Yakabe, Y., Shimyozu, T., Noguchi, M., Takatsuki M., Shimohigashi, Y., 1999. Binding characteristics of dialkyl phthalates for the estrogen receptor. Biochem. Biophys. Res. Commun. 254, 311 4. doi:10.1006/bbrc.1998.9928 Neumann, M., Schulz, R., SchŠfer, K., MŸller, W., Mannheller, W., Liess, M., 2002. The significance of entry routes as point and non point sources of pesticides in small streams. Water Res. 36, 835 842. doi:10.1016/S0043 1354(01)00310 4 Nikitin, A., Egorov, S., Daraselia, N., Mazo, I., 2003. Pathway studio -the analysis and navigation of molecular networks. Bioinformatics 19, 2155 2157. doi:10.1093/bioinformatics/btg290 Norris, D.O., Carr, J.A., 2006. Introduction to Endocrinology, in: Norris, D.O., Carr, J.A. (Eds.), Endocrine Disruption: Biological Bases for Health Effects in Wildlife a nd Humans. Oxford University Press. Oliveros, J.C., 2015. Venny. An interactive tool for comparing lists with Venn's diagrams. http://bioinfogp.cnb.csic.es/tools/venny/index.html. Oscarson, M., Zanger, U.M., Rifki, O.F., Klein, K., Eichelbaum, M., Meyer, U .A., 2006. Transcriptional profiling of genes induced in the livers of patients treated with carbamazepine. Clin. Pharmacol. Ther. 80, 440 456. doi:10.1016/j.clpt.2006.08.013 Panter, G.H., Hutchinson, T.H., Hurd, K.S., Sherren, A., Stanley, R.D., Tyler, C. R., 2004. Successful detection of (anti )androgenic and aromatase inhibitors in pre spawning adult fathead minnows ( Pimephales promelas ) using easily measured endpoints of sexual development. Aquat. Toxicol. 70, 11 21. doi:10.1016/j.aquatox.2004.06.007 Pa $ kov‡, V., Hilscherov‡, K., Bl‡ha, L., 2011. Teratogenicity and embryotoxicity in aquatic organisms after pesticide exposure and the role of oxidative stress. Rev. Environ. Contam. Toxicol. 211, 25 61. doi:10.1007/978 1 4419 8011 3_2 Perkins, E.J., Antczak, P., Burgoon, L., Falciani, F., Garcia Reyero, N., Gutsell, S., Hodges, G., Kienzler, A., Knapen, D., McBride, M., Willett, C., 2015. Adverse outcome pathways for regulatory applications: Examination of four case studies with different degrees of completen ess and scientific confidence. Toxicol. Sci. 148, 14 25. doi:10.1093/toxsci/kfv181 R Core Team, 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. https://www.R project.org. Reineke, N., Bester, K., HŸhnerfuss, H., Jastorff, B., Weigel, S., 2002. Bioassay directed chemical analysis of River Elbe surface water including large volume extractions and high performance fractionation. Chemosphere 47, 717 723. doi:10.1016/S0045 6535(01)00340 X

PAGE 82

71 S antos, L.H.M.L.M., Araœjo, A., Fachini, A., Pena, A., Delerue Matos, C., Montenegro, M.C.B.S.M., 2010. Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. J. Hazard. Mater. 175, 45 95. doi:10.1016/j.jhazmat.2009. 10.100 Schulpen, S.H.W., Pennings, J.L.A., Piersma, A.H., 2015. Gene expression regulation and pathway analysis after valproic acid and carbamazepine exposure in a human embryonic stem cell based neurodevelopmental toxicity assay. Toxicol. Sci. 146, 311 32 0. doi:10.1093/toxsci/kfv094 Schwindt, A.R., Winkelman, D.L., Keteles, K., Murphy, M., Vajda, A.M., 2014. An environmental oestrogen disrupts fish population dynamics through direct and transgenerational effects on survival and fecundity. J. Appl. Ecol. 51 582 591. doi:10.1111/1365 2664.12237 Sellin Jeffries, M.K., Mehinto, A.C., Carter, B.J., Denslow, N.D., Kolok, A.S., 2012. Taking microarrays to the field: Differential hepatic gene expression of caged fathead minnows from Nebraska watersheds. Environ. S ci. Technol. 46, 1877 1885. doi:10.1021/es2039097 Simmons, S.O., Fan, C. Y., Ramabhadran, R., 2009. Cellular stress response pathway system as a sentinel ensemble in toxicological screening. Toxicol. Sci. 111, 202 225. doi:10.1093/toxsci/kfp140 Spurgeon, D .J., Jones, O. a. H., Dorne, J. L.C.M., Svendsen, C., Swain, S., StŸrzenbaum, S.R., 2010. Systems toxicology approaches for understanding the joint effects of environmental chemical mixtures. Sci. Total Environ. 408, 3725 3734. doi:10.1016/j.scitotenv.2010 .02.038 Stamper, B.D., Garcia, M.L., Nguyen, D.Q., Beyer, R.P., Bammler, T.K. Farin, F.M., Kavanagh, T.J., Nelson, S.D., 2015. p53 contributes to differentiating gene expression following exposure to acetaminophen and its less hepatotoxic regioisomer both In vitro and in vivo. Gene Regul. Syst. Bio. 9, 1 14. doi:10.4137/GRSB.S2538 8 Steel, D.M., Whitehead, A.S., 1994. The major acute phase reactants: C reactive protein, serum amyloid P component and serum amyloid A protein. Immunol. Today 15, 81 88. doi:10.1016/0167 5699(94)90138 4 Subramanian, A., Tamayo, P., Mootha, V.K., Mukherje e, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., Mesirov, J.P., 2005. Gene set enrichment analysis: a knowledge based approach for interpreting genome wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102 15545 50. doi:10.1073/pnas.0506580102 Suja, F., Pramanik, B.K., Zain, S.M., 2009. Contamination, bioaccumulation and toxic effects of perfluorinated chemicals (PFCs) in the water environment: a review paper. Water Sci. Technol. 60, 1533. doi:10.2166/wst. 2009.504

PAGE 83

72 Suzuki, G., Sato, K., Isobe, T., Takigami, H., Brouwer, A., Nakayama, K., 2015. Detection of glucocorticoid receptor agonists in effluents from sewage treatment plants in Japan. Sci. Total Environ. 527 528, 328 34. doi:10.1016/j.scitotenv.2015.0 5.008 Tabb, M.M., Blumberg, B., 2006. New modes of action for endocrine disrupting chemicals. Mol. Endocrinol. 20, 475 82. doi:10.1210/me.2004 0513 Ternes, T.., Stumpf, M., Mueller, J., Haberer, K., Wilken, R. D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants I. Investigations in Germany, Canada and Brazil. Sci. Total Environ. 225, 81 90. doi:10.1016/S0048 9697(98)00334 9 Thomas, K. V., Hurst, M.R., Matthiessen, P., Waldock, M.J., 2001. Characterization of est rogenic compounds in water samples collected from United Kingdom estuaries. Environ. Toxicol. Chem. 20, 2165 2170. doi:10.1002/etc.5620201005 U.S. EPA, 2009. Inventory Update Reporting: Non confidential 2006 IUR records by chemical, including manufacturing processing and use information. Washington, DC. U.S. EPA, 1991. Methods for aquatic toxicity identification evaluations: Phase I toxicity characterization proceadures. Duluth, MN. Uribe, C., Folch, H., Enriquez, R., Moran, G., 2011. Innate and adaptive i mmunity in teleost fish: a review. Vet. Med. (Praha). 56, 486 503. USDA NASS, 2012. Quick stats [WWW Document]. United States Dep. Agric. Natl. Agric. Stat. Serv. URL https://www.nass.usda.gov/Quick_Stats/index.php (accessed 4.24.16). Vajda, A.M., Barb er, L.B., Gray, J.L., Lopez, E.M., Bolden, A.M., Schoenfuss, H.L., Norris, D.O., 2011. Demasculinization of male fish by wastewater treatment plant effluent. Aquat. Toxicol. 103, 213 21. doi:10.1016/j.aquatox.2011.02.007 Vajda, A.M., Norris, D.O., 2006. En docrine active phytochemicals: Environmental signaling context and mechanisms, in: Norris, D.O., Carr, J.A. (Eds.), Endocrine Disruption: The Biological Basis for Health Effects in Humans and Wildlife. Oxford University Press, pp. 390 423. 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., 2012. Ecotoxicogenomics to support ecological risk assessment: A case study with bisphenol A in fish. Envi ron. Sci. Technol. 46, 51 59. doi:10.1021/es201150a vom Saal, F.S., Hughes, C., 2005. An extensive new literature concerning low dose effects of bisphenol A shows the need for a new risk assessment. Environ. Health Perspect. 113, 926 933. doi:10.1289/ehp.7 713 Wilson, M.P., Schwarzman, M.R., 2009. Toward a new U.S. chemicals policy: Rebuilding the foundation to advance new science, green chemistry, and environmental health. Environ. Health Perspect. doi:10.1289/ehp.0800404

PAGE 84

73 Young, J., Iwanowicz, L., Sperry, A ., Blazer, V., 2014. A landscape based reconnaissance survey of estrogenic activity in streams of the upper Potomac, upper James, and Shenandoah Rivers, USA. Environ. Monit. Assess. 186, 5531 5545. doi:10.1007/s10661 014 3801 y Zipfel, P.F., Skerka, C., 20 09. Complement regulators and inhibitory proteins. Nat. Rev. Immunol. 9, 729 740. doi:10.1038/nri2620