Mapping Multiple Endocrine Disrupting Activities in Virginia Rivers using Effect-based Assays

Abstract

Water sources are frequently contaminated with natural and anthropogenic substances having known or suspected endocrine disrupting activities; however, these activities are not routinely measured and monitored. Phenotypic bioassays are a promising new approach for detection and quantitation of endocrine disrupting chemicals (EDCs). We developed cell lines expressing fluorescent chimeric constructs capable of detecting environmental contaminants which interact with multiple nuclear receptors. Using these assays, we tested water samples collected in the summers of 2016, 2017 and 2018 from two major Virginia rivers. Samples were concentrated 200X and screened for contaminants interacting with the androgen (AR), glucocorticoid (GR), aryl hydrocarbon (AhR) and thyroid receptors. Among 45 tested sites, over 70% had AR activity and 60% had AhR activity. Many sites were also positive for GR and TRβ activation (22% and 42%, respectively). Multiple sites were positive for more than one type of contaminants, indicating presence of complex mixtures. These activities may negatively impact river ecosystems and consequently human health.

Graphical Abstract

1.1. Introduction

Endocrine disruption by natural and synthetic compounds is a major concern for the health of aquatic ecosystems (Kolpin et al., 2002). Moreover, the presence of these chemicals in the environment and water sources could also affect human health (Kabir, Rahman, & Rahman, 2015; Schug et al., 2016). Most EDCs are manmade compounds and include pharmaceuticals, pesticides, industrial chemicals and personal care products (Frye et al., 2012). Synthetic hormones used as contraceptives and treatment of hormone-responsive diseases are also potential EDCs, which can easily reach water sources by natural excretion (Fent, Weston, & Caminada, 2006). In addition, synthetic androgens such as trenbolone, used to accelerate livestock growth, can contaminate water sources and affected sex determination and fecundity in fish (Ankley et al., 2003; Morthorst, Holbech, & Bjerregaard, 2010; Stephany, 2010). Steroid hormones are among the most potent EDCs and can substantially affect aquatic life (Christen, Hickmann, Rechenberg, & Fent, 2010; Jobling et al., 2006; Lange et al., 2001; Pawlowski, van Aerle, Tyler, & Braunbeck, 2004; Sumpter & Johnson, 2005). They bind to evolutionary conserved receptors at very low levels and exposure to EDCs can impact the physiology of most vertebrates, including humans (Christen et al., 2010). Epidemiological studies demonstrated that EDCs can affect prenatal growth and development, thyroid function, puberty, fertility, glucose metabolism, obesity, and are linked to the development of malignancies in humans (Ho et al., 2017; Street et al., 2018).

Standard single-analyte chemical approaches such as High-Pressure Liquid Chromatography combined with Gas Chromatography and Tandem Mass Spectrometry have been used to identify EDCs in the environment, specifically in the water (Bussy, Li, & Li, 2016; Petrovic, Eljarrat, Lopez de Alda, & Barcelo, 2002). However, structural changes induced by environmental biological and chemical processes, which transform the initial contaminants and their chemical structures, can complicate structural analysis. For example, in a recent study that screened stream water at 35 sites from 24 states in the US and Puerto Rico by both, the in vitro assays and chemical analysis concluded that estrogen receptor activity correlated with the concentrations of steroidal estrogens detected by chemical approaches; however, AR and GR activities did not correlate with chemical analyses. Furthermore, no known GR-active compounds were identified from the target-chemical analyte list (Conley et al., 2017). Similarly, a study from our laboratory failed to identify any known glucocorticoid compounds in a sample with high activity measured by GR translocation and gene expression assays (Stavreva et al., 2012). These results argue for the wider use of effect-based assays to detect endocrine disrupting potential of environmental contaminants. Moreover, water sources may contain a complex mixture of contaminants having endocrine disrupting potential which might be difficult to anticipate using single chemical identification methods (Kolpin et al., 2002).

Earlier studies successfully utilized in vitro assays to screen for relevant biological activities. Those include screening for contaminants interacting with multiple nuclear receptors (NR) including AR (Brand et al., 2013; Cavallin et al., 2014; Mehinto et al., 2015; Roberts et al., 2015), AhR (Eichbaum et al., 2014; Escher et al., 2014) and GR (Brand et al., 2013; Cavallin et al., 2014; Mehinto et al., 2015; Roberts et al., 2015) and TR (Escher et al., 2014; Jia et al., 2015).

Most of these studies were performed by transcription-based assays which are very sensitive, specific, and effect-based. However, they are time-consuming and require a 2-step validation to determine the full spectrum of endocrine disrupting activities in each sample: a first screen is performed in an agonist mode and an additional screen needs to be performed in the antagonist mode.

To account for the entire spectrum of receptor-interacting contaminants in a single screen, we developed and validated a low-cost, high-throughput imaging-based assay for quantitative imaging of translocation from the cytoplasm to the nucleus of fluorescently labeled NR chimeric constructs in mammalian cells. This translocation occurs in response to agonists or antagonists, thus reporting an unbiased endocrine disruptive activity of the tested sample in a single step (Jones et al., 2020; Stavreva et al., 2012; Stavreva et al., 2016).

Wide screening for endocrine disrupting activities in the environment is important, it does not require identification of known hormones or specific chemical structures. Thus, we employed our mammalian-cell based bioassays to screen samples from several major rivers in Virginia, some of which also supply drinking water. We assessed bioactivity of these samples across four major classes of EDCs (androgenic, glucocorticoid, aryl hydrocarbon and thyroid) and demonstrate that these rivers are frequently positive for multiple activities. The persistent presence of low-level hormonal activities may negatively affect fish and other aquatic organisms and may also indirectly affect human health.

1.2. Materials and Methods

1.2.1. Chemicals

Testosterone, Dexamethasone, 3,3′,5-Triiodo-L-thyronine sodium salt powder (T3) and the organic solvent DMSO were purchased from Sigma (catalog Nu: T1500, D1756, T6397, and D2650, respectively). CAY10465 (catalog Nu: 10006546) was purchased from Cayman Chemical.

1.2.2. River water samples collection and processing

Water samples were collected as part of the ongoing StreamSweepers Program at the Center for Natural Capital (CNC), Orange VA. They were collected between 2016 and 2018 from different geographic locations (Figure 1 and Supplemental Table 1) and included two replicas of grab water samples which were processed and concentrated according to a previously published protocol (Ciparis, Iwanowicz, & Voshell, 2012). Briefly, water samples were collected in glass bottles and stored at 4°C until processing. The water samples were filtered through a GF/F filter (0.7 μm) using a solvent rinsed all-glass apparatus. Filters were rinsed with 1 ml of methanol to liberate soluble compounds from the retained suspended solids. Filtered samples and blanks were subjected to solid phase extraction (SPE) using OASIS® HLB (200 mg) glass cartridges (Waters Corporation, Milford, MA). Cartridges were sequentially pre-conditioned, and the filtered samples were loaded onto the cartridge at a flow rate of 5–6 ml/minute (continuous vacuum). Analytes were eluted from the cartridge with 100% methanol and concentrated by evaporation. For biological testing, samples were reconstituted in DMSO and diluted in growth media to a final 1,000X concentration from the original water volume and added to cells at 200X concentration. Multiple samples were collected alongside the same river basin on the same day to examine the effect of the site location on the presence of contamination.

Figure 1.

Map showing overall and focused locations of the collected samples.

1.2.3. Cell treatment and imaging

The method for quantitative imaging of AR, GR, TR and AhR have been described previously (Gonsioroski, Mourikes, & Flaws, 2020; Jones et al., 2020; Stavreva et al., 2012; Stavreva et al., 2016). Briefly, all cell lines were maintained at 37°C in tetracycline -containing media (DMEM medium, supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO), Pen/strep, and glutamine (both from Invitrogen, Carlsbad, CA) supplemented with 5 μg/ml of Tetracycline (Sigma, St. Louis, MO) dissolved at 1,000X in EtOH) to suppress expression of the constructs prior to testing, and were not cultured for more than 3–4 weeks. Cells expressing the green fluorescent protein (GFP)-tagged AR, GR, and AhR constructs were plated in 384 well plates (Matrical, Catalog Number MGB101–1–2-LG-L) at 5,000 cells per well without tetracycline in DMEM containing 10% charcoal stripped serum (Hyclone, Logan, UT) and grown at 37°C overnight to allow expression of the GFP-tagged constructs. As GFP-GR-TRβ expressing MCF cells require longer time to express the construct, these cells were first cultured for 24 hours without tetracycline in DMEM medium containing 10% charcoal stripped serum and then plated in 384 well plates at 7,000 cells per well in the same media for additional 24 hours. Concentrated water samples (at 200X concentration), vehicle controls (1% DMSO), as well as the positive controls Dex or testosterone, respectively, were added to cells at various concentrations (up to 100 nM) for 30 min when screening for GR and AR. When screening for AhR and TR samples and vehicle controls were added for 3 hours using CAY 10465 (up to 2.5 μM) or T3 (up to 100 nM) as positive controls, respectively. Blanks, solvent controls, and positive controls hormones were included in each batch and four replicates of each sample were analyzed. Cells were then fixed with 4% paraformaldehyde in PBS for 10 min, washed 3X with PBS, stained with 4′,6-diamidino-2-phenylindole (DAPI) for 10 minutes and after 3 final washes with PBS were imaged using the Perkin Elmer Opera QEHS Image Screening System or the Yokogawa CV7000S high-throughput Imaging System. Alternatively, the plates were sealed and kept in PBS at 4°C for later imaging.

1.2.4. Automated imaging and analysis

A PerkinElmer (Waltham, MA) Opera QEHS High-Content Screening platform was used for fully automated confocal collection of images. This system employed a 40X water immersion objective lens, laser illuminated dual Nipkow spinning disk, and cooled charge-coupled device cameras to digitally capture high-resolution confocal fluorescence micrographs (323 nm pixel size with 2 X 2 camera pixel binning).

Additional imaging experiments were carried out using Yokogawa CV7000S high-throughput dual spinning disk confocal microscope. Images were acquired with a 40X Olympus PlanApoChromat air objective (NA 0.9) and two sCMOS cameras (2560 X 2160 pixels) using camera binning of 2 × 2 (Pixel size 325 nm). Samples were sequentially imaged using 488 nm and 405 nm excitation lasers and a 405/488/561/640 nm excitation dichroic mirror. Fluorescent signals were collected using the 561 nm emission dichroic mirror and BP525/50 and BP445/45 mirrors in front of the two sCMOS cameras, respectively.

1.2.5. Theory/calculation

An image analysis pipeline was customized using the Columbus software (Perkin Elmer) to segment automatically the nucleus using the DAPI channel and then construct a ring region (cytoplasm) around the nucleus mask for each cell in the digital micrograph. The pipeline automatically calculated the mean GFP intensity in both compartments using the GFP channel and translocation was calculated as a ratio of these intensities. Each value was further normalized to a control (DMSO) sample on the same plate. Data were analyzed using SigmaPlot v. 11 (SPSS Inc., Chicago, IL) as previously described (Stavreva et al., 2012; Stavreva et al., 2016). Briefly, the mean was computed across four replicates of each sample, and one-way ANOVAs compared activity of all tested samples to the negative controls following Holm-Sidak correction for multiple comparisons. The frequency for hormonal activities was based on double testing of the samples (two grab samples were tested from most locations, see Supplemental Table 1 for details). We scored any site being positive when at least one of the samples tested positive. Translocation responses to known concentration of activating hormones were applied to generate standard curves, that were used to estimate equivalent concentrations (in nM) of contaminants in the screened water samples. For a better comparison with the existing published data, in some cases, these were further converted to ng/L.

1.3. Results and Discussion

1.3.0. Samples distribution and collection

The water samples were collected over the span of 3 years at 45 locations along major water supplies in the state of Virginia (Figure 1). Samples were from the Rappahannock river and its tributaries, and the Mattaponi river (see Supplemental Table 1 for details). The Rappahannock River is the longest river in Virginia flowing from the Blue Ridge Mountains to Chesapeake Bay. The Mattaponi River is the second major watershed in Virginia and joins the York River which also flows into the Chesapeake Bay. The Mattaponi river is an excellent spawning and nursery habitat for several species of anadromous migratory fish including river herring, shad, and striped bass. In addition to being widely used for recreational activities, both rivers support key agricultural activities in Virginia, and provide water for livestock, irrigating crops, and other industries including blue crab, oyster, and fisheries. To assure the current and future health of these river ecosystems it is important to screen and monitor these river sources for the presence of emerging contaminants with endocrine disrupting potential.

1.3.1. Detection of androgen receptor-interacting contaminants

Androgen receptor at uninduced state resides in the cytoplasm and moves to the nucleus in response to binding by antagonists or agonists (Figure 2A). Nuclear AR acts as a transcription factor (Dasgupta, Lonard, & O’Malley, 2014) and is responsible for the physiological effects of testosterone and its more active metabolite, 5α-dihydrotestosterone (DHT). In mammals, testosterone is mainly produced by the testis in males and the ovary in females. In addition to male sexuality and reproductive health, androgens are also critical for the female reproductive system.

Figure 2.

Representative micrographs showing translocation of the GFP-tagged chimeric constructs in the present of their respective ligands (100nM of Testosterone, Dexamethasone and T3 for GFP-AR, GFP-GR and GFP-GR-TRβ, respectively and 2500nM CAY 19465 for the GFP-AhR-expressing cells). Scale bar 20 μm.

Many synthetic compounds potentially released into the environment have been studied for their endocrine disrupting potential. Some exhibit cross-reactivity with AR. For example, potent estrogenic compounds, such as α-zearalanol (α-ZA) and its derivatives, pesticide metabolites M2 compound and DDE, cosmetics such as benzophenone 2, and bisphenols, such as chlorinated BPA and BPC, are also anti-androgens, with affinities in the sub-to micromolar range (reviewed in (Toporova & Balaguer, 2020).To test for presence of AR-interacting contaminants, GFP-AR translocation to the nucleus in response to water samples was compared to the negative control (DMSO) and the activity quantified based on translocation induced by testosterone, the natural hormone (Figure 2A and Figure 3A). Surprisingly, most tested sites (70%) were positive for AR-interacting contaminants (red color in Figures 4, 5). Using the translocation response to known concentrations of testosterone, a standard curve was generated (Figure 3C) and the adjusted concentration of contaminants in the most positive sample (M1–17_R1, Figure 3A and B) was estimated as 1.04 ± 0.04 ng of testosterone equivalent (TestoEq)/L (N=4, ± SEM). This is in the lower range of previously reported levels of AR-agonist for water sources in US, estimated in the range of 1.6 to 4.8 ng DHTEq/L (Conley et al., 2017);

Figure 3.

Screening of water samples for AR-interacting contaminants. A) GFP-AR translocation induced by the 200x concentrated samples. B) Representative micrographs showing translocation of the GFP-AR from two positive samples (marked orange in A). Scale bar 20 μm. C) Linear response of the GFP-AR translocation to low levels (up to 0.5 nM) of Testosterone was used to estimate Testosterone equivalent (TestoEq) in the screened water samples. D) Calculated nM TestoEq in 1x samples based on the standard curve presented in (C).

Figure 4.

Detection of four classes of EDCs at different locations along Rappahannock river and its tributaries, Hughes, Robinson, and Rapidan rivers. Color indication for AR (red), GR (blue), AhR (yellow) and TR (green).

Figure 5.

Detection of four classes of EDCs at different locations along Mattaponi River. Color indication for AR (red), GR (blue), AhR (yellow) and TR (green).

In the last two decades, many studies revealed an increasing frequency of worldwide contamination of the water, soil and other environmental sources with agonists and antagonists of androgen alone or in combination with compounds which mimic other hormones, such as estrogen, thyroid and/or progesterone (Hotchkiss et al., 2008; Schug et al., 2016; Soto et al., 2004).

Monitoring environmental androgenic activities has been challenging because most efforts have been directed to identification of precise chemical structure for the various compounds which are present in small amounts in environmental samples (Backe, Ort, Brewer, & Field, 2011; Chang, Wu, Hu, Asami, & Kunikane, 2008) reviewed in (Scholz et al., 2013). Recent studies have shifted to functional detection of agonist and antagonist effects using cell-based yeast and mammalian cells reporter assays or whole organisms (amphibians and fish) (Hoffmann & Kloas, 2016; Liu, Ito, Kanjo, & Yamamoto, 2009; Scholz et al., 2013). These are important changes in providing meaningful information that can be used to establish limits for hormonal activities in the water, soil and other environmental samples. Considering the widespread presence of AR-interacting contaminants detected in most of the locations tested in our study, future studies are needed to determine whether these contaminations pose a risk for the health of the ecosystems and, thus indirectly to humans.

1.3.2. Detection of glucocorticoid receptor-interacting contaminants

Glucocorticoid receptor (GR) is expressed in most vertebrate tissues and mediates the actions of glucocorticoid hormones, a family of steroids involved in many critical physiological processes in mammals (Kadmiel & Cidlowski, 2013; Odermatt & Gumy, 2008). Previous experiments (Baxter & Tomkins, 1970); (Sibley & Tomkins, 1974) revealed that nuclear localization of the receptor-steroid complex follows steroid binding. Translocation of the GR from the cytoplasm to the nucleus (Figure 2B) was visualized using immunohistochemistry (Papamichail, Tsokos, Tsawdaroglou, & Sekeris, 1980). Subsequently, two distinct nuclear localization signals (NL1 and NL2) responsible for this translocation were defined (Picard & Yamamoto, 1987). Upon nuclear translocation, GR binds to chromatin targets to regulate gene expression (Voss & Hager, 2014).

Using a previously described highly sensitive mammalian cell line that expresses GFP-tagged GR (Walker, Htun, & Hager, 1999), we detected glucocorticoid receptor-interacting activity in 22% of 45 locations tested which are marked in Figures 4 and 5 (blue color) and detailed in Supplemental Table 1. The highest activity was detected in sample M1–11_R1 and was estimated at 2.34 ± 0.15 ng DexEq/L (N=4, ± SEM). A recent screen in US and Puerto Rico waters reported GR agonism in a range of 6.0 to 43 ng DexEq/L (Conley et al., 2017). These levels may be capable of inducing physiological effects in aquatic organisms. For example, exposure to cortisol at 16 ng/L was shown to decrease the locomotor activities of fish embryos (Zhao, Zhang, & Fent, 2018). Moreover, fludrocortisone acetate at environmental relevant concentrations decreased blood leukocyte number and altered gene expression and circadian rhythm in zebrafish (Zhao, Zhang, & Fent, 2016).

Glucocorticoids are among the most widely prescribed drugs due to their anti-inflammatory action (Barnes, 2011; Barnes & Adcock, 2009). Thus, their presence in the environment and especially in water is not surprising. Indeed, a study in the Netherlands found such activity in wastewater effluents and surface water in all tested samples (Van der Linden et al., 2008). Glucocorticoid activity was also found in river water, as well as in treated or untreated wastewater in the US, Switzerland, Czech Republic, and China (Berninger et al., 2019; Conley et al., 2017; Jia, Wu, Daniels, & Snyder, 2016; Jones et al., 2020; Macikova, Groh, Ammann, Schirmer, & Suter, 2014; Stavreva et al., 2012). In addition, activities were detected in extracts of agricultural soil in China combined with contaminations interacting with estrogen receptor (ER), AR, progesterone receptor (PR), and mineralocorticoid receptor (MR) (Zhang et al., 2018). A growing body of literature implicates many environmental contaminants (e.g. metals, metalloids, pesticides, bisphenol analogues, plasticizers, flame retardants and pharmaceuticals) and chemicals in disrupting GR (Zhang, Yang, Liu, Schlenk, & Liu, 2019).

The presence of GR-interacting contaminants in the environment is of increasing interest because dysregulated GR signaling is associated with immune diseases, allergies, mood and cognitive disorders, metabolic disorders, cardiovascular disease, and cancers (Carnahan & Goldstein, 2000; Odermatt & Gumy, 2008).

1.3.3. Detection of aryl hydrocarbon receptor-interacting contaminants

The aryl hydrocarbon receptor (AhR) is a nuclear receptor that regulates transcription through binding to DNA following activation by exogenous and endogenous ligands. Similarly to AR and GR, AhR resides in the cytoplasm in an uninduced state and translocates to the nucleus in the presence of specific ligands (Figure 2C). Initially named “dioxin” receptor for its discovery using radiolabeled TCDD in the 1970s, AhR has since been found to interact with many natural and man-made compounds (Hale et al., 2017). While first identified as the mediator of dioxin-induced toxicity, the AhR has been implicated in numerous processes including cell proliferation, adhesion and migration, birth defects, immune system function, neurotoxicity, lethality, tumor promotion, and changing enzymatic functions (e.g., CYP1A1/2 and CYP1B1) induction (Kolluri, Jin, & Safe, 2017; Larigot, Juricek, Dairou, & Coumoul, 2018; Xie et al., 2016).

We used a cell line expressing GFP-tagged AhR to test for presence of AhR-interacting activities (Jones et al., 2020). In the current study, we detected abundant AhR activity in Virginia river samples in 60% of tested locations (indicated by yellow color in Figures 4 and 5). Based on standard curve generated by using CAY19465 as a standard for AhR activation, activity in the most positive sample (M2–1_R2) was estimated to be 1342.56 ± 48.96 ng CAY 19465Eq /L (N=4, ± SEM) (see Supplemental Figure 2 for details). These high levels are a result of a relatively low translocation efficiency in response of GFP-AhR to CAY 19465 which was detected at concentrations above 150nM CAY 19465 and plateaued at 2500nM (data not shown). We refrained from using dioxin as AhR ligand because it is highly toxic and represents a human and environmental hazard.

The presence of chemicals with known affinity to the AhR in environmental water samples have been documented since the mid-1900s, and many of these chemicals persistent as contaminants today (Dyke, Foan, Wenborn, & Coleman, 1997; Zgheib et al., 2018). The most prevalent strategies for detecting AhR activity in water samples have been similar to those used for other hormone receptors. These include gas or liquid chromatography-tandem mass spectrometry (GC-/LC-MS), cell-based reporter gene assays, or in vivo models measuring physiological endpoints (Otarola, Castillo, & Marcellini, 2018).

Detection of AhR-mediated activities in waterways has been well documented in the past decade (Brack, Klamer, Lopez de Alda, & Barcelo, 2007; Chou, Matsui, & Matsuda, 2006; Dagnino et al., 2010). Most known AhR ligands are hydrophobic compounds that enter the waterways through fuel combustion, waste incineration, and runoff or leaching from landfills (Brack et al. 2007). Pesticides also represent a major source of contamination. A screen of 200 pesticides identified 11 (acifluorfen-methyl, bifenox, chlorpyrifos, isoxathion, quinalphos, chlorpropham, di- ethofencarb, propanil, diuron, linuron, and prochloraz) as AhR ligands (Kjaerstad, Taxvig, Nellemann, Vinggaard, & Andersen, 2010). In addition, many plant- and organism-derived polyphenols also modulate AhR activity, such as the isoflavone resveratrol in red wine (Nguyen et al., 2015), the flavonoid curcumin in turmeric (Ciolino, Daschner, Wang, & Yeh, 1998; Nishiumi, Yoshida, & Ashida, 2007), and epigallocatechin in green tea (Palermo, Hernando, Dertinger, Kende, & Gasiewicz, 2003). Interestingly, these molecules act as AhR antagonists, which may be beneficial (Choi et al., 2008). However, the notion that AhR agonists are toxic and antagonists are therapeutically beneficial seem simplistic in light of recent information. For example, indigo naturalis, an AhR agonist, promotes restoration and intestinal integrity in inflammatory bowel disease (Mizoguchi et al., 2018).

In sum, dioxins and dioxin-like EDCs exert their biological and toxicological activities through activation of AhR and these compounds can have biological effects even at low levels of exposure (Furue & Tsuji, 2019). Based on current information, the detected widespread contamination of Virginia rivers with AhR-interacting compounds and their possible consequences on fish and river ecosystems should be examined further.

1.3.4. Detection of thyroid receptor beta-interacting contaminants

Thyroid hormones (THs) are iodine-containing hormones indispensable for normal development, growth, and metabolism of most cells and tissues. The predominant TH secreted by the thyroid gland is 3,3′,5,5′-tetraiodothyronine (thyroxine, T4), which is the precursor for the active T3 (3,3′,5-triiodothyronine) produced by partial deiodination in peripheral tissues. THs exert their action primarily by activation of thyroid hormone receptors (TR α and β) expressed in most tissues with a specific pattern during development, depending on their function (Cheng, Leonard, & Davis, 2010; Tancevski, Rudling, & Eller, 2011; T. R. Zoeller et al., 2002).

TRs, similar to other NR, are ligand-dependent transcription factors that regulate gene expression by interaction with thyroid hormone response elements (TREs) in the promoter/enhancers DNA loci (Sap et al., 1986; Weinberger et al., 1986). Due to their wide expression, TRs regulate many processes in human physiology from metabolism, bone formation and cardiac output, to neuronal development (Cioffi, Gentile, Silvestri, Goglia, & Lombardi, 2018; Duncan Bassett & Williams, 2018; Gilbert, Rovet, Chen, & Koibuchi, 2012; Williams, 2008; Yen, 2001; R. T. Zoeller, Tan, & Tyl, 2007).

We tested TRβ translocation to the nucleus using a GFP-GR-TRβ construct which renders the receptor cytoplasmic in the absence of a ligand (Figure 2D, see inset in Supplemental Figure 3A) (Stavreva et al., 2016). Using this strategy, we have previously identified novel TRβ antagonists from the Tox21 chemical library (Paul-Friedman et al., 2019) as well as TRβ-interacting environmental contaminants (Jones et al., 2020; Stavreva et al., 2012; Stavreva et al., 2016). In the present study, we detected TRβ interacting contaminants in 42% of river samples from 45 tested locations (indicated by green color in Figures 4 and 5 see also Supplemental Figure 3 for details). The highest TR activity was detected in sample M1–18_R1, estimated to be 5.73 ± 0.6 ng T3 Eq/L (N=4, ± SEM). Using the TR antagonist amiodarone hydrochloride (AH) as a standard, a much higher TR (antagonistic) activity in the range of 21.2 to 313.9 μg AH Eq/L were reported in water samples in China (Li, Ren, Han, & Li, 2014).

Studies of disrupted thyroid function have included efforts to identify specific ECDs in environmental samples (Fini et al., 2007; Grimaldi et al., 2015; Scholz et al., 2013; Steinberg, 2013). However, thee critical issue is the detection of environmental TR-interacting compounds without the need for costly and time-consuming characterization of their molecular structures. Previous in vitro approaches used several nuclear TR transactivation assays (Murk et al., 2013). Based on an extensive review, the panel recommended a battery of methods to classify chemicals of lesser or higher concern for further hazard and risk assessment. Considering that the disruption of thyroid hormone signaling in a multicellular organism may trigger complex adverse outcomes, additional approaches to assess the disruption are being currently evaluated (Noyes et al., 2019). Addressing the central issue of TR activation, the GFP-GR-TRβ translocation assay is an alternative in vivo, activity-based method for detecting both, TRβ agonists and antagonists. Because thyroid function is particularly critical during early development, detection of TR-interacting activities in the water of major Virginia rivers is of great concern. Although the sources of such contaminants are unknown, some medications frequently used for hypothyroidism, such as Synthroid, could be present in the rivers. The sources of this contamination should be investigated further.

1.4. Conclusions

Exposures to EDCs are very costly by increasing disease and disability (Kassotis et al., 2020). Identification of substances having endocrine-disrupting potential and their regulation is carried out in EU, where substantial efforts are directed into developing comprehensive effect-based methods. At present, 205 substances, of which 16 are also endocrine disruptors, are included in the EU list of substances of very high concern (SVHC) and are subject to increased regulatory scrutiny and higher reporting standards. Moreover, ambitious multifaceted projects, such as the SOLUTIONS Project (https://www.solutions-project.eu/), coordinate efforts to develop methods for detection and monitor of EDCs as well as efficient policies to reduce potential human exposure. In the USA, regulations are strictly risk-based and pre-marketing testing is usually not required. In addition, most screening and testing efforts focus on estrogenic contaminants. Less is known about contamination with other classes of EDCs. The data presented here reveals contamination of rivers with four classes of EDCs. Multiple endocrine disrupting activities were detected in water samples collected from 45 locations along Virginia rivers. Many locations were positive for multiple activities, and 40% of samples were positive for at least three of the four tested activities. These findings suggest the presence of complex mixtures of EDCs, where individual biological activities could influence each other. Future studies are needed to address the possible effects of these activities on fish and wildlife in the screened locations. An important next step will be to also evaluate endocrine disrupting activities in samples of drinking water in the same areas to identify and characterize possible human exposure.

Supplementary Material

1

Supplemental Figure 1. Screening of water samples for GR-interacting contaminants. A) GFP-GR translocation induced by the 200x concentrated samples. B) Calculated nM Dexamethasone equivalent (DexEq) in 1x samples based on the standard curve of the GFP-GR translocation to low levels (up to 2.5 nM) of Dexamethasone (inset).

2

Supplemental Figure 2. Screening of water samples for AhR-interacting contaminants. A) GFP-AhR translocation induced by the 200x concentrated samples. B) Calculated nM CAY10465 equivalent (CAYEq) in 1x samples based on the standard curve of the GFP-AhR translocation to up to 600 nM of CAY10465 (inset).

3

Supplemental Figure 3. Screening of water samples for TRβ-interacting contaminants. A) Cells expressing the GFP-GR- TRβ chimera (inset) were used to assess the translocation induced by the 200x concentrated samples. B) Calculated nM T3 equivalent (T3Eq) in 1x samples based on the standard curve of the GFP-GR- TRβ translocation to low levels (up to 1 nM) of T3 (inset).

4

Supplemental Table 1. Location of the tested sites and day of the samples collection.

Highlights.

• A novel imaging-based assay enables characterization of individual and mixtures of endocrine disrupting activities

• Four classes of bioactivities were assessed in river water from the state of Virginia

• Androgenic and aryl hydrocarbon activities were most prevalent

• Glucocorticoid and thyroid activities were less common

• Multiple endocrine disrupting activities were present at many tested sites

Abbreviations:

AR

androgen receptor

GR

glucocorticoid receptor

AhR

aryl hydrocarbon receptor

TR

thyroid receptor

Dex

dexamethasone

Testo

Testosterone

T3

3,3′,5-Triiodo-L-thyronine sodium salt powder

CAY

CAY 10465 aryl hydrocarbon receptor agonist

DMSO

dimethyl sulfoxide