De Facto Water Reuse: Bioassay suite approach delivers depth and breadth in endocrine active compound detection

Abstract

Although endocrine disrupting compounds have been detected in wastewater and surface waters worldwide using a variety of in vitro effects-based screening tools, e.g. bioassays, few have examined potential attenuation of environmental contaminants by both natural (sorption, degradation, etc) and anthropogenic (water treatment practices) processes. This study used several bioassays and quantitative chemical analyses to assess residence-time weighted samples at six sites along a river in the northeastern United States beginning upstream from a wastewater treatment plant outfall and proceeding downstream along the stream reach to a drinking water treatment plant. Known steroidal estrogens were quantified and changes in signaling pathway molecular initiating events (activation of estrogen, androgen, glucocorticoid, peroxisome proliferator-activated, pregnane X receptor, and aryl hydrocarbon receptor signaling networks) were identified in water extracts. In initial multi-endpoint assays geographic and receptor-specific endocrine activity patterns in transcription factor signatures and nuclear receptor activation were discovered. In subsequent single endpoint receptor-specific bioassays, estrogen (16 of 18 samples; 0.01 to 28 ng estradiol equivalents [E2Eqs]/L) glucocorticoid (3 of 18 samples; 1.8 to 21 ng dexamethasone equivalents [DexEqs]/L), and androgen (2 of 18 samples; 0.95 to 2.1 ng dihydrotestosterone equivalents [DHTEqs]/L) receptor transcriptional activation occurred above respective assay method detection limits (0.04 ng E2Eqs/L, 1.2 ng DexEqs/L, and 0.77 ng DHTEqs/L) in multiple sampling events. Estrogen activity, the most often detected, correlated well with measured concentrations of known steroidal estrogens (r² = 0.890). Overall, activity indicative of multiple types of endocrine active compounds was highest in wastewater effluent samples, while activity downstream was progressively lower, and negligible in unfinished treated drinking water. Not only was estrogenic and glucocorticoid activity confirmed in the effluent by utilizing multiple methods concurrently, but other activated signaling networks that historically received less attention (i.e. peroxisome proliferator-activated receptor) were also detected.

Graphical Abstract

1. Introduction

Domestic population growth and depletion of freshwater sources have increased pressure on wastewater and drinking water treatment facilities to effectively and efficiently treat incoming waters, as well as increased the need for direct water reuse (UNWAPP, 2017). Further, the existence of complex mixtures of anthropogenically derived chemicals in freshwater environments has long been established (Bradley et al., 2017; Kolpin et al., 2002); yet wastewater (Bain et al., 2014; Jia et al., 2015; Jia et al., 2016) and drinking water (Jia et al., 2015; Snyder, 2008) treatment processes eliminate contaminants with variable removal efficiencies.

Researchers have historically relied primarily on targeted analytical chemistry approaches to measure contaminants in water sources. However, recent biotechnological innovations have provided cell- and tissue-based effects-based methods (EBMs) to detect chemical activity specific to a mechanism of action, or molecular initiating event, that are related to an adverse physiological outcome (Ankley et al., 2010; Escher and Leusch, 2012). The inherent ability of these tools to detect cumulative mixtures of chemicals with similar molecular mechanisms of action not only enables environmental sample screening for contaminant mixture effects, but also for contaminants that may be present below their analytical detection methods (Altenburger et al., 2015; Brack et al., 2019; Brack et al., 2015). Recent monitoring studies have successfully applied these EBMs, or bioassays, alongside targeted analytical chemistry methods for water quality screening (Cavallin et al., 2014; Könemann et al., 2018; Tousova et al., 2017; van der Linden et al., 2008).

In a prior collaboration between researchers at the U.S Geological Survey (USGS) and the U.S. Environmental Protection Agency (USEPA), over 700 unique organic chemical analytes were examined in impacted surface waters (Bradley et al., 2017), while endocrine (estrogen, androgen, and/or glucocorticoid) activity was also detected in nearly all tested waters using bioassays (97%, 14%, 26% of water samples, respectively) (Blackwell et al., 2019; Conley et al., 2017a). Notably, none of the selected glucocorticoid receptor ligands were measured above analytical method detection limits in the surface waters despite high glucocorticoid-like activity detected using an in vitro bioassay (Conley et al., 2017a). These results demonstrate the advantage of assessing cumulative biological activity with non-targeted effects-based methods and further support the concurrent application of the two types of screening approaches.

The function of this inter-agency De Facto Water Reuse collaboration was to identify the presence of, and potential biological activity from, contaminants of emerging concern (CECs; known endocrine disrupting compounds, perfluoroalkyl substances, inorganics, antibiotics and pharmaceuticals) emanating from wastewater effluent and to determine if natural degradation and drinking water treatment processes collectively attenuate such contaminants as water moves downstream and through a drinking water treatment plant. Although targeted and non-targeted effects-based screening approaches have been used to monitor the persistence, degradation, or remineralization of contaminants in a large river systems impacted by wastewater (Daniels et al., 2018; König et al., 2017), none, to our knowledge, have integrated the natural environmental degradation processes (microbial, UV light, etc.) on contaminants with the attenuation of drinking water treatment. The aim of this study was to assess the combined effects of the natural degradation and drinking water treatment attenuation processes on biological activity in six sites along a river in the northeastern United States (USEPA Region 1) using a suite of in vitro bioassays, generally focused on the detection of endocrine activity. However, other molecular signaling pathways also were represented herein and within the larger De Facto collaboration (Zhen et al., 2018).

The presence of endocrine disrupting compounds (EDCs) in environmental matrices continues to be a concern due to the well-characterized adverse effects on reproduction and development of resident aquatic species (Guillette et al., 1994; Kidd et al., 2007; Matthiessen and Gibbs, 1998). Also, there are more recent reports of potential human exposure to EDCs through both surface (Kolpin et al., 2002) and drinking water resources (Benotti et al., 2009; Glassmeyer et al., 2017). In addition to reproduction (Jobling et al., 1998; LaLone et al., 2012; Lee Pow et al., 2017) and development (Beverly et al., 2014; Conley et al., 2018; Wang et al., 2011) in mammals and aquatic species, EDCs can affect other physiological processes including metabolism (Inadera and Shimomura, 2005; Kanayama et al., 2005; Kugathas and Sumpter, 2011) and bone mineralization (Curkovic et al., 2013), frequently at relatively low exposure concentrations; thus illustrating the importance of broadening chemical screening suites to represent a variety of molecular signaling pathways.

2. Materials and Methods

2.1. Sample Collection

Residence time-weighted (time-integrated) samples were collected along a stream reach in the northeastern United States (USEPA Region 1; due to a client anonymity agreement the exact study location cannot be disclosed) beginning upstream from a waste water treatment plant (WWTP; 1.2 million gallons/day, serving 1,700 residents), and through a downstream drinking water treatment plant (DWTP; 4.5 million gallons/day, serving 37,000 residents; Table 1). While the WWTP used advanced secondary treatment with ultraviolet disinfection, the DWTP used a combination of ozone and chloramination for disinfection. Samples from all six sites and field blanks were collected during the fall (October 2014; streamflow: 0.08 meters³/second [m³/sec]), spring (April 2015; streamflow 0.4 m³/sec), and summer (August 2015; streamflow: approximately 0.08 m³/sec). Sampling sites included a location 1.2 km upstream from the WWTP, the effluent outfall, a downstream mixing zone (0.62 km from effluent), a site further downstream (9.1 km from effluent), at the DWTP intake (14.6 km from effluent) and unfinished treated water. Water samples were collected from the middle of the stream reach by complete submersion of 4 L silanized amber glass bottles, and the DWTP intake and treated water samples were collected from DWTP source and unfinished treated water taps, respectively. Samples were stored and transported overnight at < 4 °C.

Table 1. Sample Site Description.

Sampling was conducted at six points along a river between a WWTP and downstream DWTP. Each site was sampled three times, during fall (October 2014), spring (April 2015), and summer (August 2015) months.

Site Description	Representative Sample Season	Collection Date (MM/DD/YYYY)
upstream	Fall	10/7/2014
upstream	Spring	4/28/2015
upstream	Summer	8/3/2015
effluent pipe	Fall	10/7/2014
effluent pipe	Spring	4/28/2015
effluent pipe	Summer	8/3/2015
mixing zone	Fall	10/7/2014
mixing zone	Spring	4/28/2015
mixing zone	Summer	8/3/2015
downstream	Fall	10/8/2014
downstream	Spring	4/28/2015
downstream	Summer	8/5/2015
DWTP intake	Fall	10/7/2014
DWTP intake	Spring	4/28/2015
DWTP intake	Summer	8/5/2015
treated water	Fall	10/8/2014
treated water	Spring	4/29/2015
treated water	Summer	8/5/2015

2.2. Attagene Inc. High Content Screening Assays

Sample and field blank extracts from the first sampling season (Oct 2014; six sample sites and three field blanks) were extracted as described previously (Blackwell et al., 2019), then quantitatively assessed for the activation of 29 nuclear receptors and 52 transcription factor signatures using the Trans-FACTORIAL™ and Cis-FACTORIAL™ (Attagene Inc., Morrisville, NC) multiplexed assays, respectively (AttageneInc., 2018). This multi-endpoint assessment was limited to a single replicate of each sample extract dilution due to cost.

High-Performance Liquid Chromatography (HPLC)-grade solvents were purchased from Fisher Scientific (Waltham, MA) and dimethyl sulfoxide (DMSO; purity: >99.7%) was purchased from Sigma-Aldrich (St. Louis, MO). Approximately 1 L whole water was filtered through a baked 47 mm glass-fiber filter (0.7 μm porosity). After filtration, 0.5 L of each water sample/blank was concentrated by solid-phase extraction (200 mg sorbent, 5 mL glass cartridge; Waters Oasis HLB, Milford, MA) using preconditioned cartridges (5 mL ethyl acetate, 5 mL 50:50 methanol (MeOH): dichloromethane (DCM), 5 mL 100% MeOH, and 5 mL ultrapure water). Cartridges were aspirated for 30 min to remove residual water then eluted sequentially with 6 mL MeOH, then 6 mL of 50:50 MeOH: DCM. Extracts were evaporated to dryness under a gentle stream of nitrogen at 30 °C, reconstituted in 500 μL of DMSO and frozen at −20°C until analysis in bioassays. Final sample enrichment factor was 1000-fold, where the lowest test dilution was 41x more concentrated than ambient concentrations of water samples (1000x extraction concentration and 0.041x testing dilution).

Positive controls (20 μM chenodeoxycholic acid, 0.1 μM 1α,25-dihydroxyvitamin D3, 10 μM dexamethasone, 0.2 μM estradiol, 2.0 μM 6α-fluorotestosterone, 0.1 μM retinoic acid, 0.1 μM 9-cis-retinoic acid, 1.0 μM rosiglitazone, 10 μM rifampicin, 15 μM triiodothyronine, 1.0 μM T0901317, and 0.2 μM GW0742) were dissolved in DMSO (≤ 1% DMSO final exposure concentration). Samples and field blank extracts were assessed in a concentration-response manner with six 1:3-fold dilutions (10.0, 3.33, 1.11, 0.37, 0.12, and 0.041 relative extract concentration factor), where n = 1 replicate for each sample dilution treatment.

2.3. Estrogenic, Androgenic, Glucocorticoid Activity

2.3.1. Sample Extraction

Water from each sample (6 sites x 3 sampling events; total 18 samples) also were extracted using methods described previously (Conley et al., 2017a) for assessment in the transcriptional activation bioassays and chemical analyses. Each 600 mL aqueous sample was concentrated by solid phase extraction (47 mm C18 Empore® disks, 3M, St. Paul, MN). Disks were eluted with 10 mL 90:10 methanol:acetone (volume: volume). Two 5 mL aliquots of extract were removed (equivalent to 300 mL field sample) and concentrated to dryness. One of the paired aliquots was used for the steroidal estrogen chemical analysis and the other was shipped overnight on ice for in vitro analysis.

2.3.2. Transcriptional Activation Assays

For the transcriptional activation assays 17β-estradiol (E2; purity ≥ 98%; catalog no. E887; lot: 28H0818), 4,5αdihydrotestosterone (DHT; catalog no. D-073; lot: 104H04971), and dexamethasone (purity ≥97%; catalog no. D4902; Lot: BCBH2988V) were purchased from Sigma-Aldrich and served as reference chemicals for standard curve production. ICI 182,780 (purity ≥ 99%; catalog no. 1047 batch: 20A/116982; Torcis Bioscience, Minneapolis, MN) and hydroxyflutamide (OHF; purity ≥ 98%; catalog no. H4166; lot: 025M4732V; Sigma-Aldrich), in competition with respective reference compound, served as antagonist controls. Standards and controls were dissolved in ethanol and serially diluted for standard curve stocks. Dried water extracts were resuspended with 100 μL ethanol in original glass shipping vials. Standards, controls and samples were diluted in respective cell media for exposures with ethanol concentration ≤ 0.002%.

All cells for estrogen, androgen and glucocorticoid activity screening assays, were plated in 96-well luminometer plates (Greiner Bio-one, Germany, catalog no. 65098 for estrogen assay and Costar, Corning Inc., USA, catalog no. 3610 for estrogen, androgen and glucocorticoid assay). Each sample, including standards, controls, and water extracts were run in quadruplicate, and each sample screen was at least duplicated (i.e. different cell passage number). After 24 hr in vitro exposure, cells were washed with Dulbecco’s phosphate buffered saline (ThermoFisher Scientific, USA) and mixed with 25 μL lysis buffer (luciferase cell culture lysis 5x reagent, Promega) per well. The relative amount of luciferase production resulting from sample exposure was quantified using a luminometer (BMG FLUOstar® Omega; BMG Labtech, Cary, NC). Well readings occurred for 5 secs immediately following injections of 25 μL luciferase reaction buffer and 25 μL firefly luciferase substrate.

2.3.3. Estrogenicity

The T47D-KBluc (American Type Cell Culture, Manassas, VA; #CRL-2865) stable cell line has been developed by researchers at the USEPA (Wilson et al., 2004; Wilson et al., 2002) and successfully applied to chemical and environmental sample testing for estrogenic activity (Conley et al., 2017b; Conley et al., 2016). Cell culture maintenance and sample screening were conducted as previously described (Conley et al., 2017b; Wilson et al., 2004). Cells were maintained in RPMI media supplemented with 10% Fetal Bovine Serum (FBS) and one week prior to assay, confluent cells were transferred to withdrawal media (RPMI +10% dextran-coated charcoal treated FBS; DCC-FBS). On day 7, cells were resuspended in DCC-FBS (Dulbecco’s Modified Eagle’s Medium) DMEM and plated at a density of 1 × 10⁴ cells/well. The following day, media was replaced with serially-diluted water samples, ICI, or E2 standards (0, 0.3, 1.0, 3.0, 10, and 30 pM) in 10% DCC-FBS DMEM.

2.3.4. Androgenicity

The MDA-kb2 (ATCC, #CRL-2713) stable cell line was developed by investigators at the USEPA (Wilson et al., 2002) and has been successfully applied to environmental sample testing for androgenic activity (Conley et al., 2017a). Sample screening for androgenic activity was conducted as previously described (Cavallin et al., 2014; Conley et al., 2017a), with some exceptions. Cells were maintained in Leibovitz’s L-15 media (Invitrogen™) supplemented with 10% FBS and 100 U/mL penicillin, 100 U/mL streptomycin, and 0.25 μg/L Amphotericin B (Invitrogen™). Cells resuspended in L-15 were plated at a density of 1 × 10⁴ cells/well. After 4–5 hrs of incubation at 37°C, or after cells adhered to well bottom, media was replaced with samples, OHF, or DHT standards (0, 0.01, 0.03, 0.1, 0.3, and 1.0 nM), in L-15 media.

2.3.5. Glucocorticoid Activity

The CV1 cell line (ATCC, #CCL-70), which is naturally devoid of glucocorticoid and androgen receptors was transfected with adenoviral vector expressing the human glucocorticoid receptor (hGR) as previously described (Conley et al., 2017a; Hartig et al., 2002). Briefly, cells were maintained at 37°C in 10% DCC-FBS RPMI and sub-cultured into a 60nM petri dish one week prior to assay. Approximately 5 × 10⁶ cells were inoculated with adenovirus containing the human GR (Ad/GR4) and MMTV-luc (Ad/Luc7), with a multiplicity of infection ratio of 1 and 1.3, respectively. Cells were resuspended in 5% DCC-FBS RPMI, plated at a rate of 2.2–3.3 × 10⁵ cells/well, and exposed to dexamethasone standard curve concentrations (0, 0.003, 0.01, 0.03, 0.1, 0.3, 1.0, 3.0, 10, 30 nM) or samples for 24 hrs. Plates were stored at −80 °C with lysis buffer and thawed to room temperature prior to analysis.

2.4. Steroidogenesis Assay

Effects on steroidogenesis were assessed in vitro by incubating gonadal tissue from sexually mature female fathead minnows (Pimephales promelas) in unextracted sample water from all sampling events. Concentrations of testosterone, or estradiol, produced by the ovarian tissue were measured in the incubation media post-exposure using a radioimmunoassay as described previously (McMaster et al., 1995), with alterations (Ankley et al., 2007).

Three surface water samples and one control sample were tested in each 48-well plate. Powdered Medium 199, supplemented with powdered NaCHO₃ and L-glutamine was dissolved in a 12 mL volume of sampled water or ultrapure water (control) and then sterilized using a syringe driven 0.22 μm filter. Medium was then spiked with 1 μg 25-hydroxy cholesterol (Sigma-Aldrich catalog no. H1015)/mL to serve as a substrate for steroid production. For each sample, or control, twelve 500 μL aliquots of the appropriate medium was transferred to glass culture tubes and placed on ice. Ovary tissue from a 10 sexually mature female fathead minnow (4–6 months old) was split into the four 10–20mg subsections. Tissue from a single fish was added to each control and three treated media tubes, yielding 10 biological replicates for each control or sample.

Sample medium with no tissue served as sample-specific assay blanks (eight wells/plate accounting for steroids or other chemicals present that could interfere with radioimmunoassay). Samples were then transferred to a 2 5^oC water bath for a 12 hr incubation and placed back on ice to slow/stop steroid production. Sample medium was removed from each tube and stored at −20°C until assessed via radioimmunoassay using estradiol (Perkin-Elmer, Boston, MA, USA; Lot#: 2088404) and testosterone (Perkin-Elmer; Lot#: 2056433) radioactive tracers as described previously (Jensen et al., 2001).

2.5. Chemical Analysis

Of the 236 chemicals screened in the collaborative De Facto Water Reuse study, four known steroidal estrogen receptor agonists were quantified as using solid anhydrous analytical standards of estrone (E1), 17β-estradiol (E2), ethinyl estradiol (EE2), and estriol (E3) (Steraloids, Newport, RI). The labeled internal standard, estriol-d₃ (CDN Isotopes, Pointe-Clare, Canada), ¹³C₆-estradiol, ¹³C₆-estrone, and ¹³C₂-ethinyl estradiol (Cambridge Isotopes Laboratory, Tewksbury, MA) were purchased and applied in analytical chemical detection methods as described previously (Conley et al., 2017a; Schenck et al., 2015). Sample extracts were spiked with labeled analogues of each target analyte and concentrated to dryness. The dried residue was resuspended in 0.1 mL aqueous 0.1 M NaHCO3, and 0.1 mg dansyl chloride in 0.1 mL acetone was added and allowed to react at 70 °C for 5 min. The solution was extracted with three aliquots 0.5 mL hexane. The combined organic layers were concentrated to dryness and stored at −20 °C until resuspension and analysis. Concentrates were resuspended in 50:50 methanol:water and determination of the target analytes was performed by Liquid Chromatography/tandem Mass Spectrometry (LC-MS/MS) with chromatographic separation using a Restek Pinnacle® DB Biphenyl analytical column (Bellefonte, PA) and detection using a ThermoScientific™ Vantage Triple Stage Quadrupole (Waltham, MA) operated in the selective reaction monitoring mode. Identification of target analytes was based on the presence of both quantitation and confirmation product ions. Quantitation was performed using internal standard calibration. The lowest concentration minimum reporting level (LCMRL: typically applied to Unregulated Contaminant Monitoring Regulation for finished drinking water from DWTPs is a single laboratory-determined value that is the lowest true concentration that future analyte recovery is predicted with at least 99% confidence to fall between 50 and 100% ) (USEPA, 2010) and method detection limit (MDL) values (in ng/L) were E1: 0.28, 0.064; E2: 0.26, 0.072; EE2: 0.14, 0.064; and E3: 1.38, 0.46, respectively.

2.6. Data and Statistical Analysis

Data analyses were completed and graphs were generated using GraphPad Prism version 7.02 for Windows (GraphPad Software, LaJolla California, USA), while statistical analysis was performed in SAS (Cary, NC, USA). Results for the potential activation of 29 nuclear receptors (Trans-FACTORIAL™) and 52 transcription factors (Cis-FACTORIAL™) were filtered for endpoints with well-defined positive controls. Fold induction values (relative to DMSO controls) were normalized to their respective positive control. Water extract dilutions with fold induction increases greater than three standard deviations above the mean field blank values, in at least two consecutive sample dilutions (since n = 1 for dilution treatment), were identified as active for the specific endpoint and are presented graphically. The percentage of maximal activation for activated endpoints was plotted and the area under the sample dilution-response curve (AUC) was calculated for each endpoint.

In the estrogen, glucocorticoid, and androgen receptor transcriptional activation assays, relative light unit (RLU) values were normalized to background (mean RLU value of vehicle control treated cells), data were then log₁₀ transformed and converted to the percent maximal response of respective hormone standard for each plate replicate. Half maximal effective concentrations (EC₅₀ values) were determined by plotting the log₁₀ (standard concentration) versus the percent maximum response and fit to a four-parameter concentration-response curve using nonlinear regression. Biological equivalency (BioEq) values were determined for each sample extract using the equation,

$$\mathrm{B}\mathrm{i}\mathrm{o}\mathrm{E}\mathrm{q}=\frac{\left(\mathrm{R}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}{\mathrm{E}\mathrm{C}}_{50}\right)}{\left(\mathrm{S}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}{\mathrm{D}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{50}\right)\left(\mathrm{S}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{E}\mathrm{F}\right)}$$

where all concentrations are expressed in ng/L and the sample enrichment factor (EF) is the degrees to which the water samples were concentrated during extraction (EF = 3000-fold).

The minimum detectable concentration (MDC) for hormone activity was determined for each of the three transcriptional activation assays as described previously (Conley et al., 2017b). The MDL, which integrates sample concentration that occurred during the extraction process was determined using the study enrichment factor and assay-specific MDC where,

$$\mathrm{M}\mathrm{D}\mathrm{L}=\frac{\mathrm{M}\mathrm{D}\mathrm{C}}{\mathrm{S}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{E}\mathrm{F}}$$

Mean steroid hormone concentrations from the steroidogenesis assay tissue incubation media were normalized to a per wet tissue weight concentration. Data are presented as fold concentration changes compared to control (media without water extract) as described previously (Ekman et al., 2011). An analysis of variance, followed by Tukey’s multiple comparison procedure (GraphPad Software, 2017), was conducted to identify significant differences (p < 0.05) in steroid hormone production between tissues exposed to sample medium (i.e., prepared from river water samples) versus those exposed to control medium (i.e., prepared using ultrapure water) for each assay.

3. Results and Discussion

During this collaborative De Facto Water Reuse study, extracts from water samples collected along a river in USEPA Region 1 (between a WWTP and DWTP; Table 1) were analyzed for activation of molecular initiating events associated with potential adverse physiological outcomes, including endocrine disruption, and for the presence of over 200 selected CECs. The strategically selected sites permitted assessment of the contributions of natural (from the river) and the anthropogenic attenuation processes (related to the DWTP) that may be affecting CEC exposure concentrations.

A complete summary of the chemical analysis has been reported elsewhere (Glassmeyer et al., 2018). An abbreviated version containing concentrations of four known environmental steroidal estrogens is provided in Table 2. Also reported herein are results from concurrent water quality screens using several in vitro assays that have been applied previously to infer specific chemical toxicity (Ankley et al., 2007; Blake et al., 2010; Martin et al., 2010), as well as to detect potential endocrine disrupting activity in water extract samples (Blackwell et al., 2019; Cavallin et al., 2014; Conley et al., 2017a). As in several previous water quality screening studies (Blackwell et al., 2019; Cavallin et al., 2014; Jia et al., 2015; Jia et al., 2016; Neale et al., 2015), bioassays were chosen that would identify pathway signaling disruption by potential contaminants at multiple molecular initiating events, i.e. altered production of endogenous steroid hormones, nuclear hormone receptor binding, and receptor-mediated transcriptional activation in a single adverse outcome pathway.

Table 2. Environmental Estrogen Concentrations.

Known steroidal estrogen receptor ligands were detected in water samples collected from sites between a WWTP and downstream DWTP using LC-MS/MS. Samples with estrogen concentrations at or below the lowest concentration minimum reporting limit (LCMRL; estrone: 0.28, 17β-estradiol: 0.26, estriol: 1.38, and ethinyl estradiol: 0.14 ng/L) are italicized and estrogen concentrations below method detection limits (<MDL: estrone:0.064, 17β-estradiol: 0.072, estriol: 0.46, and ethinyl estradiol: 0.064 ng/L) are also identified.

Fall 2014	Estrone	17β-Estradiol	Estriol	Ethinyl Estradiol
upstream primary	0.25	<MDL	<MDL	<MDL
upstream duplicate	0.36	<MDL	<MDL	<MDL
effluent primary	13	1.0	3.3	<MDL
effluent duplicate	12	1.0	2.8	<MDL
mixing zone primary	0.41	<MDL	<MDL	<MDL
mixing zone duplicate	0.38	<MDL	<MDL	<MDL
downstream primary	0.25	<MDL	<MDL	<MDL
downstream duplicate	0.28	<MDL	<MDL	<MDL
DWTP intake primary	0.41	<MDL	<MDL	<MDL
DWTP intake duplicate	0.41	0.099	<MDL	<MDL
treated water primary	<MDL	<MDL	<MDL	<MDL
treated water duplicate	<MDL	<MDL	<MDL	<MDL
Spring 2015	Estrone	17β-Estradiol	Estriol	Ethinyl Estradiol
upstream primary	0.15	<MDL	<MDL	<MDL
upstream duplicate	0.15	<MDL	<MDL	<MDL
effluent primary	36	2.9	3.3	0.29
effluent duplicate	32	2.9	3.0	0.13
mixing zone primary	0.16	<MDL	<MDL	<MDL
mixing zone duplicate	0.16	<MDL	<MDL	<MDL
downstream primary	0.14	<MDL	<MDL	<MDL
downstream duplicate	0.15	<MDL	<MDL	<MDL
DWTP intake primary	0.17	<MDL	<MDL	<MDL
DWTP intake duplicate	0.17	<MDL	<MDL	<MDL
treated water primary	<MDL	<MDL	<MDL	<MDL
treated water duplicate	<MDL	<MDL	<MDL	<MDL
Summer 2015	Estrone	17β-Estradiol	Estriol	Ethinyl Estradiol
upstream primary	0.11	<MDL	<MDL	<MDL
upstream duplicate	0.13	<MDL	<MDL	<MDL
effluent primary	16	<MDL	2.4	<MDL
effluent duplicate	17	<MDL	2.6	<MDL
mixing zone primary	0.19	0.29	<MDL	<MDL
mixing zone duplicate	0.18	<MDL	<MDL	<MDL
downstream primary	0.18	<MDL	<MDL	<MDL
downstream duplicate	0.19	<MDL	<MDL	<MDL
DWTP intake primary	0.21	<MDL	<MDL	<MDL
DWTP intake duplicate	0.18	<MDL	<MDL	<MDL
treated water primary	<MDL	<MDL	<MDL	<MDL
treated water duplicate	<MDL	<MDL	<MDL	<MDL

Overall, biological activity and steroidal estrogens were detected most often, and in the highest concentrations, in the wastewater effluent. The number of positive detections for biological activity per water extract, and relative magnitude of endpoint activation, progressively decreased as water flowed downstream (away from effluent outfall) and through the DWTP.

To initially assess a broad range of signaling pathways with the potential for activation at nuclear receptors and transcription factors a two multi-endpoint bioassays (Trans-FACTORIAL™, Figure 1 and Cis-FACTORIAL™, Figure 2) was used. Significant increases in activation (three standard deviations above field blanks) of the estrogen receptor alpha (ERα), glucocorticoid receptor (GR), peroxisome proliferator-activated receptor-gamma (PPARγ), and pregnane X receptor (PXR) via ligand-receptor binding were detected using the Trans-FACTORIAL™ assay (Figure 1). Significant activation of pathway signaling through the estrogen receptor response element (ERE), pregnane X receptor response element (PXRE), and aryl hydrocarbon response element (AhRE) was detected above field blank values with the Cis-FACTORIAL™ (Figure 2). A majority of the water sample extracts generated responses, indicating either activation of receptor and/or signaling pathways, with the greatest number of activated receptors and/or signaling pathways, as well as greatest relative activation, detected in the wastewater effluent (Figure 1B and Figure 2B).

Figure 1. Trans-FACTORIAL ™ Assay for Nuclear Receptor Binding.

Water sample extracts from Fall 2014 were assessed for the ability to activate nuclear receptors including; the estrogen receptor α (ERα), glucocorticoid receptor (GR), peroxisome proliferator-activated receptor γ (PPARg), and pregnane X receptor (PXR). The area under the curves (in parentheses) and % maximal responses of respective positive control (+Cont) are reported.

Figure 2. Cis-FACTORIAL™ Assay for Nuclear Receptor Activation.

Water sample extracts from Fall 2014 were assessed for ability to activate molecular signaling networks through transcription factors including; the estrogen receptor response element (ERE), pregnane X receptor response element (PXRE) and aryl hydrocarbon receptor response element (AhRE). The area under the curves (in parentheses) and % maximal responses of respective positive control (+Cont) are reported.

Not surprisingly, activation of the pregnane X receptor (PXR), or PXR signaling pathway (PXRE), was the most frequently detected endpoint. PXR is recognized as a general xenobiotic sensor with low specificity involved in regulation of phase I, II, and III metabolic enzymes and has been shown to bind ligands specific to other nuclear receptors (Dagnino et al., 2014). PXR/PXRE activation was detected in 5 of 6 sample extracts (not in treated water) from the Fall 2014. Similar results have been previously reported in a study of nearly 40 sampled surface water sites screened with the Attagene assays (Blackwell et al., 2019). Blackwell et al. (2019) attributed the ubiquitous PXR and PXRE activity to the presence of many anthropogenic ligands within the samples and the inherent promiscuity of the receptor.

Low level PPARγ receptor activity (Trans-FACTORIAL™) also was detected in effluent (Figure 1B), which could be expected given that the PPARγ endpoint was highly active during previous screens of wastewater quality (Escher et al., 2014) and chemical toxicity (Huang et al., 2011). In fact, almost half of the 309 ToxCast phase I chemicals screened previously using the PPARγ Trans-FACTORIAL™ activated the endpoint (Martin et al., 2010). Additionally, low-level activation of the aryl hydrocarbon (AhR) signaling pathway via the AhR response element (AhRE) in effluent and downstream sites was detected (Figure 2B and 2D, respectively). Despite previous reports of high levels of activation using AhR bioassays (Escher et al., 2014; Neale et al., 2015) we detected only 3.7% peak relative activity (Figure 2B).

The activation of endocrine signaling networks detected using the multi-endpoint Attagene assays generally corresponded with the specific and validated transcriptional activation detected using the ER, GR, and AR bioassays. Although assay sensitivity cannot be directly compared using MDLs, because only one positive control concentration was used in the Attagene experimental design, the single endpoint nuclear receptor transcriptional activation bioassays were qualitatively more sensitive for detecting nuclear receptor activation. For example, 16 of 18 samples produced estrogenic activity above the T47D-KBluc bioassay method detection limit (MDL; 0.04 ng/L) (Figure 3), including effluent from the fall 2014 sampling event that activated receptor-ligand binding in the Trans-FACTORIAL™ assay (Figure 2B). Further, all six of the Fall 2014 samples tested positive for estrogenic activity in the T47D-KBluc assay (Figure 3), compared to the Attagene bioassay which resulted in 1/6 and 3/6 fall extract samples producing significant estrogenic activity (compared to positive control) in the Trans (Figure 1B) and Cis (Figure 2A, B and C) assays, respectively.

Figure 3. Estrogenic Activity by Sampling Site.

Estrogenic activity in estradiol equivalents (E2Eqs) was measured in all sampling events and all field blanks (not pictured, no significant activity), using the T47D-KBluc (estrogen receptor) transcriptional activation assay. Estrogen activity is reported as mean +/− standard error (n = 3). Study specific method detection limit (MDL) is also reported.

Estrogenic activity measured in the T47D-KBluc assay (Figure 3), relative to the concurrent estradiol standard, ranged from 0.01 to 29 ng E2Eqs/L (median: 0.16 ng E2Eq/L) and was detected above the assay MDL at the upstream site, effluent, mixing zone, site downstream from the mixing zone and the drinking water intake during every season. Estrogenic activity has been detected in treated wastewaters at concentrations ranging from 0.03–68 ng E2Eqs/L (Leusch et al., 2018); comparable to the 15 ng E2Eqs/L (mean) detected in treated wastewaters of the current study, but greater than the effects-based trigger value (EBT) for adverse ecological effects of 0.5 ng E2Eqs/L (based on European Union environmental quality standards) (Escher et al., 2018). Additionally, the 0.03 ng E2Eqs/L (mean) detected in the treated water extracts from the DWTP was comparable to previous studies using the T47D-KBluc bioassay (0.078 ng E2Eqs/L) (Conley et al., 2017b) and other bioassays (<0.03 ng E2Eqs/L) (Leusch et al., 2018), but well below EBTs for treated drinking water using other in vitro bioassays (3.8 ng E2Eqs/L) (Leusch et al., 2018).

Many CECs were detected in the De Facto Water Reuse extracts using LC-MS/MS (Glassmeyer et al., 2018). Herein only the environmental concentrations of four known environmental steroidal estrogens detected in sample extracts and duplicates were reported including; estrone, 17β-estradiol, estriol, and ethinyl estradiol (Table 2). There were 43 detects for estrogens above the MDL, with 24 at or above the LCMRL. Estrone was the most prevalent compound (detected in every sample except the treated water) with the highest detected concentration (36 ng/L; Table 2) in effluent.

An estimated E2Eq value for each detected compound was calculated by converting environmental concentrations into in vitro bioanalytical equivalent concentrations (E2Eq) using pre-established in vitro relative potency factors from the T47D-KBluc bioassay (Conley et al., 2016). We then compared the sum of the estimated E2Eq values for the four steroidal estrogens in one sampling event to observed E2Eq values from the bioassay. The correlation (± 95% confidence interval) between the estimated and observed E2Eq was determined using linear regression and resulted in an R² = 0.890 (Figure 4A). This high positive correlation indicates that the bulk of the estrogenic activity was caused by the measured steroidal estrogens; similar to a previous nationwide surface water screens using analytical chemistry and bioassays to detect estrogens and estrogenic activity (Blackwell et al., 2019; Bradley et al., 2017; Conley et al., 2017a).

Figure 4. Estimated vs. Observed Cumulative Estrogen Concentrations.

Environmental concentrations of each detected estrogen (estriol, 17β-estradiol, ethinyl estradiol, and estrone) was converted to in vitro bioanalytical equivalent concentrations (E2Eq) using pre-established in vitro relative potency factors from the T47D-KBluc bioassay. The sum of the estimated E2Eq values for each sampling event was compared to the observed E2Eq values from the T47D-KBluc bioassay using linear regression. The ± 95% confidence interval around the best fit line are represented by dotted lines. Method comparison is shown with (A) and without (B) detected estrogen concentrations below the LCMRL.

This method comparison (Figure 4A) includes all detected steroidal estrogen concentrations above the analytical chemistry MDL, including those below the LCMRL. Typically applied to finished drinking water, application of the LCMRL increases both the accuracy and precision among laboratories reporting chemicals under the Unregulated Contaminant Monitoring Regulation for DWTPs (Martin et al., 2007). The method comparison without the estrogen concentrations detected below the LCMRL resulted in fewer data points overall (37% decrease), a slight decrease in R² (0.855), and as expected a wider 95% confidence interval band (Figure 4B).

Although the estrogen concentrations detected below the LCMRL may lack precision or accuracy (compared to those concentrations detected above the LCMRL), the concurrent screening of sample extracts in the bioassay increases our confidence in reporting these qualitative values, indicated by the relatively high R² values of both correlations. Further, incorporating the low-level chemical concentrations supports our conclusions of the presence of ER activating contaminants (positive bioactivity detections above the T47D-KBluc MDL). This demonstration supports the use of the low-level concentrations despite being below the LCMRL, especially in water quality screening that incorporates methods, such as the transcriptional activation bioassays, that can account for cumulative effects of multiple contaminants present at relatively low environmental concentrations.

Glucocorticoid activity, measured using the CV1-hGR bioassay, was also detected above the assay MDL (1.21 ng/L) in the wastewater effluent extracts from all sampling events (Figure 5). Similar to the reported estrogenic activity from the T47D-KBluc bioassay, glucocorticoid activity concentrations detected in the fall (21.1 ng dexamethasone equivalents (DexEqs)/L) and spring samples (21.4 ng DexEqs/L) were an order of magnitude higher than the summer concentration (Figure 5). However, these DexEqs concentrations fall below both the EBTs generated using the GR CALUX bioassay (20 ng DexEqs/L) (van der Oost et al., 2017) and other bioassays (100 ng DexEqs/L) (Leusch et al., 2018). Only the fall samples were evaluated using the Attagene bioassays and therefore resulted in 1/6 (Trans; Figure 1B) and 0/6 (Cis: Figure 2) positive hits for glucocorticoid activity; compared to the 1/6 fall sample extracts with activity above the CV1-hGR assay MDL in the fall extracts (Figure 5).

Figure 5. Glucocorticoid Activity by Sampling Site.

Glucocorticoid activity, in dexamethasone equivalents (DexEqs), was measured in each sampling event and all field blanks (not pictured, no significant activity), using the CV1-hGR (glucocorticoid receptor) transcriptional activation assay. Glucocorticoid activity is reported as mean +/− standard error (n = 3). Study specific method detection limit (MDL) is also reported.

None of the established GR ligands included in the analytical chemical detection methods (betamethasone, fluticasone, hydrocortisone, prednisone, and prednisolone) were detected above method detection limits in samples that produced in vitro glucocorticoid activity (Glassmeyer et al., 2018). To date, all glucocorticoid receptor ligands (or glucocorticoid activity) detected in waste and surface water have been (or have been attributed to) pharmaceuticals (Chang et al., 2007; Jia et al., 2016; Schriks et al., 2010; Weizel et al., 2018). To our knowledge there are no known environmental GR ligands that weren’t designed specifically to target the receptor (i.e. pharmaceuticals). Taken together, this would suggest that additional GR ligand pharmaceuticals, excluded from analytical chemistry methods, were present in the environmental mixture and possibly responsible for the measured GR activity. For example, triamcinolone acetonide was identified as the agonist responsible for glucocorticoid activity detected using a similar GR bioassay while screening samples from the Santa Cruz River in Arizona, USA (Daniels et al., 2018). Alternatively, compounds that have yet to be recognized as environmental ligands for GR were present in the samples extracts.

Despite the absence of significant androgenic activity in either of the Trans- or Cis-FACTORIAL™ assays in the fall samples, androgenic activity determined using the MDA-kb2 bioassay was detected above the MDL (0.77 ng/L) in fall and spring effluent sample extracts (Figure 6). These results, together with a separate assessment of the two AR bioassays, using several (partial) androgens, suggests the AR Trans-FACTORIAL™ bioassay has reduced sensitivity compared to the MDA-kb2 bioassay. For example, the half maximal effective concentrations (EC₅₀) for the AR Trans-FACTORIAL™ were at least an order of magnitude greater than the MDA-kb2 EC₅₀ values for 4-androstene-3,17-dione (LogEC₅₀ = −5.592 and −9.976 M) (Figure S1), cyproterone acetate (LogEC₅₀ = −5.017 and −6.195 M) (Figure S2), mifepristone (LogEC₅₀ = −4.319 and −7.928 M) (Figure S3), progesterone (LogEC₅₀ = −2.735 and −5.274 M) (Figure S4), and spironolactone (LogEC₅₀ = not available and −4.971 M) (Figure S5) (USEPA, 2015; USEPA, 2019).

Figure 6. Androgenic Activity by Sampling Site.

Androgenic activity, measured in dihydrotestosterone equivalents (DHTEqs), was measured in each sampling event and all field blanks (not pictured, no significant activity), using the MDA-kb2 (androgen receptor) transcriptional activation assay. Androgen activity is reported as mean +/− standard error (n = 3). Study specific method detection limit (MDL) is also reported.

Further, extracts from over 35 impacted surface water sites across the United States were analyzed in a previous study for androgenic activity using the MDA-kb2 (Conley et al., 2017a) and Attagene (Blackwell et al., 2019) Trans-FACTORIAL™ bioassays. The higher percentage of positive “hits” for androgen activity permitted a correlation analysis of results from the two bioassays resulting in R² = 3.06 × 10⁻⁵ (Figure S6). However, the range of response (percent of maximum response compared to positive control) was very low in the AR Trans bioassay (<1% overall), likely contributing to the low correlation value. However, similar analyses for estrogenic and glucocorticoid activity were performed and resulted in higher R² values of 0.405 (Figure S7) and 0.431 (Figure S8), respectively.

The two sample extracts that produced activity in the MDA-kb2 bioassay, fall and spring effluent (2.1 and 0.95 ng dihydrotestosterone equivalents/L; DHTEqs/L; Figure 6), were similar to previous reports of androgenic activity in U.S. streams at 1.6–4.7 ng DHTEqs/L (Conley et al., 2017a). Testosterone concentrations were measured above analytical method detection limits (0.044 ng/L) in the Fall 2014 and Spring 2015 effluent sample extracts (0.55 and 0.16 ng/L; mean of sample duplicates). However, a linear regression analysis was not performed to compare the two detection methods due to few detects overall.

Environmental chemicals also may disrupt homeostatic signaling by altering the synthesis of endogenous compounds. For example, some have been shown to alter cytochrome P450 gene expression (Cheshenko et al., 2008). Ketoconazole, an imidazole fungicide, inhibits activity of multiple enzymes in the cytochrome P450 family (Hasselberg et al., 2005; Hegelund et al., 2004), which convert cholesterol into sexual reproductive hormones in vertebrates. This fungicide has been shown to ultimately suppress testosterone production in gonadal fish tissues (Ankley et al., 2007).

To identify potential contaminants that alter steroid hormone signaling during molecular interactions, other than agonist-receptor binding, e.g., steroidogenesis, we measured changes in estrogen and testosterone levels produced by fathead minnow ovary tissue after exposure to water samples from each sampling event. However, none of the sampled waters, and the mixture of contaminants contained therein, significantly affected the production levels of estradiol (Figure 1A) or testosterone (Figure 2B) compared to controls (tissue exposed to media prepared with ultrapure water) after in vitro exposures. While in vitro exposure to individual chemicals has been shown to alter steroid production in fathead minnow ovary tissue (Villeneuve et al., 2007), such effects have rarely been observed when screening ambient water samples (excluding the present study, over 50 samples from across the US have been screened) (Villeneuve, 2019). Consequently, results of the current study support the general conclusion that direct inhibitors of steroid biosynthesis are generally not present in US surface waters at concentrations that demonstrably impair steroid synthesis in vitro.

The strength of this ‘suite’ approach with water quality screening bioassays lies not only in the overlap, or the corroborating results from multiple methods suggesting contaminants may disrupt a single signaling pathway, but in the breadth of endpoints tested. For example, the initial broad high-content screen with the Attagene assays suggested the presence of estrogenic and glucocorticoid-active compounds in the Fall 2014 WWTP effluent. These results were supported by the high biological equivalency values produced in the single endpoint transcriptional activation assays that have greater statistical power. Further, the Attagene assays provided evidence that other singling pathways (PPARγ, Figure 1B and AhR, Figure 2B and D) may also be affected. These patterns of biological activity (positive detections of ER, GR, PPAR, PXR and AhR) are consistent with previous screening studies (Escher et al., 2014).

When considering each bioassay separately, however, limitations become more evident. Although the multi-endpoint assays screen a broad range of endpoints, this type of analysis could benefit from implementing more positive control chemicals (one positive control for each endpoint), running positive controls concurrently in a concentration-response, and increasing replication of sample extract analysis per each endpoint dilution (n > 1). These enhancements would refine the relevance of the magnitude of changes in the multi-endpoint bioassays and permit a more robust statistical analysis of the data. Our data analysis methods began with filtering out any Attagene endpoint that did not have a well-established positive control incorporated in the experimental design. This eliminated 47/52 signaling pathway and 15/29 nuclear receptor endpoints. Additionally, the cost associated with the patented high-throughput screens excluded the incorporation of replication into the experimental design and therefore reduced the overall statistical power.

Although the ER, GR, and AR transcriptional activation bioassays assess a single endpoint, the experimental design allows for increased sensitivity and higher statistical power making these three bioassays a sensitive and a less expensive option compared to their analytical chemistry and high-content counterparts. Ultimately, it is the complementarity offered by the implementation of the in vitro assay suite approach that increases confidence in reporting the potential for disruption specifically in the ER and GR endocrine signaling pathway in the WWTP effluent samples.

Conclusions

Of the designated sampling sites along a specific surface water flow path, the extracts analyzed from the WWTP effluent, as expected, resulted in the greatest number of activated endpoints, the highest relative biological activities and the highest environmental estrogen concentrations. Each measure of biological activity or chemical concentration was attenuated in the closest downstream site (mixing zone) compared to the effluent, likely due to natural processes (e.g. dilution, sorption, photo and biodegradation). In general, these measures of biological activity and chemical concentration continued to progressively decrease as the water flowed farther downstream to the drinking water intake. The additional attenuation observed between the DWTP intake and treated water can be contributed to the engineered water treatment processes of the DWTP.

Results from the quantifiable, sensitive and statistically powerful single-endpoint bioassays were corroborated using the Attagene high content assays; together these approaches confirm the presence of contaminants that activate estrogenic and glucocorticoid signaling pathways. While targeted analytical chemical analysis results correlated well with the bioassay results for estrogenicity (R² = 0.890), a comparable analysis could not be made to help identify the contaminants responsible for glucocorticoid or androgenic activity.

Research efforts worldwide have resulted in new effects-based screening methods, like the ones described herein, as well as optimized analysis methods for the in vitro screening tools. The results of this De Facto Water Reuse study may not explicitly generate a novel tool, but resolutely supports the ever-increasing trend of applying more than one bioassay, in an in vitro assay suite approach, integrated with targeted analytical chemistry approaches to determine the potential presence of biologically active contaminants in wastewater and drinking water.

Supplementary Material

S2. Figure S2 Androgenicity Bioassay Comparison: Cyproterone acetate.

Concentration response curves for cyproterone acetate in the A) Attagene Trans-FACTORAL™ and B) MDA-kb2 bioassays. Percent of maximum response values represent mean +/− standard error.

S3. Figure S3 Androgenicity Bioassay Comparison: Mifepristone.

Concentration response curves for mifepristone in the A) Attagene Trans-FACTORIAL™ and B) MDA-kb2 bioassays. Percent of maximum response values represent mean +/− standard error.

S1. Figure S1 Androgenicity Bioassay Comparison: 4-Androstene-3,17-dione.

Concentration response curves for 4-androstene-3,17-dione in the Attagene Trans-FACTORIAL™, MDA-kb2, and CV1-hAR bioassays. Percent of maximum response values represent mean +/− standard error. (Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.)

S6. Figure S6 Androgenicity Bioassay Detection Correlation: MDA-kb2 and Attagene Trans-FACTORIAL™.

Responses from MDA-kb2 bioassay (ng DHTEqs/L) were compared to responses from DHT exposure in the Attagene Trans-FACTORIAL™ for the androgen receptor (percent maximum of response of AR to DHT) using linear regression.

S5. Figure S5 Androgenicity Bioassay Comparison: Spironolactone.

Concentration response curves for spironolactone in the A) Attagene Trans-FACTORAL™, B) MDA-kb2 and C) CV1-hAR bioassays. Percent of maximum response values represent mean +/− standard error.

S4. Figure S4 Androgenicity Bioassay Comparison: Progesterone.

Concentration response curves for progesterone in the A) Attagene Trans-FACTORIAL™ and B) MDA-kb2 bioassays. Percent of maximum response values represent mean +/− standard error.

S8. Figure S8 Glucocorticoid Activity Bioassay Detection Correlation: CV1-hGR and Attagene GR Trans -FACTORIAL™.

Responses from CV1-hGR bioassay (ng DexEqs/L) were compared to responses from dexamethasone exposure in the Attagene Trans-FACTORIAL™ for the GR (percent maximum of response of GR to dex) using linear regression.

S7. Figure S7 Estrogenicity Bioassay Detection Correlation: Attagene Cis-FACTORIAL™, ERα Trans-FACTORIAL™, BLYES and T47D-KBluc.

Responses from ER response element (ERE) Attagene Cis-FACTORIAL™, ERα Trans-FACTORIAL™, and the BLYES (percent of maximum response to 17β-estradiol) were compared to responses in the T47D-KBluc bioassay (ng E2Eq/L) using non-linear correlation analysis.

Figure 7. In vitro Steroidogenesis.

Female fathead minnow ovary tissue subsamples were exposed to site water from each sampling event to assess impacts of steroidogenesis, i.e. changes in A) estradiol and B) testosterone production. Fold changes relative to hormone production in concurrent media only controls are reported as mean (n = 10) +/− standard error.

Highlights.

1. A suite of in vitro assays was applied to detect biological activity

2. Estrogenic and glucocorticoid activity in effluent was corroborated by multiple bioassays

3. Estrogenic activity verifies estrogen concentrations below chemical detection limits

4. Most biological activity was below method detection limits in unfinished treated drinking water