A range of potentially toxic substances are discharged from municipal wastewater treatment plants (WWTPs), including pharmaceuticals and personal care products (Metcalfe et al. 2003; Carballa et al. 2004; Ratola et al. 2012), illicit drugs (Rodayan et al. 2015), estrogens (Servos et al. 2005; Ternes 2006; Liu et al. 2015), pesticides (Kahle et al. 2008) and toxic metals (Marcogliese et al. 2015). For many organic contaminants, there are also transformation products of the parent compounds that are present in wastewater. Effective wastewater treatment can reduce the discharge of various substances from municipal effluents, but removals may be influenced by the type of treatment and the operating conditions, seasonal changes in temperature and other factors (Eggen et al. 2014; Baalbaki et al. 2017).

Due to advances in analytical methods, it is now possible to monitor thousands of chemical contaminants in water and wastewater (Richardson and Ternes 2018). Regulating wastewater discharges based on the analysis of a wide range of target compounds is an expensive and time-consuming approach (Jia et al. 2015). In addition, limiting contaminant monitoring to a number of known pollutants may underestimate risks to the environment and hazards to human health (Smital et al. 2013) and does not account for mixture effects (Snyder and Leusch 2018). A complementary approach is to use bioanalytical tools to screen for different modes of toxicity in samples of water and wastewater. These screening assays can be combined with subsequent analytical approaches to assess the effectiveness of wastewater treatment (Escher and Leusch 2012; Jia et al. 2015).

The objectives of the present study were to compare treatment performance using bioassays in completement to chemical analyses and to evaluate the quality of the water in the receiving streams. Extracts from samples of treated and untreated wastewater, and surface water were analyzed for the concentrations of 16 target analytes (i.e., indicator compounds), which included specific compounds selected from the classes of estrogens, androgens, antibiotics and pharmaceuticals, pesticides, artificial sweeteners and personal care products. Results of the bioassays were analyzed using two approaches described in Neale et al. (2020): comparison between influent and effluent for the evaluation of treatment process efficiency and comparison to effect-based trigger values for the evaluation of treated water quality. The extracts were tested for toxicity using four mammalian cell-based in vitro assays. In the case of in vitro toxicity testing, extracts from water and wastewater were initially screened using the Qiagen Nuclear Receptors 10-Pathway Reporter Array to provide information regarding the regulation of multiple nuclear receptors. The estrogenic activity of the extracts was subsequently tested with the ERα CALUX assay. The capacity of extracts to induce oxidative stress was tested with a Nrf2 Luciferase Luminescence Assay. Finally, the MTS assay was employed to evaluate non-specific cytotoxicity to ensure that responses observed with the reporter gene-based assays were not influenced by decreased cell viability. In tandem with chemical analyses, the in vitro toxicity data were used to evaluate whether toxic substances are removed effectively from wastewater and to evaluate the sensitivity of the assays to detect changes in the levels of specific classes of chemical contaminants in water and wastewater.

Materials and Methods

Indicator Compounds

The 16 indicator compounds analyzed in extracts of water and wastewater are listed in Table 1, along with information on the class of the contaminant, the supplier and the stable isotopically labeled surrogate used as an internal standard for quantification. The compounds were chosen to represent various classes of compounds and selected based on their prevalence in WWTP effluents, as observed in earlier studies in the area, and their presence on the list of priority substances (European Commission 2012). An external standard stock solution (1000 ppm) for these indicator compounds was prepared in methanol (Fisher Optima, LC/MS grade) and stored in the dark at 4 °C for preparation of fresh analytical standards.

Table 1 List of indicator compounds, their compound class and respective labeled surrogate, with the suppliers identified in parentheses
Full size table

Sample Collection

In 2014, samples of wastewater were collected at two WWTPs in the West Central Region of southern Ontario (WWTP 1 and WWTP 2). Both treatment plants employ conventional activated sludge wastewater treatment, but WWTP 2 also has a tertiary treatment train. At these WWTPs, 24-h composite samples of untreated wastewater (i.e., influent) and treated wastewater (i.e., effluent) were collected once a month for 5 months from April to August. A wastewater lagoon located in the Central Region of southern Ontario (WL) was also sampled in the spring and fall of 2014 during periods of intermittent discharge. In this case, grab samples of influent and effluent were collected in May, June and twice in September. Grab samples of surface water were also collected from a river approximately 2.0 km downstream of the lagoon during discharge. Information regarding the WWTPs and WL is summarized in Table S1 in Supplementary Information. A total of 7L of water was collected for each sample, 6L in 1-L HDPE bottles and 1L in 1-Lamber glass bottle for the ER-CALUX assay. Samples were frozen upon collection and sent to the lab for preparation and analysis the next day.

The Laboratory Services Branch of the Ontario Ministry of the Environment, Conservation and Parks (Etobicoke, ON, Canada) analyzed all samples of water and wastewater for a wide range of water quality parameters. Samples were analyzed for a range of cations and metals, chemical oxygen demand (COD), carbonaceous biochemical oxygen demand (cBOD), pH, nitrogen species, phosphorous species, total and dissolved organic carbon, suspended and dissolved solids, etc. Samples were analyzed according to standard protocols developed by the Ministry.

Sample Extraction

Subsamples of water and wastewater were extracted using two solid-phase extraction (SPE) methods, including extraction with Oasis® MCX cartridges (6 mL, 150 mg) to concentrate basic/neutral compounds and extraction with MAX cartridges (6 mL, 400 mg) to concentrate acidic compounds. The SPE cartridges were purchased from Waters (Milford, MA, USA). Prior to extraction, samples (110 mL) were filtered through 1 μm glass fiber filters (Fisher Scientific, Ottawa, ON, Canada) and the pH was adjusted to pH = 2.5 for extraction using MCX cartridges and to pH = 8.0 for extraction using MAX cartridges. Subsamples prepared for analysis of contaminants (n = 3) were spiked with a standard solution of labeled surrogates to account for recovery during the extraction, possible degradation during storage and matrix effects. Subsamples prepared for bioassays (n = 3) were not spiked. The methods for SPE extraction were previously described by Baalbaki et al. (2017). Following extraction with both types of cartridges, elution solvents were evaporated to near dryness and samples were reconstituted in either 0.4 mL of a 1:1 methanol–water solution (Fisher Optima HPLC grade) for chemical analyses or in 0.4 mL of DMSO (Sigma-Aldrich Bioreagent, molecular biology grade) for bioassays. The MCX and MAX extracts were analyzed separately for contaminants, as described below. Prior to conducting the in vitro bioassays, the MCX and MAX extracts were pooled for each sample. Extracts were placed into amber high performance liquid chromatography (HPLC) vials with polytetrafluoroethylene (PTFE) tops and stored at -20 °C until analysis. The overall pre-concentration factor was 275 × (i.e., 0.11 L of sample concentrated to 400 µL of extract). Chemical analysis and bioassays were run on triplicate samples.

To evaluate whether there was contamination from the extraction process and to allow for the calculation of SPE recoveries, procedural blanks were prepared by extracting 110 mL of Milli-Q water. Prior to extraction, the procedural blank was spiked with the internal standard solution of labeled surrogates and the pH was adjusted for extraction using MCX and MAX cartridges, as described above.

Chemical Analysis

The concentrations of the 16 indicator compounds in procedural blanks and in extracts from water and wastewater were quantified by liquid chromatography with high resolution mass spectrometry (LC–HRMS) or by liquid chromatography with tandem mass spectrometry (LC–MS/MS). An Accela HPLC coupled to an LTQ Orbitrap XL hybrid linear ion trap–orbital trap instrument purchased from Thermo Fisher Scientific (Waltham, MA, USA) was used for LC-HRMS analysis. Analysis by LC–MS/MS was conducted with an AB Sciex QTrap 5500 instrument equipped with an Agilent 1100 series HPLC purchased from Applied Biosystems-Sciex (Mississauga, ON, Canada) and operated in multiple reaction monitoring mode (MRM).

For LC–MS/MS, estrone, 17ß-estradiol (E2) and 17α-ethinylestradiol (EE2) were separated by liquid chromatography using a Genesis C18 column (150 mm × 2.1 mm ID; 4 mm particle size) purchased from Chromatographic Specialties (Brockville, ON, Canada), coupled with a guard column with the same packing material (4 mm × 2.0 mm) purchased from Phenomenex (Torrance, CA, USA). The solvents used for chromatographic separations were: [A] 10 mM ammonium acetate with 0.1% acetic acid and [B] 100% acetonitrile, using the gradient previously described by Metcalfe et al. (2014).

For LC-HRMS, chromatographic separation of the indicator compounds was accomplished with a Thermo Scientific Hypersil Gold C18 column (100 × 2.1 mm, 1.9 µm particle size) heated to 30 °C which was connected to an in-line UHPLC filter (0.2 µm). The chromatography solvents were [A] 0.1% acetic acid in water and [B] 100% acetonitrile. The flow rate in the column was 200 µL/min.

Three different analytical methods were used for the various indicator compounds: Method 1: LC-HRMS with electrospray ionization (ESI) and positive ion monitoring; Method 2: LC-HRMS with ESI and negative ion monitoring; and Method 3: LC–MS/MS with ESI and positive ion monitoring in MRM. Table S2 in Supplementary Information describes which of these three analytical methods was employed for each of the indicator compounds. For LC-HRMS, a 6-point calibration curve covering the range of anticipated analyte concentrations was used for external calibration. The concentrations were adjusted according to the recovery of the internal standards in order to compensate for sample matrix effects on ionization and for variations in the recoveries during sample extraction. The recoveries for the 16 indicator compounds were first evaluated using Milli-Q water. Of the 16 indicator compounds analyzed, 12 were found to have mean recoveries > 80%, including the estrogens, estrone (113 ± 47%), E2 (113 ± 48%) and EE2 (121 ± 47%). There were lower recoveries for DEET and sulfamethoxazole of 57 ± 30% and 57 ± 28%, respectively. A high recovery of 604 ± 426% for acesulfame K indicates that there was signal enhancement for this compound, probably because a constituent of the sample matrix increased the ionization efficiency. Responses to the internal standards spiked into all samples of water and wastewater were used for  the quantitation of each compound.

The limits of detection (LODs) and limits of quantification (LOQs) were determined by analyzing serial dilutions of a 1:1 methanol–water standard solution containing all compounds. LOD and LOQ are defined as the indicator compound concentration producing a peak with a signal-to-noise ratio of 3 and 10, respectively. The LODs and LOQs for all analytes are listed in Table S3 provided in the Supplementary Information. Two samples spiked with the indicator compounds at concentrations of 12 and 30 µg L−1 were extracted and analyzed for quality control purposes. For these spiked samples, the relative error between expected and measured quality control concentrations was ˂20% for all indicator compounds.

Nuclear Receptors 10-Pathway Reporter Array

The Nuclear Receptors 10-Pathway Reporter Array purchased from Qiagen (Germantown, MD, USA), hereafter referred to as the 10-Pathway Reporter Array, was used as a screening assay to determine which cellular signaling pathways are affected or regulated by exposure to sample extracts. MCF7 breast cancer cells were selected to carry out this assay. This cell line expresses a majority of the transcription factors monitored by the Qiagen 10-Pathway Reporter Array. Kittler et al. (2013) systematically “mapped the genomic binding sites of all nuclear receptors expressed in MCF-7 breast cancer cells.” In total, they “mapped the genomic binding sites of a total of 33 proteins whose corresponding genes are expressed at moderate to high levels in MCF-7 cells.” Among these proteins, 9 of the 10 transcription factors whose activity can be monitored by the Qiagen reporter array, were identified. The only transcription factor that will likely not be encountered in the MCF7 cells is HNF4 (hepatocyte nuclear factor 4). However, Escher et al. (2014) found that HNF4 (hepatocyte nuclear factor 4) is not a major player in the MOA in any of the toxicity pathways they investigated. All cell lines employed in this study were cultured at 37 °C, 5% CO2 and 95% humidity. All cell culture materials were manufactured by Life Technologies (Rockville, MD, USA) and supplied by Thermo Fisher. The MCF7 cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1 × GlutaMAX supplement, 100 units mL−1 penicillin and 100 µg mL−1 streptomycin. To carry out this assay, cells were transfected with Qiagen Reporter Array DNA. This was achieved employing the Attractene Transfection Reagent (Qiagen) as described by the manufacturer. Prior to transfection, 75–95% confluent MCF7 cells growing in T75 plates were trypsinized with 0.25% trypsin–EDTA and plated onto white, opaque 96-well plates at a density of 4 × 104 cells mL−1 (i.e., 100 μL per well) in Opti-MEM® I Reduced Serum Media (Invitrogen) supplemented with 5% FBS and 1% NEAAs. Cells were plated onto transparent polypropylene 96-well plates. Following 24-h growth, media was removed from the cells and the cells were rinsed with 100 μL of phosphate buffered saline (PBS) prior to the addition of 88 μL (per well) fresh antibiotic and Opti-MEM® I Reduced Serum Media (FBS-free). At this point, cells were exposed to the transfection materials.

Transfection materials were prepared in 15-mL polypropylene centrifuge tubes. The total volume of the transfection materials prepared was determined by the number of wells on the 96-well plate to be transfected. Briefly, into each centrifuge tube, 0.08 μg per well (i.e., 0.8 μL per well) of Reporter Array DNA was added to 12 μL per well of Opti-MEM® I Reduced Serum Media. Next, 0.4 µL per well of Attractene Transfection Reagent was added to the DNA and the mixture was vortexed. The DNA complexes were then incubated at room temperature for 15 min, before being added to the MCF7 cells in the 96-well plates. Cells were then incubated under normal growth conditions for 48-h. Following incubation with the transfection reagents, the Dual-Luciferase Reporter Assay System (Promega) was utilized to monitor luminescence. Firefly luminescence and Renilla luminescence were quantified on the luminometer (Berthold Technologies Mitras LB 940 Multimode Microplate Reader) as described by the assay supplier. Positive and negative control transfections were also prepared and included on each plate.

Responses for the various reporters in the array were quantified in terms of Induction Ratios (IR), which in luminescence-based assays is defined as the ratio of the relative light units (RLU) measured for an experimental sample relative to a suitable control sample, as described in Eq. 1 (Jia et al. 2015). For the assay used, an IR value > 5 is recommended as evidence of a significant response. This value was later compared to Escher et al. (2014) who reported an IR of 1.5 as evidence of a positive biological response in other in vitro assays and further suggest that the ratio was sufficient to ensure that a significant effect was observed.

$${\text{IR}} = \frac{{{\text{RLU}}_{{{\text{sample}}}} }}{{{\text{RLU}}_{{{\text{control}}}} }}$$
(1)

ERα CALUX Assay

The ERα Chemically Activated LUciferase eXpression® assay, hereafter referred to as the ERα CALUX assay was employed to determine the estrogenic activity of sample extracts. This reporter gene assay is based on the ERα U2OS.Luc cell line, which can be stably transfected with ERα or ERβ and a luciferase reporter gene and was obtained under license from Biodetection Systems B.V. (Amsterdam, the Netherlands). The recombinant construct for this reporter gene assay and the basis for the response in the assay has been previously described (Quaedackers et al. 2001; Sonneveld et al. 2005; Wang et al. 2014).

The ERα U2OS.Luc cells were cultured using the methods recommended by the supplier, with minor modifications. All cell lines employed in the present study were cultured at 37 °C, 5% CO2 and 95% humidity. All cell culture materials were from Life Technologies. Briefly, the cells were cultured in T75 flasks containing 22 mL of DMEM/F12 media supplemented with 10% Fetal Bovine Serum (FBS), 100 units mL−1 of penicillin and 100 µg mL−1 of streptomycin, 0.8 mg mL−1 geneticin and 1 × nonessential amino acids (NEAAs). All assays were carried out in 96-well plates containing assay medium without phenol red and supplemented with 10% USDA-approved charcoal stripped FBS, 100 units mL−1 penicillin and 100 µg mL−1 streptomycin and 1 × NEAAs. No geneticin was added to the assay media. Phenol red-free media was employed because of the ER-agonist activity of phenol red (Berthois et al. 1986). Furthermore, charcoal stripped FBS was employed to minimize exposure to any estrogens present in the culture media.

Once plated, 75–95% confluent cells were trypsinized with phenol red-free, 0.05% trypsin–EDTA. Next, 100 µL of cells at a 100,000 cells mL−1 concentration was plated onto 96-well white opaque plates (Nunc, Thermo Scientific). The Bright-Glo™ Luciferase Assay System (Promega) was employed to quantitate firefly (Photinus pyralis) luciferase expression following 24-h exposure to sample extracts. A E2 standard curve at concentrations ranging from 1.0 × 10–13 to 1.0 × 10–10 M was included on each opaque plate, allowing for quantitation of estrogenic activity in terms of ng L−1 of E2 equivalents. All luminescence measurements were made on a Berthold Technologies Mitras LB 940 Multimode Microplate Reader.

Nrf2 Luciferase Luminescence Assay

An assay with the Nrf2 luciferase reporter MCF7 stable cell line, hereafter referred to as the Nrf2 assay, was employed to determine whether exposure to sample extracts resulted in an oxidative stress response. Nrf2 cells were obtained from Signosis Inc. (Santa Clara, CA, USA) and were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS, 1 × GlutaMAX, 0.8 mg mL−1 geneticin (G418 sulfate), 100 units mL−1 penicillin and 100 µg mL−1 streptomycin. All media components were from Life Technologies. Once 75–95% confluent in T75 flasks, cells were trypsinized with 0.25% trypsin–EDTA (Life Technologies) and plated onto white, opaque 96-well plates at a density of 5 × 104 cells mL−1 (i.e., 100 μL of cells added per well) for luminescence analyses. Cells were allowed to grow for 24-h prior to the addition of sample extracts. Following 24-h exposure to the extracts, the Bright-Glo™ Luciferase Assay System (Promega) was employed to quantitate firefly luciferase expression. Once more, luminescence measurements were made on a Berthold Technologies Mitras LB 940 Multimode Microplate Reader.

For quantitation of the oxidative stress response, a dilution series of tert-butylhydroquinone (t-BHQ, Sigma-Aldrich) at concentrations ranging from 1 × 10–5 to 1.6 × 10–7 M was included on each opaque plate. Oxidative stress was expressed in terms t-BHQ equivalents.

MTS Assay

The cytotoxicity of the sample extracts to all mammalian cell lines employed in the present study was assessed with the CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS) manufactured by Promega (Madison, WI, USA). In this colorimetric assay, the MTS reagent (3-[4,5-dimethylthiazol-2-yl]-5-[3-carboxy methoxyphenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt) is bio-reduced to a soluble formazan product when it is exposed to actively respiring cells. For every 96-well plate of cells tested, 2.5 mL of MTS reagent was diluted in 12.5 mL of Opti-MEM® I Reduced Serum Media. Following 24-h exposure of the cells plated on transparent polypropylene 96-well plates (Greiner) to the sample extracts, the media was removed, and the cells were rinsed twice with sterile PBS. Next, 120 µL of the MTS reagent diluted in Opti-MEM® media was added to the cells and then allowed to incubate for 2-h. Formazan product formation was detected with a Biorad (Hercules, CA, USA) UV–visible microplate reader (Benchmark Plus with Microplate Manager 5.2.1 software) by measuring the optical density at λ = 490 nm.

Quality Control and Statistical Analysis

For all aforementioned chemical analyses and assays, the significance of results obtained was ensured using controls. Field blanks were prepared for each batch of samples collected, and laboratory blanks were run with each batch of samples processed.

The statistical analysis was performed using Prism. For the comparison between responses, the Mann–Whitney test (p < 0.05) was used. To determine whether estrogenicity or oxidative stress were significantly decreased during treatment, the Wilcoxon matched-pairs signed rank test was used (p < 0.05).

Results and Discussion

Wastewater Quality Results

A wide range of wastewater quality parameters were monitored in the influent and effluent of WWTP 1, WWTP 2 and WL. As illustrated with a selected number of these parameters, there was a marked improvement in water quality as a result of wastewater treatment (Fig. 1). WWTP2 was particularly effective at treating the wastewater, possibly because of the tertiary treatment system at this plant. The wastewater lagoon was also effective at improving wastewater quality before discharge into receiving waters (Fig. 1).

Fig. 1
figure 1

Concentrations (mg L−1, n = 1) of total suspended solids (TSS), carbonaceous biological oxygen demand (cBOD), dissolved organic carbon (DOC), total ammonia nitrogen (TAN) and total phosphorus (TP) in influent (INF) and effluent (EFF) samples collected from WWTP 1, WWTP 2 and WL during the sampling period in 2014. NA = Not analyzed

Full size image

Concentrations of Indicator Compounds

The analytical data for wastewater samples collected at WWTPs 1 and 2 are summarized in Table 2, and the data for WL and receiving waters are summarized in Table 3. The estrogens, E2 and EE2, the pharmaceutical, gemfibrozil and the herbicides, atrazine, bentazon and MCPA were not detected in any of the samples. Acesulfame K and carbamazepine were detected in all influent and effluent samples from WWTP 1 and were widely detected in samples collected from WWTP 2 (Table 2) and WL (Table 3). Carbamazepine is known to be poorly removed in WWTPs (Blair et al. 2013), and the artificial sweetener, acesulfame K, is also poorly removed by wastewater treatment (Subedi and Kannan 2014). The antibiotics, sulfamethoxazole and trimethoprim were widely detected in both influent and effluent samples from WWTP 1 and WWTP 2 (Table 2), but these compounds were detected less frequently in grab samples from WL (Table 3). The nonprescription analgesic, ibuprofen, and the antibacterial compound, triclosan, were frequently detected in influent samples collected in WWTP1, but not in effluent samples, which is consistent with the high removals usually reported for these compounds in WWTPs (Blair et al. 2013).

Table 2 Mean concentrations (ng L−1; ± %SD, n = 3) of contaminants in the influent and effluent wastewater sampled at WWTP1 and WWTP2 in 2014
Full size table
Table 3 Mean concentrations (ng L−1; ± %SD, n = 3) of contaminants in influent (INF) and effluent (EFFL) wastewater sampled in 2014 at the WL. Also included are the mean concentrations (ng L−1; ± %SD, n = 3) of contaminants in the river water (SURF WATER) 2.0 km downstream of the lagoon discharge
Full size table

It is notable that androstenedione, which is an intermediate in the biosynthesis of testosterone, was detected at concentrations > 100 ng L−1 in several influent samples from the WWTPs (Table 2). There are only limited monitoring data for androstenedione in the literature. However, Baalbaki et al. (2017) detected androstenedione in influent samples, but not in effluent samples collected from WWTPs, indicating that this compound is effectively removed by wastewater treatment. Estrone was detected in influent samples collected in June and August from WWTP 1 (Table 2) and from influent samples collected in May and September in WL (Table 3). It was present at concentrations < LOQ in influent and effluent samples collected from WWTP 2 in June and August (Table 2).

For DEET, the active ingredient in some insect repellents, concentrations in the influents were an order of magnitude higher in the months of June and July at WWTP 1 and WWTP 2, presumably because of the higher numbers of biting insects in the summer. However, at WL, DEET levels peaked in May (Table 3). Overall, DEET appears to be partially removed by wastewater treatment, with generally lower levels detected in effluent samples. The herbicide 2,4-D was detected in wastewater samples collected in June and July at WWTP 1 and occasionally detected in samples collected from WWTP 2 (Table 2) and WL (Table 3). We assume that storm water overflow is a possible source this herbicide in domestic wastewater as it is generally not used in households. As 2,4-D has been banned since 2009 in Ontario for cosmetic weed control, it is difficult to speculate on the sources of this herbicide. Acesulfame K, DEET and ibuprofen were frequently detected in the surface water samples collected from a river 2.0 km downstream of the discharge from the wastewater lagoon (Table 3). The river sub-watershed is approximately 31% agricultural, 17% urban, 3% roads, 3% golf courses and 3% industrial, with the remainder being natural heritage features.

The present study does not include calculations of contaminant removal as caution should be exercised in interpreting the relative concentrations of the target compounds in influent and effluent samples as an indicator of contaminant removal by wastewater treatment, particularly when the sampling strategy was not designed for such analysis. The hydraulic retention times for wastewater of 1–3 days in WWTPs and even longer in some wastewater lagoons means that influent and effluent samples collected on the same day are not synchronized in terms of the composition of the wastewater (Ort et al. 2010). This is especially problematic when interpreting the analytical results from the grab samples and samples of influent and effluent collected simultaneously, as shown in our earlier work (Baalbaki et al. 2017).

Results of MTS Assay

In working with human cell lines, the endocrine endpoint generally exhibits greater sensitivity than endpoints of toxicity (Kolkman et al. 2013). Nonetheless, it is important to evaluate whether a reduced response in in vitro assays is caused by cytotoxicity (i.e., decreased cell viability). The MTS assay indicated that cell viability was not affected by exposure to the sample extracts at all dilutions examined (data not presented). Note that all cell lines employed (i.e., MCF7, Qiagen transfected MCF7, U2OS, Nrf2 cells) were tested for cell viability using the MTS assay. Thus, any change in response using these cell lines can be attributed to pathway-specific impacts.

Results of 10-Pathway Reporter Array

The results of the 10-Pathway Reporter Array were used to select which cellular responses should be investigated more closely. Table 4 summarizes the mean IR values (n = 3) for the various receptors following exposure to selected wastewater extracts. As mentioned in the materials and methods section, an IR > 5 is considered as evidence of a significant response for this assay and all controls were run as part the procedure provided by the manufacturer. All sample extracts induced upregulation of the estrogen receptor and liver X receptor (Table 4). IRs of 12 to 47 were recorded for the estrogen receptor, while IRs ranged from 10 to 45 for the liver X receptor. There was also upregulation of the vitamin D and retinoid X receptors in some treatments, with IRs ranging from 2 to 16 and from 3 to 19, respectively (Table 4). For the progesterone receptor, the IR values were ≥ 5 in treatments with only four influent samples, namely WWTP1 in August 2014, WWTP2 in June 2014 and August 2014, and WL in September 2014. Upregulation of the peroxisome proliferator activation (PPAR) receptor was only observed in treatments with the June 2014 and September 2014 influent samples from WWTP1 and WL, respectively. No upregulation of the glucocorticoid and retinoic acid receptors was observed (all IR < 5, Table 4). There was also no upregulation of the androgen receptor or the retinoic acid receptor, with maximum IR values of 3 and 4, respectively (Table 4). The lack of a significant response for upregulation of the androgen receptor is worth noting considering that androstenedione was detected in several influent samples.

Table 4 Mean (± SD, n = 3) induction ratios for upregulation of cellular receptors in the 10-Pathway Reporter Array in treatments with extracts prepared from wastewater collected from WWTP 1, WWTP 2 and WL, and surface water downstream of WL
Full size table

While there was some evidence that upregulation was more frequently reported in the influent (in 50% of the responses) than in the effluent extracts (25% of the responses), there is no indication that the assay is quantitative and allows the comparison of the intensity of upregulation between samples.

In a study of in vitro bioassays to assess wastewater treatment, Escher et al. (2014) observed that 5 out of 25 nuclear receptors were activated when exposed to effluent extracts; including the pregnane X, PPARγ, liver X and glucocorticoid receptors. Based on the results from the 10-Pathway Reporter Array described in the current study, it is apparent that the estrogen and liver X receptors showed the greatest upregulation in treatments with wastewater extracts. Note that upregulation of these pathways occurred in treatments with samples of influent, effluent and surface water. Significant upregulation was also observed for the retinoid X receptor.

Upregulation of liver X and retinoid X pathways is of particular interest as the two receptors form heterodimers that can then regulate genes associated with a range of cellular processes, such as lipid metabolism and inflammation (Gage et al. 2016). The regulation of PPAR receptors by wastewater extracts is also of interest as these receptors are targeted by cholesterol-regulating drugs (Roberts et al. 2015), including the gemfibrozil drug selected for analysis in this study. Metcalfe et al. (2013) detected PPAR-agonists in extracts prepared from wastewater using an in vitro assay, but these responses were not correlated with the concentrations of cholesterol-reducing drugs targeted for analysis. A variety of other compounds that can be present in wastewaters have the capacity to bind with PPARs, including anti-inflammatory drugs (Gijsbers et al. 2011), and phthalates, perfluorinated compounds and bisphenol-based compounds (Desvergne et al. 2009; Riu et al. 2011; Chamorro-Garcia et al. 2012). Synthetic glucocorticoids such as prednisone and hydrocortisone are drugs that are widely prescribed for suppression of inflammation. Synthetic progestins are the active ingredients for hormone therapies (e.g., for endometrial hyperplasia) and in many birth-control formulations. Future monitoring of wastewater using analytical techniques could include analysis for glucocorticoid and progesterone agonists used for therapy (Schriks et al. 2010; Wu et al. 2019).

Results of ERα CALUX Assay

The activation of the estrogen receptor observed in all samples tested with the 10-Pathway Reporter Array highlighted the need for additional tests of estrogenicity using the ERα CALUX assay. The results from the ERα CALUX assay in the present study, expressed as ng L−1E2 equivalents, are reported in Table 5. When combining all three sites, the mean estrogenic response to extracts from the effluents were statistically lower than the estrogenic response to extracts from the influents (Wilcoxon matched-pairs signed rank test, P value = 0.0001). The mean estrogenic response to extracts from the WWTP 1 influent ranged from 27 to 72 ng L−1 E2 equivalents, while the mean estrogenic response to extracts from WWTP 1 effluent ranged from 1 to 10 ng L−1 E2 equivalents (Table 5). The statistical analysis indicated significantly lower estrogenic potency of effluent relative to the influent extracts for WWTP1 (Wilcoxon matched-pairs signed rank test, P value = 0.0312) (Table 5). For samples collected from WWTP2, the mean estrogenic response to influent samples ranged from 34 to 59 ng L−1 E2 equivalents, while the mean estrogenic potency of effluent samples was statistically lowered and ranged from 2 to 14 ng L−1 E2 equivalents (Wilcoxon matched-pairs signed rank test, P value = 0.0312) (Table 5).

Table 5 Mean (± %SD, n = 3) responses in the ERα CALUX assay (ng L−1 E2 equivalents) and NFR2 assay (mg L−1 tBHQ equivalents) in treatments with extracts prepared from wastewater collected from WWTP 1 and WWTP 2
Full size table

No significant difference in estrogenic activity was observed between effluent samples from WWTP1 and WWTP2 (Mann–Whitney test, P value of 0.4206). Previously, it was suggested that nitrification may enhance the degradation of steroid estrogens (Servos et al. 2005; Khanal et al. 2006), suggesting that the effluent of WWTP2, with a treatment train including a nitrification step, could have further reduced the residual estrogenic activity of the water. The absence of a difference might be explained by the mean concentration of nitrate plus nitrite in the effluent of WWTP 2 (i.e., 16.1 mg L−1) over the monitoring period that was similar to the mean concentration in the effluent of WWTP 1 (i.e., 14.8 mg L−1). It is difficult to speculate on the extent of nitrification that occurred during the sampling period and to evaluate whether nitrification was an important parameter for reducing estrogenic activity. The treatment train in WWTP 2 also includes tertiary treatment by filtration, whereas there is only secondary treatment at WWTP 1, but the additional treatment step in WWTP 2 also did not seem to enhance the reduction in estrogenicity.

For grab samples collected from the wastewater treatment lagoon (WL), the mean estrogenic responses to influent extracts, ranging from 56 to 215 ng L−1 E2 equivalents (Table 6), were statistically higher (Mann–Whitney test, P value of 0.0080) than the estrogenic responses observed in influent from WWTPs 1 and 2 (Table 5). Nonetheless, the estrogenicity of effluents from WL was comparable to the estrogenicity of the effluents from the WWTPs (Mann–Whitney test, P value = 0.2398), with mean values ranging from 4 to 13 ng L−1 E2 equivalents. This indicates that treatment of wastewater in the lagoon provided an effluent of equal quality in terms of estrogenicity. However, the estrogenicity observed for treated wastewater samples in the present study were higher than the 1–216 pg/L E2 equivalent values previously reported for wastewater (Kase et al. 2018).

Table 6 Mean (± %SD, n = 3) responses in the ERα CALUX (ng L−1 E2 equivalents) and NFR2 (mg L−1 tBHQ) in treatments with extracts prepared from wastewater collected from WL and surface water downstream of the lagoon discharge
Full size table

The results support literature demonstrating that the CALUX assays have the ability to evaluate the biological potency of wastewater samples (Roberts et al. 2015; Kase et al. 2018; Könemann et al. 2018). Könemann et al. (2018) tested wastewater samples from 17 sites in Central and Southern Europe and concluded that results from the ER-CALUX assay were also comparable to other methods, including the luciferase-transfected human breast cancer cell line (MELN) gene reporter assay, the ER-GeneBLAzer assay, the stably transfected human estrogen receptor-alpha transcriptional activation assay using hERa-HeLa-9903 cells (HeLa-9903 assay) and the planar Yeast Estrogen Screen (pYES). Often, in vitro assays are more sensitive than analytical methods for detecting the presence of agonists (Escher et al. 2012; Kase et al. 2018). This could have been the case in the present study, where significant upregulation of the estrogen receptor and positive responses in the ERα CALUX assay were detected, even though the concentrations of two of the most potent estrogens were below the limits of detection (i.e., EE2, E2). Analytical methods used for the monitoring of priority substances require a LOQ ≤ 30% of the EQS (Könemann et al. 2018). For estrogens, this can only be achieved using state-of-the-art instruments dedicated to such analyses, which was not the case here. However, estrone was detected in samples collected from WWTP 1 and WL and was present at concentrations < LOQ in samples from WWTP 2. Although Brand et al. (2014) reported that steroid estrogens (e.g., E2) were by far the most potent agonists in the ERα CALUX assay, a variety of estrogenic compounds could have contributed to the estrogenic potency observed in wastewater extracts, including alkylphenols, bisphenol A and phytoestrogens.

Finally, surface water samples collected downstream of the WL discharge were also estrogenic, with mean potencies of 5 to 17 ng L−1 E2 equivalents (Table 6). Different values ranging from 100 to 500 pg/L E2-equivalents (EEQ) have been proposed as effect-based trigger (EBT) values for wastewater. Based on the discussion presented in Kase et al. (2018), the use of an EBT of 400 pg /L EEQ seems justified. Considering that the values measured in the treated wastewater (effluent) were much higher, in the range of 1–14 ng/L E2 equivalents, the results indicate a potential ecological risk associated with the discharge of the effluent. In the EU, in the context of human risks, an EBT value for the ERα CALUX assay of 3.8 ng L−1 E2 equivalents has been proposed for drinking water and source waters (Brand et al. 2013, 2014). Since estrogenic activity was detected in surface waters downstream of the lagoon discharge (5 to 17 ng L−1 E2 equivalents, Table 6), it may be advisable to monitor drinking water for estrogenic activity using the ERα CALUX assay or another sensitive in vitro assay.

Results of Nrf2 Assay

The Nrf2 assay is an indicator of oxidative stress in cells. More specifically, the bioassay measures induction of the Nrf2-Keap-ARE pathway, which protects cells against oxidative damage resulting from spontaneous cellular processes or exposure to contaminants. Jia et al. (2015) observed that the Nrf2-Keap-ARE pathway responds to “a very wide range of chemicals,” but did not specify which classes of compounds were active. Martin et al. (2010) reported that 165 of the 309 chemicals tested in the Phase I ToxCast survey conducted by the US EPA induced oxidative stress, as detected by Nrf2 activation. After testing 19 compounds for responses in various in vitro assays, van der Linden et al. (2014) reported that 2,4-dichlorophenol, curcumin, ethyl acrylate, p-nitrophenol and propyl gallate, in addition to tBHQ, gave a positive response in an Nrf2 bioassay. Tang et al. (2013) detected significant Nrf2 activation in treatments with extracts from urban storm water and commented that, “further chemical analysis is required to identify the causative agents for the underlying toxicity.”

Overall, the responses to wastewater extracts in the Nrf2 assay used in the present study indicate that there was a significant (Wilcoxon matched-pairs signed rank test, P value = 0.0001) decrease in the capacity to induce oxidative stress in effluent extracts relative to influent extracts from samples collected at WWTP 1 and WWTP 2 (Table 5) and in WL (Table 6). In treatments with extracts from WWTP 1, exposure to untreated wastewater samples induced mean responses of t-BHQ equivalents ranging from 0.26 to 0.34 mg L−1; which is significantly (Wilcoxon matched-pairs signed rank test, P value = 0.0312) higher than the mean responses to effluent samples that did not exceed 0.12 mg L−1 and were as low as 3.5 × 10–3 mg L−1 (Table 5). Likewise, for samples from WWTP 2, exposure to extracts from influent samples resulted in mean t-BHQ equivalents ranging from 0.41 to 0.56 mg L−1, while exposure to effluent samples resulted in significantly (Wilcoxon matched-pairs signed rank test, P value = 0.0312) lower responses, with mean t-BHQ equivalents ranging from 0.17 to 0.24 mg L−1 (Table 5). Jia et al. (2015) also reported that activity was reduced in extracts from samples collected after wastewater treatment in an Nrf2 Luciferase Luminescence Assay with a different cell line.

The data for WL showed that the treatment did not significantly (Wilcoxon matched-pairs signed rank test, P value = 0.0625) reduce capacity to induce oxidative stress (Table 6). The responses to extracts from surface water samples collected downstream of WL indicated that there were compounds present that induce oxidative stress, but the responses to extracts from surface water showed no apparent correlation with the activity in the corresponding lagoon effluent samples. For example, while mean t-BHQ equivalent values for effluent and surface water samples collected in May of 2014 were 0.25 and 0.14 mg L−1, respectively, treatments with extracts from June 2014 effluent and surface water samples showed mean responses of 0.12 and 0.28 ± mg L−1 t-BHQ equivalents, respectively (Table 6). However, exposure to extracts from effluent and surface water samples collected in September resulted in very similar responses (Table 6).

Conclusions

Analysis of wastewater extracts using the 10-Pathway Reporter Array showed widespread ER, liver X, vitamin D and retinoid X receptor upregulation. Also, exposure to certain samples resulted in glucocorticoid, peroxisome and PPAR receptor upregulation. Future work should evaluatethe ability of wastewater extracts to induce these receptors, as these were not further investigated in the present study. There are CALUX reporter gene assays that can test for agonistic activity for the glucocorticoid and progesterone receptors (Brand et al. 2013), PPAR receptors (Gijsbers et al. 2011), and oxidative stress through the Nrf2 pathway (van der Linden et al. 2014). Previous studies have associated endocrine disruption (e.g., estrogenicity) to sediments, regardless of the level of endocrine activity in the surface water (Peck et al. 2004; Koyama et al. 2013). Thus, future work could also include evaluations of the activity of extracts prepared from sediments in receiving waters or sludge/biosolids from wastewater treatment plants.

The present study demonstrated that there were reductions in the estrogenic activity in treated wastewater relative to untreated wastewater collected from both conventional WWTPs and a sewage lagoon, except for the lagoon treatment that did not result in a significant decrease in oxidative stress. Based on the results obtained, the use of a nitrification process, expected based on literature to further reduced estrogenic potency, did not provide a greater removal of estrogenicity. In addition, E2 equivalent values for the treated wastewaters and surface waters were higher than previously reported for treated wastewater (Kase et al. 2018) and effect-based trigger values, suggesting that further investigations are required to minimize the ecological risks associated with the discharge of the treated effluent. The reductions in cellular responses were consistent with improvements in wastewater quality after treatment that were measured using standard wastewater monitoring methods (e.g., TSS, cBOD, total ammonia). However, analysis of targeted contaminants in wastewater from the classes of pharmaceuticals and personal care products, pesticides and steroid hormones indicated that several of these compounds were below detection limits (e.g., E2 and EE2) or not removed effectively by wastewater treatment (e.g., carbamazepine, trimethoprim, acesulfame K, DEET).

No correlations could be established between the results of chemical analyses and the cellular activity detected in in vitro assays. Similarly, Tang et al. (2014) used four in vitro cell-based assays to evaluate biological responses to extracts from wastewater and recycled water. They also analyzed these extracts for 299 organic compounds and observed that the concentrations of known chemicals explained less than 3% of the cytotoxicity and less than 1% of the oxidative stress responses to the extracts. In addition, some in vitro assays, such as the CALUX assays, can be more sensitive at detecting cellular responses from treatments with wastewater extracts than the analytical methods used to detect the target compounds that may induce these responses (e.g., steroid hormones) (Könemann et al. 2018). There are currently no regulatory applications of in vitro bioassays (Snyder and Leusch 2018) and the present study contributes further evidence for the implementation of bioassays in regulatory frameworks, such as the one proposed in Europe (Könemann et al. 2018; Snyder and Leusch 2018). Overall, the present study further demonstrates that in vitro assay systems can be valuable additions to evaluating treatment performances, especially when target analytes are below limits of quantification. Further studies are required to assess differences in treatment performance for the removal of estrogenic activity and oxidative stress.