Article pubs.acs.org/est
Occurrence and In Vitro Bioactivity of Estrogen, Androgen, and Glucocorticoid Compounds in a Nationwide Screen of United States Stream Waters Justin M. Conley,† Nicola Evans,† Mary C. Cardon,† Laura Rosenblum,‡ Luke R. Iwanowicz,§ Phillip C. Hartig,† Kathleen M. Schenck,∥ Paul M. Bradley,⊥ and Vickie S. Wilson*,† †
U.S. Environmental Protection Agency/National Health and Environmental Effects Research Laboratory/Toxicity Assessment Division, Research Triangle Park, North Carolina 27711 United States ‡ CB&I Federal Services, Cincinnati, Ohio 45212 United States § U.S. Geological Survey/Leetown Science Center, Kearneysville, West Virginia 25430 United States ∥ U.S. Environmental Protection Agency/National Risk Management Research Laboratory/Water Supply and Water Resources Division, Cincinnati, Ohio 45220 United States ⊥ U.S. Geological Survey/South Atlantic Water Science Center, Columbia, South Carolina 29210 United States S Supporting Information *
ABSTRACT: In vitro bioassays are sensitive, effect-based tools used to quantitatively screen for chemicals with nuclear receptor activity in environmental samples. We measured in vitro estrogen (ER), androgen (AR), and glucocorticoid receptor (GR) activity, along with a broad suite of chemical analytes, in streamwater from 35 wellcharacterized sites (3 reference and 32 impacted) across 24 states and Puerto Rico. ER agonism was the most frequently detected with nearly all sites (34/35) displaying activity (range, 0.054−116 ng E2Eq L−1). There was a strong linear relationship (r2 = 0.917) between in vitro ER activity and concentrations of steroidal estrogens after correcting for the in vitro potency of each compound. AR agonism was detected in 5/35 samples (range, 1.6−4.8 ng DHTEq L−1) but concentrations of androgenic compounds were largely unable to account for the in vitro activity. Similarly, GR agonism was detected in 9/35 samples (range, 6.0−43 ng DexEq L−1); however, none of the recognized GR-active compounds on the target-chemical analyte list were detected. The utility of in vitro assays in water quality monitoring was evident from both the quantitative agreement between ER activity and estrogen concentrations, as well as the detection of AR and GR activity for which there were limited or no corresponding target-chemical detections to explain the bioactivity. Incorporation of in vitro bioassays as complements to chemical analyses in standard water quality monitoring efforts would allow for more complete assessment of the chemical mixtures present in many surface waters.
■
INTRODUCTION The exposure of aquatic organisms to complex mixtures of anthropogenic chemicals in surface water is well established in the scientific research community.1 Contaminants of emerging concern (CEC) are of particular interest due to their potential biological effects (including endocrine disruption) and include relatively new persistent organic pollutants (e.g., polybrominated diphenyl ethers, perfluoro-alkyl acids), pharmaceuticals and personal care products, veterinary medicines, natural and synthetic hormones, and nanomaterials.2 In addition to wildlife exposures, there is the potential for human exposure to surface water CECs via direct dermal absorption, ingestion of persistent contaminants in treated municipal drinking water,3 or consumption of CEC-contaminated aquatic organisms. Numerous sampling and analysis efforts have been conducted © XXXX American Chemical Society
thus far to help define the scope (number and types of compounds) and severity (concentrations) of exposure to these compounds in aquatic systems. However, broader screening and monitoring efforts are needed to more fully characterize the scope and severity of CECs in surface waters, define the risk to aquatic organisms, estimate potential human exposure, and prioritize chemicals for future regulatory action. Endocrine disrupting chemicals have received considerable attention over the past several decades due to the nature of these compounds to specifically target and adversely affect Received: December 23, 2016 Revised: March 7, 2017 Accepted: March 22, 2017
A
DOI: 10.1021/acs.est.6b06515 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Table 1. Identification and Watershed Characteristics of Sampling Sites Determined from 2011 National Land Cover Database. Reference Sites (#1−3) Indicated in Bold site #
stream
state
drainage (mi2)
urban (%)
agriculture (%)
population (per mi2)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
Penn Swamp Branch West Clear Creek North Sylamore Creek New River Santa Ana River Sycamore Slough South Platte River C-111 Canal Hillsboro Canal Sope Creek Fourmile Creek Sand Run Gulch South Fork Zumbro River Sunrise River Tributary Hohokus Brook West Branch Delaware River Chisholm Creek Zollner Creek Rio Bairoa Trinity River Hawksbill Creek Fishtrap Creek Rio Fajardo Abrams Creek Mill Creek North Dry Creek Jordan Creek Blue River Tembladero Slough Deep Creek East Branch Perkiomen Creek Chicago Sanitary and Ship Canal Fall Creek Enoree River Hite Creek
NJ AZ AR CA CA CA CO FL FL GA IA ID MN MN NJ NY OK OR PR TX VA WA PR GA OH NE PA MO CA OR PA IL NY SC KY
4.60 241.42 58.66 1471.12 2261.44 64.30 4082.21 51.04 311.34 30.81 60.94 79.49 312.00 6.59 20.47 48.22 38.78 15.90 7.68 6264.63 68.52 16.45 20.38 79.58 177.93 78.74 82.32 184.19 153.97 48.98 29.35 749.34 126.35 84.80 5.54
0.2 0.0 0.2 4.9 23.5 2.6 7.5 4.9 1.6 38.6 9.5 0.2 8.5 36.5 19.9 0.7 50.4 5.0 27.2 15.4 4.2 50.4 4.2 1.0 5.4 0.2 13.8 41.4 13.7 7.9 13.5 77.3 1.3 23.6 51.9
0.0 0.0 1.5 27.3 4.6 95.0 0.6 65.4 93.4 0.6 77.7 47.2 62.3 22.4 0.1 34.8 5.4 88.0 1.1 14.1 31.5 36.3 2.4 30.2 75.5 88.3 48.5 29.0 26.7 40.4 43.7 1.9 45.3 8.5 10.3
0 0 1 52 1342 7 436 298 75 2340 393 29 317 1194 2746 63 2289 72 2733 696 140 0 300 12 167 3 1020 1696 977 392 929 5773 124 1110 2323
human and wildlife reproductive and developmental processes.4 Overt impairment of reproductive ability and clearly defined malformations resulting from exposure to synthetic chemicals that alter endocrine function have been described in free-living fish populations5 and American alligators,6 among others. In vitro and in vivo experiments have identified many chemicals as endocrine disruptors by linking mechanism of action with adverse apical effects, but many more chemicals with endocrine disrupting capability (e.g., chemical metabolites, biotransformation products, etc.) may be present in aquatic ecosystems than are typically screened for in traditional chemical analyses. In vitro bioassays provide a high degree of added value to the assessment of water quality and there is considerable interest in incorporating in vitro assays into regulatory water quality monitoring regimes.7−10 Studies in Africa,11 Australia,12−14 Europe,15,16 Japan, 17 and the United States 18−22 have successfully utilized in vitro assays to screen for a broad suite of biological activities including nuclear receptor (ant)agonism (androgen,15,19,23 aryl hydrocarbon,12,24 estrogen,15,18,19,23,25 glucocorticoid,15,23,26 peroxisome-proliferator,12,27 pregnane X,12 progesterone,15,23 thryroid12,21) as well as genotoxicity21 and mutagenicity.21,28 Further, these analyses have been applied to a wide variety of freshwater sample types including surface
water,11,12,18,19,29 wastewater influent,13 wastewater effluent,12,13,17,30 recycled water,9,12 stormwater,12 treated drinking water,12,20 and groundwater.15,21 The Chemical Mixtures and Environmental Effects study is a nation-wide interagency, collaborative investigation between U.S. Geological Survey (USGS) and U.S. Environmental Protection Agency (USEPA).31 The goals of the study are to better define what chemicals occur in streams of the U.S. and their concentrations, what land-use practices influence the composition of chemical mixtures, and whether exposure of fish and other aquatic organisms to these mixtures may be resulting in adverse health effects. Samples from 38 stream sites (34 highly developed/impacted and 4 undeveloped/reference) across 24 states and Puerto Rico were collected and analyzed for over 800 chemical compounds. In addition, a variety of in vitro bioassays were utilized to evaluate biological activity of samples and relate those activities to the chemical analyses. The results of the target-chemical analysis are discussed in detail elsewhere (Bradley et al.32). Here, we describe the results of in vitro estrogen receptor (ER), androgen receptor (AR), and glucocorticoid receptor (GR) transcriptional activation assays and compare the results with reported concentrations of known B
DOI: 10.1021/acs.est.6b06515 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
hexane. The organic layers were combined, 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 LC-MS/ MS with chromatographic separation using a Restek Pinnacle DB Biphenyl analytical column (Bellefonte, PA) and detection using a Thermo Scientific 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 daughter ions. Quantitation was performed using internal standard calibration. The lowest concentration minimum reporting limit (LCMRL)35 values (in ng L−1) were: E1, 0.28; E2, 0.26; EE2, 0.14; E3, 1.38; P, 0.66; and T, 0.092. In Vitro Transcriptional Activation Assays by USEPA. Both the estrogen-responsive T47D-KBluc36 (CRL-2865) and androgen-responsive MDA-kb237 (CRL-2713) assays were developed by USEPA and are publicly available from the American Type Culture Collection (ATCC; Manassas, VA). Both assays have been successfully utilized for more than a decade to screen chemicals and mixtures for estrogen and androgen receptor (ant)agonism. For assessment of GR activity, CV-1 cells were transduced (adenovirus) with human GR and luciferase reporter genes (details below). This specific assay has not been previously published but is similar to prior work utilizing adenoviral AR transduction.38 All in vitro assays were run in 96-well plates (Corning Costar 9610) for detection of receptor agonism. Dried concentrates were resuspended in 50 μL 100% ethanol (EtOH) and diluted into cell culture media. Exposure time was 24 h and luciferase activity was quantified in relative luminescence units (RLU) using a luminometer (BMG Fluostar Omega; BMG Labteck, Durham, NC). Throughout this process cells were visually scored for cytotoxicity39 and if cytotoxic effects were observed in a given well, data from that well were excluded from further analysis. All samples were analyzed using ≥2 biological replicate plates (i.e., unique cell passages) with 4 technical replicates per treatment on all plates. Estrogen Receptor − T47D-KBluc Assay. T47D-KBluc cells naturally express endogenous estrogen receptors (primarily ERα40) and were engineered to stably express a triplet estrogen response element-luciferase promoter-reporter gene construct.36 Cells were maintained and assayed as previously described.20 The initial sample extract dilution was 1:1000 into media (0.1% EtOH) followed by 2-fold serial dilutions in media down to a 1:12 8000 dilution (8 dilutions). A standard curve of E2 (0.3, 1, 3, 10, 30 pM), vehicle blank (EtOH), and E2 plus inhibitor (ICI-182 780) were run concurrently with 2 sample extracts on each plate. Androgen Receptor − MDA-kb2 Assay. MDA-kb2 cells naturally express endogenous androgen and glucocorticoid receptors and were engineered to stably express a mouse mammary tumor virus-luciferase (MMTV-Luc) promoterreporter gene construct.37 Cells were maintained and assayed as previously described.19 The initial sample extract dilution was 1:1000 into media (0.1% EtOH) followed by 2-fold serial dilutions in media down to a 1:8000 dilution (4 dilutions). A standard curve of DHT (10, 30, 100, 300, 1000 pM), vehicle blank (EtOH), and DHT plus inhibitor (hydroxyflutamide) were run concurrently with four sample extracts on each plate. Glucocorticoid Receptor − CV-1 Transduced with Human GR. CV-1 cells (ATCC CCL-70), which are naturally devoid of GR and AR, were transduced with human GR (Ad/
estrogens, androgens, and glucocorticoids in samples from this nationwide survey.
■
MATERIALS AND METHODS Sample Collection and Preparation. Stream sampling locations were identified from over a thousand locations previously studied by USGS and selected to cover highly developed watersheds with a wide range of contaminant sources and undeveloped reference sites. Streams were sampled between December 2012 and June 2014. At each site grab samples were collected from the center of the streamflow using either a Teflon splitting churn or directly into sample bottles prepared for specific laboratory analyses (see details in Romanok et al.33). Samples were stored at 4 °C and then shipped overnight on ice to USGS and USEPA laboratories for extraction and subsequent analyses. Samples from three sites were damaged during shipping, therefore the present analysis includes the evaluation of 32 impacted sites and 3 reference sites (Table 1). Samples were subject to a suite of three in vitro steroid hormone receptor transcriptional activation assays, a bioluminescent yeast estrogen screen (BLYES), and two independent chemical analyses. USEPA chemical analyses included an abbreviated suite of compounds (17α-ethinyl estradiol (EE2), 17β-estradiol (E2), estriol (E3), estrone (E1), progesterone (P), and testosterone (T)), while multiple USGS laboratories (see Romanok et al.33) conducted chemical analyses for a broad suite of contaminants (>800 compounds). Specifically, for the USEPA in vitro and chemical analyses reported here, 600 mL samples were preserved with 6 mg CuSO4·5H2O in the field and amended with 24 mg Na2EDTA (copper reducer) and 5 ng Bisphenol A-d16 (extraction surrogate) in the lab. The concentration of the extraction surrogate (BPA-d16; 3.6 pM) was ∼1000-fold below the lowest concentration that produces a response above background in the T47D-KBluc assay. Samples were then concentrated by solid phase extraction (47 mm C18 Empore disks, 3M, St. Paul, MN). Disks were eluted with 9 mL 90:10 methanol:acetone (v:v). A 1.5 mL aliquot of extract was removed (equivalent to 100 mL of field sample) and concentrated to dryness and sent to Research Triangle Park, NC for USEPA in vitro analyses (see below). The remaining 7.5 mL extract was used to perform the USEPA chemical analyses for steroid hormones (see below). Chemical analyses by USEPA. Solid anhydrous analytical standards of E1 (CAS# 53−16−7), E2 (CAS# 50−28−2), EE2 (CAS# 57−63−6), E3 (CAS# 50−27−1), T (CAS# 58−22− 0), and P (CAS# 57−83−0) were purchased from Steraloids (Newport, RI). The stably labeled internal standards testosterone-d3, progesterone-d9, and estriol-d3 were purchased from CDN Isotopes (Pointe-Clare, Canada); 13C6estradiol, 13C6-estrone, 13C2-ethinyl estradiol, and Bisphenol Ad16 (extraction surrogate) were purchased from Cambridge Isotopes Laboratory (Tewksbury, MA). Detailed description of the analytical methodology utilized in the present study has been previously reported.34 Briefly, sample extracts were spiked with a solution containing stably 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. As a cleanup step, the solution was extracted with three 0.5 mL portions of C
DOI: 10.1021/acs.est.6b06515 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Concentrations of in vitro estrogenic activity (panels a, b) and estrogen compounds (panels c, d) across sampling sites. In vitro estrogen receptor transcriptional activation (T47D-KBluc, panel a) and bioluminescent yeast estrogen screen (BLYES, panel b) reported as 17β-estradiol equivalents (E2Eq). Independent chemical analyses for suites of analytes were conducted by USEPA (panel c) and USGS (panel d).
GR4) and MMTV-Luc (Ad/mLuc7) promoter-reporter constructs using adenoviral vectors.41 Briefly, cells were grown by standard techniques, similar to above with T47DKBluc but in 60 mm Petri dishes. Once cells reached confluence they were transduced with GR and MMTV-Luc constructs at multiplicities of infection of 50. After 24 h incubation with adenovirus, cells were rinsed, suspended in 10 mL media, and the transduced cells from a single 60 mm dish were used to seed a single 96-well plate (100 μL cell suspension per well). Simultaneously, sample extracts were diluted at 1:500 in media (0.2% EtOH) followed by 2-fold serial dilutions in media down to 1:4000 (4 dilutions). A 100 μL aliquot of each extract dilution was added to the 100 μL of media and cells already present in the 96-well plate for an additional 2-fold dilution. A standard curve of dexamethasone (3, 10, 30, 100, 300 pM, 1, 3, 10, 30 nM) and a vehicle blank (EtOH) were run on each plate along with 4 sample extracts. All positive samples were rerun with MMTV-Luc transduced but not GR and lack of reporter activity confirmed GR specificity of the sample. In Vitro Assay Data Processing. Data analysis was performed using GraphPad Prism (v7.01, San Diego, CA) as previously described.20 Briefly, raw data were normalized to background (vehicle control), log10 transformed, and converted to % maximum response based on saturating levels of agonist standard (E2, DHT, or Dex). Dose response curves were fit with a four-parameter logistic regression using the following constraints: top =100, bottom =0, and hillslope = hillslope of concurrent reference agonist standard curve. The reference agonist (Ref) and sample EC50 values were determined from the logistic regression curves and used to calculate the bioanalytical equivalent (BioEq) concentrations (i.e., E2Eq, DHTEq, or DexEq) of the samples using the following equation:
BioEq =
Ref EC50 (Sample EC50)(Sample EF)
where all concentrations are expressed in ng L−1, the sample enrichment factor (EF) represents the degree to which the raw field sample was concentrated via solid phase extraction prior to testing (2000-fold). Reporting limits were determined using the 95% prediction interval across all standard curves from the study as previously described:20 0.032 ng E2Eq L−1 for T47DKBluc, 0.82 ng DHTEq L−1 for MDA-kb2, and 6.8 ng DexEq L−1 for CV-1 GR assay. BLYES Assay by USGS. BLYES uses Saccharomyces cerevisiae containing an estrogen-inducible bioluminescent reporter to quantitatively assess estrogenic activity relative to 17β-estradiol.42 Strain BLYES was purchased from 490 BioTech (Knoxville, TN) and run as previously described with minor modifications (see Supporting Information (SI)).43 The final sample extract dilution was 1:40 in media. Each assay plate included an E2 standard curve ranging from 4−500 pg mL−1 and media controls to establish background luminescence. The estrogenicity of each sample relative to the E2 standard curve was calculated using a four-parameter logistic regression (SoftMax Pro 6.2.2, Molecular Devices) and corrected for sample enrichment. The BLYES assay detection limit was 0.2 ng L−1. Samples from sites 2, 10, 25, and 31 were not analyzed due to sample vial breakage during shipping. Chemical Analyses by USGS. An extensive suite of >800 target-chemical compounds were monitored by multiple USGS instrumental methodologies. A comprehensive reporting of all methodological details can be found in Romanok et al.33 and results of chemical analyses are detailed in Bradley et al.32 Quality Assurance. Field quality assurance included 15 field blanks and 13 matrix spikes (for USGS chemical analyses only). Detailed laboratory QA/QC for USGS analytical D
DOI: 10.1021/acs.est.6b06515 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology chemical determinations are reported in Romanok et al.33 USEPA laboratory quality assurance included lab blanks (n = 20), lab fortified blanks (LFB) containing internal standards (n = 20, for chemical analyses only), field blanks (n = 3), and E2 standards (n = 5, for T47D-KBluc assay). We were unable to determine target analyte extraction recovery from field samples due to a lack of a matrix-matched duplicate; however, the extraction surrogate, bisphenol-d16, displayed acceptable recovery (50−150%). Further, all LFBs displayed acceptable recovery (50−150%) for all internal standards. E2 standards (mean ± sd, 3.3 ± 1.1 pM) were within 50−150% of target concentration (3 pM). Two of 20 USEPA laboratory blanks contained low ER activity in the T47D-KBluc assay (
