In vitro effects-based method and water quality screening model for use in pre- and post-distribution treated waters

Abstract

Recent urban public water supply contamination events emphasize the importance of screening treated drinking water quality after distribution. In vitro bioassays, when run concurrently with analytical chemistry methods, are effective tools to evaluating the efficacy of water treatment processes and water quality. We tested 49 water samples representing the Chicago Department of Water Management service areas for estrogen, (anti)androgen, glucocorticoid receptor-activating contaminants and cytotoxicity. We present a tiered screening approach suitable to samples with anticipated low-level activity and initially tested all extracts for statistically identifiable endocrine activity; performing a secondary dilution-response analysis to determine sample EC₅₀ and biological equivalency values (BioEq). Estrogenic activity was detected in untreated Lake Michigan intake water samples using mammalian (5/49; median: 0.21 ng E2Eq/L) and yeast cell (5/49; 1.78 ng E2Eq/L) bioassays. A highly sensitive (anti)androgenic activity bioassay was applied for the first time to water quality screening and androgenic activity was detected in untreated intake and treated pre-distribution samples (4/49; 0.93 ng DHTEq/L). No activity was identified above method detection limits in the yeast androgenic, mammalian anti-androgenic, and both glucocorticoid bioassays. Known estrogen receptor agonists were detected using HPLC/MS-MS (estrone: 0.72-1.4 ng/L; 17α-estradiol: 1.3-1.5 ng/L; 17β-estradiol: 1.4 ng/L; equol: 8.8 ng/L), however occurrence did not correlate with estrogenic bioassay results. Many studies have applied bioassays to water quality monitoring using only relatively small samples sets often collected from surface and/or wastewater effluent. However, to realistically adapt these tools to treated water quality monitoring, water quality managers must have the capacity to screen potentially hundreds of samples in short timeframes. Therefore, we provided a tiered screening model that increased sample screening speed, without sacrificing statistical stringency, and detected estrogenic and androgenic activity only in pre-distribution Chicago area samples.

Keywords: Androgen; Effects-based method; Estrogen; Tapwater.

Figure 1.. Water sampling scheme

Three sample types were taken from sites around the Greater Chicago metropolitan area including untreated Lake Michigan source water (red circle), water filtration plant (WFP)-treated pre-distribution water (green “x”), and post-distribution residential tapwater (blue triangle). Untreated intake and WFP-treated water was sampled at two WFPs in East Chicago, IN, one WFP in North Chicago, IL and one WFP in South Chicago. Residental post-distribution tapwater samples representing the North (n = 14) and South Chicago (n = 16) WFPs distribution areas were also sampled.

Figure 2.. Tiered screening and statistical analysis model

for bioassays that measure endocrine activity in tapwater sample extracts. Here the model is presented using the CV1-chAR bioassay and Chicago area tapwater samples results. In tier one: Is there activity? all samples are screened with minimal dilutions and statistically compared to vehicle control (p > 0.05) using general linear model (GLM) and multiple comparison procedure (MCP). In tier two: How much activity? only active samples from tier one were screened again using a dilution-response to determine biological equivalency values (BioEq). Each sample BioEq is then compared to the bioassay method detection limit (MDL) and relevant trigger value.

Figure 3.. Estrogenic activity

was measured in extract samples from untreated Lake Michigan intakes, WFP treated pre-distribution waters, and post-distribution waters using the A) T47D-KBluc (biological replicate data is shown as mean ± standard deviation) and B) BLYES bioassays. Method detection limit (MDL) for T47D-KBluc was 0.0044ng E2Eq/L and 0.1 ng E2Eq/L for BLYES assay.

Figure 4.. Estrogenic bioassay comparison

using estrogenic activity detected in water sample extracts. T47D-KBluc and BLYES bioassays are compared through linear regression analysis where y = 0.0439x + 0.0640 and R² = 0.82.

Figure 5.. Androgenic activity

was measured in extract samples from untreated Lake Michigan, WFP treated pre-distribution, and post-distribution waters using a CV1-chAR bioassay. Biological replicate data is shown as mean ± standard deviation and method detection limit (MDL) was 0.30 ng DHTEq/L.