Headspace-Free Setup of in Vitro Bioassays for the ... - ACS Publications

Oct 14, 2013 - trihalomethanes, 1,1-dichloroethene, bromoethane, and two haloacetonitriles) that cover a wide range of air−water partition coefficie...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/crt

Headspace-Free Setup of in Vitro Bioassays for the Evaluation of Volatile Disinfection By-Products Daniel Stalter,* Mriga Dutt, and Beate I. Escher National Research Centre for Environmental Toxicology (Entox), The University of Queensland, 39 Kessels Road, Brisbane Qld 4108, Australia S Supporting Information *

ABSTRACT: The conventional setup of in vitro bioassays in microplates does not prevent the loss of volatile compounds, which hampers the toxicological characterization of waterborne volatile disinfection by-products (DBPs). To minimize the loss of volatile test chemicals, we adapted four in vitro bioassays to a headspace-free setup using eight volatile organic compounds (four trihalomethanes, 1,1-dichloroethene, bromoethane, and two haloacetonitriles) that cover a wide range of air−water partition coefficients. The nominal effect concentrations of the test chemicals decreased by up to three orders of magnitude when the conventional setup was changed to a headspace-free setup for the bacterial cytotoxicity assay using bioluminescence inhibition of Vibrio f ischeri. The increase of apparent sensitivity correlated significantly with the air−water partition coefficient. Purge and trap GC/MS analysis revealed a reduced loss of dosed volatile compounds in the headspace free setup (78−130% of nominal concentration) compared to a substantial loss in the conventional set up (2−13% of the nominal concentration). The experimental effect concentrations converged with the headspace-free setup to the effect concentrations predicted by a QSAR model, confirming the suitability of the headspace-free approach to minimize the loss of volatile test chemicals. The analogue headspace-free design of the bacterial bioassays for genotoxicity (umuC assay) and mutagenicity (Ames fluctuation assay) increased the number of compounds detected as genotoxic or mutagenic from one to four and zero to two, respectively. In a bioassay with a mammalian cell line applied for detecting the induction of the Nrf-2mediated oxidative stress response (AREc32 assay), the headspace-free setup improved the apparent sensitivity by less than one order of magnitude, presumably due to the retaining effect of the serum components in the medium, which is also reflected in the reduced aqueous concentrations of compounds. This study highlights the importance of adapting bioanalytical test setups when volatile/semivolatile compounds are present in the sample to avoid the loss of chemicals and thus to avoid underestimating the toxicity of mixtures and complex environmental samples.

■

DBPs,10 but quantitative exposure of volatile chemicals in cellbased bioassays remains an issue. Standard in vitro test procedures in microtiter plates incur the loss of volatile chemicals and hence overestimate their effect concentrations if these are based on nominal concentrations.11−13 Accordingly, by using negligible depletion solid-phase microextraction (nd-SPME) Heringa et al.12 demonstrated that the free concentration is a better measure of exposure than the nominal (i.e., dosed) concentration. Furthermore, Tanneberger et al.14 demonstrated that the impact of solvents and dosing procedure is greater the more volatile and/or hydrophobic the test chemicals are, indicating that concentrations diminish over time due to evaporation, binding to well plate surfaces and medium constituents.11,15 Therefore, attempts have been made to minimize the loss of volatile compounds during exposure in microtiter plate setups. Kramer et al.15 developed a partition-controlled

INTRODUCTION Waterborne volatile disinfection by-products (DBPs) contribute considerably to toxic effects in chlorinated drinking water, swimming pool water, and wastewater. DBPs in chlorinated drinking water can exhibit high toxicity,1 with the prominent mode of action being mutagenicity.2 DBPs from indoor swimming pools have been found to be mutagenic using the Ames assay3 and can enter swimmers’ blood leading to DNA damage in blood lymphocytes.4 In wastewater samples, volatile substances contribute considerably to cytotoxic effects, particularly after ozonation,5 while wastewater chlorination also leads to the formation of toxic and volatile DBPs.6 Given that the major portion of DBPs in swimming pools and drinking water are not known yet,7 a range of unknown volatile compounds can be expected in environmental samples. Yet, the characterization of complex environmental mixtures (e.g., drinking water, swimming pool water, and wastewater) commonly incurs the loss of volatile chemicals due to sample preparation and extraction.3,8,9 Purge and trap methods may allow for the enrichment of volatile © 2013 American Chemical Society

Received: February 28, 2013 Published: October 14, 2013 1605

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

high volatility and mutagenicity and because it is also a known waterborne volatile chemical.20 As nonvolatile performance reference compounds, phenol was used in the Microtox assay, 4-nitroquinoline1-oxide in the umuC assay, nitrofurantoin in the Ames assay, and t-butylhydroquinone (tBHQ) in the AREc32 assay for oxidative stress response. All chemicals were purchased from Sigma Aldrich, Australia. Bioassays. Four microplate-based bioassays, which include bacterial as well as mammalian cell-based test systems, were selected because they had been shown to be sensitive to DBPs in previous studies.1,9,21 The test performances of various bioassay setups were compared to the standard procedures using standard nonvolatile reference compounds as well as the eight selected volatile or semivolatile compounds (Table 1). The bacterial cytotoxicity assay using Vibrio f ischeri bioluminescence inhibition22 was selected to validate the different dosing and incubation approaches because a QSAR was available to predict the effect concentrations of those chemicals that act as baseline toxicants in this bioassay (eq 1).18 The experimental effect concentrations of the volatile chemicals were compared to the effect concentrations predicted with the QSAR (QSAR-EC50) for a simple assessment of whether the correct EC50 had been reached.

dosing system to maintain constant concentrations of volatile and hydrophobic compounds but did not prevent volatilization. Alternatively, closed systems were applied, or dosing was accomplished via the gas phase,11,16,17 but in these systems, the dosing of mixtures of volatile and nonvolatile chemicals, as they occur in disinfected water samples, is difficult to accomplish and not suitable for routine application of bioassays in water quality monitoring. Furthermore, effect concentrations derived by gasphase in vitro exposure systems are hardly comparable to actual concentrations of DBPs in the aqueous phase. In the present study, we adopted a headspace-free setup to reduce the loss of volatile chemicals in in vitro bacterial and mammalian cell-based, high-throughput bioassays. We validated the approach with established, nonvolatile, reference compounds and compared the headspace-free setup of four bioassays with the conventional procedures with headspace to ensure a comparable performance of the cell lines with the different test designs. To assess the success of the adaptation of in vitro bioassays for use with volatile compounds, we selected eight chemicals that cover a wide range of air−water partition coefficients and measured their effect concentrations with different setups until effect concentrations reached a minimum value and, in the case of the Microtox assay, matched predictions by a QSAR model.18 If the experimental 50% effect concentration (EC50) was higher than the QSAR-EC50, the concentration of the test chemical in the bioassay was reduced, e.g., via volatilization or sorption. Vice versa, if the experimental EC50 was lower than the QSAR-EC50, the chemical must exhibit a specific mode of action18 and could therefore not be used for the QSAR analysis. Once the experimental EC50 of the nonspecifically acting chemicals converged to the QSAR-EC50 with improved incubation and dosing procedures, the test design is deemed appropriate to prevent the loss of volatile chemicals. This QSAR approach circumvented the need to measure the freely dissolved concentration, which is only a fraction of the dosed nominal concentration.12,19 This is particularly important for in vitro bioassays with small volumes of assay medium (e.g., in 96-well or 384-well plates) as well as for complex mixtures of volatile compounds where it is not possible to easily measure the concentrations in the assay medium, even with methods like ndSPME because negligible depletion cannot be assured in such small volumes.

■

− log EC50 = 0.72 × log Klipw + 1.32

(1)

The liposome−water partition coefficient (Klipw) was proposed by Vaes et al.23 as a more suitable hydrophobicity descriptor than the octanol−water partition coefficient for polar and nonpolar baseline toxicants.23 The QSAR used here (eq 1) was validated using a range of organic contaminants that are typically occurring in wastewater, and almost all of the investigated compounds, apart from antibiotics, fell within one log-unit of the baseline toxicity QSAR.18 The umuC assay with Salmonella typhimurium (TA1535/pSK1002) was selected as a simple screening tool for genotoxicity.24 The Ames was included as it has been central for mutagenicity assessment since the 1970s.25 We applied the Ames fluctuation test26 modified to a microplate version according to Reifferscheid et al.27 The oxidative stress induction (AREc32 assay)28 was included as a reactive toxicity endpoint that is particularly sensitive to DBPs.9 Each volatile DBP was run in duplicate or triplicate, and each compound was tested in at least two independent test runs. The number of test runs per compound and setup as well as replicates per test is given in the Figure and Table legends. The variability is given as the coefficient of variance CV (standard deviation divided by the mean) in the Tables and as standard deviation in the Figures. Methanol was used as solvent, and each compound was tested in a 1:2 dilution series in concentrations appropriate to derive a concentration−effect curve. The solvent to media ratio was kept constant across the treatments (1% of methanol v/v). This is a higher solvent content than typically recommended for cell-based bioassays29 but could not be avoided due to the applied dosing regime. Controls were prepared with 1% methanol as well, and results were not affected by the solvent. The assessment endpoint was the 50% effect concentration (EC50) for the Microtox assay derived from a log−logistic concentration−effect curve or the effect concentration that elicits an induction ratio of 1.5 (ECIR1.5) for genotoxicity, mutagenicity, and oxidative stress. The induction ratio IR is defined as the ratio of effect of the sample divided by

MATERIALS AND METHODS

Test Chemicals. We selected seven DBPs that cover a wide range of volatility, expressed as saturation vapor pressure (p*), and tendency to escape an aqueous phase, expressed as Henry coefficient (KH) or air− water partition coefficient (Kaw, Table 1). We also included bromoethane, which is not a DBP, in the set of volatile test chemicals due to its

Table 1. Selected Compounds and Their Physicochemical Properties at 25 °C Sorted According to Decreasing Tendency to Escape the Aqueous Phase abbreviation 1,1-dichloroethene bromoethane chloroform dichloromethane bromoform iodoform dichloroacetonitrile trichloroacetonitrile a

DCE BE CF DCM BF IF DCAN TCAN

CAS 75-35-4 74-96-4 67-66-3 75-09-2 75-25-2 75-47-8 3018-12-0 545-06-2

p* (mm Hg)a

purity (%) 99.5 ≥99.0 ≥99.0 ≥99.8 ≥99.0 99.0 99.9 99.9

600.85 454.77 199.96 448.00 5.17 0.03 21.68 68.60

KH (atm m3/mol)b −2

2.61 × 10 1.45 × 10−2 3.67 × 10−3 3.25 × 10−3 5.35 × 10−4 3.06 × 10−5 3.79 × 10−6 1.34 × 10−6

Kawc

Kowd

1.17 3.06 × 10−1 1.64 × 10−1 1.45 × 10−1 2.39 × 10−2 1.37 × 10−3 1.69 × 10−4 5.98 × 10−5

134.9 40.7 93.3 17.8 251.2 1071.5 1.9 123.0

Vapor pressure. bHenry coefficient. cAir−water partition coefficient. dOctanol−water partition coefficient. 1606

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

the average effect observed in the controls. The ECIR1.5 was derived using linear regression.29 Details on the bioassay procedures are provided in the Supporting Information, section S1.

setup, we increased the liquid volumes of all components proportionally to reach the maximum volume per well of the microplates (300−400 μL, depending on the microplate and the sealing; Table 2), and we employed a variety of sealing techniques to avoid air in the wells and volatilization of the test compounds (Figure 1B and Table 2). For the umuC and Ames assay, we proportionally adapted the bacterial numbers to the adapted liquid volume per well to ensure the same cell concentration in the different setups. This is important as in both assays after the exposure, a fraction of the cells is transferred to a second plate for a second incubation step where we applied only a setup with headspace as here the bioavailable fraction is not relevant anymore and because this enables a better comparison between the setups. In the Vibrio f ischeri assay, we aimed at the same light emission as in the conventional setup (i.e., 600,000 relative light units per well). Accordingly, the cell density was not kept constant as this would have increased the signal in the setup with higher volume, but the total number of cells was kept constant with and without headspace. In the mammalian cell assay, the total cell number was the same with and without headspace. The ratio of exposed plastic to media volume decreased in our headspace-free setup (e.g., ca. 27% decrease in the headspace-free setup of the AREc32 assay) because in the case of a volume increase, the surrounding surface always increases slower than the volume. Therefore, filling up the wells does not negatively affect the bioavailable concentration due to the sorption of test compounds to microplate plastic, which is particularly important for serum-free assays.30 We compared the dosing of volatile compounds before and after sealing. To dose after sealing, we used a Hamilton syringe to inject the chemicals (solved in methanol) through the sealing mat.

■

OPTIMIZATION OF THE HEADSPACE-FREE SETUP The volume ratios between headspace and assay medium in the conventional test setups vary typically from 0.5 to 9 (i.e., half to 9× more headspace volume compared to the liquid volume in each test well; Figure 1A and Table 2). For the headspace-free

Figure 1. Well of a 96-well plate with the different bioassay setups. (A) w/HS, conventional setup with headspace and predosing before sealing with a polyurethane sealing membrane; (B) w/o HS plastic, headspacefree setup in plastic microplates with dosing before sealing with aluminum foil and silicone rubber mat on top; and (C) w/o HS glass, headspace-free setup in glass microplates with dosing via direct injection through the preslit PTFE/silicone rubber sealing mat. Blue indicates the liquid level in the wells, which have a total volume of 360−400 μL (see Table 2 for additional information).

Table 2. Bioassay Setups setup with headspace (conventional)

bioassay abbreviation V. f ischeri

w/HS

setup without headspace in plastic w/o HS plastic

setup without headspace in glass, direct dosing w/o HS glass

setup w/o HS in plastic, direct dosing w/o HS plastic, septum

dosing

dosing before sealing

dosing before sealing

sealing

polystyrene plate lid polystyrene Greiner, 96-well, #655075 150 μL

Umu

plate material plate manufacturer no. final liquid volume per well dosing

aluminum foil + silicone rubber mat polystyrene Greiner, 96-well, #655075 396 μL

dosing before sealing

not evaluated

polyurethane membrane polystyrene Greiner, 96-well, #655180 270 μL

Ames

sealing plate material plate manufacturer no. final liquid volume per well dosing

dosing with Hamilton syringe after not evaluated sealing PTFE/silicone rubber mat glass SUN SRI, 96-well, #502423 400 μL

not evaluated

polyurethane membrane Corning, 24-well, #3524 500 μL

AReC32

sealing plate manufacturer no. final liquid volume per well dosing

dosing with Hamilton syringe after not evaluated sealing PTFE/silicone rubber mat SUN SRI, 96-well, #502423 400 μL

dosing before sealing

dosing before sealing

not evaluated

sealing

polyurethane membrane

plate material plate manufacturer no. final liquid volume per well

polystyrene Corning, 96-well, #3610 100 μL

aluminum foil + silicone rubber mat polystyrene Corning, 96-well, #3610 350 μL

dosing before sealing

1607

dosing with Hamilton syringe after not evaluated sealing PTFE/silicone rubber mat glass SUN SRI, 96-well, #502423 400 μL

dosing with Hamilton syringe after sealing silicone rubber mat polystyrene Corning, 96-well, #3610 300 μL

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

apparatus was fitted with a Vocarb 3000 trap (Sigma Aldrich, Australia) and a 5 mL purge vessel. The gas chromatograph inlet was fitted with a RESTEK 1 mm glass liner (Restek Corporation, Bellefonte, Pennsylvania, USA) and was operated in split mode with the purge and trap. The analytical column used for purge and trap was a Zebron ZB-624, 20 m × 0.18 mm diameter ×1.0 μm film thickness (Phenomenex, Torrance, California, USA). The gas chromatograph was operated in constant linear velocity mode with temperature programming. The EST auto sampler used standard 40 mL glass vials fitted with a screw cap and a PTFE-faced silicone septum. The analytes were solved in methanol and spiked to the bioassay medium in the wells of the microplates used for bioassay setups with and without headspace. The solvent concentration in the medium was 1% and two concentrations were tested which fell close to the detected effect concentrations with a dilution step of 1:2.5 (see Supporting Information, Table S6 for the nominal concentrations). The chemical quantification was done for four setups with original media but without cells in the Microtox (w/HS and w/o HS glass) and the AREc32 assay (w/HS and w/o HS plastic, septum; see Table 2 for details). We used a Hamilton syringe for dosing. After exposure (30 min, 20 °C for Microtox; 37 °C, 24 h for AREc32) 80 μL of the medium was aspirated with a Hamilton syringe and directly transferred to an auto sampler vial, filled with 40 mL ultrapure water. As medium-free comparison, we spiked ultrapure water headspace-free in a 1.8 mL amber glass vial with PTFE-lined screw caps with the same analyte concentration as in the bioassays and transferred 80 μL to an auto sampler vial. For Bromoethane, the exact concentration could not be determined as Queensland Health Forensic and Scientific Services had no analytical standard available for this compound. Therefore, to calculate the percentage of the nominal concentration of bromoethane in the bioassays the peak area of the ultrapure water spiked samples was used instead.

This direct dosing incurs the risk of artifacts due to insufficient mixing leading to uneven cell exposure. To minimize the risk of artifacts caused by uneven exposure, we first aspirated the media volume to the loaded 25 μL syringe before we added the compound in methanol to the wells followed by a thorough and immediate mixing using a Hamilton syringe (5 quick aspiration and reinjection steps with the total syringe volume). In the case of dosing before sealing, silicone rubber (PDMS) mats were used as plate sealing in combination with aluminum foil (underneath the rubber mat) to reduce sorption to the PDMS (Figure 1B). In the case of dosing after sealing, we used a PTFE/silicone sealing (web seal kit with 96-well glass plates and preslit PTFE/silicone microplate mats from SUN SRI, Rockwood, Tennessee, USA) to reduce sorption to PDMS but without aluminum foil as the Hamilton syringe would disrupt it. Accordingly, we had to compromise between the optimal sealing (aluminum foil underneath the rubber mat) but with open dosing (potential loss of volatiles during dosing) and the PDMS sealing that might lead to sorption effects but allows a direct dosing without loss of volatiles. Because of the high volume, the headspace-free setup is prone to spillage, which should be avoided before sealing but cannot be avoided in all cases during plate seal removal. Therefore, it is important to use microplates with an empty space between the wells instead of a plane surface. Spilled medium can then be captured between the wells to avoid cross contamination after exposure, while the tight sealing avoids cross-contamination of different treatments and concentrations during exposure. Besides the headspace-free test design in conventional 96-well plastic microplates (Greiner, Orlando, Florida, USA; Figure 1B and Table 2) in an alternative setup, we used glass microplates with PTFE/silicone sealing (web seal kit with 96-well glass plates and preslit PTFE/silicone microplate mats from SUN SRI, Rockwood, Tennessee, USA) and direct dosing of test compounds through the sealing mat using a glass Hamilton syringe (Figure 1C and Table 2). The use of glass aimed at minimized sorption of hydrophobic compounds to the plate material,31 and direct injection in each well reduced volatilization during the sample preparation and dosing. However, the glass setup is not compatible with mammalian cell lines, which have to adhere to a plastic surface. The plates of the headspace-free setup of the mammalian cell assay in plastic plates with dosing after sealing were sealed with silicone rubber mats without aluminum because the PTFE/silicone sealing mats were not compatible with the plastic plates. Significance of differences between the effect concentrations of each setup was tested using one-way ANOVA with Tukey’s multi comparisons test when effect concentrations were available for both setups with and without headspace. Quantification of analytes after exposure to different test setups. The concentrations of the five most volatile test chemicals were quantified after incubation in different test setups without cells but with the appropriate medium: 1,1-dichloroethene, bromoethane, chloroform, dichloromethane and bromoform. The test compounds were quantified by Queensland Health Forensic and Scientific Services (Coopers Plains, Queensland, Australia) with a purge and trap GC/MS method.32 The analytical system comprised a Shimadzu QP2010 Plus gas chromatograph/mass spectrometer/data system (Shimadzu, Kyoto, Japan), Tekmar Velocity purge and trap concentrator (Teledyne Tekmar, Mason, Ohio, USA) and EST Centurion purge and trap auto sampler (EST Analytical, Fairfield, Ohio, USA). For analysis of our target compounds, the purge and trap

■

RESULTS AND DISCUSSION Vibrio f ischeri bioluminescence inhibition assay. As a first step, we compared the % inhibition of the reference compound phenol, calculated according to the ISO guideline, with the % inhibition calculated without a measurement of luminescence prior to incubation (It0) in ten independent test runs (see Supporting Information equations S1 and S3 in section S1). The derived concentration-effect curves for phenol were almost identical (Supporting Information, Figure S1) and the calculated EC50 values differed only by an average of ±2.5% (from −3.1% to +6.1%), which confirms that It0 is very uniform across the plates and hence calculating the % inhibition without It0 is acceptable (as long as noncolored samples are measured; Supporting Information, Section S1). The concentration-effect curve for the standard reference compound phenol was comparable between the setups with and without headspace in plastic microplates (Supporting Information; Figure S2; i.e., EC50 of 2.43 mM (CV: 0.23) in the conventional setup vs 2.27 mM (CV: 0.19) in the headspace-free setup). However, in the setup in glass microplates with direct dosing the EC50 was reduced by a factor of 1.7 (i.e., EC50 of 1.4 mM) but still fell within the variability of the long-term QA/QC record of this bioassay (Supporting Information, Figure S3) while resulting in a reduced variability (CV: 0.08). Due to the low hydrophobicity of phenol (log Kow: 1.5) it is unlikely that a reduced sorption to plate material resulted in the higher apparent sensitivity, however, we observed a comparable decrease of effect concentration with the umuC assay. A probable explanation 1608

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

Figure 2. Vibrio fischeri bioluminescence inhibition assay. Effect of setup with (w/) and without (w/o) headspace (HS) and plate material (plastic or glass) on 50% effect concentration EC50. (A) Effect of HS on EC50 in relation to the air−water partition coefficient Kaw; (B) effect of different setups compared to the modeled cytotoxicity (QSAR, black line) in relation to the liposome−water partition coefficients (log Klipw); P, reference compound phenol; and (C) % of nominal concentrations of BF, DCM, CF, BE, and DCE in bioassay medium after incubation in different setups. Bioassay results from 2 (w/o HS glass) to 3 (w/HS, w/o HS) independent experiments per compound are shown, each with 2 replicates per concentration; and chemical analysis results from two concentrations with two replicates each are shown. The error bars display the standard deviation.

halogenated compounds due to increased molecule size and thus exceeding the 30 min incubation time. Without headspace, the nominal EC50 of five DBPs (1,1dichloroethene, bromoethane, chloroform, dichloromethane, bromoform) converged to the QSAR-modeled EC50 confirming a nonspecific mode of toxic action for these compounds and the suitability of the headspace-free setup to minimize the loss of volatile compounds (Figure 2B). For dichloromethane in the conventional setup (w/HS), no concentration-effect curve could be derived and accordingly no EC50 could be determined but effects measured in the HS-free setups fell close to the baseline toxicity QSAR (Figure 2B). The loss of volatile compounds in the setup with headspace was confirmed with chemical analysis (Figure 2C). The fractional loss of analyte concentrations did not differ significantly between the two doses applied (p > 0.05, one-way ANOVA with Tukey’s multi comparisons test; Supporting Information, Table S6) and were therefore plotted combined as % of nominal concentration. The concentration of the five most volatile test compounds ranged from 2−13% of the nominal concentration in the conventional setup after 30 min incubation (20 °C) while reaching 78−130% in the headspace-free setup (Figure 2C; Supporting Information, Table S6). Accordingly the loss of analyte is reduced by 89−98% (Supporting Information, Table S6). This significantly reduced loss of free aqueous concentrations (p < 0.01) corresponded with reduced EC50 values of 85−98% when comparing the headspace-free setup with the conventional setup (Supporting Information, Table S1). Trichloroacetonitrile, dichloroacetonitrile and iodoform were more toxic than predicted (Supporting Information, Figure S4). A measure of the specificity of the effect of a compound is the toxic ratio TR,36 which is defined as the quotient of the EC50 predicted with the baseline toxicity QSAR and the experimental EC50 (eq 2).

could be a decelerated mixing of test compound during direct dosing14 despite a thorough and immediate mixing after sample injection. This might have caused an uneven exposure for a couple of seconds until mixing is completed. However, the exposure concentration is not the biologically active dose19 and it is likely that uptake and or partitioning of the compound to the target compartment of the cells takes longer then the complete mixing.33 Furthermore, a pronounced effect of decelerated mixing reported by Tanneberger et al.34 was observed only when DMSO was used as solvent for the direct dosing to expose the cells attached to a microplate. Here it was assumed that the DMSO drop sank and spread across the cells leading to a transitory high peak exposure of the cells. With methanol, this effect was negligible and most likely not pronounced enough to explain the effect observed in this study. When testing the volatile test compounds in the V. f ischeri assay, the logarithm of the ratio of effect concentrations EC50 w/HS/ EC50 w/o HS plastic and EC50 w/HS/ EC50 w/o HS glass correlated significantly with the air−water partition coefficient (log Kaw; Spearman’s ρ > 0.89, p < 0.05; Figure 2A), however, only under exclusion of trichloroacetonitrile, which is discussed further below. Thus the more of a tendency a chemical has to escape to the gas phase, the more the toxicity is underestimated by the conventional bioassay setup with headspace. The comparison between the conventional setup and the headspace-free setup in plastic microplates showed a significant difference in EC50 (p < 0.05) by up to more than three orders of magnitude, demonstrating a higher apparent sensitivity in the HS-free setup (Figure 2B; Supporting Information, Figure S4 and Table S1). The setup in glass microplates with direct injection further decreased the nominal effect concentration of the volatile compounds by a factor of 1.6 to 5.1 (Supporting Information, Table S1). This is probably a result of a reduced loss of volatile test compound during the dosing. However, considering the lowered effect concentration of the reference compound also a high peak exposure of the bacteria directly after injecting the test compound could have contributed to this effect.14 A potential analyte loss via diffusion through the polystyrene microplates is not likely to explain the difference in effect concentrations as diffusion through polystyrene plates takes about 40 min for oxygen molecules35 and presumably longer for

TR =

EC50 QSAR EC50 w/o HS glass

(2)

The TR was 947 for trichloroacetonitrile, 404 for dichloroacetonitrile and 61 for iodoform, which suggests a specific or reactive mode of toxic action for these compounds (Supporting Information, Figure S4).18,36 Thus these compounds could not be used for the validation of the HS-free setup by means of the 1609

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

QSAR model but they showed the same trend of decreasing nominal EC50 concentrations without headspace (Supporting Information, Figure S4). The effect of the headspace-free setup on the toxicity of trichloroacetonitrile was unexpectedly high (EC50 ca. 1600× lower in the HS-free setup in plastic microplates and 3800× lower in glass microplates; Supporting Information, Figure S4 and Table S1), which cannot be explained by volatilization because of the relatively low Kaw. However, Glezer et al.37 demonstrated that haloacetonitriles are prone to rapid hydrolytic degradation. This quick degradation presumably increased the observed effect concentration in the conventional setup but cannot explain why this effect is less pronounced in the headspace-free setup. There was no correlation between the EC50 decrease caused by switching from plastic to glass plates and log Kow (Supporting Information, Figure S5), suggesting that sorption to plate material plays a minor role compared to volatilization. Furthermore, the plastic plates were conventionally dosed with a pipet and then sealed, while the glass plates were sealed and then dosed by injection through the silicone septum with a Hamilton syringe. Direct injection through the septum with a Hamilton syringe reduced the chance of volatilization as compared to the dosing with conventional pipettes, but the dosing is tediously slow, which results in dissimilar exposure time between

wells. The improvement by switching from plastic to glass is small (EC50 w/o HS plastic/EC50 w/o HS glass ranged from 1.6 to 5.1) compared to the effect of w/ versus w/o headspace (EC50 w/HS / EC50 w/o HS plastic ranged from 2.5 to 1600). Thus, a headspace-free setup in conventional plastic plates appears to be a pragmatic compromise between keeping the assay high-throughput and as sensitive as possible. UmuC Assay. The headspace-free setup (Table 2) met the validity criteria with regard to minimum growth in the negative control (minimum turbidity after the second exposure, 140 formazin attenuation units, FAU; w/headspace, 820 ± 29; w/o headspace, 621 ± 55), minimum induction ratio in the positive control (>2), and minimum growth factor (>0.5 for data evaluation).38 Analogous to the results of the Vibrio f ischeri assay, the headspace-free setup also increased the apparent sensitivity to cytotoxic effects in Salmonella typhimurium (Figure 3; Supporting Information, Figure S6 and Table S2). A slightly increased apparent sensitivity in the headspace-free setup in glass microplates was found for the reference compound 4-nitroquinoline-1-oxide (4NQO) compared to the conventional setup (decrease of the EC50 by a factor of 2; Supporting Information, Figure S6). Iodoform, which has a relatively low Kaw, was the only chemical that tested positive for genotoxicity in the conventional setup (Supporting Information, Table S2). The headspace-free setup in glass microplates enabled the detection of genotoxicity of four additional compounds and significantly reduced the nominal ECIR1.5 of iodoform (p < 0.001) by a factor of 5 (Table S2; Supporting Information, Figure S6). Figure 3 shows exemplarily the concentration−effect curves for bromoethane with and without headspace. Ames Fluctuation Assay. The Ames assay worked well under headspace-free conditions, and the standard reference compound nitrofurantoin exhibited a similar concentration− effect curve as the conventional setup (Figure 4A). The headspace-free setup met the validity criteria.27 No mutagenicity could be detected in the conventional setup with headspace for any of the selected volatile compounds (Supporting Information, Table S4). The setup without headspace in glass plates indicated mutagenic effects (significant dose-dependent increase of revertant wells compared to the negative control, p < 0.05) for bromoethane and dichloromethane (Figure 4B,C; Supporting Information, Table S3−S4) with ECIR1.5 of 1.6 × 10−03 M and 3.9 × 10−03 M (Table 3).

Figure 3. UmuC assay. Effect of headspace on genotoxicity (given as induction ratio IR) and growth factor GF of Salmonella typhimurium TA1535/pSK1002 (without metabolic activation) of BE. w/HS is shown as squares, and w/o HS as circles; the error bars display the standard deviation of 2 replicates.

Figure 4. Ames assay. Effect of headspace on mutagenicity (induction ratio). Concentration−effect curve of (A) the reference compound nitrofurantoin NF, (B) BE, and (C) DCM. NC, negative control; PC, positive control (0.25 μg NF/mL). The results from 2 independent experiments per compound, each with 3 replicates per concentration, are shown; the error bars display the standard deviation. 1610

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

Table 3. Effect Concentrations EC50 and ECIR1.5 in the Optimized Bioassay Setups without Headspacea

1,1-dichloroethene (DCE) bromoethane (BE) chloroform (CF) dichloromethane (DCM) bromoform (BF) iodoform (IF) dichloroacetonitrile (DCAN) trichloroacetonitrile (TCAN) a

bioluminescence inhibition EC50 (M)

genotoxicity ECIR1.5 (M)

mutagenicity ECIR1.5 (M)

oxidative stress ECIR1.5 (M)

w/o HS glass

w/o HS glass

w/o HS glass

w/o HS plastic

−03

4.4 × 10 9.7 × 10−03 2.3 × 10−03 2.2 × 10−02 5.4 × 10−04 4.3 × 10−06 6.4 × 10−05 1.3 × 10−06

−01b

n.e.≤1.2 × 10 6.9 × 10−03 n.e.≤5.8 × 10−02b n.e.≤1.4 × 10−01b 4.1 × 10−04 2.6 × 10−05 1.1 × 10−04 n.e.≤3.0 × 10−02b

−02b

n.e.≤2.5 × 10 1.6 × 10−03 n.e.≤8.2 × 10−02b 3.9 × 10−03 n.e.≤5.9 × 10−03b n.e.≤3.0 × 10−04b n.e.≤1.8 × 10−04b n.e.≤1.4 × 10−04b

n.e.≤6.3 × 10−02b 1.2 × 10−04 n.e.≤6.2 × 10−02b n.e.≤7.4 × 10−02b n.e.≤2.4 × 10−02b 3.3 × 10−05 8.1 × 10−06 1.8 × 10−05

See Table 2 for a description of the bioassay setups. bn.e., no effect up to the highest concentration tested.

AREc32 Assay for Induction of Oxidative Stress Response. The increased volume for the headspace-free setup did not change the effect concentration (ECIR1.5) for the reference compound tBHQ compared to the conventional setup with 10% FBS (Supporting Information, Figure S7). Decreasing the concentration of FBS (fetal bovine serum) was evaluated because chemicals can sorb to FBS, which reduces their bioavailability and thus presumably the sensitivity of this assay. Therefore, an attempt was made to keep the amount of FBS constant in the increased medium volume by proportionally decreasing the concentration. However, a decreased FBS concentration reduced the oxidative stress response considerably (Supporting Information, Figure S7), and therefore, all experiments were performed with the conventional culture medium containing 10% FBS. The diminished biological response with reduced FBS concentrations might be a result of the lack of serum constituents which are needed for the transcription of reporter gene expression. Bromoethane, iodoform, dichloroacetonitrile, and trichloroacetonitrile activated the oxidative stress response in AREc32 in both the headspace-free and the conventional setup. The linear concentration−effect (IR) relationship is exemplarily shown for bromoethane in Figure 5A. The full concentration−effect curves, including cell viability for each active compound, are given in the Supporting Information, Figure S8. The ECIR1.5 was significantly lower (p < 0.001) by up to a factor of 4.6 for bromoethane in the headspace-free setup as compared to the conventional

setup (Figure 5A). The ECIR1.5 also decreased significantly for iodoform (p < 0.05) by a factor of 2.7 and nonsignificantly and less pronounced for the less volatile dichloroacetonitrile and trichloroacetonitrile (factor of 2.4 to 2.6; Supporting Information, Table S5). Figure 5B displays the nominal effect concentrations (ECIR1.5) for all active compounds in relation to the air−water partition coefficient. Dosing before or after sealing had no considerable impact on the effect concentrations (Figure 5B). The nominal effect concentration decreased by less than one order of magnitude compared to three orders of magnitude in the bioluminescence inhibition assay when the headspace-free design was implemented, which is initially surprising considering the higher exposure temperature in the mammalian cell assay (37 °C) compared to the Microtox assay (20 °C), as enhanced volatilization can be expected with increasing temperature. However, the sorption of test chemicals to medium constituents might have a much higher impact on the apparent sensitivity as it reduces the freely dissolved concentration in the aqueous phase.30,39,40 Chemical analysis showed significantly reduced (p < 0.01) concentrations of the five most volatile test compounds by 72− 96% in the conventional setup compared to the headspace-free setup (Supporting Information, Table S6) confirming a reduced loss of volatile compounds. In the headspace-free setup, the concentrations ranged from 3−17% of the nominal concentration revealing a substantial concentration loss from the aqueous phase. The reduction pattern follows the trend of lipophilicity,

Figure 5. AREc32 reporter gene assay. Effect of headspace on oxidative stress induced by the selected DBPs. (A) Concentration−effect curve of BE. An induction ratio (IR) of 1.5 corresponds to the threshold of oxidative stress,29 indicated as a dotted line, and was used to derive the ECIR1.5. (B) Effect concentrations for different setups of BE, IF, DCAN, TCAN in relation to the air−water partition coefficients (Kaw). (C) Percent of nominal concentrations of the most volatile compounds BF, DCM, CF, BE, and DCE in bioassay medium after incubation in different setups. Bioassay results from 2 independent experiments per compound, each with 2 replicates per concentration (w/HS and w/o HS plastic) or from 1 experiment with 2 replicates (w/o HS plastic, septum) are shown. The linear regression lines are indicated in gray with the 95% confidence interval indicated with a dotted line. Chemical analysis results from two concentrations with two replicates each are shown; the error bars display the standard deviation. 1611

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

i.e., the higher the Kow (Table 1), the more pronounced is the loss of test compound with a rank order of BF > DCE > CF > BE > DCM (Figure 5C and Table S6, Supporting Information) indicating a loss of free test compound concentration via sorption processes and not via evaporation. As plastic plate material did not significantly alter the effect concentration in the Microtox assay (Figure 2B), we assume that the analyte concentration is primarily reduced via sorption to medium constituents like FBS.30,39,40 Toxicity Profiling of the Volatile Test Compounds with a Headspace-Free Bioassay Setup. The rank order in declining cytotoxicity measured by the Microtox assay was TCAN > IF > DCAN > BF > CF > DCE > BE > DCM (Table 3). The cytotoxic effect concentrations derived with the Microtox assay were one to three orders of magnitude lower compared to previously published effect concentrations derived by a mammalian cytotoxicity assay.41 From a toxicological perspective, the toxicity assessment with mammalian cells is much more relevant for human health risk assessment. However, a more sensitive and more rapid bioassay might be more desirable as a quick first tier screening approach to assess mixture effects of DBPs present in water samples.9 Genotoxic and mutagenic endpoints are among the most widely used toxicity measures to assess potential human health risks of DBPs. Five of the selected volatile compounds exhibited genotoxic/mutagenic effects in our test systems, while 1,1dichloroethene, chloroform, and trichloroacetonitrile were umuCand Ames-negative (Table 3). However, 1,1-dichloroethene was demonstrated to induce tumors in male mice after renal bioactivation,42 and trichloroacetonitrile led to genomic DNA damage in mammalian cells at a concentration of 1 mM41 and has been described as umu-positive in a previous publication.43 Presumably, in our test setup, the genotoxic potency of trichloroacetonitrile was masked by the high cytotoxicity as cell growth fell below the validity threshold of 0.5 already at 8 × 10−05 M (Supporting Information, Figure S6). Bromoform had been reported earlier as umuC-positive,44 which was confirmed by our findings. Bromoethane was reported as Ames positive in the TA100 strain in a previous study, which applied the agar-plate version of the Ames assay in a vapor exposure unit to avoid the loss of volatiles,11 and was shown to be carcinogenic in an inhalation-exposure study with mice.45 Dichloromethane was reported as weakly Ames positive in the TA100 strain with metabolic activation using a preincubation procedure to minimize the loss of volatile test chemicals46 and with or without S9 using a gas exposure unit.47 However, effect concentrations of these studies are hardly comparable due to the different dosing procedures. Bromoethane, iodoform, dichloroacetonitrile, and trichloroacetonitrile induced the oxidative stress response (Table 3). Dichloroacetonitrile was reported to induce oxidative stress previously as measured by the formation of reactive oxygen species.48 Nrf2 activates the transcription of genes encoding for proteins with antioxidant and detoxifying capacity in order to counteract the harmful effect of reactive oxygen species and environmental carcinogens.29 Accordingly, assessing oxidative stress via induction of the Nrf2 pathway is an important tool to assess the human health risk of environmental and drinking water contaminants.

latter is hampered by the test design (e.g., low media volume) or compound characteristics (e.g., high volatility). The four selected bacterial and mammalian cell-based bioassays that are indicative of biological endpoints relevant for disinfection by-product toxicity can easily be employed in a headspace-free setup to minimize the loss of volatile organic compounds while fulfilling validity criteria that are set for the conventional procedures. All four cell types (bacterial and mammalian cells) were not impaired during the incubation time by the lack of headspace, while the headspace-free setup significantly reduced the loss of volatile compounds by 72−98% compared to the conventional setup. Thus, if volatile or semivolatile compounds are tested or expected to be present in complex environmental mixtures, conventional setups of bacterial or mammalian cell-based assays should be adapted to a headspace-free version to avoid the loss of test chemical. For some cytotoxicity assays, the exposure time is much longer than that in the assays we used here, e.g., up to 72 h49 or 120 h.5 An elongated headspace-free exposure time might thwart the oxygen supply in cell-based bioassays as no additional oxygen can enter the media once the wells are sealed. Thus, if exposure duration was ever extended it must be checked prior to experiments if the oxygen solved in the media is sufficient for cell function and growth during the exposure time. Furthermore, if setups were changed (e.g., plate material or sealing) or other types of chemicals (e.g., hydrophobic chemicals) were tested, it is recommended to repeat the validation experiments performed here prior to application. In particular, more hydrophobic compounds could considerably sorb to the plastic and sealing material.11,30,31 This might be minimized by the use of glass microplates31 (not applicable for cells that adhere to plate material) and aluminum sealing with rubber mats on top to avoid adhesives. A simple adaptation of established bioassays might be also of interest for the regulatory community because an adaptation of conventional microplate-based bioassays to a headspace-free version enables the high throughput screening of volatile chemicals. However, a modified bioassay must be evaluated before application by the use of nonvolatile standard reference compounds as well as volatile compounds. A further point to consider is the increased workload as a result of test adaptation. This could be limited by a headspace-free setup in conventional plastic plates and dosing before sealing resulting in a feasible compromise between keeping the assay high-throughput and as sensitive as possible. However, an automated pipetting system might be the best option to resolve the workload problem.

■

ASSOCIATED CONTENT

* Supporting Information S

Details on the bioassays and effect concentrations of each bioassay setup. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +61 (0)7 3274 9009. Fax: +61-732749003. E-mail: d.stalter@ uq.edu.au.

■

Funding

CONCLUSIONS The combination of in vitro toxicity results with QSAR prediction of the target EC50 of test compounds might be a feasible alternative to chemical analysis for validation of test setups if the

This research was funded by Seqwater (bulk water supply provider in South East Queensland, Australia), the Australian Research Council (FT100100694), and the University of Queensland (Start-up Grant). 1612

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

Notes

(7) Richardson, S. D. (2003) Disinfection by-products and other emerging contaminants in drinking water. TrAC, Trends Anal. Chem. 22, 666−684. (8) Richardson, S. D., Thruston, A. D., Krasner, S. W., Weinberg, H. S., Miltner, R. J., Schenck, K. M., Narotsky, M. G., McKague, A. B., and Simmons, J. E. (2008) Integrated disinfection by-products mixtures research: Comprehensive characterization of water concentrates prepared from chlorinated and ozonated/postchlorinated drinking water. J. Toxicol. Environ. Health, Part A 71, 1165−1186. (9) Neale, P. A., Antony, A., Bartkow, M. E., Farré, M. J., Heitz, A., Kristiana, I., Tang, J. Y. M., and Escher, B. I. (2012) Bioanalytical assessment of the formation of disinfection by-products in a drinking water treatment plant. Environ. Sci. Technol. 46, 10317−10325. (10) Gawlowski, J., Gierczak, T., and Niedzielski, J. (1994) Identification of volatile organics in warsaw tap water. Chem. Anal. 39, 423−430. (11) Barber, E. D., Donish, W. H., and Mueller, K. R. (1981) A procedure for the quantitative measurement of the mutagenicity of volatile liquids in the Ames-Salmonella-microsome assay. Mutat. Res. 90, 31−48. (12) Heringa, M. B., Schreurs, R., Busser, F., Van Der Saag, P. T., Van Der Burg, B., and Hermens, J. L. M. (2004) Toward more useful in vitro toxicity data with measured free concentrations. Environ. Sci. Technol. 38, 6263−6270. (13) Riedl, J., and Altenburger, R. (2007) Physicochemical substance properties as indicators for unreliable exposure in microplate-based bioassays. Chemosphere 67, 2210−2220. (14) Tanneberger, K., Rico-Rico, A., Kramer, N. I., Busser, F. J. M., Hermens, J. L. M., and Schirmer, K. (2010) Effects of solvents and dosing procedure on chemical toxicity in cell-based in vitro assays. Environ. Sci. Technol. 44, 4775−4781. (15) Kramer, N. I., Busser, F. J. M., Oosterwijk, M. T. T., Schirmer, K., Escher, B. I., and Hermens, J. L. M. (2010) Development of a partitioncontrolled dosing system for cell assays. Chem. Res. Toxicol. 23, 1806− 1814. (16) Hughes, T. J., Simmons, D. M., Monteith, L. G., and Claxton, L. D. (1987) Vaporization technique to measure mutagnic activity of volatile organic-chemicals in the Ames Salmonella assay. Environ. Mutagen. 9, 421−441. (17) Westphal, G. A., Blaszkewicz, M., Leutbecher, M., Muller, A., Hallier, E., and Bolt, H. M. (1994) Bacterial mutagenicity of 2-chloro1,3-butadiene (chloroprene) caused by decomposition products. Arch. Toxicol. 68, 79−84. (18) Tang, J. Y. M., McCarty, S., Glenn, E., Neale, P. A., Warne, M. S. J., and Escher, B. I. (2013) Mixture effects of organic micropollutants present in water: Towards the development of effect-based water quality trigger values for baseline toxicity. Water Res. 47, 3300−3314. (19) Hermens, J. L. M., Heringa, M. B., and ter Laak, T. L. (2007) Bioavailability in dose and exposure assessment of organic contaminants in (eco)toxicology. J. Toxicol. Environ. Health, Part A 70, 727−730. (20) WHO (1991) Chlorinated Drinking-Water; Chlorination byProducts; Some Other Halogenated Compounds; Cobalt and Cobalt Compounds, in IARC Monographs, Vol. 52, pp 299−315, World Health Organization, Geneva, Switzerland, http://monographs.iarc.fr/ENG/ Monographs/vol52/mono52.pdf. (21) Farré, M., Day, S., Neale, P. A., Stalter, D., Tang, J. Y. M., and Escher, B. I. (2013) Bioanalytical and chemical assessment of the disinfection by-product formation potential: role of organic matter. Water Res. 47, 5409−5421. (22) Escher, B. I., Bramaz, N., Mueller, J. F., Quayle, P., Rutishauser, S., and Vermeirssen, E. L. M. (2008) Toxic equivalent concentrations (TEQs) for baseline toxicity and specific modes of action as a tool to improve interpretation of ecotoxicity testing of environmental samples. J. Environ. Monit. 10, 612−621. (23) Vaes, W. H. J., Ramos, E. U., Verhaar, H. J. M., and Hermens, J. L. M. (1998) Acute toxicity of nonpolar versus polar narcosis: Is there a difference? Environ. Toxicol. Chem. 17, 1380−1384. (24) Reifferscheid, G., Heil, J., Oda, Y., and Zahn, R. K. (1991) A microplate version of the Sos/Umu-test for rapid detection of

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The National Research Centre for Environmental Toxicology (Entox) is a joint venture of The University of Queensland and Queensland Health Forensic and Scientific Services (QHFSS). Eva Glenn, Shane McCarty, and Dr. Wasantha Wickramsinghe from Entox are thankfully acknowledged for experimental assistance and helpful discussions.

■

ABBREVIATIONS 4NQO, 4-nitroquinoline-1-oxide; ARE, antioxidant response element; BE, bromoethane; BF, bromoform; CF, chloroform; conc, concentration; CV, coefficient of variance; DBP, disinfection byproducts; DCAN, dichloroacetonitrile; DCE, 1,1dichloroethene; DCM, dichloromethane; DMSO, dimethyl sulfoxide; EC50, 50% effect concentration; FBS, fetal bovine serum; GC/MS, gas chromatograph/mass spectrometer; HS, headspace; IF, iodoform; IR, induction ratio; ISO, International Organization for Standardization; It0, luminescence prior to incubation; Kaw, air−water partition coefficient; KH, Henry coefficient; Klipw, liposome−water partition coefficient; Kow, octanol−water partition coefficient; nd-SPME, negligible depletion solid-phase microextraction; n.e., no effect; p*, saturation vapor pressure; PDMS, polydimethylsiloxane; PTFE, polytetrafluoroethylene; QA/QC, quality assurance, quality control; QSAR, quantitative structure−activity relationship; RLU, relative light units; tBHQ, t-butylhydroquinone; TCAN, trichloroacetonitrile; TR, toxic ratio; VOC, volatile organic compounds; w/HS, conventional bioassay setup with headspace in plastic microplates; w/o HS glass, bioassay setup without headspace in glass microplates; w/o HS plastic, bioassay setup without headspace in plastic microplates

■

REFERENCES

(1) Richardson, S. D., Plewa, M. J., Wagner, E. D., Schoeny, R., and DeMarini, D. M. (2007) Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: a review and roadmap for research. Mutat. Res., Rev. Mutat. Res. 636, 178− 242. (2) Claxton, L. D., Pegram, R., Schenck, K. M., Simmons, J. E., and Warren, S. H. (2008) Integrated disinfection by-products research: Salmonella mutagenicity of water concentrates disinfected by chlorination and ozonation/postchlorination. J. Toxicol. Environ. Health, Part A 71, 1187−1194. (3) Richardson, S. D., DeMarini, D. M., Kogevinas, M., Fernandez, P., Marco, E., Lourencetti, C., Balleste, C., Heederik, D., Meliefste, K., McKague, A. B., Marcos, R., Font-Ribera, L., Grimalt, J. O., and Villanueva, C. M. (2010) What’s in the pool? A comprehensive identification of disinfection by-products and assessment of mutagenicity of chlorinated and brominated swimming pool water. Environ. Health Perspect. 118, 1523−1530. (4) Kogevinas, M., Villanueva, C. M., Font-Ribera, L., Liviac, D., Bustamante, M., Espinoza, F., Nieuwenhuijsen, M. J., Espinosa, A., Fernandez, P., DeMarini, D. M., Grimalt, J. O., Grummt, T., and Marcos, R. (2010) Genotoxic effects in swimmers exposed to disinfection byproducts in indoor swimming pools. Environ. Health Perspect. 118, 1531−1537. (5) Stalter, D., Magdeburg, A., Wagner, M., and Oehlmann, J. (2011) Ozonation and activated carbon treatment of sewage effluents: Removal of endocrine activity and cytotoxicity. Water Res. 45, 1015−1024. (6) Watson, K., Shaw, G., Leusch, F. D. L., and Knight, N. L. (2012) Chlorine disinfection by-products in wastewater effluent: Bioassaybased assessment of toxicological impact. Water Res. 46, 6069−6083. 1613

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614

Chemical Research in Toxicology

Article

genotoxins and genotoxic potentials of environmental-samples. Mutat. Res. 253, 215−222. (25) Claxton, L. D., Umbuzeiro, G. D., and DeMarini, D. M. (2010) The Salmonella mutagenicity assay: The stethoscope of genetic toxicology for the 21st century. Environ. Health Perspect. 118, 1515− 1522. (26) Green, M. H. L., Bridges, B. A., Rogers, A. M., Horspool, G., Muriel, W. J., Bridges, J. W., and Fry, J. R. (1977) Mutagen screening by a simplified bacterial fluctuation test - use of microsomal preparations and whole liver-cells for metabolic activation. Mutat. Res. 48, 287−293. (27) Reifferscheid, G., Maes, H. M., Allner, B., Badurova, J., Belkin, S., Bluhm, K., Brauer, F., Bressling, J., Domeneghetti, S., Elad, T., FluckigerIsler, S., Grummt, H. J., Guertler, R., Hecht, A., Heringa, M. B., Hollert, H., Huber, S., Kramer, M., Magdeburg, A., Ratte, H. T., SauerbornKlobucar, R., Sokolowski, A., Soldan, P., Smital, T., Stalter, D., Venier, P., Ziemann, C., Zipperle, J., and Buchinger, S. (2012) International roundrobin study on the Ames fluctuation test. Environ. Mol. Mutagen. 53, 185−197. (28) Wang, X. J., Hayes, J. D., and Wolf, C. R. (2006) Generation of a stable antioxidant response element-driven reporter gene cell line and its use to show redox-dependent activation of Nrf2 by cancer chemotherapeutic agents. Cancer Res. 66, 10983−10994. (29) Escher, B. I., Dutt, M., Maylin, E., Tang, J. Y. M., Toze, S., Wolf, C. R., and Lang, M. (2012) Water quality assessment using the AREc32 reporter gene assay indicative of the oxidative stress response pathway. J. Environ. Monit. 14, 2877−85. (30) Kramer, N. I., Krismartina, M., Rico-Rico, A., Blaauboer, B. J., and Hermens, J. L. M. (2012) Quantifying processes determining the free concentration of phenanthrene in basal cytotoxicity assays. Chem. Res. Toxicol. 25, 436−445. (31) Miller, C. A. (1999) A human aryl hydrocarbon receptor signaling pathway constructed in yeast displays additive responses to ligand mixtures. Toxicol. Appl. Pharmacol. 160, 297−303. (32) Ketola, R. A., Virkki, V. T., Ojala, M., Komppa, V., and Kotiaho, T. (1997) Comparison of different methods for the determination of volatile organic compounds in water samples. Talanta 44, 373−382. (33) Vogs, C., Bandow, N., and Altenburger, R. (2013) Effect propagation in a toxicokinetic/toxicodynamic model explains delayed effects on the growth of unicellular green algae scenedesmus vauolatus. Environ. Toxicol. Chem. 32, 1161−1172. (34) Tanneberger, K., Knöbel, M., Busser, F. J. M., Sinnige, T. L., Hermens, J. L. M., and Schirmer, K. (2012) Predicting fish acute toxicity using a fish gill cell line-based toxicity assay. Environ. Sci. Technol. 47, 1110−1119. (35) Arain, S., Weiss, S., Heinzle, E., John, G. T., Krause, C., and Klimant, I. (2005) Gas sensing in microplates with optodes: Influence of oxygen exchange between sample, air, and plate material. Biotechnol. Bioeng. 90, 271−280. (36) Verhaar, H. J. M., Ramos, E. U., and Hermens, J. L. M. (1996) Classifying environmental pollutants. 2. Separation of class 1 (baseline toxicity) and class 2 (‘polar narcosis’) type compounds based on chemical descriptors. J. Chemometr. 10, 149−162. (37) Glezer, V., Harris, B., Tal, N., Iosefzon, B., and Lev, O. (1999) Hydrolysis of haloacetonitriles: Linear free energy relationship, kinetics and products. Water Res. 33, 1938−1948. (38) ISO (1999) Water Quality: Determination of the Genotoxicity of Water and Waste Water Using the Umu-Test, in International Standard No. 13829, ISO13829:2000, pp 1−19, International Organization for Standardization, Geneva, Switzerland. (39) Schirmer, K., Chan, A. G. J., Greenberg, B. M., Dixon, D. G., and Bols, N. C. (1997) Methodology for demonstrating and measuring the photocytotoxicity of fluoranthene to fish cells in culture. Toxicol. in Vitro 11, 107−119. (40) Seibert, H., Morchel, S., and Gulden, M. (2002) Factors influencing nominal effective concentrations of chemical compounds in vitro: medium protein concentration. Toxicol. in Vitro 16, 289−297. (41) Plewa, M. J., Wagner, E. D., Muellner, M. G., Hsu, K. M., and Richardson, S. D. (2008) Comparative Mammalian Cell Toxicity of NDBPs and C-DBPs, in Disinfection By-Products in Drinking Water:

Occurrence, Formation, Health Effects, and Control (Karanfil, T., Krasner, S. W., and Xie, Y., Eds.) Vol. 995, pp 36−50, American Chemical Society, Washington, DC. (42) Speerschneider, P., and Dekant, W. (1995) Renal tumorigenicity of 1,1-dichloroethene in mice - the role of male specific expression of cytochrome-P450 2E1 in the renal bioactivation of 1,1-dichloroethene. Toxicol. Appl. Pharmacol. 130, 48−56. (43) Kramer, M., Huebner, I., Roerden, O., and Schmidt, C. K. (2009) Haloacetonitriles − Another Important Group of Disinfection Byproducts in Swimming Pool Water, Swimming Pool & Spa International Conference 2009, London, Mar 2009, Pool Water Treatment Advisory Group: Norfolk, UK, 10.4, pp 1−12, http://www.pwtag.org/ researchdocs/Used%20Ref%20docs/63%20Paper%2010. 4%20Kramer%20et%20al.pdf. (44) Ono, Y., Somiya, I., and Kawamura, M. (1991) The evaluation of genotoxicity using DNA repairing test for chemicals produced in chlorination and ozonation processes. Water Sci. Technol. 23, 329−338. (45) Holder, J. W. (2008) Analysis of chloroethane toxicity and carcinogenicity including a comparison with bromoethane. Toxicol. Ind. Health 24, 655−675. (46) Kundu, B., Richardson, S. D., Granville, C. A., Shaughnessy, D. T., Hanley, N. M., Swartz, P. D., Richard, A. M., and DeMarini, D. M. (2004) Comparative mutagenicity of halomethanes and halonitromethanes in Salmonella TA100: structure-activity analysis and mutation spectra. Mutat. Res.-Fundam. Mol. Mech. Mutagen. 554, 335−350. (47) Jongen, W. M. F., Alink, G. M., and Koeman, J. H. (1978) Mutagenic effect of dichloromethane on Salmonella typhimurium. Mutat. Res. 56, 245−248. (48) Ahmed, A. E., Aronson, J., and Jacob, S. (2000) Induction of oxidative stress and TNF-alpha secretion by dichloroacetonitrile, a water disinfectant by-product, as possible mediators of apoptosis or necrosis in a murine macrophage cell line (RAW). Toxicol. in Vitro 14, 199−210. (49) Plewa, M. J., Wagner, E. D., Richardson, S. D., Thruston, A. D., Woo, Y. T., and McKague, A. B. (2004) Chemical and biological characterization of newly discovered lodoacid drinking water disinfection by-products. Environ. Sci. Technol. 38, 4713−4722.

1614

dx.doi.org/10.1021/tx400263h | Chem. Res. Toxicol. 2013, 26, 1605−1614