Mixture effects of drinking water disinfection by-products: implications for risk assessment

Daniel Stalter *ab, Elissa O'Malley a, Urs von Gunten bc and Beate I. Escher *ade
aQueensland Alliance for Environmental Health Sciences, The University of Queensland, Brisbane, Australia
bEawag, Swiss Federal Institute of Aquatic Science and Technology, Duebendorf, Switzerland
cSchool of Architecture, Civil and Environmental Engineering (ENAC), Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
dDepartment of Cell Toxicology, UFZ – Helmholtz Centre for Environmental Research, Leipzig, Germany. E-mail: beate.escher@ufz.de
eEnvironmental Toxicology, Center for Applied Geoscience, Eberhard Karls University, Tübingen, Germany

Received 7th November 2019 , Accepted 16th January 2020

First published on 17th January 2020


Abstract

Disinfection by-products (DBPs) in drinking water have been associated with increased cancer risk but single DBPs cannot explain epidemiological cancer occurrences. To test if combined effects of DBPs are plausible to explain epidemiological evidence for adverse health effects, we assessed if mixture effects of DBPs can be predicted using the concentration addition (CA) model. We prepared 12 mixtures of DBPs (trihalomethanes, halonitromethanes, haloacetonitriles, haloketones, haloacetic acids, chloral hydrate, haloacetamides, 3-chloro-4-(dichloromethyl)-5-hydroxy-5H-furan-2-one (MX)) in equipotent concentration ratios. We determined effect concentrations with three reporter gene bioassays (AREc32, ARE-bla, and p53-bla) based on human cell lines and one bacterial assay (Microtox). The experimental effect concentrations agreed well with the effect concentrations predicted with the CA model, which suggests that the CA model is applicable for reactive DBPs despite different molecular mechanisms because the reporter gene assays are only detecting one mechanism each. Modelling of mixture effects of DBPs in ratios detected in drinking water revealed that haloacetonitriles, haloketones, and mono-haloacetic acids contributed the most to the total effect indicating a higher health relevance of these DBP groups. In drinking water samples the sum of the detected DBPs explained <6% of effect in most cases. The CA model could be applied to prioritize DBPs for further risk assessments to potentially close the gap between toxicological cancer risk predictions and epidemiological findings.



Water impact

It is not the concentration of individual disinfection by-products (DBPs) in a drinking water sample that determines the risk but how they act together in mixtures. With designed and realistically reconstituted mixtures of DBPs, we demonstrated that the mixture concept of concentration addition applies for many in vitro bioassays, which is important for implementation of mixture risk assessment of DPBs.

1. Introduction

Despite its critical importance for public health,1 disinfection of drinking water has raised concerns because of the formation of disinfection by-products (DBPs).2,3 DBPs are formed by the reactions between disinfectants (commonly chlorine or chloramine) and natural organic matter (NOM) as well as inorganic precursors (e.g., bromide).2 Epidemiological studies suggested an increased risk of bladder cancer after life-long ingestion of chlorinated drinking water pointing toward adverse health effects of DBPs.4–6 Based on such epidemiological studies the US-EPA calculated that 2–17% of bladder cancer cases could be avoided if the exposure to DBPs were ceased.7 However, the causation of urinary bladder cancer and other diseases by DBP exposure has not been conclusively proven.6 Additionally, known DBPs cannot explain epidemiological risk estimates despite approximately 700 DBPs identified within the last decades.2,4,8–11 The majority of all identified DBPs is not yet quantifiable, let alone toxicologically characterized. Some of the known DBPs have been characterized by various in vitro bioassays12–14 and such tools have also been applied to evaluate drinking water quality.15

Given that more than 50% of total organic halogens are unknown,16,17 and that there is additionally a diverse set of unknown non-halogenated compounds formed during disinfection,18 it seems plausible that unknown DBPs contribute to adverse health effects.10 Unknown DBPs occur—most likely—at low concentrations and thus single DBPs would have to have an extreme potency if they were to close the gap between toxicological risk assessment estimates and epidemiologic estimates. Therefore, unknown and known DBPs with similar mechanisms of action or common adverse outcomes acting together as mixtures might cause the observed epidemiological evidence.

Our working hypothesis is that DBPs act in mixtures according to the model of concentration addition (CA), which is a widely used model to describe the effects of mixtures with the same mode of action.19,20 The model of CA was derived initially for receptor-mediated effects but was shown to be applicable also to apical effects provided the chemicals had similar modes of action.21,22 While strictly valid only for mixtures with components that have the same mode of action, empirical evidence points to CA being a realistic worst case even if chemicals of diverse modes of action act together and apical endpoints such as mortality or growth inhibition are assessed.23

Mixture studies with DBPs have been performed previously with somewhat conflicting results. Narotsky et al.24 demonstrated that five HAAs and four trihalomethanes (THMs) contributed to DBP-induced pregnancy loss in rats. However, a characterization of the type of interaction of the nine component chemicals was not possible. Hooth et al.25 evaluated mixtures of bromate, 3-chloro-4-(dichloromethyl)-5-hydroxy-5H-furan-2-one (MX), chloroform, and bromodichloromethane in a rat model for hereditary renal cancer. The authors found that the mixtures were not more carcinogenic than the most potent mixture constituent and concluded that application of the CA model may overestimate the carcinogenic effect of DBP mixtures. In contrast, Andrews et al.26 found that the CA model adequately predicted the observed developmental toxicity of three haloacetic acids (HAAs) in a rat whole embryo assay. In a binary mixture study of two HAAs (dichloroacetic acid and trichloroacetic acid) Hassoun et al.27 found additive or slightly greater than additive effects on oxidative stress induction in hepatic tissue of mice. Parvez et al.28 proposed a method to evaluate the contribution of unknown DBPs to mixtures for the endpoint puberty acquisition in rats by comparing whole mixtures of disinfected water with defined mixtures of regulated DBPs found in the whole mixture and concluded that the nine regulated DBPs could explain most of the effect.

Zhang et al.29 found synergistic effects of sodium chlorite on bromate-induced renal cell death in an in vitro bioassay using rat kidney cells with the most likely explanation that NaClO2 partially reversed the bromate induced and cell protective G2/M arrest. Greater than additive effects of a binary DBP mixture were also found with several in vitro assays.30 In contrast, Simmons et al.31 found an antagonistic departure from additivity for various mixtures of five HAAs in a Chinese Hamster Ovary cell chronic cytotoxicity assay. Simmons et al.32 demonstrated that the mixing ratio had a significant impact on the toxicity of mixtures of 10 HAA in the same assay.

These examples demonstrate uncertainties regarding mixture effects of DBPs. Conflicting outcomes could be a result of differing mixture designs and differing biological endpoints. It is critical to use equipotent concentrations for testing the hypothesis whether CA applies. In equipotent mixtures all components are mixed in concentration ratios of their effect concentrations, i.e., more potent chemicals have a lower concentration but equal contribution to the mixture effect as low-potency chemicals. Otherwise a combination of very abundant and/or very potent compounds may dominate the mixture effect.33 The use of equipotent concentrations is clearly stated only in one of the mentioned studies.26 More often the applied concentration ratios were equimolar or based on environmental concentrations.24,31,32

DBPs act via different molecular initiating events and cellular toxicity pathways34,35 but this knowledge on the toxicity pathways remains incomplete. Most DBPs trigger reactive mechanisms and converge into the same adaptive stress response pathways, in particular oxidative stress response.14 We applied reporter gene assays that target one mechanism only. Previous work has demonstrated that in reporter gene assays, CA is a suitable mixture model unless the specific mechanism targeted is masked by cytotoxicity.36,37

To test our hypothesis that DBPs act together in a concentration additive manner, we prepared mixtures with DBPs found in drinking water from three representative water treatment plants and additional DBPs, which are known to be particularly potent, such as mono-HAAs, haloacetamides (HAcAms) and MX (3-chloro-4-(dichloromethyl)-5-hydroxy-5H-furan-2-one).14 Mixtures were prepared in equipotent concentration ratios for the AREc32 and p53-bla assay.

All mixtures were exposed in three human cell-based bioassays for activation of oxidative stress response (AREc32, ARE-bla) and genotoxicity (p53-bla) and one bacterial assay on cytotoxicity (Microtox). We compared the full experimental concentration–effect curves from the bioassays with the calculated concentration–effect curves predicted with the CA model to evaluate also the dependence of mixture interaction on the effect level.

After confirming CA with equipotent mixtures, we calculated the effect contribution of individual DBPs in known DBP mixtures detected in various drinking water samples in literature38–43 to evaluate which detected DBPs dominate mixture effects. Finally, the mixture effects of known and detected chemicals were compared to the effect of the extract of the entire water sample measured previously in the same bioassays.43

2. Materials and methods

2.1 Chemicals

For the mixture experiments we selected 22 DBPs that were detected in drinking water from three water treatment plants (WTPs) of the greater Brisbane area, Australia.43,44 Additionally, we selected 10 HAcAms, MX, and three mono-HAAs because of their high toxicity.14 Prior to this study 50 DBPs, including the 36 of the present study, were toxicologically characterized with various cell-based bioassays,14 and these single-chemical effect data were used for the mixture effect predictions. Details about the tested chemicals, including abbreviations, supplier and purity are compiled in Table S1 of the ESI. Methanolic stock solutions of all DBPs and DBP mixtures were stored at −80 °C.

2.2 Bioassays

We applied four bioassays for this mixture study. The bacterial cytotoxicity assay using Aliivibrio fischeri (formerly termed Vibrio fischeri) bioluminescence inhibition was selected as non-specific cytotoxicity screen because of its sensitivity to DBPs.41,42,45 The human MCF7 cell-based AREc32 assay targets the activation of the oxidative stress response pathway NRf2-ARE.36 The activation of the Nrf2-ARE stress response has been demonstrated in previous studies to be an important adaptive stress response pathway of mono-HAAs35,46 and appears to play a central role for the toxicity of many more DBPs.14,47 The ARE-bla assay48 was employed in addition to the AREc32 assay to detect oxidative stress response because it is based on a different cell line (HepG2 liver cells), which might reveal cell-specific differences in the response. Additionally, different reporter gene constructs may lead to a different responsiveness of assays depending on promoter/enhancer construction, ARE orientation and other factors.49 The p53-bla assay,45,50 derived from HCT-116 human colon carcinoma cells, was applied because activation of p53 has been discussed as marker for genotoxic properties of chemicals51 and because many DBPs activate the p53 adaptive stress response.14 For all mixtures the bioassays were performed free of headspace52 or with reduced headspace for the ARE-bla and p53-bla to reduce the loss of volatile DBPs.14 Each mixture was analysed in 8-step 2-fold dilutions in two to four independent experiments to derive full concentration–response curves. All replicates were evaluated together. If two independent repeats aligned closely, they were not further repeated, if there were differences up to four experiments were performed. The concentrations of the stock solutions were >300 times higher than the EC-values and complete concentration–effect curves with constant concentration ratios were derived. We used methanol as solvent at a concentration of 1% in the bioassays because it showed the lowest effect in the AREc32 assay compared to DMSO, ethanol or MTBE.53 The methanol control did not exhibit effects different from the medium control. The bioassays were performed according to the same protocols as in the previous study on single chemicals and the effect concentrations of the chemicals included in this study are reprinted in the ESI, Table S2.14

For the Microtox assay the assessment endpoint was the effect concentration (EC) that caused 50% inhibition of bioluminescence (EC50). The EC50 was derived from a log-logistic concentration–effect curve as described in Escher et al.54 The concentrations of the mixtures were expressed as the sum of molar concentrations of all components. For the oxidative stress response, p53 pathway activation and activation of the SOS response we used the induction ratio (IR) as the measure of effect. The IR is defined as the ratio of effect of the sample divided by the average effect observed in the solvent control (medium with 1% methanol). The effect concentration ECIR1.5 that elicits an IR of 1.5 (i.e., 50% effect increase compared to the negative control) was the assessment endpoint for these assays, calculated by use of linear regression of all experimental data points up to IR 4 with a fixed intercept at IR 1, which is the IR of the negative controls.55 The standard error σECIR1.5 was derived by error propagation as outlined by Escher et al.36,56

In addition to the four applied in the experimental mixture study, we also included the bacterial umuC assay57 in the modelling part of this study. The umuC assay detects the activation of the cellular SOS-response, a global response to DNA damage to induce DNA repair mechanisms, and hence indirectly detects genotoxicity.

2.3 Mixture design

Since the single DBPs exhibit differing relative effect potencies in the tested bioassays, an equipotent mixture in one bioassay may not be equipotent in another bioassay. Therefore, we designed the equipotent mixture according to the EC values in two bioassays (AREc32 and p53-bla) but measured each mixture in all bioassays.

For six mixtures, 3 to 24 DBPs were mixed in equipotent concentration ratios, where the ratios of concentration were according to the ratios of effect concentrations in the AREc32 assay (mix1AREc32 to mix6AREc32; Table S3). Additionally, six mixtures of 3 to 21 DBPs were mixed in equipotent concentration ratios derived from effect concentrations in the p53-bla assay (mix1p53-bla to mix6p53-bla; Table S4).

2.4 Mixture toxicity prediction

We compared the experimentally derived effect concentrations (ECmix,exp) with ECs predicted by use of the CA model.19,20,58 The ECmix,CA of the mixture predicted with the model of CA can be calculated with eqn (1) for i components present in the fraction pi (∑pi = 1) from the individual effect concentrations ECi of all mixture components i.
 
image file: c9ew00988d-t1.tif(1)
The error of the CA prediction (σECmix,CA) was propagated from experimental standard deviations of the effect concentration of each mixture component (σECi) assuming no error in the fractions pi (eqn (2)).
 
image file: c9ew00988d-t2.tif(2)
The ECmix,CA can be calculated for all effect levels to construct predicted concentration–response curves of the mixtures depicted in Fig. S1–S4.

We used the index on prediction quality (IPQ, eqn (3) and (4)) as a measure for the deviation between experimental (ECmix,exp) and predicted mixture effects (ECmix,CA).36,59

 
image file: c9ew00988d-t3.tif(3)
 
image file: c9ew00988d-t4.tif(4)
If the ratio between ECmix,exp and ECmix,CA is 1, then the IPQ is 0. A ratio of 2 results in an IPQ of 1 (if ECmix,CA is greater than ECmix,exp) or −1 (if ECmix,exp is greater than ECmix,CA), a ratio of 3 yields an IPQ of ±2, and so on.

2.5 Mixture predictions using literature data

We calculated bioanalytical equivalent concentrations (BEQ)45,60 to identify the mixture risk drivers among known chemicals and to compare bioanalytical results of water samples with predicted effects based on DBPs quantified in the extracts.

Dibromoacetonitrile (dBAN) was selected as common reference compound for all bioassays due to its high potency and because it was active in all bioassays.14 Hence, BEQs were expressed as dBAN equivalent concentrations (dBAN-EQ) in units of moldBAN L−1.

The dBAN-EQ of the DBPs chemically quantified in a sample (dBAN-EQchem) was calculated from the concentration Ci and the relative effect potency REPi (eqn (5)) of all previously detected DBPs i (eqn (6)).60 The REPi for all bioassays stem from the EC values derived in our previous study14 and are reprinted for convenience in the ESI, Table S5.

 
image file: c9ew00988d-t5.tif(5)
 
image file: c9ew00988d-t6.tif(6)
Eqn (6) is based on the assumption that CA is valid and that the relative effect potency is independent of the effect level. Accordingly, BEQs are considered a special case of CA where the log sigmoidal concentration–effect curves of all mixture components are assumed to be parallel or linear concentration–effect curves but apply for all effect levels in case of linear concentration–effect curves.56

The contribution of each mixture component i, BEQi, to the total effect (BEQchem) was calculated by eqn (7).

 
image file: c9ew00988d-t7.tif(7)
The measured dBAN equivalent concentration in a sample (dBAN-EQbio) was calculated by dividing the EC of the reference compound by the EC of the sample (eqn (8)).
 
image file: c9ew00988d-t8.tif(8)
The ratio between BEQchem and BEQbio (i.e., dBAN-EQchem and dBAN-EQbio) yields the fraction of effect that can be explained by the quantified DBPs.45

2.6 Effect contribution of DBPs in a known mixture

We calculated the dBAN-EQchem (eqn (6)) in 20 known DBP mixtures based on literature data of DBP concentrations from 16 different disinfected drinking water samples38–43 plus four hypothetical mixtures. Concentrations and literature source of data are listed in Table S5, ESI.

These mixtures had not been toxicologically profiled, we just use the analytical data to predict the mixture effect and the contribution of the components to the overall predicted mixture effect. The hypothetical mixtures were included because comprehensive DBP occurrence data are rare and most studies focus on a limited number of DBP groups. Thus, some highly toxic DBPs are often not included, such as iodinated DBPs, haloacetamides, mono-HAAs or MX, wherefore we included them in the hypothetical mixtures. Hypothetical mixture component concentrations were selected based on occurrence data of the respective compounds (Table S5).

2.7 Predicted effects versus measured effects in whole samples

We used results of water samples analysed in a previous study43 to calculate the fraction of effect of whole mixtures that stem from unknown or undetected DBPs. We had sampled drinking water from three different taps (TW 1–3, two sampling campaigns) and three water treatment plants (WTP1: chloramination, WTP2: chloramination, WTP3: chlorination) in the greater Brisbane area (Queensland, Australia).43 The distribution system is connected with all three treatment plants, and thus the tap waters could be mixtures originating from different drinking water treatment plants with differing disinfection methods.

We had used solid phase extraction to enrich the non-volatile DBP fraction.43 For samples TW1–3, we additionally enriched the volatile fraction with a purge and trap method43 before applying solid phase extraction. The extracts had been tested with the AREc32, p53-bla, umuC and Microtox assays and analysed for DBPs.43 For convenience the concentrations detected43 are reprinted in Table S6.

3. Results and discussion

3.1 Does concentration addition predict mixture effects of DBPs?

The mixtures were not equipotent with respect to the Microtox assay but the comparison between experimental and CA-predicted concentration–effect curves showed a good agreement (Fig. S1, ESI) especially at higher effect levels. The IPQ (eqn (3) and (4)) is a quantitative measure of the agreement between experiments and prediction model and ranged from −1.16 to 1.15 (Table S7, ESI). In the Microtox assay the IPQ showed a larger range than in the other bioassays (Fig. S5, ESI).

Agreement with CA is consistent with previous mixture studies on organic micropollutant in the Microtox assay,33 where most of the tested chemicals were classified as baseline toxicants. In contrast, most DBPs were classified as reactive toxicants in the Microtox assay.14 Concentration additive mixture effects of reactive soft electrophiles have been observed before in the Microtox assay.61,62 The endpoint of the Microtox assay is bioluminescence inhibition, which is related to energy depletion, e.g., caused by impaired supporting physiological pathways (for example the respiratory chain) or non-specifically via narcosis. Thus, bioluminescence inhibition can be regarded as integrative endpoint and many chemicals act as baseline toxicants in the Microtox assay.63 Mixtures of soft electrophiles have resulted in at least close to concentration additive effects in the Microtox assay.61 In a previous study, a cytotoxicity assay based on Escherichia coli growth inhibition was able to differentiate between soft and hard electrophiles. Within each group electrophiles acted according to CA but the model for independent action (IA) was valid between the groups.64

For the AREc32 assay on oxidative stress response 6 of the 12 mixtures were equipotent (mix1AREc32 to mix6AREc32). For all 12 mixtures, the experimental concentration–effect curves (Fig. S2) matched the CA prediction with −0.34 ≤ IPQ ≤ 0.84 (Fig. 1 and Table S7, ESI). For ARE-bla (Fig. S3) the agreement was only slightly lower with −0.83 ≤ IPQ ≤ 1.03 (Fig. 1 and Table S7, ESI). The AREc32 and ARE-bla assays are reporter gene assays and hence there is no effect observed for chemicals not triggering this particular response. Thus, CA in mixtures can be expected and has been observed previously for diverse micropollutants.36 We observed a similar level of agreement between modelled and experimental effects for the ARE-bla assay on oxidative stress with −0.83 ≥ IPQ ≤ 1.03 (Fig. 1 and Table S7, ESI).


image file: c9ew00988d-f1.tif
Fig. 1 Index on prediction quality (IPQ) for all equipotent mixtures in four bioassays (Table S7). A ratio of predicted (ECmix,CA) and experimental EC (ECmix,exp) of 1 resulted in an IPQ of 0, a ratio of 2 yields an IPQ of ±1.

For the p53-bla assay on adaptive stress response to genotoxicity (Fig. S4) 6 of the 12 mixtures were equipotent (mix1p53-bla to mix6p53-bla) and we observed a systematic deviation of the IPQ from zero for all mixtures (experimental potency was higher than predicted by CA (Table S7 and Fig. S5, ESI)). However, the IPQs for p53-bla were ≤0.62, and hence variations were relatively small (Fig. 1, Table S7 and Fig. S5).

Generally, discrepancies between CA-prediction and experimental data were small and, apart from p53-bla, not systematic as indicated by the distribution of IPQs. In 44 out of 48 comparisons between experimental and modelled ECs (i.e., 90%; Table S7), we found −1 < IPQ < 1 (Fig. 1) and the highest IPQ deviation from IPQ = 0 (i.e., perfect agreement with CA) was −1.16 (Table S7,Fig. 1). This is in agreement with a previous study where 303 mixture effect data where analysed from literature and 88% fell within −1 < IPQ < 1.65 A similar level of agreement was also found by Escher et al.36 and Tang et al.33 for mixtures of micropollutants. Thus, our results demonstrate that the CA model satisfactorily predicts the mixture effects of DBPs for adaptive stress responses and cytotoxicity despite differences in molecular initiating events triggering the toxicity pathway.

3.2 Identifying the risk drivers in a known mixture

Given the good agreement between the experimental mixture effects and the CA model, we can calculate the contribution of each mixture component to the total effect (eqn (6)). We selected literature data of DBP concentrations from 16 different samples38–43 and used four hypothetical mixtures that we considered to be representative for drinking water samples (Table S5, ESI) to calculate the contribution of each mixture component to the total effect of the detected DBPs (BEQchem). The hypothetical mixtures were used to include some highly toxic DBPs, such as iodinated DBPs, haloacetamides, mono-HAAs, or MX, which are often not included in DBP monitoring studies.

Fig. 2 shows the DBP concentrations (A) in comparison with the resulting BEQchem and the contribution of each included DBP for the AREc32 assay (B), ARE-bla (C), p53-bla (D), umuC (E), and Microtox assay (F). The concentrations are clearly dominated by trihalomethanes (THMs, Fig. 2A) but the THMs hardly contribute to the mixture effects (Fig. 2B–F). For the mammalian cell assays (Fig. 2B–D), the results suggest not only a minor effect contribution of THMs, but also of di- and tri-haloacetic acids, halonitromethanes (HNMs), choral hydrate, haloacetamides, and MX (i.e., % effect contribution <5%, except for bromochloroacetic acid BCAA). In contrast, haloacetonitriles (HANs), haloketones (HKs, AREc32 only), and mono-HAAs, if present in a sample, are expected to make a large contribution to the mixture effect even if present in low concentrations due to their high relative effect potency (Fig. 2B–D). HKs were not active in the ARE-bla assay and hence did not contribute to the total effect for oxidative stress response activation in this assay (Fig. 2C).


image file: c9ew00988d-f2.tif
Fig. 2 (A) DBP concentrations from 16 samples based on literature data and four hypothetical mixtures that include more potent DBPs, which are often not quantified in studies (see Table S5, ESI, for numerical values of the concentrations). (B–F) Contribution of all DBPs present in a sample to the bioanalytical equivalent concentrations (dBAN-EQchem, units of moldBAN L−1) based on DBP concentrations and the relative effect potencies in the respective bioassays. B and C: Adaptive stress response to oxidative stress in human cells (AREc32 and ARE-BLA). D: Adaptive stress response to genotoxicity in human cells (p53-BLA). E: Bacterial genotoxicity (umuC). F: Bacterial cytotoxicity (Microtox).

The bacterial assays delivered a very different pattern (Fig. 2E and F). Mono-HAAs did not contribute to the total effect in the umuC assay because cytotoxic effects masked genotoxic effects and hence we could not derive ECs for these compounds.14 Additionally, genotoxic effects in the umuC assay were largely dominated by MX (>80%), which is known to be more potent in bacterial assays than in mammalian cell-based assays.14,66 In the Microtox assay, effects were dominated by HNMs (<70%), whenever present in a sample (Fig. 2F), because HNMs have a high REP in this assay.14

It needs to be emphasized that the effect contribution of compounds, which require metabolic activation, such as N-nitrosamines, may be underestimated with the applied bioassays.14 Treating DBPs or DBP mixtures with rat liver S9 fractions prior to dosing into the bioassays could help to evaluate the role of metabolic activation for toxicity.

Because of the limited data base available compared to the wide range of DBPs identified in real water samples,8 these samples do not allow for a comprehensive risk comparison but serve as blueprint on how to use DBP occurrence data and the CA model to assess which DBPs are most relevant in a mixture of known DBPs. Our approach is similar to the TIC-Tox approach proposed recently by Plewa et al.67 They also demonstrated that the THM, which dominate the concentrations (expressed as peak area of the chromatogram, total ion current TIC), are no drivers of the mixture cytotoxicity on Chinese hamster ovary (CHO) cells but that haloacetonitriles and haloacetamides are dominating the mixture effect, just like in the present study evidenced for the mammalian reporter gene assays.

3.3 Effect-contribution of unknown DBPs in whole mixtures

As opposed to the approach based on mixtures with known components, whole mixture approaches reflect the real-world scenario because the major fraction of DBPs in drinking water is unknown. Calculating the % contribution of known components in a drinking water sample to the total effect expressed as bioanalytical equivalents enables one to estimate how relevant the known DBPs are compared to the unknown DBP fraction. It also allows one to estimate if newly discovered DBPs would be able to lessen the gap between observed effect in a sample and predicted mixture effect based on the known components.

The fraction of dBAN-EQ explained by the 34 DBPs included in the chemical analysis of which a maximum of 15 was detected (Table S6, ESI) was <6% in all real drinking water samples except for the two purge and trap extracts TW2-1P&T and TW3-1P&T where 29 and 92% could be explained (Fig. 3). This demonstrates that unknown or undetected DBPs must account for the major fraction of effects in the samples, in particular in the non-volatile fraction. This is in concordance with the fact that known DBPs cannot explain adverse health outcomes.4 Among the approximately 700 DBPs reported in literature only a small fraction is routinely monitored or has been toxicologically characterized.8 If all of these known DBPs were included in an effect and occurrence database the calculated contribution of the known DBPs to the total effect would most likely be significantly larger.


image file: c9ew00988d-f3.tif
Fig. 3 Percentage of bioanalytical equivalent concentration (dBAN-EQ) in four bioassays explained by known DBPs calculated with eqn (7) detected in solid-phase extracts (SPE) and purge-and-trap (P & T) extracts of three tap water samples (TW) and in three SPE extracts from three water treatment plants (WTP). Analytical data from Stalter et al.,43 reprinted in Table S6; REP from Table S5. Tap water had been sampled twice from each sampling point at different time points (e.g., TW1-1 and TW1-2: sampled from the same tap at two different time points). Samples from the first sampling campaign (TW1-1, TW2-1, TW3-1) were extracted with a purge and trap method (P & T) before SPE to capture the volatile DBP fraction and samples from the second sampling campaign only with SPE.

4. Conclusions

We found good agreement between the experimental effects for biological endpoints of oxidative stress response, genotoxicity and cytotoxicity and the effects predicted with the model of concentration addition. Our findings support the conclusion by Kortenkamp et al.20 that it is possible to predict the toxicity of multi-component mixtures with reasonable accuracy and precision and that deviations from CA are rare and relatively small. Accordingly, the concept of CA can be employed for the assessment of DBP mixtures despite the inclusion of various chemical classes with different molecular mechanisms because they are triggering a similar adverse outcome. This supports increasing evidence that toxicants can act together in an additive manner to induce a biological effect, despite initial steps of the adverse outcome pathway—including molecular initiating events and key events—may differ profoundly.20 Another potential issue is metabolism: many DBPs are only active after metabolic activation. There is still a research gap concerning the role of metabolism in in vitro assays.

Calculating the contribution of single compounds to BEQchem of a known mixture in drinking water enables the estimation of the toxicological relevance of a compound by accounting for the potency as well as concentrations. Therefore, the CA model could be a tool to prioritize DBPs for further risk assessments. Compounds with a high contribution to the total effect could be considered as high priority candidates for further toxicological characterization. A prerequisite would be the development of a comprehensive effect database derived from standardized bioassays of known DBPs while newly discovered compounds need to be toxicologically characterized and continuously added to the database. The difference between the results of the different reporter gene and bacterial assays demonstrate the importance to consider a set of different bioassays to capture the diversity of modes of action relevant for DBPs.

Due to the large fraction of unexplained effects in drinking water samples, further research should focus on the identification of toxicologically relevant DBPs to find compounds, which may explain the burden of disease reported in positive epidemiologic studies. Possible approaches include effect directed analysis (EDA) for a targeted identification of toxic compounds,68 which has not been applied for DBPs yet. This could allow targeted mitigation strategies during drinking water treatment to reduce potential human health hazards from DBPs.

One great challenge for future research on DBPs is to characterize all known DBPs with standardized test systems to establish an effect database to prioritize DBPs for further research to assess potential health effects. ToxCast and Tox21 have set precedence for high-throughput screening (HTS) of in vitro effects of micropollutants.69 Due to the volatility of many DBPs, HTS approaches need to be adapted to the challenge of evaporative loss from the test system.43,52

Another and possibly more pragmatic strategy would be to apply a battery of in vitro assays as monitoring tools. Bioanalytical monitoring would not replace but complement chemical analysis of prominent DBPs. The only requirement for their application would be the definition of effect-based trigger values that can differentiate between acceptable and poor water quality. Attempts have been made to derive such thresholds for micropollutant mixtures in recycled and drinking water for a wide range of bioassays33,36,55 and similar approaches could be used for DBPs.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was supported by a Marie Curie International Outgoing Fellowship within the 7th European Community Framework Program (PIOF-GA-2012-329169), by SEQwater (bulk water supply provider in South East Queensland, Australia) and by the Australian Research Council (FT100100694 and DP140102672). Dr. Janet Tang is acknowledged for helpful suggestions and comments.

References

  1. E. W. Akin, J. C. Hoff and E. C. Lippy, Waterborne outbreak control - which disinfectant, Environ. Health Perspect., 1982, 46, 7–12 CrossRef CAS PubMed.
  2. S. D. Richardson, M. J. Plewa, E. D. Wagner, R. Schoeny and D. M. DeMarini, 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., 2007, 636, 178–242 CrossRef CAS PubMed.
  3. D. L. Sedlak and U. von Gunten, The Chlorine Dilemma, Science, 2011, 331, 42–43 CrossRef CAS PubMed.
  4. S. E. Hrudey and J. Fawell, 40 years on: what do we know about drinking water disinfection by-products (DBPs) and human health?, Water Sci. Technol.: Water Supply, 2015, 15, 667–674 CAS.
  5. K. P. Cantor, C. M. Villanueva, D. T. Silverman, J. D. Figueroa, F. X. Real, M. Garcia-Closas, N. Malats, S. Chanock, M. Yeager, A. Tardon, R. Garcia-Closas, C. Serra, A. Carrato, G. Castano-Vinyals, C. Samanic, N. Rothman and M. Kogevinas, Polymorphisms in GSTT1, GSTZ1, and CYP2E1, disinfection by-products, and risk of bladder cancer in spain, Environ. Health Perspect., 2010, 118, 1545–1550 CrossRef CAS PubMed.
  6. S. E. Hrudey, Chlorination disinfection by-products, public health risk tradeoffs and me, Water Res., 2009, 43, 2057–2092 CrossRef CAS PubMed.
  7. R. Odom, S. Regli, M. Messner, J. Cromwell and M. Javdan, Benefit-cost analysis of the stage 1D/DBP rule, J. - Am. Water Works Assoc., 1999, 91, 137–147 CrossRef CAS.
  8. S. D. Richardson, in Disinfection by-products: formation and occurrence in drinking water in 'Encyclopedia of Environmental Health', ed. J. O. Nriagu, Elsevier Science Inc., Burlington, MA, USA, 2011, ch. 2, pp. 110–136 Search PubMed.
  9. P. M. Bradley, D. W. Kolpin, K. M. Romanok, K. L. Smalling, M. J. Focazio, J. B. Brown, M. C. Cardon, K. D. Carpenter, S. R. Corsi, L. A. DeCicco, J. E. Dietze, N. Evans, E. T. Furlong, C. E. Givens, J. L. Gray, D. W. Griffin, C. P. Higgins, M. L. Hladik, L. R. Iwanowicz, C. A. Journey, K. M. Kuivila, J. R. Masoner, C. A. McDonough, M. T. Meyer, J. L. Orlando, M. J. Strynar, C. P. Weis and V. S. Wilson, Reconnaissance of Mixed Organic and Inorganic Chemicals in Private and Public Supply Tapwaters at Selected Residential and Workplace Sites in the United States, Environ. Sci. Technol., 2018, 52, 13972–13985 CrossRef CAS PubMed.
  10. X. F. Li and W. A. Mitch, Drinking Water Disinfection Byproducts (DBPs) and Human Health Effects: Multidisciplinary Challenges and Opportunities, Environ. Sci. Technol., 2018, 52, 1681–1689 CrossRef CAS PubMed.
  11. S. Y. Kimura, A. A. Cuthbertson, J. D. Byer and S. D. Richardson, The DBP exposome: Development of a new method to simultaneously quantify priority disinfection by-products and comprehensively identify unknowns, Water Res., 2019, 148, 324–333 CrossRef CAS PubMed.
  12. E. M. Smith, M. J. Plewa, C. L. Lindell, S. D. Richardson and W. A. Mitch, Comparison of byproduct formation in waters treated with chlorine and iodine: Relevance to point-of-use treatment, Environ. Sci. Technol., 2010, 44, 8446–8452 CrossRef CAS PubMed.
  13. E. D. Wagner, J. Osiol, W. A. Mitch and M. J. Plewa, Comparative in Vitro Toxicity of Nitrosamines and Nitramines Associated with Amine-based Carbon Capture and Storage, Environ. Sci. Technol., 2014, 48, 8203–8211 CrossRef CAS PubMed.
  14. D. Stalter, E. O'Malley, U. von Gunten and B. I. Escher, Fingerprinting the reactive toxicity pathways of 50 drinking water disinfection by-products, Water Res., 2016, 91, 19–30 CrossRef CAS PubMed.
  15. P. A. Neale and B. I. Escher, In vitro bioassays to assess drinking water quality, Curr. Opin. Environ. Sci. Health, 2019, 7, 1–7 CrossRef.
  16. S. W. Krasner, H. S. Weinberg, S. D. Richardson, S. J. Pastor, R. Chinn, M. J. Sclimenti, G. D. Onstad and A. D. Thruston, Occurrence of a new generation of disinfection byproducts, Environ. Sci. Technol., 2006, 40, 7175–7185 CrossRef CAS PubMed.
  17. S. D. Richardson and C. Postigo, A new technique helps to uncover unknown peptides and disinfection by-products in water, J. Environ. Sci., 2016, 42, 6–8 CrossRef PubMed.
  18. R. Bull, G. Rice and L. Teuschler, Determinants of Whether or Not Mixtures of Disinfection By-Products are Similar, J. Toxicol. Environ. Health, Part A, 2009, 72, 437–460 CrossRef CAS PubMed.
  19. M. C. Berenbaum, The expected effect of a combination of agents - the general solution, J. Theor. Biol., 1985, 114, 413–431 CrossRef CAS PubMed.
  20. A. Kortenkamp, T. Backhaus and M. Faust, State of the art report on mixture toxicity, European Commission 070307/2007/485103/ETU/D.1: 2009, 2009 Search PubMed.
  21. R. Altenburger, T. Backhaus, W. Boedeker, M. Faust, M. Scholze and L. H. Grimme, Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri: mixtures composed of similarly acting chemicals, Environ. Toxicol. Chem., 2000, 19, 2341–2347 CrossRef CAS.
  22. R. Altenburger, T. Backhaus, W. Boedeker, M. Faust and M. Scholze, Simplifying complexity: Mixture toxicity assessment in the last 20 years, Environ. Toxicol. Chem., 2013, 32, 1685–1687 CrossRef CAS PubMed.
  23. T. Backhaus and M. Faust, Predictive Environmental Risk Assessment of Chemical Mixtures: A Conceptual Framework, Environ. Sci. Technol., 2012, 46, 2564–2573 CrossRef CAS PubMed.
  24. M. G. Narotsky, D. S. Best, A. McDonald, E. A. Godin, E. S. Hunter and J. E. Simmons, Pregnancy loss and eye malformations in offspring of F344 rats following gestational exposure to mixtures of regulated trihalomethanes and haloacetic acids, Reprod. Toxicol., 2011, 31, 59–65 CrossRef CAS PubMed.
  25. M. J. Hooth, K. S. McDorman, S. D. Hester, M. H. George, L. R. Brooks, A. E. Swank and D. C. Wolf, The carcinogenic response of Tsc2 mutant Long-Evans (Eker) rats to a mixture of drinking water disinfection by-products was less than additive, Toxicol. Sci., 2002, 69, 322–331 CrossRef CAS PubMed.
  26. J. E. Andrews, H. P. Nichols, J. E. Schmid, L. M. Mole, E. S. Hunter and G. R. Klinefelter, Developmental toxicity of mixtures: the water disinfection by-products dichloro-, dibromo- and bromochloro acetic acid in rat embryo culture, Reprod. Toxicol., 2004, 19, 111–116 CrossRef CAS PubMed.
  27. E. Hassoun, J. Cearfoss, S. Mamada, N. Al-Hassan, M. Brown, K. Heimberger and M. C. Liu, The Effects of Mixtures of Dichloroacetate and Trichloroacetate on Induction of Oxidative Stress in Livers of Mice After Subchronic Exposure, J. Toxicol. Environ. Health, Part A, 2014, 77, 313–323 CrossRef CAS PubMed.
  28. S. Parvez, G. E. Rice, L. K. Teuschler, J. E. Simmons, T. F. Speth, S. D. Richardson, R. J. Miltner, E. S. Hunter, III, J. G. Pressman, L. F. Strader, G. R. Klinefelter, J. M. Goldman and M. G. Narotsky, Method to assess component contribution to toxicity of complex mixtures: Assessment of puberty acquisition in rats exposed to disinfection byproducts, J. Environ. Sci., 2017, 58, 311–321 CrossRef PubMed.
  29. X. L. Zhang, R. J. Bull, J. Fisher, J. A. Cotruvo and B. S. Cummings, The synergistic effect of sodium chlorite and bromochloroacetic acid on BrO3-induced renal cell death, Toxicology, 2011, 289, 151–159 CrossRef CAS PubMed.
  30. S. Wang, D. J. Tian, W. W. Zheng, S. H. Jiang, X. Wang, M. E. Andersen, Y. X. Zheng, G. S. He and W. D. Qu, Combined Exposure to 3-Chloro-4-dichloromethy1-5-hydroxy-2(5H)-furanone and Microsytin-LR Increases Genotoxicity in Chinese Hamster Ovary Cells through Oxidative Stress, Environ. Sci. Technol., 2013, 47, 1678–1687 CAS.
  31. J. E. Simmons, C. Dingus, E. D. Wagner, L. K. Teuschler, G. E. Rice and M. J. Plewa, Environmentally realistic mixtures of the five regulated haloacetic acids exhibit concentration-dependent departures from dose additivity, Presented at 2009 Annual Society of Toxicology meeting, Baltimore, MD, March 15–19, 2009, http://cfpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=200345 Search PubMed.
  32. J. E. Simmons, E. D. Wagner and M. J. Plewa, Mixing ratio influences haloacetic acid (HAA) mixture toxicity, Presented at 45th Annual Society of Toxicology Meeting 2006, San Diego, CA, March 05–09, 2006, 2006, http://cfpub.epa.gov/si/si_public_record_Report.cfm?dirEntryId=140922&CFID=4046843&CFTOKEN=54348637&jsessionid=3830a08b386e94a61db91d35345c3b7f2a52 Search PubMed.
  33. J. Y. M. Tang, S. McCarty, E. Glenn, P. A. Neale, M. S. J. Warne and B. I. Escher, Mixture effects of organic micropollutants present in water: Towards the development of effect-based water quality trigger values for baseline toxicity, Water Res., 2013, 47, 3300–3314 CrossRef CAS PubMed.
  34. E. Prochazka, B. I. Escher, M. J. Plewa and F. D. L. Leusch, In Vitro Cytotoxicity and Adaptive Stress Responses to Selected Haloacetic Acid and Halobenzoquinone Water Disinfection Byproducts, Chem. Res. Toxicol., 2015, 28, 2059–2068 Search PubMed.
  35. J. Pals, M. S. Attene-Ramos, M. H. Xia, E. D. Wagner and M. J. Plewa, Human cell toxicogenomic analysis linking reactive oxygen species to the toxicity of monohaloacetic acid drinking water disinfection byproducts, Environ. Sci. Technol., 2013, 47, 12514–12523 CrossRef CAS PubMed.
  36. B. I. Escher, C. v. Daele, M. Dutt, J. Y. M. Tang and R. Altenburger, Most oxidative stress response in water samples comes from unknown chemicals: the need for effect-based water quality trigger values, Environ. Sci. Technol., 2013, 47, 7002–7011 CrossRef CAS PubMed.
  37. R. Altenburger, M. Scholze, W. Busch, B. I. Escher, G. Jakobs, M. Krauss, J. Krüger, P. A. Neale, S. Aït-Aïssa, A. C. Almeida, T.-B. Seiler, F. Brion, K. Hilscherová, H. Hollert, J. Novák, R. Schlichting, H. Serra, Y. Shao, A. J. Tindall, K. E. Tolefsen, G. de Aragão Umbuzeiro, T. D. Williams and A. Kortenkamp, Mixture effects in samples of multiple contaminants – an inter-laboratory study with manifold bioassays, Environ. Int., 2018, 114, 95–116 CrossRef CAS PubMed.
  38. R. J. Miltner, T. F. Speth, S. D. Richardson, S. W. Krasner, H. S. Weinberg and J. E. Simmons, Integrated disinfection by-products mixtures research: Disinfection of drinking waters by chlorination and ozonation/postchlorination treatment scenarios, J. Toxicol. Environ. Health, Part A, 2008, 71, 1133–1148 CrossRef CAS PubMed.
  39. J. R. Wei, B. X. Ye, W. Y. Wang, L. S. Yang, J. Tao and Z. Y. Hang, Spatial and temporal evaluations of disinfection by-products in drinking water distribution systems in Beijing, China, Sci. Total Environ., 2010, 408, 4600–4606 CrossRef CAS PubMed.
  40. H. S. Weinberg, S. W. Kranser, S. D. Richardson and A. D. Thruston, The Occurrence of Disinfection By-Products (DBPs) of Health Concern in Drinking Water: Results of a Nationwide DBP Occurrence Study, National Exposure Research Laboratory Office of Research and Development U.S. Environmental Protection Agency Athens, GA, 2002, EPA/600/R-02/068 Search PubMed.
  41. P. A. Neale, A. Antony, M. E. Bartkow, M. J. Farre, A. Heitz, I. Kristiana, J. Y. M. Tang and B. I. Escher, Bioanalytical assessment of the formation of disinfection byproducts in a drinking water treatment plant, Environ. Sci. Technol., 2012, 46, 10317–10325 CAS.
  42. M. J. Farré, S. Day, P. A. Neale, D. Stalter, J. Y. M. Tang and B. I. Escher, Bioanalytical and chemical assessment of the disinfection by-product formation potential: Role of organic matter, Water Res., 2013, 47, 5409–5421 CrossRef PubMed.
  43. D. Stalter, L. D. Peters, E. O'Malley, J. Y. M. Tang, M. Revalor, M. J. Farre, K. Watson, U. Von Gunten and B. I. Escher, Sample enrichment for bioanalytical quality assessment of disinfected drinking water: concentrating the polar, the volatiles, the unknowns, Environ. Sci. Technol., 2016, 50, 6495–6505 CrossRef CAS PubMed.
  44. G. A. De Vera, D. Stalter, W. Gernjak, J. Keller and M. Farré, Towards reducing DBP formation potential of drinking water by favouring direct ozone over hydroxyl radical reactions during ozonation, Water Res., 2015, 87, 49–58 CrossRef CAS PubMed.
  45. R. Y. L. Yeh, M. J. Farré, D. Stalter, J. Y. M. Tang, J. Molendijk and B. I. Escher, Bioanalytical and chemical evaluation of disinfection by-products in swimming pool water, Water Res., 2014, 59, 172–184 CrossRef CAS PubMed.
  46. M. S. Attene-Ramos, E. D. Wagner and M. J. Plewa, Comparative human cell toxicogenomic analysis of monohaloacetic acid drinking water disinfection byproducts, Environ. Sci. Technol., 2010, 44, 7206–7212 CrossRef CAS PubMed.
  47. M. J. Nieuwenhuijsen, J. Grellier, R. Smith, N. Iszatt, J. Bennett, N. Best and M. Toledano, The epidemiology and possible mechanisms of disinfection by-products in drinking water, Philos. Trans. R. Soc., A, 2009, 367, 4043–4076 CrossRef CAS PubMed.
  48. Life-Technologies, CellSensor™ ARE-bla Hep G2 cell-based assay protocol, Invitrogen/Lifetechnologies assay protocol K1208.pps, 2006, http://tools.lifetechnologies.com/content/sfs/manuals/cellsensor_AREblaHepG2_man.pdf Search PubMed.
  49. S. J. Shukla, R. L. Huang, S. O. Simmons, R. R. Tice, K. L. Witt, D. VanLeer, R. Ramabhadran, C. P. Austin and M. H. Xia, Profiling environmental chemicals for activity in the antioxidant response element signaling pathway using a high throughput screening approach, Environ. Health Perspect., 2012, 120, 1150–1156 CrossRef CAS PubMed.
  50. Life-Technologies, CellSensor™ p53RE-bla HCT-116 cell-based assay protocol, Invitrogen/Lifetechnologies assay protocol K1202.pps, 2010, http://tools.lifetechnologies.com/content/sfs/manuals/cellsensor_p53HCT116_man.pdf Search PubMed.
  51. P. J. Duerksen-Hughes, J. Yang and O. Ozcan, p53 induction as a genotoxic test for twenty-five chemicals undergoing in vivo carcinogenicity testing, Environ. Health Perspect., 1999, 107, 805–812 CrossRef CAS PubMed.
  52. D. Stalter, M. Dutt and B. Escher, Headspace-free setup of in vitro bioassays for the evaluation of volatile disinfection by-products, Chem. Res. Toxicol., 2013, 26, 1605–1614 Search PubMed.
  53. B. I. Escher, M. Dutt, E. Maylin, J. Y. M. Tang, S. Toze, C. R. Wolf and M. Lang, Water quality assessment using the AREc32 reporter gene assay indicative of the oxidative stress response pathway, J. Environ. Monit., 2012, 14, 2877–2885 RSC.
  54. B. I. Escher, N. Bramaz, J. F. Mueller, P. Quayle, S. Rutishauser and E. L. M. Vermeirssen, 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., 2008, 10, 612–621 RSC.
  55. B. I. Escher, M. Allinson, R. Altenburger, P. A. Bain, P. Balaguer, W. Busch, J. Crago, N. D. Denslow, E. Dopp, K. Hilscherova, A. R. Humpage, A. Kumar, M. Grimaldi, B. S. Jayasinghe, B. Jarosova, A. Jia, S. Makarov, K. A. Maruya, A. Medvedev, A. C. Mehinto, J. E. Mendez, A. Poulsen, E. Prochazka, J. Richard, A. Schifferli, D. Schlenk, S. Scholz, F. Shiraish, S. Snyder, G. Y. Su, J. Y. M. Tang, B. van der Burg, S. C. van der Linden, I. Werner, S. D. Westerheide, C. K. C. Wong, M. Yang, B. H. Y. Yeung, X. W. Zhang and F. D. L. Leusch, Benchmarking organic micropollutants in wastewater, recycled water and drinking water with in vitro bioassays, Environ. Sci. Technol., 2014, 48, 1940–1956 CrossRef CAS PubMed.
  56. B. Escher, P. A. Neale and D. Villeneuve, The advantages of linear concentration-response curves for in vitro bioassays with environmental samples, Environ. Toxicol. Chem., 2018, 37, 2273–2280 CrossRef CAS PubMed.
  57. G. Reifferscheid, J. Heil, Y. Oda and R. K. Zahn, A microplate version of the Sos/Umu-test for rapid detection of genotoxins and genotoxic potentials of environmental-samples, Mutat. Res., 1991, 253, 215–222 CAS.
  58. M. Faust, R. Altenburger, T. Backhaus, H. Blanck, W. Boedeker, P. Gramatica, V. Hamer, M. Scholze, M. Vighi and L. H. Grimme, Predicting the joint algal toxicity of multi-component s-triazine mixtures at low-effect concentrations of individual toxicants, Aquat. Toxicol., 2001, 56, 13–32 CrossRef CAS PubMed.
  59. R. Altenburger, W. Boedeker, M. Faust and L. H. Grimme, Regulations for combined effects of pollutants: Consequences from risk assessment in aquatic toxicology, Food Chem. Toxicol., 1996, 34, 1155–1157 CrossRef CAS PubMed.
  60. B. I. Escher and F. D. L. Leusch, Bioanalytical tools in water quality assessment, 2012 Search PubMed.
  61. D. A. Dawson, G. Poch and T. W. Schultz, Chemical mixture toxicity testing with Vibrio fischeri: Combined effects of binary mixtures for ten soft electrophiles, Ecotoxicol. Environ. Saf., 2006, 65, 171–180 CrossRef CAS PubMed.
  62. D. A. Dawson, J. Jeyaratnam, T. Mooneyham, G. Poch and T. W. Schultz, Mixture Toxicity of S(N)2-Reactive Soft Electrophiles: 1. Evaluation of Mixtures Containing alpha-Halogenated Acetonitriles, Arch. Environ. Contam. Toxicol., 2010, 59, 532–541 CrossRef CAS PubMed.
  63. B. Escher, A. Baumer, K. Bittermann, L. Henneberger, M. König, C. Kühnert and N. Klüver, General baseline toxicity QSAR for non-polar, polar and ionisable chemicals and their mixtures in the bioluminescence inhibition assay with Alliivibrio fischeri, Environ. Sci.: Processes Impacts, 2017, 19, 414–428 RSC.
  64. M. Richter and B. I. Escher, Mixture toxicity of reactive chemicals by using two bacterial growth assays as indicators of protein and DNA damage, Environ. Sci. Technol., 2005, 39, 8753–8761 CrossRef CAS PubMed.
  65. J. B. Belden, R. J. Gilliom and M. J. Lydy, How well can we predict the toxicity of pesticide mixtures to aquatic life?, Integr. Environ. Assess. Manage., 2007, 3, 364–372 CrossRef CAS.
  66. M. J. Plewa, Y. Kargalioglu, D. Vankerk, R. A. Minear and E. D. Wagner, Mammalian cell cytotoxicity and genotoxicity analysis of drinking water disinfection by-products, Environ. Mol. Mutagen., 2002, 40, 134–142 CrossRef CAS PubMed.
  67. M. J. Plewa, E. D. Wagner and S. D. Richardson, TIC-Tox: A preliminary discussion on identifying the forcing agents of DBP-mediated toxicity of disinfected water, J. Environ. Sci., 2017, 58, 208–216 CrossRef PubMed.
  68. W. Brack, Effect-directed analysis: a promising tool for the identification of organic toxicants in complex mixtures?, Anal. Bioanal. Chem., 2003, 377, 397–407 CrossRef CAS PubMed.
  69. A. M. Richard, R. S. Judson, K. A. Houck, C. M. Grulke, P. Volarath, I. Thillainadarajah, C. Yang, J. Rathman, M. T. Martin, J. F. Wambaugh, T. B. Knudsen, J. Kancherla, K. Mansouri, G. Patlewicz, A. J. Williams, S. B. Little, K. M. Crofton and R. S. Thomas, ToxCast Chemical Landscape: Paving the Road to 21st Century Toxicology, Chem. Res. Toxicol., 2016, 29, 1225–1251 Search PubMed.

Footnotes

Electronic supplementary information (ESI) available. See DOI: 10.1039/c9ew00988d
Now at: German Federal Institute for Risk Assessment, Department of Chemical and Product Safety, Max-Dohrn-Str. 8-10, 10589 Berlin, Germany.

This journal is © The Royal Society of Chemistry 2020