Developmental effects and genotoxicity of 10 water disinfection by-products in zebrafish

Disinfection by-products (DBPs) are contaminants formed as a consequence of chemical disinfection of public drinking waters. The disinfectants, such as chlorine, can react with natural organic matter in surface waters leading to the formation of a complex mixture of DBPs. So far, more than 600 DBPs have been identified and reported in the literature; nevertheless, they represent less than half of all possible environmental DBPs. The most prevalent DBPs include the four trihalomethanes (THMs), chloroform, bromoform, chlorodibromomethane (CDBM) and bromodichloromethane (BDCM), the group of haloacetic acids (HAAs) and bromate anion. DBPs in drinking water are generally present at sub-μg/L or low- to mid-μg/L levels (Richardson et al. 2007) and some countries have regulated the levels of some DBPs in drinking water (Table 1, US Environmental Protection Agency (EPA) 2006). Moreover, World Health Organization (WHO) guidelines exist as well as European Union (EU) DBP standards (Table 1, EU Directive 98/83/EC 1998; WHO 2004).

Several DBPs have been confirmed as mutagenic, genotoxic and/or carcinogenic in different test systems (Richardson et al. 2007). There is an increasing concern about the association of DBPs exposure with adverse developmental effects. Some individual DBPs have been found to cause developmental toxicity in mammalian assays at high doses (Ruddick et al. 1983; Narotsky et al. 1996; Christian et al. 2001a). Some DBPs induce specific congenital malformations of the cardiovascular and neurological system, but, in general, fetal body weight reduction is often reported as the major effect (Epstein et al. 1992; Hunter III et al. 1996). Recent studies had evaluated the developmental effects of complex mixtures and these appeared to exert no adverse developmental effects (Narotsky et al. 2012). The epidemiologic studies found inconsistent results or very weak associations for congenital anomalies/birth defects, central nervous system anomalies, neural tube defects and spontaneous abortion. However, these studies suggested a positive association with some measure of growth retardation (reviewed in Graves et al. 2001; Tardiff et al. 2006; Colman et al. 2011).

Chemical treatment of public water supplies is designed to kill pathogens that may exist in the drinking water, so the risk-benefit balance of water disinfection is considered positive. However some factors, such as the huge magnitude of the population affected, the distorted perception by the population of chemical risks, the availability of several alternative water treatment methodologies, the uncertainties in the hazard characterization of DBPs and the intrinsic limitations of the epidemiological studies, warrant the requirement of a better knowledge about DBPs toxicity and mechanisms of action (Colman et al. 2011). In this study, we have explored the capabilities of the zebrafish embryo model in order to identify and characterize the potential embryotoxic and genotoxic effects of some DBPs.

Zebrafish (Danio rerio) is a prominent model vertebrate in developmental genetics, toxicology and ecotoxicology (Postlethwait et al. 2000; Hill et al. 2005; Scholz et al. 2008). Size, easy husbandry, high fecundity and fast development represent the main benefits of using zebrafish over other vertebrate species. In addition, zebrafish embryos are transparent and develop outside the mother. Hence, morphological structures and internal organs can be easily visualized (Zhang et al. 2003). Moreover, the mechanisms of embryogenesis are well conserved along the vertebrates and there are several studies that confirm the ability of the zebrafish model to predict the teratogenic potential of chemicals in mammals (Brannen et al. 2010; Padilla et al. 2012; Selderslaghs et al. 2012). Hence, the zebrafish is increasingly used for assessing developmental toxicity of chemicals.

Currently, ‘whole-mixture’ approaches are being used to address concerns related to the potential adverse health effects of DBPs exposure. Scientists from the US EPA office of Research and Development proposed an experimental design for a multigenerational reproductive/developmental bioassay to optimize the probability of detecting adverse effects (Simmons et al. 2008). In the present study, the capability of the zebrafish embryo, as an alternative model, was explored in order to investigate the developmental effects and genotoxicity of DBPs. The selected DBPs – four THMs (CDBM and BDCM), five HAAs (dichloroacetic acid [DCA], trichloroacetic acid [TCA], dibromoacetic acid [DBA], tribromoacetic acid [TBA] and bromochloroacetic acid [BCA]) and sodium bromate – represented the most prevalent compounds and data were available for comparison with other test systems and models.

DBA, TCA, chloroform, bromoform and sodium bromate were purchased from Sigma-Adrich (St Louis, MO). DCA and TBA were delivered from Tokyo Chemical Industry (Tokyo, Japan). BCA, BDCM and CDBM were purchased from Alfa Aesar (Karlsruhe, Germany). Buffered embryo medium (17.4 mM NaCl; 0.23 mM KCl; 0.12 mM MgSO₄ · 7 H₂O; 0.18 mM Ca(NO₃)₂; 1.5 mM HEPES; pH 7.4) was used as the medium for all solutions during the experiments to keep the pH stable and constant between assays (Gustafson et al. 2012).

All stock solutions were prepared with buffered embryo medium except for THMs that were initially prepared in 100% dimethylsulfoxide (DMSO) and subsequently diluted in buffered embryo medium with a final DMSO concentration of 0.1% (v/v).

For all substances, a concentration range-finding experiment was conducted with a constant spacing factor 2. The range finding test allow us to select the final tested concentrations based on the presence of a 0 and 100% effect level (for both malformation and mortality). Each substance was tested in 5–7 concentrations with a negative control, test medium only or solvent control with 0.1% of DMSO. Exposure concentrations are anticipated to be stable throughout the test duration.

Zebrafish embryos were collected by natural spawning and staged according to Kimmel et al. (1995). Fertilization success was checked and only batches of eggs with a fertilization rate of at least 80% were used. Exposures of embryos began at 4 hours post-fertilization (hpf) and were incubated at 27 ± 1 °C on a 14-h light and 10-h dark cycle for 72 hours. The exposure was semi-static and solutions were renewed every 24 hours.

Embryos were exposed to HAAs and sodium bromate in a 6-well culture plate (Greiner Bio-one, Germany). Ten embryos were randomly distributed into wells and filled with 5 mL of each solution. Each 6-well plate held five different concentrations of the test compound and the negative or solvent control. In order to prevent losses by volatilization, THMs were tested in 20 mL hermetically sealed glass vials. Ten embryos per vial and treatment were exposed with 10 mL of each test solution. For each substance, three independent exposure experiments were conducted using eggs from independent spawning events (n = 3).

At 8, 28, 52 and 76 hpf, mortality of embryos was checked using a stereo microscope (SMZ-168, Motic). According to Nagel (2002), four endpoints were considered as indicators of mortality: coagulation of eggs, non-development of somites, non-detachment of the tail and no presence of heartbeat. Dead embryos were removed daily after assessment of mortality rate. The fraction of dead embryos at the end of the test was used to calculate median lethal concentration (LC₅₀) values.

The teratogenic effects were evaluated and recorded as described in Teixidó et al. (2013) (Table 2). The fraction of abnormal embryos was determined for each concentration and control group in order to calculate median effective concentration (EC₅₀) values. The frequency of teratogenic effects in all tested groups of a test substance was also analyzed. If the following criteria were fulfilled, (a) concentration-response relationship and (b) the endpoint is observed in ≥50% of all embryos showing malformations, the effect was considered to be a distinctive or identifying malformation (fingerprint endpoint) for this substance (Weigt et al. 2011). Embryos were observed using a stereo microscope, and images were obtained with a camera (Moticam 2000, Motic). Image processing was performed in Image J 1.41 (available at http://rsb.info.nih.gov/ij/) and Adobe Photoshop CS3 (Adobe Systems Inc., USA). Test concentrations are expressed in nominal concentrations.

From 48 hpf the embryos are able to hatch. Hatching success was recorded at 76 hpf as the percentage of embryos that hatched with respect to surviving embryos. The malformations observed in each embryo could affect its movement and consequently reduce or delay the hatching. Therefore, motility was also assessed by touch evoked response as a complementary endpoint.

Embryos that had not yet hatched were dechorionated. All embryos were anesthetized with buffered tricaine methanesulfonate (MMS) (0.5 mM, Sigma-Aldrich, St Louis, MO) and photographed (Moticam 2000, Motic) positioned on their lateral side in order to measure the distance between the anus and the posterior end of the notochord, defined as tail length (Bachmann 2002). The minimum concentration to inhibit growth (MCIG) is defined as the minimum concentration to significantly produce a decrease in tail length.

After exposure for 72 h, seven surviving fish embryos for the treatments and control groups were processed for cell isolation and comet assay. Cell isolation was carried out mechanically according to the protocol by Kosmehl et al. (2006). The EC₅₀ values from teratogenic effects were used as the test concentrations for genotoxicity testing.

The alkaline single cell gel electrophoresis (SCGE) or comet assay was performed as described by Singh et al. (1988) with some modifications. Fifty microliters of cell suspension were added to a tube containing 100 μL of 0.9% low melting point agarose (37 °C). The suspension was added to an agarose pre-coated slide and gently covered with a cover slide to make a micro gel. The gel was allowed to solidify for 10 min at room temperature and 6 min at −20 °C. Immersion in lysis buffer (2.5 M NaCl, 100 mM disodium EDTA and 10 mM Tris, pH 10) containing lauryl sarcosine 1% (v/v), Triton 1% (v/v) and DMSO 10% (v/v) in the dark was performed for 1.5 h at 4 °C. The slides were then placed in a horizontal gel electrophoresis unit, immersed in cold alkaline electrophoresis buffer (300 mM NaOH and 1 Mm Na₂EDTA, pH >13.5), and left in solution for 20 min at 4 °C. After electrophoresis in the same buffer at 25 V and 300 mA for 20 min, slides were neutralized by washing three times with 0.4 M Tris buffer at pH 7.5.

DNA was stained with 20 μL of DAPI solution (4′,6-diamidino-2-phenylindole) and immediately analyzed using a Nikon E600 fluorescence microscope. DNA damage as a percentage of DNA in the tail was measured using the Comet Assay IV software (Perceptive Instruments, Suffolk, UK). Zebrafish in vivo exposure to methyl MMS during 72 h (from 4 hpf to 76 hpf) was used as a positive control for comet assay.

Concentration-response curves for mortality and teratogenicity were plotted for compounds that showed a clear concentration-response relationship. It has been demonstrated that ignoring control mortality, even at lower than 10%, can lead to biased estimation of LC₅₀ (lethal effects) and EC₅₀ (teratogenic effects) (Hoekstra 1987). Therefore, data were corrected for control mortality with Abbott's formula: Pc = (P – Pi)/(100 – Pi) × 100 where Pc is the corrected percentage, P is the percentage mortality of the treated embryos and Pi is the percentage mortality of the control embryos (Abbott 1987).

Concentration-response curves were calculated using probit analysis (SPSS 15.0). Confidence intervals (CI) were set at 95%. Based on LC₅₀ and EC₅₀ values, a teratogenic index (TI) was calculated as the ratio LC₅₀/EC₅₀. In the case where no TI could be calculated, the compound was considered to be a non-teratogen although it could still be embryotoxic. Effective concentration 20 (EC₂₀) values were calculated for comparison between zebrafish embryo and human exposure data.

Statistical analysis was performed with SPSS 15.0. One-way analysis of variance followed by post hoc multi-comparison with the Bonferroni's test was used to analyze homogeneous data of the continuous variables. The Kruskal-Wallis test was used to analyze non-homogeneous data followed by Dunnett's post hoc test. Significance was accepted when p < 0.05.

Exposure to the THMs selected in this study resulted in adverse developmental effects in zebrafish. The cumulative concentration-response curves for lethality and teratogenesis are shown in Figure 1(a). The vehicle control (0.1% DMSO) was not toxic to embryos. The estimated LC₅₀, EC₅₀, EC₂₀ and TI values at 76 hpf are represented in Table 3. Malformations that occurred with the highest frequency included cardiac edema and tail developmental abnormalities (Table 4 and Figure 1(b)) and are in accordance with those reported by Brennan et al. (2005) for tadpoles. In our study, the EC₂₀ values range between 0.11 mM (23 mg/L) for CDBM and 0.7 mM (84.7 mg/L) for chloroform. Using the EC₂₀ to compare the chemicals a ranking of the compounds in order of decreasing potency can be established as: CDBM > Bromoform ≈ BDCM > Chloroform. CDBM (LC₅₀ = 0.48 mM, 100 mg/L and EC₅₀ = 0.16 mM, 33.3 mg/L) resulted to be three times more potent than chloroform (LC₅₀ = 2.4 mM, 286.5 mg/L and EC₅₀ = 0.84 mM, 100.3 mg/L). This order of toxicity was also found by Mattice et al. (1981) after exposure of common carp embryos to THMs. All THMs inhibited the growth of the embryo (Table 3), an effect that is often observed when there is a cardiovascular impairment (Billiard et al. 1999). Delayed hatching was also observed after THMs exposure and could be due to an inhibition of enzymes involved in hatching or a decreased mobility in the embryo (Von Westernhagen 1988). The developmental abnormalities and the concentration dependent sedative effect observed after THMs exposure partially or completely inhibit movement which may be required for proper hatching (Figure 2).

The EC₅₀ values reported in this study for chloroform (EC₅₀ = 0.85 mM) and BDCM (EC₅₀ = 0.26 mM) were similar to those reported for the Frog Embryo Teratogenesis Assay with Xenopus laevis in Brennan et al. (2005) (0.92 mM and 0.4 mM respectively). Based on the TI values (from 2.5 to 3.6), all THMs were found to produce teratogenic effects in zebrafish. Studies in mammalians suggest evidence of a fetotoxic response for THMs but not a teratogenic effect (Thompson et al. 1974; Ruddick et al. 1983). Other mammalian studies have reported a decreased fetal growth, delayed ossification and craniofacial abnormalities in rats treated by chloroform inhalation (Schwetz et al. 1974; Murray et al. 1979). Rats exposed to chloroform via drinking water have only showed to have an impaired postnatal growth (Lim et al. 2004). Brown-Woodman et al. (1998) reported a no effect concentration for chloroform of about 1.05 mM in rat whole embryo culture (WEC), similar to the EC₂₀ concentration reported in our study (EC₂₀ = 0.7 mM).

The only HAAs that induced some morphological effects were DCA and TBA at high concentrations (Figure 3(b) and Table 3). Embryos exposed to DCA exhibited moderate pericardial edema whereas no lethality was observed up to 46.5 mM (Figure 3(a) and (b), Table 4). The estimated LC₅₀, EC₅₀, EC₂₀ and TI-values are represented in Table 3. DCA were previously found to be developmental toxic in zebrafish by Hassoun et al. (2005). Craniofacial abnormalities (reduced mouth and jaw formation), skeletal muscle deformations and yolk sac and cardiac edema were observed at 144 hpf after an exposure of DCA in a concentration range between 8 and 32 mM. However, Weber et al. (2004) did not find any embryotoxic effect up to a concentration of DCA of about 124 mM in Xenopus laevis. DCA have been shown to produce cardiovascular defects when administered to pregnant dams at high dose levels (Epstein et al. 1992; Johnson et al. 1998).

In our study, TBA exposure caused curvatures of the spinal column, swelling of the pericardia and small eyes (Figure 3(b), Table 4) with an EC₂₀ of 4.4 mM (1,300 mg/L, Table 3). The current data are in agreement with results from Hunter III et al. (1996) who reported that brominated acetic acids are potentially more toxic than chlorinated species to the developing embryo. Developmental toxicity studies with frogs (Bantle et al. 1999) reported a TI of about 3.86 after TBA treatment, a higher value compared to our results with zebrafish embryos (TI = 2.2). In vivo, TBA exposure was found to not produce any adverse developmental effect (NTP (National Toxicology Program) 1998a). HAAs have been shown to produce neural tube defects, prosencephalic and pharyngeal arch hypoplasia, and heart and eye defects in rats and mice WEC (Hunter III et al. 1996; Andrews et al. 2004), but at a lower range concentration (0.05–2 mM) compared to zebrafish (10–30 mM).

BCA exposure did not produce any adverse developmental effects at the highest concentration tested (30 mM, Table 3). In BCA-exposed tadpoles, some gut abnormalities and decreased growth were observed only at higher concentrations (57.7 mM and 46 mM, respectively) (Brennan et al. 2005). BCA was found to not produce any adverse developmental effect in vivo in mammals (NTP 1998b). In our study, TCA exposure did not cause developmental effects (Table 3). In vivo, administered TCA to pregnant dams at high dose levels has been shown to produce cardiovascular defects and low weight (Smith et al. 1989; Johnson et al. 1998). Zebrafish embryo exposure to DBA only produced a decrease in tail length (Table 3). DBA was found to not produce any adverse developmental effect in vivo in mammals (Christian et al. 2001b).

Sodium bromate showed embryotoxic effects on zebrafish only at very high concentration (LC₅₀ = 65.4 mM, Table 3), probably by an irrelevant non-specific effect. Small eyes and pericardial edema were the predominant sodium bromate-induced malformations in zebrafish embryos (Figure 3(b) and Table 4). As shown in Table 3, the EC₂₀ value after exposure to sodium bromate was 51.7 mM. The 96-h LC₅₀ (8.32 mM) reported for frog embryos after sodium bromate treatment resulted to be about 8 times lower than the LC₅₀ reported for zebrafish embryos in this study (65.4 mM). Studies in mammals have shown that sodium bromate did not cause any adverse developmental effect (Wolf & Kaiser 1996).

DNA-damage caused by water DBPs was investigated using the comet assay. Genotoxicity could represent a potential mechanism leading to developmental disorders and embryo mortality, and also it has been associated with carcinogenic effects provoked by DBPs. The four THMs, the TBA and sodium bromate were selected for the comet assays as they exhibit embryo toxicity in zebrafish. The test concentrations used were the EC₅₀ values from teratogenic effects. Chloroform (0.85 mM) and CDBM (0.16 mM) produced significant DNA damage compared to the control solvent group (Figure 4(b)). If compared to the DNA damage produced by the positive control (Figure 4(a)), DBPs displayed a weak positive response. Exposure to the positive control (MMS) produced a concentration-dependent increase in the mean percentage of DNA in the tail (r = 0.89, p < 0.01; Figure 4(a)) and ranged between 4 and 37%. The highest tested concentration of MMS (25 mg/L) produced less than 10% embryo mortality, however all embryos showed developmental effects (data not shown). Many in vitro techniques have been used to investigate the mutagenic and genotoxic properties of THMs and HAAs (Richardson et al. 2007). These studies have shown that THMs are weak inducers of DNA damage (Landi et al. 2003; Zhang et al. 2012) and that glutathione S-transferase-theta (GSTT1-1) activity mediated transformation of brominated THMs to mutagenic intermediates (Pegram et al. 1997). Although the homolog GSTT1-1 gene for zebrafish has not yet been found, embryos possess a lower GST activity that could play a role not only in the weak mutagenicity observed but also in the low teratogenicity observed. HAAs caused DNA breaks in Chinese hamster ovary cells (Plewa et al. 2010) and in the human derived hepatoma cell line (Zhang et al. 2012). In our study, the strongest effect was found in sodium bromate-treated embryos with a median damage of 8% of DNA in the tail (Figure 4(c)). It has been reported that bromate induced DNA damage (SCGE assay) in mammalian cells through oxidative damage (Priestley et al. 2010).

Exposure to DBPs in humans has been quantitatively assessed by measuring the concentration of DBPs and its metabolites in blood, urine and exhaled breath after the oral intake of chlorinated water and after the dermal or inhalatory exposure during shower and bath. The reported blood levels in exposed humans are in the pg/mL range, with the highest levels of about 300 pg/mL for chloroform after a shower with a tap water surpassing the current standards (Nieuwenhuijsen et al. 2000; Nuckols et al. 2005). Transplacental crossing of some DBPs in blood at concentrations equal to or greater than in maternal blood has been demonstrated in rodents and humans (Dowty et al. 1976; Danielsson et al. 1986; Christian et al. 2001a). In this study, effect concentrations of 20% were between 0.11–0.7 mM (20–100 μg/mL) after DBPs exposure in zebrafish, several orders of magnitude above the levels to which human embryos would be exposed through their mothers' blood. It should be noted that it has yet not been demonstrated that plasma concentrations in mammals relate to toxicity effect concentrations similarly to the way exposure concentrations relate to toxicity in the fish embryo test.

One of the alleged weaknesses of the zebrafish embryo as a model for teratogenicity in mammals is the difference in metabolic activity towards exogenous substances. This is especially relevant in the case of xenobiotics whose toxicity is mediated by their metabolites. It is well known that some of the toxic effects of halogenated short-chain hydrocarbons are mediated by intermediate electrophilic metabolites. Zebrafish have a total of 94 CYP genes, distributed among 18 gene families, most of which are direct orthologs of human CYPs. Most of these CYPs are expressed in embryos during various time courses along the first 48 hours after fertilization. Indeed, some maternally-derived CYPs RNA transcripts are present in the unfertilized egg (Goldstone et al. 2010). Jones et al. (2010) have further demonstrated the expression of several xenobiotic metabolizing genes similar to human genes (CYP1A, CYP2B6, CYP3A5, UGT1A1) and their functional capacity metabolizing some model compounds during the early development. Therefore, the zebrafish embryo is endowed with a wide spectrum metabolic capacity, but the capability of metabolic transformation in comparison to mammalian models is still not fully understood (Hill et al. 2012). To our knowledge, there are no data about the capacity of the zebrafish embryo to metabolize THMs or HAAs. However, there is evidence of the capacity to bioactivate THMs by other fish species (Räbergh & Lipsky 1997; Vega-López et al. 2012).

Effect concentrations in zebrafish embryos support previous observation of a weak teratogenic and genotoxic potential of DBPs. The proximity of effect concentration of lethality and malformations suggest that probably the teratogenic effects are related to unspecific embryo toxicity. The effects are observed only at concentration levels well above those that can be attained in the fetal blood in humans exposed to chlorinated water, providing further evidence for only a weak teratogenic potential of DBP products. However, more studies are needed to explore the involvement of metabolism in the potential DBP toxicity and to extend our knowledge about exposure to mixtures and the possible developmental effects and genotoxicity. Finally, our study indicates that zebrafish embryos are as sensitive as other test systems and can be used as a potential screening and prioritization tool to assess a large number of DBPs.