Development of an oxidative stress in vitro assay in zebrafish (Danio rerio) cell lines

The nuclear factor erythroid 2-related factor 2 (Nrf2) is a key regulator of cellular defense against oxidative stress and correlated with classical toxicological endpoints. In vitro methods using fish cell lines for the assessment of aquatic toxicity are needed for mechanistic studies and as an alternative to in vivo. We describe an in vitro assay to study oxidative stress using zebrafish cell lines. Transfection efficiency of twelve commercially available transfection reagents were tested in the zebrafish cell lines ZFL, ZF4, and Pac2. The most efficient reagent for each cell line was selected for further experiments. Cells were transiently transfected with an Nrf2-responsive luciferase plasmid. The assay was tested using the oxidative stress inducing chemicals tertbutylhydroquinone, hydrogen peroxide, and sulforaphane. Of the transfected cell lines, ZF4 and ZFL showed higher sensitivity. The latter were used to study potential oxidative stress induced by pesticides (diazinon, deltamethrin, atrazine, metazachlor, terbutylazine, diuron). Besides known inducers, Nrf2 activity was also significantly induced by diazinon, deltametrin, diuron, and metazachlor. Activation of Nrf2 by metazachlor is a novel finding. The described assay could be a valuable tool for research in toxicology to study the stress response of both pure chemicals and environmental water samples.

and this type of approach is still in its infancy within aquatic toxicology. A requirement for the development of reporter gene assays, is the availability of functioning transfection methods. Most transfection reagents were developed for mammalian cell lines and are based on lipid-fusion mechanisms. However, the latter might be less efficient in fish cells due to lower incubation temperatures 20 . Furthermore, the transfection efficiency for a specific transfection regent often varies greatly between cell lines 21 . Thus, in the present investigation cells of different developmental stages and tissue origin were tested.
In this study, we have established a reporter gene assay for detecting oxidative stress by measuring induction of free Nrf2 in zebrafish cell lines Pac2, ZF4, and ZFL. Initially, we tested the transfection efficiency of twelve commercially available transfection reagents for these three cell lines. The established reporter gene assays were tested with potential Nrf2 inducers and six pesticides which are suspected to cause oxidative stress 22 . The established in vitro bioassay might be a useful tool in screening for potential inducers of oxidative stress, both regarding toxicity of pure compounds and for analysis of environmental samples, e.g. for ecotoxicology and environmental monitoring.
Cell viability testing. In order to determine cytotoxic concentrations of used compounds within the exposure range, cell viability was examined using an MTS-based [3-(4,5-dimethylthiazol-2-yl)-5-(3-carb oxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium] CellTiter 96 ® AQueous One Solution Cell Proliferation Assay (Promega, Madison, USA) in accordance with the manufacturer's protocol. Viability tests were conducted in parallel to the dual reporter gene assays, applying the same arrangements and exposure set-ups. Cells were seeded in transparent 96-well plates (Corning, New York, USA) and transient transfection was omitted. Simultaneously to transfected cells being lysed for scoring of luminescence signal, nutrition medium of cells being used in viability test was discharged and replaced by PBS containing 17% (v/v) MTS-reagent. After 2 h of incubation, formazan product turnover absorbance was measured at 490 nm using a Wallac Victor2 1420 microplate reader (PerkinElmer, Massachusetts, USA). Relative effects on cell viability were calculated in relation to the vehicle control.
Statistical analyses. Results from the dual luciferase assay and the viability assay were processed using R and GraphPad Prism 7 (GraphPad Software, La Jolla, USA). Graphs and illustrations were designed using GraphPad Prism 7. For both assays data of three or four experiments (experimental unit n = 3-4), each performed with quadruplicate samples for each concentration, were pooled and log-transformed to achieve homoscedasticity, giving a total population size for every exposure group of 10-16 (observational unit N = 10-16). Data was normalized against internal background within the dual reporter gene assay and against the negative control, giving fold induction as a final output. Significant differences between the control and the exposure groups were analyzed via a mixed-model two way-ANOVA followed by Dunnett's post-hoc test 28 . Thereby, transformed output data was considered as a fixed factor whereas experiments were considered as a random factorial factor within the model in order to account for inter-experimental variation. A p < 0.05 was considered statistically significant. Normality (quantile-quantile-plot) and homoscedasticity (residuals-vs.-fitted) of data sets were checked via graphical analysis of residuals.

Results
Transfection efficiency of various commercial reagents in zebrafish cell lines. Establishment of transgenic cell lines requires knowledge on transfection efficiency of different available transfection reagents for specific cell lines. Most commercially available transfection reagents are developed for transfection of mammalian cells. We have therefore tested the transfection efficiency of twelve commercially available transfection reagents in all three zebrafish cell lines (see supplementary data). The cells were transfected with a Renilla luciferase plasmid in different reagent to DNA mass ratios, according to the manufacturer's instructions. The total Renilla luminescence was analyzed as a measurement of transfection efficiency. Additionally, increasing cell densities were tested in the best detected DNA to transfection reagent ratios (data not shown). The following transfection parameters were found to result in the highest transfection efficiency for Pac2, ZF4 and ZFL, respectively: Pac2, plating density of 10 4 cells/well and a 4 µg FHD to 1 µg DNA ratio (Fig. S1). ZF4, plating density of 1.25*10 4 cells/well and a 2 µg FHD to 1 µg DNA ratio (Fig. S2). ZFL, plating density of 2.5*10 4 cells/well and a 2 µg XHP to 1 µg DNA ratio Scientific RepORtS | (2018) 8:12380 | DOI:10.1038/s41598-018-30880-1 or jetPRIME in a 3 to 1 ratio (Fig. S3). Finally, FHD for the Pac2 and ZF4 and XHP for the ZFL cell lines were chosen for further experiments .
Activation of Nrf2 and cell viability in zebrafish cell lines. All three zebrafish cell lines were co-transfected with an Nrf2-responsive Firefly-luciferase plasmid and a normalizing Renilla-luciferase plasmid. In order to quantify cell line specific responses, cells were exposed to known or postulated Nrf2 inducers tBHQ, SFN, and H 2 O 2 in concentrations of 0.1 to 100 µM for 24 h. DMSO was used as a cytotoxic control. In parallel to detecting Nrf2-activation via the dual reporter gene assays, cell viability was measured within the same experimental set-up using the MTS-assay.
In the Pac2 cell line (Fig. 1) there was no statistically significant increase in Nrf2 activity by any of the tested compounds. Statistically significant decrease in activity was observed for SFN (Fig. 1B) at a concentration of 100 µM and for DMSO (Fig. 1D) at 7% and 10%. Statistically significantly reduced cell viability was observed at 10 and 100 µM for SFN and at 7% and 10% for DMSO.
Activation of Nrf2 by pesticides. The cell lines ZF4 and ZFL showed a higher sensitivity to the tested positive controls (section 3.2), as compared to the Pac2 cell line. Therefore, the latter was not further tested since in addition fibroblast-like cells were already covered by ZF4. To test the applicability of the assay, ZF4 and ZFL cells transfected with the Nrf2 sensitive luciferase plasmid were used to investigate the potential oxidative stress potency of six pesticides.

Discussion
The Keap1-Nrf2-ARE signaling pathway is a major balancer of oxidative homeostasis, given its role in oxidative stress response but also physiological regulation 29 . Oxidative stress may lead to toxicity e.g. via enzyme deactivation, lipid peroxidation, and DNA damage. These effects may lead to tissue damage, apoptosis or cell necrosis 1,30 and further to developmental toxicity 31,32 and cancer 33 . So far, in vivo methods for determination of Nrf2 activity have been established in zebrafish strains [34][35][36] , in vitro methods using mammalian cells are described [37][38][39][40] , and standardized high-throughput assays in human cell lines 41 have been developed. Here, we report the development of a reporter gene assay for oxidative stress response in zebrafish cell lines.
Transfection efficiency. In this study, we have tested the transfection efficiency of twelve commercially available transfection reagents in three different cell lines and found a high variability in efficiency between the transfection reagents. We found that FuGENE HD (Promega) showed the highest transfection efficiency in both Pac2 and ZF4, while X-tremeGENE HP (Roche) and jetPRIME (Polyplus) showed the highest transfection efficiency in ZFL. These are important findings for future efforts to establish in vitro assays using transfected zebrafish cell lines.  is the primary location of detoxification and show highest abundancy in critical phase II enzymes, such as GSTs 42 . In order to test single cell line feasibility, each transfected cell line was first exposed to postulated Nrf2 inducers. The results showed that the Nrf2 activity following exposure varied between cell lines. The Keap1-Nrf2-ARE signaling pathway is not functional immediately after fertilization and the responsiveness to oxidative stress seems to vary during embryogenesis and larval development 32,43,44 . Within 24 hpf, the response to oxidative stress or Nrf2 inducers is quite low, increases up to 48 hpf, exhibits high variability during hatching and post-hatching, and finally stabilizes at around 96 hpf. This could be an explanation for the observed differences in sensitivity between the three investigated cell lines, as they are derived from zebrafish in different developmental stages. In addition, an impact of total serum concentration in the culture medium and therefore less available, non-protein bound compound concentration could be an alternative explanation for the differences in response between these zebrafish cell lines. Pac2 cells are cultured at a higher serum concentration (15%) than ZF4 and ZFL cells (10 and 5% respectively), which might explain the observed differences in sensitivity. Since Nrf2 is ubiquitously expressed in most cell types and acts via a common mechanism, its activity is generally only assayed in a single cell type when using bioassays based on mammalian cells [45][46][47] . However, our result show that cell line selection is crucial for this assay and further research is needed to fully explain the observed differences in oxidative stress sensitivity in zebrafish cell lines.
H 2 O 2 does not activate Nrf2 in zebrafish cell lines. Surprisingly, we were unable to show an Nrf2 induction by H 2 O 2 in any of the cell lines studied, although H 2 O 2 is normally metabolized by GSTs which are Nrf2 regulated. However, Nrf2-recruitment failure by H 2 O 2 has previously been described during development in embryos 35,48,49 and in larvae 36 . The phenomenon is either interpreted as a systemic breakdown of the Nrf2-Keap1 axis 48 or absence of an intermediate "posthatch factor" which is induced by H 2 O 2 and then modulates the Nrf2 response 35 . On the contrary, Nrf2 induction by H 2 O 2 was reported in mammalian cell lines 50 . Species differences, methodological aspects, since H 2 O 2 is a very reactive compound with short half-life, and interactions with media components may partly explain why we did not see any Nrf2 induction by H 2 O 2.
Application of an Nrf2 responsive reporter gene assay. To test the potential application of the established assay, six pesticides suspected to cause oxidative stress 22 were analyzed. We found that four out of the six analyzed compounds significantly increased the Nrf2 activity in at least one cell line. The dose-effect relationship for the compounds differed, where some compounds showed a clear dose-dependent increase in the Nrf2 activity, while other compounds showed a threshold-like effect with an increased Nrf2 activity only observed within the highest exposure concentration. However, those compounds would need to be tested at higher, non-toxic concentrations in order to exclude dose-dependency. These different patterns may be explained by differing mechanisms by which the compounds are inducing the Nrf2 release from Keap1 and thereby trigger the oxidant stress response. Such a mechanism-related model has been proposed by Kobayashi, Yamamoto, and co-workers 9,35,51,52 .
It has been reported that deltamethrin is causing alterations in antioxidant enzymes (glutathione S-transferase (GST), superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GPx), and glutathione reductase (GR)), reduced glutathione concentration levels (GSH), and lipid peroxidation levels (LPO) in Prussian carp (Carassius gibelio) 53,54 , spotted snakehead (Channa punctata) 55,56 , carp (Cyprinus carpio) 57 , and bushynose (Ancistrus multispinis) 58 , whereas other species did not show a similar response to deltamethrin 59 . We report an effect in ZF4, but no effects were observed in ZFL. Deltamethrin is primarily metabolized by phase I enzymes CYP1A1, CYP1A2, and carboxylesterases 60,61 . The lack of effect in the ZFL cell line may therefore be due to a faster metabolism of the compound within hepatocytes, since functional phase I activity has been reported in vitro 27,62,63 on transcript and protein levels, although also other mechanisms might explain the observed difference in response in ZFL and ZF4. Alterations in antioxidant enzymes, GSH, and LPO levels have also been reported in carp (Cyprinus carpio) after exposure to diazinon 64,65 . Beyond that, histopathological endpoints had been identified in bluegill sunfish (Lepomis macrochirus) 66 . Diuron has been reported to induce oxidative stress after chronic exposure to goldfish (Carassius auratus) 67 , alternating GST and CAT levels in tropical fish (Astyanax sp.) 68 , and genotoxicity in rainbow trout liver and gill cell lines (RTL-W1, RTG-W1) 69 . An induced Nrf2 activity by diuron and diazinon has also been reported in human liver hepatoma cell line (HepG2) at comparable concentrations 11 . Metazachlor was the most potent inducer of Nrf2 activity, showing a higher potency than the known inducers used as positive controls. No studies on oxidative stress response by metazachlor were found in the literature. However, regulatory toxicity testing has demonstrated low acute toxicity in rainbow trout (Oncorhynchus mykiss), carp, and bluegill sunfish 70 . In mammals limited evidence of carcinogenic effects has been reported besides acute toxicity 71 . To our knowledge, this is a novel finding on the toxicity of metazachlor in vertebrates and prompts further investigation regarding the toxicity of metazachlor.

Conclusion
We describe the development and application of a novel Nrf2 responsive reporter gene assay to monitor oxidative stress response in various zebrafish cell lines. Further, we have shown that the developed assay is useful for chemical testing by analyzing the effect of postulated Nrf2 inducers and compounds which were suggested to induce oxidative stress in fish. We suggest that this model could be a valuable tool for future research in aquatic toxicology to study the toxicity of both pure compounds and environmental samples. Data availability. The datasets generated during the current study are available from the corresponding author upon request.