Yazhen Hong & Shuxing Liu & Xiaofeng Lin & Jiqiu Li & Zhenzhen Yi & Khaled A. S. Al-Rasheid

Received: 5 October 2014 /Accepted: 8 January 2015 /Published 29 January 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract The equivocality of dose–response relationships has, in practice, hampered the application of biomarkers as a

means to evaluate environmental risk, yet this important issue has not yet been fully recognized or explored. This paper

evaluates the potential of antioxidant enzymes in the ciliated protozoan Euplotes vannus for use as biomarkers. Dose–response dynamics, together with both the enzyme activity and the gene expression of the antioxidant enzymes, superoxide

dismutase, and glutathione peroxidase, were investigated when E. vannus were exposed to graded doses of nitrofurazone for several discrete durations. Mathematical models were explored to characterize the dose–response profiles and, specifically, to identify any equivocality in terms of endpoint. Significant differences were found in both enzyme

activity and messenger RNA (mRNA) expression in the E. vannus treated with nitrofurazone, and the interactions between

exposure dosage and duration were significant. Correlations between enzyme activity, mRNA expression, and nitrofurazone dose varied with exposure duration. Particularly, the dose–responses showed different dynamics depending on either endpoint or exposure duration. Our findings suggest that both the enzyme activity and the gene expression of the tested antioxidant enzymes can be used as biomarkers for ecotoxicological assessment on the premise of ascertaining appropriate dosage scope, exposure duration, endpoint, etc., which can be achieved by using dose–response dynamics.

Keywords  Antioxidase . Biomarker . Dose–response dynamics . mRNAexpression . Nitrofurazone

   Introduction

   In environmental studies, biomarkers are typically defined as quantitative measures of change in the biological system that respond to either exposure to, and/or doses of, xenobiotic substances (Lam and Gray 2003). Biomarkers that respond

to exposure exhibit a causal relationship between the concentration of pollutants in the environment and the response of

target cells, tissues, or the organism (Handy et al. 2003). In terms of biomarkers exhibiting dose–response relationships, it is worth noting that these are subject to many factors, such as endpoints (Ameur et al. 2012), biological indicators (Kahru et al. 2011), pollution properties (Bondarenko et al. 2013), environmental factors (Fonseca et al. 2011), monitoring technologies (Bellas et al. 2014), exposure procedures (Li et al. 2014), etc. of the key issues for the application of such biomarkers in practice, therefore, is their counterintuitive variation according to these various factors (Li et al. 2014). Despite a wealth of studies on the potential of biological responses used as biomarkers, in most cases, these have focused on ascertaining whether the above factors account for the biological responses, rather than quantifying the dose–response relationship between pollution and the biological response (Raphael et al. 2014). Historically, questions about how to best apply information from toxicity studies have led to a growing recognition that dynamic analyses, particularly dynamic models that estimate dose–response, are valuable tools for addressing such methodological problems, by helping to reinterpret existing data and to generate explanatory and predictivemodels (Li et al. 2014;McCarty 2013; Thompson et al. 2008). The optimal use of biomarkers in environmental studies, therefore, requires in-depth knowledge of the dynamics of biomarker responses (Raphael et al. 2014). As the most important free radical scavengers, antioxidant enzymes are being increasingly studied in toxicological research due to their potential to provide biochemical biomarkers (Stara et al. 2012; Zheng et al. 2013). Superoxide dismutase, catalase, and glutathione peroxidase are considered as three major antioxidant enzymes underpinning the oxidant/ antioxidant balance in aquatic species (üner et al. 2005). Many studies of aquatic organisms experiencing environmental contaminant, however, have demonstrated the disparities between activity and gene expression in these three antioxidant enzymes (Glanemann et al. 2003; Li et al. 2014). That is, these endpoints made asynchronous responses to environmental contaminant when exposure concentration or duration varied. To the best of our knowledge, relatively little attention has been paid to the response dynamics or gene expression of antioxidant enzymes as a means of illustrating such disparities.

   It is necessary, therefore, to reveal these disparities by characterizing their response dynamics based on the mathematical models; this will help to clarify the mechanisms by which antioxidant enzymes respond to environmental stress. Nitrofurazone, chosen as the model antibiotic in the present study, is a broad-spectrum antibiotic whose toxicity has been intensively investigated and demonstrated in humans, mammals, shellfish, and microorganisms (Li et al. 2014; Martinez et al. 1995; Vlastos et al. 2010; Zhou et al. 2011). Despite the introduction of policies to prohibit the use of nitrofurazone, it is still widely employed in fishery aquaculture because of its cheapness and efficiency (Du et al. 2014; Vlastos et al. 2010). For these reasons, it is crucial to develop ideal bioassays for the toxicological assessment of nitrofurazone in aquaculture. With the extraordinary advantages of high diversity, short life cycles, cosmopolitan distribution, simplicity, a high degree of reproducibility, and quick responses to environmental disturbances in an integrated and continuous manner, ciliated protozoa have been recognized as ideal biological indicators for environmental risk assessment (Li et al. 2014; Gomiero et al. 2013; Zhou et al. 2011). In previous studies, they have also been confirmed as of the ideal test organisms for assessing the toxicological effects of pollutants.

   In order to evaluate the potential of antioxidant enzymes in ciliated protozoan for use as biomarkers through a consideration of the importance of dose–response dynamics, both the enzyme activity and gene expression of the antioxidant enzymes superoxide dismutase and glutathione peroxidase were investigated in E. vannus exposed to graded doses of nitrofurazone for several discrete durations. Specifically, mathematical models were explored to characterize the dose–response dynamics in order to identify difference in response due to enzyme species and endpoint level. It is anticipated that the information in this study will provide strategies to evaluate the best potential biomarkers for risk assessment in aquatic ecosystems and to improve the ability to use them reliably in practice.

 Materials and methods

  Nitrofurazone

  The tested nitrofurazone (5-nitro-2-furfural semicarbazone) was bought from China Sigma-Aldrich Shanghai Trading

Co., Ltd., Shanghai, China (CAS No.: 59870). The stock solutions (100 mg l?1) of nitrofurazone were made by dissolving

them in sterilized artificial marine water (AMW, 28 g of NaCl, 0.8 g of KCl, 5 g of MgCl2·6H2O, and 1.2 g of CaCl2 in

1000 ml of distilled water, pH 8.2, salinity 30‰), and then, test solutions of different concentrations were prepared by

further diluting with AMW.

  Organisms and media

  The ciliated protozoan Euplotes vannus, a common species in coastal waters, was used as the test model organism. The test E. vannus (salinity 30‰) was obtained from the Laboratory of Protozoology Ocean University of China, Qingdao, China. Species identification had been previously completed based on morphology and molecular information (Chen and Song

2002). Clonal cultures were established and maintained in the AMW (prepared as defined in “Nitrofurazone” section) at

25 °C. Rice grains were added to enrich the natural bacteria which are the food source for the ciliated protozoa.

Exposure of ciliates to nitrofurazone E. vannus cells in the exponential growth phase were inoculated into nitrofurazone solutions at gradated doses of 0, 3, 6,12, and 24 mg l?1, with each dose series being tested at fourdiscrete time intervals (6, 12, 18, and 24 h). Each treatment was performed in triplicate in 50-ml flasks. The final volume of the culture media was 20 ml, with the density of E. vannus cells being about 4×103 ind.ml?1. After being exposed into the nitrofurazone solutions, the cells were collected in triplicate from each treatment at each discrete time point and then condensed by centrifugation at 3000 rpm, 4 for 3 min in order to determine antioxidant enzyme activities and messenger RNA (mRNA) relative expression level. After centrifugation, the cell density in each sample was no less than 8×104 ind. ml?1. Next, the cells in sample were lysed by ultrasonication in cell lysis buffer for Western and IP (in 1 % PMSF; Beyotime Co., China). During the preprocessing procedure, the samples were kept in the dark and on ice or in ice water mixture.

  To minimize contamination through food (mainly bacteria) of the tested E. vannus, the test groups were not fed throughout the entire experimental period, and the samples were further treated by adding lysozyme (1.6 ml, at a dose of 10 mg ml?1) for 12 h before extracting the total RNA.

   Antioxidant enzyme activity assays

   The homogenates of ciliate cells were centrifuged again at 14,000 rpm for 10 min at 4 °C, and the supernatants were collected for antioxidant enzyme activity assays. Superoxide dismutase (SOD) (EC1.15.1.1) activity was measured by the ferricytochrome c method using xanthine/xanthine oxidase as a source of superoxide radicals,monitored at 550 nm (McCord and Fridovich 1969). unit of SOD activity is defined as the amount of enzyme required to cause a 50 % inhibition of cytochrome c reduction, with the activity being expressed as units per milligram protein. Glutathione peroxidase (GPx) (EC1.11.1.9) activity was examined according to the method of Lawrence and Burk (1976), in which the rate of NADPH oxidation in a coupled reaction with glutathione reductase is tracked. The absorbance was recorded at 412 nm. unit of GPx activity was defined as the amount of enzyme required to deplete 1 μmol GSH in 1 min, with this activity also being expressed as units permilligram protein.

  The protein level was estimated following the Bradford method (Bradford 1976). Briefly, bovine serum albumin was used as a standard and absorbance was recorded at 595 nm. Allmeasurements were performed using an ultraviolet visible spectrophotometer (Pucci, China). Quantitative analysis of the antioxidant enzymes’ mRNA expression by real-time PCR Total RNAwas extracted from the ciliates, further treated with RQ1 RNase-Free DNase (Promega,Madison, USA), and then reverse-transcribed into complementary DNA (cDNA) using the M-MLV RTase cDNA Synthesis Kit (TaKaRa, Dalian, Japan). The quality of RNA samples was determined by using Nano Drop ND-1000 spectrophotometric measurements of the ratio of absorbance 260 and 280 nm (A260/280) in a range from 1.8 to 2.0 and 1 % agarose gel electrophoresis, based> the integrity of 18S and 28S rRNA bands. Full-length cDNAs of manganese superoxide dismutase (MnSOD) (GenBank accession no. KJ619484) and GPx (GenBank accession no. KF049698) from E. vannus were cloned in this study. The α-tubulin gene of E. vannus (GenBank accession no. Z11769) was used as an internal control gene, based on our preliminary study that the amplicon efficiencies of the target and reference gene were approximately equal. Two MnSOD-specific primers (SOD-RT-F and SOD-RTR; Table S1) were used to amplify a 104-bp fragment of MnSOD. Primers GPx-RT-F and GPx-RT-R were used to amplify a 111-bp fragment of GPx. Primers tubulin-RT-F and tubulin-RT-R (Li et al. 2014) were used to amplify a 122-bp fragment of α-tubulin, as the internal control for quantitative RT-PCR. Real-time PCR was performed on an ABI 7500 (Applied Biosystems, USA) to study the mRNA expression of MnSOD and GPx in ciliates treated under different conditions. The PCR reaction was performed on a 20-μl volume with a SYBR? Premix Ex Taq? (Tli RNaseH Plus) Kit (TaKaRa, Dalian, Japan), 10 μl SYBR? Premix Ex Taq? (2×), 0.4 μl ROX Reference Dye (50×), 3.27 μl of sterile deionized water, 0.16 μl (25 μM) of each specific primer, and 1 μl (400 ng μl?1) of cDNA using the following procedure: initial denaturation at 95 °C for 30 s followed by 40 cycles of amplification (95 °C for 5 s and 60 °C for 34 s). In order to test the primers’ specificity and to have the best efficiency of amplification, in our pre-experiments, a gradient PCR was performed to detect the optimum annealing temperature. To make sure that each PCR product was unique, following the real-time quantitative PCR experiment, we examined whether the dissolution curve was normal and whether or not there was a single peak. The relative expression levels of different genes were calculated using the 2?ΔΔCT method (Livak and Schmittgen 2001). Statistical analyses and dose–response modeling Statistical analyses were performed by using SPSS software (version 17.0 forWindows). Results are shown as means±SE (standard error of means; n=3). All tests used a statistical significance level of P<0.05. In order to determine whether significant variation existed, data from each treatment was subjected to repeated measures of analyses of variance (rANOVA). Dosage (0, 3, 6, 12, and 24 mg l?1) was used as the between-subjects factors, and exposure duration (6, 12, 18, and 24 h) was used as the withinsubjects factors. When overall differences existed, a least significant difference (LSD) test was used to compare the means between individual treatments.

  To analyze possible correlations between enzyme activity and gene relative expression level, and biological responses and nitrofurazone dose, Pearson’s correlation analyses were performed on the variables measured at discrete exposure durations, separately. Dose–response models were developed to characterize the dose–response dynamics. The method has been successfully used and detailedly described in our previous study (Li et al.2014).

   Results

   Enzyme activities of SOD and GPx Results of rANOVA showed that significant nitrofurazone treatment and duration effects, and treatment×duration interaction, were detected for enzyme activities of both SOD and GPx of E. vannus that were exposed to gradated doses of nitrofurazone. In particular, there were significant differences in the enzyme activities of both SOD and GPx in the E. vannus groups that were exposed to graded doses of nitrofurazone at different exposure durations (Fig. 1). As for SOD activity, the magnitude of the difference between the nitrofurazone-treated E. vannus and the control groups was reduced as the exposure duration increased; to the extent that when exposure duration reached 24 h, no significant differences were found in SOD activity between the different groups of tested E. vannus (Fig. 1A). In regard to GPx activity, the nitrofurazone-treated E. vannus were significantly higher than the control groups (P<0.05), and significant dose–response relationships were presented at the discrete exposure durations with exception of the 6 h. In terms of exposure duration, at 12 h, GPx activities significantly increased and presented a significant dose–response relationship (P<0.05); when exposure duration reached 24 h, however, no significant differences were found in GPx activities among the nitrofurazone-treated E. vannus and the differences with the control groups also reduced (Fig. 1B). These results indicate that exposure to nitrofurazone caused an alteration of enzyme activity in both SOD and GPx and that the dose–response relationships depend on the exposure duration.

   mRNA relative expression levels of SOD and GPx Results of rANOVA showed that significant nitrofurazone treatment, duration effects, and treatment×duration interaction were detected on the mRNA relative expression levels of both SOD (Fig. 2A) and GPx (Fig. 2B) from the E. vannus that were exposed to gradated doses of nitrofurazone. With respect to SOD, the mRNA relative expression levels generally increased as the dose of nitrofurazone increased. the exposure duration reached 18 h, however, the mRNA relative expression

   Fig. 1 Enzyme activity of superoxide dismutase (SOD, A) and glutathione peroxidase (GPx, B) in Euplotes vannus exposed to different doses of nitrofurazone for exposure durations ranging from 6 to 24 h. Columns bearing the same superscript letter are not significantly different determined by LSD test (P=0.05). Data are presented as means ±SE (standard error of themean), and the error bars represent the standard errors of the means (n=3) levels decreased, particularly, at 24 h. The maximum mRNA relative expression levels were recorded at a dose of 12 mg l?1, which were significantly higher than those at 24 mg l?1 (P<0.05).

   Similarly, the mRNA relative expression levels of GPx generally showed a positive dose relationship at exposure durations from 6 to 24 h, with the maximum expression levels recorded for the 24 mg l?1 dose, with significant differences (P<0.05). In regard to the relationship for exposure duration, mRNA relative expression levels of GPx decreased as the exposure durations increased, with no significant differences being found between the 12 and 24 mg l?1 groups at exposure durations of 6, 12, and 18 h.

   These results indicate that a significant dose–response relationship exists between the mRNA relative expression levels of both SOD and GPx in E. vannus and the doses of nitrofurazone, a relationship which also changes according to the exposure duration.

   Fig. 2 mRNA relative expression levels of superoxide dismutase (SOD,A) and glutathione peroxidase (GPx, B) in Euplotes vannus exposed to different doses of nitrofurazone for exposure durations ranging from 6 to 24 h. Columns bearing the same superscript letter are not significantly different determined by LSD test (P=0.05). Data are presented as means ±SE (standard error of themean), and the error bars represent the standard errors of the means (n=3)

   Dose–response dynamics for enzyme activities of SOD and GPx

   Dose–response dynamics for enzyme activities showed that the activities of both tested enzymes generally increased as the nitrofurazone dose increased. At each of the discrete exposure durations, the dose–response curves for SOD activity presented a Napierian logarithm relationship (Fig. 3A), which were best described by Eq. (1), where “y”=SOD activity (U), “x”=dose of nitrofurazone (mg l ? 1 ) , “ x 0” = specific dose of nitrofurazone for SOD activity infinite tends to zero (mg l?1), and “a”=variance to measure the slope for enzyme activities’ curve. As the nitrofurazone dose increased, the slope (a) for the dose–response curves reduced; that is, the positive effect of nitrofurazone dose on SOD activity weakens (Table S2).

   Dose–response dynamics for GPx activity presented different relationships at different discrete exposure durations (Fig. 3B). At exposure durations of 6 and 24 h, the dose–response curves were best described by Eq. 1. Here, the value of a increased from 254.9 to 583.7 as the exposure duration changed from 6 to 24 h (Table S3). At exposure durations of 12 and 18 h, however, the dose–response curves presented linear relationships which were best described by Eq. 2, where y=enzyme activity (U) or mRNA relative expression level, x= dose of nitrofurazone (mg l?1), and “k”=variance to measure the slope for linear relationship. Here, the value of k increased from 59.95 to 64.34 as the exposure duration changed from 12 to 18 h (Table S3). The dose–response dynamics for enzyme activities, therefore, showed that for both tested enzymes, their activities exhibited a prominent positive dose–response relationship, but that the shape of this relationship depended on the exposure duration.

   Dose–response dynamics for the mRNA relative expression levels of SOD and GPx

   At exposure durations of 6, 12, and 18 h, the dose–response dynamics for mRNA relative expression of SOD presented linear relationships, which were best described by Eq. 2. The values of k were similar for the dose–response curves relating to exposure durations of 6 and 12 h, but noticeably less when the exposure duration increased to 18 h (Table S4). At 24 h, however, the dose–response curves presented a sigmoid relationship (Fig. 4a; Table S5), which was best described by Eq. 3, where y=mRNA relative expression level, “ymax”= the maximum value of the mRNA relative expression level in cells which were tested in specific series of doses of nitrofurazone, x=dose of nitrofurazone (mg l?1), x0=specific dose of nitrofurazone for “ymax/2” (mg l?1), and “b”=a variance to measure the dispersion degree for a set of mRNA relative expression levels (Table S4).

   Similarly, at each of the discrete exposure durations, the dose–response curves for the mRNA relative expression level of GPx overall presented a sigmoid relationship (Fig. 4B), which were best described by Eq. 3 (Table S5). According to Eq. 3, the theoretical peak values of mRNA relative expression levels decreased from 53.96-fold to 5.308-fold with the increasing of exposure durations. It suggests that mRNA relative expression levels of GPx in nitrofurazone-treated E. vannus reached a platform as the exposure doses increased to a specific dose and the peak values decreased as the exposure durations were extended.

   Briefly, these results show that the dynamics of the nitrofurazone-induced dose–response for mRNA relative expression levels in both the tested antioxidant enzymes (SOD and GPx) were different and depended on the enzyme and the exposure durations.

Correlation between antioxidant enzyme activities, mRNA expression level, and nitrofurazone dose The correlation between antioxidant enzyme activity, mRNA expression level, and nitrofurazone dose showed a variation according to the duration of exposure to nitrofurazone (Table S6). Generally, both enzyme activity and gene expression level showed a positive correlation with nitrofurazone dose at different exposure durations. Nitrofurazone dose was significant correlated with SOD activity at 6 h (P<0.01) and 12 h (P<0.01) and with GPx activity at 12 h (P<0.01) and 18 h (P<0.01). However, gene expression level of both SOD and GPx showed no significant correlation with nitrofurazone

   Fig. 3 Dose–response dynamics for enzyme activities of superoxide dismutase (SOD, A) and glutathione peroxidase (GPx, B) in Euplotes vannus exposed to different doses of nitrofurazone for exposure durations ranging from 6 to 24 h. Continuous lines are the best fit to the data in the figures following equations in Tables S2 and S3, respectively dose during exposure (P>0.05). The correlation between SOD activity and its gene expression level was positive at 6 h (P>0.05), 12 h (P>0.05), and 18 h (P<0.05), but negative at 24 h (P>0.05). A positive correlation was detected between GPx activity and its gene expression level according to exposure duration, but without significance (P>0.05). SOD activity showed a significant positive correlation with GPx activity at exposure times of 6 h (P<0.01) and 12 h (P<0.01). The positive correlation between SOD and GPx in respect to the strength of gene expression was significant for all the exposure durations (P<0.01). The results of these correlation analyses indicate that the dose–response relationships of these two antioxidant enzymes depend on the endpoint and exposure duration.

   Discussion

   The metabolic activation of nitrofurazone leads to the production of the superoxide anion radical (O2?) which is then dismutated to hydrogen peroxide (H2O2) (Ghersi-Egea et al. 1998). This in turn reacts with Cu(I) to produce the primary reactive species capable of causing DNA damage (Hirakuet al. 2004). As the most important antioxidant enzymes,

   Fig. 4 Dose–response dynamics for mRNA relative expression levels of superoxide dismutase (SOD, A) and glutathione peroxidase (GPx, B) in Euplotes vannus exposed to different doses of nitrofurazone for exposure durations ranging from 6 to 24 h. Continuous lines are the best fit to the data in the figures following equations in Tables S4 and S5, respectively

   SODs are the precursors to protect against the oxidative damage by catalyzing the dismutation of the O2? to H2O2 plus water (Khan et al. 2012). Subsequently, catalase (CAT) and GPx are activated by the accumulation of H2O2 and, then, work in cooperation to convert the H2O2 into water and molecular oxygen (Wang et al. 2011). In this present study, the tested antioxidant enzymes, SOD and GPx, showed significant differences in both activity and mRNA expression level in E. vannus exposed to graded doses of nitrofurazone. Together with the similar result observed in our previous study, where CATwas chosen as the biomarker (Li et al. 2014), these results confirm that exposure to nitrofurazone causes an antioxidative response in the ciliated protozoan, e.g., enzyme activity and mRNA expression level. As suggested by Handyet al. (2003), response to xenobiotic stress is of the prerequisites if biological responses are to be used as biomarkers.

   On the other hand, the results of the correlation analyses in the present study showed that the dose–response relationships vary according to both exposure durations and endpoints, and this can be of great importance for toxicological evaluations. Exposure dose and duration were considered as the most significant factors that determine biological response dynamics and chemical toxic effect (Belkebir et al. 2011). Firstly, the dose–response relationships generally varied as the exposure duration increased. Additionally, the interaction between these two variables and the biological responses has been confirmed by an increasing number of investigations, which usually result in counterintuitive dose–response relationships (Brown and Foureman 2005; Li et al. 2014). In the present study, interactions between nitrofurazone dose and exposure duration were detected on all the tested biological responses in that E. vannus that was exposed to graded doses of nitrofurazone. In agreement with the observations of Calabrese and Baldwin (2001), both nonlinear (including logarithmic, biphasic, and sigmoid) and linear dose–response relationships were frequently found in the tested biological responses to pollutant doses. It is worth mentioning that the nonlinear dose–response relationships were observed in a number of cases, which generally makes the interpretation of biological response more complex (Richardson et al. 2008; Li et al. 2014). Additionally, the mRNA relative expression level of GPx presented a typical sigmoid relationship in E. vannus exposed to different doses of nitrofurazone for all discrete exposure durations. This kind of counterintuitive dose–response relationship has been of the main problems preventing the use of antioxidant enzymes as biomarkers to evaluate environmental risk.

   As discussed above, the detoxification brought about by the antioxidant enzyme system is a complex bioprocess. During this process, each enzyme plays a specific role, but it also works in sequence, and the activity of impacts on that of the others (Li et al. 2012; Nie et al. 2013; Zheng et al. 2013). From the perspective of biological monitoring, therefore, a single antioxidant enzyme is probably not sufficient to indicate the toxic effects comprehensively. In addition, difference in the biological response patterns between endpoints is a common occurrence, involving different biological levels or components in identical systems, which leads to an element of equivocality in the correlations (Arukwe 2002; Glanemannet al. 2003; Jayaraj et al. 2006; Park et al. 2006). Consequently, these disparity and equivocality make the application of biomarkers more complicated (Arukwe 2002; Jayaraj et al.2006; Glanemann et al. 2003). Previous studies have also shown differences in response pattern between the activities of antioxidant enzymes and their mRNA expression levels, with almost no significant correlations being found (Peixotoet al. 2006; Stephensen et al. 2000; Timofeyev et al. 2008). Researchers, therefore, have been caught in a dilemma as to whether to select a single enzyme or multiple endpoints for use as a biomarker. benefit of a dose–response dynamic is to articulate a system through which it is possible to understand the modes of action on biomarkers (Auger and Poggiale 2006; Fukushima et al. 2005; Brown and Foureman 2005). In terms of this present study, the information collected from the dose–response dynamics, involving dose–response relationship characteristics (the model parameters in equations), as well as their variation tendencies, will facilitate a better understanding of the modes of action for the observed effects and help to interpret the correlations between the responses of the enzymes in the antioxidant system.

   The dose–response dynamics in the present study showed that SOD activities increased with logarithmic dose–response relationships, while mRNA relative expression levels present positive linear shapes, but the slopes (growth ratios) of each decreased as the exposure time increased. This provides further evidence for a feedback mechanism in which an excess amount of H2O2 provided by SOD can in turn inhibit SOD activity (Zheng et al. 2013). These results are supported by previous studies in which ciprofloxacin and sulfamethoxazole significantly induced SOD activity in fish and microalgae (Liet al. 2012; Nie et al. 2013).

   GPx activities, however, increased sharply, with higher slopes in the intermediate exposure durations (12 and 18 h).

   Compared to the linear dose–response shapes of SOD, sigmoid growth shapes were found in the mRNA relative expression level of GPx, which were characterized by a more gradual initial increase, which then accelerated sharply to a peak as the exposure doses increased. This indicates that both the enzyme activity and the mRNA relative expression of GPx were regulated by the amount of hydrogen peroxide (H2O2) (Zhang et al. 2013). On the other hand, dynamics with different characters probably resulted from that each antioxidant enzyme plays a specific role and works in sequence during the detoxification process. Based on this principle, it is easy to understand that the differences in specific biological function together with working in sequence are bound to lead to complexity and equivocality in correlations between different endpoint responses.

   Interestingly, the two tested antioxidant enzymes were themselves positively and significantly correlated in terms of their mRNA expression levels through all the exposure durations (P<0.01). Based on the dose–response dynamics, however, it can be concluded that their correlations will become more uncertain as the durations are increased. In brief, the dose–response dynamics can provide a framework to understand the mode of action for antioxidant enzymes and the detoxification process; it also can help clarify the interpretation of the complex correlation in dose–responses between different endpoints.

   Knowing in what context a toxicological dataset is to be used is important in order to establish a relevant yardstick for judging whether or not the data is acceptable (McCarty 2012).

  The dynamics of biomarker responses based on the biological models probably provide a practical approach to quantify the relationship between contacted toxicants or other stressors> the hand and various elicited biological or toxicological responses in affected populations on the other hand (Mushak 2013). In the present study, variable dose–response relationships can be characterized by the parameters from the corresponding dose–response dynamic models. Considering the different ranges of exposure and different levels of stimulatory response, it cannot be expected that a single model will be a good fit for all the different sets of dose–response data (Belz and Piepho 2012). With the aid of dose–response dynamics, however, researchers can choose the appropriate endpoints which present a suitably straightforward dose–response relationship for use as a biomarker (Handy et al. 2003). In terms of this study, the dose–response models for eithermRNArelative expression level of SOD at 6 or 12 h or GPx activities at 12 or 18 h are more ideal than the others. As for the nonlinear dose– response relationship, such asmRNA relative expression level of GPx, the dividing points (peak values) between the two kinds of linear relationships can be established by the model equations, which could then be used to disaggregate any nonlinear curve into two linear curves (Kendig et al. 2010). Therefore, the dose–response dynamic model can provide a basis for selecting a suitable endpoint and for determining the exposure duration, as well as the pollutant dose for risk assessment.

   Conclusions

   In brief, our findings have demonstrated the equivocality of dose–response relationships between antioxidant enzymes

and the exposure dosage of nitrofurazone, and further analysis confirmed the importance of dose–response dynamics for

characterizing dose–response relationships by using mathematical model. Our results, therefore, strongly suggest that

analyzing the dose–response relationship by using dose–response dynamics is an effective strategy to evaluate a potential biomarker for risk assessment in aquatic ecosystems and thus will improve the ability to use such a marker reliably in practice.

   There are many other factors impacting the dose–response model, however, e.g., pollutant, organism-specific factors

and so hence, it is clear that further work centered>these factors is necessary. Furthermore, the analytical approach to dynamic model recognition of dose–response data provides a structure to strategic thinking in that has to condense thoughts into a set of mathematical relationships. Enforcing this strongly analytical approach is another key advantage of adopting the kind of mathematic models previewed in this paper.

   Acknowledgments This work was supported by the National Science Foundation of China (Project Nos. 41476128, 31222050, and 31172060) and King Saud University, Deanship of Scientific Research, Research Group Project (No. RGP-083).

References

Ameur WB, de Lapuente J, Megdiche YE, Barhoumi B, Trabelsi S,

Camps L, Serret J, Ramos-López D, Gonzalez-Linares J, Driss

MR, Borràs M (2012) Oxidative stress, genotoxicity and histopathology

biomarker responses in mullet (Mugil cephalus) and sea

bass (Dicentrarchus labrax) liver from Bizerte Lagoon (Tunisia).

Mar Pollut Bull 64:241–251

Arukwe A (2002) Complementary DNA cloning, sequence analysis and

differential organ expression of b-naphthoflavone-inducible cytochrome

P4501A in Atlantic salmon (Salmo salar). Comp Biochem

Physiol C 133:613–624

Auger P, Poggiale JC (2006) Mathematical modeling is a necessary step

in biology and in environmental sciences. Geoscience 338:223–224

Belkebir E, Rousselle C, Duboudin C, Bodin L, Bonvallot N (2011)

Haber’s rule duration adjustments should not be used systematically

for risk assessment in public health decision-making. Toxicol Lett

204:148–155

Bellas J, Albentosa M, Vidal-Li?án L, Besada V, Franco Má, Fumega J,

González-Quijano A, Vi?as L, Beiras R (2014) Combined use of

chemical, biochemical and physiological variables inmussels for the

assessment of marine pollution along the N-NWSpanish coast.Mar

Environ Res 96:105–117

Belz RG, Piepho HP (2012) Modeling effective dosages in hormetic

dose-response studies. PLoS 7:e33432

Bondarenko O, JugansonK, IvaskA, Kasemets K,MortimerM,Kahru A

(2013) Toxicity of Ag, CuO and ZnO nanoparticles to selected environmentally

relevant test organisms and mammalian cells in vitro:

a critical review. Arch Toxicol 87:1181–1200

BradfordMM(1976) A rapid and sensitive method for the quantitation of

microgramquantities of protein utilizing the principle of protein-dye

binding. Anal Biochem 72:248–254

Brown KG, Foureman GL (2005) Concentration-time-responsemodeling

for acute and short-term exposures. Regul Toxicol Pharmacol 43:

45–54

Calabrese EJ, Baldwin LA (2001) The frequency of U-shaped dose responses

in the toxicological literature. Toxicol Sci 62:330–338

Chen Z, Song W (2002) Phylogenetic positions of Aspidisca steini and

Euplotes vannus within the Order Euplotida (Hypotrichia:

Ciliophora) inferred from complete small subunit ribosomal RNA

gene sequences. Acta Protozool 41:1–9

Du N, Chen M, Sheng L, Chen S, Xu H, Liu Z, Song C, Qiao R (2014)

Determination of nitrofuran metabolites in shrimp by high performance

liquid chromatography with fluorescence detection and liquid

chromatography–tandem mass spectrometry using a new derivatization

reagent. J Chromatogr A 1327:90–96

FonsecaVF, Fran?a S,VasconcelosRP, SerafimA, Company R, Lopes B,

Bebianno MJ, Cabral HN (2011) Short-term variability of multiple

biomarker response in fish from estuaries: influence of environmental

dynamics. Mar Environ Res 72:172–178

Fukushima S, Kinoshita A, Puatanachokchai R, Kushida M, Wanibuchi

H, Morimura K (2005) Hormesis and dose-response-mediated

mechanisms in carcinogenesis: evidence for a threshold in carcinogenicity

of non-genotoxic carcinogens. Carcinogenesis 26:1835–

1845

Ghersi-Egea JF, Maupoil V, Ray D, Rochette L (1998) Electronic spin

resonance detection of superoxide and hydr?l radicals during the

reductive metabolism of drugs by rat brain preparations and isolated

cerebral microvessels. Free Rad Biol Med 24:1074–1081

Glanemann C, Loos A, Gorret N, Willis LB, O’Brien XM, Lessard PA,

Sinskey AJ (2003) Disparity between changes in mRNA abundance

and enzyme activity in Corynebacterium glutamicum: implications

for DNA microarray analysis. ApplMicrobiol Biotechnol 61:61–68

Gomiero A, Dagnino A, Nasci C, Viarengo A (2013) The use of protozoa

in ecotoxicology: application of multiple endpoint tests of the ciliate

E. crassus for the evaluation of sediment quality in coastal marine

ecosystems. Sci Total Environ 442:534–544

Handy RD, Galloway TS, DepledgeMH (2003) A proposal for the use of

biomarkers for the assessment of chronic pollution and in regulatory

toxicology. Ecotoxicology 12:331–343

Hiraku Y, Sekine A, Nabeshi H, Midorikawa K, Murata M, Kumagai Y,

Kawanishi S (2004) Mechanism of carcinogenesis induced by a

veterinary antimicrobial drug, nitrofurazone, via oxidative DNA

damage and cell proliferation. Cancer Lett 215:141–150

Jayaraj R, Anand T, Rao PV (2006) Activity and gene expression profile

of certain antioxidant enzymes to microcystin-LR induced oxidative

stress in mice. Toxicology 220:136–146

Kahru A, Bondarenko O, Kajankari P, K?kinen A, Ivask A, Kasemets K,

MortimerM, Blinova I (2011) Bioavailability and toxicity of copper

oxide and silver nanoparticles to bacteria, yeasts, crustaceans and

protozoa. Toxicol Lett 205(Suppl):S284–S285

Kendig EL, Le HH, Belcher SM(2010) Defining hormesis: evaluation of

a complex concentration response phenomenon. Int J Toxicol 29:

235–246

Khan RA, Khan MR, Sahreen S (2012) Brain antioxidant markers, cognitive

performance and acetylcholinesterase activity of rats: efficiency

of Sonchus asper. Behav Brain Funct 8:1–7

Lam PKS, Gray JS (2003) The use of biomarkers in environmental monitoring

programmes. Mar Pollut Bull 46:182–186

Lawrence RA, Burk RF (1976) Glutathione peroxidase activity in

selenium-deficient rat liver. Biochem Biophys Res Commun 71:

952–958

Li J, Zhou L, Lin X, Yi Z, Al-Rasheid KAS (2014) Characterizing dose–

responses of catalase to nitrofurazone exposure in model ciliated

protozoan Euplotes vannus for ecotoxicity assessment: enzyme activity

andmRNAexpression. Ecotoxicol Environ Safe 100:294–302

Li Z, Lu G, Yang X, Wang C (2012) Single and combined effects of

selected pharmaceuticals at sublethal concentrations on multiple

biomarkers in Carassius auratus. Ecotoxicology 21: 353–361

Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression

data using real-time quantitative PCR and the 2-ΔΔC(T) method.

Methods 25:402–408

Martinez PG, Winston GW, Metashdickey C, Ohara SCM, Livingstone

DR (1995) Nitrofurantoin-stimulated reactive oxygen species production

and genotoxicity in digestive gland microsomes and cytosol

of the common mussel (Mytilus edulis L.). Toxicol Appl Pharmacol

131:332–341

McCarty LS (2012) Model validation in aquatic toxicity testing: implications

for regulatory practice. Regul Toxicol Pharmacol 63:353–362

McCarty LS (2013)Are we in the dark ages of environmental toxicology?

Regul Toxicol Pharmacol 67:321–324

McCord JM, Fridovich I (1969) Superoxide dismutase, an enzymatic

function for erythrocuprein. J Biol Chem 244:6049–6055

Mushak P (2013) Limits to chemical hormesis as a dose–response model

in health risk assessment. Sci Total Environ 443:643–649

Nie X, Liu B, Yu H, LiuW, Yang Y (2013) Toxic effects of erythromycin,

ciprofloxacin and sulfamethoxazole exposure to the antioxidant system

in Pseudokirchneriella subcapitata. Environ Pollut 172:23–32

Park J, Hecker M, Murphy MB, Jones PD, Solomon KR, Kraak GVD,

Carr JA, Smith EE, du Preez L, Kendall RJ, Giesy JP (2006)

Development and optimization of a Q-RT PCR method to quantify

CYP19 mRNA expression in testis of male adult Xenopus laevis:

comparisons with aromatase enzyme activity. Comp Biochem

Physiol B 144:18–28

Peixoto F, Alves-Fernandes D, Santos D, Fontaínhas-Fernandes A (2006)

Toxicological effects of oxyfluorfen on oxidative stress enzymes in

tilapia Oreochromis niloticus. Pestic Biochem Physiol 85:91–96

Raphael S,Mélissa P, Aurélien B, Emmanuelle V, Jean Marc P, Sylvie B,

Alain D, Wilfried S (2014) Kinetic response of a genotoxicity biomarker

in the three-spined stickleback and implication for environmental

monitoring. Ecotoxicol Environ Safe 102:6–11

Richardson BJ, Mak E, De Luca-Abbott SB, Martin M, McClellan K,

Lam PKS (2008) Antioxidant responses to polycyclic aromatic

hydrocarbons and organochlorine pesticides in green-lippedmussels

(Perna viridis): do mussels “integrate” biomarker responses? Mar

Pollut Bull 57:503–514

Stara A,Machova J, Velisek J (2012) Effect of chronic exposure to simazine

on oxidative stress and antioxidant response in common carp

(Cyprinus carpio L.). Environ Toxicol Pharmacol 33:334–343

Stephensen E, Svavarsson J, Sturve J, Ericon G, Adolfson-EriciM, F?rlin

L (2000) Biochemical indicators of pollution exposure in shorthorn

sculpin (Myoxocephalus scorpius), caught in four harbours on the

south-west coast of Iceland. Aquat Toxicol 48:431–442

Thompson CM, Sonawane B, Barton HA, DeWoskin RS, Lipscomb JC,

Schlosser P, Chiu WA, Krishnan K (2008) Approaches for applications

of physiologically based pharmacokinetic models in risk assessment.

J Toxicol Environ Health B 11:519–547

Timofeyev MA, Shatilina ZM, Bedulina DS, Protopopova MV,

Pavlichenko VV, Grabelnych OI, Kolesnichenko AV (2008)

Evaluation of biochemical responses in Palearctic and Lake Baikal

endemic amphipod species exposed to CdCl2. Ecotoxicol Environ

Saf 70:99–105

üner N, Oru? E, Sevgiler Y (2005) Oxidativestress-related and ATPase

effects of etoxazole in different tissues of Oreochromis niloticus.

Environ Toxicol Pharmacol 20:99–106

Vlastos D, Moshou H, Epeoglou K (2010) Evaluation of genotoxic effects

of semicarbazide on cultured human lymphocytes and rat bone

marrow. Food Chem Toxicol 48:209–214

Wang HF, Zhong XH, ShiWY, Guo B (2011) Study of malondialdehyde

(MDA) content, superoxide dismutase (SOD) and glutathione peroxidase

(GSH-Px) activities in chickens infected with avian infectious

bronchitis virus. Afr J Biotechnol 10:9213–9217

Zhang D, Hu C, Li D, Liu Y (2013) Lipid peroxidation and antioxidant

responses in zebrafish brain induced by Aphanizomenon flos-aquae

DC-1 aphantoxins. Aquat Toxicol 144–145:250–256

Zheng Q, Feng M, Dai Y (2013) Comparative antioxidant responses in

liver of Carassius auratus exposed to phthalates: an integrated biomarker

approach. Environ Toxicol Pharmacol 36:741–749

Zhou L, Li J, Lin X, Al-Rasheid KAS (2011) Use of RAPD to detect

DNA damage induced by nitrofurazone in marine ciliate, Euplotes