The present study shows a novel approach to measure toxicological effects in fish cells. This is not only specific for certain toxins, but also reveals a concentration-dependent performance. The use of whole cells for MALDI-TOF mass spectrometry is well established for identification of bacteria [10]. In recent years, this technology has been increasingly deployed for applications to create characteristic profiles of eukaryotic cell lines [11–18]. The use of IC-MS to produce specific expression profiles of fish cells and identify toxins through their specific patterns as a biomarker is a new approach in ecotoxicology.
For that purpose, we investigated the effects of three toxins belonging to three different chemical classes on the rainbow trout liver cell line RTL-W1: the metal compound copper sulfate, the NSO heterocyclic compound acridine, and the PAH compound BNF.
Firstly, a concentration range of each toxin was determined with different viability assays. The CellTiter-Blue® and CellTiter-Glo® record the cell viability according to the metabolic impairment in consequence to the toxin treatment. Whereas the CellTiter-Blue® measures the redox activity and hence the content of NADPH and NADH [27], the CellTiter-Glo® focuses on kinase activity and detects ATP levels in the cell [28]. Two different cell parameters are considered providing further information about the mechanisms of cellular actions affected by the respective substance [29]. The chosen cell number of 20,000 cells per well in a 96-well microtiter plate format is in accordance with the cell number usually applied in viability assays with fish cells [26, 30, 31]. The final DMSO concentration never exceeded 1 % (v/v) because in former studies, this concentration revealed first cytotoxic effects [32]. As acridine is a volatile substance, a gastight cover sheet was used for viability studies (cf. [23, 33]). Nevertheless, a loss of substance over time should be considered due to adsorption onto the well surface or precipitation. Peddinghaus et al. and Hinger et al. reported a loss of substance although a gastight cover sheet was used [5, 23].
The exposure of RTL-W1 with copper sulfate and acridine led to lower EC50 values for CellTiter-Blue® than CellTiter-Glo® viability assays. This could be related to a greater impairment of the redox activity of the cells. Primarily, cellular reductases, particularly diaphorases, are responsible for the conversion of resazurin into resorufin in the CellTiter-Blue® viability assay [27]. In contrast to that, the CellTiter-Glo® viability assay detects the amount of ATP in the cells via a luciferase reaction. As ATP is a common substrate of kinases, its detection delivers information about the kinase activity in the cell [28]. Consequently, the exposure with copper sulfate and acridine affected the reductases prior to the ATP content of the cell. The higher signals of low copper sulfate concentrations in contrast to the control treatment could be related to a higher metabolic activity as a response to the toxin exposure.
The molecular mechanisms of action for copper sulfate are still unknown. Nevertheless, a study with RTgill-W1 cells proved a significant increase in reactive oxygen species (ROS), as well as the induction of DNA strand breaks, as a consequence of the copper sulfate exposure [34].
The cellular effects of acridine could have a similar background. However, only few in vitro viability studies examined NSO-HETs so far and further investigations are crucial [23, 33].
The third substance, BNF, showed different results than copper sulfate and acridine. The CellTiter-Glo® viability assay was markedly more sensitive than the detection of reductases activity. This leads to the conclusion that the mode of action is unlike copper sulfate or acridine. Already, a concentration of 10 μM BNF resulted in a signal decrease. It can be assumed that a distinct EROD activity is induced by this concentration [30] and hence an increased kinase activity. Possibly not only the cytotoxic effect of BNF but also the increased ATP consumption consequent to the enzyme induction attributed to the decreased luminescence signals.
High concentrations of BNF resulted in precipitation in the culture medium due to its low solubility. Consequently, the effective concentration of BNF could be lower and should be considered for the estimation of its viability impairment or induction potency.
The calculated EC50 values for RTL-W1 cells with copper sulfate were a bit higher than literature values, e.g., 3.69 mg/L (redox activity) and 6.56 mg/L (membrane integrity) [26]. However, several test series confirmed the concentration range in this work and the influence of different assays should be considered. An EC20 of 40.9 mg/L (228 μM) was recorded for acridine with neutral red retention assay in preparation for micronucleus assays with RTL-W1 [33]. This indicates a much higher EC50 value than presented in this work, which is also ascribed to the different detection principle of the viability assays. For BNF, no reference values could be found, but likewise, several test series were performed. Therefore, the listed values are an indication for further investigations.
The performance of viability studies resulted in concentration ranges that revealed cytotoxic effects of copper sulfate, acridine, and BNF on RTL-W1 cells and were a starting point for IC-MS investigations. For that purpose, concentrations were chosen that resulted in a viability decline of roughly 10, 40, and 60 %. All three model compounds induced profile variations of RTL-W1. Furthermore, the six independent measurements for each concentration revealed similar profile changes. At least seven peaks were detected that changed significantly through the toxin exposure (acridine). For copper sulfate (eight peaks) and BNF (nine peaks), even more signals showed significant alterations. Particularly striking is the progressive increase of ion abundances at m/z 6731, 11,295, 13,448, and 13,627 as a result of copper sulfate and acridine exposure of RTL-W1. Those peaks were also increased in a study with preapoptotic substances with HeLa cells and other mammalian cell lines [14]. Microscopic studies support the assumption that the alteration of those peaks is also related to apoptosis of the fish cell line RTL-W1. Even for the lowest concentration of 2 mg/L copper sulfate, the IC-MS showed an apoptotic behavior, whereas the viability studies did not suggest a viability decline for this concentration. Cells treated with acridine showed additional variations in the low mass range. The signals at m/z 4562, 4576, 4633, and 4647 decreased notably compared to the control treatment. In contrast, the BNF-treated cells showed increased ion abundances in the lower mass range, whereas a significant decrease in the higher mass range could be observed.
Consequently, the results of the IC-MS reveal a different molecular mode of action when comparing BNF to acridine and copper sulfate likewise, which was confirmed by PCA. Microscopic investigations of BNF treated cells did not show typical signs of necrosis (e.g., cellular swelling) [35]. However, results of several studies indicate that the various cell death pathways (apoptosis, necrosis, and autophagy) correlate more than previously thought [36]. Hence, the information of microscopy and IC-MS are not sufficient to make a final statement on cell death of RTL-W1 after treatment with BNF.
The IC-MS of cells that were treated with mono-substances or complex samples offer a great potential for typing cytotoxic effects. A characteristic set of peaks could be used as a biomarker to identify different ecotoxicologically relevant substances in soil, sediment, or groundwater samples.
As of now, there have been few attempts made to identify the intracellular mechanisms that are correlated with altered peaks of IC-MS profiles. In particular, the identification of certain proteins or peptides behind the peaks in cell profiles is a desirable objective [12].
Bergquist et al. found altered ion abundances between m/z 4000 and 18,000 in profiles of the rat cell line PC 12 that could be related to stimulation with a nerve-growth factor [37]. Hanrieder et al. used SDS-PAGE and ESI-MS to identify proteins of glial cells that showed up in the IC-MS profiles [13]. Zhang et al. were able to identify a few proteins in the lower mass range that are either common between different mammalian cell lines or appeared specifically [12]. Some of the peaks that were found in this work were also mentioned by Zhang, Bergquist and Hanrieder, e.g., m/z 5655, 6275, and 11,295. Most notably, the latter was found in all three studies and identified as a modified form of histone 4 [13]. Furthermore, two forms of the structuring protein thymosin were identified in the mass range below 5000 Da. Since the peaks m/z 4885 and 4902 of RTL-W1 were significantly lower after the exposure with BNF, this could be related to changes in the organization of the cytoskeleton [38]. Similarly, other peaks could stand for certain molecular changes that were caused by the toxin exposure. However, the identification of peptides and proteins was not attempted in this work. Besides, one has to keep in mind that no prior purification steps were performed and hence, the peaks most likely do not refer to single proteins. Additionally, posttranslational modifications, e.g., glycosylation and phosphorylation, have to be considered, and thus only the mass is insufficient for protein identification [37].
Conclusion
This study outlines an initial attempt to identify toxic effects on RTL-W1 cells by IC-MS. The changes of certain mass peaks were not only toxin specific, but also depended on the used concentration. Consequently, a set of peaks could be identified as a biomarker for a certain toxin or chemical class. Hence, IC-MS could be a valuable tool for the profiling of toxic effects and risk assessment of ecosystems. Since it is a fast, easy-to-use, and automatable method, it is suited for the high throughput of environmental samples as well. The fast identification of the substances that are responsible for the contamination could prevent further pollution or initiate environmental remediation. However, there are still several steps necessary to evaluate complex samples that could eventually identify molecular mechanisms that are related to peak changes in IC-MS. Other metal compounds, PAH, NSO-Hets, and structural analogs, have to be analyzed to look for specific or common changes of spectral patterns. As for now, it is not possible to quantify the toxic effects with the presented method, but new matrices or internal standards may offer the opportunity to directly relate toxin concentration and ion abundances. Further research is needed to identify the potential whether the developed method could be used as an effect-based tool in the context of the different monitoring programs linking chemical and ecological status assessment within the water framework directive [39].