Introduction

The use of pesticides has improved agricultural productivity by controlling weeds, pests and pathogens that are harmful to agricultural products. However, long-term and intensive pesticide use has caused environmental pollution and ecological side effects [1].

Endosulfan, the broad-spectrum organochlorine insecticide, has been widely used on various crops for many years [2], and it produces endosulfan sulfate as a major oxidative metabolite. Due to the semivolatility and persistence of endosulfan and endosulfan sulfate, they are still currently detected in large areas of various environments [3, 4]. Endosulfan sulfate has a similar toxicity and similar or higher persistence than endosulfan [5,6,7].

Metabolomics explores the dynamic and broad response of organisms to reveal the effects and toxic mechanisms of certain chemical compounds, and enables understanding of the impacts of these chemicals on organisms and the environment [8]. Conventional toxicological assessment variables, such as behavioral changes and mortality, are inefficient in distinguishing among toxic substances that produce similar phenotypes [9]. Changes in metabolic products indicate toxicity, genetic modification, and disturbance due to environmental factors and disease conditions [10]. Metabolomics studies are typically performed using mass spectrometry or nuclear magnetic resonance (NMR) for the analysis of metabolites [11]. NMR is a relatively simple and convenient method but has low sensitivity and resolution [12, 13]. Mass spectrometry, on the other hand, is able to detect many metabolites with high sensitivity and selectivity [12].

Fish are often used as test organisms for measuring water pollution among various ecotoxicity models [14]. Carp (Cyprinus carpio) has a number of advantages over other small fish models, such as easy rearing over a long period [15]. In the present study, carp was exposed to endosulfan sulfate to understand its toxicological effects at the molecular level. Therefore, the metabolic profiles of dead or live carp were investigated using targeted GC–MS analysis, and the results were compared via metabolic pathway analysis.

Materials and methods

Chemicals and reagents

Endosulfan sulfate was obtained from ChemService Inc. (West Chester, PA). Ribitol and the MSTFA reagent (1% TMCS) were purchased from Wako (Tokyo, Japan) and Thermo Fisher Scientific Co. (Pittsburgh, PA), respectively. Methoxyamine hydrochloride, pyridine, dimethyl sulfoxide (DMSO) and N,N-dimethylformamide (DMF) were purchased from Sigma-Aldrich (St. Louis, MO). Methanol of HPLC grade was from Merck (Darmstadt, Germany), and all of the reagents were the highest grade products.

Experimental animals

Carp approximately 3–5 cm in total length were acclimated for 2 weeks in a glass water tank with dechlorinated tap water at 26 ± 1 °C. The carp were fed a commercial fish food once a day during the acclimation period. They were not fed for 1 day prior to the chemical exposure. A constant 16:8-h light:dark cycle photoperiod was used during experiment.

Comparison of the metabolite extraction efficiency of different extraction solvents

The acclimated carp without any treatment were frozen in liquid N2 and then ground into a fine powder using a mortar and pestle. Each 100 mg of the homogeneous sample powder was vortexed for 10 min at 4 °C with 1 mL of 50% methanol, 80% methanol, or dichloromethane/methanol/water (2/5/2) solutions. The mixtures were then centrifuged at 13,000 rpm for 10 min, and 100 μL of the supernatant was dried using a speed vacuum concentrator (Hanil Modulspin 40). The dried residue was reacted with 50 μL of the methoxyamination reagent (20 mg/mL of methoxyamine hydrochloride in pyridine) at 37 °C for 90 min with shaking, and then treated with 50 μL of the MSTFA reagent (MSTFA + 1% TMCS) at 37 °C for 30 min.

Chemical exposure

Endosulfan sulfate was dissolved in DMSO and diluted to concentrations of 8 mg/L for the stock solution. Either 1 mL of the stock solution or 1 mL of DMSO as a control was added to 1 L of water in 1 L beakers. After 15 carp per test group were distributed into each beaker, the live carp in each group were sampled after 2 days, and the dead carp from the exposed group were sampled immediately after death for up to 2 days.

Sample preparation

The whole body of the live or dead carp sample was frozen in liquid N2 and ground into a fine powder using a mortar and pestle as previously described [1]. 100 mg of the ground sample was vortexed with 1 mL of 50% methanol solution containing 0.2 μg/mL of ribitol as an internal standard for 10 min at 4 °C. After extraction of the samples, the derivatization process was conducted as described in the ‘comparison of the metabolite extraction efficiency of different extraction solvents’ section above.

Profiling and identification of metabolites by GC–MS

The metabolites were scanned using a GC–MS (Shimadzu QP2010) and a SIM-library of 373 metabolites (311 metabolites from the Smart Metabolites Database, Shimadzu, and 62 metabolites from an in-house library made by our group) was applied to identify metabolites from the live and dead carp. Split (10:1) mode was used on an injection of a 1.0 μL aliquot into a DB5 column (30 m × 0.25 mm i.d., 1 μm film thickness, Agilent). The GC–MS parameters were set to optimal conditions using an injector temperature of 280 °C, and ion source and interface temperatures of 230 °C and 280 °C, respectively. The initial oven temperature was 100 °C for 4 min, and the temperature was then increased to 320 °C at a rate of 4 °C/min and held there for 11 min. The flow rate of the helium carrier gas was 1 mL/min, and the electron ionization energy was 70 eV.

GCMS Solution software (version 4.3, Shimadzu) was used for data processing. The peak reconfirmation was carried out by manually examining the selected ions and the peak retention times of the corresponding metabolites. The relative area of the individual metabolites was calculated by comparing the peak areas of the metabolites and the internal standard, ribitol.

Metabolic pathway analysis

The processed data were analyzed using a one-way ANOVA (analysis of variance) and the fold changes of metabolites in dead carp versus control was calculated to identify the metabolites that were significantly affected (P < 0.001, fold change > 1.5 or < 0.667) by exposure to endosulfan sulfate. A heat map was generated with MetaboAnalyst 4.0 (http://www.metaboanalyst.ca) using the relative areas of the metabolites that had been selected as biomarkers in the ANOVA analyses. Metabolic pathway analysis plots were produced with MetaboAnalyst 4.0 and the metabolic pathways were identified based on the Cyprinus carpio KEGG library.

Results and discussion

Chemical exposure level

The exposure concentration of endosulfan sulfate was set at 8 μg/L to ensure the death of carp and understand the toxicological phenomenon. This concentration was twice the concentration of the 48-h LC50 value (2.6–5.0 μg/L) for four fish species [5].

Selection of the metabolite extraction solvent

Three extraction solvents, 50% methanol, 80% methanol and dichloromethane/methanol/water (2/5/2), have been frequently used in metabolomics studies of fish [1, 16, 17]. Therefore, we selected those three extraction solvents to compare their extraction efficiency of metabolites in carp. The 50% methanol was chosen as the extraction solvent in this study because it showed better peak shape and extraction efficiency than the other two solvents when peak shape and area were compared (Table 1). A previous study comparing the extraction method of metabolites in fish also showed that the yield of metabolites was higher with 50% methanol compared with the other various solvents [18].

Table 1 Total peak area of 146 metabolites in each extraction solvent
Full size table

Metabolite profiling using GC–MS

After the GC–MS analyses, metabolites were identified using an SIM-library of 373 metabolites included 16 alcohols, 7 amides, 7 amines, 63 amino acids, 74 carbohydrates, 2 esters, 44 fatty acids, 1 glyceride, 4 indoles, 8 nucleosides, 107 organic acids, 9 phenols, 1 phosphoric acid, 13 purines, 4 pyridines, 3 pyrimidines, 6 steroids, 1 terpene and 3 tocopherols. In the live and dead carp, 146 metabolites were detected as confirmed metabolites (Table 2, Fig. 1). The relative area of the individual metabolites compared to ribitol was subjected to a one-way ANOVA and fold change calculation for the determination of biomarkers.

Table 2 Metabolites detected in whole-body carp samples by targeted metabolite analyses using GC–MS
Full size table
Fig. 1
figure 1

Total ion chromatogram (TIC) of metabolites in live carp sample under exposure to 8 μg/L of endosulfan sulfate (a) and individual chromatograms of representative metabolites and the internal standard, ribitol (b)

Full size image

Metabolic pathway and function analysis

ANOVA analyses (P < 0.001) of the 146 metabolites was performed and fold change of them in dead carp compared to control (fold change > 1.5 or < 0.667) was calculated to obtain a total of 30 metabolites as biomarkers that contributed significantly to distinguish the metabolic profiles among the test groups (Table 3). The heat map with dendrograms of the 30 metabolites was generated using MetaboAnalyst 4.0 software, allowing for an intuitive identification of changes in the altered metabolite levels between the control and live and dead carp after exposure (Fig. 2). The metabolite levels are shown ranging in color from red (highest) to blue (lowest). In the hierarchical clustering analyses of the differential levels of metabolites associated with those groups, which were performed by the sample (columns) and variables (rows), a clear separation between test groups was confirmed.

Table 3 Significantly altered metabolites in carp exposed to endosulfan sulfate
Full size table
Fig. 2
figure 2

Heat map and hierarchical clustering of differentially expressed metabolites in the control, live and dead groups after exposure

Full size image

Metabolic profiling is able to show changes in each metabolite level and reveal a comprehensive interpretation of the metabolic processes due to toxic xenobiotics [19]. The pathway analysis module of MetaboAnalyst 4.0 contains new algorithms and concepts with the high-quality Kyoto Encyclopedia of Gene and Genomes (KEGG) and well-established methods [16]. In this study, metabolic pathway analyses were performed with 30 biomarkers using the pathway analysis module of MetaboAnalyst 4.0 (Fig. 3) [20]. As a result, five metabolic pathways (starch and sucrose metabolism, galactose metabolism, glycerolipid metabolism, the citrate cycle and linoleic acid metabolism) were identified as the major disturbance pathways. To compare the relative amounts of the metabolites that showed significant changes between the groups, the mean relative area of the individual metabolites that related to each metabolic pathway were graphed as histograms (Fig. 4). In the live and dead carp exposed to endosulfan sulfate, the metabolites associated with energy metabolism were significantly altered, and all of the metabolites associated with each metabolic pathway have a consistent trend that decreases from the control group to the dead group.

Fig. 3
figure 3

Summary of pathway analyses with MetaboAnalyst 4.0, as visualized by bubble plots. The bubble size is proportional to the impact of each pathway, and the bubble color denotes the degree of significance, from the highest (red) to the lowest (white)

Full size image
Fig. 4
figure 4

Average relative areas of typical significant metabolites that are related to each metabolic pathway: starch and sucrose metabolism (a), galactose metabolism (b), glycerolipid metabolism (c), the citrate cycle (d), and linoleic acid metabolism (e). In the graphs, the red bars are the control groups, the green bars are the live groups and the blue bars are the dead groups. Significantly different changes of P < 0.01 and P < 0.001 in the relative areas are labeled by “*” and “**”, respectively

Full size image

In starch and sucrose metabolism, the levels of glucose, maltose and trehalose were decreased in the live carp after endosulfan sulfate exposure. This significant downregulation in the live carp showed that their energy production increased to try to survive in the toxic environment [9]. However, those metabolites were almost exhausted in the dead group (Fig. 4a).

The levels of galactose and mannose in galactose metabolism decreased in the exposure groups (Fig. 4b). These sugars may be broken down into lactate and malate for producing additional energy [21]. Reduction of these sugars has also been demonstrated in the lungs of asthmatic mice, which may contribute to the development of airway inflammation and the severity of the disease [21].

In glycerolipid metabolism, glyceric acid and glycerol-3-phosphate were reduced in the carp after endosulfan sulfate exposure (Fig. 4c). Glyceric acid is a branch point between the glycerolipid and carbohydrate metabolic pathways, and this metabolite might be utilized to mobilize glycerolipids as an energy source through β-oxidation [22, 23].

With respect to the citrate cycle, citric acid and phosphoenolpyruvic acid were significantly decreased in the dead carp, while in the live carp there were no significant differences compared to the control group (Fig. 4d). Citric acid actually increased slightly in the live group, indicating that the citrate cycle was affected by endosulfan sulfate exposure [19]. In a previous study of zebrafish infection [24], citrate cycle intermediates showed the same tendency to decrease in the dead group and increase in the live group. These interesting results suggested that the status of the citrate cycle could be a biomarker for the survival or death of carp.

In linoleic acid metabolism, linoleic acid was significantly reduced in both the live and dead groups (Fig. 4e). Linoleic acid plays important physiological and biochemical functions in freshwater fish [25]. In a study of gilthead seabream, low levels of linoleic acid in plasma were related to the lean muscle phenotype [26]. Therefore, decreasing levels of linoleic acid indicated the dysfunction of energy metabolism.