Metabolic disturbance in hippocampus and liver of mice: A primary response to imidacloprid exposure

Imidacloprid (IMI) is one of the most frequently used neonicotinoid insecticides, but recent studies have shown adverse effects on mammals. IMI was found to be neurotoxic and hepatotoxic. In the present study, the effects of repeated oral administration of two doses of IMI (5 and 20 mg/kg/day) for 28 days on hippocampus and liver of female KM mice were studied. The histopathological and biochemical experiments indicated obvious damages to the hippocampus and liver of mice in the high-dose group (20 mg/kg/day). Using a high-throughput metabolomics platform based on ultrahigh performance liquid chromatography/hybrid quadrupole time-of-flight mass spectrometry (UPLC/Q-TOF MS), we studied effects of IMI on metabolic profiles in the hippocampus and liver of mice. Significant differences among the control group, the low-dose group and the high-dose group were clearly presented using multivariate analysis. The changed metabolic profile in the low-dose group (5 mg/kg/day) revealed that the metabolic disturbance in the hippocampus and liver of mice had been induced by low-dose of IMI, although no significant histopathological changes were observed in the low-dose group. Six differential metabolites in the hippocampus and 10 differential metabolites in the liver were identified as the possible biomarkers to distinguish IMI exposure from the control group using the variable importance in projection (VIP) value and receiver operating characteristic (ROC) analysis. The metabolism disturbances of important biochemical pathways in the hippocampus and liver of mice in the exposed groups were elucidated, mostly concentrated in lipid metabolism, amino acid metabolism, nucleotide metabolism, carbohydrate metabolism, and energy metabolism (p < 0.05). Such investigations give out a global view of IMI-induced damages in the hippocampus and liver of mice and imply a health risk associated with early metabolic damage in mice.

Behavior analysis. All mice underwent a Morris water maze to test their ability of learning and memory.
The test process was divided into the six-days of space exploration training (Fig. 1A) and probe test on the seventh day ( Fig. 1A-D). The results were analyzed and evaluated by recording the escape latency and total distance of   Table 1. Body weight and relative organ weight of mice. Data presented as mean ± SE; N = 10 per group. Relative organ weight = organ weight/body weight × 100. *p < 0.05; ***p < 0.001.
Histopathological examination. The histopathological examinations of the hippocampus and liver of mice were carried out using hematoxylin-eosin staining. The results of the hippocampus were presented in Fig. 2A-C. Compared with the control group, there was no significant change in the low-dose group. However, the cell band of the hippocampal formation was significantly absent in the high-dose group, suggesting an IMI-induced toxicity in the brain of mice at the dose of 20 mg/kg/day. The histopathological examination results of the liver were presented in Fig. 2D-F. The results clearly showed that there was no significant change between the control group and the low-dose group. Compared with the control group, the liver appeared obvious vascular swelling in the high-dose group. The hepatic cords arranged around the central vein were disordered and liver cell necrosis and nuclear pyknosis were observed. The results above indicated that IMI produced significant subacute toxicity in the brain and liver of mice at 20 mg/kg/day which were in agreement with previous findings in Badgujar's study 31 and Arfat's study 19 .
Biochemical analysis. To investigate the effects of IMI exposure on the functional activity of liver enzymes, the plasma biochemical parameters of mice orally administered IMI including alanine aminotransferase (ALT), aspartate aminotransferase (AST) and alkaline phosphatase (AKP) were examined. The results were listed in Table 2. Compared with the control group, there was no significant change in the low-dose group. However, oral administration of IMI to mice at the high dose for 28 days resulted in significant increases of the plasma ALT, AST and AKP, implying the liver cell injury and degeneration.
Quality control validation and UPLC/Q-TOF MS data processing. Pearson correlation analysis was performed on quality control (QC) samples to evaluate the system stability during UPLC/Q-TOF MS instrument analysis process and the results were presented in Supporting Information Fig. S1. As shown in Fig. S1, all the correlation coefficients (r) for the hippocampus and liver QC samples in positive and negative models were more than 0.9, indicating that the analysis system was stable and all the data were under control. www.nature.com/scientificreports www.nature.com/scientificreports/ According to the UPLC/Q-TOF MS results, 6444 ion peaks for hippocampus and 11652 ion peaks for liver were detected and processed using XCMS program (https://xcmsonline.scripps.edu/landing_page.php?pgcon-tent=mainPage) for feature detection, retention time (t R ) correction and alignment. The identification of the metabolites was operated in terms of accurate m/z value (<25 ppm) and MS/MS spectra by comparing with available reference standard mass spectral databases, Mass Bank (http://www.massbank.jp/) and Metlin library (http://metlin.scripps.edu/), and an in-house database (Shanghai Applied Protein Technology) established with authentic standards. In total, 116 endogenous metabolites from hippocampus and 50 endogenous metabolites from liver were identified which exhibited statistically significant differences (p < 0.05) among 3 groups. Heap  www.nature.com/scientificreports www.nature.com/scientificreports/ maps produced by clustering of these differential metabolites were presented in Figs. S2 and S3. Among them, 47 metabolites from hippocampus and 31 metabolites from liver exhibited statistically significant differences (p < 0.05) between the control group and the exposed groups (Tables S1 and S2).

Multivariate analysis of UPLC/Q-TOF MS data.
The supervised partial least squares-discriminant analysis (PLS-DA) was used for modeling differences among 3 groups. The PLS-DA score plots and permutation tests for the hippocampus and liver samples in positive and negative models were shown in Figs. 3 and 4. It could be seen that all data points in the PLS-DA score plots were located within the Hotelling T2 (95%) ellipse. Permutation tests were performed to evaluate the fitting degree of PLS-DA after modeling the data. R 2 stands for the explanation capacity of the model and Q 2 stands for the predictive capacity of the model 32 . The low values of intercepts, R 2 (<0.520) and Q 2 (<−0.185), showed that the built models were not over-fitting (Figs. 3 and 4)). The correct rates of these models were more than 83.33%. The score plots presented a clear separation among 3 groups both in the hippocampus and liver, indicating that exposure to IMI induced metabolism disturbances in the hippocampus and liver of mice. Note that, compared with the control group, the metabolic profile in the low-dose group of 5 mg/kg/day had changed, although no significant histopathological changes were observed between the control group and the low-dose group (Fig. 2). This revealed a metabolism disruption and a health risk associated to early metabolic injury by exposure to low-dose of IMI.
Potential biomarkers identification. The potential biomarkers for IMI exposure have been comprehensively analyzed and filtered by the variable importance in projection (VIP) value and receiver operating characteristic (ROC) analysis. The ROC curve was constructed by plotting the sensitivity against the corresponding false-positive rate (1-specificity). The differential metabolites with the area under curve (AUC) > 0.70 and with VIP value > 1.0 were selected as potential biomarkers. The sensitivity, specificity, and youden index for the selected cut-off of each parameter were given in Table 3.
As described in Table 3, in the hippocampus, 3 metabolites including acetylcholine, sn-glycerol 3-phosphoethanolamine and L-phenylalanine were identified as the possible biomarkers to distinguish low-dose of IMI exposure from the control group. When the exposure dose of IMI increased, the metabolic profile changed. This led to significant changes in the concentration of 4 metabolites including acetylcholine, D-lyxose, adenine and GMP, which could be used as the potential biomarkers. It's worth noting that with the increase of the exposure dose of IMI, the difference of acetylcholine level was more obvious. Compared with the low-dose group, the higher AUC (1.000) and youden index (1.000) of acetylcholine were obtained between the control group and the high-dose group. This revealed that acetylcholine as the potential biomarker could provide a satisfactory distinction of IMI exposure from the controls and a better prediction of early metabolic damage of an organism induced by IMI exposure. The orthogonal partial least squares-discriminant analysis (OPLS/O2PLS-DA) loading plots were employed to further identify the variables that strongly contribute to the separation of classes. The x axis, the www.nature.com/scientificreports www.nature.com/scientificreports/ loading p, denotes the influence of the variables and the y axis (po) denotes the credibility of the variables 33 . From Figs. S4 and S5, we could still detect these potential biomarkers concentrated in the lower left corner of the plots, which exhibited a reliability of identified biomarkers.   Table 3. Potential biomarkers for IMI exposure discovered by VIP value and ROC analysis. Abbreviations: AUC, area under curve; GMP, Guanosine 5′-monophosphate; AMP, Adenosine monophosphate. *p < 0.05; **p < 0.01; ***p < 0.001.
In the liver, 7 metabolites including L-glutamine, S-lactoylglutathione, inosine, AMP, L-pyroglutamic acid, adenine and 2-hydroxyadenine were identified as the possible biomarkers to distinguish low-dose of IMI exposure from the control group. Among them, L-glutamine was discovered with the highest AUC (1.000) and youden index (1.000) between the control group and the low-dose group. The excellent sensitivity and specificity of L-glutamine revealed that L-glutamine was expected to be a biomarker to predict early metabolic damage by exposure to low-dose of IMI. With the increase of the exposure dose of IMI, the metabolic profile changed. Six metabolites including guanosine, UDP-N-acetylglucosamine, L-pyroglutamic acid, adenine, 2-hydroxyadenine and 1-stearoyl-2-hydroxyl-sn-glycero-3-phosphocholine were identified as the possible biomarkers to distinguish high-dose IMI exposure from the control group. Among them, 3 metabolites including L-pyroglutamic acid, adenine and 2-hydroxyadenine were selected at two exposed concentrations at the same time, implying their good monitoring and early warning for IMI exposure. In addition, as shown in Figs. S6 and S7, these potential biomarkers were found to be furthest from the origin, exhibiting a reliability of identified biomarkers.

Metabolic pathway and function analysis.
Based on the results of differential metabolites, we further investigated the relevant metabolic pathways affected by IMI in the hippocampus and liver. The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis [34][35][36] and Fisher's Exact Test were employed to analyze and calculate the significance level of differential metabolite enrichment in each pathway, so as to determine the metabolism pathways that were significantly affected. The pathway enrichment analysis results (Tables S1 and S2) showed that in the low-dose group, 27 differential metabolites in the hippocampus participated in 6 metabolic cycles including lipid metabolism, amino acid metabolism, nucleotide metabolism, carbohydrate metabolism, energy metabolism, and metabolism of cofactors and vitamins (p < 0.05), while 22 differential metabolites in the liver mainly participated in 4 metabolic cycles including amino acid metabolism, nucleotide metabolism, carbohydrate metabolism, and metabolism of other secondary metabolites (p < 0.05). The alteration in these metabolic pathways might be the initial response to IMI exposure, since histopathological changes had not been induced yet. The disruption of certain metabolic pathways might return to normal through compensatory responses. However, certain metabolic disorders might be exacerbated with the increase of exposure time, eventually leading to tissue damage in mice. It could be seen from Fig. 5, when exposed to high-dose of IMI, glycerophospholipid metabolism, 5 amino acid metabolism pathways, pyrimidine metabolism, and pantothenate and CoA biosynthesis were still affected, whereas the disturbance of biosynthesis of unsaturated fatty acids and purine metabolism were appeared in the hippocampus. In the liver (Fig. 6), purine metabolism was still affected by exposure to high-dose of IMI, whereas the disturbance of biosynthesis of unsaturated fatty acids, D-glutamine and D-glutamate metabolism, glyoxylate and dicarboxylate metabolism, and pyruvate metabolism were appeared.
The glycerophospholipid metabolism in the hippocampus of mice was found to be affected by exposure to IMI at two exposed concentrations. In this metabolic pathway, phosphocholine, choline, phosphoethanolamine, and triethanolamine were significantly downregulated (p < 0.05), whereas acetylcholine was significantly upregulated (p < 0.05) in the high-dose group. This suggested a disruption of cholinergic system relating with the mode of www.nature.com/scientificreports www.nature.com/scientificreports/ action of IMI 28 . Acetylcholine was identified as a potential biomarker for IMI exposure due to its excellent sensitivity and specificity. The up-regulation of acetylcholine might be due to that IMI bound competitively to the acetylcholine receptors (nAChRs) on the postsynaptic membrane of neuronal cells, leading to the accumulation of acetylcholine. The decrease of choline and the increase of acetylcholine implied the inhibition of acetylcholinesterase (AchE) and choline O-acetyltransferase (ChAT).
The dysregulation of biosynthesis of unsaturated fatty acids in the hippocampus and liver were appeared in the high-dose group. The levels of arachidonic acid and eicosapentaenoic acid were found to be decreased, whereas an increase in oleic acid and linoleic acid were found. Derivatives of long-chain polyunsaturated fatty acids such as arachidonic acid and eicosapentaenoic acid were important mediators of inflammation 37 . Inflammatory mediators derived from eicosapentaenoic acid such as resolvins and protectin had anti-inflammatory effects 38 . Arachidonic acid was mainly used to synthesize pro-inflammatory mediators. After binding to inflammatory cell receptors, arachidonic acid activated intracellular inflammatory signal transduction and promoted the synthesis and release of inflammatory factors such as tumor necrosis factor and interleukin, leading to amplification of the inflammatory response 39 .
An alteration in many amino acids in the hippocampus and liver were observed, involving arginine biosynthesis, alanine, aspartate and glutamate metabolism, D-glutamine and D-glutamate metabolism, glycine, serine and threonine metabolism, valine, and leucine and isoleucine biosynthesis. The metabolism including amino acids was known as an important part of neurotransmitter processing 40 . The levels of 9 amino acid neurotransmitters involving L-aspartate, L-glutamate, L-glutamine, L-isoleucine, L-leucine, taurine, L-serine, L-phenylalanine and γ-aminobutyric acid (GABA) were found to be significantly downregulated (p < 0.05). Especially, L-phenylalanine could be used as a potential biomarker to assess the metabolic disturbance in the hippocampus caused by exposure to low-dose of IMI.
Additionally, the disturbance of amino acid metabolism could affect the gut-associated immune system 41-43 and initiate gut inflammation [44][45][46] via the modulation of key metabolic signaling pathways. Changes in some amino acids such as L-glutamine, L-glutamate, L-aspartate, L-leucine and L-serine could effectively impact bacterial composition, diversity and activity, and modulate macrophages and dendritic cells 47 . Meanwhile, L-glutamine could regulate the compartmental amino acid metabolism via a variety of pathways, thus further affecting the amino acid composition and content in the gut 48 .
The metabolism of other amino acids was also dysregulated by exposure to IMI involving β-alanine metabolism, taurine and hypotaurine metabolism, and the glutathione metabolism. The glutathione metabolism was usually correlated to oxidative stress 49 . Glutathione was involved in the process of cell membrane protection and free radical scavenging. It could help cells maintain normal immune system functions, and had antioxidant and integrative detoxification effects. The alteration of glutathione metabolism suggested that IMI could trigger the overproduction of free radicals leading to oxidative stress in the hippocampus and liver of mice, which was in agreement with previous findings in Duzguner's studies 50,51 . The levels of L-pyroglutamic acid, Figure 6. Disturbed biochemical pathways for the comparison between the control group and the exposed groups in the liver. The color and shape of the nodes represent the fold changes between different groups: Red, up-regulation; Green, down-regulation; Gray, no statistical difference; Triangle, Control-Low; Diamond, Control-High; Dot, Control-Low and Control-High. The bold without borders represents the selected biomarkers.
In addition, nucleotides were involved as well and, as a consequence, purine metabolism and pyrimidine metabolism were dysregulated by exposure to IMI. In the hippocampus, the levels of adenine and GMP were found to be increased, whereas the levels of adenosine, hypoxanthine and xanthine were found to be decreased in the high-dose group. In the liver, the levels of adenine, guanosine and ADP-ribose were found to be increased, whereas the level of xanthine was found to be decreased in the high-dose group. Among them, 3 differential metabolites involving adenine, GMP and guanosine were identified as the potential biomarkers for IMI exposure due to their good sensitivity and specificity.
From the metabolomics results, there was no significant difference in hormones between the control group and the exposed groups, and the disturbance of hormone-related pathways were not appeared by GEGG pathway enrichment analysis. These exhibited the intergroup consistency of hormone levels and indicated that repeated oral administration of IMI at two different doses did not cause changes in hormone levels, which were in agreement with previous findings in Vohra's study 16 and Kapoor's study 52 .

conclusions
In the present study, the histopathological, biochemical and metabolic alterations induced by oral administration of two doses of IMI (5 and 20 mg/kg/day) for 28 days on hippocampus and liver of mice were studied. Compared with the control group, the hippocampus and liver of mice appeared significant histopathological changes in the high-dose group (20 mg/kg/day). Although no obvious damages were observed in the low-dose group, multivariate statistical analysis presented 3 different metabolic profiles for the control group, the low-dose group and the high-dose group. The altered metabolic profile in the low-dose group (5 mg/kg/day) indicated that the metabolism disturbance in the hippocampus and liver of mice had been induced by low-dose of IMI, implying a health risk associated with early metabolic damage in mice. Six major metabolic cycles in the hippocampus and 4 major metabolic cycles in the liver were affected by exposure to IMI. Six metabolites in the hippocampus and 10 metabolites in the liver were discovered as the possible biomarkers for IMI exposure. The neurotransmitter acetylcholine in the hippocampus and L-glutamine and S-lactoylglutathione associated with oxidative stress in the liver as the potential biomarkers (AUC > 0.9; youden index > 0.8) could provide a satisfactory prediction of early metabolic damage of an organism induced by IMI exposure. Such investigations give out a global view of IMI-induced damages in the hippocampus and liver of mice. Our results may facilitate further research aimed at understanding the toxic mechanisms of IMI on mammals and developing the early intervention of IMI exposure combined with useful biomarkers.

Materials and Methods
Chemicals and reagents. Technical grade IMI (>98% purity), MS-grade ammonium acetate, and MS-grade ammonium hydroxide were purchased from Sigma-Aldrich (Schnelldorf, Germany). HPLC-grade acetonitrile and methanol were purchased from Merck (Darmstadt, Germany). Milli-Q water was obtained from a Millipore purification system (Millipore, Milford, MA, USA).

Animals and diet. Animal experiments were approved by the Ethics Committee of Chongqing Medical
University (Chongqing, China), and all experiments were carried out in accordance with the approved guidelines. Female KM mice (6-8-week-old), individual-ventilated-cages grade, were obtained from Chongqing Medical University Laboratory Animal Center. Female mice were chosen because female mice were reported more sensitive to the toxic effects of neonicotinoids 18,53,54 and have been widely used in metabolomics studies of the toxic effects induced by environmental pollutants [55][56][57][58] . The mice were housed in a climate-controlled room with a temperature of 20-25 °C, a humidity of 30-60%, and a 12 h light/dark cycle. The mice were given standard diet containing pelleted food and water ad libitum. All mice were acclimatized for one week before using them for experiments.
Experimental design and sample collection. After acclimatization for one week, the mice were randomly divided into 3 groups. One group was served as control and was given corn oil as vehicle through oral intubation. The other two groups were given 5 and 20 mg/kg/day (corresponding to 1/30th and 1/7.5th LD 50 of IMI) IMI suspended in corn oil to mice for 28 days, respectively. Body weights were recorded throughout the period of experiments. After 28-day exposure, the mice of each group were used in behavioral study. Then, blood samples were collected from the retro-orbital sinus in heparin anticoagulation tube. Finally, all the mice were sacrificed by cervical dislocation. The mice used in the metabolomics analysis were in the stage of estrus by visual observation of the external genitalia (the vaginal opening was swollen, pink and moist) described by Champlin's group 59 . The brain and liver were quickly harvested and weighted. The hippocampus and liver samples were rinsed with phosphate-buffered saline (PBS) and stored at −80 °C until analysis.
Morris water maze test. A circular stainless steel poor was divided into 4 quadrants marking with square, triangle, circle and tree, respectively. Flour was used to turn the water white and the water temperature was kept at 23 ± 2 °C. A 9-cm-diameter escape platform was positioned in one of the quadrants submerged 0.5-1 cm above the surface of water. Screened mice had vision problems or did not swim were excluded for following experiments. The mice were trained 4 times a day at 4 different starting positions facing the pool wall with 20-min intervals for 6 days. In each trial, mice were given 60 s to find the platform. If the mice could not find the platform within the specified time, they were placed directly on the platform for 15 s. The swim speed, distances (path length), and time (escape latency) to the platform were monitored and calculated with a video camera linked to a computer system using MWM software (WMT-100S Morris water maze, Techman, Sichuan, China). On day 7, a probe test was performed to evaluate spatial memory ability. The platform was removed from the pool and Statistical analysis. Statistical analyses were performed using SPSS 22.0 software (IBM Corporation, Armonk, NY, USA). After the determination of the normality of the data distribution, one-way "analysis of variance (ANOVA)" with post-hoc Fisher's least significant difference (LSD) test was applied to assess statistical differences among multiple groups. A p < 0.05 was considered to be statistically significant. The results were reported as the mean value ± standard error (SE) and displayed using GraphPad Prism version 5.0 statistical software (GraphPad, Inc., La Jolla, USA).
Partial least squares-discriminant analysis (PLS-DA) was carried out using SIMCA-P 14.1 (Umetrics, Umea, Sweden) to model differences among the control group and two exposed groups. The potential biomarkers for IMI exposure were comprehensively analyzed and filtered by variable importance in projection (VIP) value, receiver operating characteristic (ROC) analysis and orthogonal partial least squares-discriminant analysis (OPLS/ O2PLS-DA) loading plots. Hierarchical clustering analysis was performed using the open-source cross-platform software Cluster 3.0 (http://bonsai.hgc.jp/~mdehoon/software/cluster/software. htm) coupled with Java Treeview package (http://www.java.com/). The altered metabolism pathways were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The disturbed biochemical pathways and key metabolites were displayed by Microsoft Office Visio 2007 (Microsoft Corporation, Redmond, WA, USA).