NMR-based metabolomics approach to study the toxicity of lambda-cyhalothrin to gold

Aquatic Toxicology 146 (2014) 82–92Contents lists available at ScienceDirectAquaticToxicologyj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /a q u a t oxNMR-based metabolomics approach to study the toxicity of lambda-cyhalothrin to goldfish (Carassius auratus )Minghui Li a ,Junsong Wang b ,∗∗,Zhaoguang Lu a ,Dandan Wei a ,Minghua Yang a ,Lingyi Kong a ,∗aState Key Laboratory of Natural Medicines,Department of Natural Medicinal Chemistry,China Pharmaceutical University,24Tong Jia Xiang,Nanjing 210009,PR China bCenter for Molecular Metabolism,School of Environmental &Biological Engineering,Nanjing University of Science &Technology,200Xiao Ling Wei Street,Nanjing 210094,PR Chinaa r t i c l ei n f oArticle history:Received 30August 2013Received in revised form 22October 2013Accepted 24October 2013Keywords:Lambda-cyhalothrin Metabolomics 1H NMR spectroscopyGoldfish (Carassius auratus )a b s t r a c tIn this study,a 1H nuclear magnetic resonance (NMR)based metabolomics approach was applied to inves-tigate the toxicity of lambda-cyhalothrin (LCT)in goldfish (Carassius auratus ).LCT showed tissue-specific damage to gill,heart,liver and kidney tissues of goldfish.NMR profiling combined with statistical methods such as orthogonal partial least squares discriminant analysis (OPLS-DA)and two-dimensional statisti-cal total correlation spectroscopy (2D-STOCSY)was developed to discern metabolite changes occurring after one week LCT exposure in brain,heart and kidney tissues of goldfish.LCT exposure influenced levels of many metabolites (e.g.,leucine,isoleucine and valine in brain and kidney;lactate in brain,heart and kidney;alanine in brain and kidney;choline in brain,heart and kidney;taurine in brain,heart and kidney;N-acetylaspartate in brain;myo -inositol in brain;phosphocreatine in brain and heart;2-oxoglutarate in brain;cis -aconitate in brain,and etc.),and broke the balance of neurotransmitters and osmoregulators,evoked oxidative stress,disturbed metabolisms of energy and amino acids.The implica-tion of glutamate–glutamine–gamma-aminobutyric axis in LCT induced toxicity was demonstrated for the first time.Our findings demonstrated the applicability and potential of metabolomics approach for the elucidation of toxicological effects of pesticides and the underlying mechanisms,and the discovery of biomarkers for pesticide pollution in aquatic environment.© 2013 Elsevier B.V. All rights reserved.1.IntroductionEnvironmental pollution caused by pesticides is a big problem in many developing countries,e.g.China and other agricultural countries all over the world (Zhang et al.,2011),where pesti-cides are excessively used to gain agricultural productivity.Most pesticides ultimately find their way into rivers,lakes and ponds (Arjmandi et al.,2010),showing detrimental effects on non-target organisms in natural habitats nearby agricultural fields.Pesticide residue in water environment is known to have ill effects on the growth,survival and reproduction of aquatic animals,causing seri-ous aquaculture loss.Even worse,through bioaccumulation,those at the top of the food chain,e.g.humans,are especially threatened by intake of contaminated aquatic creatures.∗Corresponding author.Tel.:+862583271405;fax:+862583271405.∗∗Corresponding author.Tel.:+862583271402.E-mail addresses:wang.junsong@ (J.Wang),cpu lykong@ ,lykong@ (L.Kong).Fishes,rich in proteins,lipids and other nutritions,are impor-tant for global food supply.Besides,they also play an essential role in ecological balance and energy flow in aquatic ecosystems.Fishes are particularly sensitive to water contamination,thus are suitable to act as bioindicators of environmental pollutants,monitoring the quality of aquatic systems (Adams and Greeley,2000).Exposure of fish to pesticides produced biochemical disturbances in fish (Gül et al.,2004).Pesticides could significantly damage physiological functions and disturb biochemical processes in fishes (Banaee et al.,2011).There have increasing reports on the death of fish in vari-ous streams,lakes and ponds around the world due to pesticides runoff associated with intense agricultural or industrial practices.Recently,many studies have been conducted to determine the mechanisms for the toxicity of pesticides in fishes,with the ulti-mate aim to monitor,control and possibly intervene in the effects of xenobiotics exposure on aquatic ecosystem.Pyrethroids are a class of insecticides derived of pyrethrins,naturally occurring insecticidal components in flowers of chrysan-themums (Chrysanthemum cinerariaefolium ).LCT (Fig.1)is one of the pyrethroid insecticides widely used worldwide in agricultural pest control due to its strong activity against a broad spectrum of0166-445X/$–see front matter © 2013 Elsevier B.V. All rights reserved./10.1016/j.aquatox.2013.10.024Edited by Foxit ReaderCopyright(C) by Foxit Software Company,2005-2008For Evaluation Only.M.Li et al./Aquatic Toxicology146 (2014) 82–9283Fig.1.The chemical structure of two isomers of lambda-cyhalothrin. pests(Mathirajan et al.,2000).Though in relatively lower toxicity to mammals than other insecticides,they can also do severe harm to aquatic organisms and humans(Haya,1989).The major toxicity of LCT is due to its interaction with membrane bound ion channels in neurons and disruption of nerve function by prolonging the open phase of the sodium channel gates,thus leading to hyperactivity of the nervous system,and eventually resulting in paralysis and/or death.Despite of wealthy information on the toxicity of LCT available, its toxic effects on the whole body,especially its disturbance on the metabolism of animals are still scarce.Following genomics, transcriptomics and proteomics,metabolomics is an imporatant ‘omic’science in systems biology,which holistically determines the global metabolites and thus detects metabolic events of a bio-logical system in response to disease,genetic and environmental perturbations.Metabolomics is widely applied in toxicity screening (Shockcor and Holmes,2002),disease diagnosis(Gowda et al., 2008),mechanistic study(Song et al.,2013;Tian et al.,2013),ther-apy and physiological monitoring(Lindon et al.,2004;Spratlin et al., 2009)as well as environmental sciences(Robertson et al.,2011). Metabolomics technologies offered new possibilities in the clar-ification of the modes of action of chemicals(Aliferis and Jabaji, 2011),and in the discovery of new biomarkers(Miracle and Ankley, 2005).Despite the promise of metabolomics,its application to evalu-ate the toxic effects of LCT has not been reported previously.The aim of the present study is to investigate under laboratory condi-tions the global metabolic response and specific changes in brain, heart,kidney and liver tissues of goldfish after sub-lethal exposure to LCT.2.Materials and methods2.1.ChemicalsLCT is a synthetic pyrethroid insecticide(C23H19C l F3NO3): CAS chemical name␣-cyano-3-phenoxybenzyl-3-(2-chloro-3,3,3-trifluoro-1-propenyl)-2,2-dimethyl cyclopropanecarboxylate,CAS registry number91465-08-6.A commercial formulation of micro-elusive LCT,named“Tiffe®”(Taifeng agrochemicals,Hangzhou, China)was used in the experiments.All other chemical products used in this study were purchased from Sigma Chemical Co.(St. Louis,MO,USA).2.2.Experimental animalsGoldfish(Carassius auratus∼12cm long,∼30g each)were pur-chased from a local market.The goldfish were fed with commercial fish food once a day and acclimated for one week in polycarbo-nate tanks(60L)filled with activated charcoal-filtered tap water of Nanjing city,China,refreshed every24h.2.3.Dosefinding experimentsExperimental concentrations of LCT were determined consid-ering the sensitivity of goldfish to the insecticide as observed in previous studies(Haya,1989;Köprücüand Aydın,2004).In a preliminary experiment tofind an appropriate exposure concen-tration,goldfish were exposed to a series of concentrations of LCT at7.5,1.5,0.3,0.06and0.012␮g/L for96h and the mortality was recorded.Goldfish exposed to0.012␮g/L LCT were all survived, while higher concentrations of LCT produced mortality over50% in3days.Therefore,0.012␮g/L was selected as thefinal exposure concentration of LCT.2.4.Exposure experimentAfter the acclimation period,goldfish were randomly sepa-rated into two groups of ten goldfish each,in the tankfilled with water containing0.012␮g/L LCT(LCT group)or not(control group). Seven days after chemical exposure,goldfish were sacrificed in accordance with the Animal Ethics Committee of China Pharmaceu-tical University and the Chinese council on animal care guidelines. Organs of liver,brain,heart and kidney were removed from the goldfish body,immediately immersed in liquid nitrogen and then stored at−80◦C until use for metabolomics analyses.2.5.Haematoxylin and eosin(H&E)stainingAn aliquot of organs were quickly removed,rinsed with cold phosphate buffered saline(PBS)and then immersed in10%neutral-buffered formaldehyde for24h,embedded in paraffin,and sliced into5␮m thickness.The sliced sections were stained with H&E, and examined under light microscopy.2.6.Sample preparation for1H NMR analysisMetabolites were extracted based on the reported protocol(Beckonert et al.,2007).Briefly,each sample was weighted,homogenized with an icy cold solvent (methanol/H2O/chloroform=4/2.85/4,v/v),and centrifuged at 1000×g,4◦C.Subsequently,the upper aqueous layer(containing polar metabolites)of each sample was transferred into fresh Eppendorf tubes and was then frozen and lyophilized until dryness on a vacuum concentrator.The dried samples were stored at−80◦C until use for NMR analysis.The dried samples were dissolved in 600␮L99.8%D2O phosphate buffer(0.2M,pH7.0)containing 0.05%(w/v)sodium3-(trimethylsilyl)propionate-2,2,3,3-d4(TSP). After vortexing and centrifugation to remove any debris,the supernatant was then transferred to a5mm NMR tube for1H NMR analysis.2.7.1H NMR spectroscopyAll1H NMR spectra were recorded at298K on a Bruker AV 500MHzflow-injection spectrometer.D2O was used forfield fre-quency locking and TSP was used as a chemical shift reference(1H,ı0.00).A transverse relaxation-edited Call-Purcell-Meiboom-Gill (CPMG)sequence(90( –180– )n-acquisition)with a total spin-echo delay(2n )of10ms was used to suppress the signals of84M.Li et al./Aquatic Toxicology146 (2014) 82–92proteins.1H NMR spectra were measured with128scans into32K data points over a spectral width of7500Hz.The spectra were Fourier transformed after multiplied the FIDs by an exponential weighting function corresponding to a line-broadening of0.5Hz.2.8.Pre-processing of NMR dataThe spectra for all samples were manually corrected for phase and baseline,and referenced to TSP(1H,ı0.00),using Bruker Topspin3.0software(Bruker GmbH,Karlsruhe,Germany).The 1H NMR spectra were automatically exported to ASCIIfiles using MestReNova(Version8.0.1,Mestrelab Research SL),which were then imported into“R”(/),and aligned with an in-house developed R-script.The one-dimensional(1D) spectra were converted to an appropriate format for statistical anal-ysis by automatically segmenting each spectrum into0.003ppm integrated spectral regions(buckets)between0.2and10ppm. Regions between4.5and5.0ppm,containing the residual water resonance,were excluded.All spectra were mean-centered and the integral values of each spectrum were probability quotient normal-ized to account for different dilutions of samples.2.9.Peaks assignmentsResonances were assigned by quering publicly accessible metabolomics databases such as Human Metabolome Database (HMDB,http://www.hmdb.ca),Madison-Qingdao Metabolomics Consortium Database(MMCD,/) and Kyoto Encyclopedia of Genes and Genomes(KEGG, http://www.kegg.jp),aided by Chenomx NMR suite7.5(Chenomx Inc.,Edmonton,Canada),andfinally confirmed by two-dimensional NMR techniques TOCSY(total correlation spectroscopy),HSQC (heteronuclear single quantum correlation).The assignments and integrations of peaks were aided by two-dimensional STOCSY (Cloarec et al.,2005;Maher et al.,2008),performed by a suite of in-house developed scripts running in“R”.2.10.Statistical analysisMultivariate statistical data analysis,including unsupervised principal component analysis(PCA)and supervised OPLS-DA meth-ods were performed by a suite of in-house developed scripts running in“R”software(/).OPLS-DA iden-tified most significant variations between the treatment groups. To assess the statistical significance of selected predictive quality parameters of the established OPLS-DA model,permutation testing and two-fold cross-validation(2CV)was carried out.Permutation testing is based on the comparison of the predictive capabilities of an OPLS-DA model using real class assignments to a number of models calculated after random permutation of the class labels. The validity of the models against overfitting was assessed by the parameters R2Y,and the predictive ability was described by Q2Y.Negative or very low Q2Y values indicate that the differ-ences between groups are not statistically significant.To ensure that discrimination in the OPLS-DA model was not the result of data overfitting,a validation of the model was performed using permutation testing(1000times).The observed statistic P val-ues via permutation testing which were less than0.05confirmed the significance of the OPLS-DA model at a95%confidence level. The OPLS-DA model parameters for brain,heart,kidney and liver are presented in the supporting information(Table S1,Fig.S1).A parametric Student’s t-test or a nonparametric Mann–Whitney test(dependent on the conformity to normal distribution)was performed on the signal integrals to evaluate the difference of metabolites between groups.Firstly,the integration areas of the detected metabolites with marked differentiating ability werefirst tested for the normality of the distribution.If the distribution followed the normality assumption,a parametric Student’s t-test was applied;otherwise,a nonparametric Mann–Whitney test was performed to detect statistically significant metabolites that were increased or decreased between groups,the associated P values were summarized in Table1.3.Results3.1.Effects of LCT on behavior and histopathologySelected tissues including gill,brain,heart,liver and kidney were histopathologically investigated(Fig.2).H&E stained tis-sue sections revealed tissue specific pathological changes in LCT treated goldfish.Fish gills,directly exposed to the environment, are especially vulnerable:shortening and lamellar fusion of sec-ondary lamellae,epithelial necrosis,desquamation were noticed after one week exposure to LCT(Fig.2F).For heart after seven days exposure,cardiac pericardium can be seen as inflammatory cell infiltrations(Fig.2H).Photomicrographs of the liver revealed that histopathological changes were not evident for treated group: slight inflammatory cell infiltrations in LCT group were observed (Fig.2I).One week of LCT exposure significantly damaged the renal histological structure(Fig.2J),including dilation of glomeru-lar capillary,vacuolar variations in endothelial cellular size,and interstitial inflammatory cell infiltrations.Behavior is the most reliable indicator of potential toxic effects on central nervous system.Goldfish exposed to LCT at0.06␮g/L or higher dosage showed behavioral abnormalities as evidenced by spastic movements(sudden and uncontrolled body movements), hyperactivity,erratic swimming,convulsion,partial/complete loss of equilibrium,jaw spasms and gulping respiration.3.2.STOCSY2D-STOCSY was used to identify correlations between spectral resonances of interest to assist metabolite identification.Different resonances from a same molecule are highly correlated(correlation coefficient r=1theoretically).Furthermore,molecules in the same biochemical pathway may also exhibit a secondary high correlation coefficient because of their similar or even codependent responses to a stimulus.This would be meaningful for our study to confirm the ambiguous peaks due to overlap and to analyze the disturbed metabolic pathways under LCT insult.2D-STOCSY is very useful to decipher the structures of metabolites from biological samples, taking glutamate and glutamine(Fig.3B)and taurine(Fig.3C)as examples.Glutamate and glutamine are the most important amino acids regarding neurological function.As the major toxicity of LCT is its neurotoxicity,their detection and monitoring in this study is cru-cial.Both glutamate and glutamine have two methenes and one methine,presenting multiplets between2.0and2.5ppm,and at 3.8ppm in the1H NMR spectrum.Due to partial overlapping,glu-tamate is still difficult to be resolved from glutamine.In the2D STOCSY map,distinct correlations from the peak at2.05ppm to peaks at2.36and3.75ppm were observed,and hence,they were assigned to glutamate.The peak at2.14ppm also has STOCSY cor-relations with peaks at2.46and3.77ppm,and therefore,they were assigned to glutamine.3.3.Multivariate statistical analyses of1H NMR spectra3.3.1.Aqueous liver extractsA principal component analysis(PCA)was performedfirstly to examine intrinsic pattern in the data set and to gain an overview of variation among the groups;however,PCA could not sufficientlyM.Li et al./Aquatic Toxicology146 (2014) 82–9285Fig.2.Histopathological examination of control and LCT treated gill,brain,heart,liver and kidney tissues by H&E staining.(A–E)Control goldfish with normal gill,brain, heart,liver and kidney(×400).(F)Gill of LCT group showing epithelial necrosis,desquamation,lamellar fusion,and shortening of secondary lamellae(×400).(G)Brain of LCT group showing no significant change(×400).(H)Heart of LCT group showing inflammatory cell infiltrations of cardiac pericardium(×400).(I)Liver of LCT group showing interstitial inflammatory cell infiltrations(×400).(J)Kidney of LCT group showing dilation of glomerular capillary,vacuolar variations in endothelial cellular size, and interstitial inflammatory cell infiltrations(×400).86M.Li et al./Aquatic Toxicology146 (2014) 82–92Table1Metabolites identified from the brain,heart and kidney tissue extracts,and their variations of LCT group versus control group.No.Metabolites Assignments Chemical shifts aı1H(ppm)Brain Heart KidneyChange b P c Change P Change P 1LeucineıCH3,ıCH3, CH,˛CH0.95(d),0.96(d),1.70(m),3.73(t)↓***––↓*** 2IsoleucineıCH3, CH3,˛CH0.94(t),1.00(d),3.66(d)↓**––↓** 3Valine CH3, CH30.97(d),1.04(d)↓***––↓*** 4MethylsuccinateˇCH3 1.09(d),//––5Lactate CH3,CH 1.32(d),4.10(q)↑*↓*↓–6AlanineˇCH3,˛CH 1.47(d),3.77(q)↑*––↓–7LysineıCH2, 1.72(m),↓–/↓*** 8GABA˛CH2,ˇCH2, CH2 2.30(t),1.89(m),3.01(t)↑–// 9Acetate CH3 1.91(s)––//10Acetamide CH3 1.98(s)––↓–↓–11NAA CH3,CH2,CH2,CH 2.02(s),2.49(dd),2.70(dd),4.38(m)↓–//12GlutamateˇCH2,ˇCH2, CH2,˛CH 2.05(m),2.12(m),2.36(m),3.75(m)↑**↑–↑**13GlutamineˇCH2, CH2,˛CH 2.14(m),2.45(m),3.76(t)↓*↑–––142-Oxoglutarate CH2,ˇCH2 2.40(t),3.01(t)↓*//15Succinate CH2 2.41(s)––/––16Methylamine CH3 2.58(s)––/↑**17Dimethylamine CH3 2.72(s)––/––18Creatine CH2,CH3 3.01(s),3.91(s)↑**––↓**19Phosphocreatine CH2,CH3 3.03(s),3.93(s)↓–↓–––20Malonate CH2 3.13(s)↓*/↑***21Choline CH3 3.18(s)↓–↓–↑**22Phosphocholine CH3 3.21(s)↓***↓–↓–23Taurine NH2–CH2,SO3–CH2 3.26(t),3.43(t)↑*↓–↑*24TMAO CH3 3.26(s)↑–/–↑–25myo-Inositol CH 3.27(t),3.53(dd),3.62(t),4.05(t)↑–//26Glucose 3.65–3.92(m)/↓/↑/27Maltose 5.40(m),5.23(d),3.96(t),3.56–3.92(m)/↓/↑/28cis-Aconitate CH 5.90(s)↓**/––29Inosine O–CH–N,N–CH N,N–CH N 6.10(d),8.23(s),8.34(s)↓–––↑***30Histidine N–CH C,N–CH N7.10(s),7.89(s)––//31Tyrosine CH,CH 6.90(d),7.20(d)↓*/↓***32Phenylalanine CH,CH,CH7.33(d),7.37(m),7.43(m)//↓***33AXP CH 6.15(d)↑*–*↓*34Nicotinurate CH,CH8.93(s),8.71(d)––//35Adenosine CH,CH8.25(s),8.34(s)//↑***36AMP N CH–N,N CH–N8.23(s),8.56(s)↑–/↓***a Multiplicity:s singlet,d doublet,t triplet,dd doublet of doublets,q quartets,m multiplets.b Metabolites with“↑/↓”means increased/decreased of LCT group versus control group,“/”means not detected or data not shown.c*represents P<0.05,**represents P<0.01and***represents P<0.001,“–”means no significant differences.P-values were calculated based on a parametric Student’s t-test or a nonparametric Mann–Whitney test(dependent on the conformity to normal distribution).Fig.3.Two-dimensional STOCSY analysis of brain extraction regions from1.8ppm to4.0ppm used to identify peaks of glutamate(blue)and glutamine(green),and heart extraction regions from3.1ppm to3.6ppm used to identify peaks of taurine(black).The degree of correlation across the spectrum has been color coded and projected on the spectrum.(A)Partial spectrum range from1.9to4.0ppm of brain extraction;(B)partial spectrum for the rangeı1.8–2.6of brain extraction,(C)partial spectrum for the rangeı3.1–3.6of heart extraction.The STOCSY enabled the assignments of these three metabolites as glutamate,glutamine and taurine,respectively.(For interpretation of the references to color in thisfigure legend,the reader is referred to the web version of the article.)87Fig.4.Typical500MHz CPMG1H NMR spectra of brain,heart and kidney tissue extracts with the recognized metabolites labeled.1Leucine,2Isoleucine,3Valine,4Methyl-succinate,5Lactate,6Alanine,7Lysine,8GABA(4-Aminobutyrate),9Acetate,10Acetamide11NAA(N-Acetylaspartate),12Glutamate,13Glutamine,142-Oxoglutarate, 15Succinate,16Methylamine,17Dimethylamine,18Creatine,19Phosphocreatine,20Malonate,21Choline,22Phosphocholine,23Taurine,24TMAO(trimethylamine-N-oxide),25myo-Inositol,27Maltose,28cis-Aconitate,29Inosine,30Histidine,31Tyrosine,32Phenylalanine,33AXP(adenosine di/tri/mo-phosphate),34Nicotinurate,35 Adenosine,36AMP(adenosine monophosphate).separate the two groups(see Fig.S2in Supporting information).The supervised OPLS-DA model were performed to obtain clear separa-tion,but also failed(R2Y=0.723,Q2Y=0.286,P=0.190),suggesting no significant metabolites change in liver.3.3.2.Aqueous brain extractsRepresentative1H NMR spectra of aqueous brain extracts from control and LCT treated groups were shown in Fig.4A,respec-tively(see the detailed metabolite assignments in Table1).The 1H NMR spectra of aqueous brain extracts contained a number of metabolites including branched-chain amino acids(BCAAs,e.g. leucine,isoleucine and valine),lactate,alanine,lysine,gamma-aminobutyric acid(GABA),acetamide,N-acetylaspartate(NAA), glutamate,glutamine,phosphocreatine(PCr),malonate,choline, phosphocholine,taurine,trimethylamine-N-oxide(TMAO),myo-inositol,inosine,and etc.PCA was performedfirstly,however,no significant difference(P>0.05)was found for thefirst two principal components between the two groups.OPLS-DA were then performed to gain better group separation and to reveal differential metabolites between the two groups.In the OPLS-DA score plot(Fig.5A),the two groups were well separated (R2Y=0.960,Q2Y=0.735,P=0.020).From the OPLS-DA score plot (Fig.5A)and color-coded loading plots(Fig.5C and D),metabolites e.g.glutamate,glutamine and etc.were found markedly differ-ent in the two groups.Student’s t test or Mann–Whitney test of metabolite concentrations between both groups was used to cal-culate the P values,the associated P values were summarized in Table1.3.3.3.Aqueous heart extractsThe representative500MHz CPMG1H NMR spectra of the aque-ous heart fractions obtained from the control and LCT treated88M.Li et al./Aquatic Toxicology146 (2014) 82–92Fig.5.Scores plot,S-plot and color-coded loadings plot(with the metabolites labeled in Fig.4A)according to the OPLS-DA analysis of NMR data from brain tissue extracts of goldfish:(A)Scores plot from OPLS-DA analysis of control and LCT group(n=9,8,respectively);(B)S-plot from OPLS-DA analysis of control and LCT group;(C and D) color-coded loadings plot from OPLS-DA analysis shows the metabolite components that differ between the control and LCT groups.Blue:Lowest,no statistically significant difference between the groups;Red:highest,statistically significant.Positive peaks indicate a relatively decreased metabolite level in the LCT group,while negative peaks indicate an increased metabolite level in the LCT group.(For interpretation of the references to color in thisfigure legend,the reader is referred to the web version of the article.)Fig.6.Scores plot,S-plot and color-coded loadings plot(with the metabolites labeled in Fig.4B)according to the OPLS-DA analysis of NMR data of heart tissue extracts of goldfish:(A)Scores plot from OPLS-DA analysis of control and LCT group(n=7);(B)S-plot from OPLS-DA analysis of control and LCT group;(C and D)color-coded loadings plot from OPLS-DA analysis shows the metabolite components that differ between the control and LCT groups.Blue:Lowest,no statistically significant difference between the groups;Red:highest,statistically significant.Positive peaks indicate a relatively decreased metabolite level in the LCT group,while negative peaks indicate an increased metabolite level in the LCT group.(For interpretation of the references to color in thisfigure legend,the reader is referred to the web version of the article.)M.Li et al./Aquatic Toxicology146 (2014) 82–9289Fig.7.Scores plot,S-plot and color-coded loadings plot(with the metabolites labeled in Fig.4C)according to the OPLS-DA analysis of NMR data from kidney tissue extracts of goldfish:(A)Scores plot from OPLS-DA analysis of control and LCT group(n=9);(B)S-plot from OPLS-DA analysis of NMR data of control and LCT group;(C and D)color-coded loadings plot from OPLS-DA analysis shows the metabolite components that differ between the control and LCT groups.Blue:Lowest,no statistically significant difference between the groups;Red:highest,statistically significant.Positive peaks indicate a relatively decreased metabolite level in the LCT group,while negative peaks indicate an increased metabolite level in the LCT group.(For interpretation of the references to color in thisfigure legend,the reader is referred to the web version of the article.)groups were shown in Fig.4B,with metabolites assigned(Table1). The dominant metabolites in the aqueous heart fractions included BCAAs,lactate,acetamide,glutamine,glutamate,PCr and etc.PCA wasfirstly employed to examine the data.One sample was consid-ered as outlier based on PCA modeling of the binned NMR data and removed before further analyses.The subsequent OPLS-DA analysis (R2Y=0.966,Q2Y=0.786,P=0.030)gave a clear separation between the control and LCT treated groups(Fig.6A).The P values calculated by Student’s t test or Mann–Whitney test of detected metabolites were summarized in Table1.3.3.4.Aqueous kidney extractsFig.4C presented typical1H NMR CPMG spectra for kidney sam-ples from control and LCT treated groups with major metabolites labeled.Visually,the metabolites like BCAAs,lactate,PCr,creatine and taurine constituted the most intense resonances in the spectra. Firstly,PCA was employed to examine all kidney data of the two groups and it produced a good separation between both groups. Since one sample was far away from the great majority of sam-ples(outlier)based on PCA modeling of the binned NMR data and therefore,it was removed before further analysis.Subsequently, an OPLS-DA model was used to discriminate the samples accord-ing to their class membership(R2Y=0.897,Q2Y=0.523,P=0.025). The scores plot(Fig.7A)of kidney data illustrates a distinct sep-aration of the control group and the LCT treated group along the t p,1component.In the loadings plot(Fig.7C and D),some dif-ferential metabolites were found:the primary differences in LCT treated group were the decrease of BCAAs,alanine,and lactate, and the increase of glutamate,taurine and etc.,as compared with the control group.The P values calculated by Student’s t test or Mann–Whitney test of detected metabolites were summarized in Table1.4.DiscussionThe toxicity of LCT was investigatedfirstly by histopathologi-cal inspection.The goldfish gill,heart,liver and kidney exhibited tissue specific impairments by LCT,as shown in Fig.2.How-ever,no significant histological alterations were found in brain of LCT group though the primary toxicity of LCT is its neu-rotoxicity,which showed the disadvantage of traditional tissue inspection:insensitive to toxic effect.NMR profiling of LCT treated and normal goldfish,aided with multivariate statistical analy-sis,revealed metabolic disturbance in neurotransmitter balance, oxidative stress,energy metabolism,amino acid metabolism and osmolyte balance induced by LCT.4.1.Neurotransmitter disturbanceThe primary mechanism of LCT toxicity is well known:it binds to the membrane lipid phase in the immediate vicinity of the sodium channel,thus blocking the closing of the sodium gates and prolong-ing the return of the membrane potential to its resting state,leading to hyperactivity of the nervous system which can eventually result in paralysis and/or death(Lim et al.,2010).The significantly increased glutamate level in the brain of LCT group was observed,which was consistent with the result obtained using rats(Breckenridge et al.,2009).With the glutamate con-centration substantially elevated,the glutamate receptors,known as excitatory amino acid receptors,were excessively activated exhibiting behavioral abnormalities such as hyperactivity and spas-tic movements.The brain level of glutamine,another amino acid of equal neural importance as glutamate,however,was significantly decreased.The two amino acids could be interconverted in neurons and glial cells(Bak et al.,2006;Peng et al.,1993).。