Journal of Hazardous Materials 299(2015)249–259Contents lists available at ScienceDirectJournal of HazardousMaterialsj 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 /j h a z m atA novel method to analyze hepatotoxic components in Polygonum multiflorum using ultra-performance liquidchromatography-quadrupole time-of-flight mass spectrometryLongfei Lin a ,Hongmei Lin a ,Miao Zhang a ,Boran Ni b ,Xingbin Yin a ,Changhai Qu c ,Jian Ni a ,∗aSchool of Chinese Materia Medica,Beijing University of Chinese Medicine,Beijing,China bSchool of Basic Medical Science,Beijing University of Chinese Medicine,Beijing,China cModern Research Center for TCM,Beijing University of Chinese Medicine,Beijing,Chinah i g h l i g h t s•Investigated the hepatotoxicity ofPolygonum multiflorum and its con-stituents.•The run time of each sample was 25min by UPLC-Q-TOF/MS.•Used an analytical method based on Progenesis QI and Makerlynx XS soft-ware.•Identified 9potential hepatotoxic components in P.multiflorum.g r a p h i c a la b s t r a cta r t i c l ei n f oArticle history:Received 17March 2015Received in revised form 21May 2015Accepted 6June 2015Available online 9June 2015Keywords:Polygonum multiflorumPolygonum multiflorum Praeparata Hepatotoxicity ToxicityUPLC-Q-TOF/MSa b s t r a c tPolygonum multiflorum ,called Heshouwu in China,is a traditional Chinese medicine used to treat var-ious diseases .However,the administration of P.multiflorum (PM)and P.multiflorum Praeparata (PMP)causes numerous adverse effects.This study sought to analyze the toxic components of PM using ultra-performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF/MS),and their hepatotoxicity in L02human liver cells.Toxicity was evaluated by using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT),lactate dehydrogenase (LDH)leakage,and liver enzyme secretion (aspartate aminotransferase [AST]and alanine aminotransferase [ALT])assays.Furthermore,UPLC-Q-TOF/MS,Progenesis QI,and Makerlynx XS software analyses were used to differentiate extracts and analyze the toxic components.The order of toxicity was P.multiflorum ethanol extract (PME)>P.multiflorum water extract (PMW)>P.multiflorum Praeparata ethanol extract (PMPE)>P.multiflorum Praeparata water extract (PMPW),which was determined by MTT assay,LDH leakage,and liver enzyme secretion levels.The analysis methods suggest that PM toxicity may be associated with anthraquinone,emodin-O -(malonyl)-hex,emodin-O -glc,emodin,emodin-8-O -glc,emodin-O -(acetyl)-hex,and emodin-O -hex-sulphate.The toxic mechanisms of these components require further study.©2015Elsevier B.V.All rights reserved.∗Corresponding author.Fax:+861084738607.E-mail address:njtcm@ (J.Ni).1.IntroductionPolygonum multiflorum is a popular traditional Chinese medicine used in many patent medicines and prescriptions.Recent stud-ies showed that PM has antioxidant activity,due to its flavonoid/10.1016/j.jhazmat.2015.06.0140304-3894/©2015Elsevier B.V.All rights reserved.250L.Lin et al./Journal of Hazardous Materials299(2015)249–259Table.1The dried extract rate and diluted concentrations of different extracts.Name Dried extract rate(%)Concentrations for cell tests(␮g/mL)PME29.201000.00704.00352.00176.0088.0044.0022.0011.00 PMW35.601000.00859.00429.5214.75107.3853.6926.8413.42 PMPE20.681000.00500.00250.00125.0062.5031.2515.637.81 PMPW41.441000.00500.00250.00125.0062.5031.2515.637.81and phenolic acid compounds[1].Additionally,stilbene from PM has anti-tumor,anti-aging,and liver-protective effects[2,3]. Furthermore,anthraquinones have many biological effects,includ-ing antimicrobial,antioxidant,and antihuman cytomegalovirus activity[4].However,researches have reported PM-induced hepa-totoxicity,nephrotoxicity,and embryonic toxicity.Hepatic adverse events,including acute toxic hepatitis,have been reported in many countries[5–9].PM also induces nephrotoxicity[10,11].Emodin, rhein,and physcion,the major components of PM,significantly inhibit the proliferation of a human proximal tubular epithelial cell line(HK-2)[12].Moreover,PM can also induce lung toxicity[13].Although PM toxicity is of great concern,the hepatotoxic com-ponents remain unknown.In this study,the toxicity of water and ethanol extracts from PM and PMP wasfirstly compared.Secondly, the components in these four extracts were comprehensively ana-lyzed by UPLC-Q-TOF/MS,tofind out the variation tendency of the components in these four extracts based on Progenesis QI and Makerlynx XS software.Thirdly,the toxic components in PM were speculated through the combination of the toxicity results and data analysis.The well-designed experiments in vitro to predict the toxi-city induced by risky material and medicine foe humans.MTT assay is the most common method for detecting cell growth and survival, and it is also widely used to evaluate the cytotoxicity.So,the toxi-city induced by PM and PMP was investigated on human liver cells (L-02)in vitro in this study,and used a novel analytical method to analyze the toxic components.2.Materials and methods2.1.ChemicalsMethanol(HPLC-grade)was purchased from Fisher Scientific (Waltham,MA,USA).HPLC-quality water was obtained from a Cascada TM IX-water Purification System(Pall Co.,Port Washing-ton,NY,USA).Ammonia was purchased from Guangdong Xilong Chemical Co.,Ltd.(Shenzhen,China).PM and PMP were provided by Beijing Tongrentang Co.,Ltd.(Beijing,China).Standard samples of emodin and2,3,5,4 -tetrahydroxystilbene-2-O-␤-glucoside were purchased from Shanghai Standard Biotech Co.,Ltd.(Shanghai,China).Emodin-8-O-␤-d-glucopyranoside,cat-echin,l-epicatechin,P-hydroxybenzaldehyde,rhaponiticin,and gallic acid were purchased from Shanghai Yuanye Bio-Technology Co.,Ltd.(Shanghai,China).2.2.Preparation of P.multiflorum extractsPM and PMP were decocted three times with10-fold water or 70%ethanol for1.5h.The resultingfiltrates were mix together and concentrated to dryness under reduced pressure.The dried extract rate of each extraction was shown in Table1.The concentrations of the major components were displayed in Supplementary data 1.The stock solutions for the cellular evaluations were made by dissolving the extracts in50%dimethylsulfoxide(DMSO).Extracts were further diluted with0.2%fetal bovine serum(FBS,Hyclone, Logan,UT,USA)in Dulbecco’s modified Eagle’s medium(DMEM, Gibco Invitrogen Corp.,Grand Island,NY,USA)to the concentra-tions shown in Table1.The diluted concentrations of the different extracts were not same with each other,because the purpose is to ensure the quantity of the crude herbs was equivalent,allowing for the comparison of extract toxicity.For example,704␮g/mL of PME,859␮g/mL of PMW,500␮g/mL of PMPE,and1000␮g/mL of PMPW contained the same quantity of crude herbs.In this section, the diluted manner was different from previous studies,because we are not only to obtain the IC50of each extracts,but also to make a parallel-group study to compare the hepatotoxicity(ALT,AST and LDH)about these four extracts at the same amount of crude herbs.2.3.Culture conditionsL02cells,a human-derived fetal hepatocyte cell line,were pur-chased from China Infrastructure of Cell Line Resources.Cells were cultured in DMEM supplemented with10%FBS.Cultures were maintained in a humidified incubator with5%CO2and95%air at 37◦C.Trypsin(0.5%,Sigma,St.Louis,MO,USA)was used to passage cells at80–90%confluence.2.4.MTT cell viability assayThe MTT reduction assay was used to determine cell viability. The L02cells were seeded into96-well plates(1×104cells/well, Corning Incorporated,Corning,NY,USA)and incubated for24h. And then,cells were exposed to fresh medium with various extract concentrations for36h.Following incubation,5mg/mL of MTT (10␮L,BioTOP,Suzhou,China)dissolved in PBS(pH7.4)was added and incubated for4h.The medium was then removed,and100␮L of DMSO was added to dissolve the formazan product.Crystals were dissolved in DMSO after gently swirling the96-well plates for 5min at room temperature.The absorbance change of specimen was measured under570nm using a Multiskan GO spectrometer (Molecular Devices,Thermo,Waltham,MA,USA).2.5.Enzyme leakage from L02cell after extract treatmentThe activity of the AST,ALT,and LDH releasing,were regarded as the index of hepatotoxicity.L02cells were seeded in24-well plates(2×104cells/well,Corning Incorporated).After24h,cells were exposed to fresh medium with the test compounds or extrac-tions for36h.The Culture medium was then analyzed for AST,ALT and LDH using a kit(Beijing Leadman Biochemistry Co.,Ltd.,Bei-jing,China)on the Glamour4000biochemical analyzer(Glamour Co.,Ltd.,USA).Six plates were read in parallel,with4wells per condition.2.6.UPLC-Q-TOF/MS analysis of PME,PMW,PMPE and PMPW2.6.1.Sample preparationThe PME,PMW,PMPE and PMPW samples(365.1,445.0,258.3, and518.1mg,respectively)were accurately weighed respectively for6times.The PMW and PMPW were dissolved in50mL of water, and the PME and PMPE were dissolved in50mL of70%ethanol. Varying concentrations were diluted to1mL.Then,they were cen-trifuged at14,000rpm for10min to separate the supernatant for UPLC-Q-TOF/MS analysis.L.Lin et al./Journal of Hazardous Materials299(2015)249–259251 Table.2MTT assay results for PME,PMW,PMPE and PMPW.PME PMW PMPE PMPW␮g/mL Cell proliferation(%)␮g/mL Cell proliferation(%)␮g/mL Cell proliferation(%)␮g/mL Cell proliferation(%)11.0095.16±11.1313.42100.93±5.137.8198.58±3.707.81100.85±2.4522.0097.58±8.4926.8496.87±4.4715.6397.65±13.2615.63101.21±0.9444.00101.00±9.8053.6997.08±5.9331.2594.16±2.7031.2599.86±10.8488.0087.04±25.54107.3895.09±4.9162.5088.46±2.3462.5099.00±8.41176.0082.26±4.64214.7594.37±5.52125.0089.46±0.89125.0097.86±12.70 352.0042.31±3.39429.5068.95±6.72250.0078.42±3.67250.00103.77±6.33 704.0011.47±0.63859.0020.58±1.85500.0073.22±5.28500.00101.42±5.671000.008.62±1.281000.0020.16±0.631000.0056.20±6.621000.0086.89±4.50Table.3Enzyme secretion of AST,ALT and LDH from L02cell after exposure of PME,PMW,PMPE and PMPW.Group AssayPME Concentration(␮g/mL)1000.00703.00351.50175.7587.8843.9421.9710.98Control AST(U/L)23.08±9.27***21.10±4.85**15.85±0.8313.98±2.8113.38±2.6414.03±1.1512.00±2.3314.60±0.4514.11±2.84 ALT(U/L)10.65±3.25*** 6.85±1.14*** 4.43±0.41** 3.03±0.83 2.65±0.88 2.10±0.55 2.30±0.92 2.20±2.03 2.01±0.54 LDH(U/L)62.73±8.37***60.75±11.48***46.45±4.80***39.68±8.02***29.13±1.5923.55±1.5222.98±3.4421.85±6.4022.16±5.52PMW Concentration(␮g/mL)1000.00916.00458.00229.00114.5057.2528.6214.31 AST(U/L)21.50±2.40***17.93±1.70*15.13±1.8014.15±2.0711.70±2.1912.50±1.2713.08±3.8611.48±3.0714.11±2.84 ALT(U/L)7.23±2.99*** 5.25±0.99*** 3.90±0.57** 2.30±0.48 2.20±0.59 2.00±0.41 1.90±0.42 2.20±0.43 2.01±0.54 LDH(U/L)55.05±4.78***56.73±0.57***40.63±7.04***30.13±3.13*23.23±4.5521.85±2.2420.58±2.9315.93±5.7422.16±5.52 PMPE Concentration(␮g/mL)1000.00500.00250.00125.0062.5031.2515.637.81 AST(U/L)15.48±3.5715.08±3.6614.18±0.7712.50±1.6812.03±2.4812.98±0.9312.40±3.0612.33±1.9714.11±2.84 ALT(U/L) 4.05±1.75*** 2.85±1.18 2.78±0.71 2.10±0.32 2.10±1.31 2.05±0.66 1.75±0.71 2.05±0.52 2.01±0.54 LDH(U/L)44.25±2.30***37.38±2.40***31.25±2.22**23.90±4.8419.58±4.4322.13±7.8621.48±2.5722.40±3.2422.16±5.52PMPW Concentration(␮g/mL)1000.00500.00250.00125.0062.5031.2515.637.81 AST(U/L)15.90±2.5114.63±2.9915.90±2.0913.50±1.5113.53±1.4414.63±1.0114.13±2.6614.10±1.1614.11±2.84 ALT(U/L) 3.30±1.21* 2.08±0.40 2.13±0.54 2.15±0.90 2.13±0.28 2.00±0.98 1.85±n0.72 2.15±0.96 2.01±0.54 LDH(U/L)31.78±4.21**24.33±5.7824.93±3.9521.28±3.0021.83±1.5722.93±6.9322.85±3.9319.45±2.6622.16±5.52*,**,***Significant difference compared with control cells,p<0.05,p<0.01,p<0.001.2.6.2.Chromatographic and mass spectrometric conditionsAnalysis was performed on a Waters Acquity Ultra Performance LC system coupled with a Xevo G2Q-TOF mass spectrometry equipped with an electrospray ionization(ESI)source(Waters Corp.,Milford,MA,USA).The separation of all samples was per-formed on an ACQUITY UPLC HSS T3(100×2.1mm,1.8␮m).The gradient elution employed0.1%of formic acid-acetonitrile solu-tion as solvent A and0.1%of formic acid aqueous solution as solvent B.The gradient program was as follows:0–3min,0–10%A; 3–10min,10–20%A;10–20min,20–70%A;20–21min,70–100% A;21–21.1min100–0%A;21.1–25min0%A.Theflow rate was set at0.3mL/min,the injection volume was5␮L.Negative ion ionization was used for the detection of more com-pounds and it had higher sensitivity than the positive mode[14]. The cone gas and the source temperature were set to50L/h and 120◦C.The capillary was2.5kV.High-purity nitrogen served as both the nebulizing and dry gas.The desolvation temperature was held at350◦C,and the gasflow was600L/h.The full-scan range was from50to1200m/z.The validation of the UPLC-Q-TOF/MS method was displayed in Supplementary data2.2.7.Statistics and DATA analysisAll cytotoxicity data were expressed as the mean±SD.The data were subjected to one-way analysis of variance(ANOVA)fol-lowed by multiple group comparisons test,and were calculated by using the Probit function in SPSS17.0.The half-maximal inhibitory concentration(IC50)of each tested compound or extraction was cal-culated with GraphPad prism5.Results were classified into three significance levels using the p-values0.05,0.01,and0.001.The LC–MS runs were loaded on Progenesis QI software,which can visualize the raw data at each step in the analysis without missing values.Ion intensity maps shows a2D representation of retention time,m/z,and feature intensity,as well as mass spectra and chromatogram views,provided quality assurance of the auto-matic alignment,peak picking,and compound deconvolution on every run.Wefiltered the data as necessary,and then imported into Makerlynx XS software to analyze.3.Results and discussion3.1.Extract toxicityIn order to evaluate the hepatotoxicity of PM and PMP,more than one assay should be used to determine cell viability in vitro, in order to increase the reliability of the results.The MTT and LDH assays reflect the proliferation and the function of cell membranes. In contrast,the ALT and AST activity levels are the important indi-cator of hepatotoxicity.The ALT level is3times greater than normal indicate liver injury[15,16].Changes in cell proliferation after PME,PMW,PMPE and PMPW exposure are listed in Table2.The AST,ALT and LDH levels are shown in Table3.PME showed significant cytotoxicity and hepatotoxicity,inhibit-ing cell growth57.69%at352␮g/mL and88.53%at704␮g/mL,as determined by MTT assay.LDH leakage was elevated significantly following the176␮g/mL treatment.ALT leakage was significantly increased in the352and704␮g/mL groups.Furthermore,the ALT and LDH levels were3times higher than those in the control group after the704␮g/mL treatment.PMW cytotoxicity was weaker than PME.Cell survival rates after PMW treatment were higher than the rates in MTT assay,with 31.05%and79.42%inhibition at429.5and859␮g/mL,respectively.252L.Lin et al./Journal of Hazardous Materials299(2015)249–259Table.4The constituents detected in PME by UPLC-Q-TOF/MS.No.t R[M-H]−or[M+FA-H]−Error(ppm)Formula Identification10.889227.03214C8H9N2O9Unknown20.941195.0508 1.5C6H11O7Gulonic acid3 1.0311025.3427 1.9C36H65O33Trehalulose4 1.065341.1082/683.2263−0.6C12H21O11Glucosyl-glucose5 1.065404.1038−0.7C12H22NO14Unknown6 1.116439.0771−1.6C14H19N2O14Unknown7 1.416387.1136−0.8C13H23O13Unknown8 1.433133.0135−1.5C4H5O5Malic acid9 1.454475.12990C16H27O16Unknown10 2.524503.1588−4.8C18H31O16Glucopyranosyl–glucopyranosyl-glucose11 2.579128.03490.8C5H6NO3N-Acryloyl glycine12 2.648191.0187−2.6C6H7O7Citric acid13 3.529169.0132−3C7H5O5Gallic acid14 4.021331.0654−3.3C13H15O10Glucosyl gallate15 4.162419.1667−3.3C22H27O8cis-rhaponitin16 4.179419.1666−4.5C22H27O8Rhaponitin17 6.164531.1502−0.2C26H27O12Unknown18 6.202142.0654−2.1C10H8N Unknown19 6.322531.1499−0.8C26H27O12Unknown20 6.763289.0718 2.1C15H13O6Catechin217.289549.1605−0.5C26H29O13Unknown227.73567.173 2.8C26H31O14Tetrahydroxystilbene-O-di-hex238.312289.0704−2.8C15H13O6l-Epicatechin248.556729.145−0.8C37H29O16Mono-o-galloylprocyanidin258.979405.1184−0.5C20H21O9Tetrahydroxystilbene-O-hex269.172729.1453−0.4C37H29O16Mono-o-galloylprocyanidin279.33729.1454−0.3C37H29O16Mono-o-galloylprocyanidin289.419405.1185−0.2C20H21O9cis-2,3,5,4 -Tetrahydroxystilbene-2-O-glc 299.595729.1446−1.4C37H29O16Mono-o-galloylprocyanidin3010.07421.1145 2.4C20H21O10Catechin-O-furanoside3110.263567.1722 1.4C26H31O14Tetrahydroxystilbene-O-di-hex3210.755567.17140C26H31O14Tetrahydroxystilbene-O-di-hex3311.477405.1197 2.7C20H21O92,3,5,4 -Tetrahydroxystilbene-2-O-glc 3411.863557.12960.2C27H25O13Tetrahydroxystilbene-O-(galloyl)-hex 3512.513557.12960.2C27H25O13Tetrahydroxystilbene-O-(galloyl)-hex 3612.551447.12920.2C22H23O10Tetrahydroxystilbene-O-(acetyl)-hex 3712.727555.11390C27H23O13Unknown3812.988431.0984 1.4C21H19O10Emodin-O-glc3913.129447.1284−1.6C22H23O10Tetrahydroxystilbene-O-(acetyl)-hex 4013.219557.1302 1.3C27H25O13Tetrahydroxystilbene-O-(galloyl)-hex 4113.779312.1245 2.9C18H18NO4N-trans-feruloyl tyramine4213.819551.1559 1.1C36H23O6Tetrahydroxystilbene-2-O-(coumaroyl)-hex 4313.821581.16610.3C30H29O12Tetrahydroxystilbene-2-O-(feruloyl)-hex 4413.869511.05460C21H19O13S Emodin-O-hex-sulphate4513.886431.09810.7C21H19O10Emodin-O-glc4614.078407.1339−0.7C20H23O9Torachrysone-8-O-glc4714.185431.0980.5C21H19O10Emodin-8-O-glc4814.536559.1458 1.1C27H27O13Torachrysone-O-glucogallin4914.75517.0979−0.6C24H21O13Emodin-O-(malonyl)-hex5014.767473.1083−0.2C23H21O11Emodin-O-(acetyl)-hex5115.045283.0604−0.7C16H11O5Anthraquinone5215.383473.10840C23H21O11Emodin-O-(acetyl)-hex5315.486329.2327−0.3C18H33O5Trihydroxy-octadecenoic acid5416.786283.0609 1.1C16H11O5Anthraquinone5517.492283.06−2.1C16H11O5Anthraquinone5618.39269.0445−1.9C15H9O5Emodin5719.515269.0445−1.9C15H9O5Anthraquinone5819.952277.1809 1.8C17H25O3Heptadecanoic acid-ester5920.093329.2326−0.6C18H33O5Trihydroxy-octadecenoic acidThe ALT and LDH leakage also significantly increased at these two concentrations,although to a smaller extent than PME.The toxicity of PMPE was obviously weaker than that of PME and PMW,as the growth inhibition rates were markedly lower. Although the LDH leakage significantly increased with250␮g/mL treatment,ALT levels were not significantly altered.PMPW did not inhibit cell proliferation in the MTT assay,except where they were at the condition of highest concentration(1000␮g/mL).The cyto-toxicity and hepatotoxicity of this extract was the lowest in four extracts.In summary,the PME(IC50=345␮g/mL in the MTT assay), PMWE(IC50=555␮g/mL),PMPE(IC50=1536␮g/mL),and PMPW (IC50>10,000␮g/mL)showed severe,moderate,less and no cyto-toxicity,respectively.The cytotoxicity of PM was stronger than that of PMP,and the ethanol extract exerted greater cytotoxicity than the water extract.Considering that the elevated enzyme secre-tion affects liver function,patients should be monitored during the long-term use of PM.3.2.UPLC-Q-TOF/MS analysis of PMW,PME,PMPW and PMPEIn this study,59major ingredients in PMW,PME,PMPW and PMPE were separated and detected by using a UPLC-Q-TOF/MS system(Table4).49of the59compounds were identified byL.Lin et al./Journal of Hazardous Materials299(2015)249–259253 Table5The difference between every two groups using OPLS-DA.Group1vs Group2m/z Retention time(min)Group1Group2PME vs PMW341.1081 1.36640545738565475.1298 1.435054104482133.0135 1.4290354181835557.129411.86316547201540231.065412.816478812085431.097213.0122276674154245.08114.0872097188886230.05714.0812824717141407.133714.0811********.19431.097714.2223004701118380863.202614.2213853617641.3473.108414.73129229079405.8517.098414.73142514526.5861035.20514.73909360.0015245.081214.81487792117.35283.060415.01665031131789919.229915.05584530.001283.059915.682296227751.92269.044519.5232710051968.1 PMW vs PMPE341.1082 1.03666478161144683.2248 1.0342942932758341.1081 1.3673856561138.8683.2244 1.36213840159531.1497 6.1537316285792531.1499 6.2936169212242243.065711.4439327001883750405.118611.4447948601470430811.245411.44101500099166431.097913.853*********511.054813.852*********.55431.097714.221118380376589269.044519.5251968225075PMW vs PMPW341.1082 1.03666478138351683.2248 1.0342942928648341.1081 1.3673856575162683.2244 1.36213840113133.0135 1.4218183588001503.0854 2.980.0004131238289.0707 6.76154848805405.118611.4447948602098970811.245411.441015*********557.129411.8620154061124.8431.097913.853*********511.054813.852********431.097714.221118380177239269.044818.35223263820PME vs PMPE341.1082 1.03712067161144683.2248 1.0345459832758341.1081 1.3664054561138683.2244 1.36227135159531.1497 6.157806285792531.1499 6.297936212242549.16057.297580170192405.118611.4443458801470430811.245411.4494793899166557.129411.8631654762642431.097913.855261368941511.054813.852*********245.08114.0872097159594431.097714.222300470376589473.108414.7312922901283.88283.060415.01665031104527283.059915.6822962211130293.178118.18150920352149269.044818.3526474112517PME vs PMPW277.03210.897464.06243919341.1082 1.03712067138351683.2248 1.0345459828648341.1081 1.3664054575162503.0854 2.980.0004131238405.118611.4443458802098970811.245411.44947938226532431.097913.855261362604511.054813.852********245.08114.0872097124817254L.Lin et al./Journal of Hazardous Materials 299(2015)249–259Table 5(Continued )Group1vs Group2m /zRetention time (min)Group1Group2431.097714.222300470177239473.108414.731292290182283.060415.0166503135021293.178118.18150920322613269.044818.35264741820269.044519.5232710028539PMPE vs PMPW531.1497 6.1528579293664531.1499 6.2921224273385549.16057.2917019283491531.158.158949542089811.245411.4499166226532431.097714.22376589177239283.060415.0110452735021269.044519.5222507528539595.287620.78132010666712.536321.41756430.003Fig.1.The translation process by Progenesis QI software.the comparison with standards or investigating -pounds 8,15,16,20,25,35,and 47were attributed to gallic acid,rhaponitin,catechin,l -epicatechin,2,3,5,4 -tetrahydroxystilbene-2-O -glc,emodin-8-O -glc,and emodin,respectively,by comparing the retention times and mass spectral data with reference pounds 2and 4were attributed to gulonic acid and glucosyl-glucose,whereas compound 12was identified as citric acid by using references [17].Compound 11was characterized as N -acryloyl glycine [18].Compound 24,26,27,and 29showed the same [M–H]−ions at m /z 729.145,so they may be isomers of mono-o -galloylprocyanidin [14].Compounds 22,25,28,31,32,34,35,36,39,40,42,and 43gave diagnostic ions at m /z 405and 243,suggesting that they are tetrahydroxystilbene-O -hex derivatives.There are four groups of compounds (compounds 22,31,32;com-pounds 25,28;compounds 34,35,40and compounds 36,39)that produced the same [M–H]−ions at m /z 567.17,557.13,and 447.13,indicating they may be isomers of tetrahydroxystilbene-O -di-hex,2,3,5,4 -tetrahydroxystilbene-2-O -glc,tetrahydroxystilbene-O -(galloyl)-hex,and tetrahydroxystilbene-O -(acetyl)-pounds 42and 43were identified as tetrahydroxystilbene-2-O -(coumaroyl)-hex and tetrahydroxystilbene-2-O -(feruloyl)-hex [19–21],pounds 46and 48gave [M–H]−ions at m /z 407.1339(C 20H 23O 9)and m /z 559.1458(C 27H 27O 13),which were torachrysone-8-O -glc and torachrysone-O -glucogallin,respectively.[20]Compounds 44,45,49,50and 52showed [M–H]−ions at m /z 511.0546,431.0981,517.0979,473.1083,and 473.1084,which further yielded the same ions at m /z 431and 269.By investigating the reference data,the compounds were ten-tatively characterized as emodin-O -hex-sulphate,emodin-O -glc,emodin-O -(malonyl)-hex,emodin-O -(acetyl)-hex,and emodin-O -(acetyl)-hex [14,17,22,23].Compounds 51,54,55,and 56showed [M–H]−ions at m /z 283.0604,283.0609,283.06,and 269.0445.These four compounds were anthraquinone constituents;how-ever,they did not match the standards for physcion,rhein,and emodin.Thus,compounds 35,37,and 38may be isomers of physcion,rhein,or pounds 58and 59were lipids in PM,which were identified as heptadecanoic acid–ester and trihydroxy-octadecenoic acid [24].Some components were diffi-cult to identify,because the levels were low in PM.L.Lin et al./Journal of Hazardous Materials299(2015)249–259255Fig.2.The peak picking-ion map.Fig.3.The normalization graphs of the PME,PMW,PMPE and PMPW.256L.Lin et al./Journal of Hazardous Materials 299(2015)249–259Fig.4.OPLS-DA scatter plots of the difference between two groups.A =PME vs PMW;B =PMW vs PMPE;C =PMW vs PMPW;D =PME vs PMPE;E =PME vs PMPW;F =PMPE vsPMPW.Fig.5.The dendrogram of the correlation analysis.3.3.DATA analysis of PMW,PME,PMPW and PMPEIn order to observed all differences among the four extracts,fur-ther sample profiling of the gelatins required the use of multivariate statistical tools.In this study,the first step of the multivariate sta-tistical analysis was to convert the 3D LC/MS data into a 2D matrix,expressed as an Exact Mass Retention Time (EMRT)pair by using Progenesis QI3.3.1.Step 1:DATA analysis by Progenesis QIFirstly,the TIC spectrum (Fig.1A)created by UPLC-Q-TOF/MS for all samples were imported into the Progenesis QI software andL.Lin et al./Journal of Hazardous Materials 299(2015)249–259257Fig.6.The results of presumed toxic components.(Note:m /z at 517and 1035were one compound.407,m /z at 230and 245also were one compound in A.m /z at 863and 431,t R at 14.2,m /z at 473and 245also were one compound in B,respectively).translated into a 2D ion intensity map,as shown in Fig.1B.The ordinate represented the retention time,and the abscissa was m /z .Secondly,six QC samples were evaluated to determine whether all four group samples were aligned.The alignment vector is shown in Fig.1C.The results of the alignment quality evaluation revealed thatthe scores of all samples had scores greater than 90%,as shown in Fig.1D.Not all peaks in the total ion chromatogram (TIC)spectrum represent a single compound,due to high background,chromatog-raphy condition,or sample handling methods.Thus,we set the sensitivity value to three,and the minimum peak width to 0.15min,Fig.7.Suggested method of presumed the toxic ingredients in TCM.258L.Lin et al./Journal of Hazardous Materials299(2015)249–259Table6The constituents selected by correlation analysis.Name No.t R[M-H]−Identification6-A115.68283.0599Anthraquinone215.01283.0604Anthraquinone314.08407.1337Torachrysone-8-O-glc414.73517.0984Emodin-O-(malonyl)-hex6-B513.01431.0972Emodin-O-glc618.35269.0448Emodin713.85431.0979Emodin-O-glc811.86557.1294Tetrahydroxystilbene-O-(galloyl)-hex914.22431.0977Emodin-8-O-glc1014.73473.1084Emodin-O-(acetyl)-hex1113.85511.0548Emodin-O-hex-sulphatethis setting which has been found to give the best balance of detect-ing as many real feature ion signals as possible while detecting as little random noise as possible.Under these conditions,1403 peaks were observed in the2D ion intensity map,as shown in Fig.2.The normalization graphs for the four groups are shown in Fig.3.Although we identified1403compounds in the extracts, not all compounds are targets for our analysis.To identify the rea-sons for the differences in toxicity,we must identify difference in the composition between the four extracts.In order to obtain more accurate results,we included two restriction conditions to choose analytes:an ANOVA p-value≤0.05and a maximum fold change≥2fold change,these two parameter using for selecting significantly changing compounds,setting like these will reduce the“false discovery rate(FDR)”.Under these conditions,1103of the1403compounds satisfy the requirements.Finally,the selected compounds were exported into EZinfo for analysis.3.3.2.Step2:DATA analysis by Makerlynx XSThe data set was visualized by using unsupervised PCA to check for outliers and classification trends among the extracts.In the score plot obtained by PCA,the four extracts were located farther from each other,indicating a clear differentiation between the extracts. The difference between two groups was evaluated by using OPLS-DA.The obtained scatter plots are shown in Fig.4.In the scatter plot, each point represents accurate mass-retention time data.The X-axis represents the variable,and the distance of the data point from the origin indicates the relative contribution.The data at both ends of the S-curve represent the highest credibility characteristic ions of the groups.So,it is suggested tofilter out the significantly dif-ferent variables(Markers)between two groups.These plots clearly display the observations with a high absolute value of p(corr)[1] and a high absolute value of the coefficients.The observations dif-ferentiating between every two groups are listed in Table5.3.3.3.Step3:Correlation analysis the DATAThe differentiated compounds which found in step2were re-imported into Progenesis QI software for correlation analysis,as shown in Fig.5.This analysis correlates with the content of these differentiated compounds in the four extracts.bined results of the toxicity and data analysis to identify toxic componentsAs shown in Fig.6A,upon combining the results of the correla-tion analysis,the compound content inside the box was abundant in PME,but low in the other three extracts.The description of these ingredients is shown in Table6A(Combined the different ions produced by the same compound).This may be the rea-son why the toxicity of PME was stronger than other extracts. Based on the cytotoxicity analysis,the order of extract toxicity was PME>PMW>PMPE>PMPW.The AST,ALT,and LDH leakage also reflected this trend.Furthermore,the results of the correlation anal-ysis(Fig.6B)showed that the levels of the compounds inside the box were consistent with this trend.Characterization of these com-pounds is shown in Table6B(Combined the different ions produced by the same compound).The toxicity of PMPE was stronger than PMPW.The results of the OPLS-DA may explain why PMPE toxic-ity was stronger than PMPW.The analysis revealed11compounds that may be associated with PM hepatotoxicity(Table6).In the investigation of the reference samples,emodin showed severe cytotoxicity in L-02cells,is an important chemical con-stituent that induces liver cell damage[25–28].Our results coincided with these conclusions,as9of the11compounds were anthraquinone or its’derivatives,and7of the9compounds were emodin and its’derivatives,suggesting that PM toxicity may be caused by emodin and its derivatives.The results of the toxic-ity test of emodin showed that emodin has strong cytotoxicity indeed(Supplementary data3).Emodin derivatives are inher-ently toxic,or they produce toxicity by their metabolites,such as emodin,which requires further research.The other two compounds were torachrysone-8-O-glc and tetrahydroxystilbene-O-(galloyl)-hex.However,research regarding these compounds is limited and they require further study.4.ConclusionsPM hepatotoxicity is of great concern;however,the com-ponents that induce hepatotoxicity are unknown.Besides,the present studies also have many contradictions,such as some stud-ies demonstrated that PMW has greater toxicity than other extracts [29,30],but another study considered as PME[31].In this assays,it is proved that the water extractions were less toxic than the70% ethanol extractions which indicated that water decoction was a more rational extractive agent for PM and PMP.Judging from the results,the processing produce could reduce the cytotoxicity of PM to a certain extend.The difference in the toxicity above-mentioned was related to the ingredients differences(both in species and con-tents)in the four extracts.The omics research method,which was based on PCA and OPLS-DA,was introduced into the study.The purpose is tofind out the toxic ingredients instead of biomarkers. Because traditional Chinese medicine(TCM)is complex compo-nents,it is hardly to obtain and research each ingredient in TCM. But,we provide an idea can greatly narrow the scope of the tar-get compounds and also greatly reduce the effort and cost in the research,just shown as Fig.7.Under this method,the results of our data analysis and toxicity study suggest that PMP toxi-city may be related to anthraquinone,emodin-O-(malonyl)-hex, emodin-O-glc,emodin,emodin-O-glc,emodin-8-O-glc,emodin-O-(acetyl)-hex,and emodin-O-hex-sulphate.The toxic of these components require further study.Appendix A.Supplementary dataSupplementary data associated with this article can be found,in the online version,at /10.1016/j.jhazmat.2015.06. 014References[1]S.J.Song,F.F.Li,H.Yue,Z.W.Yin,Study on the anti-aging effects of radixpolygnum multiflorum,J.Hebei Med.Univ.24(2003)90–91.[2]G.Y.Lv,Z.H.Lou,S.H.Chen,H.Gu,L.T.Shan,Pharmacokinetics and tissuedistribution of2,3,5,4-tetrahydroxystilbene-2-O-␤-glucoside from traditional Chinese medicine Polygonum multiflorum following oral administration to rats,J.Ethnopharm.137(2011)449–456.[3]L.Zhang,Y.C.Rui,Y.Qiu,T.J.Li,H.J.Liu,W.S.Chen,Expression of VEGF inendothelial cells and the effects of2,3,5,4 -tetrahydroxystilbene-2-O-␤-d-glucoside,Acta Pharm.Sin.39(2004) 406–409.。