Evaluation of sample extracting methods of FCSM by Lactobacillus acidophilus based on a UPLC-Q-TOF-MS global metabolomics analysis

(1)

h ttp : / / w w w . b j m i c r o b i o l . c o m . b r /

Biotechnology

and

Industrial

Microbiology

Evaluation

of

sample

extracting

methods

of

FCSM

by

Lactobacillus

acidophilus

based

on

a

UPLC-Q-TOF-MS

global

metabolomics

analysis

Yongqiang

Wang

a,b

_,

_Wenju

_Zhang

a,∗

_,

_Cunxi

_Nie

a

_,

_Cheng

_Chen

a

_,

_Xiaoyang

_Zhang

a

_,

Jianhe

Hu

b

a_Shihezi_University,_College_of_Animal_Science_and_Technology,_Shihezi,_Xinxiang,_PR_China

b_Henan_Institute_of_Science_and_Technology,_College_of_Animal_Science_and_Technology,_Xinxiang,_Henan,_PR_China

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received29March2017 Accepted3August2017 Availableonline31October2017 AssociateEditor:RosaneSchwan

Keywords: Metabolomics Extract Fermentation FCSM Lactobacillusacidophilus UPLC-Q-TOF-MS

a

b

s

t

r

a

c

t

Thestudyofmetabolomicsrequiresextractingasmanymetabolitesaspossiblefroma biologicalsample.Thisstudyaimedtodeterminetheoptimalmethodfortheextractionof metabolitesfromsolid-statefermentedcottonseedmeal(FCSM).TheUPLC-Q-TOF-MSglobal metabolomicstechnologywasusedtodetectthemetabolitesinFCSM,andtheextraction quantityandextractionefficiencyofsevendifferentextractionmethods,specificallythe WA,50MeOH,50MeOHB,50MeCNB,80MeOHB,80MeOHandAMFmethodswereevaluated. TheresultsshowedthatthenumberofVIPmetabolitesextractedbyAMFmethodare196 and184inESI+andESI−moderespectively,itisthelargestnumberofallexacted meth-ods;andtheAMFmethodsalsoprovidedahigherextractionefficiencycomparedwiththe othermethods,especiallyinindoleacrylicacid,dl-tryptophanandepicatechin(p<0.01).As aresult,AMF/−4◦_C_method_was_identified_as_the_best_method_for_the_extraction_of

metabo-litesfromFCSMbyLactobacillusacidophilus.Ourstudyestablishesatechnicalbasisforfuture metabolomicsresearchoffermentedfeed.

Introduction

Cottonseedmeal(CSM)isthemostimportantplantproteinin theworld.However,itsapplicationinanimalhusbandryis lim-itedduetoitscomplexcontent,whichincludesfreegossypol (FG),cyclopropenefattyacids(CPFA)andotheranti-nutritional factors,suchasahighﬁbrecontent,poorproteinproﬁle,etc.

∗ _{Corresponding}_author.

E-mail:zhangwj1022@sina.com(W.Zhang).

ThemostcriticalprobleminCSMisFG,whichhasanegative effect onanimalproduction.1–3 _Microbial_fermentation_was usedtopromotetheapplicationandpopularizationofCSM because itcould reducethe FGeffectively andimprove the nutritionalvalueofCSM,asreportedrecently.4–7_{Lactobacillus}

acidophilus isan importantlactic acid bacteria and intesti-nalmicrobethatregulatestheintestinalmicro-ﬂorabalance, enhancesimmunity,andreducescholesterollevels.8–10_But_its useforfeedfermentationremainsrareandfewstudieshave investigatedsolid-statefermentedcottonseedmeal(FCSM)by

L.acidophilus.

https://doi.org/10.1016/j.bjm.2017.08.008

(2)

brazilian journal of microbiology49(2018)392–400

393

A prerequisite for a metabolomics study of the target metabolitesofFCSMbyL.acidophilusistheoptimizationof the pretreatment of the biological samples, including sci-entificsampling, reasonablequenching and high-efficiency extraction.11,12 _The_ideal_method _should_produce_a _sample ofallintracellularand extracellularmetabolites simultane-ouslyfortheanalysisofcomplexbiologicalsamplesforglobal metabolomics.13_Both_{intracellular}_and_{extracellular} metabo-lites exert a significant effect on the results of microbial metabolomicstudies.14_Fermented_feed_is_a_complex biolog-icalsample;thefermentationproductsnotonlyincludethe metabolitesofmicrobial fermentation but alsoincludethe fermentationsubstratenutrients.Totrulyreflectthe fermen-tationprocess,thepretreatmentmethodsofquenchingand extractingmetabolitesmustbeoptimized.

Previousstudiesofthepretreatmentofbiologicalsamples havedemonstrated thatdifferentquenchingandextracting methodsweresuitablefordifferentmicroorganisms,suchas

Escherichiacoli,Yeast,Lacticacidbacteria,etc.15–19 _These stud-iesmainlyincludedextractingmicroorganismmetabolitesby screeningdifferentpretreatmentmethodsandexaminingthe effectsofdifferentquenching solventsontheextractionof metabolitesandsamplegrindingdegree.However,these eval-uationsstudiedhowtousedifferentresearchplatformsand indicatorsoftheextractionmethod.Themajorityofstudies wereoftheextractionofdifferentquenchingsolvents,with aresearchplatformbasedontraditionalGC–MSandLC–MS. RecentdevelopmentshaveprovedthatUPLC-Q-TOF-MScould detect complex multi-component substances more effec-tively,faster andmoreaccuratelythan other methods.20–22 Atthesametime,theUPLC-Q-TOF-MSplatformcanalsobe analysed quantitatively for the number of metabolites. So far,afewapplicationsofL.acidophilustometabolomicshave beenstudied. SinceL.acidophilusfermentedproteinfeed is rarefortraditionalfeedproduction,themethodofextracting metabolitesfromFCSMhasnotreceivedsufﬁcientattention. Therefore,itwasdesirabletoestablishanextractionmethod thatcanbeappliedtoawide rangeofmetabolitesfrom L. acidophilus.

ToidentifytheﬁnalmetabolitesinFCSMbyL.acidophilus,

bothintracellularandextracellularmetabolitesmustbe anal-ysed. We used the global metabolomics profiling way to evaluate the efficiency of seven extraction methods (WA, 50MeOH,50MeOHB,50MeCNB,80MeOHB,80MeOHandAMF) systematicallywiththeaimofidentifyingthebestmethodfor theextractionfromtheFCSM.Sevendifferentpretreatment methodswereevaluatedthroughaglobalmetabolomics anal-ysisbasedonUPLC-Q-TOF-MS. Thequantity andefficiency ofthemetabolitesextractedfromFCSMwereevaluated.This studywillprovideagoodbasisforfutureanalysesofthetotal metabolitesintheFCSM.

Materials

and

methods

Chemicals

LC–MS-grademethanolandacetonitrilewerepurchasedfrom FisherChemical(FisherScientiﬁc,USA)orSigma–Aldrich(St.

Louis,MO,USA).ThewaterwaspuriﬁedwithanEPED-E2-T device(Nanjing,China).

Samplepreparation

L.acidophilus(JJ-0787)wasacquiredfromtheChinaCenterof IndustrialCultureCollection(CICC).Fermentationsubstrates werecomposedof65.89%CSMand34.11%corn,withatotal proteinof30%.JJ-0787strainswereculturedat37◦Cfor24h inMRSbroth(SolarbioLifeSciencesLtd.,Beijing,China),and theninoculated6mLJJ-0787strainscultureﬂuidin100g ster-ilizedfermentationsubstrate (themoisturecontentis40%), mixedthoroughly,sealedinthetrianglebottle,andthen stat-ically placed in an incubator at 37◦C for48h. Each group containedthreerepetitions.TheresultingsamplesofFCSM wereimmediatelyfrozeninliquidnitrogenandthenstoredat −80◦_C_until_analysis.

Metaboliteextraction

Toinvestigatetheoptimalmethodfortheextractionofthe metabolites inFCSM, thedifferent quenchingsolvents and pretreatment methods used in the studies conducted by TokuokaandChenweremodiﬁedasfollows.18,19

WA(J1group):AsampleofFCSMwastreatedwithhigh-speed homogenization (HHM).Then,a0.2-g samplewas rapidly quenchedanddissolvedin2mLofultra-purewater(4◦C). Thesampleswerequicklyhomogenizedfor10minusinga high-fluxtissuehomogenizer(TisssueLyserII,Qiagen, Ger-many).Thecentrifugetubeswereincubatedfor30minonice. Theextractsolutionwascentrifuged(−20◦C,10,000r/min, 5min), and the residual debris was removed. The super-natantwasdissolvedin10volumesofultrapurewater(4◦C) andfilteredwitha0.22-␮msyringefilter(004022NL-SEPTFE, Shanghai,China).Thesampleswerestoredat−80◦_C_until

furtheranalysis.

50MeOH (J2 group): Theextraction process wasthe same as that usedin the WAmethod exceptthat the quench-ingsolventwasreplacedbya50%coldmethanolsolution (methanol:water=50:50,−20◦_C).

50MeOHB (J3 group): The extraction process was identi-cal tothat used in50MeOH methodexcept that the 50% coldmethanol solution (−20◦_C) _was _replaced _with_a _50%

coldmethanolsolution(room temperature),and afterthe addition ofquenchingsolventto thesample,the method incorporatedaboilingprocedure(70◦C,5min)immediately followedbythesameprocessasusedin50MeOHmethod. 50MeCNB (J4 group): The extraction procedures were identicaltothoseusedin50MeOHBmethodwiththe excep-tion that the 50% methanol solution (room temperature) was replaced by a 50% acetonitrile solution (acetoni-trile:water=50:50,roomtemperature).

80MeOHB (J5 group): The 50% methanol solution used in 50MeOHBmethodwasreplacedwitha80%methanol solu-tion(methanol:water=80:20,roomtemperature),followedby thesamemethodasin50MeOHB.

80MeOH(J6Group):Theextractionprocedureswereidentical tothoseusedin50MeOHmethodexceptthatthe50%cold

(3)

methanolsolutionwasreplacedwithan80%methanol solu-tion(methanol:water=80:20,−20◦_C),_followed_by_the_same

methodasin50MeOH.

AMF(J7group):The50%coldmethanolsolutionusedinthe 50MeOHmethodwasreplacedbyacoldAMFsolution (ace-tonitrile:methanol:water=2:2:1,formicacid0.1mol/L,−4◦_C),

andtheothermethodologicaldetailsarethesameasthose usedinthe50MeOHmethod.

Thesampleswerediluted20-foldwiththecorresponding extractionsolvent,vortexedandcentrifugedat13,200r/min for2min. A150␮Lvolumeofsupernatantfrom each sam-plewasanalysedbyUPLC-Q-TOF-MS.Finally,themetabolites wereidentiﬁedandanalysedbasedontheobtaineddata.

UPLC-Q-TOF-MSdetectionmethod

Thechromatographicconditionswereasfollows:Theanalysis wasperformedonanACQUITYUPLCBEH,whichwas conﬁg-uredwithdirectinjectionusingaC18column(100m,2.1mm, 1.7mm,Waters,USA).Thecolumnvolumewas10␮L,andthe columntemperaturewassetto35◦C.ThemobilephaseAwas ultrapurewater,whilemobilephaseBwasHPLC-grade ace-tonitrile with0.1%(v/v)formicacid,and the ﬂowratewas 0.35mL/minstartingwith2%acetonitrileuptothefull gra-dientof98%inmobilephaseB(0–12min).Afterequilibration for1min,mobilephaseBrangedfrom98%to2%for1min andequilibrationfor1minafterthesamplewascollected.The injectionvolumewas2␮L,thecolumntemperaturewasset to35◦C,andtheautosamplertemperaturewasmaintainedat 4◦C.

Massspectrometryconditions:Anelectrosprayionization (ESI) sourcewas usedto scaninthe positive and negative ionization modes. Theion source temperature was 120◦C, and the desolvation temperature was 500◦C.The nitrogen ﬂowrateforsolventevaporationwas 800L/h,and theﬂow rate of tapered blowing nitrogen was 50L/h. The positive-andnegative-ionmodecapillaryionizationvoltageswere3.0 and2.5kV,respectively;thesamplingconevoltageswere27 and30V,respectively;thecollisionenergywas6eV;andthe samplingconevoltageswere27and30V,respectively.The col-lisionenergywas6eV,andthequadrupolescanningrangewas between50and1000m/z.

Datastatisticsandanalysis

TherawdatadetectedbyUPLC-Q-TOF-MSwereanalysedby XCMSsoftware(http://xcmsonline.scripps.edu)forpeak iden-tification and matching.23 _The _{chromatographic} _peak _data werenormalizeduniformly,andthe multidimensionaldata wereanalysedusingSIMCA-Psoftware(version13.0,Sweden). Datapre-processingwasperformedusingtheParetoScaling method.A partialleastsquares-discriminateanalysis (PLS-DA)wasthenusedtoperformsupervisedmultidimensional statisticalanalysis,andthepermutationtestwasusedto pre-vent over-fittingofthe PLS-DA model.24 _The_{differentiated} expressedmetabolites(VIPmetabolites)wereidentifiedand thenumberofdifferentmetaboliteswerecomparedthrougha

t-test(pvalue<0.05)andfold-changes(foldchange>2or<0.5), revealingamultiplerelationshipbetweenthetwogroups.22

Tandemmassspectrometrywasusedtoidentify metabo-litesbasedonexactmolecularweights(molecularweighterror of<30ppm).TheMasslynx i-FITalgorithmwasusedto fur-therconfirmthepossiblestructuralformulasaccordingtothe isotopeprinciple. Subsequently,accurateMS/MS’s fragmen-tation informationwasqueriedinthe HumanMetabolome Database(HMDB,http://www.hmdb.cawebcite)andtheMetlin database (http://metlin.scripps.edu/website) forthe screen-ingandidentificationofpotentialbiomarkersubstances.25–27 ThedatawereanalysedbyANOVAusingtheSPSS22.0 statis-tical software.Multiple comparisonswereperformed using the Duncan method. Thetest data were expressed as the mean±SD,andp<0.05wasusedasthecriterionforsignificant differences.

Results

PLS-DAanalysis

ThePLS-DA3Dscoring chart(Fig. 1)shows thatthe differ-ent metabolite extraction methodsare clearly divided into three-dimensionalspatialarrangements.Theresultsshowed signiﬁcant differences in type, amount and concentration ofmetabolitesextractedfromdifferentFCSMsamplesusing thedifferentpretreatmentmethods.Allofthesampleswere withinthe95%conﬁdenceintervalandalmostallofthegroups weredistinguishablefromtheothers.

Analysisofthenumberofmetabolites

Thenumbersofmetabolitesextractedusingdifferent meth-odsareshowninFig.2.Preliminaryevaluationofthemerits oftheextractionmethoddependingontheamountof small-moleculemetabolitesdetectedbythedifferentionscan(ESI+ andESI−)modescanallowacomprehensively characteriza-tionofthesamples.

AsshowninFig.2A,ananalysisofthemetabolitesrevealed byscanningintheESI+modeshowedthattheJ1grouphadthe lowestandtheJ6grouphadthehighestnumbersof metabo-lites. The number ofextracted metabolitesin the J6,J2, J3 andJ4groupswereincreased12.55,11.70,11.08and10.98%, respectively,comparedwiththatintheJ1group(p<0.05).The numbersofmetabolitesintheJ5andJ7groupswerehigher thanthatintheJ1group,butthedifferencewasnotsigniﬁcant (p>0.05).Thegroupscanbesortedasfollowsbasedonthe numberofmetabolites,J6>J2>J3>J4>J5>J7>J1.After scan-ning inthe ESI− mode, thenumbers ofmetabolites inthe J1groupisthefewestinallgroups,andtheJ2groupwasthe highestinallgroups.Thenumbersofextractedmetabolitesin theJ2,J5,J3andJ6groupswereincreasedby20.91,18.85,17.86 and17.15%,respectively,comparedwiththeJ1group(p<0.05). TheJ4andJ7groupshadmoremetabolitesthantheJ1group, butthedifferenceswerenotsigniﬁcant(p>0.05).Thegroups canbeorderedbasedonthenumberofmetabolitesasfollows: J2>J5>J3>J6>J4>J7>J1.

Fig. 2B shows the numbers of VIP metabolites in each group compared withthe control group. Afterscanning in both theESI+ andESI− modes,the J7group hadmore VIP metabolites than the other groups, and the J3 group had

(4)

395

-10 10 20 30 0 -20

A

50

B

40 30 20 10 -10 -20 20 0 15 10 5 0 30 20 10 -10 -20 0 -20 0 -10 10 20 30 40 50 20 15 10 5 0 t[3] t[2] t[1] t[1] t[3] t[2] -10 -5 0 5 10 30 20 10 10 -20 -30 -40 -5015 10 5 -5 -10 -10 -5 0 5 10 -15 -20 0 0 15 10 5 0_-5 -10 -20 -15 30 20 10 -10 -20 -30 -40 -50 0 J1-WA J2-50MeOH J3-50MeHB J4-50MeCNB J5-80MeCNB J6-80MeOH J7-AMF

Fig.1–PLS-DA3Dscoreplotsfortheextractedmetabolitesofdifferentpretreatments.(A)PLS-DA3DscoreplotinESI+.(B) PLS-DA3DscoreplotinESI−.t[1]–t[3]indicatingtheseparationbetweendifferentgroups.

J1 J2 J3 J4 J5 J6 J7

A

2500

B

2250 2000 1750 1500 1250 1000 750 500 Number of metabolites Number of VIP metabolites ESI+ ESI+ ESI- ESI-b b b a a a _a a a a a ab ab _ab ₂₀₀ 150 100 50 0 J2-J1 J3-J1 J4-J1 J5-J1 J6-J1 J7-J1 34 104 20 59 44 117 30 56 135 142 196 184

Fig.2–ComparisonofthenumberofmetabolitesextractedwithdifferentpretreatmentinESI+andESI−mode.(A) Comparisonofthenumberofmetabolitesaftermolecularfeatureextraction.Differentlowercaselettersmeansthatthere aresigniﬁcantlydifferent(p<0.05)ineachsameseriousbar.(B)ComparisonofthenumberofVIPmetabolites(foldchange >2or<0.5)betweeneveryexperimentgroupandcontrolgroup.J1–WA,J2–50MeOH,J3–50MeOHB,J4–MeCNB,J5– 80MeOHB,J6–80MeOH,J7–AMF.

the fewest VIP metabolites. The groups can beordered as follows basedon the number of VIP metabolitesidentiﬁed inthe EDI+ and ESI− modes was J7>J6>J4>J2>J5>J3 and J7>J6>J5>J4>J2>J3,respectively.

ComparisonofextractionefﬁciencyofcommonVIP metabolites

Themetabolitesamplesofdifferentextractionmethodswere analysedbyUPLC-Q-TOF-MS.Fifteencommonpeakvariables werescreenedintheESI+andESI−modes,and15kindsof small-moleculemetaboliteswereidentifiedbym/zandRT cor-relationdatabases,asshowninTable1.Comparedwiththe normalizedpeakareaofthemetabolites,thepeakarea val-uesofdl-tryptophan,indoleacrylicandepicatechininseven groupswere very significantly different(p<0.01). Thepeak areavalueofdl-tryptophan washigher intheJ6,J5, J7,J2, andJ3groups.TheJ6groupshowedthehighestvalue,butthe differencebetweenthefivegroupswasnotsignificant.There weresignificantdifferencesbetweenthefivegroupscompared

withthoseingroupsJ1andJ4(p<0.01).Thelargestpeakarea forthe indoleacrylic acid levelwas foundforthe J7 group, andthisgrouprevealedverysignificant(p<0.01)differences comparedwiththeJ1andJ4groups,butthedifferenceswith theJ6andJ2groupswerenotsignificant(p>0.05).Thehighest peakareaofepicatechinwasfoundfortheJ7group,butthis areawasnotsignificantly(p>0.05)differentcomparedwith the J6group.TheJ7grouppresentedsignificant differences comparedwiththeJ5,J2,J3,andJ4groups(p<0.05)andavery significantdifferencecomparedwiththeJ1group(p<0.01).In additiontotheabove-mentionedmetabolites,therewasa sig-nificantdifferencebetweencAMPandlacticacid.Forthepeak areaofcAMP,theJ2andJ3groupwerehigherthantheJ1groups (p<0.05),buttherewasnosignificantdifferencewiththeother groups(p>0.05).Forthepeakareaoflacticacid,theJ1group wasthehighestandtherewasasignificantdifferencebetween theJ7andJ1groups(p<0.05).However,theothergroupsand the J1 group had nosignificant differences (p>0.05). There wasnosignificantdifferencebetweentheothermetabolites (p>0.05)besidesabovemetabolites.

(5)

b r a z i l i a n j o u r n a l o f m i c r o b i o l o g y 4 9 (2 0 1 8) 392–400

Table1–Normalizedpeakareaofmetabolitesextractedbydifferentquenchingmethod.

ID Scanning mode m/z RT Metabolite identity Normalized peakarea

J1 WA J250MeOH J350MeOHB J4 MeCNB J580MeOHB J6 80MeOH J7 AMF p sem

11 ESI− 89.03 54.06 Lacticacid 1.89±1.62a 1.33±0.05ab 1.52±0.13ab 1.34±0.13ab 1.52±0.13ab 1.56±0.10ab 0.67±0.10b 0.41 0.14

84 ESI− 146.05 51.94 l-Glutamate 1.71±1.27 2.17±0.21 2.04±0.18 1.62±0.19 1.76±0.19 1.78±0.23 1.76±0.23 0.83 0.11

122 ESI− 164.07 118.8 dl-Phenylalanine 0.11±0.09 0.14±0.02 0.18±0.01 0.13±0.01 0.15±0.01 0.18±0.02 0.14±0.02 0.29 0.01

125 ESI− 165.06 219.26 Phenyllacticacid 1.01±0.82 1.75±0.09 1.38±0.40 1.85±0.51 1.77±0.51 1.68±0.49 1.60±0.49 0.43 0.11

142 ESI− 173.11 47.45 l-Arginine 4.34±2.74 6.02±0.28 6.08±0.41 5.14±0.58 5.28±0.58 5.31±0.56 5.05±0.56 0.55 0.24

152 ESI− 179.06 53.12 ␣-d-Glucose 4.80±3.83 3.99±0.36 4.20±0.22 3.77±0.53 4.07±0.53 4.21±0.12 3.13±0.12 0.9 0.32

186 ESI− 195.05 53.11 Gluconicacid 3.42±2.88 4.03±0.43 3.73±0.59 2.68±0.66 1.86±0.66 1.93±0.37 1.59±0.37 0.11 0.25

206 ESI− 203.08 144.56 dl-Tryptophan 1.30±1.09b 2.79±0.08a 2.68±0.37a 1.14±0.1b 3.17±0.10a 3.22±0.35a 2.80±0.35a <0.01 0.11

245 ESI+ 175.12 47.65 l-Asparticacid 0.56±0.55 0.74±0.09 0.61±0.05 0.55±0.14 0.50±0.14 0.59±0.15 0.61±0.15 0.92 0.05

275 ESI+ 188.07 148.17 Indoleacrylic

acid

2.79±2.55bc 5.15±0.45ab 4.77±0.51b 1.44±0.31c 4.40±0.31b 5.67±1.04ab 7.18±1.04a <0.01 0.27

531 ESI− 328.05 88.37 cAMP 0.88±0.76b 1.63±0.18a 1.60±0.19a 1.40±0.38ab 1.37±0.38ab 1.49±0.26ab 1.26±0.26ab 0.27 0.08

585 ESI− 344.04 92.14 cGMP 2.58±2.19 4.11±0.21 3.79±0.41 3.73±0.75 3.24±0.75 3.46±0.45 2.35±0.45 0.25 0.2

723 ESI+ 291.08 153.99 Epicatechin 0.04±0.06c 0.21±0.04b 0.16±0.07b 0.11±0.03bc 0.22±0.03b 0.27±0.04ab 0.36±0.04a <0.01 0.01

973 ESI+ 343.12 54.15 Beta-d-lactose 0.42±0.31 0.57±0.05 0.54±0.07 0.47±0.09 0.44±0.09 0.51±0.05 0.39±0.05 0.64 0.03

1487 ESI+ 451.36 355.6 Phylloquinone 7.90±2.31 7.26±0.26 5.12±4.40 5.18±4.43 8.09±4.43 7.62±0.36 3.71±0.36 0.3 0.56

RT,RetentionTime;ESI,ElectronicSprayIon;cAMP,cyclicAdenosinemonophosphate;cGMP,cyclicguanosinemonophosphate.

Note:p<0.05indicatethatthereissigniﬁcantlydifferenceinnormalizedpeakarea.p<0.01indicatethatthereisverysigniﬁcantlydifferenceinnormalizedpeakarea.Underlinednumberrepresent

(6)

397

Phylloquinone D-Glucose L-Arginine Gluconic acid Indoleacr ylic acid cGMP Lactic acid L-Glutamate DL-T ryptophan cAMP Phen yllactic acid L-Aspar tic acid BET A-D-LA CT OSE DL-Phen ylalanine Epic atechin F

old change (log2)

3 2 1 0 -1 J2-J1 J3-J1 J4-J1 J5-J1 J6-J1 J7-J1

Fig.3–Peakareasfoldchangeof15kindsofmetabolitesextractedfromFCSMwithdifferencepretreatments.Foldchange >0meansthattherelativecontentofmetabolitesisimproved,while<0meansthatitisreduced.J1–WA,J2–50MeOH,J3– 50MeOHB,J4–MeCNB,J5–80MeOHB,J6–80MeOH,J7–AMF.

Thefoldchangesinpeakareawerecalculatedusingthe formulalog2fortheratioofnormalizedpeakareavaluesof themetabolitesextractedbydifferentmethodstothecontrol group.Thefoldchangeinthepeakareareﬂectedthedegree ofchangeinmetaboliteconcentrationbetweenthetrialgroup and the controlgroup.As Fig. 3shows, the comparisonof ploidy areavariation isseen inapproximately 15 common smallmoleculemetabolitesfromFCSMbydifferentextraction methods.Itcanbeseenthatthechangeinthefoldchangearea betweenJ7andJ1groupwasthemostsigniﬁcant(p<0.05).

Discussion

Samplepretreatmentisakeystepinthestudyofbiological system metabolites because it directly affects the extrac-tionefﬁciencyofintracellularandextracellularmetabolites fromfermentedfeedsamples.Soitisnecessarytostudythe pretreatment methods usedin metabolomics study of fer-mentedfeeds.Manypretreatmentmethodsforthequenching andextracting ofmetabolitesfrom biologicalsampleshave beenproposed;however,sinceKoningandDamproposeda rapidquenchingmethodinvolvingthedirectlyaddition60% methanol at −40◦_C _and _followed _by _{centrifugation} _of _the

cells,28 _this _technique _has _not _been _changed _and _remains the most common method for the sampling of microbial cultures.15,29,30 _By _comparing _different_quenching _methods forE.coli,Winderalsodemonstratedthat60%coldmethanol solution(−48◦_C)_is_the_most_suitable_E._coli_quench_method

andrecommendedmultiplecyclesof−48◦_C_100%_methanol

freeze–thaw extraction to extract metabolites effectively.17 However,thestudyfoundthatdifferentmethanol/waterratios

have differentmetaboliceffects signiﬁcantly,with thecold methanol/water extractionmethod extracting triphosphate compoundsandotherenergeticcompoundsinE.coli metabo-lites(e.g.,ATP,CTP,UTP,etc.)andthatcoldmethanol/water extractionwasnotnecessarilythebestwaytoextractE.coli

intracellular metabolites.Basedon thestudy conductedby Kimball,31 _Rabinowitz _compared_the _effects _of ₁₀ quench-ing solventson the extractionofE. colimetaboliteson the basisofLC/MSanalysisofthepeakheightofeachpeak com-paredtothe totalmetaboliteyield.32 _The_study _found_that theoptimalextractionyieldwasobtainedusingAMF(acidic acetonitrile/methanol/aqueoussolution).Tokuokaconducted a systematic study of 11 methods for the quenching of

Aspergillus oryzae-fermented samples using 50MeCNB (50% acetonitrile/water)astheextractionsolvent.18 _The_use _of_a waterbathat70◦Cwaterbathfor5minwasidentiﬁedasthe mostefﬁcientextractionmethod.Methodsinvolving quench-ingwith60%methanol/water,80%methanol/waterand80% methanol/glycerol at−40◦_C _were _used _to _study _the _effect

of 80% methanol/water on extraction. These solvents can effectivelyreducethedegreeofmetaboliteleakageand intra-cellular metabolite levels more suitable for the quenching of Lactobacillusbulgaricus.19 _Different_samples _and _research objectives are associated withdifferent extraction require-ments, and the optimal extraction method will vary from that recommendedbythe abovestudy.Atpresent,thereis nostandardoperatingprocedure(SOP)fortheextractionof differentbiologicalsamples.Toextractthemetabolitesfrom different samples,the corresponding pretreatment method wasused.Itrequiredoptimizationandstandardizationofthe samplequenchingandextractionmethod,whichistheonly wayforthefuturedevelopmentofmetabolomics.Basedon

(7)

theresultsoftheabovestudies,methanol,acetonitrileand acidicacetonitrile/methanol/water(AMF)werechosenasthe threemainquenchingsolventsforourstudy.Wealsosetup twogradient(50%,80%)quenchingsolventstostudy. Simulta-neously,thedifferencesinextractionmethodsbetweencold methanolandhotmethanolwereinvestigated.Thecontrol groupwasselectedbyultra-purewaterextraction.Thus,in thisexperiment,wehavechosensevenextracting methods forsystematicassessment.

The development of LC/MS, GC/MS and other technol-ogyplatforms allowsfor the simultaneous qualitative and quantitative determination of a large number of probiotic fermentation metabolites. In this study, UPLC-Q-TOF-MS was used as the research platform. UPLC and Time-of-FlightMassSpectrometry(TOF-MS)canbeusedtoperform multi-componentanalysesofcomplexmaterials. Ultra-high-performanceliquidchromatography(UPLC)wouldbeagreat boost in the field of microbial fermentation metabolites research.20–22,33,34 _The _platform _ensures _the _accuracy, _high efficiency and reliability ofthe test results. In the present study, we foundthat 15 different small molecule metabo-litescanbeextractedbysevendifferentextractionmethods through the global metabolomic analysis, including amino acids,organicacids,vitaminsandsmallmolecularactive sub-stancesassociatedwiththemetabolitesofbeneficialmicrobial fermentationproteinfeeds.Thesemetabolitesplayan impor-tantrole inthegrowth,developmentand immunizationof animals.

The number of extracted metabolites and the number ofVIP metaboliteswereusedasthepreliminaryevaluation indexes,especiallythenumberofVIPmetabolite,itcanreflect the extractionaccuracyforthe metabolites. From the PLS-DA3D score(Fig. 1),it wasfoundthat differentquenching extractionmethodshaddifferentextractioneffects.The num-berofdetectablepeakswasalsosignificantlydifferent,which meansthattherewasacleardistinctionbetweenthenumber ofdetectablemetabolitesandhowmuchthenumberofVIP metabolitesreflectedthenumberofsmallmolecule metabo-litesthatmayplayaroleinthedetectionofmetabolites.In contrasttothenumberofdetectablemetabolites,itwasclear fromFig.2Athat80MeOH(J6)wasoptimalinthepositiveion mode(ESI+)and50MeOH(J2)wasoptimalinthenegativeion mode(ESI−).Thepositiveandnegativeionmodemetabolite sumrankingtrendwas50MeOH(J2)>80MeOH(J6)>50MeOHB (J3)>80MeOHB (J5)>50MeCNB (J4)>AMF(J7)>WA (J1). How-ever,incontrasttothe numberofVIP metabolites(Fig.2B), we found that the ranking trend was AMF (J7)>80MeOH (J6)>80MeOHB (J5)>50MeCNB (J4)>50MeOH (J2)>50MeOHB (J3) in positive and negative modes. The AMF method (J7) has a significant advantage in either positive or negative ion modes. To compare the extraction efficiency of dif-ferent methods of quench extraction, the fifteen common metabolitesextractedfromsevenmethodswerescreenedand the normalized peak area and fold change index of these metaboliteswere compared(Table1andFig.3).Theresults showed that significantly different metabolites included indoleacrylicacid,dl-tryptophanandepicatechin.The extrac-tionefficiencyofthesemetaboliteswasthebestinAMF(J7), followedby80MeOH(J6)and50MeOH (J2).Itisworth men-tioningthatfor50MeOHB(J3),thepeakareavalueofextracts

ofthesemetaboliteswasalsohigher,withanextraction efﬁ-ciencysecondonlytotheabovethreemethods.

The differences in the extraction efficiency of various methodsmaybecloselyrelatedtothepolarity,watercontent and temperatureoftheextractionsolvent.Allthe reagents used in this experimentwere analytical-gradeorganic sol-vents,inwhichthepolarityofmethanolishigherthanthat ofacetonitrile.However,thewatercontentoftheextraction solvent willalsoaffectthe polarity.Thecontentofsolvent moisture inthe AMF, 80MeOH and 80MeOHBsolvents was 20%;the50MeOH,50MeOHBand50MeCNBsolventmoisture contentwas50%;andthewatercontentofthecontrolgroup (J1)reached100%.Thestudyshowedthatthewatercontent inthesolventdirectlyaffectstheextractionefficiency,which wereconsistentwithKimball’sreport.31_Kimball_proved_that thehighyieldofnucleosidewasrelatedtothedecomposition ofnucleotidesintheextractionsolventenrichedinwater.The differentconcentrationsofcoldmethanol/waterleadahuge deviation inthe extracted metabolites. A higher thewater content,aworsetheextractionefficiencyofmetabolites.In thisstudy,theextractiontemperaturesofthedifferent extrac-tion methods werealso different,50% acetonitrile at−4◦_C

and a water bath at70◦C during the extraction. The 50% and80%hotmethanolsolutionwerealsoheatedinawater bathat70◦C,and thetemperature ofthe ultrapurewater was−4◦_C._Some_studies _have_conﬁrmed_that_low

tempera-tureextractioncanextractmetabolitesmoreefﬁciently.15,17 Winderevenproposedamethodofmulti-cyclefreeze–thaw extractionofmetabolitesusing−48◦_C_100%_methanol.

How-ever,somestudieshaveconﬁrmedthatheatextractionisalso aneffectivemethodofextractingmetabolites.17_For_example, amethodforquenchingfermentedsamplesofA.oryzaewas studiedusing50%acetonitrile/water(v:v=50:50)for5minin a70◦Cwaterbath,18_but_the_speciﬁc_reason_for_the tempera-turedifferenceintheextractionofmetabolitesremainstobe elucidated.

The results showed that the quality performance of the seven quenching and extraction methods was AMF>80MeOH>50MeOH>80MeOHB>50MeOHB>50%MeCNB>WA. Therefore, wecan see that the AMF method was the best methodtoextractFCSMwithL.acidophilus.Thesetestresults were consistent with Rabinowitz.32 _His _experiments con-ﬁrmed thatthe acidic acetonitrile/methanol/water solution not onlyreduced the degradation of triphosphatebut also reduced the lossofhigh-energymetabolites and promoted its conversion to low-energy derivatives, which ultimately improved the extraction concentration of metabolites. It hasahigherextractionefﬁciency,butRabinowitz’sresearch objectwasE.coli,andthisstudywasofL.acidophilusFCSM, indicatingthatthemethodalsohasapplicability.

Conclusions

In conclusion, this study only focused on the comparison of metabolite extraction methods from FCSM with L. aci-dophilusformetabolomicprotocols.Theexperimentalresults showed thatthe−4◦_CAMF _solution_extraction_method_was

themostaccuraterepresentationandthehighestextraction ofthemetabolitesofFCSMbyLactobacillus.Sincemetabolites

(8)

399

differremarkablydependingonmicroorganismsandgrowth conditions,agoodextractionmethodmightonlyworkin cer-tainsituations.Astandardextractionmethodwasdifﬁcultto establish,anddifferentextractionmethodsmayneedtobe usedfordifferentstudyobjects.

Conﬂicts

of

interest

Theauthorsdeclarenoconﬂictsofinterest.

Acknowledgments

Weare very grateful to Suzhou BioNovoGene (http://www. bionovogene.com)fortechnicalsupportsinthe comprehen-siveanalysisoftheMS/MSdata.Thisstudywassupportedby theNationalNaturalScienceFoundationofChina(31360564) and Graduate Research & Innovation Project in Xinjiang AutonomousRegionofChina(XJGRI2014058).

r

e

f

e

r

e

n

c

e

s

1.YinYL,HuangRL,ZhangHY,etal.Nutritivevalueof feedstuffsanddietsforpigs.I.Chemicalcomposition, apparentilealandfecaldigestibility.AnimFeedSciTechnol. 1993;44(1–2):1–27.

2.TabatabaiF,GolianA,SalarmoeiniM.Determinationand DetoxiﬁcationMethodsofCottonseedMealGossypolforBroiler ChickenRations.Yazdi;2002.

3.NagalakshmiD,RamaRaoSV,PandaAK.Cottonseedmealin poultrydiets:areview.JPoultSci.2007;44(2):119–134. 4.ZhangWJ,XuZO,SunJY,etal.Studyonthereductionof

gossypollevelsbymixedculturesolidsubstrate

fermentationofcottonseedmeal.AsianAustralasJAnimSci. 2006;19(9):1314–1321.

5.ZhangWJ,XuZR,ZhaoSH,etal.Developmentofamicrobial fermentationprocessfordetoxiﬁcationofgossypolin cottonseedmeal.AnimFeedSciTechnol.2007;135(1–2): 176–186.

6.JiaXF,LiAK,YaoJH,etal.Effectofsolid-statefermentation ongossypoldetoxiﬁcationandproteindegradationin cottonseedmeal.JNorthwestA&FUniv.2009;28(264): 245–250.

7.VellaichamyM,ParvezN.Gossypoldetoxiﬁcationandlysine enrichmentincottonseedcakebysolidstatefermentation.J PureApplMicrobiol.2016;10(2):1333–1339.

8.VincentJG,VeomettRC,RileyRF.Antibacterialactivity associatedwithLactobacillusacidophilus.JBacteriol. 1959;78(4):477–484.

9.KashketER.Theprotonmotiveforceinbacteria:acritical assessmentofmethods.AnnuRevMicrobiol.1985;39:219–242. 10.JackRW,TaggJR,RayB.BacteriocinsofGram-positive

bacteria.MicrobiolRev.1995;59(1):171–200.

11.Taymaz-NikerelH,DeMM,RasC,etal.Developmentand applicationofadifferentialmethodforreliablemetabolome analysisinEscherichiacoli.AnalBiochem.2009;386(1):9–19. 12.CanelasAB,RasC,PierickAT,etal.Leakage-freerapid

quenchingtechniqueforyeastmetabolomics.Metabolomics. 2008;4(3):226–239.

13.HajjajH,BlancPJ,GomaG,Franc¸oisJ.Samplingtechniques andcomparativeextractionproceduresforquantitative determinationofintra-andextracellularmetabolitesin

ﬁlamentousfungi.FEMSMicrobiolLett.1998;164(1): 195–200.

14.SchaubJ,SchieslingC,ReussM,etal.Integratedsampling procedureformetabolomeanalysis.BiotechnolProg. 2006;22(5):1434–1442.

15.MaharjanRP,FerenciT.Globalmetaboliteanalysis:the inﬂuenceofextractionmethodologyonmetabolomeproﬁles ofEscherichiacoli.AnalBiochem.2003;313(1):145–154.

16.FaijesM,MarsAE,SmidEJ.Comparisonofquenchingand extractionmethodologiesformetabolomeanalysisof Lactobacillusplantarum.MicrobCellFactories.2007;6(1):27. 17.WinderCL,DunnWB,SchulerS.Globalmetabolicproﬁlingof

Escherichiacolicultures:anevaluationofmethodsfor quenchingandextractionofintracellularmetabolites.Anal Chem.2008;80(8):2939–2948.

18.TokuokaM,SawamuraN,KobayashiK,etal.Simple metaboliteextractionmethodformetabolicproﬁlingofthe fermentedAspergillusoryzae.JBiosciBioeng.

2010;110(6):665–669.

19.ChenMM,LiAL,SunMC,etal.Optimizationofthe

quenchingmethodformetabolomicsanalysisofLactobacillus bulgaricus.JZhejiangUnivSciB.2014;15(4):333–342.

20.RainvillePD,StumpfCL,ShockcorJP,etal.Novelapplication ofreversed-phaseUPLC-oaTOF-MSforlipidanalysisin complexbiologicalmixtures:anewtoolforlipidomics.J ProteomeRes.2007;6(2):552–558.

21.LacinaO,UrbanovaJ,PoustkaJ,etal.

Identiﬁcation/quantiﬁcationofmultiplepesticideresiduesin foodplantsbyultra-high-performanceliquid

chromatography-time-of-ﬂightmassspectrometry.J ChromatogrA.2010;1217(5):648–659.

22.TheodoridisG,GikaH,FranceschiP,etal.LC–MSbased globalmetaboliteproﬁlingofgrapes:solventextraction protocoloptimisation.Metabolomics.2012;8(2):

175–185.

23.FragaCG,ClowersBH,MooreRJ,etal.Signature-discovery approachforsamplematchingofanerve-agentprecursor usingliquidchromatography–massspectrometry,XCMS,and chemometrics.AnalChem.2010;82(10):4165–4173.

24.ErikssonL,JohanssonE,Kettaneh-WoldN,etal.Multi-and MegavariateDataAnalysis.PartI.BasicPrinciplesand

Applications.Vol45(4),2ndrevisedandenlargeded;2006:362. 25.BijlsmaS,BobeldijkI,VerheijER,etal.Large-scalehuman

metabolomicsstudies:astrategyfordata(pre-)processing andvalidation.AnalChem.2006;78(2):567–574.

26.WishartDS,JewisonT,GuoAC,etal.HMDB3.0—TheHuman MetabolomeDatabasein2013.NucleicAcidsRes.

2013;41:801–807.

27.ZhuZJ,SchultzAW,WangJ,etal.Liquidchromatography quadrupoletime-of-ﬂightmassspectrometry

characterizationofmetabolitesguidedbytheMETLIN database.NatProtoc.2013;8(3):451–460.

28.KoningWD,DamKV.Amethodforthedeterminationof changesofglycolyticmetabolitesinyeastonasubsecond timescaleusingextractionatneutralpH.AnalBiochem. 1992;204(1):118.

29.OldigesM,TakorsR.Applyingmetabolicproﬁlingtechniques forstimulus-responseexperiments:chancesandpitfalls. AdvBiochemEng/Biotechnol.2005;92:173.

30.MashegoM,RumboldK,DeMeyM,etal.Microbial metabolomics:past,presentandfuturemethodologies. BiotechnolLett.2007;29(1):1–16.

31.KimballE,RabinowitzJD.Identifyingdecomposition productsinextractsofcellularmetabolites.AnalBiochem. 2006;358(2):273–280.

32.RabinowitzJD,KimballE.Acidicacetonitrileforcellular metabolomeextractionfromEscherichiacoli.AnalChem. 2007;79(16):6167–6173.

(9)

33.SpaggiariD,GeiserL,RudazS.Couplingultra-high-pressure liquidchromatographywithmassspectrometryforin-vitro, drug-metabolismstudies.TrACTrendsAnalChem.

2010;29(1):15–27.

34.YuanL,ZhangZ,HouZ,etal.Rapidclassiﬁcationand identiﬁcationofcomplexchemicalcompositionsin

traditionalChinesemedicinebasedonUPLC-Q-TOF-MS coupledwithdataprocessingtechniquesusingtheKuDieZi injectionasanexample.AnalMethods.2015;7(12):