LC-MS/MS analysis of the central energy and carbon metabolites in biological samples following derivatization by dimethylaminophenacyl bromide

biological

samples

following

derivatization

by

dimethylaminophenacyl

bromide

Cornelius

C.W.

Willacey

_,

_Martijn

_Naaktgeboren

_,

_Edinson

_Lucumi

_Moreno

_,

Agnieszka

B.

Wegrzyn

_,

_Daan

_van

_der

_Es

_,

_Naama

_Karu

_,

_Ronan

_M.T.

_Fleming

_,

Amy

C.

Harms

_,

_Thomas

_Hankemeier

a_{AnalyticalBiosciencesandMetabolomics,DivisionofSystemsBiomedicineandPharmacology,LeidenAcademicCentreforDrugResearch,Leiden}

University,Leiden,theNetherlands

b_{DivisionofDrugDiscoveryandSafety,LeidenAcademicCentreforDrugResearch,LeidenUniversity,Leiden,theNetherlands}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received25April2019

Receivedinrevisedform27July2019 Accepted30July2019

Availableonline31July2019 Keywords:

Dimethylaminophenacylbromide Derivatization

Urine LC–MS

N-Acetylatedaminoacids TCAcycle

a

b

s

t

r

a

c

t

Recentadvancesinmetabolomicshaveenabledlargerproportionsofthehumanmetabolometobe analyzedquantitatively.However,thisusuallyrequirestheuseofseveralchromatographicmethods coupledtomassspectrometrytocoverthewiderangeofpolarity,acidity/basicityandconcentration ofmetabolites.Chemicalderivatizationallowsinprincipleawidecoverageinasinglemethod,asit affectsboththeseparationandthedetectionofmetabolites:itincreasesretention,stabilizestheanalytes andimprovesthesensitivityoftheanalytes.Themajorityofquantitativederivatizationtechniquesfor LC–MSinmetabolomicsreactwithamines,phenolsandthiols;however,thereareunfortunatelyvery fewmethodsthatcantargetcarboxylicacidsatthesametime,whichcontributetoalargeproportion ofthehumanmetabolome.Here,wedescribeaderivatizationtechniquewhichsimultaneouslylabels carboxylicacids,thiolsandaminesusingthereagentdimethylaminophenacylbromide(DmPABr).We furtherimprovethequantitationbyemployingisotope-codedderivatization(ICD),whichusesinternal standardsderivatizedwithanisotopically-labelledreagent(DmPABr-D6).Wedemonstratetheability

tomeasureandquantify64centralcarbonandenergy-relatedmetabolitesincludingaminoacids, N-acetylatedaminoacids,metabolitesfromtheTCAcycleandpyruvatemetabolism,acylcarnitinesand medium-/long-chainfattyacids.Todemonstratetheapplicabilityoftheanalyticalapproach,weanalyzed urineandSUIT-2cellsutilizinga15-minutesingleUPLC-MS/MSmethodinpositiveionizationmode. SUIT-2cellsexposedtorotenoneshoweddefinitivechangesin28outofthe64metabolites,including metabolitesfromall7classesmentioned.ByrealizingthefullpotentialofDmPABrtoderivatizeand quantifyaminesandthiolsinadditiontocarboxylicacids,weextendedthecoverageofthemetabolome, producingastrongplatformthatcanbefurtherappliedtoavarietyofbiologicalstudies.

©2019TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Metabolomics,theyoungersiblingofgenomicsandproteomics, isafast-evolvingfieldwhichhasestablisheditselfasapromising approachforunderstandingbiologicalvariationswithinarangeof matricesinhumans,animals,microbesandplants[1–8].The quan-titativeprofilingofmetabolitesinbiologicalsamplesischallenging

∗ Correspondingauthors.

E-mailaddresses:C.c.w.willacey@lacdr.leidenuniv.nl(C.C.W.Willacey),

hankemeier@lacdr.leidenuniv.nl(T.Hankemeier).

duetothevastnumberofmetabolites,variationin physicochemi-calpropertiesandthewiderangeofconcentrationsinsamples.All ofthesefactorsresultinlargedifferencesintherecovery, sensitiv-ityandmatrixinterferencesofthesemetaboliteswhenanalyzed byvariousmethods.Nevertheless,recentadvancesinmass spec-trometry havegivenscientists theabilitytofurtherunderstand thehumanmetabolomeandfocusmorecloselyonselected path-wayanalysis.Whenwestudymetabolicpathways,weexperience thecomplexityastheycancompriseof manychemical conver-sionsandareintertwined,makingatargetedassaywithcoverage ofover50relevantmetaboliteshighlybeneficialforresearchersin metabolism.

https://doi.org/10.1016/j.chroma.2019.460413

Massspectrometry(MS)hastheabilitytoidentifyand quan-tifythemetabolomewithcurrentmethodsreachingsensitivities downtopicomolarconcentrations,evenwithoutanyprior sep-aration[9]. However,in themajorityof cases, chromatography priortoMSisusedtobetteraddressthechallengesintroduced byionsuppression,separationofisomersandin-source fragmen-tation. The three most common separation techniques, LC, GC andCE,haveprovidedrobustmethodologiestobettercoverthe humanmetabolome.Eachofthesetechniqueshasbeenapplied tonumerous typesof metabolites, and each technique has tai-loredadvantagesforspecifictypes ofmetabolites.For example, UPLC-MS(RP&HILIC)providescoverageforalargeproportionof themetabolomewiththeadvantagesofhigh-throughput, sensitiv-ity,reliabilityandrobustness[10].Still,inLC–MS,metabolitescan sufferfromlimitedsensitivity,orpoorseparationofparticularly polarmetabolites.Quantificationofmetaboliteswithelectrospray ionization(ESI)-MScansufferfromionsuppression[11].This inter-ferencecanbecorrectedforbyusingcoelutingisotopically-labelled internalstandards,whichareoflimitedavailabilityandexcessive costs.

Methodshavebeendevelopedtocombattheseproblemsusing advancedseparationtechniquesandalsochemicalderivatization, whichisthefocusofthisarticle.Chemicalderivatizationcanbe usedtoincreasetheseparationresolution,sensitivityorto stabi-lizethemetabolites,resultinginanincreasedmetaboliccoverage ofMS-basedmetabolomicsmethods.Forinstance,benzoyl chlo-rideis usedtoderivatize catecholaminesand theirmetabolites topreventoxidation and increase sensitivity in LC–MS [12]. In a recent review, Higashi and Ogawa [13] summarise the cur-rent techniques that are used for derivatization and conclude thatisotope-codedderivatization(ICD)hastheabilitytoenhance quantification in LC–MS(/MS). ICD is the process of labelling metabolitesin a firstsample withan unlabelled derivatization reagentandthenusinganisotopicallylabelledreagentto deriva-tizethesame metabolitestandards ina neat solution,i.e.pure solvent.Thismixture,whenaddedtothesample,canactasthe correspondinginternalstandard (IS)for allanalytesof interest. Thebenefitofthistechniqueistheabilitytointroducean isotopi-callylabelledequivalentforallmetabolitesregardlessofchemical structure complexity, which corrects for eventualion suppres-sion.ApproachessuchasICDareimportantduringderivatization workflowstocompensateforpossiblematrixeffectsasthenative matrix is altered due to derivatization. However, havingan IS for each metabolite provides a tool toadjust for matrix inter-ferencesindependent of thestarting matrices. ICD can provide acost-effectivealternativewhenstableisotopeISarenot avail-ablewhilestillenablingimprovedtruenessandprecision.Inthis way,thederivatizationreactionmethodisexploitedinan addi-tionalmannernexttomodifyingtheseparationandionizationof metabolites.

Severalstudieshaveutilizedarangeofreagents,sometaking advantageoftheICDstrategytoimprovethequantitative perfor-mance[12,14,15].Typicalexamplesarebenzoylchloride[12,14] anddansylchloride[15]whichbothlabelamines,thiols,phenols andsomealcohols.Anotherreagent,dimethylaminophenacyl bro-mide(DmPABr),hasbeenappliedpreviouslytolabelcarboxylic acidgroups[16].Therewereinconsistentreportsaboutthe reac-tivityofDmPABr.GuoandLi[16]reportedthatDmPABrreactsonly withcarboxylicacids(i.e.,notaminesandthiols),andinafollow-up studyPengandLiacknowledged thatit reactsalsowith nucle-ophilesatcertainreactionconditions[17].However,toconform withtheaimsoftheirmethod,liquid-liquidextraction(LLE)was appliedtoreducetheinterferencefromaminoacidsand deriva-tives,byexcludingthemaltogether.Theneedforareliablemethod thatcombineslabellingoftheamine,thiolandcarboxylicacid func-tionalgroupshasbeenhighlightedbypreviouspapersthathave

Table1

Listoftheabbreviationsforthemetabolitesanalyzedinthismethod.

Metabolite Abbreviation Metabolite Abbreviation

Alanine Ala N-acetylmethionine NA-Met

Arginine Arg N-acetylphenylalanine NA-Phe

Asparagine Asn N-acetylproline NA-Pro

Asparticacid Asp N-acetylserine NA-Ser

Cysteine Cys N-acetylthreonine NA-Thr

Glutamine Gln N-acetyltryptophan NA-Trp

Glutamicacid Glu N-acetyltyrosine NA-Tyr

Glycine Gly N-acetylvaline NA-Val

Histidine His Alpha-Ketoglutaricacid AKG

Isoleucine Ile Citricacids CITS

Leucine Leu Fumaricacid FUM

Lysine Lys Lacticacid LAC

Methionine Met Malicacid MAL

Phenylalanine Phe Oxaloaceticacid OXA

Proline Pro Pyruvicacid PYR

Serine Ser Succinicacid SUCC

Threonine Thr Acetylcarnitine AC

Tryptophan Trp Decanoylcarnitine DC

Tyrosine Tyr Hexanoylcarnitine HC

Valine Val Lauroylcarnitine LC

N-acetylalanine NA-Ala Myristoylcarnitine MC

N-acetylarginine NA-Arg Octanoylcarnitine OC

N-acetylasparagine NA-Asn Palmitoylcarnitine PC N-acetylasparticacid NA-Asp Propionylcarnitine PPC

N-acetylcysteine NA-Cys Stearoylcarnitine SC

N-acetylglutamine NA-Gln Arachidonicacid AA

N-acetylglutamicacid NA-Glu Capricacid DCA

N-acetylglycine NA-Gly Caprylicacid OCA

N-acetylhistidine NA-His Dodecanoicacid DDA

N-acetylisoleucine NA-Ile Oleicacid OLA

N-acetylleucine NA-Leu Undecanoicacid UDA

N-acetyllysine NA-Lys Creatinine CR

requiredtwoseparatederivatizationmethods(DmPABranddansyl

chloride)toachievethesamecoverage[18].

Inthecurrentpaper,weexpandtheutilizationofthereagent DmPABrtosimultaneouslyderivatizemetaboliteswithcarboxylic acid, amineand thiolfunctional groups. We didnotapply LLE, andanalyzed aminoacids,N-acetylatedaminoacids, carnitines, andorganicacidsusingLC–MSinpositive ionizationmode. We haveexaminedandoptimisedthereactionconditionstoreliably andrepeatablyderivatizearangeofmetabolitesandanalyzethem in a single, highly sensitive quantitative method. The reaction mechanismisidenticaltothatofthereagentphenacylbromide withprimaryamines[19],secondaryamines[20,21],thiols[22,23] andcarboxylicacid-containingmetabolites(derivatization exam-pleshowninFig.1).First,we madeadaptationstothemethod publishedbyGuoandLi[16],PengandLi[17],Stanislaus,GuoandLi [24]toimprovethemetabolitecoveragetoincludeawiderangeof centralcarbonandenergy-relatedmetabolites.Then,wedeveloped atargetedquantitativeUPLC-MS/MSmethodtoallowforthe sen-sitiveanalysisofthesemetabolitesinasingle10-minuteanalysis. Thefinalappliedmethodwassuccessfullyvalidatedforlinearity, precision,limitsofdetection(LOD)andquantification(LOQ).By applyingthis methodtohumanurine and invitroexperiments usinghumanpancreaticcancercells(SUIT-2),wecouldconfirmthe broadapplicabilityofthismethodologyandbiologicalrelevancefor thescientificcommunity.

2. Materialsandmethods

2.1. Chemicals

Fig.1.ReactionschemeofDmPABrwithcysteineshowingthelocationofderivatizationonthethiol(green)andcarboxylicacid(blue),andtwiceontheprimaryamine (red).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle).

usingaMerckMilli-poreA10purificationsystem(Raleigh,USA). Stocksolutionsof5mg/mLAla,Arg,Asn,Asp,Cys,Gln,Glu,Gly, His,Ile,Leu,Lys,Met,Phe,Pro,Ser,Thr,Trp,Tyr,Val;10mMNA-Ala, NA-Arg,NA-Asn,NA-Asp,NA-Cys,NA-Gln,NA-Glu,NA-Gly,NA-His, NA-Leu,NA-Lys,NA-Met,NA-Phe,NA-Pro,NA-Ser,NA-Thr,NA-Trp, NA-Tyr,NA-Val;1mg/mLAKG,CIT,FUM,ICIT,LAC,MAL,OXA,PYR, SUCCweremadein1:1DMSO/DMFandstoredat−80◦_C.Stock

solutionsof2mg/mLCR;1mg/mLAA,AC,BTA,DEA,DDA,EA,EIA, FOR,OLA,OCA,PA,PPA,SA,UDAin100%ACN(v/v)andstoredat −80◦_C.

2.2. Derivatizationreagent

TheDmPABrreagentwaspurchasedfromBioConnectBV (Huis-sen, The Netherlands) and the internal standard DmPABr was synthesisedfollowingthepublishedprotocolbyGuoandLi[16] usingdimethylsulphate-D6insteadofdimethylsulphate-13C2.The

structureofthereagentwasconfirmedusingnuclearmagnetic res-onance(NMR).Also,withreferencetothepaperfromGuoandLi [16],itisnotedthatthestabilityofthemetabolitesafterreaction withDmPABrlastsforupto6monthsinasolution,anddoesnot alterquantitativeresults[24].TheDmPABrreagentwasstoredin ACNat−80◦_{Ctopreventthenucleophilicsubstitutionreaction.}

2.3. Methodvalidationandbiologicalapplication 2.3.1. Methodoptimizationandvalidation

Thefollowing performance parameters wereassessed onall 64metabolitesintriplicate.Methodoptimizationstartedwiththe selectionofanappropriatealkalinesolutionatarangeof concen-trations,comparingtriethylamine(TEAat0,50,100,150,250,300, 500and750mM)andtriethanolamine(TEOA,at0,200,400,650, 700,750,800and1000mM).Thereactiontimewasassessedfor theselectedtimepoints0,5,15,30,45,60,90,180and240min usingTEOA(750mM)incubatedforonehourat65◦C.Duetothe abilityofwatertoreactwiththereagentactingasanucleophile thereactionwasassessedinthepresenceofwaterat0%,20%,40%, 60%,80%and100%.Thefinaloptimizedmethodused750mM solu-tionofTEOAforderivatizationat65◦Cforonehourinashaking incubator.Themethodwascharacterisedbyamatrix-free8-point calibrationline, andbydeterminingthecarry-overbyasolvent injectionblankafterinjectingthehighestcalibrationlevel (cali-brationpoint7:SupplementarymaterialTableS6).Thecalibration experimentwasreplicated(n=5).Matrixeffect(ME)isdefinedas “thedirectorindirectalterationorinterferenceinresponsedueto thepresenceofunintendedanalytes(foranalysis)orother inter-feringsubstancesinthesample”[25].TheMEwascalculatedasthe areaoftheinternalstandardsintheneatsolutionagainstthearea

oftheinternalstandardinthepresenceofthematrix.Themethod wasalsoassessedfor linearityofthecalibrationline(n=5)and LOD/LOQ.TheLODandLOQwerecalculatedusingthefollowing equationsaccordingtotheICHQ2R1guidelines–␴beingstandard deviationofthesignalintheblankinjection:

LOD= (3.3∗␴)/slope LOQ = (10∗␴)/slope

ME% ₌(Internalstandardinneatsolution_/ internalstandardinmatrix)∗100 2.3.2. Urinevalidationsamples

Urinefrom10healthyvolunteers(aged20–30)wascollected andpooledand usedfor methodoptimizationandvalidation.A volumeof10␮LofurinewastransferredtoanEppendorfsafe-lock vial(0.5mL).Theurinewasdried inaLabconcoSpeedVac(MO, UnitedStates).The driedcontentwasreconstitutedin 10␮Lof DMSO/DMFtodissolvetheremainingcontent.Then,10␮Lof tri-ethanolamine(750mM)wasaddedtothevial,followedby10␮L ofDmPABr (82mM).Thesealed Eppendorfvialwasplacedinto ashakingincubatorfor60minat65◦Ctocompletethe derivati-zation.Atotal of10␮Lofformicacid(30mg/mL)wasaddedto thevialtoquenchthereactionwithanadditional30mininthe shakingincubator.Then,5␮LofDmPABr-D6-labelledmetabolites

werethenadded(concentrationsinSupplementarymaterialTable S6).Beforevortexing,45␮LofACNwasalsoaddedtothevial.The contentwasthentransferredtoanHPLCvialforanalysis.The true-nessandprecisionofthemethodwasgeneratedbyusingapooled sampleofurinecollectedfromhealthyurinedonors.Sampleswere analyzedinrepeatedexperimentson3separatedaysinreplicates eachday(n=5).Usingthisdata,RSDcalculationswereperformed todemonstratethelackofvariationinthederivatizationconditions onseparatedays.

2.4. SUIT-2oxidativestressanalysisandvalidation

Human pancreatic cancer cells (SUIT-2) were cultured and placedintoa24-wellplate,eachcontaining1*106_{cellsin0.4mLof}

13,000rpmtoproduceaproteinprecipitation;thesupernatantwas transferredtoanEppendorfsafe-lockvial(1.5mL)without disturb-ingthepellet.Avolumeequivalentto2.5*105_{cellsupernatantwas}

takentototaldrynessinaspeedvacuumconcentrator.The follow-ingwereaddedtothevialandvortexedbetweenadditions:10␮L ofDMSO/DMF(to firstdissolvethedriedcontent), 10␮Lof tri-ethanolamine(750mM)And10␮LofDmPABr(82mM).Thesealed Eppendorfvialwasplacedintoashakingincubatorfor60minat 65◦Ctocompletethederivatization.Avolumeof10_␮Lofformic acid(30mg/mL)wasaddedtothevialtoquenchthereactionwith anadditional30minintheincubation.Finally,5␮LofDmPABr-D6

-labelledmetabolitesweredilutedin45␮LofACNandaddedtothe vial.ThecontentwasthentransferredtoanHPLCvialforanalysis.

2.5. LC–MS/MSanalysis

Samples were analyzed by LC–MS using a Waters Acquity UPLCClassII(Milford, USA)coupledtoan ABSciexQTrap6500 series(Framingham,USA).Thesampleswererunusingscheduled multi-reactionmonitoring(MRM)inpositivemodewithselected time windows. An injection of 1␮L was made per sample to minimisedetectorsaturationandmaintaindesirablepeakshape. Theanalytical columnusedwasaWatersAccQ-tagC18column (2.1mm×100mm, 1.8␮m, 180Å), maintained at 60◦C. Mobile phasesolventAwas0.1%v/vformicacidand10mMammonium formateinwaterandmobilephasesolventBwas100% acetoni-trile.Usingtheflowrateof700␮L/min,thegradientprofileisas follows:initial,0.2%B;1.5min,20%B;4min,50%B;6min,90% B;10min,99.8%B;13min,99.8%B;13.1min,0.2%Band15min, 0.2%B.Thelast6minallowforcolumnwashingandequilibration priortothenextinjection.Thefollowingparameterswereusedfor theABSciexQTrap6500analysis(MRMtransitionsshowninTable S2oftheSupplementarymaterials);electrosprayionizationwas usedinpositivemodeat4.5kV.Thegastemperaturewas600◦C. AutomatedpeakintegrationwasperformedusingABSciex Multi-QuantWorkstationQuantitativeAnalysisforQTrap;allpeakswere visuallyinspectedtoensureadequateintegration.

3. Results&discussion

3.1. Novelderivatizationapproach

DmPABrderivatizationhasbeenasuccessfulmethodto sup-portuntargetedandtargetedmetabolomicsplatformsusingICD forcarboxylicacid-containingmetabolites.Itwaspreviously high-lightedbyPengandLi[17]thatDmPABralsoreactswiththeamine groupofasparagineandlabelsittwice(onceontheacidandonce ontheaminegroup),howeverthedoublelabelledmetabolitewas reportedtobetheminorpeakcomparedtothesinglederivatized form.Asdemonstratedinthispaper,DmPABrcanreactwithan aminegroup(onceortwice)viaanucleophilicreactionina quanti-tativemanner,whichisusefulforLC–MSanalysis.Incomparisonto anothercommonreagentinLC–MS/MS,benzoylchloride[12,14], thereagentDmPABroffersamoreversatileanduniversalsolution forderivatization.Thisreaction,however,isslowerandforms a morestablebond,asDmPABrhastheabilitytoreactnexttothe aminegroup.It alsoreactswithcarboxylicacidswithout form-inganunstableanhydrideasthebromineisattachedtoamethyl grouprathertotheacylgroup.Therefore,westudiedtheability ofDmPABrtoreactwithmultiplefunctionalgroupssuchasthe amine,carboxyandthiolgroupsandtouseitforthemetabolomics analysisofurineandcellsamples.

3.2. Selectionofmetabolitesandbiologicalrelevance

UtilizingthefullcapabilityofDmPABrallowsustoextendfrom onlyderivatizingcarboxylicacidstoalsotargetingaminesand thi-olswhichbroadenstheapplicabilityofthemethodsignificantly. To demonstrate this, we choseto measure central carbon and energy-relatedmetabolitesrelatedtomitochondrialdysfunction asitrequirestheanalysisofabroadrangeofchemicallydiverse metabolites.Thekey metabolitesin aerobicrespirationthatare imperativeformammaliansurvival,suchas␣-ketoglutaricacid, citrates,succinicacid,fumaricacid,malicacidandoxaloaceticacid, arefrequentlyusedreferencestodeterminechangesin mitochon-drialfunctionandcellularhealthinmetabolomicsstudiesinurine, plasmaandinvitromodels[26].N-Acetylationhasbeenknownto increaseduringmitochondrialdysfunctionduetotheelevationof acetyl-CoA.Therefore,N-acetylatedaminoacidsreflect mitochon-drialdysfunctionandenergymetabolism,andinadditionarehighly relevantinurineastheyremoveexcessaminoacidsfromthebody [27,28].Shiftingoftheenergybalancefromaerobicrespirationto anaerobicrespirationcanalsobenotedbymeasurementof pyru-vicacidandlacticacidwhichwillbothdrasticallyincrease[27]. Anadditionaltargetgroup,acylcarnitines,wasalsoselected,asit reflectsenergyprocesses,particularlyfollowingbeta-oxidationand whenfattyacidsaretransportedintothemitochondria[27,29].The lasttargetgroup,fattyacids,isincludedtorepresentanalternative energysourcebythemitochondriawhensugarsareinaccessibleor depleted.Threecommonconditionsthatareoftenassociatedwith mitochondrial dysfunction are Parkinson’s disease [30], Leigh’s syndrome[27]anddiabetes[31];allofwhichholdextensive inter-estwithinthescientificcommunity.

3.3. Optimizationofreactionconditions

Weinvestigatedandoptimisedthemethodtoanalyzeamines, thiolsandcarboxylicacidstoinformoncentralcarbonandenergy metabolism.Wehavefoundthreekeyfactorsthataffectthe deriva-tizationofthefunctionalgroupsmentioned above:alkalinityof reactionsolution,reactiontime,andthepresenceofwater dur-ingthereaction.Fig.2Ademonstratesthatafter60mintherelative peakareadidnotincreasesignificantlyanymore,withhigh per-formanceparameters(indicatedinTable2).Thisappliedtoallof thetargetedfunctionalgroups:Ala(1◦amine&carboxylicacids), NA-Asp(carboxylic acids),NA-Cys(thiol &carboxylicacid),PYR (␣-keto acid) andAC (carboxylic acid).Thisderivatization time wasconsidered acceptablein terms of metaboliccoverage. We have utilized similarinertbasecatalystsas inprevious articles published for DmPABr that target the carboxylic acid function group,whichutilizedeither750mMtriethanolamine(TEOA)[16] or200mMtriethylamine(TEA)[24].Variationsinresponseusing thesebases aredepictedin Fig.2BandC, leading tothe selec-tionoftheappropriateconditionsforderivatizationofamineand thiol groups. Additional experiments indicated that 750mM of TEOAwastheoptimumconditionforconsistentderivatizationof metabolitesinurineandcells(datanotshown).TEOA(750mM) alsoprovidedthemostconsistentderivatizationindicatedby iden-ticalvalues overtheconcentrationrange of650–800mM.TEOA waspreferredbecauseaccordingtoliterature itcausesless ion suppressioninmassspectrometers,thanTEAattheconcentration tested[32].

OftheoptionsforsynthesizedisotopicallylabelledDmPABr,D6

wasusedin placeof 13_C

2 ontheamineresidue of DmPABr,as

utilizedforhigh-resolutionMSbyGuoandLi[16].Withthis,we havebeenabletointroduce amassdifferenceof6Da, whichis preferableforlow-resolutionMScomparedtotheprevious addi-tionof2Da,andlesscostly.Themassdifferenceof2Dawiththe internalstandardDmPABr-13_C

Fig.2.DmPABrreactionoptimizationshownfor5metabolites(Ala–Blue;NA-Asp–Red;NA-Cys–Green;PYR–Purple;AC–Orange;n=3percondition)representing themajorclassesselectedinthemethod.(A)theeffectofreactiontimewith750mMtriethanolamine;(B)useofTEAasbase;(C)useofTEOAasbase;and(D)thereaction efficiencyinthepresenceofwaterwith750mMtriethanolamine.Thedataispresentedaspeakareanormalizedtothehighestpeakarea.(Forinterpretationofthereferences tocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle).

Table2

Informationrelatingtothecarry-over,limitofdetection(LOD),limitofquantification(LOQ)andlinearityofan8-pointcalibrationlineinaqueoussolution.RSDwascalculated usingcalibrationpoint4(concentrationshowninSupplementaryTableS2).

Analyte LODs(nM) LOQs(nM) Fit(R2_{) RSD(%) Carryover(%) Analyte LODs(nM) LOQs(nM) Fit(R}2_{) RSD(%) Carryover(%)}

Table3

Methodperformanceinurineofhealthymenaged18–30tocalculateconcentration,intradayandinterdayprecisioncalculatedas%RSD.ND=NotDetected,N/A=Not Applicable. Analyte Urine Concentration (␮mol/mmol creatinine) Intraday precision(%) Interday precision(%) Matrixeffect (%) Analyte Urine Concentration (␮mol/mmol creatinine) Intra-Day precision(%) Inter-Day precision(%) Matrixeffect (%) Ala 3.81 4.7 5.1 92.3 NA-Met 0.03 2.3 4 37.7 Arg 0.32 5.2 9.5 68.45 NA-Phe 0.05 3.1 4.2 51.7 Asn 1.11 4.9 7 74 NA-Pro 1.31 29.3 23.5 31 Asp 0.04 18.9 16.5 95 NA-Ser 0.25 2.4 4.1 66.5 Cys 11.5 12.9 21 91.2 NA-Thr 1.7 3.2 3.6 53.9 Gln 3.67 22.8 18.5 83.5 NA-Trp 0.2 3.6 3.7 45.5 Glu 0.12 5.5 6.4 81 NA-Tyr 0.047 5 9.8 36.8 Gly 136 1.4 4.2 67 NA-Val 2.07 3.7 13.2 16.6 His 76.3 5.2 5.7 42.7 AKG 0.5 10.6 12.8 71.3 Ile 0.23 6.7 7.7 80.3 CITS 7.47 13.1 11.5 73.9 Leu 0.52 6.4 7 85 FUM 0.13 6.7 7.7 144.4 Lys 1.22 23.1 28 42.2 LAC 0.89 3.3 8.9 43.5 Met 0.11 7.7 11.9 78.2 MAL 0.13 7.1 7.4 42.8 Phe 1.35 4.1 9.2 80.2 OXA 0.3 3.9 6.6 57.2 Pro 0.11 6.4 6.7 129.5 PYR 0.17 11.5 10.2 50.3 Ser 4.43 7.3 15.5 72.1 SUCC 5.31 2.7 3.4 39.4 Thr 1.21 6.3 16.1 70.2 AC 1.93 4.2 5 83.38 Trp 1.27 7.7 7.6 77.3 DC 0.001 9.5 7.8 93.1

Tyr 1.33 5.4 6.8 85 HC ND N/A N/A 77.9

Val 0.62 5 7.3 78.9 LC ND N/A N/A 83.6

NA-Ala 0.54 3.6 6.7 43.4 MC ND N/A N/A 91.9

NA-Arg 0.67 1.3 1.5 67.5 OC 0.007 5.3 5 58.2

NA-Asn 2.01 7.7 7.4 54.8 PC ND N/A N/A 96.3

NA-Asp 2.29 2.2 5.4 55.1 PPC 0.07 2.2 2.8 88.7

NA-Cys 1.45 4.1 10.4 27.4 SC ND N/A N/A 95.4

NA-Gln 1.26 5.5 6.4 34.7 AA 0.02 9.4 13.5 103.6

NA-Glu 0.67 3.6 3 62.3 OCA ND N/A N/A 66.7

NA-Gly 0.13 3.6 7.1 33.3 DCA ND N/A N/A 90.7

NA-His 1.28 3 4.8 63.9 DDA 0.008 10.9 24.1 83.2

NA-Ile 0.14 11.5 10.3 46.7 OLA 0.006 11.2 16.3 95.8

NA-Leu 0.13 4.9 6.4 47.6 UDA 0.004 21.4 39.1 98.3

NA-Lys 0.3 24 22 60 CRa _{N/A 4.5 7.4 38.1}

a_{Creatininewasusedfornormalization.} Table4

MethodperformanceinSUIT-2celltocalculateconcentrationpermgandintra-dayvariability.

Analyte SUIT-2Conc. (fmol/mg)

Intra-Day precision(%)

Matrixeffect (%)

Analyte SUIT-2Conc.

Fig.3.DemonstrationofMSanalysisofpyruvicacid:(A)inconventionalnegativeionizationmode,holdinganegativechargeontheoxygen(greencircle);(B)following DmPAlabelling,producingahighlyionisablegroup(tertiaryamine–redcircle)andahigherretentiongroup(bluecircle).(Forinterpretationofthereferencestocolourin thisfigurelegend,thereaderisreferredtothewebversionofthisarticle).

Fig.4.Thecommonfragmentformationof180.0Da&134.1Da:theproductionofthederivatizationlabelandmetabolite-specificproductions.

thetriple-quadrupoleMSwiththemetaboliteslabelledonce,such

aslong-chainfattyacidsandN-acetylatedaminoacids.However,

thiswasnotasdetrimentaltothemetaboliteslabelledmorethan

once(suchasaminoacids)asamassdifferencegainof18Dawas

observedformetabolitessuchasAla(labelledthrice)whenusing

DmPABr-D6.

Otheradaptations of thederivatization procedurewerealso

evaluated, including the total elimination of theaqueous

con-tent prior to derivatization with DmPABr to improve reaction

efficiencyanddecreasereactionvariability.Thiswasexpectedto

beneededtocreatea quantitativemethod,unliketheprevious

published methodthat focused onidentification. This

variabil-ity and poor labelling efficiency may arise from the ability of

water toact as a nucleophile under basicconditions

(deproto-nation). It haspreviously beennoted that aslittle as 5%water

contentduringthederivatizationreactionhastheabilitytoreduce

thelabellingefficiencybyhydrolysingDmPABr [23]. Itwasalso

Fig.5. LC–MS/MSanalysisof64metabolitesafterDmPABrderivatizationinSUIT-2cells.DmPAlabellingpatternisalsoincluded(*=labelledonce;†=labelledtwice;#= labelledthrice).(Forinterpretationofthereferencestocolourinthisfigure,thereaderisreferredtothewebversionofthisarticle).

100%water,thereactionisseverelycompromised.Therefore,we chose to conduct the reaction without the presence of water to avoid the complications due to possible hydrolysis of the reagent.

3.4. TargetedLC–MS/MSmethod

Theaimsofthechromatographicmethodweretocombine high-throughputanalysiswithsensitivemeasurementofawiderange ofchemicalclasses.AfterderivatizationwithDmPABr,polar com-pounds which arehardly retainedin RP couldberetained and separated,henceeliminatingtheneedforHILICseparation. More-over,derivatizationwithDmPABrallowssensitiveanduniversal analysisinpositiveionizationmode,insteadofintwoionization modes.ByusingDmPABr,weintroducethetertiaryaminegroup (Fig.3thatimprovesionizationhenceenhancessignals.Further improvementinintensityofmeasuredmetabolitescanbegained bycarefulselectionofMRMs.InFig.4AandBweillustratethatthe metabolitesthatarelabelledonthecarboxylicacidshowcommon andprominentfragmentsof180.0Daor134.1Da.Theseproduct fragmentsare idealwhenmeasuringmetabolitessuchas those involvedintheTCAcycleastheylacknitrogenandaredifficultto analyzeinpositiveionizationmodewithoutlabelling.For metabo-liteswhicharelabelledmultipletimes,suchasaminoacids,the fragments180.0Daand134.1Daareusuallypresent(Fig.4B)but arenotselectedas theirsignalis lower.Instead,a highermass productiongivingabettersignalisoftenseen.Thisresultsfrom derivatizationtwiceontheaminegroup,andonceontheacid moi-ety,yieldingmoresensitivefragmentslike319.2Daand366.2Da observedforarginine&alanineinFig.4CandD,respectively. Hav-inga common fragmentationpattern reduces thespecificity of metabolitespeciesbutwithadequatechromatography,thisissue canbenegated.Additionally,aqualifiertransitioncanalsobeset whichwillprovideauniquefragmentationpatterntoidentifythe specificmetabolitebutthiswillprovidealowersensitivitydueto moreMS/MSevents.Forthecompletemethod,thelabelling

pat-ternandthechosenMRMtransitionsaredetailedinSupplementary TableS2.

To demonstrate the applicability of the method on biologi-calsamples,SUIT-2cellextractsweresubjectedtoderivatization andanalysis,resultinginwiderepresentationofvariouschemical classes(Fig.5).Thefiguredemonstratesthatowingtostrong reten-tionofpolaranalytes(PYR,Gly)all64metabolitesaredetectedin oneruninpositiveionizationmodeonlywithin8.4min(latest elu-tion,ofarachidonicacid,AA).Thederivatizationleadstounique retentionprofile,suchasthecloseelutionbetweenundecanoic acid(derivatizedonce)andleucine(derivatizedthreetimes),yet withbaselineresolutionbetweentheisomersIleandLeu.Another criticalpairofisomers,CITandICITpresentacommonchallenge inchromatography,andarenotbaselineresolvedhere(seeFig.5), thereforearereportedastotalcitrates(CITS).Incontrast,good sep-arationwasobservedforN-acetylatedaminoacids,manyofwhich eluteearly.Thefirstpeaktoelutewascreatinine(Cr)whichisoften usedtonormalizeandreportmetabolitesinurine[34].DmPABr cansuccessfullyderivatizecreatinine,unlikethereagentsutilized incommercially-availablekits,thatquantifynon-derivatized crea-tinine(BiocratesAbsoluteIDQ®p180Kit;WatersAccQ-TagTM_).

3.5. Methodperformanceinneatsolutions

Themethodsperformanceincorporatesthederivatization effi-ciencyandtheinstrumentalresponse.Themethodwasvalidated for64metabolitesthatweredeemedtobebiologicallyrelevantto assessthecentralenergyandcarbonmetabolism.UsingtheICD strategy,eachmetabolitehaditscorrespondingDmPABr-D6

inter-nalstandardtocorrectforionsuppression.Thisresultedinlinear calibrationlinesforallmetabolitesinneatsolutions(Table2).All metabolites,includingtheaminoacidswhicharederivatizedby reacting2–5timeswithDmPABrshowedasatisfactorylinear cali-bration(R2_{>0.99)exceptforCys(R}2_{=0.98)andPro(R}2_=0.98).The

Fig.6. VolcanoplotofSUIT-2cellsexposedto100nMrotenonefor24hvscontrol.Allofthemetabolitesinvolvedinthemethodshowbiologicalchangesacrossallclasses oncetreatedwithrotenone(aminoacids–lightpurple;carnitines–blue;glycolysis–red;long-chainfattyacids–darkgreen;medium-chainfattyacid–lightgreen; N-acetylatedaminoacids–darkpurple;andTCAmetabolites–orange).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtotheweb versionofthisarticle).

N-Acetylatedaminoacidsalsoshowedgoodanalytical perfor-mancesimilartotheirfreeaminoacidcounterparts.Table2also showsthatthecarry-overofthemethodwasnegligible(<0.05%). Lookingatlimitsofdetection(whichareaffectedbythe deriva-tizationprocessitself),N-acetylatedaminoacidshaveaverylow LOD,asrecordedforNA-Asp(4nM),NA-Cys(34nM)andNA-Phe (0.3nM),whicharesufficientfortheiranalysisinurineandcells. N-Acetylatedaminoacidsareusedasthetransportmechanismto excreteexcessaminoacids(particularlyintheurine)thatoccurin relativelylowconcentrationswhencomparedtofreeaminoacidsin urine[28,35].Wehavecircumventedtheissuesoflimiteddynamic rangeofthedetectorinordertoallowgoodquantitationofawide concentrationrangeofmetabolites.AsshowninTable2,theLOD ofGly,His,Ser,CITSandLAC,werehighercomparedtotheother metabolitesinthismethod.Thisisduetointentionalchoiceofless sensitiveMRMchannel,toreducethesignalandpreventdetector saturation,counteractingthehighphysiologicalconcentrationsin urineorcells.Anotherinterventiontopreventdetectorsaturation tookplace,namelynon-optimalionizationsprayvoltage through-out(4.5kVvs.theoptimal5.5kV).Theapplicationofthismethod tovariousmatricescouldbenefitfromtailoringtheMSparameters aswellassamplehandlingtoimprovetheLOD.

3.6. Methodperformanceinurineandcells

We applied the quantitative DmPABr method to urine and SUIT-2cells(Tables3and4).Table3showstheendogenous concen-trationofthemetabolitesmeasuredinurinefromhealthymales, afternormalizationtocreatinine(measuredinthesamemethod). Atotalof57compoundsweredetectedandquantifiedinurine. Thecompoundsthat werenotdetectedincludesomecarnitines andmedium chainfattyacids astheydonot occuroroccurin lowconcentrations inhealthyurine. Allof theamino acidsand N-acetylatedaminoacidsthathavebeenstudiedfallwithinthe expectedconcentrationscuratedinHMDB[3].Urinewasassessed

forintra-dayandinter-dayvariability.AminoacidssuchasAla,Ile, Trphadverylowintra-andinter-dayvariability(allbelow10%) andN-acetylatedaminoacidsincludingNA-Asp,whichiscrucial forneurologicalstudies,hadanintra-dayandinter-day variabil-ityof2.2%and5.4%,respectively.Creatininehadanintra-dayand inter-dayvariabilityof4.5%and7.4%,respectively,whichprovides consistentnormalizationfactor,ifdesired.Overall,theaminoacids hadahigherderivatizationvariabilitythanotherclasses,whichis probablyduetothesecondstepofderivatizationonthe2◦amine requiringmoreenergy,comparedtothesinglereactionwiththe carboxylicacidgroup.

The application of the method to cells was conducted by measuringuntreatedSUIT-2cells.Thecellswereassessedfor intra-dayvariabilityandnotinter-dayduetopractical considerations (Table4).All64metabolitesweredetectedfromtheintra-cellular environmentincelllysate.Thisprovidedanexcellentreadouton theenergystateofthecellsusingmetabolitesinvolvedintheTCA cycle(i.e.,CITS,FUM&SUCC)andglycolysis(PYR&LAC).

Matrixeffectwascalculatedforbothurineandcellsamples.The matrixeffectwassignificantduringtheearlyelutingpeakssuchas theN-acetylatedaminoacidsinurine.However,thematrix inter-ferenceswerenotashighduringtheanalysisofcells.Thepresence ofamatrixeffectshowstheimportanceofusingtheICDtechnique toprovideaninternalstandardforallmetabolites.

showedsignificantchangesincludingmetabolitesfromallofthe7 classes(additionaldatashowninSupplementaryTableS3).Thetop tenmostdistinguishingmetaboliteswere:Asn(p=0.0001);AKG (p=0.0001);Pro(p=0.0004);CITS(p=0.0004);PYR(p=0.0004); OXA(p=0.001);FUM(p=0.002);AC(p=0.002);SUCC(p=0.002) andMC(p=0.003)asshowninFig.6.Thetwometaboliteswiththe largestfoldchange,CITSandAKG,coincidewiththeshutdownof theTCAcyclebyrotenoneinhibitionofcomplexIoftheelectron transportchain[36].ThesamereductionwasseeninSUCC and FUMbuttoalesserextent.Interestingly,wealsoidentifiedchanges intheN-acetylatedaminoacidssuchasNA-Glu(p=0.01),NA-Ala (p=0.003).N-Acetylationofaminoacidshasbeendocumentedin mitochondrialdysfunctionbuthasnotbeenextensivelystudied duetodifficultywithanalysis[27,37,38].Theresultsobtainedwith ournovelmethoddemonstrateitspotentialinstudyingtherole ofcentralcarbonand energymetabolismsuchasmitochondrial dysfunctionandParkinson’sdisease[39].

4. Conclusions

The presented work expands the metabolite coverage of DmPABr by implementation of changes to the reaction condi-tions. Actually, the derivatization of several functional groups includingcarboxylicacids,primaryamines,secondaryaminesand thiol groups was achieved in a consistent and robust way for thefirst timeusingDmPABr.Thisvastlyimprovesthecoverage of the methodallowing for a higher proportion of thehuman metabolome tobe targeted. We have demonstrated that using DmPABrderivatizationtoitsfullabilityallowsustocreatea sin-gleRPLC-MS/MSanalysiswithin10minacquisitiontimeusingonly positiveionizationmode.Sinceweusedatargetedmetabolomics methodemployhinginternal standards which werederivatized withstableisotope-labelledreagent,wecanreporteach metabo-litereliablywithitsabsoluteconcentration.Thegreatversatilityof thisapproachwasdemonstratedbyquantificationofurine metabo-lites(normalizedtoDmPABr-derivatizedcreatinine).Applyingthe methodtoSUIT-2cells exposedtorotenoneshowedsignificant changesinalmost50%ofthemetabolitescoveredinthismethod, includingcommon TCA and glycolysis metabolites and not-so-commonlystudiedN-acetylatedaminoacids.Understandingand documentingthesebiologicalandbiochemicalchangesinthebrain couldproveinvaluableforfutureresearchintoneurodegenerative diseases,andrequiresinvestigationwithapreciseandrobust quan-titativeanalyticalapproach.Acomputationalapproachtowardsthe predictionofderivatizationofmetabolites,andthepredictionof retentionfor newmetabolites,will furthersupportthemethod applicationtocoverawiderrangeofmetabolitesincomplex matri-ces.

DeclarationofCompetingInterest

Theauthorsdeclarethattheyhavenoknowncompeting finan-cialinterestsorpersonalrelationshipsthatcouldhaveappearedto influencetheworkreportedinthispaper.

Acknowledgements

Theauthorexpressesthanksto:JacovanVeldhovenforsupport duringthesynthesisofDmPABr-D6;AlisaL.Willaceyforadviceand

guidanceduringthefinalizationofthestudyandAlidaKindtfor statisticalsupport.ThisprojectwassupportedbytheSysMedPD project,which hasreceivedfundingfromtheEuropeanUnion’s Horizon2020research andinnovation programme under grant agreementno,668738.

AppendixA. Supplementarydata

Supplementarymaterialrelatedtothisarticlecanbefound,in theonlineversion, atdoi:https://doi.org/10.1016/j.chroma.2019. 460413.

References

[1]L.M.Bayona,M.Videnova,Y.H.Choi,Increasingmetabolicdiversityinmarine spongesextractsbycontrollingextractionparameters,Mar.Drugs16(2018).

[2]N.Karu,L.Deng,M.Slae,A.C.Guo,T.Sajed,H.Huynh,E.Wine,D.S.Wishart,A reviewonhumanfecalmetabolomics:methods,applicationsandthehuman fecalmetabolomedatabase,Anal.Chim.Acta1030(2018)1–24.

[3]D.S.Wishart,Y.D.Feunang,A.Marcu,A.C.Guo,K.Liang,R.Vazquez-Fresno,T. Sajed,D.Johnson,C.Li,N.Karu,Z.Sayeeda,E.Lo,N.Assempour,M.Berjanskii, S.Singhal,D.Arndt,Y.Liang,H.Badran,J.Grant,A.Serra-Cayuela,Y.Liu,R. Mandal,V.Neveu,A.Pon,C.Knox,M.Wilson,C.Manach,A.Scalbert,HMDB 4.0:thehumanmetabolomedatabasefor2018,NucleicAcidsRes.46(2018) D608–d617.

[4]T.Li,E.Li,Y.Suo,Z.Xu,Y.Jia,J.G.Qin,L.Chen,Z.Gu,Energymetabolismand metabolomicsresponseofPacificwhiteshrimpLitopenaeusvannameito sulfidetoxicity,Aquat.Toxicol.183(2017)28–37.

[5]N.Karu,R.Wilson,R.Hamede,M.Jones,G.M.Woods,E.F.Hilder,R.A.Shellie, DiscoveryofBiomarkersforTasmanianDevilCancer(DFTD)bymetabolic profilingofserum,J.ProteomeRes.15(2016)3827–3840.

[6]S.A.Goldansaz,A.C.Guo,T.Sajed,M.A.Steele,G.S.Plastow,D.S.Wishart, Livestockmetabolomicsandthelivestockmetabolome:asystematicreview, PLoSOne12(2017)e0177675.

[7]L.W.Sumner,Z.Lei,B.J.Nikolau,K.Saito,Modernplantmetabolomics: advancednaturalproductgenediscoveries,improvedtechnologies,and futureprospects,Nat.Prod.Rep.32(2015)212–229.

[8]M.Ramirez-Gaona,A.Marcu,A.Pon,A.C.Guo,T.Sajed,N.A.Wishart,N.Karu, Y.DjoumbouFeunang,D.Arndt,D.S.Wishart,YMDB2.0:asignificantly expandedversionoftheyeastmetabolomedatabase,NucleicAcidsRes.45 (2017)D440–d445.

[9]V.Kantae,S.Ogino,M.Noga,A.C.Harms,R.M.vanDongen,G.L.Onderwater, A.M.vandenMaagdenberg,G.M.Terwindt,M.vanderStelt,M.D.Ferrari,T. Hankemeier,Quantitativeprofilingofendocannabinoidsandrelated N-acylethanolaminesinhumanCSFusingnanoLC–MS/MS,J.LipidRes.58 (2017)615–624.

[10]B.Zhou,J.F.Xiao,L.Tuli,H.W.Ressom,LC–MS-basedmetabolomics,Mol. Biosyst.8(2012)470–481.

[11]A.S.Lopes,E.C.Cruz,A.Sussulini,A.Klassen,Metabolomicstrategiesinvolving massspectrometrycombinedwithliquidandgaschromatography,Adv.Exp. Med.Biol.965(2017)77–98.

[12]P.Song,O.S.Mabrouk,N.D.Hershey,R.T.Kennedy,Invivoneurochemical monitoringusingbenzoylchloridederivatizationandliquid

chromatography–massspectrometry,Anal.Chem.84(2012)412–419.

[13]T.Higashi,S.Ogawa,Isotope-codedESI-enhancingderivatizationreagentsfor differentialanalysis,quantificationandprofilingofmetabolitesinbiological samplesbyLC/MS:areview,J.Pharm.Biomed.Anal.130(2016)181–193.

[14]J.M.Wong,P.A.Malec,O.S.Mabrouk,J.Ro,M.Dus,R.T.Kennedy,Benzoyl chloridederivatizationwithliquidchromatography-massspectrometryfor targetedmetabolomicsofneurochemicalsinbiologicalsamples,J. Chromatogr.A1446(2016)78–90.

[15]K.Guo,L.Li,Differential12C-/13C-isotopedansylationlabelingandfastliquid chromatography/massspectrometryforabsoluteandrelativequantification ofthemetabolome,Anal.Chem.81(2009)3919–3932.

[16]K.Guo,L.Li,High-performanceisotopelabelingforprofilingcarboxylic acid-containingmetabolitesinbiofluidsbymassspectrometry,Anal.Chem. 82(2010)8789–8793.

[17]J.Peng,L.Li,Liquid-liquidextractioncombinedwithdifferentialisotope dimethylaminophenacyllabelingforimprovedmetabolomicprofilingof organicacids,Anal.Chim.Acta803(2013)97–105.

[18]Y.Sun,D.Chen,J.Liu,Y.Xu,X.Shi,X.Luo,Q.Pan,J.Yu,J.Yang,H.Cao,L.Li,L. Li,Metabolicprofilingassociatedwithautophagyofhumanplacenta-derived mesenchymalstemcellsbychemicalisotopelabelingLC-MS,Exp.CellRes. 372(2018)52–60.

[19]M.Egorov,B.Delpech,G.Aubert,T.Cresteil,M.C.Garcia-Alvarez,P.Collin,C. Marazano,AconciseformationofN-substituted3,4-diarylpyrroles– synthesisandcytotoxicactivity,Org.Biomol.Chem.12(2014)1518–1524.

[20]Q.Hu,Z.Zhang,Y.Liu,T.Imamoto,W.Zhang,ZnCl2-promotedasymmetric hydrogenationof␤-secondary-aminoketonescatalyzedbyaP-chiral Rh–bisphosphinecomplex,Angew.Chem.Int.Ed.54(2015)2260–2264.

[21]J.He,L.Shi,S.Liu,P.Jia,J.Wang,R.Hu,Ultrasound-mediatedsynthesisof N,N-bis(phenacyl)anilineundersolvent-freeconditions,Monatsheftefür Chemie–Chem.Mont.145(2014)213–216.

[22]T.I.Yousaf,E.S.Lewis,Grouptransfers.1.Enolatestructurescontributingto thetransitionstatefornucleophilicsubstitutionon.alpha.-Substituted carbonylcompounds,J.Am.Chem.Soc.109(1987)6137–6142.

[23]H.Loghmani-Khouzani,M.R.Poorheravi,M.M.M.Sadeghi,L.Caggiano,R.F.W. Jackson,␣-Fluorinationof␤-ketosulfonesbyselectfluorTM_{F–TEDA–BF4,}

monogenicformofLeighsyndrome,CellRep.13(2015) 981–989.

[28]A.Mardinoglu,S.Shoaie,M.Bergentall,P.Ghaffari,C.Zhang,E.Larsson,F. Backhed,J.Nielsen,Thegutmicrobiotamodulateshostaminoacidand glutathionemetabolisminmice,Mol.Syst.Biol.11(2015)834.

[29]J.C.Schoeman,J.Hou,A.C.Harms,R.J.Vreeken,R.Berger,T.Hankemeier,A. Boonstra,Metaboliccharacterizationofthenaturalprogressionofchronic hepatitisB,GenomeMed.8(2016)64.

[30]I.Celardo,L.M.Martins,S.Gandhi,Unravellingmitochondrialpathwaysto Parkinson’sdisease,Br.J.Pharmacol.171(2014)1943–1957.

[37]M.deTommaso,E.Ceci,C.Pica,M.Trojano,M.Delussi,G.Franco,P.Livrea,M. Ruggieri,SerumlevelsofN-acetyl-aspartateinmigraineandtension-type headache,J.HeadachePain13(2012)389–394.

[38]E.Gerlo,R.VanCoster,W.Lissens,G.Winckelmans,L.DeMeirleir,R.Wevers, Gaschromatographic-massspectrometricanalysisofN-acetylatedamino acids:thefirstcaseofaminoacylaseIdeficiency,Anal.Chim.Acta571(2006) 191–199.