硅烷衍生化

AnalBioanalChem(2013)405:9193–9205

DOI10.1007/s00216-013-7341-z

RESEARCHPAPER

Theuseoftrimethylsilylcyanidederivatization

forrobustandbroad-spectrumhigh-throughputgas

chromatography–massspectrometrybasedmetabolomics

BekzodKhakimov

&

MohammedSaddikMotawia

&

SørenBak

&

SørenBallingEngelsen

Received:22July2013/Revised:22August2013/Accepted:2September2013/Publishedonline:4October2013

#

Springer-VerlagBerlinHeidelberg2013

AbstractReproducibleandquantitativegaschromatography–

massspectrometry(GC-MS)-basedmetabolomicsanalysisof

complexbiologicalmixturesrequiresrobustandbroad-

evaluatedderivatizationof

complexmetabolitemixturesusingtrimethylsilylcyanide

(TMSCN)andthemostcommonlyusedsilylationreagentN-

methyl-N-(trimethylsilyl)trifluoroacetamide(MSTFA).Forthe

comparativeanalysis,twometabolitemixtures,astandard

complexmixtureof35metabolitescoveringarangeofamino

acids,carbohydrates,smallorganicacids,phenolicacids,fla-

vonoidsandtriterpenoids,andaphenolicextractofblueberry

fferentderivatizationmethods,(1)

directsilylationusingTMSCN,(2)methoximationfollowed

byTMSCN(M-TMSCN),(3)directsilylationusingMSTFA,

and(4)methoximationfollowedbyMSTFA(M-MSTFA)were

comparedintermsofmethodsensitivity,repeatability,and

ivatizationmethodswere

observedat13differentderivatizationtimes,5minto60h,for

utomatedsamplederivatiza-

tionandinjectionenabledexcellentrepeatabilityandprecise

ptimalsilylationtimes,peak

intensitiesof34outof35metabolitesofthestandardmixture

wereuptofivetimeshigherusingM-TMSCNcomparedwith

ectsilylationofthecomplexstandardmix-

ture,theTMSCNmethodwasupto54timesmoresensitive

rly,allthemetabolitesdetectedfromthe

blueberryextractshowedupto8.8timeshigherintensities

er,

TMSCN-basedsilylationshowedfewerartifactpeaks,robust

profiles,andhigherreactionspeedascomparedwithMSTFA.

Amethodrepeatabilitytestrevealedthefollowingrobustness

ofthefourmethods:TMSCN>M-TMSCN>M-MSTFA>

MSTFA.

KeywordsMetabolomics

.

Gaschromatography–mass

spectrometry

.

Trimethylsilylderivatization

.

Methoximation

.

Trimethylsilylcyanide

Abbreviations

BSABis(trimethylsilyl)acetamide

BSTFAN,O-Bis(trimethylsilyl)trifluoroacetamide

EIElectronimpact

HCNHydrogencyanide

MEOXMethoxiamine

M-MSTFAMethoximationfollowedbyMSTFA-based

silylation

M-TMSCNMethoximationfollowedbyTMSCN-based

silylation

MPSMulti-purposesampler

MSTFAN-methyl-

N-(trimethylsilyl)trifluoroacetamide

PARAFAC2ParallelFactorAnalysis2

PCAPrincipalcomponentanalysis

RIRetentionindex

ElectronicsupplementarymaterialTheonlineversionofthisarticle

(doi:10.1007/s00216-013-7341-z)containssupplementarymaterial,which

isavailabletoauthorizedusers.

ov

:

en

QualityandTechnology,DepartmentofFoodScience,

FacultyofScience,UniversityofCopenhagen,Rolighedsvej30,

FrederiksbergC,1958Copenhagen,Denmark

ov(*)

:

a

:

PlantBiochemistry,DepartmentofPlantandEnvironmental

Sciences,FacultyofScience,UniversityofCopenhagen,

Thorvaldsensvej40,FrederiksbergC,1871Copenhagen,Denmark

e-mail:bzo@

9194

TMCSTrimethylchlorosilane

TMSTrimethylsilyl

TMSCNTrimethylsilylcyanide

Introduction

Gaschromatography–massspectrometry(GC-MS)hasbe-

comeoneofthefavoriteanalyticalplatformsappliedin

metabolomicsbecauseofitshighreproducibilityandresolu-

tionpower[1–3].IncontrasttoNMRandLC-MS,GC-MS

analysisrequiresmetabolitestobethermallystableandvola-

rtheboilingpointofmetabolites

andincreasevolatilityforGC-MSanalysis,complexbiolog-

icalsamples,suchasplantandanimaltissueextractsandbio-

fluidsneedtobederivatizedbychemicalderivatizationto

tcommonlyused

derivatizationmethodinvolvesmethoximationfollowedby

silylation[4–6].Duringthemethoximationstep,metabolites

withcarbonyl(>C=O)functionalgroupreactwiththereagent

(20–40mgmL

−1

solutionsofO-methylhydroxylaminehy-

drochlorideinpyridine)andformoxime(>C=N–O–CH

3

)

derivatives[7–9].Themainpurposeofoximationistoform

thermallystablederivativesthatpreventcyclizationofreduc-

ingsugars,formationofketo-enoltautomersofaldehydesand

ketoneswithaprotonintheα-position,andtoprotectother

carbonylgroupcontainingmetabolitesfromdecarboxylation

[10–12].Silylationservestosubstituteactivehydrogenatoms

ofhydroxyl(−OH),carboxylicacid(−COOH),primaryand

secondaryamines(R′NH

2

,R′R″NH),andthiols(−SH)witha

trimethylsilyl(−Si(CH

3

)3)group[13].

InGC-MSmetabolomics,differentsilylationreagentshave

beenappliedandtheydifferbytheirreactivity,selectivity,side

reactions,andbyproducts[13–16].Theefficiencyandspeed

ofthesilylationreactiondependonreactiontemperature,

time,andphysicochemicalpropertiesofboththesilylation

ilylationreagents,N-meth-

yl-N-(trimethylsilyl)trifluoroacetamide(MSTFA)hasbecome

themostcommonlyappliedreagentbecauseofitshighreac-

l

studiesonGC-MSmethodoptimizationsuggesttheuseof

MSTFAaloneortogetherwith1%ofTMCSasacatalystfor

silylationofcomplexmixtures[5,17,18].Silylationreactions

arereversibleandhavebeenshowntoproceedviabimolecu-

larnucleophilicsubstitution(S

N

2-mechanism)atthesilicon

atom[19–22].

AssilylationfollowsaS

N

2-mechanism,thesilylationde-

pendsonconcentrationofboththeelectrophile(silylation

reagent)andthenucleophile(substrate).Besidesconcentra-

tion,reactiontime,andtemperature,silylationreactionyield

ovetal.

andratedependon(1)natureoftheleavinggroupofthe

silylationreagent,(2)chemistryofthesubstratethatreacts

withsilylationreagent,(3)stericeffects,(4)influenceof

solvents,and(5)eralsilylation

activityorderofthedifferentfunctionalgroupsisasfollows:

alcohols(primary>secondary>tertiary)>phenols>carboxylic

acids>thiols>amines(primary>secondary)>

ylationrateincreasesby

increasingtheaffinityofthesiliconatomtothenucleophile

inityofthe

siliconatomishighesttoanoxygenatom[23,24]and

trimethylsilylationreagentswithgoodleavinggroupsare

moreactiveinexchangingatrimethylsilylgroupwithan

activehydrogenatomorametalatom[19,25].Inaddition,

thenucleophilicattackbyasubstrateontheelectrophile

siliconatombecomeseasierwhenthecovalentbondbetween

thesiliconatomandtheleavinggroupisweakandeasily

dissociable[20,23].Solventsalsoinfluencetherateandthe

neisthemostlyused

solventduringmethoximationandsilylationreactions,andas

itisaweakbase,itcanincreasesilylationreactionrateby

scavengingactiveprotons(H

+

)-

cessibilityofthenucleophiliccenterofthesubstrateiscrucial,

andstudieshavedemonstratedaninfluenceofstericeffectsin

silylationofbranchedsecondaryaminesandprimaryamines

[26].Todate,themostthoroughlystudiedsilylationreagents

arethosewithaSi–Nbond,andsilylatedamidessuchasN,O-

bis(trimethylsilyl)acetamide,N,O-bis(trimethylsilyl)trifluoro

acetamide,andMSTFAhavebecomecommonsilylationre-

rast,veryfewstudieshave

beenconductedonsilylationcapabilitiesofcompoundswitha

Si–Cbond,suchastrimethylsilylcyanide(TMSCN).Two

recentreviewsonderivatizationreagentsandtheirreactions

usedforGC-MSanalysisofabroadrangeofmetabolites

discusstheadvantagesandlimitationsofavarietyofdifferent

derivatizationreagents[13,21].However,TMSCNwasnot

showthatTMSCNpossessesseveraladvantagesoverthe

majorityofthereagentsuseduptodate,includingahigh

silylationreactivityandreproducibility.

Inthepresentstudy,theuseofTMSCNforderivatization

nic

chemistry,TMSCNismainlyusedasasourceofcyanide

groupforvarioussyntheticreactions[27].AlthoughMai

andPatil[28]haveshownhighsilylationreactivityof

TMSCNtowardmanyfunctionalgroups,itspotentialasa

derivatizationreagentincomprehensiveGC-MSanalysisof

onlypublished,toourknowledge,exampleoftheuseof

TMSCNforGC-MSanalysiswaspublishedintheearly

1990s,whereitwasusedforthederivatizationofthe

Theuseoftrimethylsilylcyanidederivatization

prostacyclinanalogI[29].Inthisstudy,wecomparedGC-MS

analysisofvariousclassesofcompoundssilylatedbyusing

TMSCNandMSTFA.

MaiandPatil[28]haveconductedacomprehensivestudy

onsilylationofalcohols,phenols,carboxylicacids,amines,

tudy,theyshowedthat

TMSCNoutperformedmanyotherreagentsandprovided

dyshowedthatunder

mildconditions:5minat25°Cformostofalcohols,phenols,

andcarboxylicacidsand5–30minat25–100°Cforsecond-

aryamines,thiols,andcarbohydrates,TMSCN-based

silylationreactionyieldreachedupto98%.Thestudyillus-

tratedhighersilylationefficiencyandthereactionrateof

TMSCNwhencomparedwithotheralkylcyanidederivatiza-

tionreagentstowardsstericallyhinderedfunctionalgroups

tion,they

showedhighsilylationreactionyieldwithneat(solventfree)

TMSCNcomparedwiththeuseofsolventsinceallmetabolites

werereadilysolubleandratherneutralpHofTMSCNenables

non-destructivesilylationofbasesensitivecompounds.

Byproductformationisoneofthelimitationsofmost

uctsareformedduringthe

silylationreactionsandmayresultinformationofmultiple

artifactpeaksthatdecreaseprofilereproducibility,hamperthe

detectionofearlyelutingmetabolites,anddegradationof

mple,oneofthemostcommonly

usedsilylationreagentinconjunctionwithMSTFAisTMCS

thatformhydrogenchlorideasabyproduct,whichisan

encyanide

(HCN)istheonlybyproductformedinTMSCN-based

ooweakanacidtohydrolyzetheTMS-

derivatizedproducts,butbycontrast,itcanprotonateTMSCN

thatleadtoincreasedelectrophilicityandthusservestofurther

increasesilylationefficiency.

Thepurposeofthisstudyistoassessthesilylationcapa-

bilitiesofTMSCNtowardsvariousclassesofmetabolitesthat

areoftendetectedinGC-MSmetabolomicstudiesofcomplex

biologicalsamples,andtocomparewiththesilylationeffi-

abolite

mixtures,astandardmixturethatcompiled35differentcom-

poundsincludingaminoacids,carbohydrates,smallorganic

acids,phenolicacids,flavonoidsandtriterpenoids,anda

phenolicextractofblueberryfruitswereusedtoevaluatethe

nd

MSTFAsilylationperformanceswereevaluatedinconjunc-

tionswithamethoximationstep(usingpyrimidineasasol-

vent)andwithoutmethoximation(directsilylation)where

fferentderiv-

atizationmethods:(1)directsilylationusingTMSCN,(2)

methoximationfollowedbyTMSCN(M-TMSCN),(3)direct

silylationusingMSTFA,and(4)methoximationfollowedby

9195

MSTFA(M-MSTFA)wereevaluatedat11–13different

silylationtimepointsintherangeof5minto60h.

Materialsandmethods

Preparationofmetabolitestandardmixtureandextraction

ofblueberryfruits

Individualsolutionsof35standardcompoundscontaining6

aminoacids(valine,serine,threonine,glycine,aspartic

acid,andphenylalanine),6carbohydrates(ribitol,ribose,glu-

cose,glucose-6-phosphate,maltose,andsucrose),13organic

acidsandphenoliccompounds(benzoicacid,succinicacid,

malicacid,palmiticacid,phenyllacticacid,4-hydroxybenzoic

acid,2-hydroxy-2-methoxibenzoicacid,vanillicacid,2-

hydroxycinnamicacid,p-coumaricacid,caffeicacid,4-

hydroxiacetophenone,andvanillin),2polyphenols(naringenin

andcatechin),and8triterpenes(cholesterol,β-amyrin,α-

amyrin,lupeol,oleanolicacid,hederagenin,betulinicacid,and

α-epoxi-β-amyrin)werepreparedintheconcentrationof2or

0.25mgmL

−1

(onlyfortriterpenes).Mostofthestandard

compoundsweresolubleinwater,apartfrompolyphenols,

caffeicacid,benzoicacid,2-hydroxy-3-methoxybenzoicacid,

palmiticacid,andtriterpeneswhichweresolubilizedindimethyl

icacid,p-coumaricacidandcholesterolwere

solubilizedin96%olitestandardmixturewas

preparedbycombining200-μLaliquotsofallthesolutionsof

standardcompounds,besidessolutionsoftriterpenes,which

wereaddedindoubleamount(400μL).Thisstandardmixture

icandorganicacids

ofblueberryfruitswereextractedessentiallyaccordingtothe

protocoldescribedbyZadernowskietal.[30].Toincreasethe

extractionyield,slightmodificationswereintroducedtothe

rryfruitsoflow-bushblueberry(Vaccinium

myrtillus)werepurchasedfromthegroceryshopIrma(Copen-

hagen,Denmark);100goffrozenblueberryfruitswereextracted

fivetimeswith100mLof80%(vol/vol)methanolatroom

temperaturefor40minbyusinganorbitalshakerat500rpm.

Theextractswerecentrifugedat3,000×gfor5min,andtheclear

supernatantscombinedin1,000mLroundbottomflaskand

driedusingarotaryvacuumevaporatorfollowedbyfreeze-

drying;1.5goffreeze-driedextractwasdissolvedin100mL

of4Msodiumhydroxide,hydrolyzedunderthenitrogengas

atmospherefor4hatroomtemperaturewhilemixingat300rpm

esolutionwasacidifiedwith

6MhydrochloricacidtopH2,andextractedfivetimeswith

diethylether(1:1,vol/vol)for15minwhilemixingat300rpm

ntheetherextractfromfatty

acidsandothernonpolarcompounds,thecombinedetherfrac-

tionsweredriedusingrotaryvacuumevaporator,re-dissolvedin

9196

150mLof5%(wt/vol)sodiumbicarbonatesolution,andthen

remainingwaterphasewasacidifiedwith6Mhydrochloricacid

topH2,extractedfivetimeswithdiethylether,thecombined

etherfractionsdriedfinallyre-dissolvedin17mLof80%(vol/

vol)ultingextractreferredasphenolicextract

Amainlycontainsfreephenolicandorganicacidsaswellas

blueberryphenolicextractBwasobtainedinessentialthesame

wayasphenolicextractA,exceptthathydrolysisof1.5gof

freeze-driedextractwasperformedusing100mLof2M

hydrochloricacidandstirringfor40minat94°the

freephenolicacids,organicacids,andphenolicacidsderived

fromhydrolysisofesterbonds,thisextractalsocontainthe

phenolicacidsderivedfromthehydrolysisofglycosidicbonds.

Chemicals

Allthecompoundsusedforpreparationofstandardmixture

andsolventswerepurchasedfromSigma-Aldrich,exceptfor

4-hydroxiacetophenone,hydrochloricacid(37%),andsodi-

umbicarbonatethatwereobtainedfromMerck,inbestavail-

ylsulfoxide,diethylether,2-hydroxy-3-

methoxybenzoicacid,trimethylsilylcyanideandN-methyl-

N-(trimethylsilyl)trifluoroacetamidewerepurchasedfrom

penoid12α,13α-epoxy-

3β-a.

Waterusedthroughoutthestudywaspurifiedusinga

MilliporeMilli-Qlabwatersystemequippedwith0.35μm

filermembrane.

Samplederivatization

Derivatizationandinjectionofsampleswasfullyautomated

byuseofaGERSTELMultiPurposeSampler(MPS)with

DualRaitWorkStationintegratedtoaGC-MSsystemfrom

enabledreproduciblesamplederivatization

inahigh-throughputmannerandwasfullyoperatedbyMAE-

STROsoftwareintegratedwithAgilent'sChemStationsoft-

abledautomationofindividualandparallel

samplederivatizationstepsfromasinglesequencedeveloped

fortheanalysisofseveralsamplesindependentlyfromchro-

matographicsystemandprovidedprecisederivatizationtime

esofsyringeswereinstalled,

theleftMPSwasequippedwitha10-μLsyringeandused

onlyforGC-MSinjection,whereastherightMPSwas

equippedwitha100-μLsyringeandusedinsamplederivati-

oderivatization,100-μLaliquotofthe

blueberryextractand70-μLaliquotofstandardmixturesam-

pleswerelyophilizedin150-μLglassinsertsunderreduced

pressureatroomtemperaturebyuseofavacuumcentrifuge,

transferredinto1.5mLGC-MSvials,andsealedwithmag-

neticcapswithsiliconeseptumundernitrogengastoprevent

ovetal.

magneticcaps(ML33032Afrom

)withsiliconseptumenabledMPSto

moveGC-MSvialsandpreventedsolventand/orreagent

evaporationduringthederivatizationevenafterpenetration

lizedandsealedsam-

pleswereplacedonaMPSsampletrayandfurthersample

handlingwasfullyautomated.

Methoximationofsampleswasperformedbyadditionof

40μLfreshlyprepared20mgmL

−1

methoxiaminehydro-

chloride(CH

3

ONH

2

·HCl)inpyridineandincubatedfor

90minat30°ethoximation,

samplesweresilylatedbyadditionof40μLsilylationreagent,

andatotalof80μLreactionmixturewasincubatedat37°C

thevolumeofthereaction

mixtureconstantinallfourderivatizationmethods,direct

silylationmethods,TMSCNandMSTFAwereperformedby

additionof80-μLpuresilylationreagentandincubatedat

37°nthereactionstoichi-

ometry,theamountofappliedsilylationreagentsexceededat

least400(TMSCN),200(M-TMSCN),250(MSTFA),and

125(M-MSTFA)timestheamountneededforsilylationofall

theavailableactivehydrogenatomspresentinthe70-μL

-MSprofilesofthestandard

mixturewasevaluatedat11differentsilylationtimesbyusing

allfourderivatizationmethods,whereasderivatizationof

blueberryextractwasperformedat13differentsilylation

timesbyusingthree(TMSCN,M-TMSCN,andM-MSTFA)

antpracticalconsiderationsof

theautomatedsamplederivatization,GC-MSanalysisare

describedindetailintheElectronicsupplementarymaterial

(ESM;text).

Dataacquisition

Analiquotof1.0μLderivatizedsamplewasinjectedeitherin

split(splitratioof3:1wasusedinblueberryextractanalysis)

orinsplitlessmode(foranalysisofstandardmixture)intoa

Gerstelcooledinjectionsystem(CIS)equippedwithaglass

edsettingsofleftandrightMPS

syringes,sampleincubatingagitator,andCISinjectionport

parameterscanbefoundintheESM(text).TheGC-MS

consistedofanAgilent7890AGCandanAgilent5975C

rationwasperformedonanAgilentHP-

5MScolumn(30m×250μm×0.25μm)byusinghydrogen

carriergasattheconstantflowrateof1.2mLmin

−1

.TheGC

oventemperatureprogramwasasfollows:initialtemperature,

60°C;equilibrationtime,1.0min;heatingrate,

12.0°Cmin

−1

;endtemperature,310°C;holdtime,6.0min;

andpost-runtime,5minat60°ectrawererecorded

intherange

1

of50–750m/zwithascanningfrequencyof2.3

scanss

−

,andtheMSdetectorwasswitchedoffduringthe3-

nsferline,ionsource,and

quadrupoletemperaturesweresetto280,230,and150°C,

Theuseoftrimethylsilylcyanidederivatization

sspectrometerwastunedaccordingtothe

manufacturerrecommendationsbyusingperfluorotributylamine.

Dataanalysis

RelativeabundancesofmetaboliteswerecalculatedbyParal-

lelFactorAnalysis2(PARAFAC2)modelingoftherawGC-

MSdata[31,32].Themethodallowedprecisequantification

ofwell-resolved,co-eluted,overlapped,andlowS/Nratio

peaksusingfullmassspectraormarkerm/zions(for

completelyembeddedpeaks)come

ofPARAFAC2modelswere(1)scoresofeachresolvedpeak

thatcorrespondtotherelativeareasofthedetectedpeaks,(2)

spectralloadingsofeachresolvedpeakwhichrepresentpure

massspectraofthedetectedpeaks,and(3)elutiontime

explorativeanalysisofthederivatizationmethods,principal

componentanalysis(PCA)[33]wasappliedtoamatrixcon-

tainingrelativeabundancesofdetectedmetabolites(variables)

indifferentderivatizationmethods(rows).Metaboliteswere

identifiedbasedontheirretentionindices(RI)andEI-MS

librarymatchusingcommercialWiley08andNIST05librar-

-MSlibrary

searchwasperformedusingoriginalmassspectraor

PARAFAC2resolvedmassspectraofchromatographicpeaks.

Retentionindicesofeachmetabolitewerecalculatedusingin-

houseMATLABfunctionbasedontheVandenDooland

Kratzequation[34]andretentiontimesofC10–C40alkanes

thatwereanalyzedusingthesameGC-MSmethod.

Software

GC-MSchromatographicdatawasanalyzedusingAgilent

TechnologiesChemStationsoftware(version:E.02.02.1431).

PARAFAC2andPCAmodelingwereperformedbyusing

PLSToolbox(Version6.0.1,EigenvectorResearchInc.

USA)workingunderMATLAB(Version7.13.0.564,

R2011b,theMathworksInc.,USA)es

ofrawGC-MSdatawereimportedintoMATLABusingthe

function[35].

Safetyconsiderations

Allderivatizationreagents(includingMSTFAandTMSCN)

ingly,derivatizationre-

agentsmustbehandledundertheinertgasatmosphereina

tentionmustbepaidtoavoidcontactwith

riateviallidsmustbetestedfor

theirabilitytosealthevialsandtopreventevaporationofthe

tizedsamplesand

reagentsmustberemovedfromtheautosamplershortlyafter

injectionandshouldbedisposedaccordingtotheinstructions

ustbeawareofthesafety

9197

precautions,preventmoisturethatcausesformationof

rthtomention

thatoneofthepossiblebyproductsoftheTMSCNishydro-

tizationmustbeperformedunderinertgas

ifmanualhandlingisnecessary,ifanautosamplerisemployed

tightlysealedvialsthatpreventevaporationofboth,there-

agentandbyproducts,shouldbeused,andfinallydisposaloff

leftoversilylationreagentandvialswithsilylatedsamples

tailed

safetyconsiderationsofusingTMSCNandMSTFAaswell

asreagentevaporationstestsaredescribedindetailinthe

ESM(text).

Resultsanddiscussion

Derivatizationmethodcomparisonofstandardmixture

Globalanalysis

TheGC-MSprofilesobtainedfromthefourderivatization

methodsdiffered,bothqualitativelyandquantitatively,and

optimalsilylationtimeofeachofthederivatizationmethod

wasdefinedasthetimeofincubationatwhichamaximum

numberofmetabolitesreachedtheirhighestpeakintensity.

Optimalsilylationtimesofthederivatizationmethodswereas

follows:TMSCN,40min;M-TMSCN,60min;MSTFA,

30min;andM-MSTFA,1illustratestotal

ionchromatogramsofthestandardmixtureattheoptimal

1lists41derivatives

thatoriginatedfrom35metabolitesofthestandardmixture

andtheirrelativeratiosattheoptimalsilylationtimeofeach

-

atizationproductsofallthe35compoundsusedinthestan-

dardmixture,usingboththemethoxiaminationfollowedby

trimethylsilylation(M-TMSCNandM-MSTFA)anddirect

trimethylsilylation(TMSCNandMSTFA)methods,are

highlightedintheESM,eatabilityofthe

derivatizationmethodswascalculatedfromtherelativestan-

darddeviationsofabundancesof41derivativesmeasuredin

fourreplicates,foreachderivatizationmethodattheiroptimal

nerrorsofthederivatizationmethods

TMSCN,M-TMSCN,MSTFA,andM-MSTFAwere3.8%

(varyingfrom1.2to10.1%forallmetabolites),6.4%(1.2to

17.8%),26.2%(7.7to41.6%),and13.9%(2.4to30.1%),

uatethesignificanceofthedifferences

observedinthefourdifferentderivatizationmethods,PCA

analysiswasperformedonthereplicatedatamatrix(16×41)

thatinclude16samples,4replicatespermethod,andabun-

uently,

ANOVAanalysiswasperformedtoevaluatethe

Fstatistics

(variationsbetweentreatments/variationswithintreatments)

9198

Fig.1Totalioncurrent

chromatogramsofGC-MSdata

obtainedfromthecomplex

standardmixtureusingthefour

differentderivatizationmethods

attheoptimalsilylationtimes.

Peaksarenumberedinthesame

orderaspresentedintheESM,

TableS1

ovetal.

Retention time (min)

andthepvaluesbyusingthescoresofPC1andPC2that

explainedmorethan75%S1(ESM)

showsthePC1versusPC2scoresplotofthePCAanalysis

andthecorrespondingboxplotoftheANOVAanalysis

ggestsrejectionofthe

nullhypothesis(p<0.01)with95%oftheconfidenceand

showsthesignificanceofthedifferencesobservedbetween

xploratoryevaluation

ofthefourdifferentderivatizationmethodsatdifferent

silylationtimepoints,twoPCAmodelsweredeveloped.

ThefirstPCAmodelwasdevelopedonaX(56×34)

metabolomicdatathatcompiled34metabolitesdetectedin

all56samples(fourderivatizationmethodsevaluatedatelev-

endifferentsilylationtimes,includingreplicatesattheoptimal

derivatizationtimes).ThescoresandloadingsplotofthePCA

1versusPC2scoresplot

ofthePCAmodelrevealapartialseparationofthederivati-

zationmethods(Fig.2a),wheresamplesofeachmethodform

itstrajectoryfromthelefttotherightoftheplotbyincreasing

sthatarelocatedtotheveryrightside

oftheplotrepresentthesilylationtimepointswhenTMS-

derivativesreachedtheirhighestintensities(optimalsilylation

time).AsPC1versusPC2loadingsplotofthemodelshow

thatallmetabolitesarelocatedintherightsideoftheplot

havingpositiveloadingsinPC1(Fig.2b).However,further

increaseofsilylationtimeresultedindecreaseofmetabolites'

intensities,thusafteroptimalsilylationtime,samplesare

the

optimalsilylationtimes,therelativeabundancesofmostofthe

metaboliteswerehigherwhenusingTMSCNandMSTFA

methodscomparedwithM-TMSCNandM-MSTFA,and

accordingly,theyshowedhigherscoresinPC1(Fig.2a).

Theloadingsplotalsoshowsapartialseparationofvariables

thatassistedtocomparedifferentderivatizationmethodsfor

detectionofvariousmetabolitesbyvisualobservationof

theorganicandphenolic

acidsareclusteredontheupperrightsideoftheplothaving

positiveloadingsinPC2,whiletriterpenesandmetabolites

,poly-

phenols,sucrose-8tms,ribitol-5tms,serine-3tms,glycine-

3tms,andthreonine-3tms)havenegativeloadingsinPC2

andthusformsclustersonthelowerrightsideoftheplot.A

Theuseoftrimethylsilylcyanidederivatization9199

Table1Trimethylsilyl(TMS)andmethoxime-trimethylsilyl(MEOX-TMS)derivativesderivedfrom35compoundsofstandardmixturesorted

accordingtotheirretentiontime

No.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

Substance

Valine-2TMS

Benzoicacid-1TMS

Serine-2TMS

Threonine-2TMS

Glycine-3TMS

Succinicacid-2TMS

Serine-3TMS

Threonine-3TMS

Asparticacid-2TMS

4-hydroxyacetophenone-1TMS

Malicacid-3TMS

Vanillin-1TMS

Phenylalanine-1TMS

Phenyllacticacid-2TMS

4-hydroxybenzoicacid-2TMS

Vanillin-MEOX-1TMS

2-hydroxy-3-methoxybenzoicacid-2TMS

(trans)-ribose-MEOX-4TMS

Ribitol-5TMS

Vanillicacid-2TMS

2-hydroxycinnamicacid-2TMS

(trans)-glucose-MEOX-5TMS

p-coumaricacid-2TMS

(cis)-glucose-MEOX-5TMS

Palmiticacid-1TMS

(trans)-caffeicacid-3TMS

(trans)-glucose-6-phosphate-MEOX-4TMS

(cis)-glucose-6-phosphate-MEOX-4TMS

Sucrose-8TMS

(trans)-maltose-MEOX-8TMS

(cis)-maltose-MEOX-8TMS

Naringenin-3TMS

d

Catechin-5TMS

Cholesterol-1TMS

β-amyrin-1TMS

α-amyrin-1TMS

Lupeol-1TMS

Oleanolicacid-2TMS

Betulinicacid-2TMS

α-epoxi-β-amyrin-1TMS

Hederagenin-3TMS

d

TMSCN

a

66.93(40)

3.50(50)

1.73(10)

0.49(20)

2.46(150)

2.18(40)

29.34(40)

12.86(50)

23.41(50)

3.94(40)

1.94(30)

0.42(150)

34.51(50)

1.83(50)

1.92(30)

No

5.36(40)

No

1.41(50)

4.48(50)

2.35(40)

No

4.23(40)

No

1.31(50)

2.31(30)

No

No

2.32(30)

No

No

7.85(40)

1.77(150)

2.17(40)

1.74(50)

2.04(50)

1.44(60)

1.50(40)

2.47(50)

4.02(50)

1.83(50)

M-TMSCN

a

4.56(60)

2.05(60)

1.66(40)

0.49(40)

1.37(30)

1.93(150)

1.10(40)

1(150)

2.07(150)

2.24(150)

1.59(150)

No

0.69(150)

1.55(150)

1.48(150)

2.23(60)

4.48(150)

3.01(60)

1.16(60)

2.23(150)

2.16(150)

2.97(60)

1.77(150)

3.62(60)

1.86(60)

1.52(150)

4.50(150)

3.62(150)

2.15(300)

4.51(300)

4.61(300)

2.86(150)

1.35(30)

2.33(150)

1.92(150)

2.07(150)

1.61(150)

2.07(150)

2.16(150)

3.43(60)

1.93(150)

MSTFA

a

37.55(30)

5.31(50)

26.27(10)

13.46(20)

1.86(20)

2.32(40)

0.54(40)

0.42(60)

21.88(40)

9.83(20)

2.25(50min)

1(30)

58.55(40)

2.04(20min)

3.43(40)

No

4.32(30)

No

1.35(20)

5.78(30)

7.62(30)

No

5.74(30)

No

1.80(30)

2.93(30)

No

No

1.60(30)

No

No

6.11(20)

1.14(30)

4.20(30)

1.81(30)

1.94(30)

1.11(50)

1.67(30)

2.04(40)

2.58(30)

1.29(50)

M-MSTFA

a

(150)

(150)

(30)

(60)

(150)

(300)

(60)

No

(150)

(150)

(300)

No

(150)

(150)

(150)

(300)

(150)

(150)

(150)

(150)

(150)

(150)

(300)

1(60)

(60)

(150)

(150)

(150)

(300)

(300)

(150)

(150)

(40)

(60)

(150)

(150)

(150)

(300)

(150)

(150)

(150)

RI

b

1,208

1,242

1,251

1,289

1,302

1,308

1,353

1,375

1,419

1,464

1,489

1,535

1,548

1,584

1,626

1,648

1,692

1,699

1,745

1,768

1,811

1,925

1,940

1,948

2,044

2,151

2,375

2,395

2,709

2,814

2,843

2,905

2,926

3,176

3,402

3,443

3,454

3,665

3,687

3,755

3,790

EI-MS

c

90

94

93

92

92

95

95

95

94

89

91

91

86

90

94

91

94

87

98

96

95

89

88

89

91

93

88

88

90

87

87

85

86

97

98

99

98

99

99

97

99

Nonopeakwasobserved

a

PARAFAC2-basedrelativeabundancesofderivativesattheiroptimalsilylationtimesinminutes(indicatedinthebrackets),foreachderivatization

methodareillustratedastheirratiototherelativeabundanceofthecorrespondingderivativesinmethodM-MSTFA

RetentionindicesofmetabolitederivativeswerecalculatedbasedonVandenDoolandKratzequationbyusingretentiontimesofC10–C40alkanes

EI-MS-basedlibrarymatchofthemetabolitesbyusingWiley08andNIST05

PARAFAC2-basedquantificationofderivativesincludedonlycharacteristicm/zionbutnotfullmassspectra

b

c

d

9200

Scores Plot

6

4

5 m

ovetal.

0.4

20 m

50 min

40 m

30m

Loadings Plot

(a)

10 m

(b)

0.3

0.2

0.1

0

-0.1

-0.2

serin_2tms

threonin_2tms

phenylalanine_1tms

aspartic acid_2tms

p-coumaric acid_2tms

cinnamic acid_2tms

aspartic acid_3tms

vanillic

malic acid_1tms

acid_2tms

valin_2tms

phenyllactic acid_2tms

cholesterol_1tms

succinic acid_2tms

naringenin_n tms

caffeic acid_3tms

P

C

2

(

1

5

%

)

2

0

-2

-4

-6

-10

60 m

150 m

r3_30 m

5 h

r2_30 m

30 h

15 h

5 m

5 m

20 m

10 m

5 m

10 m

30 h

30 m

20 m

15 h

30 m

r1_30 m

10 m

40m

5 h

r1_40 m

40 m

20 m

r3_60 m

r2_40 m

150 m

50 m

r3_40 m

15 h

60m

r1_60 m

30 h

30 h

5

50 m

150 m

30 m

r2_60 m

h

60 m

40 m

15 h

5 h

50 m

r1_60 m

150 m

r2_60 m

60m

r3_60 m

P

C

2

(

1

5

%

)

serin_3tms

threonin_3tms

catechin_n tms

ribitol_5tms

betulinic acid_2tms

epoxi-b-amyrin_1tms

glycine_3tms

lupeol_1tms

a-amyrin_1tms

palmitic acid_1tms

sucrose_8tms

hederagenin_3tms

oleanolic acid_2tms

-5

TMSCN

0

PC 1 (58%)

MSTFA

5

M-MSTFA

10

0.1

Amino acids

0.120.140.16

PC 1 (58%)

0.180.2

M-TMSCN

Triterpenes

Phenolic and Organic acids

Sucroseand ribitol

Naringenin and Catechin

Fig.2Scores(a)andloadings(b)plotsofPCAmodeldevelopedonamatrixcontainingPARAFAC2scores(relativeabundances)ofTMSderivativesof

thestandardmixturedetectedbyfourderivatizationmethods,at12differentsilylationtimepoints(includingfourreplicatesattheoptimalsilylationtimes)

visualobservationofscoresandloadingsplotsuggeststhat

detectionoftriterpenesandmetaboliteswithseveralex-

changeablehydrogensismoreefficientwhenusingTMSCN

y

beduetotherelativelysmallermolecularsizeoftheTMSCN

reagentthanMSTFA,whichenableamorerapidandefficient

silylationofstericallyhinderedexchangeablehydrogenatoms

ofsucrose(hydroxylgroupsatC2′),catechin,andaminoacids

(e.g.,glycine-3tmsandserine-3tms).Bycontrast,detectionof

someofthephenolicandorganicacidswasslightlymore

efficientwhenusingMSTFA,suggestingthatMSTFAsam-

comparethesilylationcapacityofTMSCNandMSTFAto-

wardsstericallyhinderedmetabolites2,6-diphenylphenolwas

owedarapid

heless,aconsiderable

TMSCNsilylation,amaximumintensityoftheTMS-product

wasobservedat5s,whereaswithMSTFA,amaximumwas

reachedafter5minandonlyreachedanintensityofabout

80%oftheTMSCNsignal(seeESM,Fig.S2).

ThescoresplotofthefirstPCAmodelalsoassistedin

evaluationofthederivatizationmethods'

sumsoftheinnerdistances(Euclideandistances)ofthefour

replicatestothecenterofaclusterthattheyformwerecalcu-

rameterprovided

-

clideandistancesofreplicatesofderivatizationmethodsin-

creasedinfollowingorder:TMSCN(1.2)

TMSCNmethodwasthemostreproduciblewhileMSTFA

servationwasinagreementwiththe

repeatabilitytestofthemethodscalculatedbasedonthe

relativestandarddeviationsofmetabolites.

ThesecondPCAmodelcomparedM-TMSCNmethod

withM-MSTFA(28samples)usingallcommonmetabolites

includingmethoxime-trimethylsilylated(MEOX-TMS)deriv-

ativesofreducingsugars(41variables).Thescoresplotofthis

PCAmodel(seeESM,Fig.S3)showedbetterseparationof

theM-TMSCNsamplesfromtheM-MSTFAsampleswhen

plescorre-

spondingtotheM-TMSCNmethodhadsignificantlyhigher

scoresonthePC1thanM-MSTFAsamplesandshoweda

generaltrendindevelopmentofderivatizationreactionsover

dingsplotofthemodelalso

showedhigherloadingsofallvariablesinPC1comparedwith

esultssuggestthattheM-TMSCNmethod

outperformedM-MSTFAintermsofsilylationreactionspeed,

efficiency,andrepeatability.

Derivatizationofsugars

AcomparisonoftheM-TMSCNandM-MSTFAmethodsfor

GC-MSdetectionofMEOX-TMSderivativesofreducing

sugars,vanillinandtheTMS-derivativeofsucroseat

bothmethods,therelativepeakabundancesofmetabolites

increasegraduallytoreachtheirmaximumbetweena

ptimalsilylationtime,

abundancesofallmetaboliteswere1.5–6timeshigherinthe

ingly,

theM-TMSCNmethodoutperformedM-MSTFAinterms

completenessandthesilylationreactionratesofmono-and

disaccharides,includingthestericallyhinderedC2′hydroxyl

Theuseoftrimethylsilylcyanidederivatization

Fig.3Relativepeakabundances

ofmethoximetrimethylsilyl

derivatives(MEOX-TMS)of

glucose,maltose,glucose-6-

phosphate,ribose,vanillin,and

TMSderivativeofsucroseinfour

differentderivatizationmethods

(TMSCN,M-TMSCN,MSTFA,

andM-MSTFA)oversilylation

timerangeof5minto30h.

Relativepeakabundancesof

TMSderivativeswerecalculated

usingPARAFAC2modelingand

ln(min)scaleofsilylationtime

wasusedforbettervisualization

x 10

3

2.5

7

9201

x 10

2

6

(a)

P

e

a

k

a

r

e

a

(b)

P

e

a

k

a

r

e

a

2

1.5

1

0.5

0

1.5

1

0.5

0

234567

234567

Log(min)

Glucose methoxime-5tms (M-TMSCN)

Glucose methoxime-5tms (M-MSTFA)

Maltose methoxime-8tms (M-TMSCN)

Maltose methoxime-8tms (M-MSTFA)

Sucrose-8tms (M-TMSCN)

Sucrose-8tms (M-MSTFA)

Log(min)

Glucose-6-phosphate-methoxime-4tms (M-TMSCN)

Glucose-6-phosphate-methoxime-4tms (M-MSTFA)

Ribose-methoxime-4tms (M-TMSCN)

Ribose-methoxime-4tms (M-MSTFA)

Vanillin-methoxime-1tms (M-TMSCN)

Vanillin-methoxime-1tms (M-MSTFA)

ptimalsilylationtimes

thepeakabundancesoftrans-andcis-glucose-MEOX-5TMS

derivativeswere3-to4-foldhigherwhenusingtheM-

TMSCNmethodascomparedwiththeM-MSTFAmethod.

Similarly,attheoptimalderivatizationtimethepeaksof

trans-andcis-glucose-6-phosphate-MEOX-4TMSwere3.6-

to4.5-foldhigherinthecaseoftheM-TMSCNmethodas

se,inthecases

oftrans-andcis-maltose-MEOX-8TMS,methoximation

followedbyTMSCN(M-TMSCN)outperformedM-

MSTFA,asthedetectedderivativespeakabundanceswere

upto4.6timeshigherattheoptimalsilylationtimes.

Toevaluatethederivatizationefficiencyofcarbohydrates,

e.g.,anumberofderivativesandtheirratios,sixdifferentsugars

wereindividuallyderivatizedbythefourdifferentmethods(see

ESM,Figs.S11,S12,S13,andS14).Directsilylationusing

eitherTMSCNorMSTFAresultedinfourderivativesofglucose,

α-andβ-glucopyranose-5TMS(represented90%inaratioof

1:1)andα-andβ-glucofuranose-5TMS,whereasonlytwo

derivatives,cisandtransisomersofglucose-methoxime-

sucroseTMS-derivativesweredetectedusingallderivatization

imesucrosederivativeswerenotobserved

sincesucroseisanon-reducingsugarandcannotreactwiththe

hreesucroseTMSderivatives,

sucrose-8TMSwasthemostabundantoneandelutedlastwhile

thetwoearlierelutingpeakscorrespondtosucroseTMS-

creasingsucrose

silylationtime,theratiosbetweentheTMS-derivativeschanged

inallderivatizationmethods,andinthecaseofTMSCN,sucrose

wasfullyconvertedtothesucrose-8TMSderivative,atthe

rast,MSTFAwasnotable

tofullysilylatesucrose,andevenafter15hsilylation,allthree

lusion,directsilylation

ofsucrosewithTMSCNandMSTFAderivatizationwasmore

efficientintermsofreactionrateanddetectionascomparedwith

theM-TMSCNandM-MSTFAmethods.

Derivatizationoftriterpenes

Attheoptimalsilylationtimes,detectionofsilylatedtriterpenes

wascomparablewhenusingTMSCNandMSTFAmethods,

whiletheM-TMSCNmethodhassignificantlyoutperformedthe

M-MSTFAandabundancesofpeakswere1.4–3.4timeshigher

(Fig.4;Table1).TMSCN-basedsilylationdepictedamaximum

peakintensitiesofTMS-derivativesoflupeol,β-amyrin,and

oleanolicacidafter50–60minofderivatizationandpeakinten-

sitiesremainedstableevenafterthesilylationtimeof15h.

Whereas,inthecaseofMSTFA-basedsilylationmetabolite

peaksreachedtheirmaximumat30minofsilylationandfurther

increaseofsilylationtimeresultedinasignificantdecreaseofthe

peakintensities.

Derivatizationofpolyphenols,organicandaminoacids

Relativeabundancesofcaffeic,malic,p-coumaric,and

phenyllacticacids'TMS-derivativesweretwotosixtimeshigher

usingM-TMSCNcomparedwithM-MSTFAmethodinthe

silylationtimerangeof10minto15h(seeESM,Fig.S4).These

TMS-derivativesreachedmaximumabundanceswithinthefirst

2.5hofsilylationandafter15hofderivatization,peakabun-

dancesstarttodecrease,presumablyduetodegradationofthe

ndMSTFAmethodsperformed

almostequallywellforthedetectionofcaffeic,p-coumaric,

andphenyllacticacids,andpeakintensitiesincreasedmuchfaster

andreachedmaximumat30

–ghthe

9202

Fig.4Relativepeakabundances

oftrimethylsilylderivativesof

triterpenes,suchaslupeol,

oleanolicacid,β-amyrin,and

betulinicacidinfourdifferent

derivatizationmethods(TMSCN,

M-TMSCN,MSTFA,andM-

MSTFA)oversilylationtime

ve

peakabundancesofTMS

derivativeswerecalculatedusing

PARAFAC2modelingand

ln(min)scaleofsilylationtime

wasusedforbettervisualization

x 10

6

5

5

ovetal.

x 10

6

5

5

(a)

P

e

a

k

a

r

e

a

234567

(b)

P

e

a

k

a

r

e

a

4

3

2

1

0

4

3

2

1

0

234567

Log(min)

Lupeol-1tms (TMSCN)

Lupeol-1tms (M-TMSCN)

Lupeol-1tms (MSTFA)

Lupeol-1tms (M-MSTFA)

Oleanolic acid-2tms (TMSCN)

Oleanolic acid-2tms (M-TMSCN)

Oleanolic acid-2tms (MSTFA)

Oleanolic acid-2tms (M-MSTFA)

Log(min)

β-Amyrin-1tms (TMSCN)

β-Amyrin-1tms(M-TMSCN)

β-Amyrin-1tms(MSTFA)

β-Amyrin-1tms(M-MSTFA)

Betulinic acid-2tms (TMSCN)

Betulinic acid-2tms (M-TMSCN)

Betulinic acid-2tms (MSTFA)

Betulinic acid-2tms (M-MSTFA)

stabilityofTMS-derivativesofthesemetabolitesweresignifi-

ptimal

silylationtimes,abundancesofTMS-derivativeswere1.5–4

timeshigherwhenusingTMSCNandMSTFAcomparedwith

dicatesahighersilylation

efficiencyofthesolventfreereagentscomparedwiththesolvent,

pyridine,interactionthatisusedduringthemethoximation.

Highersilylationefficiencyofthereagentalonemayalsobe

duetothebetteravailabilityofthesilylatingreagenttothe

exchangeablehydrogenatoms.

HighersensitivityandstabilityoftheTMS-derivativewas

observedindetectionof2-hydroxy-3-methoxybenzoicacid,

naringenin,catechin,succinicacid,and2-hydroxycinnamicacid

byusingTMSCN-basedmethods(TMSCNandM-TMSCN)

thanMSTFAandM-MSTFAmethods(Table1;andseeESM,

Figs.S5,6,7,9,andS9).Bycontrast,optimalpeakintensityof

2-hydroxycinnamicacidwas3.2timeshigherusingMTFSA

r,theintensityofthismetabolitewas

reducedupto37%whenMSTFAsilylationtimewasprolonged,

whichindicatesapossiblenon-stabilityofTMS-derivativesina

silylationofvanillinusingTMSCN

resultedindetectionofthevanillin-1TMSderivative,wherethe

hydroxylfunctionalgroupwassilylated,whereasthealdehyde

ptimalsilylationtime,

intensityofthevanillin-1TMSpeakwasalmosttwofoldhigher

whenMSTFAwasappliedascomparedwithTMSCN(Table1).

However,inthecaseofM-TMSCNmethod,theintensityof

vanillin-MEOX-1TMSwas2.23timeshigherthaninM-

MSTFAderivatization.

Attheoptimalsilylationtimesofaminoacids,valine,gly-

cine,asparticacid,andphenylalanine,thedirectsilylation

methodsperformedmuchbetterthanmethoximationfollowed

tensitieswere1.6-58timeshigherwhen

usingMSTFAcomparedwithM-MSTFA,and1.5-15times

higherwhenusingTMSCNcomparedwithM-TMSCNmeth-

tizationofserineandthreoninewerereagentspecific

(seeESM,Figs.S10andS11),inthecaseofMSTFA,serine

wasmainlydetectedintheformofserine-2TMS,whereaswith

TMSCN,theabundanceofserine-3TMSexceededserine-

2TMSseveraltimes,asserine-3TMSincreasedandserine-

r,bothserine

derivativesweredetectedwhenusingM-TMSCNandM-

ylationreactionrateandpeakabundanceswere

higherinM-TMSCNmethodcomparedwithM-MSTFA.A

similarpatternwasobservedinthederivatizationofthreonine.

Choosingtheoptimalderivatizationtime

OneofthecompromisemeasuresinGC-MSanalysisofcomplex

lsituationwouldbe

thattheTMS-derivativesofcomplexmixturemetabolitesreach

theirmaximumatthesamederivatizationtimeandremainstable

unately,thisisnotthecaseasTMS-

derivativesofdifferentclassesofmetabolitesreachmaximumat

thedifferentderivatizationtimes(eveninthesamereaction

conditions)whichvariespeakabundancesasafunctionof

degradationofthederivativesover

ore,itiscriticaltochoose

anoptimalderivatizationtimeforsimultaneousdetectionofa

widevarietyofmetabolitesincomplexmixturesandwitha

r,untilkineticderivatization

samplingbecomesfeasible,thederivatizationtimemustbe

keptconstantoverallsamplesdespitenotanoptimalderivati-

uently,thestabilityofthe

TMS-derivativesdependsonthechemistryoftheTMSderiv-

ative,moisturecontent,derivatizationreagents,time,and

Theuseoftrimethylsilylcyanidederivatization9203

study,thesignificanceofthesilylationtime

infourdifferentderivatizationmethodswereevaluatedand

variationsindetectionatthethreesilylationtimepointsclosest

erved

variationswerebetween7–21,6–18,16–39,and9–24%for

TMSCN,M-TMSCN,MSTFA,andM-MSTFA,respectively.

ForMSTFAthehighpeakvariationscanbeexplainedbyan

e

thanhalfofthemetabolitesofthestandardmixture,MSTFA

r,amoresignificant

influenceofthesilylationtimeandrelativelyfasterproduct

ative

lowvariationswhenusingM-TMSCNorM-MSTFAmaybe

duetoaprotonscavengingactivityofthesolventpyridineasit

emanymetabolites

ofstandardmixturewerederivatizedanddetectedwellusing

MSTFA,thelevelofreagentderivedunexpectedpeaksand

variationswererelativelyhigherthantheotherderivatization

methods.

Derivatizationmethodcomparisonofblueberryextracts

AsdirectsilylationbasedonMSTFAexhibitlowrepeatability,

itwasomittedfromtheGC-MSanalysisofthetwoblueberry

stextractisthephenolicextractA,whichis

obtainedfrombasichydrolysis,whilethesecondone(pheno-

licextractB)enknown

andfiveunknownmetabolitesweredetectedfromtheGC-MS

semetaboliteswere

quantifiedfromeachderivatizationmethod,TMSCN,M-

TMSCNandM-MSTFAthatwereevaluatedatthe13differ-

the

identifiedphenolicacidssuchasvanillic,caffeic,syringic,p-

coumaric,m-coumaric,protocatechuic,andgallicacidhave

previouslybeenfoundinaGC-MSstudyofsmallpolish

berries[30].Moreover,metaboliteslike,succinic,malic,

vanillic,p-coumaric,andcaffeicacidsthatareidentifiedfrom

thePphenolicextractAwerealsoincludedinthestandard

cilitatedacomparisonofderivatizationmeth-

odperformancesfromtwodifferentcomplexsamplematrices.

Forallthedetectedmetabolites,TMSCNmethodprovedtobe

superiortoMSTFAintermsofsensitivity,derivatizationrate

ptimalsilylationtimes,metab-

oliteabundanceswere1.5–3.0timeshigherwithM-TMSCN

r,direct

silylationwithTMSCNoutperformedthemethoximationbased

methodsandatoptimalsilylationtimes,peakintensitieswere

1.8–8.8timeshigher(Table2).

Table2Trimethylsilyl(TMS)derivativesidentifiedfromblueberryphenolicextractA,sortedaccordingtotheirretentiontime

No.

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

a

Substance

Glycerol-3TMS

Succinicacid-2TMS

Maleicacid-2TMS

Lacticaciddimer-2TMS

Malicacid-3TMS

Unknown-1

Vanilicacid-2TMS

Unknown-2

m-coumaricacid-2TMS

Protocatechuicacid-3TMS

Unknown-3

Unknown-4

Unknown-5

Syringicacid-2TMS

p-coumaricacid-2TMS

2-methyl-2-methoxy-mandelate-2TMS

Gallicacid-4TMS

(cis)-caffeicacid-3TMS

(trans)-caffeicacid-3TMS

TMSCN

a

4.96(40)

2.72(40)

3.84(40)

4.97(60)

2.71(50)

2.84(50)

4.65(50)

2.10(40)

6.71(60)

7.43(50)

2.91(50)

8.48(40)

9.6(50)

8.81(50)

5.15(40)

5.30(50)

4.87(50)

4.69(50)

2.46(60)

M-TMSCN

a

2.64(40)

2.00(40)

1.73(50)

1.96(60)

1.73(60)

1.60(40)

2.14(150)

1.66(40)

2.72(40)

2.94(40)

1.23(60)

2.25(30)

1.5(50)

2.33(40)

1.51(60)

1.82(40)

2.09(40)

2.13(60)

1.81(150)

M-MSTFA

a

(150)

(300)

(150)

(300)

(300)

(50)

(150)

(150)

(150)

(40)

(150)

(60)

(60)

(150)

(150)

(60)

(50)

(150)

(150)

RI

b

1,267

1,308

1,351

1,391

1,489

1,760

1,768

1,784

1,807

1,830

1,850

1,866

1,886

1,907

1,942

1,954

1,975

1,992

2,151

EI-MS

c

94

95

92

90

91

96

93

96

89

91

86

89

92

93

PARAFAC2-basedrelativeabundancesofderivativesattheiroptimalsilylationtimesinminutes(indicatedinthebrackets),foreachderivatization

methodareillustratedastheirratiototherelativeabundanceofthecorrespondingderivativeinthemethodM-MSTFA

RetentionindicesofmetabolitederivativeswerecalculatedbasedonVandenDoolandKratzequationbyusingretentiontimesofC10–C40alkanes

EI-MS-basedmassspectracomparisoninvolvedWiley08,NIST05aswellasin-housetriterpeneslibraries

b

c

ovetal.

Figure5displaystherelativeabundancesTMS-derivatives

ofeightmostabundantmetabolitesatdifferentsilylationtime

ureshowsthatthedirectsilylationusing

TMSCNresultedinhighermetabolitepeakintensitiesthan

M-TMSCNandMSTFAmethods,atallsilylationtimepoints.

Likewise,ateverysilylationtimepoint,abundancesofall

metaboliteswerehigherwhenusingthederivatizationmethod

5alsoshows

anincreaseinmetaboliteabundancesatthelatersilylation

times(45and60h).Thismightbeduetoanunintendedup-

concentrationbecauseoftheevaporationofthesolventpyri-

,itisimportant

tofindthebestcompromiseinsettingthederivatizationtime

toobtainanoptimalprofilewithhighrepeatability.

RepeatabilityofGC-MSanalysisoftheblueberryextracts

wasevaluatedbycalculatingtherelativestandarddeviations

of19quantifiedmetabolitesinfourreplicatesofeachderiv-

n

errorsofthederivatizationmethodsTMSCN,M-TMSCN

andM-MSTFAwere4.6(varyingfrom2.1to8.6%forall

metabolites),9.0(3.5to18.1%),and15.2%(3.7to21.3%),

epeatabilitytestsareinagreementwith

therobustnessofthederivatizationmethodsevaluatedinthe

irectsilylationmethod

TMSCNperformedbestintermsofsilylationreactionspeed,

efficiency,andrepeatability,thismethodwasusedfortheGC-

tractcontained

phenolicandotherorganicacidsthatarepresentinfreeforms,

andconjugatedformswithothercellmembranecomponents

tizationofthisextracts

withTMSCNfor40minenabledidentificationof27metab-

olitesbasedonRIandEI-MSpatterns,includingthe

triterpenoidssuchas,β-amyrin,α-amyrin,andoleanolicacid.

Thisresultconfirmedthepresenceoftriterpenoidsaponins

derivedfromβ-amyrin,α-amyrin,andoleanolicacid,as

previouslyreportedinastudyofanticancerpropertiesof

blueberryfruits(lus)[36,37].Thedevelopedproto-

colpromisesmoreinsightintosecondarymetabolitesofthe

Fig.5Relativepeakabundances

oftrimethylsilylderivativesof

somephenolicandorganicacids

identifiedfromblueberry

phenolicextractinfourdifferent

derivatizationmethods(TMSCN,

M-TMSCN,MSTFA,andM-

MSTFA)oversilylationtime

ve

peakabundancesofTMS

derivativeswerecalculatedusing

PARAFAC2modelingand

ln(min)scaleofsilylationtime

wasusedforbettervisualization

x 10

8

6

4

2

0

6

x 10

15

5

(a)

P

e

a

k

a

r

e

a

(b)

P

e

a

k

a

r

e

a

10

5

24

Log(min)

68

0

24

Log(min)

68

Protocatechuic acid-3tms (TMSCN)

Protocatechuic acid-3tms (M-TMSCN)

Protocatechuic acid-3tms (M-MSTFA)

Syringic acid-2tms (TMSCN)

Syringic acid-2tms (M-TMSCN)

Syringic acid-2tms (M-MSTFA)

p-Coumaric acid-2tms (TMSCN)

p-Coumaric acid-2tms (M-TMSCN)

p-Coumaric acid-2tms (M-MSTFA)

2-methyl-2-methoxy-mandelate-2tms (TMSCN)

2-methyl-2-methoxy-mandelate-2tms (M-TMSCN)

2-methyl-2-methoxy-mandelate-2tms (M-MSTFA)

x 10

15

5

x 10

2

5

(c)

P

e

a

k

a

r

e

a

(d)

P

e

a

k

a

r

e

a

10

1.5

1

5

0.5

0

0

2468

2468

Log(min)

Vanilic acid-2tms (TMSCN)

Vanilic acid-2tms (M-TMSCN)

Vanilic acid-2tms (M-MSTFA)

Caffeic acid-3tms (TMSCN)

Caffeic acid-3tms (M-TMSCN)

Caffeic acid-3tms (M-MSTFA)

Log(min)

Lactic acid dimer-2tms (TMSCN)

Lactic acid dimer-2tms (M-TMSCN)

Lactic acid dimer-2tms (M-MSTFA)

2,5-dihydroxycinnamic acid-3tms (TMSCN)

2,5-dihydroxycinnamic acid-3tms (M-TMSCN)

2,5-dihydroxycinnamic acid-3tms (M-MSTFA)

Theuseoftrimethylsilylcyanidederivatization

plant-derivedsamplesandenablesdetectionoflowconcen-

trationmetabolites.

Conclusions

TheGC-MSprofilesofcomplexmetabolitemixturesobtained

fromfourderivatizationmethodsweredifferentandsignifi-

ultssuggestthat

forthemajorityoftheinvestigatedmetabolitesofthecomplex

mixtures,TMSCN-basedmethodsoutperformedMSTFA-

basedmethodsintermsofsilylationreactionspeed,sensitiv-

ity,r,direct

silylationmethods,TMSCNandMSTFAperformedequally

-

though,theMSTFAmethoddisplayedsignificantlylower

TMS-derivativestabilityovertimeandlowerrepeatabilityof

ral,directsilylationmethods

providedbettersensitivityandmorerapidsilylation,though

methoximationbasedmethodsillustratedhighermetabolite

,basedontheresultsofthisstudy,itisrecom-

mendedtopayaspecialattentiontotheconsistencyofthe

samplepreparationandderivatizationpracticepriortocom-

aseofthedirectsilylation,itis

advisedtouse50–100μLpureTMSCNforcompletedried

extractsofthe100-to200-μLsampleandtoincubateat37°C

itialmethoxiaminationisrequiredthenthe

trimethylsilylationtimemustbeincreasedto120–150minto

allowsilylationofallthelabileprotonsinthepresenceof

ragesilylationvariationsofmethodsin-

creasedinthefollowingorder,M-TMSCN

MSTFA

almostinoppositedirection,MSTFA

TMSCN

silylationofvariousclassesofmetabolitesbyusingTMSCN

eitheraloneorinconjunctionwithapriormethoximationstep.

TheseresultsillustratehighpotentialofTMSCNasanalter-

nativesilylationreagentforderivatizationofcomplexbiolog-

icalmixturesinGC-MSmetabolomics.

AcknowledgmentsTheauthorsthanktheFacultyofScienceforsupport

totheelite-researcharea“Metabolomicsandbioactivecompounds”witha

ovandTheMinistryofScienceandTechnol-

ogyforagranttoUniversityofCopenhagen(en)withthetitle

“Metabolomicsinfrastructure”underwhichtheGC-MSwasacquired.

References

ntiKK,HoP,ChanE(2008)JChromatogrB-AnalTechnol

BiomedLifeSci871:202–211

,DavidF,LynenF,RumpelK,DugardeynJ,VanDer

StraetenD,XuG,SandraP(2011)JChromatogrA1218:

3247–3254

9205

(2008)Trac-TrendsAnalChem27:261–269

erU,WagnerC,KopkaJ,TretheweyRN,WillmitzerL(2000)

PlantJ23:131–142

,KopkaJ,TretheweyRN,WillmitzerL(2000)AnalChem

72:3573–3580

,SchauerN,KopkaJ,WillmitzerL,FernieAR(2006)Nat

Protoc1:387–396

F(1978)Recentadvancesinthesilylationoforganic

:BlauK,KingG(eds)

&Son

Inc,Philadelphia,PA,pp152–200

MandLuukkainT(1965)AnalyticalChemistry37:955

gMG,MossAM,HorningEC(1968)AnalBiochem

22:284–294

CW,NakamotoH,andZumwaltRW(1969)Journalof

Chromatography45:24–51

rH(1982)JChromatogr236:355–360

H,ChrysanthopoulosPK,KlapaMI(2008)JChromatogrB-

AnalTechnolBiomedLifeSci871:191–201

F(2013)JChromatogrA1296:2–14

JL(1999)JChromatogrA844:1–22

tographyCatalogue1998–99,RegisTechnologies,

1998;pp.86–88

vatization,Pierce2003–ationshandbookand

catalog

rgJ,JonssonP,NordstromA,SjostromM,MoritzT(2004)

AnalBiochem331:283–295

ssonAPH,MoritzT,MulderH,SpegelP(2012)Metabolomics

8:50–63

AE(1968)Chemicals

Co,Rockford,IL,p58

inaMV,,ovskii(1975)RassianChem.

Rev.44,733

(2012)Derivatizationreactionsandreagentsforgas

chromatographyanalysis,advancedgaschromatography—

progressinagricultural,biomedicalandindustrialapplications.

In:MustafaAliMohd(ed),:978-953-51-

0298-4

LH,ParkerGA,LloydNC,FreyCL,MichaelKW(1967)J

AmerChemSoc89:857

H(2002)Resonance7:48

(1961)JOrgChem26:232

ffA,LingemanH(1984)JPharmBiomedAnal2:337–

380

erLandBrokmeieD(1968)TetrahedronLetters9:

1325–1328

awaS,FujikawaS(2012)TetrahedronLett53:1075–

1077

,PatilG(1986)JOrgChem51:3545–3548

PP,KarasiewiczRJ,RosenP,ToomeV(1992)JChromatogr

Sci30:29–

31

owskiR,NaczkM,NesterowiczJ(2005)JAgricFoodChem

53:2118–2124

M,PopielarzMJ,CallejonRM,MoralesML,TroncosoAM,

PetersenMA,Toldam-AndersenTB(2010)JChromatogrA1217:

4422–4429

ovB,AmigoJM,BakS,EngelsenSB(2012)JchromatogrA

1266:84–94

ingH(1933)JEducPsychol24:417–441

doolHandKratzPD(1963)JournalofChromatography11:

463–471

,BroR(2008)AnalBioanalChem390:281–285

,KotoM,KomatsuH,IgoshiK,KobayashiH,ItoY,Nohara

T(2004)FoodSciTechnolRes10:56–59

lA,PaczkowskiC,KoivuniemiH,HuttunenS(2012)JAgric

FoodChem60:4994–5002