Contents lists available atScienceDirect
Journal of Pharmaceutical and Biomedical Analysis
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / j p b a
Comprehensive phospholipid and sphingomyelin profiling of different brain regions in mouse model of anxiety disorder using online
two-dimensional (HILIC/RP)-LC/MS method
Róbert Berkecz
a,∗, Ferenc Tömösi
a, Tímea Körmöczi
a, Viktor Szegedi
b, János Horváth
b, Tamás Janáky
aaDepartmentofMedicalChemistry,FacultyofMedicine,UniversityofSzegedDómtér8,H-6720,Szeged,Hungary
bDepartmentofPhysiology,AnatomyandNeuroscience,UniversityofSzeged,Középfasor52,H-6726,Szeged,Hungary
a r t i c l e i n f o
Articlehistory:
Received11September2017
Receivedinrevisedform29October2017 Accepted30October2017
Availableonline8November2017
Keywords:
Phospholipid Sphingomyelin Brain 2D-LC/MS Lipidomics Anxietydisorder
a b s t r a c t
Anovelonlinesystemincludingtwo-dimensionalliquidchromatographycoupledtohigh-resolution massspectrometry(2D-LC/MS)wasdevelopedandappliedforcomprehensivephospholipid(PL)and sphingomyelin(SM)profilingofdorsalhippocampus(DHPC),ventral(VHPC)andprefrontalcortex(PFC) brainregionsinamousemodelofanxietydisorder.Inthefirstdimension,lipidclassesweredistinguished byhydrophilicinteractionliquidchromatography(HILIC),whiletheseconddimensionalseparationof individualPLandSMspecieswasachievedbyreversed-phase(RP)chromatography.Fortheenrichment oflipidspeciesindilutedHILICeffluent,twoRPtrappingcolumnswereusedseparately.Thedeveloped fully-automated2Dmethodallowedthequantitativeanalysisofover150endogenousPLandSMspecies inmousebrainregionswithin40min.Thedevelopedmethodwasappliedinapilotstudy,whichaimed tofindalterationofPLandSMcompositioninamousemodelofanxietydisorder.Inthecaseof37PL andSMspecies,significantdifferenceswereobservedbetweenhighanxiety-relatedbehavior(AX)and lowanxiety-relatedbehavior(nAX)mice.Inmicehavingelevatedanxiety,themosttypicaltrendwas thedownregulationofPLspecies,inparticular,inVHPC.
©2017ElsevierB.V.Allrightsreserved.
1. Introduction
Non-targetedanalysisofbiomoleculesisanimportanttrendin thefieldofomicssuchasproteomics,genomics,transcriptomics andmetabolomics.Lipidomics,anemergingfieldofmetabolomics, aims for the comprehensive analysis of lipids in cells, tissues andbiologicalfluidsandmonitorsthelipidresponsestovarious externalor internal effects/events [1–4]. Chemically,eight lipid
Abbreviations: 1D,one-dimensional;2D-LC/MS,two-dimensionalliquidchro- matography coupled to high resolution mass spectrometry; AGC, automatic gaincontrol; AX, anxiety-related behavior; C18, octadecylsilyl; DHPC, dorsal hippocampus;FA,fattyacyl;HILIC,hydrophilicinteractionliquidchromatogra- phy;IS,internalstandard;LPL,lysophospholipid;ME, matrixeffect;nAX,low anxiety-relatedbehavior;NP,normalphase;OPLS-DA,orthogonalpartialleast squarediscriminant analysis; PA, phosphatidic acid; PC, phosphatidylcholine;
PE,phosphatidylethanolamine;PFC,prefrontalcortex;PG,phosphatidylglycerol;
PI,phosphatidylinositol; PL, phospholipid;PS, phosphatidylserine; RE,extrac- tionrecovery;RP,reversed-phase;RS,chromatographicresolution;SEM,standard errors;SM,sphingomyelin;VHPC,ventralhippocampus.
∗Correspondingauthor.
E-mailaddress:berkecz.robert@med.u-szeged.hu(R.Berkecz).
categoriesareknown,namely,fattyacyls(FA),glycerolipids,sac- charolipids,polyketides,sterollipids,prenollipids,sphingolipids andPLs.AccordingtothepolarheadgroupofPLs,phosphatidyl- choline(PC),phosphatidylethanolamine(PE),phosphatidylinositol (PI),phosphatidylserine(PS),phosphatidylglycerol(PG)andphos- phatidicacid(PA)classesaredistinguished[5,6].
Nowadays,thecouplingofLCtoMSisoneofthemostpower- fulandwidespreadtechniquestoanalyzelipidmolecularspecies in complex samples. The main advantages of the use of this hyphenatedtechnique,comparedtodirectinfusioncalled“shot- gun”methods,areitsabilitytodistinguishisomers(isobars)and obtainhighersensitivityoflow-abundancelipidsowingtoreduced ionsuppressioneffect[7–9].InLC,lipidsareusuallyseparatedby RP,HILICornormalphase(NP)methods.InRP-LC,theretention behaviorof lipidsareincorrelationwiththeequivalentcarbon number,whileinthecaseofNP-LCandHILIC,theretentionoflipid classesdependsonthehydrophilicpropertiesofthepolarhead group[4,10,11].ThesolventsystemusedinNP-LCusuallyprovides lowionizationefficiencyinMSdetection,whileRP-LCandHILIC mobilephasesareMScompatibleandresultingoodionization[12].
https://doi.org/10.1016/j.jpba.2017.10.043 0731-7085/©2017ElsevierB.V.Allrightsreserved.
Lipidomicsdealswithlargesamplecomplexity;therefore,one- dimensional(1D)chromatographicseparationcanprovidelimited selectivityforlipidspeciesinanychromatographicmoderesulting indifficultiesinidentificationandquantification[13].Thecombi- nationofdifferentLCmodescouldprovideanumberofpossibilities toimproveseparationoflipidmolecularspeciesthroughenhanced chromatographicresolutionandhigher peakcapacity.Twocon- nectedorthogonalLCmodes,suchasNP-LCorHILICorsilver-ion chromatography(fornonpolarlipids)withRP-LC,eitheronlineor offline,havealreadybeenusedinlipidomics.[13–21].Themain benefitofoffline2Dtechniquesisthatthechromatographiccon- ditionsin bothdimensionscanbecompletelyoptimized,which helpstoimprovetheseparationoflipidspecies.Ontheotherhand, itisnotautomatedandadditionalsamplepreparationsteps,such asfractioncollection,evaporationofmobilephase,reconstitution andreruninseconddimension,makethismethodtime-consuming andlabor-intensiveandmayresultindegradationofthesample.
Online2Dtechniques,inturn,giveanopportunityforthedevel- opmentofcompletelyautomatedmeasurementswithalowrisk ofsamplelossanddegradation.However,thistechniqueprovides compromisingchromatographicresolutioninthefirstand/orsec- onddimensionduetothesynchronizationofbothdimensionsand requiresahighlevelofinstrumentation[14,16,19].
Althoughanxietydisorders are among themostwidespread affectivediseases,theirpathogenesisisstillpoorlyunderstood.The generationandregulationofthesustainedanxiousmoodareacom- plexprocess,in whichseveralbrainregions(alsoreferred toas thefearcircuitry)areinvolvedinthegenerationoffear.Keyareas ofthebrainarethemedialPFCandVHPC,sharingamonosynap- ticconnection,whichfunctionallyinteractsduringinnateanxiety tasks[22–25].It istheVHPC,whichis alsointerconnectedwith theamygdalaandtheentorhinalcortexandmediatestheeffectof glucocorticoidsonanxietyinthebrain[26–29].Theuseofanimal modelsofanxietydisordermayhelptounderstandhumandis- orders.Twomousestrainswithextremesintheiranxiety-related phenotypehavebeenestablishedbyusinganintra-strainanda selectivebidirectionalinbreedingapproachthatledtoanaccumu- lationofgeneticmaterialassociatedwiththerespectiveanxiety phenotype.Ourgrouphaspreviouslystudiedneurophysiological andbrainproteomedifferencesinAXandnAXmousestrainsand nowwepresentourlipidomicsresultsonthisanimalmodel[30,31].
Themain goalofthis studywasthedevelopmentof a new, online2D-LC/MS methodforthecharacterizationof thePL and SMcompositionof differentbrainregionsofnAXand AXmice.
Comprehensivelipidomicsanalysiswasachievedbyseparationof lipidclassesbyHILICchromatography(firstdimension)followedby reversed-phasechromatography(seconddimension)onaUHPLC columnpackedwithfullyporoussub–2mC18particlestosep- aratelipidsspecies.Theidentificationofbrainpolarlipidspecies wasbasedontheretentiontimesinbothdimensions,theaccurate m/zvaluesandisotopepatternsofthedetectedions([M−H]−or [M+HCOO]−,dependingonPLclasses).QuantificationofPLspecies identifiedwascarriedoutbyusinginternalstandards.
2. Materialsandmethods 2.1. Animalsandtissuesamples
InbredmousestrainshavingeitherAXor nAXwerebred in ouranimalfacility.ThesestrainswereoriginallydevelopedatEGIS PharmaceuticalsPLC(Budapest,Hungary)bybidirectionalinbreed- ingbasedonanticipatoryanxiety[30].2.5–3-month-oldmalemice werehousedindividuallyunderalight/dark12-hcycle(lightson at08:00)at24±1◦Ctemperaturegivingadlibitumaccesstofood andwater.Fortheexperiments,80–85thgenerationswereused.
Themiceweresacrificed,theirbrainswererapidlyremovedand PFC,VHPCandDHPCregionsweredissected.Thewetsampleswere weighed,snap-frozeninliquidnitrogenandstoredat−80◦Cbefore homogenization.ThestudywasincompliancewithEUdirective 2010/63/EUandwasapprovedbytheregionalStationforAnimal HealthandFoodControlunderProjectLicenseXXXI/2012.
2.2. Chemicalsandstandards
1-Heptadecanoyl-2-hydroxy-sn-glycero-3-phosphate (sodium salt) (LPA 17:0), 1,2-dimyristoyl-sn-glycero-3-phosphate (sodium salt) (PA 14:0/14:0), N-lauroyl-d-erythro- sphingosylphosphorylcholine (SM d18:1/12:0), 1-myristoyl-2-hydroxy-sn-glycero-3-phospho-(1-rac-glycerol) (sodium salt) (LPG 14:0), 1,2-diheptadecanoyl-sn-glycero- 3-phospho-(1-rac-glycerol) (sodium salt) (PG 17:0/17:0), 1-nonadecanoyl-2-hydroxy-sn-glycero-3-phosphocholine (LPC 19:0), 1,2-dimyristoyl-sn-glycero-3-phosphocholine (PC 14:0/14:0), 1,2-dimyristoyl-sn-glycero-3-phospho-l-serine (sodium salt) (PS 14:0/14:0), 1-myristoyl-2-hydroxy- sn-glycero-3-phosphoethanolamine (LPE 14:0), and 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine (PE 14:0/14:0)were purchased fromAvanti Polar Lipids (Alabaster, USA)and usedasinternal standards(IS).Water,methanol,ace- tonitrile,ammonium formate (all LC–MS grade), n-hexane, and chloroform (HPLC grade) were purchased from VWR (Radnor, USA).LC–MS-grade2-propanolwaspurchasedfromFluka(Buchs, Switzerland) and acetone (GC–MS grade) was from MERCK (Darmstadt,Germany).
2.3. Samplepreparation
Theweighedbrainsampleswereplacedinto1.7mLmicrocen- trifugetubesand anappropriate volumeofammoniumformate buffer(50mM)wereaddedinordertoobtain5g/Lconcentra- tionofeachhomogenate.Thesampleswereindividuallysonicated withaBioLogicsModel150VTultrasonichomogenizer(BioLogics Inc,Manassas,VA,USA)for1minusingfullpowersettingwitha 50%pulseonthemicro-tipprobe.
Extraction of lipids from brain homogenate was performed according to a slightly modified Folch procedure [32]. Prior to extraction, 10L brain homogenate was spiked with 10L lipidstandard mixture(100pmol/LPC14:0/14:0,100pmol/L LPC 19:0, 50pmol/L PE 14:0/14:0, 30pmol/L LPE 14:0, 5pmol/LPG 17:0/17:0, 5pmol/L LPG 14:0, 50pmol/L PA 14:0/14:0, 50pmol/L LPA 17:0, 50pmol/L PS 14:0/14:0 and 125pmol/LSMd18:1/12).Aftervortexmixing,5Lofbutylated hydroxytoluene (2mg/mL in ethanol) and 450L of chloro- form/methanol(2:1,v/v)wereadded,followedbyvortexmixing.
Themixturewasshakenfor15minatroomtemperature.After theadditionof115Lofammoniumformate(50mM),thesample wasvortexedfor20s.Upon5minofincubationatroomtempera- ture,thesamplewascentrifugedat1000gfor10min200Lofthe lowerphasewascollected,andtheupperphasewasre-extracted with200Lofchloroform.Aftercentrifugation,300Lofthelower phasewascombinedwiththefirstportionoforganicphaseand driedbyanitrogenstreamatambienttemperature.For2D-LC/MS measurements,the driedextracts werereconstitutedin 100L chloroform/methanol(2:1,v/v)mixture.
2.4. 2D-LC/MSconditions
2D-LC/MSanalysiswasperformedbyusingaWatersAcquityI- ClassUPLCTMsystem(Waters,Manchester,UK),equippedwithtwo binarysolventmanagers,anauto-samplerandacolumnmanager withtwosix-port,two-positionautomaticswitchingvalves.The
Fig.1.Flowschemeoftheestablished2D-LC/MSsysteminbothvalvepositions:Position1andPosition2.
eluentofthefirst-dimensionalHILICcolumnwasdilutedusinga HITACHIL-7100pump(Hitachi,Tokyo,Japan)andahigh-pressure staticmixingtee(IDEX,OakHarbor,WA,USA).Thefirstandsecond
dimensionalanalyticalcolumnswerethermostatedinthecolumn manageroftheUPLCsystem,whilefortheenrichmentcolumns,a L-7350LaChromcolumnoven(Merck,Darmstadt,Germany)was
used.The ultrahigh-performance liquidchromatography (UPLC) systemwascoupledtoThermoScientificQExactivePlushybrid quadrupole-Orbitrap(ThermoFisherScientific,Waltham,MA,USA) massspectrometer.Theexperimentalconfigurationofouronline 2D-LC/MSsystemisillustratedinFig.1.
In the 2D-LC/MS system, the first- and second-dimension columnswereconnectedthroughtwosix-portvalvesandtwoLuna C18(20×2.0mm,5m,100Å,Phenomenex)enrichmentcolumns (Fig. 1). By thescheduled switching of the six-portvalves,the dilutedmobilephasefromtheHILICcolumninfirstdimensionwas trappedonC18enrichmentcolumn,whilethetrappedcompounds ontheotherC18enrichmentcolumnwereanalyzedinthesec- onddimensioninsync(Table1).Topreventtheelutionoflipid speciesfromtheenrichmentcolumnsduringthetrappingprocess, thediluterpumpdelivered5mMammoniumformateeluentata flowrateof0.6mL/minintothestaticmixingtee,whichwascon- nectedaftertheHILICcolumnandinfrontofthefirstsix-portvalve.
Thetrapcolumnsweremaintainedat50◦C.
Fig.1demonstrates theschematicrepresentation ofour2D- LC/MSsystemin“Position1”arrangement,wherethe“RPtrap1”
columnwasfirstusedtoenrichanalytesfromthedilutedefflu- entoftheHILICcolumn.Thesystemwaskeptinthisconfiguration for8min.Thenswitchingvalveswereturnedinto“Position2”as showninFig.1.ThedilutedmobilephasefromtheHILICcolumn wasthentrappedonthe“RPtrap2”columnand,atthesametime, theanalytestrappedonthe“RPtrap1”columnwerewashedonto andseparatedonthesecond-dimensionalRPanalyticalcolumn.
ThecompletescheduleofeventtimesisdetailedinTable1.The numberofvalveeventswasodd(5),whichwouldhaveresultedin thesameinitialvalvepositionsinconsecutiveanalysis.Namely, inthefirstHILICeffluenttrapping section(0–8min),theeluted analytesofthegivenmeasurementwouldhaveenrichedonthe sametrapcolumnthatwasusedinlastsectionofpreviousmea- surement.Therefore,topreventtheundesired“doubletrapping”
ontheenrichmentcolumns,twomethodswiththesameLCand MSparametershadtobeestablished,whichdifferedonlyinthe order ofvalve positions (Table1). Thetwo methods had tobe strictlyalternatedinsamplesequence.Componentsthattrapped in the washing and re-equilibration steps of theHILIC column (32–40min)weredetectedinthefirstRP-LCrunofthefollowing samplebutwereignoredindataevaluation(Fig.2).
Lipid classes were separated in the first dimension after injecting 10L of sample/standard on a Kinetex HILIC column (150×2.1mm, 2.6m, 100Å, Phenomenex) using programmed gradientofeluentA(50mMammoniumformatesolution)andelu- entB(acetone)(Table1).Theflowrateofthemobilephaseand thetemperaturewerekept,respectively,at0.4mL/minand50◦C duringtheanalysis.Theinjectorneedlewaswashedwithhexane- 2-propanol-water(2:2:0.1,v/v/v)mixtureaftereachinjection.
Intheseconddimension,eluentAandBwerewater/acetonitrile (50:50,v/v)andwater/acetone(5:95,v/v),respectively,bothcon- taining5mMammoniumformate.Themobilephasewashedthe substancestrappedona givenC18enrichmentcolumnontoan AcquityUPLC BEHC18analytical column (2.1×50mm, 1.7m, 130Å,Waters)ataflowrateof0.4mL/min.Thecolumnwasmain- tainedat 50◦C. Thegradient programsofboth dimensionalare detailedinTable1.
Themassspectrometerwasoperatedinthenegative-ionmode usingaheatedESIsourcewiththefollowingconditions:capillary temperature250◦C,S-LensRFlevel50,sprayvoltage2.5kV,sheath gasflow45,sweepgasflow2andauxiliarygasflow10,andfull scanwithamassrangeof100–1000andaresolutionof35,000.
Theautomaticgaincontrol(AGC)settingwasdefinedas3×106 chargesandthemaximuminjectiontimewassetto100ms.
TheLCsystemwascontrolledbyMassLynx4.1SCN901(Waters, Milford,MA,USA).ThecontrolofMSsystemandMSdataacquisi-
tionwereconductedbytheXcaliburTM4.0software(ThermoFisher Scientific,Waltham,MA).
2.5. Dataprocessing
Thealignmentofretentiontimes,peakpicking,deconvolution, determinationofpeakarea,aswellaspreliminaryidentification usingLipidBlastdatabasewereperformedfromrawdatabyProge- nesisQI(NonlinearDynamics,Newcastle,UnitedKingdom)[33].
Then, the processed data including peak area, m/z value, and retention timewereimported intoMicrosoftExcelsoftwarefor normalizationofthepeakareas,whichwasbasedoncalculation oftheanalyte/ISpeakarearatios.Identificationoflipidspecieswas accomplishedviaaccuratemass(<3ppm)databasesearchingof LIPIDMAPSandourhomemadedatabasebuiltonourownmea- surements,literaturesourcesandthepredictionofretentiontimes basedonretentionbehaviorofPLclassesandPLspeciesinboth dimensions[6,34–39].
Multivariatedataanalysisofdataincludingthenormalizedpeak areaandtheabbreviationoftheobtainedidentifiedlipidswasper- formedbySIMCAsoftware14.1(Umetrics,Umeå, Sweden).The obtainedresultsoftheorthogonalpartialleastsquarediscriminant analysis(OPLS-DA)andthecorrespondingS-plotsallowedthepre- dictionofsignificantfeatures(potentialbiomarkers)ofwhichthe alterationwasinrelationtotheAXmousemodel.Boxplotsand columndiagramswithstandarderrors(SEM)wereusedtoillus- tratethedataandthedifferencesbetweenthenAXandAXmice.
Thesewereassessedwitht-testusingGraphPadPrism5statistical software(GraphPadSoftware,Inc.,LaJolla,CA,USA).
3. Resultsanddiscussion
3.1. CouplingofHILICandRP-LCin2D/LC
Severalchromatographicparameters,suchascolumnlength, mobile phase composition, types and concentration of mobile phaseadditives,columntemperature,gradientsteepnessandflow ratewereinvestigatedbothinthefirstHILICandthesecondRP-LC dimensions.
During optimization of HILIC, 100mm and 150mm length narrow-borecolumns(2.1mm)withporousshellparticleswere compared, and the latterwas selectedfor further optimization to obtain betterseparation of PL classes. To achieve this goal, methanol,acetonitrileandacetoneweretestedastheorganiccom- ponentofthemobilephase. Theapplicationofacetoneresulted in thehighest retentionand improvedchromatographic resolu- tion,in particular,for PCsand SMs,thusit wasselectedin the final method.In HILICmode, theconcentrationofwater in the mobilephasehasasignificantinfluenceontheretentionmech- anismandthereproducibilityofretentiontimebyestablishingof awater-enrichedlayerofsemi-stagnanteluentonthestationary phase[40,41].Theminimumwatercontentofthemobilephase wasdeterminedtobe3%(v/v)forthereproducibleretentionof lipids.Thechromatographicbehavioroftheanalytesisinfluenced bythetypeandconcentrationofthemobilephasemodifiers.The volatileandMS-compatibleammoniumformatewasselectedand addedat50mMconcentrationtowaterforpreparationofeluent A.Inordertoimprovethetrappingefficiency,theflowrateinfirst dimensionalseparationandthediluterpumphadtobeharmo- nized.Usually,theapplicationoflowflowrateinthefirstdimension promotestrappingefficiencyoftheanalytes.However,itresultedin seriouspeakbroadeninginourcase,thusasanoptimumflowrate of0.4mL/minwasselected.Lipidprofilingofmousebrainextract withoptimizedHILICchromatographyisshowninFig.2(A).11PL classes,specifically,PG,PI,LPG,LPI,PE,PS,LPE,PC,PA,SMandLPC,
Table1
Thedetailedgradientprogramandvalveeventsoffinal2D-LC/MSmethod.
Firstdimension(HILIC) Valveposition Seconddimension(RP-LC)
Lipidclasses Retentiontime[min] SolventB[%] Method1 Method2 Lipidspecies Retentiontime[min] SolventB[%]
nonpolarlipids 0.0 97 1 2 0.0 30
FA 2.0 95
LPG 5.0 95
PG 5.1 30
8.0 30
LPI 2 1 nonpolarlipids 8.0 30
PI FA 10.0 95
PE LPG 13.0 95
PG 13.1 30
16.0 30
LPE 1 2 LPI 16.0 30
PC PI 18.0 95
PS PE 21.0 95
21.1 30
24.0 30
PC 2 1 LPE 24.0 30
PS PC 26.0 95
SM PS 29.0 95
LPC 30.0 82 29.1 30
PA 31.0 50 32.0 30
1 2 PC 32.0 30
35.0 50 PS 34.0 95
35.5 97 SM 37.0 95
LPC 37.1 30
40.0 97 PA 40.0 30
weredistinguishedwithin40min.Identificationoftheseclasses wasperformedina separate1D experimentby theapplication ofthemethoddetailedabove.Foranalysisoflipidspeciesinthe seconddimension,theeluate oftheHILICchromatographywas dividedintofivefractionswith8minofretentiontimewindows, asdemonstratedinFig.2(A).
Bytheuseofreversed-phaseC18stationaryphase,theretention andseparationmechanismoflipidsarebasedonthelengthsoffatty acyl(oralkyl)chainsandthenumberandthepositionofdouble bonds[4],whichallowthediscriminationoflipidspecieswithin thesameclass.Fortheseparationoflipidspeciesinseconddimen- sion,anAcquityUPLCBEHC18micro-borecolumnwithaparticle sizeof1.7mwasselected.Inordertoachieveappropriatesepara- tionoflipidspecieswithinlipidclassesatpropersystempressure, flowrateandtogethighionizationefficiency,acarefulselectionof themobilephasewasalsoneeded.Theuseofwater/acetonitrile (50:50,v/v) asAeluent andwater/acetone (5:95,v/v) asBelu- ent(bothcontaining5mMammoniumformate)provedtobea goodcompromiseinpointofviewofsystempressure,retention, selectivityandionizationefficiencyoftheanalytes.Forobtaining appropriateretentionandchromatographicresolutionwithinthe 8-minruntimeincludingwashingandre-equilibrationofthecol- umn,thetemperatureofC18column,theflowrateandthegradient steepnesswereadjusted.Incaseofalllipidfractions,sameRPgra- dientprogramwasusedinordertoproducesameretentiontime forthoselipidspecieswhichwerepresentintwoHILICfractions (e.g.PCs).Inthisway,thesamechromatographicandmassspec- trometricconditionswereguaranteedforthesespecies,whichisa requirementforappropriatequantitativeanalysiscombiningpeak specieselutingintwoconsecutivefractions.
The collection/trap of analytes from the eluate of the first- dimensionalcolumnina2D-LCsystemisusuallyperformedbythe applicationofloop(s)ortrapcolumn(s)[13–21].Fortheenrichment oftheanalytesinaloop,themobilephaseshouldbeevaporated bytheuseofvacuumand temperaturemodulation.In thecase of trapcolumn, especially in highly orthogonalseparation (e.g.
NP×RPorHILIC×RP),thesolventstrengthoftheeffluentoffirst- dimensionalcolumnshouldbeadjustedtoretainanalytesonthe enrichmentcolumn.UsingHILIC×RP-LCcombination,wedidnot havetoreckononthesolventincompatibilityproblem,whichis knowninNP-LC×RP-LCmethods.Acetone,appliedaseluentBin HILICmode,isastrongeluentinRPmode;therefore,ithadtobe dilutedwithasolventwithaverylowelutingpowerinRPmode.An aqueoussolutionofammoniumformate(5mM)waschosenforthis purpose.TheflowrateratiooftheHILICmobilephaseandthedilut- ingsolventwascriticalfortheappropriateretentionoflipidspecies onthetrapcolumns,especiallyatthebeginningofanalysis,when theeffluentofHILICcolumnhadhighorganicconcentration.The effectofflowrateofthedilutingeluentontheretentionwasinves- tigatedintherange0.2–1mL/min.Aflowrateof0.6mL/minfor thedilutingeluentwasthebestcompromiseconsideringsufficient trappingoflipidspeciesandpropersystempressure.
3.2. Evaluationofextractionrecoveryandmatrixeffect
Priortoanalysis,theenrichmentofPLsthroughremovalofthe polarmatrixisacrucialsteptoobtainreliableresults.Thedetermi- nationofextractionrecovery(RE)andmatrixeffect(ME)couldhelp todecidewhetherananalyticalmethodisfeasibleforquantitative analysis.TheREandMEofthedevelopedmethodweredetermined byusing10lipidclassstandardsinaccordancewiththeprocedure ofMatuszewskietal.andCappielloetal.[42,43].Accordingtodata onFig.S1,theREandMEofstandardlipidspecieswererepro- ducibleandcomparable.ThehighmeanvaluesofRE(above85%) andrepeatabilitiesforalllipidstandardssuggestthatourextraction methodiswellestablished.
The matrix effect values ranged from63.0 to 88.0%, which, becauseofthecomplexityofbraintissue,isanacceptablerange.
MEwasbelow100%forallstandardsexaminedindicatingthelack ofionizationenhancementduringMSdetection.
Fig.2.BasepeakchromatogramofPLsinVHPCofnAXmousewasdetectedbynegativemodewith1D-HILICmethod(A)and2D-LC/MSmethod(B).Orangelinesindicatethe gradientLCprofilein1Dand2Dmeasurements.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
3.3. IdentificationandquantificationofPLsinmousebrain regions
As shown in Fig. 2, HILIC wasemployed to differentiate PL classesinthefirst-dimensionalrun.Theeffluentobtainedinthe firstdimension wasdividedinto fivefractions in orderto sep- arateindividualPLspecies withinclassesaccordingtothefatty acyl chain lengths and the number of double bonds using RP mode.Theenhancedchromatographicresolutionandhigherpeak capacitiesobtainedbythe2Dmethodisillustratedthroughthe exampleof PE36:1,PE36:2,PE 36:3,PE 34:4andPE 35:5on Fig.3,bypresentingextractedionchromatograms(negativeions) ofHILIC/MS(A)and2D-LC/MS(B).Theadvantageof2D-LC/MSis demonstratedonFig.3(B)showingbaselineseparationoftrapped phosphatidylethanolamine species. In lysophospholipids (LPLs), thefattyacylchainsareattachedateitherthesn-1orsn-2position oftheglycerolbackbone[4].Inseveralcases,separationofthese isomersinLPI,LPEandLPCclassescanbeobserved(TableS1).
TableS1lists151endogenousPLsandSMs(3LPG,7PG,5LPI, 13PI,35PE,14LPE,37PC,11PS,8SM,14LPCand4PA),which
wereidentifiedbyusingourmethodinmousebrainregions.For quantitativeanalysis oftheidentified lipid molecules,thepeak areaofeachindividuallipidspecieswascorrectedbythecorre- spondinginternalstandard.RelevantchromatographicandMSdata alongwiththerelativeabundanceofparticularlipidspecieswithin classesdeterminedindifferentbrainregions(DHPC,VHPCandPFC) arepresentedinTableS1.Ourresultssupportthefindingthatner- voustissuescontainhighamountsofplasmanylandplasmenylPLs [44].Fig.4illustratesthedistributionofbrainPLspeciescontain- ingacyl,alkyletherandvinylethersidechainswithinlipidclasses.
PlasmalogensandetheranalogsweredecisivelypresentinPEand PCclasses.
3.4. ComparinglipidprofilesinthebrainofnAXandAXmice
Seventeen micefromtwo strainswereinvolved in thepilot study,9withlowanxiety-relatedbehaviorand8withhighanxiety- related behavior. Normalized peak areas of all identified lipid speciesinDHPC,VHPCandPFCofnAXandAXmicewerestatis- ticallyevaluatedusingtheSIMCA(multivariatedataanalysis)and
Fig.3. ExtractedionchromatogramsofselectedPEspeciesdetectedbynegativemodewith1D-HILICmethod(A)and2D-LC/MSmethodwithobtainedchromatographic resolutions(RS)(B)inVHPC.RS=2×(tR2-tR1)/(w1+w2).
Fig.4. IdentifiedPLspeciesinmousebrain.
GraphPadPrism5(t-test)software.Atfirst,byusingt-test,statis- ticalsignificanceswerecalculatedforthenormalizedpeakareas oflipidclasses(sumofnormalizedpeakareasoflipidspeciesin agivenclass)innAXandAXbrainregions.Significantalterations wereobservedinfourlipidclassesandtwobrainregions.Asshown inFig.S2(A),thenormalizedpeakareaofthePEclasssignificantly decreasedinVHPCoftheAXgroup.Asimilartrendwasobserved forthePSclassinPFC;however,inthecaseofLPGandPA,therewas anincreaseinthenormalizedpeakareasintheAXgroupcompared tothoseofthenAXgroup(Fig.S2(B)).
Thecomparisonofnormalizedpeakareasof151individuallipid moleculesinthreebrainregionsbetweennAXandAXgroupswas
carriedoutusingmultivariatedataanalysis.DataonFig.S3(A,C,E) clearlydemonstrate theproperseparationof nAXand AXmice throughtheobtainedsupervisedOPLS-DAscoreplots.
TherelatedS-plots(Fig.S3(B,D,F))helpedtoselectlipidspecies, whicharepresent insignificantlydifferentconcentrationinthe twogroups.Lipidspecies,whichdidnotdiffersignificantly(dys- regulated)innAXandAXmice,arelabeledwithgreen.Lipidspecies downregulatedsignificantlyarelocatedonthelowerleftsectionof S-plot,whiletheupregulatedonesareontheupperrightsection (reddots).S-plotsofOPLS-DAmethodsunveil37lipidspecieswith significantdifferences(P<0.05)betweenthenAXandAXgroups (Fig.S3(B,D,F)).
Table2
IdentifiedPLspecieshavingsignificantlydifferentconcentrationinnAXandAX mice.
Mousebrainregion Phospholipidspecies Probability Foldchange
DHPC PE40:5 <0.001 0.27
DHPC PEO-36:3, 0.011 0.60
P-36:2
DHPC PEO-38:3, 0.017 0.62
P-38:2
DHPC PE34:1 0.020 0.84
DHPC PEO-40:7, 0.023 0.75
P-40:6
DHPC SM36:1 0.041 1.23
VHPC PE40:5 0.001 0.16
VHPC PC38:2 0.002 0.56
VHPC PCO-34:1, 0.004 0.58
P-34:0
VHPC PC40:2 0.007 0.34
VHPC PC40:4 0.009 0.65
VHPC PCO-36:2, 0.010 0.16
P-36:1
VHPC PEO-40:7, 0.013 0.77
P-40:6
VHPC PE34:1 0.015 0.78
VHPC PC36:3 0.015 0.30
VHPC PEO-40:8, 0.016 0.80
P-40:7
VHPC PC38:1 0.020 0.61
VHPC PE38:4 0.021 0.79
VHPC PI36:5 0.029 0.59
VHPC PC40:6 0.030 0.73
VHPC PE40:6 0.032 0.84
VHPC PG32:0 0.037 0.76
VHPC PC38:7, 0.037 0.50
O-38:0
VHPC PC34:2 0.040 0.76
VHPC PEO-38:5, 0.040 0.79
P-38:4
VHPC PCO-32:1, 0.040 0.65
P-32:0
PFC PE40:5 <0.001 3.84
PFC PE38:0 0.003 0.24
PFC PA40:6 0.006 1.59
PFC PS36:1 0.007 0.40
PFC PA36:2 0.019 2.21
PFC PC42:9 0.019 2.56
PFC PS40:6 0.026 0.71
PFC PEO-38:3, 0.031 0.61
P-38:2
PFC LPE20:1 0.031 0.34
(sn-2)
PFC PG34:2 0.039 1.79
PFC LPG18:0 0.042 1.74
PLsandSMscompositionofDHPCwerecomparedinnAXand AXgroups,too.TheresultsshowedthatonlySM36:1wasupreg- ulated,while5PEswerefoundtobedownregulated.Inparticular, PE40:5withthelowestPvalueshadaconcentrationinAXmice almost4timeslower(Table2).Interestingly,SM36:1wasthemost abundantlipidspeciescomparedwithotherSMsinDHPC,whilePE 40:5waspresentinarelativelylowconcentration(TableS1).
SignificantlyalteratedlipidprofileswereobservedinVHPCof AXmice:20lipidspeciesfromPE,PC,PIandPGclasseswerefound tobedownregulated(inrangeof0.16-0.84foldchange)(Table2).
Fig.5(A,B)showthemeanvalueswithSEMofnormalizedpeak areasofsignificantly alteredpolarlipidspecies fromDHPCand VHPCinthenAXandAXgroups.IncontrastwithDHPC,onlysig- nificantdownregulationof VHPC lipid species wasobserved in micehavingelevatedanxiety.Again,PE40:5showedthehigh- estconcentrationdifferences inthetwogroups.Similarly, large downregulationswerefoundforPCO-36:2;P-36:1(6.25×),PC36:3 (3.33×)andPC40:2(2.9×)species(Table2).
ThecomparisonofnormalizedpeakareasofpolarlipidsinPFC ofnAXandAXmicerevealed11statisticallysignificantdifferences
Fig.5.NormalizedpeakareaofthesignificantlydifferentPLsofnAX(bluecolumns) andAXgroups(orangecolumns)inthethreebrainregions:(A)DHPC,(B)VHPCand (C)PFC.*P≤0.05;**P≤0.01;***P≤0.001.(Forinterpretationofthereferencesto colourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
Fig.6.DistributionofPLsthatfoundtobeup-ordownregulatedinDHPC,VHPC andPFCofAXmice.
inLPE,PE,PC,PS,PG,LPGandPAclasses(Table2,Fig.5(C)).PE38:0, PS36:1,PS40:6,LPE20:1andPEO-38-3;PEP-38:2lipidspecies weredownregulatedinPFCoftheAXgroup(Fig.S3(F)).Incontrast toDHPCandVHPC,PE40:5washighlyupregulatedinPCFofthe AXmice.Concerningthevaluesofup-anddownregulationoflipid speciesintheAXgroup,higherthantwo-foldincreaseswerefound forPE40:5,PA36:2andPC42:9.ForPE38:0,LPE20:1(sn-2)and PS36:1,incontrast,higherthantwo-folddecreaseswereobserved.
Table2 revealsthatthemajorityofPLspecies havingstatis- ticallydifferentconcentrationsinthenAXandAXgroupscontain unsaturatedfattyacids.Furthermore,plasmanylandplasmenylPLs exhibitingsignificantalterationwerealsopresentinhighpercent- age.
Finally,Fig.6illustratesthedistributionofsignificantlyaltered PLspeciesinDHPC,VHPCandPFCbrainregionsoftheAXmice.Only PE40:5showedsignificantalterationinalloftheinvestigatedbrain regions.
4. Conclusion
This paper describes the development of a comprehensive online2D-LC/MSmethodfortheanalysisofPLandSMspeciesin mousebrainregions.Inournovelsystem,thecouplingofHILICand RP-LCwasdesignedbyusingtwoRPtrapcolumns,thusenabledthe enrichmentofHILICeffluentonthetrapcolumnandtheseparation ofthetrappedlipidsintheseconddimensionsynchronously.Inthe firstdimension,theHILICseparationoflipidclasseswasdivided intofivefractionsand theanalysisoftheindividualPLand SM specieswithinfractionswasperformedbyfourRPruns.Thefinal methodprovidedthequantificationofmorethan150PLandSM speciesintheDHPC,VHPCandPFCbrainregionswithin40min run-time.Withtheestablishedmethod,thedifferencesofPLcom- positioninbrainregionsofnAXandAXmicewerecompared.To ourbestknowledge,thisisthefirsttimethatPLandSMalterationof differentbrainregionsinmousemodelofanxietywasreported.Our studyrevealedthat37PLandSMspecieshadsignificantlyaltered concentrationintheAXgroup:20werefoundinVHPC,6inDHPC and11inPFC.Itisworthnotingthatsignificantunidirectionalalter- ationoftheconcentrationofPLspecieswasfoundinVHPCofthe AXgroup.Overall,thedevelopedfullyautomated2Dmethodwas successfullyappliedinprofilingbrainPLsandSMsinmousemodel ofanxietydisorder.
Acknowledgement
RóbertBerkecz thanksforthefinancial supportof theJános BolyaiResearchScholarshipoftheHungarianAcademyofSciences.
ThisresearchwassupportedbytheEU-fundedHungariangrant EFOP-3.6.1-16-2016-00008.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttps://doi.org/10.1016/j.jpba.2017.10.043.
References
[1]J.VanderGreef,S.Martin,P.Juhasz,A.Adourian,T.Plasterer,E.R.Verheij,R.N.
McBurney,Theartandpracticeofsystemsbiologyinmedicine:mapping patternsofrelationships,J.ProteomeRes.6(2007)1540–1559.
[2]S.Hanash,Diseaseproteomics,Nature422(2003)226.
[3]E.A.Dennis,Lipidomicsjoinstheomicsevolution,Proc.Natl.Acad.Sci.U.S.A.
106(2009)2089–2090.
[4]K.Sandra,A.dosSantosPereira,G.Vanhoenacker,F.David,P.Sandra, Comprehensivebloodplasmalipidomicsbyliquid
chromatography/quadrupoletime-of-flightmassspectrometry,J.
Chromatogr.A1217(2010)4087–4099.
[5]E.Fahy,S.Subramaniam,R.C.Murphy,M.Nishijima,C.R.Raetz,T.Shimizu,F.
Spener,G.vanMeer,M.J.O.Wakelam,E.A.Dennis,UpdateoftheLIPIDMAPS comprehensiveclassificationsystemforlipids,J.LipidRes.50(Suppl)(2009) S9–S14.
[6]E.Fahy,M.Sud,D.Cotter,S.Subramaniam,LIPIDMAPSonlinetoolsforlipid research,NucleicAcidsRes.35(Supplement2)(2007)W606–W612.
[7]X.Han,R.W.Gross,Shotgunlipidomics:electrosprayionizationmass spectrometricanalysisandquantitationofcellularlipidomesdirectlyfrom crudeextractsofbiologicalsamples,MassSpectrom.Rev.24(2005)367–412.
[8]P.T.Ivanova,S.B.Milne,D.S.Myers,H.A.Brown,Lipidomics:amass spectrometrybasedsystemslevelanalysisofcellularlipids,Curr.Opin.Chem.
Biol.13(2009)526–531.
[9]S.Pati,B.Nie,R.D.Arnold,B.S.Cummings,Extraction,chromatographicand massspectrometricmethodsforlipidanalysis,Biomed.Chromatogr.30 (2016)695–709.
[10]E.Sokol,E.Almeida,H.K.Hannibal-Bach,D.Kotowska,J.Vogt,J.Baumgart,K.
Kristiansen,R.Nitsch,J.Knudsen,C.S.Ejsing,Profilingoflipidspeciesby normal-phaseliquidchromatography,nanoelectrosprayionization,andion trap–orbitrapmassspectrometry,Anal.Biochem.443(2013)88–96.
[11]E.Cífková,M.Holˇcapek,M.Lísa,M.Ovˇcaˇci´ıkova´ı,F.Lyˇcka,P.Sandra, Nontargetedquantitationoflipidclassesusinghydrophilicinteractionliquid chromatography–electrosprayionizationmassspectrometrywithsingle internalstandardandresponsefactorapproach,Anal.Chem.84(2012) 10064–10070.
[12]H.P.Nguyen,K.A.Schug,TheadvantagesofESI–MSdetectioninconjunction withHILICmodeseparations:fundamentalsandapplications,J.Sep.Sci.31 (2008)1465–1480.
[13]M.Li,B.Feng,Y.Liang,W.Zhang,Y.Bai,W.Tang,T.Wang,H.Liu,Lipid profilingofhumanplasmafromperitonealdialysispatientsusingan improved2D(NP/RP)LC-QToFMSmethod,Anal.Bioanal.Chem.405(2013) 6629–6638.
[14]Y.S.Ling,H.J.Liang,M.H.Lin,C.H.Tang,K.Y.Wu,M.L.Kuo,C.Y.Lin, Two-dimensionalLC–MS/MStoenhanceceramideandphosphatidylcholine speciesprofilinginmouseliver,Biomed.Chromatogr.28(2014)1284–1293.
[15]J.F.Li,X.Yan,Y.L.Wu,M.J.Fang,Z.Wu,Y.K.Qiu,Comprehensive two-dimensionalnormal-phaseliquidchromatography×reversed-phase liquidchromatographyforanalysisoftoadskin,Anal.Chim.Acta962(2017) 114–120.
[16]M.Holˇcapek,M.Ovˇcaˇcíková,M.Lísa,E.Cífková,T.Hájek,Continuous comprehensivetwo-dimensionalliquidchromatography–electrospray ionizationmassspectrometryofcomplexlipidomicsamples,Anal.Bioanal.
Chem.407(2015)5033–5043.
[17]Y.Shan,Y.Liu,L.Yang,H.Nie,S.Shen,C.Dong,Y.Bai,Q.Sun,J.Zhao,H.Liu, LipidprofilingofcyanobacteriaSynechococcussp.PCC7002using two-dimensionalliquidchromatographywithquadrupoletime-of-flight massspectrometry,J.Sep.Sci.39(2016)3745–3753.
[18]S.Wang,X.Shi,G.Xu,Onlinethreedimensionalliquidchromatography/mass spectrometrymethodfortheseparationofcomplexsamples,Anal.Chem.89 (2017)1433–1438.
[19]M.Lísa,E.Cífková,M.Holˇcapek,Lipidomicprofilingofbiologicaltissuesusing off-linetwo-dimensionalhigh-performanceliquidchromatography?mass spectrometry,J.Chromatogr.A1218(2011)5146–5156.
[20]M.Li,X.Tong,P.Lv,B.Feng,L.Yang,Z.Wu,X.Cui,Y.Bai,Y.Huang,H.Liu,A not-stop-flowonlinenormal-/reversed-phasetwo-dimensionalliquid chromatography–quadrupoletime-of-flightmassspectrometrymethodfor comprehensivelipidprofilingofhumanplasmafromatherosclerosispatients, J.Chromatogr.A1372(2014)110–119.
[21]D.R.Stoll,Recentadvancesin2D-LCforbioanalysis,Bioanalysis7(2015) 3125–3142.
[22]M.A.Parent,L.Wang,J.Su,T.Netoff,L.L.Yuan,Identificationofthe hippocampalinputtomedialprefrontalcortexinvitro,Cereb.Cortex20 (2009)393–403.
[23]R.W.H.Verwer,R.J.Meijer,H.F.M.VanUum,M.P.Witter,Collateral projectionsfromtherathippocampalformationtothelateralandmedial prefrontalcortex,Hippocampus7(1997)397–402.
[24]S.Adhikari,M.A.Topiwala,J.A.Gordon,Singleunitsinthemedialprefrontal cortexwithanxiety-relatedfiringpatternsarepreferentiallyinfluencedby ventralhippocampalactivity,Neuron71(2011)898–910.
[25]A.Adhikari,M.A.Topiwala,J.A.Gordon,Synchronizedactivitybetweenthe ventralhippocampusandthemedialprefrontalcortexduringanxiety, Neuron65(2010)257–269.
[26]L.Jacobson,R.Sapolsky,Theroleofthehippocampusinfeedbackregulation ofthehypothalamic-pituitary-adrenocorticalaxis,Endocr.Rev.12(1991) 118–134.
[27]N.Maggio,M.Segal,Differentialcorticosteroidmodulationofinhibitory synapticcurrentsinthedorsalandventralhippocampus,J.Neurosci.29 (2009)2857–2866.
[28]A.Pitkänen,M.Pikkarainen,N.Nurminen,A.Ylinen,Reciprocalconnections betweentheamygdalaandthehippocampalformation,perirhinalcortex,and postrhinalcortexinrat:areview,Ann.N.Y.Acad.Sci.911(2000)369–391.
[29]A.Albrecht,G.C¸alıs¸kan,M.S.Oitzl,U.Heinemann,O.Stork,Long-lasting increaseofcorticosteroneafterfearmemoryreactivation:anxiolyticeffects andnetworkactivitymodulationintheventralhippocampus,
Neuropsychopharmacology38(2013)386.
[30]É.M.Szeg"o,T.Janáky,Z.Szabó,A.Csorba,H.Kompagne,G.Müller,G.Lévay,A.
Simor,G.Juhász,K.A.Kékesi,Amousemodelofanxietymolecularly characterizedbyalteredproteinnetworksinthebrainproteome,Eur.
Neuropsychopharmacol.20(2010)96–111.
[31]J.Horváth,T.Szögi,G.Müller,V.Szegedi,Theanxiolyticbuspironeshifts copingstrategyinnovelenvironmentalcontextofmicewithdifferent anxiousphenotype,Behav.BrainRes.250(2013)32–38.
[32]J.Folch,M.Lees,G.H.Sloane-Stanley,Asimplemethodfortheisolationand purificationoftotallipidsfromanimaltissues,J.Biol.Chem.226(1957) 497–509.
[33]T.Kind,K.H.Liu,D.Y.Lee,B.DeFelice,J.K.Meissen,O.Fiehn,LipidBlastinsilico tandemmassspectrometrydatabaseforlipididentification,Nat.Methods10 (2013)755–758.
[34]L.Abdullah,J.E.Evans,S.Ferguson,B.Mouzon,H.Montague,J.Reed,G.
Crynen,T.Emmerich,M.Crocker,R.Pelot,M.Mullan,F.Crawford,Lipidomic analysesidentifyinjury-specificphospholipidchanges3moaftertraumatic braininjury,FASEBJ.28(2014)5311–5321.
[35]C.Bascoul-Colombo,I.A.Guschina,B.H.Maskrey,M.Good,V.B.O’Donnell,J.L.
Harwood,DietaryDHAsupplementationcausesselectivechangesin phospholipidsfromdifferentbrainregionsinbothwildtypemiceandthe Tg2576mousemodelofAlzheimer’sdisease,Biochim.Biophys.ActaMol.Cell.
Biol.Lipids1861(2016)524–537.
[36]S.M.vanLiempd,D.Cabrera,F.Y.Lee,E.González,E.C.Dell’Angelica,C.A.
Ghiani,J.M.Falcon-Perez,BLOC-1deficiencycausesalterationsinaminoacid profileandinphospholipidandadenosinemetabolisminthepostnatal mousehippocampus,Sci.Rep.7(2017)1–11.
[37]T.G.Oliveira,R.B.Chan,F.V.Bravo,A.Miranda,R.R.Silva,B.Zhou,F.Marques, V.Pinto,J.J.Cerqueira,G.DiPaolo,N.Sousa,Theimpactofchronicstresson theratbrainlipidome,Mol.Psychiatry.21(2016)80.
[38]C.H.Cortie,A.J.Hulbert,S.E.Hancock,T.W.Mitchell,D.McAndrew,P.L.Else,Of mice,pigsandhumans:ananalysisofmitochondrialphospholipidsfrom mammalswithverydifferentmaximallifespans,Exp.Gerontol.70(2015) 135–143.
[39]T.Zhang,S.Chen,X.Liang,H.Zhang,Developmentofa mass-spectrometry-basedlipidomicsplatformfortheprofilingof phospholipidsandsphingolipidsinbraintissues,Anal.Bioanal.Chem.407 (2015)6543–6555.
[40]A.J.Alpert,Hydrophilic-interactionchromatographyfortheseparationof peptides,nucleicacidsandotherpolarcompounds,J.Chromatogr.A499 (1990)177–196.
[41]P.Hemström,K.Irgum,Hydrophilicinteractionchromatography,J.Sep.Sci.
29(2006)1784–1821.
[42]B.K.Matuszewski,M.L.Constanzer,C.M.Chavez-Eng,Strategiesforthe assessmentofmatrixeffectinquantitativebioanalyticalmethodsbasedon HPLC-MS/MS,Anal.Chem.75(2003)3019–3030.
[43]H.Trufelli,P.Palma,G.Famiglini,A.Cappiello,Anoverviewofmatrixeffects inliquidchromatography–massspectrometry,MassSpectrom.Rev.30(2011) 491–509.
[44]N.E.Braverman,A.B.Moser,Functionsofplasmalogenlipidsinhealthand disease,Biochim.Biophys.ActaMol.BasisDis.1822(2012)1442–1452.