Contents lists available atScienceDirect
International
Journal
of
Mass
Spectrometry
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 / i j m s
Liquid
extraction
surface
analysis
for
native
mass
spectrometry:
Protein
complexes
and
ligand
binding
Victor
A.
Mikhailov
a,1,
Rian
L.
Griffiths
b,1,
Helen
J.
Cooper
b,∗aDepartmentofChemistry,UniversityofOxford,Oxford,OX13TA,UK
bSchoolofBiosciences,UniversityofBirmingham,Edgbaston,BirminghamB152TT,UK
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received26May2016
Receivedinrevisedform1September2016 Accepted20September2016
Availableonlinexxx Keywords:
Liquidextractionsurfaceanalysis(LESA) Nativemassspectrometry
Non-covalentproteincomplexes Protein-ligandinteractions Drugbinding
a
b
s
t
r
a
c
t
Nativeliquidextractionsurfaceanalysis(LESA)massspectrometryenablesthedirectsamplingof pro-teincomplexesfromasolidsurface.WehavepreviouslydemonstratednativeLESAmassspectrometry ofholomyoglobin(∼17kDa)fromglassslidesandtetramerichaemoglobin(∼64kDa)fromdriedblood spotsandthintissuesections.Here,wefurtherexplorethecapabilitiesofthisemergingtechniqueby investigatingarangeofproteinswhichexistinvariousoligomericstatesinvivo.Tetramericavidin (∼64kDa),octameric(∼190kDa)andhexadecameric(∼380kDa)CS2 hydrolase,and tetradecameric
GroEL(∼800kDa)werealldetectedbynativeLESAmassspectrometry.Moreover,trimericAmtB,a mem-braneprotein,couldalsobeobservedbynativeLESAmassspectrometry.ThesuitabilityofLESAmass spectrometryforprobingprotein-ligandbindingwasalsoinvestigated.Non-covalentcomplexesofthe ligandbiotinwiththeproteinsavidin,haemoglobinandbovineserumalbuminweredetected.Theresults indicatethatnon-specificbindingisminimalandthatnativeLESAmassspectrometryisapromisingtool fortheinvestigationofbiologicallysignificantligandbinding.
©2016TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).
1. Introduction
Knowledgeofproteintertiaryandquaternarystructureis gen-erallyrequiredforunderstandingproteinfunctionormalfunction. Experimentalmethodsthatprovidethisknowledgearecapableof maintainingnoncovalentbonds,suchashydrogenbondsand- interactions,thuspreservingfoldedproteinsintheirnative struc-tureandthenoncovalentbondingwiththosespecieswithwhich theyinteract.Nativeelectrospraymassspectrometry(MS)enables analysisoffoldedproteinsandnon-covalentproteincomplexesin thegasphasebychoosinga non-denaturingelectrospraybuffer andcarefuloptimisationofelectricfieldsandpressuresinthemass spectrometer[1–4].IncombinationwithotherMStechniques,such asionmobilitymeasurements[5–7],gas-phasedissociation meth-ods[8,9],hydrogen-deuteriumexchange[10–12]andnon-covalent labelling[12–14]morespecificinformationontheoverall topol-ogyof non-covalent complexes, connectivity betweendifferent subunitsinthecomplexanddrugbindingsitescanbeobtained. Morerecently,anativeMSapproachhasbeendevelopedfor mem-braneproteincomplexes[15,16].Stabilisationofthesecomplexes
∗ Correspondingauthor.
E-mailaddress:[email protected](H.J.Cooper). 1 Theseauthorscontributedequallytothiswork.
insolutionandthegas-phaserequiresthattheyarepurifiedand electrosprayedindetergentmicelles.ThenativeMSapproachalso allows investigationof interactions of membrane protein com-plexeswithlipidsanddrugs[17,18].
Nativemassspectrometryischaracterisedbyionsinrelatively lowchargestates–fewerprotonationsitesareavailableinfolded proteins–andthushighermass-to-chargeratiosarerecorded(for gas-phaseproteincomplexestypicalm/z≥3000)[1,2].Increasing theupper limit ofthemass range toprobe everlargerprotein assemblieshasbeenoneofthemajormotivesindevelopmentofthe nativeMSapproach.Q-TOFmassanalysershavebeentheanalysers ofchoiceinnativeESI,duetotheirhighm/zrange[2,19];however, morerecentlynativeMShasbeensuccessfullyimplementedon FTICRandOrbitrapinstruments[20,21],whichoffermass resolu-tionsuperiortoQ-TOF.Todate,massspectraofintactcomplexes aslargeas18MDa (anentire capsidassembly ofbacteriophage HK97)havebeenreported[22],althoughthemassresolutionof suchlargespeciesislimitedbyincompletedesolvationoftheir gas-phaseionsratherthanbytheinherentinstrumentresolutionofthe massanalysersemployed[23].
MostnativeMSexperimentsarecarriedoutbydirect electro-sprayinfusionofthepurifiedsampleintothemassspectrometer. AnalyticalapplicationsofthenativeMSapproachcanbeexpanded considerablywithsamplingofnon-covalentprotein–proteinand protein-ligandcomplexesdirectlyfromsurfaces:theavenuesfor http://dx.doi.org/10.1016/j.ijms.2016.09.011
future explorations include probing diverse biological samples, suchasdriedbloodspotsandthintissuesections,protein/ligand assays and biomimicking membranes. Surface samplingcan be coupledto electrospray ionisation (essential for native MS) by desorptionelectrosprayionisation(DESI)orliquidmicrojunction basedtechnologiessuchasliquidextractionsurfaceanalysis(LESA) [24,25].InDESI,ajetofelectrosprayedsolventisdirectedata sur-facecontaininganalytemoleculesinordertodesorbanddeliver themintothemassspectrometer[26–28].Thistechniqueproduces multiplychargedionssimilartodirectESIinfusion[29,30].DESIhas beensuccessfullyappliedtothesurfaceanalysisofsmallmolecules, peptidesandproteinsupto∼25kDa [27,31,32],however direct analysisofintactproteinsfrombiologicallyrelevantsampleshas notbeenreportedtodate.AvariantofDESIinwhichliquid sam-plesareprobed(liquid-DESI)hasdemonstratedthatintactproteins andproteincomplexesupto∼150kDacanbedetected[29].
Inanotherapproach,DESIcanbeusedtoprobenon-covalent interactions between immobilisedproteins and ligands. In this methodproteinsareimmobilisedonasurfaceandtreatedwitha mixtureofligands.Thenon-bindingligandsarethenwashedaway, andtheassayissubjectedtoDESIanalysiswithdenaturingsolvents leadingtothedetachmentof theboundligandmoleculesfrom theproteinandtheirdeliverytothemassspectrometer[33].This approachcanbeusedforhigh-throughputscreeningforbinding ligands,buttheindirectassessmentofnon-covalentinteractions betweentheproteinandtheligandsviadetectionoftheligand moleculesbyDESI is not entirelyconclusive,becausethe pres-enceoftheimmobilisingagentcanproducefalsepositiveresults [34].Furthermore,these studiescannotprovide informationon thestoichiometryofbinding,and arenotapplicabletosamples withheterogeneousproteintargets,suchasbloodspotsandtissue sections.
Alternativesurface sampling techniques that are potentially capableofextractingadiverserangeofnon-covalentcomplexesfor MSanalysisareliquidmicrojunctionbasedmethodssuchasliquid extractionsurfaceanalysis(LESA)[24,25]orcontinuousflow liq-uidmicrojunctionsampling[35].LESAenablessamplingofdiscrete locationsonasurfacebyplacingasmalldropletofthesolventona spotwhichisthenre-aspiredwithaconductivepipettetippriorto ESIMS.LESAMSanalysisofintactproteinshasbeenachievedfrom avarietyofsubstrates:driedbloodspotsonfilterpaper[36–39], thintissuesectionsthaw-mountedontoglassslides[40–43]and E.colicoloniesgrownonagargel[44].Recently,proteinspeciesup to∼15kDawerereportedby‘nanoDESI’imagingoftissuesections [45].Despiteitsname,nano-DESIisactuallybasedoncontinuous flowliquidmicrojunctionsurfacesampling[46].
We recently reported the use of LESA with non-denaturing ammoniumacetatebasedsolventfortheMSanalysisof nonco-valentproteincomplexesfromglassandpolyvinylidenefluoride (PVDF) substrates [47]. Using this ‘native’ LESA MS method, intact∼17.5kDaholomyoglobin,anon-covalentmyoglobin-heme complex, and a range of non-covalently bound species from a haemoglobinsample:␣HandH-globinsubunitsassociatedwith
heme,∼16kDaeach;the(␣)2Hheterodimers(∼32kDa)andthe
intact(␣)2 4H tetrameric complexes(∼64kDa)weredetected.
Furthermore,theintacthaemoglobintetramercouldbedetected fromdriedbloodspotsonfilterpaper.Griffithsetal.subsequently demonstrated direct detection of the tetrameric haemoglobin complexfromvasculaturein thintissuesectionsofmouseliver thaw-mountedontoglassslides[48].
Intheworkpresentedhere,wesoughttofurthercharacterise thecapabilitiesofnativeLESAMS.Wedemonstrateasignificant extensionofthemassrangeofourmethod,and showthatit is possibletodetectintactproteinassembliesaslargeas∼800kDa. Wealsodemonstratethatmembraneproteincomplexes,thatare preparedinmicelle-containingsolutions,areamenabletonative
LESAMSandcanbeprobeddirectlyfromdriedspotsonasurface. ThemassspectraproducedbynativeLESAMSaresimilartothose obtainedfrom‘standard’nativemassspectrometry.Furthermore, protein-ligandinteractionshavebeenprobedbyLESAMS,andthe detectionofbiotinnoncovalentlyboundtoavidin,haemoglobin andBSAispresented.
2. Materialsandmethods
2.1. Materials
Protein standards (avidin from chicken egg white, bovine serumalbumin(BSA),humanhaemoglobin,Chaperonin60from Escherichiacoli(GroEL)),biotinandammoniumacetatewere pur-chasedfromSigma-Aldrich(Gillingham,UK). GroELwasfurther purified as described elsewhere [1]. CS2 hydrolase from
Acidi-anusA1-3wasagiftfromJasminMecinovic,RadboudUniversity Nijmegen, Netherlands [49]. Ammonium channel (AmtB) was expressedand purified in-houseat theUniversity of Oxford as describedelsewhere[50].Octyltetraglycol(C8E4)waspurchased fromAvanti(Alabaster,AL,USA).
2.2. Samplepreparation
Proteins were prepared in 200mM ammonium acetate (pH=7.5) resulting in a final protein concentration 1–10M. GroELandBSAproteinswerefurtherdesaltedusingMicro-BioSpin columns(BioRad,Hercules,CA,USA).AmtBwaspreparedin ammo-niumacetatesolutioncontainingC8E4attwicethecriticalmicelle concentration.1lofproteinsolutionwasspottedontoacleaned glassslideforLESA.Thesamplespotswereair-driedatroom tem-peratureandprobedusingLESAwithadelayof20–90minbetween thedepositionofthesampleanditsextractionforMSanalysis. 2.3. Surfacesampling
LESAwascarriedoutusingtheTriversaNanomatechip-based ESI source (Advion Biosciences, Ithaca, USA) coupled with the SynaptG2Smassspectrometer(Waters,Manchester,UK).The sam-plewasmounted ontotheLESAuniversaladaptorplateandan imagewasacquiredusinganEpsonPerfectionV300photoscanner. TheexactlocationtobesampledwasselectedusingtheLESAPoints software(Advion).Theextraction/ionisationsolventwas200mM ammoniumacetateinwater(HPLCgrade,JTBaker,Deventer,The Netherlands).Forbiotinbindingstudies,theammoniumacetate solutionwasusedtospottheproteinsamplesontheslide,the extractionsolutioncontainedeither1.0or0.1mMbiotinin200mM ammoniumacetate.Intheextractionprocess,thesolventwas aspi-ratedfromthesolventwell,beforetheroboticarmrelocatedtoa positionabovethesampleslide.Someofthesolventwasthen dis-pensedandaliquidmicrojunctionwasmaintainedbetweenthe pipettetipand thesamplesurfacebeforeaspiratingthesolvent anddeliveringittotheelectrospraychip.Thedistancebetweenthe nanospraychipandtheentrancetothesampleconewas3–6mm. ThedetailsoftheLESAextractionprocedureandionsource (Nano-mate)parametersforeachproteinaregiveninTable1.
2.4. Massspectrometry
Aspeedivalve(BOCEdwards,Crawley,Sussex,UK)wasinstalled betweentherotaryand turbopumpsthat pumptheion source regioninSynaptG2S(Waters,Wilmslow,UK).Thepressureinthe ionsourcecanbecontrolledbycarefulthrottlingofthepumpingby thespeedivalvewiththeaimofimprovingthetransmissionofions withhighm/zthatarecharacteristicforintactproteincomplexes.
Table1
ParametersofLESAextractionprocedureandionsourceconditions.
Protein Volumeaspirated
fromsolventwell (l)
Heightabovethe surface(mm)
Volume dispensedonto thesurface(l)
Delaytime(secs) Volume reaspirated(l) Gaspressure (psi) ConeVoltage (kV) AmtB 2.0 0.8 1.5 60 2.0 0.3 1.9 CS2hydrolase 2.0 0.8 1.5 60 2.0 0.3 1.9 GroEL 4.0 0.6 2.0 30 2.5 0.15 1.8 Haemoglobin 4.0 0.6 2.0 30 2.5 0.15 1.8 Avidin 4.0 0.6 2.0 30 2.5 0.15 1.8 BSA 4.0 0.6 2.0 30 2.5 0.15 1.8
Thebackingpressurewasalteredfrom3to9mbarforthese
exper-iments.MassspectrawereacquiredinthefullscanTOFmodeon
theSynaptG2S.Thescantimewassetat2–5s,them/zrangewas
altereddependingonthespeciesinvestigated(usually≥10,000).
Thecapillarytemperaturewassetto30◦Candtheconevoltagewas
setat45V.Insomeexperiments(seetext)collisioninduced
disso-ciation(CID)wascarriedoutontheionsinthetrapregionofthe
massspectrometer.Collisionvoltagewasvariedfrom50to200V.
AlldatawereanalysedbyMassLynxsoftware(version4.1,Waters).
Therawmassspectrahavebeenaveragedfor20–100scansand
presentedherewithminimalsmoothingandbackground
subtrac-tion.ThenativeLESAmassspectrahavebeencomparedagainst
themassspectraofthesameproteinsusingdirectinfusion
elec-trospraywithNanomateonthesamemassspectrometerandthe
massspectrafromthepreviouswork.Forselectedexamples(e.g.
avidin-biotinbinding),LESAMSspectrawerecomparedwith
high-resolutionnativeMSspectraobtainedonanOrbitrapQ-Exactive
instrument(Thermo,Bremen,Germany)intheUniversityofOxford
modifiedforhighm/zrange,similartothemodificationspreviously
reportedbyRoseetal.[21].
3. Resultsanddiscussion
3.1. Protein-proteincomplexes
Arangeofproteinswhichexistinoligomericstatesinvivoand innon-denaturingsolutionswereinvestigated.Theirnon-covalent complexesvaryinmassfrom∼60kDaupto∼800kDaandexistin differentoligomericstates,fromtrimers(AmtB)tohexadecamers (CS2 hydrolase).Fig.1Ashowstheresultsobtainedforavidin, a
solubletetrameric glycoprotein,which hasbeenlong-usedas a standardfor nativeMS[51].Themassoftheavidintetrameris similartothatofthehaemoglobintetramerdescribedinLESAMS studiesin ourpreviousworks [48,52].Sampling thedried pro-teinonglasswith200mMammoniumacetatesolutionresulted in the detection of thetetrameric species (∼64kDa) in charge statesrangingfrom12+to16+ atm/z∼5300,4900,4600,4300 and4000respectively,andareinaccordancewiththechargestate distributionusuallyobservedinthemassspectraofavidinfrom non-denaturingelectrospray[53].Thetetramerpeaksshowninthe spectrumarerelativelybroad(∼200Th).Usuallypeakbroadening innativeMSiscausedbyincompletedesolvationofthegas-phase molecularionsandthepresenceofsalts[47],butcanalsobeasign ofproteindegradation.Ahigh-resolutionmassspectrumofavidin, producedbydirectnano-electrosprayonanOrbitrapQ-Exactive massspectrometerthatwasmodifiedforhighm/zrange,shows thepresenceofpeakscorrespondingtothesemodificationswithin theenvelopofeachchargestateofthetetramer(Fig.S1A, Support-inginformation).Massspectraofthechargestates(14+to18+)of theavidintetramerobtainedontheNanomate/SynaptG2S appa-ratusbydirectinfusiondemonstratepeakbroadeningsimilarto LESAMS(Fig.S1BversusFig.1A).Furthermore,collisioninduced dissociation(CID)oftheavidintetramerrevealsmultiplepeaksfor eachchargestateofthefragmentmonomer(Fig.S1B),indicating
thepresenceofmultiplemodificationsontheproteinbackbone. TheseresultssuggesttheavidintetramerionsobservedinLESA MShavethesamestructureasthoseobservedin‘standard’native MS.Furtherevidencethatavidinretaineditsbiologicallyrelevant stateintheLESAMSapproachcamefromourresultsforitsbinding tobiotin(seebelow).
Carbon disulphide hydrolase (CS2 hydrolase) is an enzyme
that converts carbon disulphide to carbon dioxide and hydro-gensulphide.Itexistsasa190kDaoctamericringanda380kDa hexadecamericcatenane(locked-ringstructure).Thecatenane for-mationwasreportedasbiologicallyrelevantin2013byEldijketal. [54].Surfacesamplingofdriedsolutionsofthisproteinledtothe detectionofboth oftheseoligomericforms.Theoctamericring structurewasdetectedinchargestatesrangingfrom26+to31+ betweenm/z6000to8000asshowninFig.1B.Inthesamemass spectrum,thehexadecamericcatenaneisrecordedinchargestates between35+to42+(m/zrange9000–12000).Similarmassspectra oftheCS2hydrolasewerereportedinearlierdirectinfusionnative
MS[49],andwereobtainedinthiswork:directinfusion nanoelec-trosprayofthesameproteinsolutionusedforLESAMSresulted inthemassspectrumshowninFig.S1C.Thefactthatboth bio-logicallyrelevantoligomericformsofCS2hydrolasesurvivedthe
dryingstageoftheprocedureandcouldthenbedetectedbyLESA MSemphasisesthebiologicalrelevanceofthismethod.
The largestsoluble complexanalysed herebyLESA MS was GroEL,achaperoneproteinwhichformsa∼800kDatetradecameric complexconsistingoftworingheptamers.GroELtetradecameris asolubleproteincomplexwhichhasoftenbeenusedtotestthe uppermassrangeinnativeMSapplications[21,55–57].GroELwas detectedbyLESAMSfromdriedproteinsamplesinchargestates ranging from61+ to68+ at m/z∼13100,12900, 12750, 12500, 12350,12200,12000and 11850respectively(Fig.1C).Previous nativeelectrosprayionisationstudiesinammoniumacetatebased solventsystemshave reportedthedetectionof theGroEL com-plexinthesimilarm/zrangewithslightvariationsinthecharge statedistributionthatwasapparentlydependentonthetypeof mass spectrometerused [21,56,57]. We also detectedunfolded monomeric subunits of GroELat m/z<4000(∼58kDa, datanot shown),whichmightbeanindicationofeitherpartialdegradation ofthecomplexinthedryspotsbeforeLESAextraction,orcollision induceddissociationintheionsource.Nevertheless,mostofthe GroELtetradecamerwasretainedduringtheLESAprocedureand producedmassspectrasimilartonativeMSdirectlyfromsolution, indicativeofthesimilarityofthegas-phaseGroELionsformedin thesetwomethods.
ThenativeLESAMSapproachwasalsoappliedtoa different classof proteincomplexes:membrane proteins.The encapsula-tionofhydrophobicmembranecomplexesindetergentmicelles isrequiredtoretaintheirstabilityinsolutionandthegasphase. NativeESImassspectraofmembraneproteincomplexesinmicelles result inunstructured features due totheheterogeneity of the micelles.Collisionactivationisusedtoliberatethemembrane com-plexfromthegas-phasemicellesandresolveitschargestates[16]. WeusedasimilarapproachinourLESAMSexperiment.AmtBis
Fig.1.NativeLESAmassspectraofproteincomplexesfromglasssubstrate.A)Massspectrumofavidintetramer(64kDa)aftercollisionactivationofavidinionsinthetrap region(100Vcollisionvoltage).
Fig.2. NativeLESAmassspectraofBSAfromglasssubstrate.A)BSAspotssampledwith200mMammoniumacetate.B)BSAspotssampledwith200mMammonium acetate/1mMbiotin.
amembranetransporterproteinwhichformsatrimericcomplex withamassofapproximately140kDa.AmtBinC8E4micellesin ammoniumacetatesolutionwasspottedontheglassslidesand, afterthespotsweredried,wasextractedandanalysedbyLESAMS usingpureammoniumacetatesolution.Toresolvethechargestates ofthis trimercomplex weusedanelevatedcollisionvoltageof 200VinthetrappartoftheSynaptG2S.Wedetectedchargestates oftheintacttrimerrangingfrom17+to20+atm/z∼8400,7900, 7400and7000respectively.Thesepeakswererelativelynarrow, ≤50Thwide(Fig.1D),incomparisonwiththewidthofthepeaks ofthesolublecomplexes(Fig.1A–C).Thiseffectisprobablydueto thefactthatthemembraneproteinisnotlikelytoretainmany sol-ventmoleculesbecauseofitslimitedhydrophilicsurfaceandthe largeactivationenergythathadtobeappliedtoitsgas-phaseions inordertoremovethedetergentmoleculesandrevealthecharge states.AsimilarmassspectrumofAmtBtrimerwasproducedby directinfusionnanoelectrospray(Fig.S1D).
3.2. Protein-ligandinteractions
Biotin(244Da)bindstoavidinspecificallyandwithveryhigh affinity(onebiotinmoleculepereachavidinmonomer).This prop-ertyisexploitedinbiotechnologytoisolatebiotinylatedmolecules inaffinityassaysorchromatography,whereimmobilisedavidinor homologousproteinsareusedtoanchorthemoleculesofinterest [58].Asbiotinbindingisfast,asolutionofbiotininammonium acetatewasusedtosampledriedspotsofavidinbyLESAinorder totestwhetherthebindingtoavidincantakeplaceduringthis procedure. Samplingwith 200mM ammonium acetatesolution containing1mM biotinledtoamass shiftforthemajorpeaks ofthechargeofavidintetramer,asshowninFig.1Ainset.This massshift,averagedfor14+to16+chargestate,relatestoan addi-tionof∼960Da,indicatingthatasmanyasfourbiotinmolecules (244Da×4=976Da)areboundtotheavidintetramer.Atpresent wearelimitedbytheresolvingpoweroftheSynaptmass spectrom-eterandtherelativelysmallm/zrangeoftheOrbitrapcoupledto LESAinBirmingham(m/z<4000)todemonstratespecificbinding ofbiotintoavidintetramerwithhigherresolutionbythismethod. WehavealsotestedbiotinbindingtoavidinontheOrbitrap Q-Exactive(upperm/zlimit30,000Th,UniversityofOxford)bythe directinfusionofavidin/biotinsolutionusingananospray capil-lary,Fig.S2AandB.Asthefinestructureofthechargestatesof avidincanberesolvedontheOrbitrap,themassshiftofthepeaks correspondingtobiotinbindingwascalculatedtobeexactly976Da (fourbiotinmoleculesbound).Moreover,CIDoftheavidin/biotin non-covalentconjugateat100Vcollisionvoltageresultedinthe productionofapo-avidintetramer(−976Damassshift,Fig.S2C). Therefore,weconcludethatimplementingLESAMSona
higher-resolution/highermassrangeinstrumentinfutureshouldbeable toresolvethefinestructureofmultiplymodifiedproteinsand pro-teincomplexesandtheexactstoichiometryofthedrugbindingto them.
Determinationoftheoverallpharmacokineticprofileofadrug requiresmeasurementsofitsaffinitytowardplasmaproteins[59]. NativeLESAMSwasappliedtotheinvestigationofbindingofbiotin totwobloodproteins,BSAandhumanhaemoglobin,inafashion similartotheavidinexperiments.Thedegreeofmodificationof wildtypeBSAissmallerthanthatofmultiplyglycosylatedavidin. It wasthereforepossibletoresolve peakscorrespondingtothe bindingofindividualbiotinmolecules(atleastuptofour)toBSA (chargestates13+to17+)ontheSynaptG2ScoupledtoLESA,Fig.2, althoughtheresolutionwasnothighenoughtoestimatethe max-imumnumberofboundmolecules.Biotinbindingtohaemoglobin dimers,(␣)2Hwasalsodetected,withbettermassresolutionand
relativelyhighsignalintensity,inchargestatesrangingfrom12+ to8+betweenm/z2500to4250,Fig.3.Atleastsevenmolecules ofbiotinboundtothehaemoglobindimerweredetectedatzero collisionenergyinthetrapregionofSynaptG2S.Thetetrameric haemoglobin((␣2)4H)boundtobiotinwasalsodetectedwitha
lowermassresolutioninthechargestatesfrom14+to16+,and withapositivemassshiftinrespecttothepeaksofthecharge statesofunboundhaemoglobintetramers(datanotshown),but theexactnumberoftheboundbiotinmoleculeswasimpossibleto estimatewiththislevelofmassresolution.Peakscorrespondingto ␣-and-globinmonomerandmonomer-hemespeciesincharge statesrangingfrom8+to16+werealsoobserved,butitisnot appar-entinthemassspectrawhetheranybiotinbindingtookplacefor themonomers(Fig.3B):itis,however,possibletoconcludethat apo-proteinmonomericionsaremuchmoreabundantthanthe ionswhichmightbecorrespondingtoprotein/biotincomplexes, sothebindingofbiotintothemonomerichaemoglobincannotbe asstrongastothedimers.Theseresultsareinaccordwitha pre-viouslyreportedstudyonbiotinbindingtohumanbloodplasma proteins[60].Althoughtheauthorsdidnotinvestigatethebiotin bindingtospecificproteins,theyfoundoutthat12%ofbiotinin plasmaiscovalentlyboundand7%isboundreversibly.
Non-specificbindingofligandstoproteinsornon-specific for-mationofnon-covalentproteincomplexesissometimesobserved innativemassspectrometry[1,2].Inordertoexcludenon-specific bindingfromthemassspectra,theyareoftencollectedeitherina serieswithdiminishingligandorproteinconcentrationsinthe elec-trospraysolutioninordertodecreasethechancesofnon-specific associationin theelectrospraymicrodroplets,orwithextraion activation(e.g.,carefullyelevatedsamplingconeorcollision volt-age)todissociatenon-specificcomplexesinthegas-phase.LESA massspectraofhaemoglobinandhaemoglobin/biotinwitha
colli-Fig.3.NativeLESAmassspectraofhaemoglobinfromglasssubstrate.A)andC)Haemoglobinspotssampledwith200mMammoniumacetate.B)andD)Haemoglobinspots sampledwith200mMammoniumacetate/1mMbiotin.C)andD)100Vtrapcollisionvoltagewasused.
sionvoltageof100VinthetrapregionofSynaptG2Sareshown inFig.3C andD.Thesemass spectrademonstrate thatsomeof thebiotinmoleculesremainboundtohaemoglobindimerseven undertheharshactivationconditions.Thisresultindicatesthat atleastsomeofthebiotinbindingtohaemoglobinheterodimers observedbyLESAMSisspecificandthereforebiologically signif-icant.Assumingthatbiotindoesnotbindspecificallytounfolded haemoglobinmonomers,theabsenceofstrongMSsignals corre-spondingtobiotinbindingtothemonomershaemoglobin(Fig.3B andD)alsoindicatesthatnon-specificbindinginourLESA experi-mentswasminimal.
4. Conclusions
Liquidextractionsurfaceanalysis(LESA)ofproteinspecieson glasssubstrateshasbeenshowntobesuccessfulforprotein com-plexesupto∼800kDa(GroEL).Thismassissignificantlygreater thanpreviouslyreported(∼64kDa,haemoglobin).Remarkably,the investigatedcomplexescouldsurvivearelativelylongtime,from 20mintooveranhour,ontheglasssubstratebeforetheirextraction
byLESA.TheresultsalsoshowthatnativeLESAMShaspotential forapplicationsinmembraneproteinstudiesthatinvolvetheuse ofstabilisingdetergentmicelles,asdemonstratedbythesuccessful LESAMSdetectionoftheAmtBmembranecomplex.Furthermore, thepotentialofnativeLESAMStoprobeprotein-ligand interac-tionshasbeenshownforthefirsttimewiththebindingofbiotinto avidin,BSAandhaemoglobin.Theresultsobtainedforbiotin bind-ingtoavidinconfirmedthat avidinremainedinthebiologically activestateduringLESA.Asourpreviousresultsforhaemoglobin demonstrated[47],improvedmassresolutioncanbeachievedby interfacingthismethodwithmassspectrometersofhigher resolv-ingpower,e.g.Orbitrap.
Insummary,nativeLESAmassspectrometryis(1)suitablefor theanalysisofhighmassproteincomplexes;(2)interrogating com-plexesofboth solubleandmembraneproteins;and(3)probing protein-ligandinteractions.Theresultsarehighlyencouragingfor futureattemptstoextendtheuseofthenativeLESAMSmethod totheanalysisof protein/ligand bindingassays, theanalysisof membraneproteincomplexesfrombiomimickingmembranes,and
insituanalysisofheterogeneousbiologicalsurfaces,suchasdry bloodspotsandtissuesections.
Acknowledgments
RLG and HJC are funded by EPSRC (EP/L023490/1). Waters Corporation is acknowledged for a Fellowship to VAM. The Waters Synapt G2S mass spectrometer was funded by EPSRC (EP/K039245/1).Supplementarydatasupportingthisresearchis openlyavailablefromtheUniversityofBirminghamdataarchive athttp://findit.bham.ac.uk/.JosephGaultandDenisShutin(Carol Robinson’sgroup)fromtheUniversityofOxfordareacknowledged forprovidingasampleofAmtBandtheirhelpwithcollectingMS dataontheOrbitrapQ-Exactive.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.ijms.2016.09.011.
References
[1]H.Hernandez,C.V.Robinson,Determiningthestoichiometryandinteractions ofmacromolecularassembliesfrommassspectrometry,Nat.Protoc.2(3) (2007)715–726.
[2]A.J.Heck,R.H.vandenHeuvel,Investigationofintactproteincomplexesby massspectrometry,MassSpectrom.Rev.23(5)(2004)368–389. [3]M.Sharon,StructuralMSpullsitsweight,Science340(6136)(2013)
1059–1060.
[4]J.Marcoux,C.V.Robinson,Twentyyearsofgasphasestructuralbiology, Structure(Lond.,Engl.:1993)21(9)(2013)1541–1550.
[5]Z.Hall,A.Politis,C.V.Robinson,Structuralmodelingofheteromericprotein complexesfromdisassemblypathwaysandionmobility-massspectrometry, Structure20(9)(2012)1596–1609.
[6]A.Politis,etal.,Integratingionmobilitymassspectrometrywithmolecular modellingtodeterminethearchitectureofmultiproteincomplexes,PLoSOne 5(8)(2010).
[7]E.vanDuijn,etal.,Chaperonincomplexesmonitoredbyionmobilitymass spectrometry,J.Am.Chem.Soc.131(4)(2009)1452–1459.
[8]J.L.P.Benesch,Collisionalactivationofproteincomplexes:pickingupthe pieces,J.Am.Soc.MassSpectrom.20(3)(2009)341–348.
[9]M.Zhou,C.Huang,V.H.Wysocki,Surface-induceddissociationofion mobility-separatednoncovalentcomplexesinaquadrupole/time-of-flight massspectrometer,Anal.Chem.84(14)(2012)6016–6023.
[10]T.E.Wales,J.R.Engen,Hydrogenexchangemassspectrometryfortheanalysis ofproteindynamics,MassSpectrom.Rev.25(1)(2006)158–170.
[11]J.R.Engen,Analysisofproteinconformationanddynamicsby
hydrogen/deuteriumexchangeMS,Anal.Chem.81(19)(2009)7870–7875. [12]L.Konermann,X.Tong,Y.Pan,Proteinstructureanddynamicsstudiedby
massspectrometry:H/Dexchange,hydroxylradicallabeling,andrelated approaches,J.MassSpectrom.43(8)(2008)1021–1036.
[13]K.Takamoto,M.R.Chance,Radiolyticproteinfootprintingwithmass spectrometrytoprobethestructureofmacromolecularcomplexes,Annu. Rev.Biophys.Biomol.Struct.35(2006)251–276.
[14]Y.Pan,etal.,Structuralcharacterizationofanintegralmembraneproteinin itsnaturallipidenvironmentbyoxidativemethioninelabelingandmass spectrometry,Anal.Chem.81(1)(2009)28–35.
[15]N.P.Barrera,etal.,Micellesprotectmembranecomplexesfromsolutionto vacuum,Science321(5886)(2008)243–246.
[16]A.Laganowsky,etal.,Massspectrometryofintactmembraneprotein complexes,Nat.Protoc.8(4)(2013)639–651.
[17]A.Laganowsky,etal.,Membraneproteinsbindlipidsselectivelytomodulate theirstructureandfunction,Nature510(7503)(2014),p.172-+.
[18]J.Marcoux,etal.,Massspectrometryrevealssynergisticeffectsof nucleotides:lipids,anddrugsbindingtoamultidrugresistanceeffluxpump, Proc.Natl.Acad.Sci.U.S.A.110(24)(2013)9704–9709.
[19]F.Sobott,etal.,Atandemmassspectrometerforimprovedtransmissionand analysisoflargemacromolecularassemblies,Anal.Chem.74(6)(2002) 1402–1407.
[20]H.Li,etal.,Nativetop-downelectrosprayionization-massspectrometryof 158kDaproteincomplexbyhigh-resolutionfouriertransformioncyclotron resonancemassspectrometry,Anal.Chem.86(1)(2013)317–320. [21]R.J.Rose,etal.,High-sensitivityOrbitrapmassanalysisofintact
macromolecularassemblies,Nat.Methods9(11)(2012),p.1084-+. [22]J.Snijder,etal.,Studying18MDavirusassemblieswithnativemass
spectrometry,Angew.Chem.Int.Ed.52(14)(2013)4020–4023.
[23]P.Lossl,J.Snijder,A.J.R.Heck,Boundariesofmassresolutioninnativemass spectrometry,J.Am.Soc.MassSpectrom.25(6)(2014)906–917.
[24]D.Eikel,etal.,Liquidextractionsurfaceanalysismassspectrometry (LESA-MS)asanovelprofilingtoolfordrugdistributionandmetabolism analysis:theterfenadineexample,RapidCommun.MassSpectrom.25(23) (2011)3587–3596.
[25]W.Rao,etal.,AmbientDESIandLESA-MSanalysisofproteinsadsorbedtoa biomaterialsurfaceusingin-situsurfacetrypticdigestion,J.Am.Soc.Mass Spectrom.24(12)(2013)1927–1936.
[26]Z.Takáts,etal.,Massspectrometrysamplingunderambientconditionswith desorptionelectrosprayionization,Science306(5695)(2004)
471–473.
[27]Y.-S.Shin,etal.,Desorptionelectrosprayionization-massspectrometryof proteins,Anal.Chem.79(9)(2007)3514–3518.
[28]A.A.Stokes,etal.,Top-downproteinsequencingbyCIDandECDusing desorptionelectrosprayionisation(DESI)andhigh-fieldFTICRmass spectrometry,Int.J.Massspectrom.289(1)(2010)54–57.
[29]C.N.Ferguson,etal.,Directionizationoflargeproteinsandproteincomplexes bydesorptionelectrosprayionization-massspectrometry,Anal.Chem.83 (17)(2011)6468–6473.
[30]P.Liu,etal.,Measuringprotein–Ligandinteractionsusingliquidsample desorptionelectrosprayionizationmassspectrometry,Anal.Chem.85(24) (2013)11966–11972.
[31]K.A.Douglass,A.R.Venter,Proteinanalysisbydesorptionelectrospray ionizationmassspectrometryandrelatedmethods,J.MassSpectrom.48(5) (2013)553–560.
[32]J.M.Wiseman,etal.,Desorptionelectrosprayionizationmassspectrometry: imagingdrugsandmetabolitesintissues,Proc.Natl.Acad.Sci.U.S.A.105 (47)(2008)18120–18125.
[33]C.Yao,etal.,Screeningofthebindingofsmallmoleculestoproteinsby desorptionelectrosprayionizationmassspectrometrycombinedwithprotein microarray,J.Am.Soc.MassSpectrom.26(11)(2015)1950–1958.
[34]D.A.Annis,etal.,Affinityselection-massspectrometryscreeningtechniques forsmallmoleculedrugdiscovery,Curr.Opin.Chem.Biol.11(5)(2007) 518–526.
[35]G.J.VanBerkel,V.Kertesz,R.C.King,High-throughputmodeliquid
microjunctionsurfacesamplingprobe,Anal.Chem.81(16)(2009)7096–7101. [36]R.L.Edwards,etal.,Hemoglobinvariantanalysisviadirectsurfacesamplingof
driedbloodspotscoupledwithhigh-resolutionmassspectrometry,Anal. Chem.83(6)(2011)2265–2270.
[37]R.Edwards,etal.,Top-downproteomicsanddirectsurfacesamplingof neonataldriedbloodspots:diagnosisofunknownhemoglobinvariants,J.Am. Soc.MassSpectrom.23(11)(2012)1921–1930.
[38]R.L.Edwards,etal.,Compoundheterozygotesandbeta-thalassemia: top-downmassspectrometryfordetectionofhemoglobinopathies, Proteomics14(10)(2014)1232–1238.
[39]R.L.Griffiths,etal.,Liquidextractionsurfaceanalysisfieldasymmetric waveformionmobilityspectrometrymassspectrometryfortheanalysisof driedbloodspots,Analyst140(20)(2015)6879–6885.
[40]J.Sarsby,etal.,Top-downandbottom-upidentificationofproteinsbyliquid extractionsurfaceanalysismassspectrometryofhealthyanddiseasedhuman livertissue,J.Am.Soc.MassSpectrom.25(11)(2014)1953–1961.
[41]J.Sarsby,etal.,Liquidextractionsurfaceanalysismassspectrometrycoupled withfieldasymmetricwaveformionmobilityspectrometryforanalysisof intactproteinsfrombiologicalsubstrates,Anal.Chem.87(13)(2015) 6794–6800.
[42]RianL.Griffiths,etal.,LESAFAIMSmassspectrometryforthespatialprofiling ofproteinsfromtissue,Anal.Chem.UnderRevision(2016).
[43]R.L.Griffiths,etal.,Formallithiumfixationimprovesdirectanalysisoflipids intissuebymassspectrometry,Anal.Chem.85(15)(2013)7146–7153. [44]E.C.Randall,J.Bunch,H.J.Cooper,Directanalysisofintactproteinsfrom
Escherichiacolicoloniesbyliquidextractionsurfaceanalysismass spectrometry,Anal.Chem.86(21)(2014)10504–10510.
[45]C.-C.Hsu,P.-T.Chou,R.N.Zare,Imagingofproteinsintissuesamplesusing nanospraydesorptionelectrosprayionizationmassspectrometry,Anal. Chem.87(22)(2015)11171–11175.
[46]P.J.Roach,J.Laskin,A.Laskin,Nanospraydesorptionelectrosprayionization: anambientmethodforliquid-extractionsurfacesamplinginmass spectrometry,Analyst135(9)(2010)2233–2236.
[47]N.Martin,etal.,Nativeliquidextractionsurfaceanalysismassspectrometry: analysisofnoncovalentproteincomplexesdirectlyfromdriedsubstrates,J. Am.Soc.MassSpectrom.2015(2016)1–8.
[48]R.L.Griffiths,H.J.Cooper,Directtissueprofilingofproteincomplexes:toward nativemassspectrometryimaging,Anal.Chem.88(1)(2016)606–609. [49]M.B.vanEldijk,etal.,EvidencethatthecatenaneformofCS2hydrolaseisnot
anartefact,Chem.Commun.49(71)(2013)7770–7772.
[50]E.Reading,etal.,Theroleofthedetergentmicelleinpreservingthestructure ofmembraneproteinsinthegasphase,Angew.Chem.Int.Ed.54(15)(2015) 4577–4581.
[51]B.L.Schwartz,K.J.Light-Wahl,R.D.Smith,Observationofnoncovalent complexestotheavidintetramerbyelectrosprayionizationmass spectrometry,J.Am.Soc.MassSpectrom.5(3)(1994)201–204.
[52]N.J.Martin,etal.,Nativeliquidextractionsurfaceanalysismassspectrometry: analysisofnoncovalentproteincomplexesdirectlyfromdriedsubstrates,J. Am.Soc.MassSpectrom.2015(2016)1–8.
[53]M.F.Bush,etal.,Collisioncrosssectionsofproteinsandtheircomplexes:a calibrationframeworkanddatabaseforgas-phasestructuralbiology,Anal. Chem.82(22)(2010)9557–9565.
[54]M.B.vanEldijk,etal.,EvidencethatthecatenaneformofCS2hydrolaseisnot anartefact,Chem.Commun.49(71)(2013)7770–7772.
[55]J.L.P.Benesch,etal.,Proteincomplexesinthegasphase:technologyfor structuralgenomicsandproteomics,Chem.Rev.107(8)(2007)3544–3567. [56]V.A.Mikhailov,etal.,Mass-selectivesoft-landingofproteinassemblieswith
controlledlandingenergies,Anal.Chem.86(16)(2014)8321–8328. [57]A.Dyachenko,etal.,Allostericmechanismscanbedistinguishedusing
structuralmassspectrometry,Proc.Natl.Acad.Sci.U.S.A.110(18)(2013) 7235–7239.
[58]E.Morag,E.A.Bayer,M.Wilchek,Immobilizednitro-avidinand
nitro-streptavidinasreusableaffinitymatricesforapplicationinavidin-biotin technology,Anal.Biochem.243(2)(1996)257–263.
[59]C.Brauckmann,etal.,Theinteractionofplatinum-baseddrugswithnative biologicallyrelevantproteins,Anal.Bioanal.Chem.405(6)(2013)1855–1864. [60]D.M.Mock,M.I.Malik,Distributionofbiotininhumanplasma:mostofthe