BIROn - Birkbeck Institutional Research Online
Sivalingam, G.N. and Yan, Jun and Sahota, Harpal and Thalassinos,
Konstantinos (2013) Amphitrite: a program for processing travelling wave
ion mobility mass spectrometry data.
International Journal of Mass
Spectrometry 345 , pp. 54-62. ISSN 1387-3806.
Downloaded from:
Usage Guidelines:
Please refer to usage guidelines at
or alternatively
ContentslistsavailableatSciVerseScienceDirect
International
Journal
of
Mass
Spectrometry
j o ur na l ho me 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
Amphitrite:
A
program
for
processing
travelling
wave
ion
mobility
mass
spectrometry
data
Ganesh
N.
Sivalingam
a,
Jun
Yan
a,
Harpal
Sahota
b,
Konstantinos
Thalassinos
a,∗aInstituteofStructuralandMolecularBiology,DivisionofBiosciences,UniversityCollegeLondon,London,UK bInstituteofStructuralandMolecularBiology,Crystallography,BirkbeckCollege,London,UK
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received3June2012
Receivedinrevisedform4September2012 Accepted13September2012
Available online 6 October 2012
Keywords:
Ionmobilitymassspectrometry Travellingwaveionmobility Software
Dataprocessing
a
b
s
t
r
a
c
t
Sincetheintroductionoftravellingwave(T-Wave)-basedionmobilityin2007alargenumberofresearch laboratorieshaveembracedthetechnique,particularlythoseworkinginthefieldofstructuralbiology. Thedevelopmentofsoftwaretoprocessthedatageneratedfromthistechnique,however,hasbeen limited.WepresentanovelsoftwarepackagethatenablestheprocessingofT-Waveionmobilitydata. Theprogramcandeconvolutecomponentsinamassspectrumandusesthisinformationtoextract correspondingarrivaltimedistributions(ATDs)withminimaluserintervention.Itcanalsobeusedto automaticallycreateacollisioncrosssection(CCS)calibrationandapplythistosubsequentfilesof inter-est.Anumberofapplicationsofthesoftware,andhowitenhancestheinformationcontentextracted fromtherawdata,areillustratedusingmodelproteins.
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
Ionmobilityisagas-phasetechniquethatseparatesionsasthey travelthroughacounterflowingneutraltargetgasunderthe influ-enceofanappliedelectricfield.Thetimeittakesaniontotraverse thecellisrelatedtoitsmass,charge,andtherotationallyaveraged collisioncrosssection(CCS)ofanion[1–3].Ionmobilitycoupledto massspectrometry(IM-MS)isapowerfulanalyticaltechniquethat wasinitiallyonlyavailableinafewlaboratoriescapableof build-ingsuchspecialisedinstruments.Theprimarymeansofperforming IM-MSseparationswasbasedondriftcelltechnology[4].
Shortlyafterthedescriptionofacommercialinstrumentthat wasmodified forIM-MSmeasurements[5],theintroductionof travellingwave(T-Wave)ionmobilityseparation[6],incorporated in a commercial quadrupole time-of-flight instrument (Synapt HDMS, WatersCorp.) [7],furtherpopularised thetechnique. In additiontothehighmassaccuracyobtainable,theSynaptcanbe usedtocarryoutionmobility-tandemmassspectrometry exper-imentsbyperformingcollisioninduceddissociation(CID)before and/orafter themobility cell. A second generationinstrument, theSynaptG2,wasintroducedin2011withanuptofour-fold increasein theT-Waveionmobility resolution,asexpressedin terms of / [8],where is the rotationallyaveraged CCS. AnotherattractivefeatureoftheSynaptinstrumentsisthatthey canbemodifiedforhighmassoperationbytheincorporationofa
∗Correspondingauthor.Tel.:+442076792197;fax:+442076797193. E-mailaddress:[email protected](K.Thalassinos).
32Kquadrupole,allowingtheselectionandtransmissionofhigh
m/zspecies.
T-Waveionmobilitymassspectrometry(TWIM-MS)hassofar beenusedtostudyanumberofsyntheticandbiologicalmolecules suchaspolymers[9–11],carbohydrates[12],peptides[13,14]and lipids[15,16].Themajorityof applications,however,have been withinthestructuralbiologyfieldasTWIMS-MShasclear advan-tagesoverotherestablishedtechniqueswithinthisarea.Proteins thatexhibittoomuchconformationalflexibilityorthataretoolarge tostudybyestablishedtechniquessuchasX-rayandNMR respec-tively,canstillbeamenabletoanalysisbymeansofTWIM-MS.In addition,TWIM-MScanbeusedtoseparateandstudyco-existing populationspresent insolution[17] incontrasttothemajority of otherbiophysical techniquesthat can onlyprovide informa-tionregardingthepopulationaverage.TWIM-MShasbeenused toprobetheconformationofsolubleproteinsandproteinsbound tovariousligands[18–22],protein complexes[23–26],proteins involvedinmisfoldingandaggregation[18,27,28]intactviruses,
[29,30],andmembraneproteins[31,32].InconjunctionwithCID, TWIM-MShasalsobeenusedtoprobethestructuralstabilityof suchmolecules[20,33,34].
Foralargenumberoftheapplicationsmentioned,thereisno requirementtoconvertarrivaltimedistributions(ATDs)toCCSin ordertoanswerthebiologicalquestionstudied.ObtainingaCCS, however,isessentialincaseswheretheCCSsareusedasaway offilteringcomputergeneratedmodels[35–37].ClassicalIM-MS instrumentationusesadriftcellmobilityseparationdevice.While thephysicalprinciplesbehinddriftcellIM-MSarewellunderstood andcanbeusedtoobtainaCCSforeachionstudied[38],thesame 1387-3806/$–seefrontmatter© 2012 Elsevier B.V. All rights reserved.
isnottruefortheT-Wave-baseddevice.Despiteinitialattemptsto characterisetheT-Wavedevice,[39]theionmotioninthedeviceis stillnotfullyunderstoodandcannotbeusedtoderiveCCSsdirectly fromthearrivaltime(td)data,however,anumberofprotocolshave
beendevelopedwhichallowthecalibrationoftheT-Waveagainst standardsofknowncrosssection[13,22,25,40].Anumberofsuch standardsareavailableintheformofpeptides[41],proteins[41], proteincomplexes[42,43]anddrug-likemolecules[44].
DespitetheadvancesinbothTWIM-MSinstrumentation devel-opmentandthegrowingapplications,advancesinthesoftwareto processsuchdatahasbeenlimited.Theonlysoftwarecurrently availableis Driftscope (Waters Corp.)which involvesextensive manualuserinteraction.Auserhastoidentifythepeaksinthe massspectrum,whichcanbechallengingespeciallywhen deal-ingwithspectracontainingmorethanonecomponentssuchas heterogeneousproteincomplexes,thenusethesetoreconstruct thecorrespondingATD.Fromthisdistributionthedrifttime(s)of maximumintensityareextractedforfurtheranalysis.Thismanual interventioncanbelabourintensiveandcanalsointroduceerrors intheanalysis.Whileprogramstoprocessintactproteinand pro-teincomplexMSdata[45–48]areappearingintheliterature,there isstillnoprogramfortheautomaticprocessingofTWIM-MSdata. Inthisworkwepresentanovelsoftwarefortheprocessingof TWIMS-MSdata.Thesoftwareautomatesthedeconvolutionofthe MSdataandautomaticallyextractsATDsfromtherawdatafiles.It alsoallowsforthefacilecreationofacalibrationthatcanthenbe appliedtoentiredatasetsautomatically.Thesoftwarecanbeused tocreateCCSvs.m/zheatmapsthatcanbeoverlaidbetween dif-ferentexperimentalconditions,somethingthatallowsforamore in-depthprobingofthestructuralchangestakingplacebetween differentconditions.Havingaprogramdotheseanalysesallows forthestandardisationofthedataprocessing,makingtheentire processmoreobjectiveandreproduciblebetweendifferent practi-tioners.Anumberofdifferentusesoftheprogram,withaparticular focus,oncommonlyencounteredstructuralbiologyapplications areillustratedusingmodelproteins.
2. Materialsandmethods
2.1. Samplesourcesandpreparationprocedures
cytochrome c from equine heart, myoglobin from equine heart, alcohol dehydrogenase (ADH) from Saccharomyces cere-visiae, bovine serum albumin (BSA), and concanavalin A from
CanavaliaensiformiswerepurchasedfromSigmaAldrich(St.Louis, MO).SerumamyloidPcomponent(SAP)fromhumanserumwas purchasedfromCalBioChem,MerckBiosciencesLtd.(Darmstadt, Germany).For nativeexperiments,protein sampleswerebuffer exchanged into 250mM ammonium acetate, and concentrated to20MusingAmiconUltra0.5mlcentrifugalfilters(Millipore UKLtd,Watford,UK).Fordenaturingexperiments,protein sam-pleswerebufferexchangedintoa49:49:2(v:v:v)ratioofH2O:
methanol:aceticacid,andconcentratedto20MusingAmicon Ultra0.5mlcentrifugalfilters.
2.2. TWIMS-MS
Mass spectrometry experiments were carried out ona first generationSynapt HDMS (Waters Corp., Manchester, UK)mass spectrometer[7]. The instrument wasmass calibrated using a 33MsolutionofCesiumIodidein250mMammoniumacetate. 2.5laliquotsofsamplesweredeliveredtothemass spectrome-terbymeansofnanoESIusinggold-coatedcapillaries,preparedin house[49].Typicalinstrumentalparameterswereasfollowsunless otherwise specified: sourcepressure 5.5mbar, capillary voltage
1.10kV, conevoltage 40V,trapenergy 8V, transferenergy 6V, biasvoltage15V.IMSpressure5.18×10−1mbar,IMSwavevelocity
250m/s,IMSwaveheight6V,andtrappressure4.07×10−2mbar.
2.3. Experimentalprocedures
cytochromecwasanalysedwithabiasvoltageof30V,with denaturedmyoglobinbeingusedasacalibrantforobtainingCCS. FortheheatingexperimentADHwasincubatedat60◦Cfor30min inaheatblock.Thesamplewasremovedfromtheheatblockand immediatelyintroducedtothemassspectrometer.Instrumental parameterswereoptimisedasfollows:sourcepressure4.50mbar, conevoltage60V,trapenergy15V,transferenergy12V,andIMS waveheight7V.BSAandconcanavalinAwereusedasCCS cal-ibrants.Forthecollisionunfoldingexperimentthenativefoldof cytochromecwasdisruptedbyincreasingthebiasvoltageby10V atatime,from10V,untilreaching80V.Instrumentalparameters wereoptimisedasfollows:sourcepressure3.55mbar,conevoltage 30V,andIMSwaveheight7V.DenaturedmyoglobinandADHwere usedasCCScalibrants.Forthemixingexperiment,ADH,BSA,and concanavalinAweremixedinanequimolarratio,andthe instru-mentalparameterswereoptimisedasfollows:trapenergy60V, transferenergy30V,biasvoltage22V.IMSwaveheight7V.SAP, BSA,andconcanavalinAwereusedasCCScalibrants.
2.4. Softwaredevelopment
DuringaTWIM-MSexperimentionarrivaltimedistributions (ATDs)arerecordedbysynchronizingtheoa-TOFacquisitionwith thegatedreleaseofapacketofionsfromthetrapT-Wave.Foreach packetofions200massspectraareacquiredataratedependent onthepusherfrequency.
Amphitritehandlesthedatain theformof a n×200matrix (wherenisthenumberofm/zbin increments),withindividual vectorstodescribetheassociatedaxes.Thismatrixcanbeusedto generatethefullmassspectrumofallarrivaltimesbysumming downtoan-lengthvector.Additionalmanipulationscanbe car-riedoutbyselectingsectionsofthematrixbyindex,forexample thearrivaltimesofaparticularioncouldbeextractedby supply-ingthelowerandupperm/zlimits,andthensummingalongthe
m/zaxis.Themanipulationsofthismatrixformsthebasisofthe functionalityoftheprogram.
Thesoftwarewasdeveloped usingthePythonprogramming language[50].SeveralPythonmoduleswereutilisedfordata anal-ysisNumPy,SciPy[51]andMatplotlib[52],andthegraphicaluser interfacewasdevelopedusingwxPython[53].Theinitial conver-sionofarawTWIM-MSfiletoanAmphitriteprojectfilecanonly berununderMicrosoftWindows,however,allotheraspectsof Amphitriteare cross platformcompatible and installerbinaries for Linux and Mac OS X systems are available on thewebsite
http://www.homepages.ucl.ac.uk/∼ucbtkth/amphitrite.html. The softwarewasdevelopedona3.4GHzquad-coreprocessormachine with 16GB memory running Ubuntu 12.04. Processing times quotedarefora2011MacBookAirwitha1.7GHzdual-core pro-cessorand4GBmemory.
3. Resultsanddiscussion
3.1. Massspectrumsimulation
Programscapableofautomaticandsemi-automaticanalysisof mass spectrometrydata ofproteins and proteincomplexes are becomingincreasinglyavailable[45–48].Amphitritealsoincludes analgorithmforthedeconvolutionofmassspectra,asfittingpeaks tothemassspectrumisthefirstprocessintheautomatic extrac-tionofcorrespondingATDstothesepeaks.AGaussianmodel(Eq.
(1))isusedtorepresentthedistributionofpeak heightsofthe ionpeakswithinachargestatedistributionofagivenmolecular species,andthisisusedasaconstraintinmassspectralsimulations. Theindividualpeaks,correspondingtoaspecificchargestate,can bemodelledasGaussian(Eq.(1)),Lorentzian(Eq.(2))orahybrid peakshapewhichconsistsofGaussianandLorentzianregions(Eq.
(3))whereAistheamplitude,isthemean,andisthefullwidth halfmaximum(FWHM)ofthepeak.
f(x)=Ae−(x−)2/2
/2√2ln2
2
(1)
f(x)=A 1
[1+((−x)/(/2))2] (2)
f(x)=
⎧
⎪
⎨
⎪
⎩
Ae−(x−)2/2(/2√2ln2)2 :x≤
A 1
[1+((−x)/(/2))2] :x>
(3)
Throughouttheexamplespresentedherethehybridpeakshape wasusedandthemodelusedtogenerateacompletechargestate seriesis describedin Eq.4. z0 andzn representthelowest and
highestchargestateintheseries,Az,zandzrepresentthe
ampli-tude,meanandFWHMparametersforthechargestatedistribution GaussianrespectivelyandH+isthemassofaproton.Additionally
themasshasbeendenotedas“mass”todistinguishmass/zifrom
mass-to-chargeratio(m/z).
f(x)=
zn
zi=z0 Aze
−
(mass/zi)+H+
−z 22·
z/2√
2ln2
2·
⎧
⎪
⎪
⎨
⎪
⎪
⎩
e− (x−)
2
2·(/2√2ln2)2 :x≤ 1
[1+((x−)/(/2))2] :x>
(4)
Afterminimaluserinputtheprogramcansimulateamass spec-trumasshownin Fig.1C withacomputationalprocessingtime ofunder2s.Toassess thequalityofthefit anerrorstatisticis calculatedbysummingtheabsoluteerrorofalldatapointsand averagingperm/zunit.InthecaseofFig.1Ctheerrorwas0.47% (ofbasepeakintensity)perm/z.Therearetwowaysinwhichone canspecifytheinputrequired.Ifthemassofthecomponentsin thespectrumisknown,itcanbemanuallyentered,alongwiththe chargestaterangeoverwhichtosimulatethatparticularmasse.g., +22to+27(Fig.1D).Morethanonemasscanbeentered,andafter thistheprogramusesaleastsquaresoptimisationtominimisethe differencebetweenthesimulatedandexperimentaldata.If, how-ever,themassofthecomponentsinthespectrumisunknown,the programaidstheuserinthisprocess.Theprogramusesa gradi-entmethodtoautomaticallyidentifypeaks(wheref(m/z)=0and
f(m/z)<0)inthemassspectrumwhicharethengivenarbitrary uniquenumericalidentifiersasshowninFig.1E.
Theuserthenselectspeakscorrespondingtosequentialcharge statepeaksofaparticularspecies.Themassofthespeciesis cal-culatedusingthem/zvaluesofthepeaktops.Thetheoreticalm/z
valuesforchargestatesarecalculated(default1+to100+)andare displayedasverticalmarkersalongwiththecalculatedmassand error(Fig.1D).Bothofthesefeatureshelptoensurethatpeaks
Fig.1.Differentstagesinextractingarrivaltimedistributionplotsofserumamyloid P(pentamer).TheuserselectspeaksofthechargestateseriesinpanelE(numbersare uniquepeakidentifiers),themassiscalculatedandthetheoreticalchargestatesare thenplottedoverthespectrum(panelD),withthesubsequentsimulatedspectrum plottedinpanelC.PanelBshowsamassmobilityplot,andusingthesimulated spectrumeachchargestatecanbeidentifiedandtheATDsextracted.TheATDs havebeendisplayedasoverlaidATDdistributionsforeachchargestateasshownin panelA.
werecorrectlyidentified,asincorrectpeakpickingwouldresult inmisalignedtheoreticalchargestatesandlargemasserrors.The userthensuppliesthechargerangetosimulate,basedoncharge stateionpeakintensities.Afterthisprocesshasbeencompleted foreachspecies,theprogramcanfitsimulateddatatothesupplied spectrumusingleastsquaresoptimisationwiththeresultshownin
Fig.2.CreationofaCCScalibrationusingdenaturedmyoglobin.Amphitriteautomaticallyselectschargestates(panelA),whichcorrespondtopublishedCCS[41].Fromthe selectedpeaks,theATDsareextractedandplottedinpanelB,andthepeaktopsareautomaticallypickedanddisplayed.Acalibrationcurve,usingapowerfittothedata, isthencalculatedandplotted.Poorfitscanberecalculatedbymanuallyadjustingthepeaktopsselectedinthepreviousstage.Thecalibrationprocedureusedhasbeen describedin[13].
provideamoreaccuratemeasureofthepeakintensityfor overlap-pingpeaks.
3.2. ATDextraction
Standardm/zagainstarrivaltime(td)plotslikethosedisplayed
by Drifscope can be drawn by Amphitrite (Fig. 1B) and a key improvementistheresolutionoftheseimages.InAmphitritethe usercandeterminethespacebetweeneachdatapointinthem/z
spacei.e.,howwideaparticularm/zbinis.Forthefiguresshown hereaspacingof2m/zunitswasused.
ExtractingATDsacrossallchargestatesofagivenspectrumhas nowbeenstreamlinedasthefittingprocedurepreviouslydescribed determinestheFWHMandpeakcentreofeachofthepeaksinthe massspectrum,andusesthisinformationtoautomaticallyextract the corresponding ATD for each charge state, withthe results showninFig.1A.
Inexperimentswheremultiplespectraareobtainedofthesame proteinunderdifferentconditions,theATDscorrespondingtoa singlecharge statecanbeextractedforallthefilesinasimilar mannerasexemplifiedinFig.7.
3.3. Calibration
ProtocolstoconvertTWIM-MSarrivaltimestoCCShavebeen describedpreviously[13,22,25,40],andAmphitriteautomatesthis procedure,therebyreducing subjectivitythat canbeintroduced duringtheATDextractionandsubsequenttdpeakselection.
InFig.2,theprocessofcreatingacalibrationisshown.From auserinputperspective,theprogramisgiventhecalibrantraw datafileandthenameofthatcalibrant,inthiscase myoglobin. Creatingacalibrationwithmorethanonecalibrantproteinisalso
possible.It thenautomaticallyselectsthechargestates(vertical bandsinFig.2A)thathavecorrespondingpublishedCCSs[41,42]. Lowabundancechargestatepeakscanbedeselectedandignored inordertoimprovethefit.Theprogramautomaticallytakesthe highestintensitytdtouseinthecalibrationandproducestheoutput
showninFig.2C.Thecalibrationprocedureusedtocreatethisfigure hasbeenpublishedpreviously[13],however,alternative proce-dures[42]canalsobecalculatedusingAmphitrite.Outlierscanbe addressedbyspecifyingalternatepeaktops(whicharealso auto-maticallydetected),byspecificallyprovidingthetdasinput,orby
removingitfromthecalibration.
3.4. Applyingacalibration
Theabilitytoreadtherawdatafileshasallowedamorefine grainedapproachtoapplyingacalibrationtoTWIM-MSdata.Since theprogramcanidentifythepeakspresentintheMSdimension andcalculatethecorrespondingchargestateforeachpeak,a cal-ibrationcanbeappliedtoeachm/z“slice”ratherthanonlytoa specifictd.Herewerefertoam/z“slice”aspartofthematrixthat
holdstheextractedrawdatasuchthatM[i−j][1−200],whereiis thelowestm/zandjisthehighestm/zdescribingapeakintheMS. Thecalibrationcalculationdependsonthem/z,zandtdofadata
point.Insteadofextractingapeakandapplyinganoverall calibra-tion,theprogramrecalculatesthecalibrationforeachm/zvaluein thedataset,whichcouldincreasetheprecisionoftheCCSvalues determined,aspeakswithlargeramountsofadductionwouldhave theadditionalmasscorrectedfor.
Fig.3. Chargestatecollisioncrosssectionplotsofcytochromec.Acalibrationsimilar tothatshowninFig.2wasappliedtothetddatatogeneratetheCCSdata.Adotin
panelArepresentseachpeaktopinthetdforthatparticularchargestate.Thesame
dataareshowninpanelBasaheatmapwithpeakintensitybeingrepresented bythecolourintensityandtheCCSofpeaktopsshownasawhitedash.PanelC showsthesedatanormalisedbyindividualpeakvolume.(Forinterpretationofthe referencestocolourinthisfigurelegend,thereaderisreferredtothewebversion ofthisarticle.)
conformationexistsfora givenchargestate,suchasforthe+8 chargestateshowninFig.3,thereisnowayofdeducingthe rela-tiveintensitiesofthedifferentconformations.Itisalsonotpossible toinferthewidthofeachconformationintheCCSdirection. Bio-logically,thiscanbeveryinformativeasanincreaseinthewidth ofaCCSdistributioncanindicateincreasedconformational flexi-bility[27,32]andasshownrecentlycan,incertaincases,alsolimit theobservedATDresolutionofhigherresolvinginstruments[37]. Comparingtwoproteinsbyoverlayingfigures likethose shown inFig.3Acanmissimportantconformationalchangesasthepeak CCScanremainthesame,whilethewidthofeachCCScanchange betweentwoconditions[27].ThefiguresgeneratedbyAmphitrite (Fig.3BandC)alsoprovideinformationregardingthewidthofthe MSpeak(inthex-axisdirection)thatwasusedtoreconstructthe ATDs.TheprogramdisplaystheCCSdimensionpeaktops,asshown inFig.3A,andthesearecalculatedautomatically(wheref()=0 andf()<0).
Featuresoflowabundancecanbevisualisedbyusingadifferent methodofnormalisingthepeakintensities.InFig.3Bthecolour isnormalisedtotheintensityofthebasepeakinthemass spec-trum,i.e.,tothemaximumintensityintheentirematrixholding
Fig.4. IM-MSanalysisofamixtureofBSA,concanavalinAandalcohol dehydroge-nase.Themassspectrumwasdeconvolutedintoitscomponentparts(panelC),with therawarrivaltimedistributionshowninpanelB.Usingthedeconvolutiondataand CCScalibration(likethatshowninFig.2),therawarrivaltimescanbeseparated andconvertedintoCCSvs.m/zinformationforeachmolecularcomponent(panel A).ThecolouringisconsistentbetweenpanelAandC(concanavalinAmonomer– red,dimer–blue,tetramer–purple,BSAmonomer–green,dimer–brown,ADH tetramer–magenta).
thedata,sothatonecanseethatthe+7chargestateisthemost intensepeakinthemassspectrum.InFig.3Ctheintensityofeach chargestateisnormalisedtothetotalintensityforthatm/z“slice”. Thisallowsforlessabundantfeaturestobevisualisedby increas-ingthedynamic rangedisplayforeachchargestateand forthe conformationalflexibilityandadductiontobereadilyassessed.
3.5. Complexmixtureanalysis
Fig.5.Spectralaveraging.PanelA,showsthevariationforthesamesamplemeasuredusingdifferentcapillaryneedles.Eachindividualspectrumhasbeenoverlaidand coloureddifferently.Apeakatapproximately5150m/zhasbeenenlargedtoportraymoreclearlythevariationbetweeneachexperiment(panelB).Theaverageofthethree spectrainpanelAisplottedinpanelC.ThesameenlargedpeakinpanelBisagainshowninpanelD,withtheminimumandthemaximumofthethreespectraplottedas lightbluelinesandthemeanplottedinblack.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
easily(seeFig.4).TotesttheperformanceofAmphitritewhen deal-ingwithcomplexmixturesanequimolarmixtureofbovineserum albumin,concanavalinA,andalcoholdehydrogenasewasprepared. AsseeninFig.4theprogramsuccessfullydeconvolutesthe spec-trumbyidentifyingandcalculatingthemassandthechargestate distributionparametersforallspecies.Additionallytheindividual
ionpeakwidthsaredetermined(Fig.4C).Atypicaltdvs.m/zplot
isshowninFig.4Bandwhenobservedinisolationdoesnot eas-ilyallowonetoidentifythenumberofspeciespresent.Usingthe parametersdeterminedinFig.4C,acalibrationliketheoneshown inFig.2canbeapplied,transformingthetdvs.m/zplotintoaCCS
vs.m/zplot(Fig.4A).Thisconversionintoabsolutecrosssection
Fig.6. Comparisonofheatingexperimentsofalcoholdehydrogenaseat20◦Cand60◦C.DataateachtemperaturewasreplicatedandaveragedasshowninFig.5.PanelA
andBshowthedistributionofcollisioncrosssectionsfor20◦Cand60◦Crespectively.AdifferenceplotisshowninpanelC,whereoverlappingCCSdistributionsinpanelA
Fig.7. Collisioninducedunfoldingofthe6+chargestateofcytochromec.TheCCSdistributionofthe6+chargestateisshowninpanelAasthecollisionenergyisincreased from10Vto80Vat10Vincrements.IntensitiesforpanelAhavebeennormalisedtothetotalionintensityforeachthreedimensionalpeak.ThecorrespondingCCSplotfor eachvoltageincrementisshowninpanelB.TherelativeintensityofeachpeaktopidentifiedfromtheCCSplotsisalsomonitoredasthebiasvoltageisincreased(panelC).
separatesouttheindividualspeciesandresultsinaplotwhichcan bemorereadilyanalysed.
3.6. Spectralaveraging
Collectingmultiplemassspectracanhelpreducetheerrorand variationcausedbycertainfactorssuchasneedle-to-needle vari-ation andneedle positioning.Fig.5A andBshows theeffectof obtainingthreemassspectraofthesamesampleusingthesame instrumental conditions but withthree different needles.From theanalysisofmassspectraofsamplesunderdifferentconditions (e.g.,temperature),peakintensitiesandareascanofferinformation additionaltothemassoftheions.
Bycomparingtheareasofdifferentoligomericspeciesunder differingconditions (e.g.,temperature), theformation or disap-pearanceofoligomersinresponsetoconditionsofinterestcanbe inferred.Itisadvantageoustobeabletoaveragetechnicalreplicates andcomparethosebetweenconditionsinordertoassesswhether changesareduetodifferingconditionsratherthantechnical vari-ability.
Usingtheprogramonecanaveragespectrainthemass spec-trum,tdandCCSspace.Fig.5Dshowsthebandbetweenminimum
andmaximumintensitiesforapeak,withtheaverageinthe cen-tre.Theprogramcanalsobesettodisplaytheerrorrangeinterms ofstandarddeviation,quartilerangesandpercentageerrors.Peak areasand heights canbeautomaticallyextractedtobe usedin furtheranalyses.
3.7. Comparingdifferentconditions
Spectralaveragingwasperformedontheheatingexperiments usedinFig.6.IM-MScanbeusedtomonitortheeffecton confor-mationofadiscreetstressorsuchasheating[18].Todemonstrate thisweacquiredaspectrumofADHatroomtemperature(20◦C), andafterheatingat60◦Cfor30min.Foreachconditionthree tech-nicalreplicateswereacquired.InadditiontotheCCSvs.m/zplot ofthesampleat20◦C(Fig.6A)and60◦C(Fig.6B),adifferenceplot canalsobedrawn.ThisisshowninFig.6C,withthecolours match-ingthoseusedinFig.6AandB.Inthisfigurethe20◦Cand60◦C spectrahavebeennormalisedtothevolumewithintheplot,as thishelpstomakethecomparisonmorerepresentative.Thedata showthattheprocessofheatingADHcausesittoadoptamore openconformationasdemonstratedbytheincreaseinCCS.
3.8. Collision-inducedunfolding
adifferentrawdatafileisrecorded.Thisiswheretheprocessing ofsuchdatasetsgreatlybenefitsfromtheautomationofferedby Amphitrite.
Fig.7showstheresultsofthecollisioninducedunfoldingforthe +6chargestateofcytochromec.Datawererecordedateight dif-ferentcollisionenergies.Theprogramusesthedatafilewhichwas acquiredatthelowestcollisionenergytoidentifyandperformafit (inordertocalculateaFWHMforthatpeak)tothemassspectrum aspreviouslydescribedinthematerialsandmethodssection. Alter-natively,massrangescanalsobeenteredmanually.Them/zrangeis thenusedtoautomaticallyextractATDsfromallfilesinthedataset. IfaCCScalibrationisprovidedatthisstage,allATDsareconverted toCCSvalues.Thereducedamplitudeforthepeaksseenfor 10 and20Vareduetothepeakbroadeningeffectsofboundadducts whichlessensasthecollisionenergyisincreased.Thishighlights thebenefitofthenewplot(Fig.7A);changesinboththem/zandthe ATD/CCSdimensionscanbevisualisedsimultaneouslyprovidinga meansofgloballymonitoringionstructuralchangesduring colli-sioninducedunfoldingexperiments.Finally,theprogramcantrack thepeakintensitiesofgivenCCSortdvaluesasshowninFig.7C.
4. Conclusions
AmphitritesubstantiallyenhancestheprocessingofTWIM-MS data,makingtheprocessautomatedandlesspronetouser sub-jectivity.Italso allowsfor a moredetailed analysisof thedata acquiredand thefacilecomparisonofentireTWIM-MSdatasets betweendifferentexperimentalconditions.Anumberofcommon applicationsinstructuralmassspectrometrywerepresented;we arehowever,planningfuturereleasesofthesoftwarewhichwill enableapplicationinotherfields.Suchanexamplewouldbeto combinethefunctionalityfoundinAmphitritewithavailable soft-warethatexpeditestheannotationoftandemmassspectrometry (MS/MS)datafromsyntheticpolymers[55],tobenefitresearchers workinginthefieldofpolymerstructurecharacterisation.Another areathatweexpectAmphitritetohavealargeimpactisinthe processingofTWIM-MSdataobtainedaspartoflargescale pro-teomicsexperiments,asrecentlyTWIM-MSseparationhasbeen coupledtoliquidchromatographyseparationandadata indepen-dentmodeofacquisition(MSE)[56].Ithasbeenpreviouslyshown
thatinamass-mobilityplotclassesofmoleculessuchas phospho-peptides,lipids,carbohydratesandnucleotidespopulatedifferent regionsofsuchaplot[57]andAmphitritewillallowforthe auto-maticclassificationofsuchcompoundclasses.
Acknowledgements
WewouldliketothankKeithRichardson(WatersCorp. Man-chester)forprovidinguswithhelpinaccessingtherawTWIM-MS datafile and Adam Cryar for helpfuldiscussions regarding the manuscript.ThisworkwassupportedbyanInstituteofStructural andMolecularBiologystartupgranttoK.T.,MRCstudentshipsto G.N.S.,H.S.,andaWellcomeTruststudentshiptoJ.Y.K.T.wouldlike tothankJimandKeithforintroducinghimtothewonderfulworld ofmassspectrometry.
References
[1]M.F.Jarrold,Peptidesandproteinsinthevaporphase,AnnualReviewof Phys-icalChemistry51(2000)179–207.
[2]C.S.Hoaglund-Hyzer,A.E.Counterman,D.E.Clemmer,Anhydrousproteinions, ChemicalReviews99(1999)3037–3080.
[3]G.vonHelden,T.Wyttenbach,M.T.Bowers,Conformationofmacromolecules inthegasphase:useofmatrix-assistedlaserdesorptionmethodsinion chro-matography,Science267(1995)1483–1485.
[4]A.B.Kanu,P.Dwivedi,M.Tam,L.Matz,H.H.Hill,Ionmobility-mass spectrom-etry,JournalofMassSpectrometry:JMS43(2008)1–22.
[5]K.Thalassinos,S.Slade,K.Jennings,J.Scrivens,K.Giles,J.Wildgoose,J.Hoyes,R. Bateman,M.Bowers,Ionmobilitymassspectrometryofproteinsinamodified commercialmassspectrometer,InternationalJournalofMassSpectrometry 236(2004)55–63.
[6]K.Giles,S.D.Pringle,K.R.Worthington,D.Little,J.L.Wildgoose,R.H. Bate-man,Applicationsofatravellingwave-basedradio-frequency-onlystacked ringionguide,RapidCommunicationsinMassSpectrometry:RCM18(2004) 2401–2414.
[7]S.D.Pringle,K.Giles,J.L.Wildgoose,J.P.Williams,S.E.Slade,K.Thalassinos,R.H. Bateman,M.T.Bowers,J.H.Scrivens,Aninvestigationofthemobilityseparation ofsomepeptideandproteinionsusinganewhybridquadrupole/travelling waveIMS/oa-ToFinstrument,InternationalJournalofMassSpectrometry261 (2007)1–12.
[8]K.Giles,J.P.Williams,I.Campuzano,Enhancementsintravellingwaveion mobilityresolution,RapidCommunicationsinMassSpectrometry:RCM25 (2011)1559–1566.
[9]C.A.Scarff,J.R.Snelling,M.M.Knust,C.L.Wilkins,J.H.Scrivens,New struc-turalinsightsintomechanicallyinterlockedpolymersrevealedbyionmobility massspectrometry,JournaloftheAmericanChemicalSociety134(22)(2012) 9193–9198.
[10]G.R.Hilton,A.T.Jackson,K.Thalassinos,J.H.Scrivens,Structuralanalysisof syn-theticpolymermixturesusingionmobilityandtandemmassspectrometry, AnalyticalChemistry80(2008)9720–9725.
[11]X. Li,Y.-T. Chan,G.R.Newkome,C. Wesdemiotis, Gradienttandem mass spectrometryinterfacedwithionmobilityseparationforthe characteriza-tionofsupramoleculararchitectures,AnalyticalChemistry83(2011)1284– 1290.
[12]D.J.Harvey,F.Sobott,M.Crispin,A.Wrobel,C.Bonomelli,S.Vasiljevic,C.N. Scan-lan,C.A.Scarff,K.Thalassinos,J.H.Scrivens,Ionmobilitymassspectrometryfor extractingspectraofN-glycansdirectlyfromincubationmixturesfollowing glycanrelease:applicationtoglycansfromengineeredglycoformsofintact, foldedHIVgp120,JournaloftheAmericanSocietyforMassSpectrometry22 (2011)568–581.
[13]K.Thalassinos,M.Grabenauer,S.E.Slade,G.R.Hilton,M.T.Bowers,J.H.Scrivens, Characterizationofphosphorylatedpeptidesusingtravelingwave-basedand driftcellionmobilitymassspectrometry,AnalyticalChemistry81(2009) 248–254.
[14]J.P.Williams,K.Giles,B.N.Green,J.H.Scrivens,R.H.Bateman,Ionmobility augmentstheutilityofmassspectrometryintheidentificationofhuman hemoglobinvariants,RapidCommunicationsinMassSpectrometry:RCM22 (2008)3179–3186.
[15]W.B. Ridenour, M.Kliman, J.A. McLean, R.M. Caprioli, Structural charac-terization ofphospholipids andpeptides directlyfromtissuesections by MALDItraveling-waveionmobility-massspectrometry,AnalyticalChemistry 82(2010)1881–1889.
[16]H.I.Kim,H.Kim,E.S.Pang,E.K.Ryu,L.W.Beegle,J.A.Loo,W.A.Goddard,I. Kanik,Structuralcharacterizationofunsaturatedphosphatidylcholinesusing travelingwaveionmobilityspectrometry,AnalyticalChemistry81(2009) 8289–8297.
[17]D.P. Smith,K.Giles,R.H.Bateman,S.E.Radford, A.E.Ashcroft,Monitoring copopulatedconformationalstatesduringproteinfoldingeventsusing electro-sprayionization-ionmobilityspectrometry-massspectrometry,JAM18(2007) 2180–2190.
[18]M.P.Nyon,L.Segu,L.D.Cabrita,G.R.Lévy,J.Kirkpatrick,B.D.Roussel,A.O.M. Patschull,T.E.Barrett,U.I.Ekeowa,R.Kerr,C.A.Waudby,N.Kalsheker,M.Hill,K. Thalassinos,D.A.Lomas,J.Christodoulou,B.Gooptu,Structuraldynamics asso-ciatedwithintermediateformationinanarchetypalconformationaldisease, Structure(London,England:1993)20(2012)504–512.
[19]T.Wyttenbach,M.Grabenauer,K.Thalassinos,J.H.Scrivens,M.T.Bowers,The effectofcalciumionsandpeptideligandsontherelativestabilitiesofthe calmodulindumbbellandcompactstructures,TheJournalofPhysical Chem-istry:B114(2010)437–447.
[20]M.R.Schenauer,J.A.Leary,Anionmobility-massspectrometryinvestigationof monocytechemoattractantprotein-1,InternationalJournalofMass Spectrom-etry287(2009)70–76.
[21]C.A.Scarff,V.J.Patel,K.Thalassinos,J.H.Scrivens,Probinghemoglobinstructure bymeansoftraveling-waveionmobilitymassspectrometry,Journalofthe AmericanSocietyforMassSpectrometry20(2009)625–631.
[22]C.A.Scarff,K.Thalassinos,G.R.Hilton,J.H.Scrivens,Travellingwaveion mobil-itymassspectrometrystudiesofproteinstructure:biologicalsignificanceand comparisonwithX-raycrystallographyandnuclearmagneticresonance spec-troscopymeasurements,RapidCommunicationsinMassSpectrometry:RCM 22(2008)3297–3304.
[23]J. Freeke,M.Bush,C. Robinson,Gas-phase proteinassemblies: unfolding landscapesandpreservingnative-likestructuresusingnoncovalentadducts, ChemicalPhysicsLetters524(2012)1–9.
[24]J.Freeke,C.V.Robinson,B.T.Ruotolo,Residualcounterionscanstabilisealarge proteincomplexinthegasphase,InternationalJournalofMassSpectrometry 298(2010)91–98.
[25]B.T.Ruotolo,J.L.P.Benesch,A.M.Sandercock,S.-J.Hyung,C.V.Robinson,Ion mobility-massspectrometryanalysisoflargeproteincomplexes,Nature Pro-tocols3(2008)1139–1152.
[26]B.T.Ruotolo,Evidenceformacromolecularproteinringsintheabsenceofbulk water,Science310(2005)1658–1661.
ofprionproteinsbymeansofdriftcellandtravelingwaveionmobilitymass spectrometry,JournaloftheAmericanSocietyforMassSpectrometry21(2010) 845–854.
[28]D.Smith,S.Radford,A.E.Ashcroft,Elongatedoligomersin2-microglobulin amyloidassemblyrevealedbyionmobilityspectrometry-massspectrometry, ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica 107(2010)6794.
[29]C.Uetrecht,I.M.Barbu,G.K.Shoemaker,E.vanDuijn,A.J.R.Heck,Interrogating viralcapsidassemblywithionmobility-massspectrometry,NatureChemistry 3(2011)126–132.
[30]T.W.Knapman, V.L. Morton,N.J.Stonehouse, P.G. Stockley,A.E.Ashcroft, Determiningthetopologyofvirusassemblyintermediatesusingionmobility spectrometry-massspectrometry,RapidCommunicationsinMass Spectrom-etry:RCM24(2010)3033–3042.
[31]S.C.Wang,A.Politis,N.DiBartolo,V.N.Bavro,S.J.Tucker,P.J. Booth,N.P. Barrera,C.V.Robinson,Ionmobilitymassspectrometry oftwotetrameric membraneproteincomplexesrevealscompactstructuresanddifferencesin stabilityandpacking,JournaloftheAmericanChemicalSociety132(2010) 15468–15470.
[32]M.Zhou,N.Morgner,N.P.Barrera,A.Politis,S.C.Isaacson,D.Matak-Vinkovic,T. Murata,R.A.Bernal,D.Stock,C.V.Robinson,MassspectrometryofintactV-type ATPasesrevealsboundlipidsandtheeffectsofnucleotidebinding,Science334 (2011)380–385.
[33]J.T.S.Hopper,N.J.Oldham,Collisioninducedunfoldingofproteinionsinthe gasphasestudiedbyionmobility-massspectrometry:theeffectofligand bindingonconformationalstability,JournaloftheAmericanSocietyforMass Spectrometry20(2009)1851–1858.
[34]J.A.Leary,M.R.Schenauer,R.Stefanescu,A.Andaya,B.T.Ruotolo,C.V. Robin-son, K.Thalassinos,J.H.Scrivens,M.Sokabe,J.W.B.Hershey,Methodology for measuring conformation of solvent-disrupted protein subunits using T-WAVEionmobilityMS: aninvestigationinto eukaryoticinitiation fac-tors, Journal of the American Society for Mass Spectrometry 20 (2009) 1699–1706.
[35]A.Politis,A.Y.Park,S.-J.Hyung,D.Barsky,B.T.Ruotolo,C.V.Robinson, Integrat-ingionmobilitymassspectrometrywithmolecularmodellingtodetermine thearchitectureofmultiproteincomplexes,PLoSONE5(2010)e12080. [36]M.R.Ninonuevo,J.A.Leary,Ionmobilitymassspectrometrycoupledwithrapid
proteinthreadingpredictorstructurepredictionandcollision-induced dissoci-ationforprobingchemokineconformationandstability,AnalyticalChemistry (2012),120316130633000.
[37]Y.Zhong,S.-J.Hyung,B.T.Ruotolo,Characterizingtheresolutionandaccuracy ofasecond-generationtraveling-waveionmobilityseparatorforbiomolecular ions,Analyst136(2011)3534–3541.
[38]E.A.Mason,E.McDaniel,TransportPropertiesofIonsinGases,Wiley,NewYork, 1988.
[39]A.A.Shvartsburg,R.D.Smith,Fundamentalsoftravelingwaveionmobility spectrometry,AnalyticalChemistry80(2008)9689–9699.
[40]D.P.Smith,T.W.Knapman,I.Campuzano,R.W.Malham,J.T.Berryman,S.E. Radford,A.E.Ashcroft,Decipheringdrifttimemeasurementsfromtravelling waveionmobilityspectrometry-massspectrometrystudies,EuropeanJournal ofMassSpectrometry15(2009)113–130.
[41]D.E. Clemmer, Clemmer Protein Database – http://www.indiana.edu/ clemmer/Research/cross%20section%20database/cs%20database.htm(2012).
[42]M.F.Bush,Z.Hall,K.Giles,J.Hoyes,C.V.Robinson,B.T.Ruotolo,Collisioncross sectionsofproteinsandtheircomplexes:acalibrationframeworkanddatabase forgas-phasestructuralbiology,AnalyticalChemistry82(2010)9557–9565. [43]R.Salbo,M.F.Bush,H.Naver,I.Campuzano,C.V.Robinson,I.Pettersson,T.J.D.
Jørgensen,K.F.Haselmann,Traveling-waveionmobilitymassspectrometry ofproteincomplexes:accuratecalibratedcollisioncross-sectionsofhuman insulinoligomers,RapidCommunicationsinMass Spectrometry:RCM 26 (2012)1181–1193.
[44]I.Campuzano,M.F.Bush,C.V.Robinson,C.Beaumont,K.Richardson,H.Kim,H.I. Kim,Structuralcharacterizationofdrug-likecompoundsbyionmobilitymass spectrometry:comparisonoftheoreticalandexperimentallyderivednitrogen collisioncrosssections,AnalyticalChemistry83(2011)8596–8603. [45]B.vanBreukelen,A.Barendregt,A.J.R.Heck,R.H.H.vandenHeuvel,Resolving
stoichiometriesandoligomericstatesofglutamatesynthaseproteincomplexes withcurvefittingandsimulationofelectrospraymassspectra,Rapid Commu-nicationsinMassSpectrometry:RCM20(2006)2490–2496.
[46]N.Morgner,C.V.Robinson,Massign:anassignmentstrategyformaximizing informationfromthemassspectraofheterogeneousproteinassemblies, Ana-lyticalChemistry84(2012)2939–2948.
[47]F.Stengel,A.J.Baldwin,M.F.Bush,G.R.Hilton,H.Lioe,E.Basha,N.Jaya,E. Vier-ling,J.L.P.Benesch,Dissectingheterogeneousmolecularchaperonecomplexes usingamassspectrumdeconvolutionapproach,Chemistry&Biology19(2012) 599–607.
[48]Y.-H.Tseng,C.Uetrecht,A.J.R.Heck,W.-P.Peng,Interpretingthechargestate assignmentinelectrospraymassspectraofbioparticles,AnalyticalChemistry 83(2011)1960–1968.
[49]H.Hernandez,C.V.Robinson,Determiningthestoichiometryandinteractions ofmacromolecularassembliesfrommassspectrometry,NatureProtocols2 (2007)715–726.
[50]G.VanRossum,VanRossum:PythonReferenceManual.PythonSoftware...– GoogleScholar,2006.
[51]T.Oliphant,Pythonforscientificcomputing,ComputinginScience& Engineer-ing9(2007)10–20.
[52]J.Hunter,Matplotlib:a2Dgraphicsenvironment,ComputinginScience& Engineering9(3)(2007)90–95.
[53]H.Talbot,wxPython,aGUIToolkit,LinuxJournal,2000.
[54]S.-J.Hyung,C.V.Robinson,B.T.Ruotolo,Gas-phaseunfoldinganddisassembly revealsstabilitydifferencesinligand-boundmultiproteincomplexes, Chem-istry&Biology16(2009)382–390.
[55]K.Thalassinos,A.T.Jackson,J.P.Williams,G.R.Hilton,S.E.Slade,J.H.Scrivens, Novelsoftwarefortheassignmentofpeaksfromtandemmass spectrome-tryspectraofsyntheticpolymers,JournaloftheAmericanSocietyforMass Spectrometry18(2007)1324–1331.
[56]E.Rodríguez-Suárez,C,Hughes,L.Gethings,K.Giles,J.Wildgoose,M.Stapels, K.E.Fadgen,S.J.Geromanos,J.P.Vissers,F.Elortza,J.I.Langridge,Anionmobility assisteddataindependentLC–MSstrategyfortheanalysisofcomplexbiological samples,CurrentAnalyticalChemistry,SpecialIssue.Ionmobility spectrome-try:usingsizeandshapetounderstandreal-worldsystemsatthemolecular level,HT-SBJ-CAC-0005(2012).
