BenchtopIRImagingofLiveCells:MonitoringtheTotalMassofBiomoleculesinSingleCells

Yow-RenChang,Seong-MinKim,andYoungJongLee*

BiosystemsandBiomaterialsDivision,NationalInstituteofStandardsandTechnology,Gaithersburg,MD20899,USA

Email:

youngjong.lee@

Abstract:Absolutequantityimagingofbiomoleculesonasinglecellleveliscriticalformeasurementassuranceinbiosciencesandbioindustries.Whileinfrared(IR)transmissionmicroscopyisapowerfullabel-freeimagingmodalitycapableofchemicalquantification,itsapplicabilitytohydratedbiologicalsamplesremainschallengingduetothestrongIRabsorptionbywater.TraditionalIRimagingofhydratedcellsreliesonpowerfullightsources,suchassynchrotrons,tomitigatethelightabsorptionbywater.However,weovercomethischallengebyapplyingasolventabsorptioncompensation(SAC)techniquetoahome-builtbenchtopIRmicroscopebasedonanexternal-cavityquantumcascadelaser.SAC-IRmicroscopyadjuststheincidentlightusingapairofpolarizerstopre-compensatetheIRabsorptionbywaterwhileretainingthefulldynamicrange.IntegratingtheIRabsorbanceoveracellyieldsthetotalmassofbiomoleculespercell.Wemonitorthetotalmassofthebiomoleculesoflivefibroblastcellsovertwelvehours,demonstratingpromiseforadvancingourunderstandingofthebiomolecularprocessesoccurringinlivecellsonthesingle-celllevel.

Characterizingandquantifyingbiomoleculesincellsiscriticalforunderstandingcellularfunctions,1macromolecularcrowding,2,3advancedbiomanufacturingoftherapeutics,4anddiseaseprogressionanddiagnostics.5,6However,thequantificationofabsolutemassinsinglelivecellsremainsameasurementchallenge.Traditionalbiochemicalassays7andmassspectrometry-basedanalysesareextremelysensitive,8buttheirensemble-based,destructiveapproachisunsuitableforcharacterizinghighlyheterogeneous,livecellsystems.Opticalimagingtechniquessuchasfluorescencemicroscopycandiscriminateindividualcellswithhighmolecularspecificitytotargetmarkers.Still,thequantificationofcell-consistingmoleculesiscomplicatedbyalackofcontrollabilityinlabelingefficiencyandphotobleaching.9,10Alternatively,spontaneousorcoherentRamanimaging11–13canidentifymoleculeswithoutlabelingbyvibrationalsignatures;however,weakemissionsignalsaresystem-dependentand,thus,yieldonlyrelativequantitationandrequireinternalreferencesofknownconcentrationsforabsolutequantification.13

Infrared(IR)absorption-basedimagingcansimultaneouslyidentifylabel-freeandquantifytheabsoluteconcentrationsofmoleculesofacell,includingproteins,fattyacids,andnucleicacids,bydetectingtheirspectralsignatureswithrelativelyhighcross-sections.14Moreover,themeasuredvalue,absorbance,issystem-independent,makingthevalueSI(internationalsystemofunits)-traceableand,thus,interlaboratory-comparable.However,afundamentalchallengeinIRabsorptionmeasurementsofbiologicalmoleculesisthebroadandstrongIRabsorptionbywater.15Inparticular,thewaterbendingmodeat1650cm-1overlapswiththeamideIabsorptionpeak,acharacteristicpeakforproteinquantificationandsecondarystructureanalysis,16makingIRunsuitableforquantifyingthiskeybiomolecule.

ThischallengetoIRimaginghasbeenpassivelymitigatedeitherbyreducingthetransmissionpathlength,17,18employingattenuatedtotalreflectance(ATR)configurations,19,20orusingasynchrotronforincreasedlightintensity.21However,thinnersamplingchamberssufferfromphysicalcompressionofcellsanddifficultyinmicrofluidiccontrolforlivecellimagingapplications.DespitetheeasycouplingofATRdetectionwithalive-cellchamber,ATR-IRcansenseonlyafewmicrometersnearasubstratesurface,representingonlypartofacell.BrightsynchrotronsourcestoenabletransmissionthroughwaterhavebeenusedforlivecellIRimaging.22–24However,thelimitedaccessibilityhindersthebroadapplicabilityofIRimaging.ChanandKazarianreportednon-synchrotrontransmissionFourier-transformIR(FT-IR)imagesoflivecellsinmicrofluidicchannelsusingapairofhemispherelenses.25,26However,theydidnottracklivecellstomonitortheirspectralvariationsovertime.

Recently,readilyavailablebenchtopexternal-cavityquantumcascadelasers(EC-QCLs),whichemitmonochromatictunablelightinmid-andfar-IR(800–3000cm-1),haveenablednewdiscretefrequencyinfraredmicroscopedesigns.27,28Notably,QCL-basedIRabsorptionmicroscopydemonstrateshighspeed,aberrationcorrection,andhighsignal-to-noise(SNR).29,30OtherQCL-basedIRmicroscopymethodsbasedonindirectphotothermalIRsignalshavebeenreported.31,32Forexample,IR-pumpvisible-probe(refraction32–34andfluorescence32,35,36)approachesenablednon-contactIRabsorptionimagingwithsub-micrometerspatialresolutionandwereusedforlivecellimaging.However,thesignalchangescausedbyaphotothermaleffectaresample-andsystem-dependentand,thus,arenotreadilyconvertibletoabsoluteconcentrationsormass.

Ontheotherhand,adirecttransmission-basedIRimagingapproachsufferschallengesoriginatingfromthestrongIRabsorptionbywaterinthemid-IRregion.ThetransmittedlightintensityvariesbymorethanathousandtimesneartheamideIband.However,itischallengingtoadjustthedynamicrangeofthedetectionsystembymultipleordersofmagnitudeduringrapidwavelengthscanning.Toaddressthiscriticallimitation,werecentlyintroducedasimpleopticaltechniquethatcouldadjusttheincidentlightintensityrapidlywithagreatdynamicrange.37Themethod,calledsolventabsorptioncompensation(SAC),modifiedtheincidentlighttoequalizethetransmissionspectrumthroughareferenceorsolvent.Thus,SACcouldusethemaximumdynamicsrangeovertheentirescanningrange,eliminatingwaterabsorptioncontributionopticallyandenhancingSNRby>100timesfortheamideIabsorptionpeakcomparedtoaconventionalIRtransmissionmeasurement.37InsubsequentSAC-IRspectroscopypapers,38,39weusedacousto-opticmodulators(AOMs)toadjustthelightintensityrapidlyduringwavelengthscanning.However,AOM-basedunitsforSACrequirebulkyfootprintsandcansufferfromwavelength-dependentbeamdirectionshifts,whichcanposeasignificantchallengetohyperspectralimaging.Thus,inthepresentwork,weintroduceasetofrotatingandfixedpolarizersforSACcontrol.40BecausetheresponseoftherotationalstageofapolarizerisslowerthantheAOMresponse,thistwo-polarizermethodmaybeunsuitableforveryrapidwavelength-scanningspectroscopysystems.However,thedeviation-freeapproachusingtwoflatpolarizerscanworkforamicroscopysystemwherespatialscanning(X,Y)precedesspectralscanning(w).Thetwo-polarizerSACunitiscoupledtoahome-builttransmissionQCL-IRmicroscopeinasample-scanningmodewiththeentirelaserwavenumberrangespanningfrom900cm-1to1776cm-1.Thiswiderangecoversthefingerprintpeaksofmultiplebiomolecules,includingprotein,fattyacid,andnucleicacid,insideasinglecell.WecompareproteinmasspercellbetweenfixedandlivecellsusingtheamideIabsorptionpeakofproteinsnear1650cm-1,whichwasnotavailablebyconventionalnon-SAC-IRapproaches.Wecharacterizesystemperformanceanddemonstratethatasimple,directtransmission-basedIRmeasurementcanbeusedforlive,intercellularabsoluteproteinmassmeasurements.

Figure1.SchematicofsolventabsorptioncompensationIR(SAC-IR)microscopy.(a)OpticalsystemoftheSAC-IRmicroscopysetup.Thelaserintensityiscontrolledasafunctionofwavelengthbyapairofrotatingandfixedpolarizers.EC-QCL:external-cavityquantumcascadelaser;RP:rotatingpolarizer;FP:fixedpolarizer;andLN-MCT:liquidnitrogen-cooledmercury-cadmium-telluridedetector.Theinsetillustratesamicrofluidicsamplechamberconsistingoftwo1-mmthickCaF2windowsanda25μmspacer.OneoftheCaF2windowshasdrilledholesforintroducingcellsandmedia.(b,c)Spectraoftransmittedlightintensity(IT)inafixedfibroblastcellregion(red)andaphosphatebuffersaline(PBS)region(blue)withoutSAC(b)andwithSAC(c).Theacquisitiontimeforalinespectrum(X,w)was5min.(d,e)AbsorbancespectraofaPBS(reference,blue)andafixedfibroblastcell(red)withoutSAC(d)andwithSAC(e).(f,g)Absorbanceimagesoffixedfibroblastcells,measuredat1650cm-1withoutSAC(f)andwithSAC(g).

Figure1ashowstheschematicoftheSAC-IRmicroscope(SIfordetaileddescription).Briefly,forSAC,theIRbeampassedthroughanIRpolarizer(ISPOptics)mountedonarotationstage(Newport)andthenbackthroughafixedIRpolarizer;rotatingthefirstpolarizermodulatedtheincidentlightintensityasafunctionofwavelength.Samplechambersusedinthisworkconsistedofa1mmthickCaF2slide(Crystran),a25μmspacertape(3M),anda1mmthickCaF2slidewithdrilledholesforcellsandmediaintroduction.ForSAC,thetransmittedbeamwasmovedtoafluid-onlyblankregion(termedreferenceregion),andforeachwavenumber,therotatingpolarizeranglewasadjustedsothatthetransmissionintensitybecameaconstantsetpointovertheentirewavenumberscanningrange.(seeFigureS1).

TodemonstratetheneedforSAC,wecomparedthetransmittedlightintensityITofareferenceregion(Figure1b,theblueline)tothetransmissionintensityfromanimagepixelnearthecenterofafixedNIH3T3fibroblastcell(Figure1b,theredline,seeSIforcellculturedetails).Figure1bshowsthatwithoutSAC,waterstronglyabsorbsIRintheregionof1600cm-1–1700cm-1,andthetransmissionintensitydifferencebetweenareference(PBS-onlyregion)andsample(PBSandfibroblastcell)isclosetoorbelowthedynamicrangenoiselimitofthedetectionsystem.Incomparison,Figure1cdemonstratestheSACimplementation;thereferenceintensityspectrum(PBS,theblueline)isnearconstant,andthereisaresolvableintensitydifferencebetweenthereferenceandsample.WealsonotethatIRbelow1000cm-1isweakduetostrongabsorptionbytheCaF2windows,butSACcompensatesforsubstrateIRabsorption,too.

Figures1d,eshowabsorptionspectrainareferenceregion(blue)andnearacellcenter(red).TheSAC-implementedIRspectraofthecellinthe1600cm-1–1700cm-1regionareobservablewithahighSNR.Wethenconstructedabsorptionimagesofhydratedfixedfibroblastcells(seeFigureS2fordetails).TheabsorptionimagewithoutSACinFigure1fshowsthatcellsareunresolvedfromthebackgroundbecausethestrongabsorptionofwateroverlapswithabsorptionfrombiomoleculeswithinthefibroblastcell.InSAC(Figure1g),waterabsorptioniscompensated,andtheabsorptionbyanalytemoleculesinthecellsbecomesresolvablefromthebackground.FigureS3showshistogramsofabsorbancevalues;thebroaddistributionofthebackgroundinnon-SACoverwhelmsthesignalofcells,whereas,inSAC,thecells’absorbancevaluescanbereadilyresolvedfromthenarrowedmaindistribution.Weestimatethepixel-to-pixel(spatial)absorbancenoiseat1645cm-1inthebackgroundregion(FigureS3)ofnon-SACabsorptionimagestobe0.025.ThespatialabsorbancenoiseinSACimagesis0.002,a>10-foldimprovementcomparedtonon-SAC.Similarly,thespectralnoisesmeasuredinthe1623cm-1–1672cm-1regionare0.020and0.002fornon-SACandSAC,respectively(FigureS4).

Figure2.SystemperformanceofSAC-IRmicroscopy.Imagingresultsusingamixtureof5-μmdiameterpolystyrene(PS)andpoly(methylmethacrylate)(PMMA)microparticlesinwaterwitha25μmspacer.(a)IRspectraofthePSandPMMAmicroparticlesusingSAC.(b)Compositeimagewithabsorbancesat1493cm-1(green)and1712cm-1(red)usingSAC.(c)LinescansoftheindicatedPSparticleinpanelb.

Figure2demonstratessystemperformanceinspaceandfrequencyusingamixtureofpolystyrene(PS,Phosphorex)andpoly(methylmethacrylate)(PMMA,Phosphorex)particlesdispersedinwater.Thenominaldiametersofbothparticlesare5μm.SimilartoFigure1,theabsorbanceofeachimagepixelwascalculatedbythereference(water)region.Thus,theapparentIRabsorbancespectraofthetwotypesofmicroparticlesareshowninFigure2a.Bothspectrashowastrongnegativeabsorbanceregioncenterednear1645cm-1,indicatingthatthetransmissionishigheratapixelthanthebackgroundregion.Thisapparentnegativeabsorbanceoccurswhenthematterinthebeampathabsorbslesslightofthespecificwavenumberthanthereplacedwater.41,42Inadditiontothewaterexclusioneffect,theMiescatteringmakesitdifficulttoquantitativelyseparatetheinherentvibrationalabsorptionspectrumandthescatteringfromanobservedtransmissionspectrum.Evenwiththecomplexspectralmixing,thestrongabsorptionpeakcenterednear1712cm-1observedfromPMMAparticlesisassignabletotheC=OstretchingmodefromtheeastergroupofPMMA.43SimilartoPMMAparticles,theapparentIRabsorptionspectrumofPSparticlesinFigure2ashowsthenegativedipat1645cm-1andthepositivebaselinedriftduetowaterexclusionandlightscattering,respectively.Still,thesignaturepeaksat1493cm-1and1453cm-1ofthebenzeneringofPSwereobservable.44

Figure2bshowsapseudo-labeledcompositeimageofamicroparticlemixtureconstructedwithtwoabsorbanceimagesacquiredat1493cm-1(green)and1712cm-1(red),correspondingtoPSandPMMA,respectively.AllparticlesinthecompositeimagewerereadilyidentifiableaseitherPSorPMMA.Wemeasuredthefull-width-half-maximum(FWHM)fromlinescansofmultipleparticles.ThemeanFWHMsfromparticlesare7.3±0.2μminthefast-scanningdirection(X,horizontal)and7.0±0.4μmintheslow-scanningdirection(Y,vertical),wheretheuncertaintiesindicatethestandarddeviation.Thisresolvingpowerisclosetothediffraction-limitedresolution(0.5λ/NA)at1650cm-1witha0.4NAreflectiveobjective.

Figure3.SAC-IRspectraandimagesoffixedfibroblastcells.(a)Line-averagedspectraofthreedifferentfixedfibroblastcellscrossingthecellcenters.(b)IRspectraofbovineserumalbumininwater(solidgreen,richina-helix)and-lactoglobulininwater(dottedgreen,richinb-sheet)forprotein,andherringDNAinwater(blue)fornucleicacid.TheabsorptionspectraoftheproteinsandDNAsolutionsweremeasuredwithareferenceofdeionizedwater.TheabsorptionspectrumofglyceroltrioctanoateinCCl4andCS2(red)forfattyacidesterwasdownloadedfromtheNISTChemistryWebBook().Allabsorptionspectrawerescaledtotheconcentrationof10mg/mL.(c–e)Absorbanceimagesatwavenumbersrepresentativeof(c)proteinat1656cm-1,(d)fattyacidat1748cm-1,and(e)nucleicacidat1085cm-1and1053cm-1.(f)Compositeimageofpanelsc–e.Theverticaldashedlinesina,bindicatethewavenumbersusedtoconstructtheimagesofpanelsc–e.

Next,wedemonstratelabel-freechemicalimagingofkeybiomoleculesinfixedcells.Figure3ashowstheabsorbancespectraofthreefixedfibroblastcells.Foreachspectrumofacell,thesamplestagewasX-scannedatafixedYlocation,anditwasrepeatedwhilethewavenumberchanged.Theline-averagedspectraofthreedifferentcellsshowsimilarityinIRpeakpositionandwidth,althoughpeakheightsvaryamongcells.Thepeakheightvariationmaybeduetovariationsineithercellheightormolecularconcentrationinsideacellorboth.Unfortunately,theorigincannotbeknownonlywithanIRmeasurement.InFigure3b,IRspectrawerepresentedforcomparisonfromrepresentativecellularmolecules,includingprotein,nucleicacid,andfattyacidester.Thepeaksat1650cm-1and1550cm-1fromthecellsshowastrongcorrelationwiththeamideIandIIpeaksofprotein,respectively.Ontheotherhand,thepeaksobservedat1085cm-1and1230cm-1correspondtothesymmetricandantisymmetricPO2-stretchingmodesofnucleicacid,respectively.45Theweakpeakat1745cm-1canbeattributedtotheestergroupinphospholipidorfattyacidester.Althoughvarioustypesofestermoleculesexistincells,wewillcallthe1745cm-1peakforfattyacid,forsimplicity,inthispaper.Figures3c–eshowabsorbanceimagesconstructedatthemostrelevantwavenumbersofprotein,fattyacid,andnucleicacid,respectively.Becauseoftherelativelyweakabsorbanceofnucleicacidcomparedtothebaseline,Figure3efornucleicacidwasconstructedbyabsorbancedifferencebetween1085cm-1and1053cm-1.ThecompositeimageinFigure3fshowsthespatialdistributionofthethreebiomolecules.Theblue-coloredfeatureislocatedatthecenterofcells,whileproteinisdistributedovertheentirecell.Ontheotherhand,fattyacidislocalizedinnon-nuclearregions.

AmongtheIRpeaksthatrepresentbiomoleculesinthecellspectra,wefocusedontheabsorbancepeakat1650cm-1,whichisdominantlyduetotheamideIpeakofprotein.Also,theabsorptioncrosssectionsofmajorbiomolecules,suchasnucleicacidandfattyacid,arelowerthanthatofproteinat1650cm-1,asshowninFigure3b.Somecarbohydratescontainpeptidebonds,e.g.,sialicacid,buttheirconcentrationinacellismuchlowerthanproteins.Thus,forsimplicity,weassumedtheabsorbanceat1650cm-1originatedfromproteinsandusedittodeterminetheabsoluteproteinmasspercell.

Ifcellheightisknown,measuredabsorbancewillleadtotheconcentrationofmoleculesinacell.However,thecurrentimagingsystemcouldnotmeasureabsorbanceandheightsimultaneouslyduringlivecellimaging.Instead,wedeterminethetotalmasspercell(mc)fromtheabsorbancesum(SA)andtheabsorptioncoefficient(e)withthefollowingrelation,

mc=c×V=(A/εl)×(lS)=1εA×S

(1)

wherecistheconcentration;Visthecellvolume;eistheabsorptioncoefficient;listhemeancellheight;Sisthecellarea;andAisthemeanabsorbanceoverS.Acellarea,S,wasdeterminedbysegmentingcellsinabsorptionimageswiththethresholdof99%ofthecumulativedistributionofthebackgroundabsorbancevalues(seeSIfordetails).

Figure4.ComparisonofSAC-IRimagesoffixedandlivefibroblastcells.(a,b)IRimagesoffixedandlivecellsat1650cm-1.(c)Scatterplotsofproteinmasspercellfor74fixedand94livecells.Themedian(mean)valuesare105pg(126pg)and122pg(131pg)forfixedandlivecells,respectively.Theboxesindicatethe25%and75%,andthehorizontallinesindicatethemedianvalues.

Tocalculateproteinmc,weusedtheabsorptioncoefficient,e=0.28mm2pg-1at1650cm-1,calculatedfromthespectraoftwonon-glycosylatedproteinswithdifferentsecondarystructures,showninFigure3b.However,wenotethatusingasingleevaluemeasuredfromonlytwoproteinsolutionsmayyieldanon-negligibleuncertaintyindeterminingtheabsolutemassofproteinbecausenon-proteinmoleculescanstillcontributetothe1650cm-1absorbance,andthepeakshapeoftheamideIpeakcanvarydependingonproteinspecies.Thus,aslongasimagingspeedallows,evaluesatmorewavenumberscanbeusedforcalculation,improvingtheaccuracyinquantifyingproteinmc.Backtothestraightforwardandquickapproach,thesingleevaluewasusedinEq.(1)tocalculatetheproteinmcofbothfixedandlivecells,whoseSAC-IRimagesareshowninFigure4a,b.Thecalculatedproteinmcvalues,showninFigure4c,spanfrom20pgto350pg,andthedistributionsseemnon-Gaussian.Themedianvalueofthefixedcells’mcis105pgfrom74cells.Thisisslightlylowerthanthatoflivecells’mc(122pgfrom94cells).However,consideringthebreadthofthedistributions,itisdifficulttoconcludewhetherproteinmcisreducedbyfixationornot.Asasanitytest,wecomparedtheproteinmcbyvaryingthethresholdlevelforcellsegmenting.FigureS5showsthatthemeanproteinmcisreducedby6%whenthethresholdiselevatedfrom1sto3s,wheresisthestandarddeviationofthebackgroundabsorbancedistribution.Therangeofproteinmcisconsistentwithpreviouslyreportedvaluesfromwhole-cellproteinmassmeasurements,46UVabsorptionmicroscopy,47,48andstimulatedRamanscattering(SRS).13Thus,despitethemanyassumptionsandsimplifications,theabsorbanceat1650cm-1canbeusefulforthequantificationofproteinmassinlivecells.

Next,weusedSAC-IRmicroscopytoimagelivecellsevery12minfor12hatthreerepresentativewavenumbers(1656cm-1,1745cm-1,and1230cm-1)forprotein,fattyacidester,andnucleicacid,respectively.Fornucleicacid,weusedabsorbanceatasinglewavenumberof1230cm-1insteadofabsorbancedifference,asshowninFigure3,toincreasetheimagingframerateandtakeadvantageofslightlyhigherabsorbancesignals.Theabsorbanceimagessequentiallyacquiredforthethreewavenumberswerecombinedintoacompositeimageevery12min.Thetime-lapseofentireIRabsorbancecompositeimagesfor12hcanbefoundinSIMovie1.Mostcellswereagile,andsomeunderwentcelldivision.Figure5ashowsIRimagesatthreedifferenttimesamongthetotal62frames.Figures5b–dshowthetrajectoriesofthreeisolatedandin-framecellsduringtheentireimaging.Thecelllabeledasbshowedcelldivision.Thecellcwasabouttofinishcelldivisioninthelastframes.Ontheotherhand,cellddidnotshowdivisionbutshowedfluctuationinsizewhilemigrating.Wemonitoredthefluctuationoftheabsorbancesatthethreewavenumberscorrespondingtoprotein,fattyacid,andnucleicacid.Figures5e–gshowthetotalabsorbancepercell(A×S)asafunctionoftimeforthethreewavenumbers.Theabsorbancetracksofthethreebiomoleculesshowastrongcorrelation.Forexample,whenthesamplechamberwasreplenishedwithaCO2-saturatedmediumpreparedinanincubatoratt=7h,theabsorbancetracksrespondedtothemediumreplacement.InFigures5e,d,onehourbeforecelldivision,allthreeabsorbancesbegantoincrease,likelytopreparecelldivision.Ontheotherhand,thenon-dividingcelldshowslittlecorrelationbetweenthethreeabsorbanceprofiles.Forproteins,thetotalabsorbancewasconvertedintotheabsolutemasspercellusingEq.(1)andlabeledontheright.Unlikeabsorbanceat1656cm-1forproteins,single-frequencyabsorbancesat1745cm-1and1230cm-1canbecontributedbyotherchemicalsthanfattyacidandnucleicacid.Thus,wedidnotconvertthetotalabsorbancesatthetwofrequenciesintothetotalmassesoffattyacidandnucleicacid.However,ifsufficientlymorespectralimagesareavailable,theirabsoluteper-cellmassescouldalsobemonitored.AnIRtransmissionimagingsystemwithafasterimagingspeedandmorefrequenciescouldprovidemorequantitative,time-dependentinformationontheabsoluteper-cellmassesforvariousbiomoleculesinlivecells.

Figure5.LivefibroblastSAC-IRimages.(a)Three-colorSAC-IRimagesacquiredeverytwelveminutesfortwelvehours.Yellowcolorrepresentsabsorbanceassociatedwithproteinat1656cm-1;magentaforfattyacidat1745cm-1;andcyanfornucleicacidat1230cm-1.(b–d)Enlargedimagesofthedividingcells(bandc)andthenon-dividingcell(d),indicatedinpanela.Thecentersofmassabsorbanceofthemarkedcellsareplottedastimetraces.(e–g)Trackingtotalabsorbancepercell(A×S)ofthecellsinpanel(b–d),respectively,measuredatthethreedifferentwavenumbers.In(e),theredandblacksolidlinesafter10.8haretheA×Svaluesofthedaughtercells.

Insummary,wehavedemonstratedthenon-synchrotronIRtransmissionimagingoflivecellsusingthehome-builtSAC-IRmicroscopytechnique.TheSACapproachsuccessfullymitigatedthestrongwaterabsorptionneartheamidebands.BasedonSAC-IRimagesmeasuredatmultiplefrequencies,wehavesuccessfullydemonstratedanopticalmethodtomeasuretheabundanceanddistributionofkeybiomolecules(e.g.,protein,fattyacid,andnucleicacid)infixedandlivecells.Inparticular,usingthetotalamideIbandabsorbancepercell,wedeterminedtheabsoluteper-cellmassofproteininmultiplelivefibroblastcells.Byimaginglivefibroblastcellsovertwelvehours,wemonitortheper-cellmasschangeofthethreemolecularspeciesduringvariousphases,includingcelldivision.Thecurrentdemonstrationwasbasedononlythreewavenumbers.Ifwecouldrecordmorewavenumberimageswhilemaintainingbiologicallyrelevantframeratesforlivecells,wewouldbeabletoretrievemorereliableanddetailedinformationonmolecularcompositions.Asaneconomicalandwidelyadoptableimagingmodality,SAC-IRmicroscopycouldemergeasastandardimagingmethodofquantifyingbiomoleculeswithSI-traceabilityinlivecellsandhydratedtissuesacrossfieldssuchasbiology,biotechnology,andmedicine.

SupportingInformation.

TheSupportingInformationisavailablefreeofchargeat

/doi/10.1021/acs.analchem.4c02108

.

DetailedexperimentaldescriptionsofSAC-IRmicroscope,samples,anddataprocessing.ExampleofpolarizeranglesettingasafunctionoffrequencyforSACpre-compensationofwaterabsorption.DescriptionofIRimagepre-processingsteps.Comparisonbetweennon-SACandSACsignalandnoisedistributions.Comparisonbetweennon-SACandSACabsorbancespectra.Descriptionofthethresholdlevelselectionprocesstodefinethecellboundary(PDF)

LivecellIRabsorptionmoviesofNIH3T3fibroblastcells(MOV)

Acknowledgments

WethankCharlesCampforhelpfuldiscussionsandJoyDunkersforassistancewithconfocalmicroscopy.WethanktheNISTBiomanufacturingProgramforfinancialsupport.Yow-RenChangthankstheNationalResearchCouncilforfinancialsupport.

Authorcontributions

Themanuscriptwascreatedwithcontributionsfromallauthors.Allauthorshaveapprovedthefinalversionofthemanuscript.

Dataandmaterialsavailability

Dataareavailableuponrequestfromthecorrespondingauthor.

Notes

Theauthorsdeclarenocompetingfinancialinterest.

Certaincommercialequipment,instruments,ormaterialsareidentifiedinthispapertofosterunderstanding.SuchidentificationdoesnotimplyendorsementbyNIST,nordoesitimplythatthematerialsorequipmentidentifiedarenecessarilythebestavailableforthepurpose.

Official