Particlesizedistributionofexosomesandmicrovesiclesdeterminedbytransmissionelectronmicroscopy,flowcytometry,nanoparticletrackinganalysis,andresistivepulsesensing-Pol-2014-JournalofThrombosisandHaemostasis-WileyOnlineLibrary
JournalofThrombosisandHaemostasisVolume12,Issue7OriginalArticleFreeAccessParticlesizedistributionofexosomesandmicrovesiclesdeterminedbytransmissionelectronmicroscopy,flowcytometry,nanoparticletrackinganalysis,andresistivepulsesensingE.vanderPol

CorrespondingAuthor

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

Correspondence:EdwinvanderPol,BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Meibergdreef9,POBox22660,1100DD,Amsterdam,theNetherlands.

Tel.:+31205664386;fax:+31205669569.

E‐mail:e.vanderpol@amc.uva.nl

SearchformorepapersbythisauthorF.A.W.Coumans

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

SearchformorepapersbythisauthorA.E.Grootemaat

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

SearchformorepapersbythisauthorC.Gardiner

NuffieldDepartmentofObstetricsandGynaecology,JohnRadcliffeHospital,Oxford,UK

SearchformorepapersbythisauthorI.L.Sargent

NuffieldDepartmentofObstetricsandGynaecology,JohnRadcliffeHospital,Oxford,UK

SearchformorepapersbythisauthorP.Harrison

SchoolofImmunityandInfection,UniversityofBirminghamMedicalSchool,Birmingham,UK

SearchformorepapersbythisauthorA.Sturk

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

SearchformorepapersbythisauthorT.G.vanLeeuwen

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

SearchformorepapersbythisauthorR.Nieuwland

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

Searchformorepapersbythisauthor
E.vanderPol

CorrespondingAuthor

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

Correspondence:EdwinvanderPol,BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Meibergdreef9,POBox22660,1100DD,Amsterdam,theNetherlands.

Tel.:+31205664386;fax:+31205669569.

E‐mail:e.vanderpol@amc.uva.nl

SearchformorepapersbythisauthorF.A.W.Coumans

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

SearchformorepapersbythisauthorA.E.Grootemaat

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

SearchformorepapersbythisauthorC.Gardiner

NuffieldDepartmentofObstetricsandGynaecology,JohnRadcliffeHospital,Oxford,UK

SearchformorepapersbythisauthorI.L.Sargent

NuffieldDepartmentofObstetricsandGynaecology,JohnRadcliffeHospital,Oxford,UK

SearchformorepapersbythisauthorP.Harrison

SchoolofImmunityandInfection,UniversityofBirminghamMedicalSchool,Birmingham,UK

SearchformorepapersbythisauthorA.Sturk

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

SearchformorepapersbythisauthorT.G.vanLeeuwen

BiomedicalEngineeringandPhysics,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,theNetherlands

SearchformorepapersbythisauthorR.Nieuwland

LaboratoryofExperimentalClinicalChemistry,AcademicMedicalCenter,UniversityofAmsterdam,Amsterdam,TheNetherlands

Searchformorepapersbythisauthor

Giveaccess

Sharefulltextaccess

Sharefull-textaccess

PleasereviewourTermsandConditionsofUseandcheckboxbelowtosharefull-textversionofarticle.IhavereadandaccepttheWileyOnlineLibraryTermsandConditionsofUseShareableLink

Usethelinkbelowtoshareafull-textversionofthisarticlewithyourfriendsandcolleagues.Learnmore.

CopyURL

Sharealink

ShareonEmailFacebookTwitterLinkedInRedditWechatSummaryBackground

EnumerationofextracellularvesicleshasclinicalpotentialasabioMarkerfordisease.Inbiologicalsamples,thesmallestandlargestvesiclestypicallydiffer25‐foldinsize,300000‐foldinconcentration,20000‐foldinvolume,and10000000‐foldinscatteredlight.Becauseofthisheterogeneity,thecurrentlyemployedtechniquesdetectconcentrationsrangingfrom104to1012vesiclesmL–1.

Objectives

Toinvestigatewhetherthelargevariationinthedetectedconcentrationofvesiclesiscausedbytheminimumdetectablevesiclesizeoffivewidelyusedtechniques.

Methods

Thesizeandconcentrationofvesiclesandreferencebeadsweremeasuredwithtransmissionelectronmicroscopy(TEM),aconventionalflowcytometer,aflowcytometerdedicatedtodetectingsubmicrometerparticles,nanoparticletrackinganalysis(NTA),andresistivepulsesensing(RPS).

Results

Eachtechniquegaveadifferentsizedistributionandadifferentconcentrationforthesamevesiclesample.

Conclusion

Differencesbetweenthedetectedvesicleconcentrationsareprimarilycausedbydifferencesbetweentheminimumdetectablevesiclesizes.Theminimumdetectablevesiclesizeswere70–90nmforNTA,70–100nmforRPS,150–190nmfordedicatedflowcytometry,and270–600nmforconventionalflowcytometry.TEMcoulddetectthesmallestvesiclespresent,albeitafteradhesiononasurface.Dedicatedflowcytometrywasmostaccurateindeterminingthesizeofreferencebeads,butisexpectedtobelessaccurateonvesicles,owingtoheterogeneityoftherefractiveindexofvesicles.Nevertheless,dedicatedflowcytometryisrelativelyfastandallowsmultiplexfluorescencedetection,makingitmostapplicabletoclinicalresearch.

Introduction

Extracellularvesicles,suchasexosomesandmicrovesicles,arereleasedbycellsintotheirenvironmentassubmicrometerparticlesenclosedbyaphospholipidbilayer1.Thesevesiclescontributetomanyhomeostaticprocesses,e.g.coagulationandinflammation2-4,andthereforehavepotentialclinicalapplications5-8.Unfortunately,mostsinglevesiclesarebelowthedetectionrangeofmanytechniques,owingtotheirsmallsizeandlowrefractiveindex9,10,leadingtomisinterpretationofdataandreportedconcentrationsrangingfrom104to1012vesiclesmL–1inplasma9-15.

In2010,wereviewedthetheoreticalperformanceof13methodstodeterminetheparticlesizedistribution(PSD)ofvesicles9.ThePSDdescribestheconcentrationasafunctionofsize,anddefineswhichvesicletypesaremeasured4.Unexpectedly,oursimulationspredictedthateachmethodwouldobtainadifferentPSD,therebyhamperingdatainterpretation,datacomparison,andstandardization.

Inthisstudy,weperformedanexperimentalevaluationoffiveofthe13methods.Weselectedthemostwidelyusedmethodscapableofdetectingsinglevesicles:transmissionelectronmicroscopy(TEM),aconventionalflowcytometer,aflowcytometerdedicatedtodetectingsubmicrometerparticles,nanoparticletrackinganalysis(NTA),andresistivepulsesensing(RPS).ThePSDsofastandardpopulationofreferencebeadsandastandardpopulationofvesiclesweremeasuredwithallmethods.

Throughoutthisarticle,wedefine‘size’asthediameterofaparticle,andthePSDasthehistogramofparticlesizes,providingthemeannumberofparticlespermilliliterper10‐nmbin16.Dataprocessingandrepresentationwereperformedwithoriginpro(v8.0724;OriginLabCorporation,Northampton,MA,USA).

TocreateareferencesamplewithaknownPSD,amixtureoftraceablepolystyrenebeads(Nanosphere;ThermoFisher,Waltham,MA,USA)waspreparedinde‐ionizedwater.RPSmeasurementsrequireaconductivemedium;therefore,thebeadsweresuspendedinelectrolytebuffer(Izon,Christchurch,NewZealand).ThesizeandconcentrationofthereferencebeadswereselectedtoresemblethoseofpreviouslyreportedvesiclePSDs9,11-13,15,17.Table1showsthesizeofthereferencebeadsaccordingtoTEMdataofthemanufacturer.Theconcentration(beadsmL−1)wasderivedfromthespecifications.Figure1AshowsthePSDofthereferencebeadsundertheassumptionthateachsubpopulationhasaGaussiandistribution.Thereferencesamplecontainedfivesubpopulations,amongwhichlargerbeadshavelowerconcentrations.Thetotalconcentrationwas3.1×109beadsmL−1.Priortoanalysis,thereferencesamplewassonicatedfor10sandvortexedfor10s.Table2liststhediametersofsilicabeads(SilicaOxideSizeStandards[Corpuscular,ColdSpring,NY,USA];PlainSilica[Kisker,Steinfurt,Germany])usedtocalibratetheflowcytometersandNTAinstrument.

Table1.Catalognumbers,diametersandconcentrationsofthesubpopulationsofpolystyrenereferencebeadsasdeterminedbyfivemethods
NTA,nanoparticletrackinganalysis;RPS,resistivepulsesensing;TEM,transmissionelectronmicroscopy.Diameterisexpressedasmean±standarddeviation.
Table2.Manufacturers,catalognumbersanddiametersofsilicabeadsobtainedbyimagingatleast500beadswithtransmissionelectronmicroscopy

Figure1OpeninfigureviewerPowerPointParticlesizedistribution(PSD).Concentrations(onalogarithmicscale)ofthereferencebeads(left)andthevesiclestandard(right)detectedby(A,B)transmissionelectronmicroscopy(TEM),(C,D)conventionalflowcytometry,(E,F)dedicatedflowcytometry,(G,H)nanoparticletrackinganalysis(NTA)and(I,J)resistivepulsesensing(RPS)areshown.Thebinwidthis10nm.PSDsofthereferencebeads(blackline)werefittedbyasumofGaussianfunctions(dottedredline).PSDsofthevesiclestandard(blackline)werefittedbyapower‐lawfunction(dashedgreenline).ThePSDofthereferencebeadsdeterminedbyTEMisbasedondatafromthemanufacturer.PSDsgivenbyNTAandRPSoriginatefromtwomeasurementswithrelativelyhigh‐sensitivity(black)andlow‐sensitivity(blue)settings.

Asisolationofvesiclesfrombloodischallenging18,weselectedurinaryvesiclesforourbiologicalstandardsample.Urinecontainsarelativelyhighconcentrationofvesicleswithlowcontamination9.Urinefromfivehealthymaleindividualswascollected,pooled,andcentrifugedtwice(8×50mL,10min,180×g,4°C;and20min,1550×g)toremovecells.Cell‐freeurinealiquots(12mL)werefrozeninliquidnitrogenandstoredat–80°C.Priortoanalysis,sampleswerethawedonmeltingicefor1h,centrifuged(10min,1550×g,4°C)toremoveprecipitatedsalts,anddilutedin0.2‐μm‐filtered(MilliPore,Billerica,MA,USA)phosphate‐bufferedsaline.

DatafromthemanufacturerwereusedtocreatethePSDofreferencebeadsbyTEM.ForanalysisofthevesiclestandardbyTEM(CM‐10;Philips,Eindhoven,TheNetherlands),vesicleswerepreparedandanalyzedasdescribedinDataS1.Toobtainthevesicleconcentration,wemultipliedthemeannumberofvesiclespersurfaceareabythegridarea,dividedbythesamplevolume.Here,weassumedthatallvesiclesadheredtothegridandweredistributeduniformly.

Aflowcytometer(FACSCalibur;BD,FranklinLakes,NJ,USA)witha15‐mW488‐nmlaserwasusedtodetectside‐scatteredlight(SSC)for10minataflowrateof~60μLmin−1.ThedetectorsettingsaredescribedinDataS1.Tocalculatetheparticleconcentration,theflowratewasdeterminedbyweightingthesamplevolumeaspiratedduring10min.Topreventswarmdetection15,thereferencebeadsandvesiclestandardwerediluted1000‐fold(1.7×105countsvs.1.0×105backgroundcounts)and100‐fold(2.6×105countsvs.1.5×105backgroundcounts),respectively.Theabsenceofswarmdetectionwasconfirmedbyserialdilutions.

TorelateSSCtoaparticlesize,wecalibratedtheflowcytometerwithbeadsofknownsizeandrefractiveindex.Figure2AshowstheSSChistogramofpolystyrenebeads.Figure2BshowstheSSCofpolystyreneandsilicabeadsvs.theirsize.ThedatawerefittedbyMietheory,incorporatingthesizeandrefractiveindexofthebeadsandtheopticalconfigurationoftheinstrument19.MiecalculationswereperformedwiththescriptsofMätzler20inmatlab(v7.9.0.529).ThesolidcurveinFig.2BwasusedtorelateSSCtothesizeofthepolystyrenereferencebeadswitharefractiveindexof1.6115.ThedashedcurveinFig.2BwasusedtorelateSSCtovesiclesize,withtheassumptionthatvesiclesaresphereswitharefractiveindexof1.40,whichwaspreviouslyestimated14andcorrespondstotherefractiveindexofcells21,22.

Figure2OpeninfigureviewerPowerPointRelationshipbetweenscatteringandthediameterofvesicles.(A)Side‐scatteredlight(SSC;logarithmicscale)vs.concentrationforpolystyrenebeadsmeasuredbyconventionalflowcytometry.(B)Measured(symbols)andcalculated(lines)SSC(logarithmicscale)vs.diameterforpolystyrenebeads(black),silicabeads(red),andvesicles(green).TheSSCincreaseswithincreasingparticlediameter,andislowerforvesiclesthanforbeads.(C)Forward‐scatteredlight(FSC;logarithmicscale)vs.concentrationforpolystyrenebeadsmeasuredbydedicatedflowcytometry.Theconcentrationof102‐nmbeadswas100‐foldhigherthanthatoftheothersizestodiscriminatebetweensignalandbackgroundcounts.(D)Measured(symbols)andcalculated(lines)FSC(logarithmicscale)vs.diameterforpolystyrenebeads(black),silicabeads(red),andvesicles(green).Miecalculationsareinexcellentagreementwiththedata,exceptforthe799‐nmand994‐nmpolystyrenebeads,owingtodetectorsaturation.a.u.,arbitraryunits.

Throughoutthisarticle,weuse‘dedicatedflowcytometry’asagenerictermforflowcytometersdedicatedtodetectingsubmicrometerparticles.Aflowcytometer(A50‐Micro;Apogee,HemelHempstead,UK)witha20‐mW488‐nmlaserwasusedtodetectforward‐scatteredlight(FSC)andSSC.ThedetectorsettingsaredescribedinDataS1.Thesamplevolumeinjectedbytheinternalmicrosyringewasusedtocalculatetheconcentrationofparticles.Intotal,1.4×105referencebeadsand0.8×105vesicleswereanalyzed.Analogouslytoourapproachforconventionalflowcytometry,werelatedFSCtothevesiclesizebyusingbeadsandMietheory,asillustratedfordedicatedflowcytometryinFig.2C,D.

Adark‐fieldmicroscope(NS500;Nanosight,Amesbury,UK)witha45‐mW405‐nmlaserandanelectronmultiplyingcharge‐coupleddevice(EMCCD)wasusedtodeterminethePSDbytrackingtheBrownianmotionofsingleparticles12,23.Measurementswereperformedwithtwodilutionsandtwodetectionsettingstoincreasetheeffectivesizerange,whichisneededbecauselightscatteredfromthesmallestandthelargestbeadsdiffersbyfiveordersofmagnitude,whereasthedynamicrangeoftheEMCCDisonlyapproximatelythreeordersofmagnitude.Consequently,settingssuitablefordetecting46‐nmbeadswouldresultinextremesaturationfor596‐nmbeads.Twodilutionsareneeded,becausea50‐folddilutionisrequiredtodetectthesmallestbeads23,butatthisdilutiontheprobabilityofdetectinga596‐nmbeadis0.5%.

Referencebeadswereanalyzedwithhigh‐sensitivitysettings(diluted1:50;shutter,26.67ms;gain,650;threshold,22;1.8×103beadstracked)andlow‐sensitivitysettings(undiluted;shutter,1.67ms;gain,100;threshold,10;1.1×104beadstracked).WemultipliedtheconcentrationasprovidedbytheNTAsoftwarebytheratiobetweentheexpectedandmeasuredconcentrationsofcalibrationbeads23.Thisconcentrationcalibrationwasperformedwith102‐nmand203‐nmpolystyrenebeadswithconcentrationsof2×107and1×108beadsmL−1forthehigh‐sensitivityandlow‐sensitivitysettings,respectively.Thevesiclestandardwasanalyzedwithhigh‐sensitivitysettings(diluted1:500;shutter,26.67ms;gain,650;threshold,19;1.0×103vesiclestracked)andlow‐sensitivitysettings(diluted1:100;shutter,26.67ms;gain,400;threshold,10;1.1×103vesiclestracked).Concentrationcalibrationwasperformedwith105‐nmand206‐nmsilicabeadswithaconcentrationof1×108beadsmL−1forboththehigh‐sensitivityandlow‐sensitivitysettings23,astherefractiveindicesofsilicaandvesiclesareclose.

Persample,20videosof30swerecapturedat22.0°CandanalyzedbyNTAv2.3.0.17(Nanosight),assumingamediumviscosityof0.95cP.ToobtaintheoverallPSDO(d),thePSDsobtainedwithhigh‐sensitivitysettings,H(d),andlow‐sensitivitysettings,L(d),werecombinedatthesized0,wheretheconcentrationsweresimilar[H(d0)≈L(d0),O(d)=H(d)foralld≤d0;O(d)=L(d)foralldd0].

RPS(qNano;Izon)determinesthePSDfromresistancepulsescausedbyparticlesmovingthroughapore.Measurementswereperformedwithtwoporesizes,fortworeasons.First,forasinglepore,thedetectablesizerangeisatbestfive‐fold,whereasoursmallestandlargestreferencebeadsdiffer12‐foldinsize.Second,theanalyzedsamplevolumedependsontheporesize.WiththeNP100pore,only0.9nLofsamplewasanalyzed,containing1beadof596nmonaverage.WiththeNP400pore,80nLwasanalyzed,containing77beadsof596nm.

ThereferencebeadswereanalyzedbyRPSwithhigh‐sensitivitysettings(NP100;voltage,0.70V;stretch,47.0mm)andlow‐sensitivitysettings(NP400;voltage,0.26V;stretch,46.5mm).Thevesiclestandardwasdiluted1:1,andanalyzedwithhigh‐sensitivitysettings(NP100;voltage,0.60V;stretch,46.0mm)andlow‐sensitivitysettings(NP400;voltage,0.40V;stretch,43.5mm).Thepressurewassetat7.0mbar.Calibrationwasperformedwithbeadssuppliedbythemanufacturer.Atleast1000particlespersamplewereanalyzed.Acustom‐madeMicrosoftVisualBasic2008applicationinterruptedthemeasurementwhentheroot‐mean‐squarenoiseexceeded10pA.WerequiredtheR2‐correlationofcumulativecountswithtimetoexceed0.999,andthebaselinecurrentdriftnottoexceed5%.ThetwoPSDsarecombinedinasimilarwayasforNTA.

Figure1showsallPSDsofthereferencebeadsandvesiclestandard.Figure1AshowsthereferencebeadPSDbasedonthepreparedconcentrationsandmanufacturer‐supplieddata.Figure1BshowsthevesiclestandardPSDmeasuredbyTEM.ThecombinedPSDhasapeakat45nm,andforlargervesiclestheconcentrationdecreaseswithincreasingsize.Thespikesontheright‐handsidecorrespondtosinglevesicles.TEMcandetectthesmallestvesiclespresent,owingtoanimagingresolutionof~1nm.However,samplepreparationmaycauseareductioninvesiclesize18,24,25.Inaddition,limitedandnon‐uniformadhesionofvesiclesonthesurfacemayaffectthePSD.

Figure1Cshowsthatthesmallestpolystyrenebeaddetectedbyconventionalflowcytometrywas203nm,andthatthepeakswerebroadenedascomparedwiththereferencebeadPSD.Figure1DshowsthatthefirstbinofthevesiclestandardPSDcorrespondsto340nm,whichis140nmlargerthanthesmallestdetectedpolystyrenebead,owingtorefractiveindexdifferences.Thedetectedconcentrationwas1.8×107vesiclesmL−1.

Dedicatedflowcytometryiscapableofdetectingsingle102‐nmpolystyrenebeads,asshowninFig.1E.ThewidthofthepeaksiscomparabletothereferencebeadPSD.Figure1FshowsthatthefirstbinofthevesiclestandardPSDcorrespondsto160nm.Consequently,dedicatedflowcytometrydetectedapproximatelytwiceassmallandthereby18‐foldmorevesiclesthanconventionalflowcytometry.Thedetectedconcentrationwas3.3×108vesiclesmL−1.Figure1D,Fwasproducedontheassumptionofavesiclerefractiveindexof1.4014,21,22.

Figure1GshowsthereferencebeadPSDasdetectedbyNTA.Bycombiningtwomeasurementswithdifferentsettings,NTAdetectedallreferencebeadsizes,althoughonly5%ofthe46‐nmbeadsweredetected.Trackingof46‐nmbeadswashinderedbythepresenceoflargerbeadsthatsaturatedthecamera.Figure1Galsoshowsthatthepeaksoverlapbecauseofbroadening,whichweattributetotheuncertaintyinthemeasureddiffusioncoefficient,resultingfromalimitedtracklengthandtheuncertaintyintheparticleposition.Figure1HshowsthevesiclestandardPSDobtainedbycombiningtwodifferentsettings.Thepeakat95nmisbroadascomparedwithothervesiclePSDs.Thesmallestdetectablevesiclesappeartobe10nm,whichweattributetobroadeningofthePSD.Usingidenticalsettings,wecoulddetectonly5%ofthe46‐nmpolystyrenebeads,whichhavecomparablelightscatteringtoa70–90‐nmvesicle.

Figure1IshowsthereferencebeadPSDasdetectedbyRPS.ThroughcombinationofmeasurementswithanNP100andNP400pore,beadsof102nmandlargerweredetected.ThepeaksarebroadenedascomparedwiththereferencebeadPSD,whichmaybecausedbyparticleaggregation,electronicnoise,andavaryingporedimensionduringthemeasurement.Figure1JshowsthevesiclestandardPSDwithapeakat75nm.

ThePSDsofvesiclesarefittedbyamathematicalfunctiontoenablequantitativecomparison.Toselectthemostappropriatefunction,wefittedthevesiclestandardPSDwithsixempiricalfunctionsthatarefrequentlyusedtodescribePSDsofparticlesinsuspension16,andperformedgoodness‐of‐fittests(DataS1).TheGammafunction,Weibulldistributionandpower‐lawfunctionresultedinthebestfits.Ofthesefunctions,weselectedthepower‐lawfunction,asitisleastsusceptibletominimumdetectablevesiclesize.TherightpanelsofFig.1showPSDsofthevesiclestandardsfittedbythepower‐lawfunction(dashedlines).

ThemeasuredreferencebeadPSDsinFig.1werefittedbyasumofGaussianfunctions(dottedlines)toderivethemeanandstandarddeviationofthesizeandtheconcentrationforeachsubpopulationofbeads(Table1).ThesymbolsinFig.3Aindicatetherelativemeasurementerrorofthesizeasthepercentagedifferencebetweenthemeasurementandthemanufacturerspecification.BecauseTEMdatawereusedasreference,itsrelativesizeerrorissetat0%.Therelativesizeerroroftheothermethodswas9%.Dedicatedflowcytometryhadthelowesterrorinsizingbeads,followedbyRPS,conventionalflowcytometry,andNTA.Weattributethelowerrorofflowcytometrytothehomogeneousrefractiveindexofpolystyreneandthestrongrelationshipbetweensizeandscatteringpower(Fig.2).TheerrorofRPSwaslimitedbecauseofspecificmeasurementrestrictions,asdescribedinMaterialsandmethods.WeattributetherelativelylargeerrorofNTAtotheuncertaintyinthemeasureddiffusioncoefficient.

Figure3OpeninfigureviewerPowerPointRelativeerror(symbols)andcoefficientofvariation(CV)(errorbars)indeterminingthesize(A)andconcentration(B)ofTEM,conventionalflowcytometry,dedicatedflowcytometry,nanoparticletrackinganalysis(NTA)andresistivepulsesensing(RPS)for46‐nmbeads(black),102‐nmbeads(red),203‐nmbeads(blue),400‐nmbeads(green)and596‐nmbeads(brown)fromthereferencemixture.Subpopulationsthatcouldnotbedetectedareindicatedbyredcrosses.

TheerrorbarsinFig.3AindicatetheCV,whichisthepercentageratiobetweenthestandarddeviationandthemeansize,andisthusameasureofthewidthofthepeaksinFig.1.OwingtothehighresolutionofTEMascomparedwiththemeasuredstandarddeviationofthebeadsizes,thisstandarddeviationisacloseapproximationoftheactualsizeofthebeads.ThelowestCVswereobtainedbydedicatedflowcytometry,followedbyconventionalflowcytometry,RPS,andNTA.

Figure3Bshowstherelativemeasurementerrorindeterminingtheconcentrationofsubpopulationsofreferencebeads.RPSwasmostaccurateindeterminingtheconcentrationofbeads,followedbyconventionalflowcytometry,dedicatedflowcytometry,andNTA.TheerrorofRPSwaslimitedbecauseofspecificmeasurementrestrictions.Withflowcytometry,theconcentrationwasderivedfromtheflowrate,whichhasanuncertaintyof10%.Dedicatedflowcytometryunderestimatedtheconcentrationof102‐nmbeads,asthesebeadswereclosetothedetectionthreshold.NTAwastheleastaccuratemethodfordeterminingtheconcentrationofbeads,possiblybecauseofbroadeningofthePSDandcrosstalkbetween203‐nmbeadsand400‐nmbeads.Theconcentrationof46‐nmbeadswasunderestimated,astrackingof46‐nmbeadswashinderedbythepresenceoflargerbeadsthatsaturatedthecamera.

Figure4showsthedetectedconcentrationofvesiclespertechnique.AscomparedwithRPSandNTA,conventionalflowcytometryunderestimatestheconcentrationofvesiclesalmost300‐fold,whereasthemoresensitivededicatedflowcytometerunderestimatesthevesicleconcentration15‐fold.WithTEM,thedetectedconcentrationwasaffectedbysamplepreparationlosses.

Figure4OpeninfigureviewerPowerPointTotaldetectedconcentration(logarithmicscale)ofvesiclesdetectedbytransmissionelectronmicroscopy(TEM),conventionalflowcytometry,dedicatedflowcytometry,nanoparticletrackinganalysis(NTA),andresistivepulsesensing(RPS).

Inthisstudy,wecomparedtheabilitiesoffivecommonlyusedmethodstodeterminethePSDofvesiclesinsuspension.AreferencemixtureofpolystyrenebeadswithknownPSD(Fig.1A)andavesiclestandardfromurine(Fig.1B)weremeasuredbyeachmethod.Inagreementwithourtheoreticalreview9,eachtechniquegivesadifferentPSDforthesamesample.BycomparingthevesiclePSDsandcombiningtheseresultswiththeknowledgeobtainedfromreferencebeads,however,manydifferencesarenowexplained.

ThroughouttheResultssection,wehavediscussedtherequirementsandassumptionsinvolvedinthemeasuredPSDs.Table3summarizestheserequirementsandassumptions,andalsoliststheminimumdetectablevesiclesize,themeasurementtimes,andthecapabilitiestoobtainfunctionalinformation,suchasfluorescence.Inthenextsection,wewilldiscussourapproachandtheresultsinmoredetail.

Table3.Assessedcapabilitiesoftechniquesforthedetectionofsinglevesiclesinsuspension

Calibrationwithbeads

Sphericalparticle,n=1.40±0.02

Calibrationwithbeads

Sphericalparticle,n=1.40±0.02

NTA,nanoparticletrackinganalysis;RPS,resistivepulsesensing;TEM,transmissionelectronmicroscopy.Foreachtechnique,theminimumdetectablevesiclesize,abilitytomeasurethesizeandconcentrationandtherequirementsforthis,abilitytodetectadditionalfeatures,andmeasurementtimeareestimated.WederivedtheminimumdetectablevesiclesizeofRPSfromsixmeasurementsperformedwithdifferentNP100pores.disthevesiclediameter,ΔDistheuncertaintyinthediffusioncoefficient,IvandIbarethescatteringintensitiesofavesicleandacalibrationbead,respectively,ηistheviscosityofthesolvent,nisthevesiclerefractiveindex,Qistheflowrate,QPistheflowratecausedbyexternalpressure,σvareσmaretheelectricalconductivitiesofavesicleandthemedium,respectively,andTisthetemperatureofthesolvent.ThemeasurementtimeisindicatedbyS,M,andH,meaning1min,between1minand1h,and1h,respectively.

Toobtainthesizeofthereferencebeads(Fig.1A),weusedtraceableTEMmeasurementsofthemanufacturer.TraceabilitymeansthatthemeasurementresultisrelatedtoSIunitsthroughanunbrokenchainofcomparisonswithknownuncertainties26,27.Tocharacterizebeads,TEMisparticularlyuseful,asbeadsarenotaffectedbysamplepreparation,andtheresolutionofTEMishigherthanthesizeofthebeads.Forcomparisonpurposes,wesettherelativesizeerrorofTEMto0(Fig.3A).However,therelativesizeerrorofthereferencebeadsrangesfrom1.0%forthe596‐nmbeadsto4.3%forthe46‐nmbeads.TheTEMdataalsoprovidetheCVofthebeads,whichisameasureofthespreadinbeadsizes.Consequently,theerrorbarsinFig.3Arepresentnotonlytheimprecisionoftheinstrument,i.e.thebroadeningofthereferencebeadPSDcausedbytheinstrument,butalsotheCVofthereferencebeads.

Wederivedthereferencebeadconcentrationsfromthemanufacturer‐specifiedmassconcentration,densityandsizeofthebeads.Notethattheconcentrationofsubmicrometerbeadsisnottraceable,asuncertaintiesinthemassconcentrationanddensityofthebeadsareunknown.Themassconcentrationisoftenprovidedwithsingle‐digitprecision,andthedensityofsilicabeadsmayrangefrom1.8to2.5gcm−3.Consequently,thebeadconcentrationanderrorthereofareunknown,andtherelativeconcentrationerrorscanonlybemutuallycompared(Fig.3B).

TEManalysisofvesiclesinvolvestwocentrifugationstepsandextensivesamplepreparation.ToquantifytheinfluenceofthesepreanalyticvariablesontheobtainedPSD(Fig.1B),weoverlappedthepower‐lawfunctionsofTEMandRPS,whichrequiredhorizontalandverticalstretchingoftheRPSdatawithfactorsof0.88and0.21,respectively.ConsideringRPStobethemostreliablemethodfordeterminingthePSDofvesicles,wehypothesizethatvesiclesshrinkby12%,owingtofixationanddehydration,andthat21%ofthevesiclesarerecoveredaftercentrifugationandbindingtotheformvarcoating.

Torelatethemeasuredlightscatteringtoaparticlesize,wecalibratedflowcytometersbyusingbeadsandMietheory(Fig.2),assumingsphericalparticlesofknownrefractiveindex.Asbeadsmeetthesecriteria,MietheorycanbeusedtodeterminetheirPSD(Fig.1C,E),resultingindedicatedflowcytometrybeingthemostaccurateinsizingbeads(Fig.3A).However,therefractiveindexofvesiclesisprobablyheterogeneousandnotexactlyknown28,therebyaffectingthePSDofvesiclesobtainedbyflowcytometry(Fig.1D,F).Forexample,undertheassumptionthatthevesiclerefractiveindexis1.40±0.02,theminimumdetectablevesiclesizesare270–600nmforconventionalflowcytometryand150–190nmfordedicatedflowcytometry(Table3).Weattributethehighconcentrationofvesicles340nmobtainedbyconventionalflowcytometryrelativetoothertechniquestobackgroundcounts.Anadvantageofflowcytometryisknowledgeoftheanalyzedsamplevolume,suchthattheparticleconcentrationcanbedeterminedwithoutcalibrationwithbeads.

ThePSDofbeadsdeterminedbyNTAshowsextensivebroadeningascomparedwiththeothertechniques(Figs.1Gand3A).Inaddition,thedeterminedconcentrationofvesiclesrequirescarefulinterpretation.Themanufacturerorusercalibratestheinstrumentwithbeadstorelatethemeannumberofscatterersinthefield‐of‐viewtotheconcentration23.Thiscalibrationisvalidforavesiclesizethatscattersthesameamountoflightasthecalibrationbeads.Theconcentrationofsmallervesiclesisunderestimated,whereastheconcentrationoflargervesiclesisoverestimated.Moreover,theconcentrationofbeadsisnottraceable.Toobtaintheminimumdetectablevesiclesizeof70–90nm,werelatedthescatteringof46‐nmpolystyrenebeads,whichwereatthelimitofdetection,tothediameterofvesiclesbyusingMietheory.

Softwareoftenappliesunknownandundesiredoperationstothedata.Forexample,Fig.5showsthereferencebeadPSDdetectedbyNTAwiththelow‐sensitivitysettingsandprocessedbyntav2.3.5.16(Nanosight).AnalysisofthevideoswiththisnewersoftwareresultsinaPSD(blueline)differentfromthatinFig.1G(blueline).Thesoftwaregeneratesabatchsummaryfile,whereinarollingaverageisappliedtotherawdata,resultinginasmootherbutlesscorrectrepresentationofthedata(grayline).Applicationoffinitetracklengthadjustment(FTLA)resultsinnarrowerpeaks,adecreasedaccuracyofthedeterminedmeandiameters,andthepresenceofanadditionalpeakat445nm(brownline).AsFTLAintroducesartefacts,theapplicationofFTLAtopolydispersesamplesisnotrecommended.

Figure5OpeninfigureviewerPowerPointParticlesizedistribution(PSD)ofthereferencebeadsdetectedbynanoparticletrackinganalysis(NTA)withlow‐sensitivitysettingsandprocessedbyntav2.3.5.16software(Nanosight).Thebinwidthis10nm.Thesoftwareprovidesrawdata(blueline),arollingaverageofthedata(grayline),anddataprocessedwiththefinitetracklengthadjustment(FTLA)algorithm(brownline),resultingindifferentPSDs.TheFTLAalgorithmresultsinapeak(brownarrow)thatisabsentintherollingaverageofthedata.

AccuratesizingofvesiclesbyRPSrequiresthattheelectricalconductivityofaparticleisnegligibleascomparedwiththeconductivityoftheelectrolyte16,29.Aspolystyrenebeads,cellsandintactvesiclesmeetthisrequirement30,webelievethatthedetectedvesiclesizeisrepresentativeforurinaryvesicles(Fig.1J).However,ourmeasurementrestrictionsmadetheRPSmeasurementsimpractical.ThemajorconcernswithRPSareporecloggingandporestability.Inthecaseofporeclogging,wereversedthepressureortemporarilyappliedahighpressurewithaplunger.Astheporesarestretchable,plungingmaychangetheporedimensions,asobservedbyachangeinthebaselinecurrent.Ifthebaselinecurrentchangedby5%,werepeatedthemeasurementandcalibration,resultinginameasurementtimeofseveralhours.

Theconcentrationisobtainedbycalibrationwithbeads29,whichisinaccurate,becausetheconcentrationofusedbeadsisnottraceable.Astheflowrateismainlydeterminedbypressureacrossthepore,andnotbyelectro‐osmosisorelectrophoresis,thedifferencesbetweenthezetapotentialsofvesiclesandcalibrationbeadsarenegligible.Consequently,theaccuracyindeterminingthevesicleconcentrationisexpectedtobecomparabletothatforthemixtureofbeads.

Abiomarkerbasedonvesicleenumerationshoulddeterminetheconcentrationofaspecificvesicletype.Forthisdetermination,thetechniquemustobtainbiochemicalinformationtoidentifyspecificvesicles,andthemeasurementtimeshouldnotexceedseveralminutes.Furthermore,sizeaccuracyandprecisionareimportant,e.g.todistinguishvesiclesfromplatelets.Ourfindingsdemonstratethatanyreportedconcentrationneedstobeaccompaniedbytheminimumdetectablevesiclesize.Forexample,theshadedareainFig.1Jshowsthatadecreaseintheminimumdetectablevesiclesizefrom80nmto60nmwouldresultina2.4‐foldincreaseintheobtainedconcentration.Therefore,weproposedailymonitoringoftheminimumdetectablevesiclesize,asday‐to‐dayvariationisexpectedforeachinstrument.Alternatively,apower‐lawfitmaybeappliedtocompareconcentrationsobtainedwithdifferentminimumdetectablevesiclesizes.Anadditionalrequirementforcomparisonofconcentrationsistraceabledeterminationofbothsizeandconcentration,whichisproblematicfortechniquesthatcalibratetheconcentrationwithuntraceablebeads.

Inconclusion,eachtechniquegaveadifferentPSDforthesamevesiclesample.Differencesbetweenthedetectedvesicleconcentrationsareprimarilycausedbydifferencesbetweentheminimumdetectablevesiclesizes.Theminimumdetectablevesiclesizeswere70–90nmforNTA,70–100nmforRPS,150–190nmfordedicatedflowcytometry,and270–600nmforconventionalflowcytometry.TEMcoulddetectthesmallestvesiclespresent,albeitafteradhesiononasurface.Dedicatedflowcytometrywasmostaccurateindeterminingthesizeofreferencebeads,butisexpectedtobelessaccurateonvesicles,owingtoheterogeneityoftherefractiveindexofvesicles.Areliableestimateofthevesiclerefractiveindexisrequiredtoconverttheopticalscattersignaldetectedbyflowcytometrytosize.Nevertheless,dedicatedflowcytometryisrelativelyfastandallowsmultiplexfluorescencedetection,makingitmostapplicabletoclinicalresearch.

E.vanderPol,R.Nieuwland,andT.G.vanLeeuwenconceivedanddesignedtheresearch.E.vanderPolandA.E.Grootemaatacquiredthedata.C.Gardiner,E.vanderPol,F.A.Coumans,R.Nieuwland,andT.G.vanLeeuweninterpretedthedata.E.vanderPol,F.A.Coumans,andR.Nieuwlandwrotethemanuscript.Allauthorsreviewedandmadecriticalrevisionstothemanuscript.

TheauthorswouldliketoacknowledgeE.vanderPol,WageningenUniversity,theNetherlands,forstatisticalsupport.PartofthisworkwasfundedbytheEMRP(EuropeanMetrologyResearchProgramme)undertheJointResearchProjectHLT02(www.metves.eu).TheEMRPisjointlyfundedbytheEMRPparticipatingcountrieswithintheEuropeanAssociationofNationalMetrologyInstitutesandtheEuropeanUnion.

jth12602-sup-0001-Supplementary_material.pdfapplication/PDF,606.4KBDataS1.Methodsandmathematicalfunctiontofittheparticlesizedistributionofvesicles.

Pleasenote:Thepublisherisnotresponsibleforthecontentorfunctionalityofanysupportinginformationsuppliedbytheauthors.Anyqueries(otherthanmissingcontent)shouldbedirectedtothecorrespondingauthorforthearticle.

1Conde‐VancellsJ,Rodriguez‐SuarezE,EmbadeN,GilD,MatthiesenR,ValleM,ElortzaF,LuSC,MatoJM,Falcon‐PerezJM.Characterizationandcomprehensiveproteomeprofilingofexosomessecretedbyhepatocytes.JProteomeRes2008;7:5157–66.
2RatajczakJ,WysoczynskiM,HayekF,Janowska‐WieczorekA,RatajczakMZ.Membrane‐derivedmicrovesicles:importantandunderappreciatedmediatorsofcell‐to‐cellcommunication.Leukemia2006;20:1487–95.
3SimonsM,RaposoG.Exosomes–vesicularcarriersforintercellularcommunication.CurrOpinCellBiol2009;21:575–81.
4vanderPolE,BöingAN,HarrisonP,SturkA,NieuwlandR.Classification,functionsandclinicalrelevanceofextracellularvesicles.PharmacolRev2012;64:676–705.
5BerckmansRJ,SturkA,SchaapMC,NieuwlandR.Cell‐derivedvesiclesexposingcoagulanttissuefactorinsaliva.Blood2011;117:3172–80.
6ManlyDA,WangJG,GloverSL,KasthuriR,LiebmanHA,KeyNS,MackmanN.Increasedmicroparticletissuefactoractivityincancerpatientswithvenousthromboembolism.ThrombRes2010;125:511–12.
7RautouPE,LeroyerAS,RamkhelawonB,DevueC,DuflautD,VionAC,NalboneG,CastierY,LesecheG,LehouxS,TedguiA,BoulangerCM.MicroparticlesfromhumanatheroscleroticplaquespromoteendothelialICAM‐1‐dependentmonocyteadhesionandtransendothelialmigration.CircRes2011;108:335–43.
8ELAndaloussiS,MaegerI,BreakefieldXO,WoodMJA.Extracellularvesicles:biologyandemergingtherapeuticopportunities.NatRevDrugDiscovery2013;12:348–58.
9vanderPolE,HoekstraAG,SturkA,OttoC,vanLeeuwenTG,NieuwlandR.Opticalandnon‐opticalmethodsfordetectionandcharacterizationofmicroparticlesandexosomes.JThrombHaemost2010;8:2596–607.WileyOnlineLibrary
10vanderPolE,CoumansF,VargaZ,KrumreyM,NieuwlandR.Innovationindetectionofmicroparticlesandexosomes.JThrombHaemost2013;11:36–45.WileyOnlineLibrary
11deVrijJ,MaasSL,vanNispenM,Sena‐EstevesM,LimpensRW,KosterAJ,LeenstraS,LamfersML,BroekmanML.Quantificationofnanosizedextracellularmembranevesicleswithscanningionocclusionsensing.Nanomedicine2013;1–16.
12DragovicRA,GardinerC,BrooksAS,TannettaDS,FergusonDJP,HoleP,CarrB,RedmanCWG,HarrisAL,DobsonPJ,HarrisonP,SargentIL.Sizingandphenotypingofcellularvesiclesusingnanoparticletrackinganalysis.Nanomedicine2011;7:780–8.
13YuanaY,OosterkampTH,BahatyrovaS,AshcroftB,GarciaRP,BertinaRM,OsantoS.Atomicforcemicroscopy:anovelapproachtothedetectionofnanosizedbloodmicroparticles.JThrombHaemost2010;8:315–23.WileyOnlineLibrary
14KonokhovaAI,YurkinMA,MoskalenskyAE,ChernyshevAV,TsvetovskayaGA,ChikovaED,MaltsevVP.Light‐scatteringflowcytometryforidentificationandcharacterizationofbloodmicroparticles.JBiomedOpt2012;17:0570061–8.
15vanderPolE,vanGemertMJC,SturkA,NieuwlandR,vanLeeuwenTG.Singleversusswarmdetectionofmicroparticlesandexosomesbyflowcytometry.JThrombHaemost2012;10:919–30.WileyOnlineLibrary
16JonaszM,FournierGR.LightScatteringbyParticlesinWater,1stedn.London:AcademicPress,2007.
17NieuwlandR,vanderPolE,GardinerC,SturkA.Platelet‐derivedmicroparticles.In:ADMichelson,ed.Platelets,3rdedn.SanDiego,CA:AcademicPress,2012:453–67.
18YuanaY,BertinaRM,OsantoS.Pre‐analyticalandanalyticalissuesintheanalysisofbloodmicroparticles.ThrombHaemost2011;105:396–408.
19BohrenCF,HuffmanDR.AbsorptionandScatteringofLightbySmallParticles.NewYork,NY:Wiley,1983.
20MätzlerC.MATLABfunctionsforMiescatteringandabsorption.2002–11.2002.InstitutfürAngewandtePhysik.
21vanManenHJ,VerkuijlenP,WittendorpP,SubramaniamV,vandenBergTK,RoosD,OttoC.Refractiveindexsensingofgreenfluorescentproteinsinlivingcellsusingfluorescencelifetimeimagingmicroscopy.BiophysJ2008;94:L67–9.
22BeuthanJ,MinetO,HelfmannJ,HerrigM,MullerG.Thespatialvariationoftherefractiveindexinbiologicalcells.PhysMedBiol1996;41:369–82.
23GardinerC,FerreiraYJ,DragovicRA,RedmanCWG,SargentIL.Extracellularvesiclesizingandenumerationbynanoparticletrackinganalysis.JExtracellVesicles2013;2:1–11.
24ThéryC,AmigorenaS,RaposoG,ClaytonA.Isolationandcharacterizationofexosomesfromcellculturesupernatantsandbiologicalfluids.CurrProtocCellBiol2006;30:3.22.1–22.29.WileyOnlineLibrary
25JensenOA,PrauseJU,LaursenH.ShrinkageinpreparatorystepsforSEM–astudyonrabbitcornealendothelium.AlbrechtVonGraefesArchKlinExpOphthalmol1981;215:233–42.
26MeliF,KleinT,BuhrE,FraseCG,GleberG,KrumreyM,DutaA,DutaS,KorpelainenV,BellottiR,PicottoGB,BoydRD,CuenatA.Traceablesizedeterminationofnanoparticles,acomparisonamongEuropeanmetrologyinstitutes.MeasSciTechnol2012;23:125005.
27VargaZ,YuanaY,GrootemaatAE,vanderPolE,GollwitzerC,KrumreyM,NieuwlandR.Towardstraceablesizedeterminationofextracellularvesicles.JExtracellVesicles2014;3:1–10.
28IssmanL,BrennerB,TalmonY,AharonA.Cryogenictransmissionelectronmicroscopynanostructuralstudyofshedmicroparticles.PLoSOne2013;8:e83680.
29KachelV.Electricalresistancepulsesizing:Coultersizing.In:RAHoffman,ed.FlowCytometryandSorting,2ndedn.NewYork,NY:Wiley‐Liss,1990:45–80.
30MontesinosE,EsteveI,GuerreroR.Comparisonbetweendirectmethodsfordeterminationofmicrobialcell‐volume–electron‐microscopyandelectronicparticlesizing.ApplEnvironMicrobiol1983;45:1651–8.

Thefulltextofthisarticlehostedatiucr.orgisunavailableduetotechnicaldifficulties.

Pleasecheckyouremailforinstructionsonresettingyourpassword.Ifyoudonotreceiveanemailwithin10minutes,youremailaddressmaynotberegistered,andyoumayneedtocreateanewWileyOnlineLibraryaccount.

Can"tsignin?Forgotyourusername?

Enteryouremailaddressbelowandwewillsendyouyourusername

Iftheaddressmatchesanexistingaccountyouwillreceiveanemailwithinstructionstoretrieveyourusername