University of Groningen
Coupled adhesion of bacteria to surfaces
Skogvold, Rebecca van der Westen
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Skogvold, R. V. D. W. (2018). Coupled adhesion of bacteria to surfaces. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
C H A P T E R
2
Quantification of the Viscoelasticity of the
Bond of Biotic and Abiotic Particles
Adhering to Solid-Liquid Interfaces using
a Window-equipped Quartz Crystal
Microbalance with Dissipation
This chapter is published with permission from Elsevier:
Rebecca van der Westen, Henny C. van der Mei, Hans De Raedt, Adam L. J. Olsson, Henk J. Busscher and Prashant K. Sharma
ABSTRACT
The quartzͲcrystalͲmicrobalanceͲwithͲdissipation (QCMͲD) has become a powerful tool for studying the bond viscoelasticity of biotic and abiotic colloidal particles adhering to substratum surfaces. A windowͲequipped QCMͲD allows highͲthroughput analysis of the average bond viscoelasticity,measuringover106particlessimultaneouslyinonesingleexperiment.Othertechniques requirelaboriousanalysesofindividualparticles.Inthisprotocol,thequantitativederivationofthe springͲconstantanddragͲcoefficientofthebondbetweenadheringcolloidalparticlesandsubstratum surfacesusingQCMͲDisexplainedforbacteriaandsilicaparticles,usingtheparticleͲmassderivedfor validation.Bondviscoelasticityiscalculatedusingacoupledresonatormodel,payingspecialattention totheprotocolformathematicalfittingneededtoobtainreliablequantitativeoutput.Knowledgeof the viscoelasticity of the bond between colloidal particles and substratum surfaces facilitates developmentofnewstrategiestodetachadheringparticlesfromorretainthemonasurface.
INTRODUCTION
Controlovertheadhesionofbiotic(suchasbacteria)andabioticcolloidalparticles(suchas silica, polystyrene or latex particles) is a key concern in engineering and medicine. In particular, adhesionofbacteriatosurfacescanformahazardtohumanhealth,1whileadhesioncontrolofabiotic
particlesisessentialinareassuchassensinganddatastorage.2,3Thebondbetweenacolloidalparticle
andasubstratumsurfaceisseldomrigidandmostlycomprisesanelasticandviscouscomponent.4,5
Theviscoelasticityofabondisnotonlydeterminantforparticleadhesion,butalsoforthemechanism of particle detachment.6,7 For bacteria, the viscoelastic properties of their bond with a substratum
surface often allow adhering bacteria to remain adhering under shear conditions through gradual elongationofthebond.8
Severalexperimentaltechniqueshavebeendevelopedtostudytheadhesivebondbetween adhering colloidal particles and a substratum surface, such as the use of atomic force microscopy (AFM)9–11andopticalormagnetictweezers.12Inthesetechniques,asinglecolloidalparticleisforced
tocontactasubstratumsurfaceafterwhichitispulledoffandtheforcerequiredtobreakthebondis takenastheadhesiveforce.ElasticityandviscosityofthebondcanbemeasuredusingAFMbypressing aparticleonasubstratumsurfaceunderaconstant,appliedforceandmeasuringdeformationorby measuring the force resulting from an applied, constant deformation of the particle. Using this approach,Luetal.13describedthebondcomponentsofbacterialcellsurfacesasaspring,representing
theelasticcomponentasrecognizedinthestandardsolidmodel,placedinaserieswithacombination ofaspringandadashpotinparallel.Inordertoensurethatsuchmodelingonlycomprisesthebond betweenanadheringparticleandasubstratumsurfaceandnotthebulkofaparticleasintraditional Hertz, JohnsonͲKendallͲRoberts or DerjaguinͲMullerͲToropov models, Chen et al.14 suggested to
considerthebondbetweenanadheringparticleandasubstratumsurfaceasacylindricalvolumethat deforms under conditions of constant volume to provide a method allowing to confine traditional analysistothebonditself.AnalysisoftheBrownianmotioninducednanoscopicvibrationsexhibited bybiotic15andabiotic7,16colloidalparticlesoffersacompletelydifferentwaytoobtaintheelasticityof
thebond,withsmallervibrationalamplitudesbeingindicativeofhigherelasticity.15Apartfromthe
assumptions involved in each of the aboveͲmentioned techniques,they all possess the common limitationthatadheringparticlesmustbestudiedoneͲbyͲoneonanindividualbasis,whichmakesit hardtoobtainstatisticallyreliable,quantitativedata.
ThequartzͲcrystalͲmicrobalanceͲwithͲdissipation(QCMͲD) effectively avoids this “oneͲbyͲ one”drawback,andhasbeenappliedtoanalyzetheviscoelasticityofthebondbetweenbioticand abioticparticlesadheringtosolidͲliquidinterfacesoverlargenumbersofadheringparticles,typically intheorderof1010perm2ofasensorsurface,approximatelyequivalentto106particlesonthesensor
surface.NanometerͲscale shear oscillations of the sensor cause deformation of the bonds with an adheringparticle,withanopposingforcearisingfromthesurroundingliquid.TheQCMͲDregistersthe shiftinresonancefrequencyofthesensor(ȴf)duetoparticleadhesionaswellastheenergylossto the surrounding liquid (change in “dissipation”(ȴD)). Moreover, for studies involving colloidal particles,itisadvantageoustouseaQCMͲDequippedwithawindowchamber,allowingsimultaneous microscopicregistrationofthenumberofparticlesadheringtothesensorsurface.
Traditionally,QCMͲDhasbeenmostlyusedtodetermineadsorptionofmolecularmasstoa sensor surface, assuming the adsorbed mass directly couples to the sensor surface. According to Sauerbrey’s relation,17 an adsorbed coupled mass increases the effective sensor mass, yielding a
reductioninthesensorresonancefrequency,orinQCMͲtermsanegative(resonance)frequencyshift. Since the penetration depth for the shear wave in QCMͲD is less than 250 nm (at 5 MHz), this representsthemaximumthicknessofadsorbedfilmsthatcanbereliablymeasured.Thesensitivityof the mass detection in QCM is in the nanogram range. Particles however, do not necessarily massͲ coupletothesensorsurface,butinsteadmayadhereascoupledresonators.18–21Thesensorresonates atdifferentfixedfrequencies.A5MHzsensorresonatesnotonlyat5MHzbutalsoatitsovertones (15,25upto65MHz).Providednottouchingeachother,allcolloidalparticlesadheringtothesensor surfaceactasindividual,coupledresonators(Figure1a)withanimpactontheresonancefrequency shift(ȴf)measured.Energydissipationchange(ȴD)ismaximalwhentheparticleresonancefrequency (fp)matchesthesensorresonancefrequency(fs).Thedevelopmentofthecoupledresonatormodel hasgreatlywidenedthepossibilitiesofQCMͲD,whichwerepreviouslyconfinedtomolecularmass adsorption.Itisinterestingtonotethatwhereasbondviscoelasticitiesofindividualparticlesofthe same kind often show large variations,8,15,22 QCMͲD identifies a wellͲdefined zeroͲvalue in sensor
frequency (fZC), see Figure 1b). Zero crossing frequencies are only observed when the adhering colloidalparticlesoscillateatfrequencieswithinthewindowofthesensorresonancefrequencyand itsobservableovertones,whichrangefrom5MHzto65MHz(seeFigure1b).Positivefrequencyshifts observedintheliterature,couldnotbeexplainedpriortotheintroductionofthecoupledresonator model.4,5,19 Figure1.Schematicdiagramofthecoupledresonatormodel. a)Colloidalparticlewithmass,mpadheringtoasensorsurfacethroughaviscoelasticbondcomprised ofaspringwithspringconstant(k),anddashpotwithadragcoefficient(ʇ),yieldingaparticleresonance frequencyfp.Inthecurrentscheme,springanddashpotareplacedinparallel,i.e.asintheKelvinͲVoigt modelofviscoelasticresponse. b)Theoreticalshiftsinthesensorresonancefrequency(ȴfs)anddissipationchanges(ȴD)forparticles
adhering with different adhesive bond stiffnesses to a QCMͲD sensor, as explained in the coupled resonatormodel,asafunctionoftheQCMͲDresonancefrequencyfs.Particleresonancefrequencyfp
increaseswithadhesivebondstiffness,andthefrequencyofzeroͲcrossing(fs=fp)canonlybeobserved
withinwhatisreferredtoas‘thewindowofobservableQCMͲDfrequencies’,includingitsresonance frequencyandovertones,indicatedbythedashedrectangles.Theasymptoticshiftfromanegativeto apositiveȴfoccurswhenfpequalsfsandisaccompaniedbyamaximuminȴD.
ʇ
k
m
pa
b
2
Inthecoupledresonatormodel,thefrequencyshiftsanddissipationchangesderivedfromthe QCMͲDcanberelatedinaKelvinͲVoigtmodeltotheviscoelasticityofthebondaccordingto ο ୧ୈ౩ ଶ ൌ ూ୫౦ ౧ ή ୮ ன౩య൫ன౦మିஓమ൯ିன౩ன౦ర ሺன౩మିன౦మሻమାன౩మஓమ ன౩రஓ ሺன౩మିன౦మሻమାன౩మஓమ൨ [1]
where ȴf (Hz)istheshift inQCMͲDresonance frequency,ȴDis thechangeindissipation,fFisthe
fundamentalresonancefrequencyofthesensor(5MHz),fsistheQCMͲDsensorsurfaceresonance
frequency,mpistheparticlemass(kg),ʘsisthesensorresonanceangularfrequency(ʘs=2ʋfs),ʘpis
theparticleresonanceangularfrequency(ʘp=2ʋfp),ZqistheacousticimpedanceofanATͲcutquartz
crystal(8.8x106kgmͲ2sͲ1),N
pisthenumberofadheringparticlesperunitsensorarea(m2),ɶequals
ʇ/mpwithʇbeingthedragcoefficient(seeFigure1a),indicativeoftheviscouscomponentofthebond.
The elastic component of the bond follows from ʘp equaling with ට ൗ ୮ . Note that the coupled
resonatormodelasdescribedinEq.1assumescompletecouplingofadheringparticleswiththesensor surface,includingtorsionandsheardeformationofthesensorsurfaceduetoparticleoscillation.23
Moreover,itisassumedthatthespringanddashpotareindependentfromthefrequencyaccording to the coupled resonance model, which means that the frequency should not be causing any perturbationsoftheadhesivebond.Stillitshouldbenotedthatbondstiffnessesdeterminedusingthe QCMͲDmaybeinfluencedbytheexperimentalconditionofhavingbeendeterminedintheMHzrange. Inthisprotocolpaper,weexplaintheusetheQCMͲDanddataanalysistoobtainelasticities (thespringconstant,k)andviscosities(thedragcoefficient,ʇ)ofthebondofbioticandabioticparticles adhering to sensor surfaces, as calculated with the coupled resonator model. Knowledge of the elasticityandviscosityofthebondthroughwhichcolloidalparticlesadheretosubstratumsurfaceswill assist in a better understanding of the mechanisms of particle adhesion and in developing new strategiestodetachadheringparticlesfromorretainingthemonasurface.
MATERIALSANDPREPARATORYPROCEDURESAPPLIED Reagentsused x 1Ͳoctadecanothiol(SigmaͲAldrich,Zwijndrecht,TheNetherlands) x 11ͲmercaptoͲ1Ͳundecanol(SigmaͲAldrich,Zwijndrecht,TheNetherlands) x Ammonia(NH3)(Merck,Darmstadt,Germany) x Calciumchloride(CaCl2)(Merck,Darmstadt,Germany) x Ethanol100%(VWRChemicals,FontenayͲSousͲBois,France) x Hydrogenperoxide(H2O2)(Merck,Darmstadt,Germany) x Ultrapurewater(>18Mɏcm) x Potassiumchloride(KCl)(SigmaͲAldrich,Zwijndrecht,TheNetherlands) x Potassiumphosphate(KH2PO4)(Merck,Darmstadt,Germany) x Sodiumdodecylsulphate(SDS)2%(w/v)(Merck,Darmstadt,Germany) x Silicaparticles(radius0.5ʅm)(Bangslaboratories,Inc.,Fischer,IN,USA) x ToddHewittbroth(THB)(Oxoid,Basingstoke,UK) Mainequipmentused x Centrifuge(JͲlite,JLA16.250FixedAngleRotor,BeckmanCoulter,CA,USA) x Sonicatorbath(TranssonicTP640,ElmaGmbH&CoSingen,Germany) x Sonicator(VibraCellmodel375:Sonicaandmaterials,DanburyCT,USA) x UV/Ozone(BioforceNanosciences,SloughBerkshire,UnitedKingdom)
x Quartz Crystal Microbalance with Dissipation Monitoring (QCMͲD) E1 system (QͲSense AB, Stockholm,Sweden)
x CCDcamera(ModelA101,Baslervisiontechnologies,Ahrensburg,Germany)
x Metallurgicalmicroscopewith20xobjective(LeicaDM2500M,Rijkswijk,TheNetherlands) x PeristalticPump(Ismatec,Wertheim,Germany)
x QCMcrystalswithsiliconoxideandgoldcoatings(QͲSenseAB,Stockholm,Sweden)
NOTE Most QCMͲD systems operate at a fixed driving power to bring the crystal into resonance.Currentlytheimpactofdrivingpowerontheoscillatorybehaviorofparticlesadheringto aQCMͲDcrystalsurfaceisnotknown.
Preparatoryproceduresapplied Preparationofbacterialsuspensions.Twodayspriortotheactualadhesionexperiment,makeapreͲ cultureofthebacterialstrain,selectasinglecolonyfromanagarplateandinoculateinto10mlTHB, andincubateat37°C.Twentyfourhourslater,pourthe10mlsolutionofTHBandbacteriaintoa200 mlsolutionofTHB,andgrowat37°Cfor16h,afterwhichharvestingofthebacteriacanbegin. Harvestthebacteriabycentrifuging(5minat5000g)andbywashingthesuspensionin100 mlbuffer(50mMpotassiumchloride,2mMpotassiumphosphateand1mMcalciumchloride,pH6.8) twotimes,followedbysonicating10mlofbacterialsuspensionthreetimesfor10sat30W,while cooling in an ice bath. Sonication is particularly needed for streptococci as they grow in chains, consideringsinglebacteriaarepreferredformostexperiments.Afterwards,thebacterialsuspension iscentrifuged,andwashedonelasttimein100mlbufferbeforebeingdilutedinthebuffertoafinal concentrationof3x108bacteriaperml. CrucialstepBywashingthebacteriaaftersonication,itisensuredthatfreemoleculesthat mighthavebeenreleasedduringthesonicationareremovedfromthesuspension.Freemolecules yielddirectmasscouplingwhenadsorbingtoaquartzcrystal,givingrisetonegativefrequencyshifts duringparticulateQCMͲDmeasurements,therebyseverelycomplicatingdataanalysis. Preparationofsilicaparticlesuspensions.Silicaparticleswitharadiusof0.5ʅmwerewashedtwiceby centrifugationin10mlofultrapurewater,anddilutedtoafinalconcentrationof2x108particlesper ml.Forthesilicaparticleadhesionexperiment,asuspensionin50mMKCl,pH6.8wasprepared. SetͲupoftheQCMͲD.ColloidalparticleadhesionisstudiedunderflowusingaQͲsenseE1window chamber.Foropticalmonitoring,aCCDcameraisconnectedtothemicroscopeinordertofacilitate realͲtimemonitoringofparticleadhesionandtheirenumeration(seeFigure2).TheQCMͲDwindow chamberisdiscͲshaped(diameter14mm)encompassingavolumeofapproximately100ʅlcombined withaninletandoutletarea.Fluidflowisestablishedusingaperistalticpumpandcanbeswitched frombuffertoacolloidalparticlesuspensionbyinsertingtheattachedtubingsintoacontainerwith thedesiredfluid.
Figure2.SchematicsoftheQCMͲDsetͲup.ThesetͲupconsistsofaninletandanoutletwheretubings canbeattachedtoleadingdifferentfluidsthroughaperistalticpumpthroughthewindowchamber intoawastecontainer.Inthewindowchamber,aquartzcrystalsensorisplacedbetweenapairof electrodesandbyapplyinganACvoltageovertheelectrodes,thecrystalisbroughttooscillationatits acousticresonancefrequencyfs.Whenthisvoltageisturnedoff,theoscillationdecaysexponentially fromwhichthedissipationcanbedetermined. METHODSACCORDINGTOPROTOCOL QCMͲDmeasurementprocedure 1.Cleanthequartzcrystalsbasedonsupplier’sinstructions.Goldcoatedcrystalsaregenerallycleaned byimmersionina3:1:1mixtureofultrapurewater,NH3(28%)andH2O2(30%)at70°Cfor10min.Silica
coatedcrystalsmustbecleanedbysubmergingtheminto2%(w/v)sodiumdodecylsulphate(SDS)for 15mininasonicatingbath,followedbysubmersioninultrapurewateralsoinasonicatingbathfor15 min.Asafinalstep,independentofapossiblesurfacecoatingofthecrystals,itisessentialtoremove molecularcontaminantsfromthecrystalsurfacebyputtingtheminanUV/ozoneenvironmentfor15 min.
Crucial step H2O2 should not be added until the temperature of ultrapure water and NH3
reaches70°CasthereactionbetweenNH3andH2O2requiredforpropercleaningofthequartzcrystals
islesseffectiveatlowertemperature.
2. Freshly cleaned goldͲcoated quartz crystals are hydrophilic by nature, but can simply be made hydrophobicbyleavingthecleanedgoldcrystalsovernightinambientair.Thecrystalmayrequirea further coating to adjust its surface composition and associated physicoͲchemical properties. SelfͲ assembled monolayers (SAMs) for instance, can be applied rendering the gold coated crystals hydrophobicbysubmersingthefreshlycleanedcrystalsintoasolutionof0.001M1Ͳoctadecanethiol orhydrophilicbysubmersingthemintoasolutionof0.001M11ͲmercaptoͲ1Ͳundecanoldissolvedin 100%ethanolfor18h.
Crucial step During the coating process, it is important in order to obtain a homogeneous coatingonthecrystals,thatthecrystalsareleftinthesolutionforthefull18h,sincethealkylchains needtimetoalignthemselvestogiveahomogeneouscoatingonthesurface.Itisadvisabletoperform somesortofcharacterizationofthecrystalsurfaceinordertoruleoutthatpossibledeviatingresults are due to aberrant properties of the crystal surface. Water contact angle measurements usually suffice to this end, but when more chemical confirmation of the crystals surface composition is required, XͲray Photoelectron Spectroscopy may be considered, amongst other surface chemical analysistechniques.
3.MountthecleanedorcoatedcrystalintheQCMͲD.
Crucial step After mounting of the crystal, it is crucial that the window chamber is not tightenedtoostrongly,sincecrystalsmaybreakduetotheexcessivepressure.
4. Switch on the QCMͲD and controlling software. Adjust the setting to the desired temperature. Throughoutthisprotocol,wehavedoneallourmeasurementsat21°C. 5.Closethewindowchamberandobtainthecrystal’sresonancefrequenciesanddissipationvaluesat thefundamentalfrequencyandobservableovertonesinordertoensurethatthecrystalisingood condition. CrucialstepThedissipationvalueatthecrystalsfundamentalfrequencyshouldbearound40 x 10Ͳ6 or less (oral communication with application specialist, QͲSense, Biolin Scientific AB Västra
Frölunda,Sweden).Ifthedissipationvaluedeviatesmorethan5x10Ͳ6fromthis“normal”value,its causeshouldfirstbedeterminedbeforecontinuingtheexperiment.Sometimesithelpstoloosenor tightenthescrewsthatholdthecrystalinthewindowchamberinordertoregain“normal”values.In othercasesitmayhelptotaketheentirewindowchamberapartandremountthecrystal.Ifnothing helps,thecrystalhastobereplaced.
6.ConnectthetubingstotheQCMͲDchamberandstarttheperistalticpumptointroduceaflow(300 ʅl/min)ofbufferthroughtheQCMͲDwindowchamber.
Crucial step It is crucial to ensure that filling of the system is done free of air bubbles. Air bubblescanbepreventedbytiltingthewindowchamberduringfillingwithbuffer,therebyallowing thebuffertoslowlyrunintothechamberanddisplaceallair.Incaseairbubblesareformed,theycan oftenberemovedbyincreasingthespeedofthefluidflowtoforcethebubblesoutofthesystem.Itis alsopossibletodeͲaeratethebuffersbysonicationpriortousingthemintheexperiment.Airbubbles canspecificallyarisewhenworkingwithcrystalscoatedwithahydrophobiccoating. 7.Findresonancefrequenciesanddissipationvaluesofthesensorcrystalinbufferbylettingthebuffer flow through the system until the frequency is stabilized, typically requiring approximately 5 min. Stabilityisacceptedwhenthedriftinfrequencyshiftsforallobservableovertonesislessthan2Hzper 10min. CrucialstepItisadvisedtosetupalogͲjournalofcrystalresonancefrequenciesanddissipation valuesinbufferaswellasinairtodefine“normal”valuesinbothmediaforthespecificusemade. Afterfillingthechamberwithbuffer,dissipationvaluesusuallyincrease,butthisincreaseshouldnever bemorethantenͲfold(oralcommunicationwithapplicationspecialist,QͲSense,BiolinScientificAB Västra Frölunda, Sweden). Aberrant resonance frequencies and dissipation values after filling the chamberwithbuffercanbeduetothepresenceofminorairbubblesinthechamber.
8. Introduce the suspension containing either biotic or abiotic particles, typically at particle concentrations of around 108particles per ml. Particle adhesion proceeds at a speed that mainly
dependsontheirsedimentationinsuspensionandisusuallyfasterforparticleswithahigherspecific density.Asurfacedensityof1010particlesadheringperm2shouldbeaimedfor,yieldingasurface
coverageformicronͲsizedcolloidalparticlesthatpreventsthemfromtouchingeachother.Particle adhesionshouldbemonitoredusingthemicroscopemountedCCDcamera.Followingtheappropriate adhesiontime,thenumberofadheringparticlesmustbedeterminedusingimageanalysissoftware, whichcanbeeasilywrittenusingtheMatlabplatform.Appropriateprogramsforbiologicalanalysis canalsobedownloadedforfreefromtheinternet,suchasImageJ,CellProfilerandFiji.24 CrucialstepInordertocountthenumberofadheringcolloidalparticlesproperlyusingthe CCDcamera,itisimportantthatbufferisrunningthroughthesysteminordertoremoveanynonͲ adhering particles from the chamber and prevent them from intervening with the enumeration of adhering ones. Alternatively for kinetic analysis of particle adhesion during flow with a particle suspension,imageanalysissoftwarethatfiltersoutparticlesthatchangepositionalongwiththefluid flowshouldbeuse.25
9. The frequency shifts and dissipation changes are retrieved from the QCMͲD at the crystals fundamentalfrequencyanditsovertones,andaresubsequentlyusedinEq.1.Forthepurposeoflater fittingitisadvantageoustoconvertdissipationchanges(ȴD)intoȴȳvalues((ȴDfs)/2),thatisthehalf bandwidthathalfheight(“bandwidth”)oftheresonancecurve.26Thesedataarethenusedtofitthe parametersoccurringinEq.1usingabruteforcefittingprograminFortran90(seeSupplementary Material)toderivevaluesforthespringconstantkthedragcoefficient,ʇ,theparticlemassmp,and thequalityofthefitortheRootMeanSquaredDeviation(RMSD). CrucialstepThenumberofdatapointsthatcanbeobtainedperexperimentisrestrictedby thenumberofovertonesthatcanbeobserved.Withtheinstrumentusedinourstudies,amaximum ofsevendatapointsforȴfandȴȳasafunctionofthecrystalsresonancefrequencycanbeobtainedin eachexperiment.Thisisnotnecessarilyenoughtoreliablycalculatealloutputparameterssincethe iterative numerical procedure can easily get trapped in local minima yielding physically unrealistic results,mostnoticeablyfortheparticlemassandoutputdatashouldbecriticallyevaluatedforbeing physicallyrealistic.Forvalidationoftheoutput,wesuggesttocomparethemassderivedfromthe QCMͲDoutputwiththeparticlemasscalculatedfromitsdimensionsandspecificdensity. ANALYSISOFBRUTEFORCEFITTINGOFQCMͲDTOACOUPLEDRESONATORMODEL GiventhatonlysevenvaluesofȴfandȴDareavailableandthat,also,thescatterinthedata isfrequentlyconsiderable,itisessentialtodeviseafittingalgorithm,whichisrobusttoavoidphysically unrealisticresults.AbruteforcealgorithmisagenericproblemͲsolvingalgorithminwhichallpossible data points are employed to obtain the best fitting parameters, while checking whether each fit satisfiestheproblem'sstatement.Byimplementingthisbruteforcetechnique,physicallyunrealistic, erroneousoutputispreventedduetosolutionsgettingtrappedinlocalminimaofthefittingalgorithm. The algorithm achieves robustness by defining the lower and upper bound of the three output parametersi.e.mp,kandʇandthenthealgorithmdefinesagridofchoicesforparameters(typically
100x100x100parametercombinations)andsearchestheglobalminimumoftherootͲmeanͲsquared deviationbetweenthedataandthefit.Careshouldbetakenhowever,thatfittedparametersdonot hittheboundariesinstalledandwheneverthisisthecase,boundariesshouldbewidened.
To demonstrate advantages and disadvantages of using seven data points of, for instance, triplicateexperimentsinseparatebruteforcefitsversustheuseofall21datapointsinonebruteforce fitforthederivationofthespringconstantk,thedragcoefficientʇandtheparticlemassmpfromthe
QCMͲDoutput,wefirstlypresentresultsfromtriplicateexperimentswithseparatebacterialcultures eachcomprisingsevendatapoints.Inthespecificexamplechosen(seeFigures3aͲcandTable1)we aimed to analyze the properties of the bond between a streptococcal bacterium, Streptococcus salivarius HB7 possessing 91 nm long fibrillar surface appendages, and a hydrophobic SAM on the
crystalsensorsurface.ResultsoffittingthethreeparametersoccurringinEq.1tothedatapointsin eachofthethreeexperimentscanbeseeninFigures3aͲc,whileresultingoutputparametersincluding therootmeansquaredeviation(RMSD)valuesdescribingthequalityofthefitsarepresentedinTable 1.Onaverage,physicallyrealisticresultsareobtainedwiththeaverageparticlemass(theonlyoutput parameterthatcanbecomparedwithanexpectedvalueascanbefoundintheliterature)coinciding wellwiththeonebasedonparticledimensionsandspecificdensity(5x10Ͳ16kg).Qualitiesofthefit arevariableacrossthethreeexperimentsthough.Next,all21datapointsweresubjectedtoasingle bruteforceanalysis(seeFigure3dandTable1),yieldingnearlyidenticalresultsforthespringconstant, dragcoefficientandparticlemassasobtainedafteraveragingthethreedatasetscomprisingseven datapoints.Thustherearenooverridingargumentstoeitherusesevendatapointsfrommultiple experimentsinseparatebruteforcefitsversustheuseofalldatacombinedinonebruteforcefit,else thanthatfitting21pointsdirectlyyieldsanRMSDvalue,oppositetoaveragingthreeindividualfits. Figure3.Comparisonofusingsevendatapointsoftriplicateexperimentsinseparatebruteforcefits (Figs.3aͲc)versustheuseof21datapointsinonebruteforcefit(Fig.3d)forthederivationofthe spring constant k, the drag coefficient ʇ and the particle mass mp from the QCMͲD output for S.
salivariusHB7adheringonahydrophobicSAMonthecrystalsensorsurface.
Notethatin previousstudiesQCMͲD dataand fittingwerepresentedinsoͲcalledpolarplotsofȴf versus ȴȳ,27 we prefer to present both ȴf and ȴȳ versus f as two separate functions, the blue line
representingȴfandtheredlinerepresentingȴȳasafunctionofthesensorresonancefrequency c
a
b
c
d
2
Table1.Valuesforthespringconstantk,dragcoefficientʇandmassoftheparticlesmpobtainedusing sevendatapointsoftriplicateexperimentsinseparatebruteforcefits(seealsoFigs.3aͲc)versusthe useof21datapointsinonebruteforcefit(seeFig.3d)inaKelvinͲVoigtmodel.RMSDindicatesthe qualityofthebruteforcefit. S.salivariusHB7 onahydrophobicSAM k(kg/s2) Ɍ(10Ǧ9kg/s) mp(10Ǧ16kg) RMSD (Hz) ͳǡ ͲǤʹͶ ͷ ʹ͵ ʹǡ ͲǤͶͺ ͳʹ ͺ ʹ ͵ǡ ͲǤ͵ͷ ͳͳ ʹͲ ͳǦ ͵ ͲǤ͵άͲǤͳʹ ͳͲά͵ άʹ Ǧ ͳǦ͵ ʹͳ ͲǤ͵ ͳͲ Ͷʹ TheassumptionunderlyingEq.1thatalladheringparticlescouplecompletelywiththesensor surface to cause torsion and shear deformation is not necessarily true, which reflects in minor deviationsoftheparticlemassderivedfromthetrueparticlemass.23Anotherassumptionunderlying
Eq.1isthatalladheringparticlespossessidenticalsizeandshape.Polydispersityhowever,maygive rise to a distribution in angular particle frequency,20,23 that can be accounted for by adding a
polydispersityparametertoEq.1.Althoughthismayincreasethequalityofthefit,andpossiblyavoid overlysmallspringconstantsasobservedforsilicaparticlesonsilicacrystalsurfacesandbiotinylated crystalsurfaces(seeTable2),fittingoffourparameterstoacomplicatedequationasEq.1withonlya limitednumberofdiscretedatapointbearstheriskofyieldingphysicallyunrealisticvalues. DURATIONOFEXPERIMENT Inthisestimateofthetimerequiredforexperiments,weneglectthetimetoprepareparticle suspensions, as preparation of biotic particle suspensions usually requires much more time due to culturingthanabioticparticlessuspensions,especiallywhencommerciallypurchased.Onceaparticle suspensionhasbeenprepared,atypicalQCMͲDexperimentasdescribedabove,shouldnottakemore than4htoperform.Bruteforcefittingofthedatacanbedonewithin30minaftermeasurements.
EXPECTEDRESULTS Allquantitativepropertiesoftheadhesivebondreportedinthissectionhavebeenobtained byfittingk,ʇ,andmp,asoccurringinEq.1totheQCMͲDfrequencyshiftsanddissipationchangesat theobservablefrequenciesusingabruteforcefittingalgorithmandusingalldatapointsavailableina singlefitwithoutaveragingdataobtainedinseparateexperimentsatthesamefrequency.Figure4 depictsthefittoEq.1ofexperimentaldataforabioticsilicaparticlesandsilicaparticlescoatedwith streptavidin,adheringtoasilicasensorcrystalorasilicasensorcoatedwithbiotinylatedpolyethylene glycol(PEG)alkanethiol.Visualinspectionofthegraphsshowsagoodfittothedatapoints.
Figure4. ȴf and ȴȳ versusfsforabioticsilicaparticlesandsilicaparticles coatedwithstreptavidin,
adheringtoasilicasensorcrystalorasilicasensorcoatedwithabiotinylatedPEGalkanethiol. a)silicaparticlesonasilicacrystal b)silicaparticlesonabiotinylatedPEGalkanethiolcoatedsilicacrystal c)streptavidincoatedsilicaparticlesonbiotinylatedPEGalkanethiolcoatedsilicacrystal. GoodqualityofthefitisconfirmedbytherelativelylowRMSDvaluesofthefitssummarized in Table 2, with the exception of the fit for the experiments comprising streptavidin coated silica particlesonbiotinylatedPEGalkanethiolcoatedsilicacrystal.Valuesforkandʇreflecttheinfluence ofanadsorbedproteinfilmonthecrystalsurfaceversusabarecrystalsurfaceontheadhesivebond properties(higherdragcoefficient)aswellastheimpactofspecificligandͲreceptorbindingversusnonͲ specificbinding(higherspringconstantsanddragcoefficients).Importantly,althoughvaluesforthe particlemassmpobtainedvaryoverafactorofthree(seealsoTable2),theyareofthesameorderof magnitudeasexpectedfor1μmdiametersilicaparticles(14x10Ͳ16kg).
a
b
c
2
Table2.Springconstantsk,dragcoefficientsʇandmassesofabioticandbioticparticles,mp, adheringtovariouscrystalsurfaces,includingtheRMSDindicatingthequalityofthebruteforcefitto theQCMͲDoutput. k(kg/s2) Ɍ(10Ǧ9kgsǦ1) mp(10Ǧ16kg) RMSD(Hz) Abioticparticles:silicaparticles Ǧ Ͳ ͳ ͳ͵ ͳͺ Ǧ Ͳ ͳͲʹ Ͷ ͳͺ Ǧ ͲǤͲͺ ʹͻ ͺ Bioticparticles:S.salivariusHB7 ͲǤʹͳ ͳͲ ͳͻ ͲǤ͵ͷ ͳͲ Ͷʹ ͲǤʹ͵ ͷ ͷ ͺͻ Table2alsoillustratestheanticipatedresultsfortheadhesionofS.salivariusHB7adheringto differentcrystalsurfaces.Mostnoticeably,springconstantskofthestreptococcalbondsarehigher thanforabioticsilicaparticleswithlittleinfluenceofthecrystalsurfaceproperties.Thedragcoefficient ʇhowever,isordersofmagnitudesmallerthanforabioticsilicaparticles,indicatingthatlowerfriction lossesduetooscillationsasaresultofthelowerweightanddensityofthebioticparticles.Notethat inallcases,thebacterialmassmpobtainedinthefitmatchesverywellwiththemassexpectedfor bacteria(5x10Ͳ16kg). Insummary,theproposedprotocolinvolvestheuseofacoupledresonatormodeltoobtain valuesforthespringconstantanddragcoefficientofthebondbetweenadheringbioticandabiotic particlesonQCMͲDcrystalsurfacesaswellastheirmasses.Thesethreeoutputparametersarefitted totheQCMͲDoutputȴfandȴȳasafunctionofthecrystalsresonancefrequencieswithinthewindow ofobservablefrequencies.KnowledgeofthebondpropertiesinparticularadhesiontoasolidͲliquid interfaceallowsbetterunderstandingonhowtoinfluenceparticleadhesionanddetachment.
FUTUREPROSPECTS
QCMͲD data analysis for particle adhesion is neither trivial nor beyond dispute. Other viscoelastic modelsthan the KelvinͲVoigt modelimplementedinEq.1,such asthe Maxwellmodel (springanddashpotplacedinseries)mightimprovethequalityofthefitandyieldlowerRMSDvalues (D.Johannsmann,privatecommunication).Asmost“oneͲbyͲone”methodstoderiveviscoelasticbond properties indicate large standard deviations over individual particles, inclusion of a polydispersity parameterinEq.1isalsoworthwhiletoattemptinordertoimprovethequalityoffitting.Finally,as molecularadsorptionofbacterialbiͲproductssuchaspolysaccharides,proteinsandDNAishardto avoid,analysesofcombinedmolecularadsorptionandresonatorcouplingbyintroducinganoffsetin frequencyanddissipationinEq.1todifferentiatebetweenmolecularadsorptionandmassͲcoupling mightincreasetheeaseofuseofQCMͲDpreparatorystepsforbacterialadhesionstudies. ACKNOWLEDGEMENTS ThisworkwassupportedbytheUniversityMedicalCenterGroningen,TheNetherlands. COMPETINGFINANCIALINTERESTS HJBisalsodirectorofaconsultingcompany,SASABV(GNSchutterlaan4,9797PCThesinge, TheNetherlands).Theauthorsdeclarenopotentialconflictsofinterestwithrespecttoauthorship and/orpublicationofthisarticle.Opinionsandassertionscontainedhereinarethoseoftheauthors andarenotconstruedasnecessarilyrepresentingviewsoftheirrespectiveemployers Supplementarymaterial ThebruteforcefittingprogramoftheQCMͲDoutputcanbefoundinthesupplementary information.
2
REFERENCES
(1) Darouiche, R. O. DeviceͲAssociated Infections: A Macroproblem that Starts with Microadherence.Clin.Infect.Dis.2001,33,1567–1572. (2) Serra,B.;Gamella,M.;Reviejo,A.J.;Pingarrón,J.M.LectinͲModifiedPiezoelectricBiosensors forBacteriaRecognitionandQuantification.Anal.Bioanal.Chem.2008,391,1853–1860. (3) Hayward,R.C.;Saville,D.A.;Aksay,I.A.ElectrophoreticAssemblyofColloidalCrystalswith OpticallyTunableMicropatterns.Nature2000,404,56–59. (4) Pomorska,A.;Shchukin,D.;Hammond,R.;Cooper,M.A.;Grundmeier,G.;Johannsmann,D. Positive Frequency Shifts Observed upon Adsorbing MicronͲSized Solid Objects to a Quartz CrystalMicrobalancefromtheLiquidPhase.Anal.Chem.2010,82,2237–2242. (5) Johannsmann,D.ViscoelasticAnalysisofOrganicThinFilmsonQuartzResonators.Macromol. Chem.Phys.1999,516,501–516. (6) Peterson,B.W.;He,Y.;Ren,Y.;Zerdoum,A.;Libera,M.R.;Sharma,P.K.;VanWinkelhoff,A.ͲJ.; Neut,D.;Stoodley,P.;VanderMei,H.C.;BusscherH.J.ViscoelasticityofBiofilmsandtheir RecalcitrancetoMechanicalandChemicalChallenges.FEMSMicrobiol.Rev.2015,39,234–245. (7) Dabros,T.;Warszynski,P.;vandeVen,T.G.M.MotionofLatexSpheresTetheredtoaSurface. J.ColloidInterfaceSci.1994,162,254–256.
(8) Stoodley, P.; Lewandowski, Z.; Boyle, J. D.; LappinͲScott, H. M. Structural Deformation of BacterialBiofilmsCausedbyShortͲTermFluctuationsinFluidShear:AninSituInvestigationof BiofilmRheology.Biotechnol.Bioeng.1999,65,83–92. (9) Dufrêne,Y.F.TowardsNanomicrobiologyUsingAtomicForceMicroscopy.Nat.Rev.Microbiol. 2008,6,674–680. (10) Muller,D.J.AFM:ANanotoolinMembraneBiology.Biochemistry2008,47,7986–7998. (11) Wright,C.J.;Armstrong,I.TheApplicationofAtomicForceMicroscopyForceMeasurements totheCharacterisationofMicrobialSurfaces.Surf.InterfaceAnal.2006,38,1419–1428. (12) Neuman, K. C.; Nagy, A. SingleͲMolecule Force Spectroscopy: Optical Tweezers, Magnetic
TweezersandAtomicForceMicroscopy.Nat.Methods2008,5,491–505.
(13) Lu, S.; Walters, G.; Parg, R.; Dutcher, J. R. Nanomechanical Response of Bacterial Cells to CationicAntimicrobialPeptides.SoftMatter2014,10,1806–1815. (14) Chen,Y.;Norde,W.;VanderMei,H.C.;Busscher,H.J.BacterialCellSurfaceDeformationunder ExternalLoading.mBio2012,3,e00378Ͳ12. (15) Song,L.;Sjollema,J.;Sharma,P.K.;Kaper,H.J.;VanderMei,H.C.;Busscher,H.J.Nanoscopic VibrationsofBacteriawithDifferentCellͲWallPropertiesAdheringtoSurfacesunderFlowand StaticConditions.ACSNano2014,8,8457–8467. (16) Kamiti,M.;VandeVen,T.G.M.MeasurementofSpringConstantsofPolyacrylamideChains
BridgingParticlestoaSolidSurface.Macromolecules1996,29,1191–1194.
(17) Sauerbrey, G. Verwendung von Schwingquarzen Zur Wägung Dünner Schichten Und Zur Mikrowägung.ZeitschriftfürPhysik.1959,206–222. (18) D’Amour,J.N.;Stålgren,J.J.R.;Kanazawa,K.K.;Frank,C.W.;Rodahl,M.;Johannsmann,D. CapillaryAgingoftheContactsbetweenGlassSpheresandaQuartzResonatorSurface.Phys. Rev.Lett.2006,96,058301. (19) Dybwad,G.L.ASensitiveNewMethodfortheDeterminationofAdhesiveBondingbetweena ParticleandaSubstrate.J.Appl.Phys.1985,58,2789–2790.
(20) Olsson, A. L. J.; Van der Mei, H. C.; Johannsmann, D.; Busscher, H. J.; Sharma, P. K. Probing ColloidͲSubstratumContactStiffnessbyAcousticSensinginaLiquidPhase.Anal.Chem.2012, 84,4504–4512.
(21) Johannsmann,D.Viscoelastic,Mechanical,andDielectricMeasurementsonComplexSamples withtheQuartzCrystalMicrobalance.Phys.Chem.Chem.Phys.2008,10,4516–4534. (22) VadilloͲRodriguez,V.;Schooling,S.R.;Dutcher,J.R.InSituCharacterizationofDifferencesin
the Viscoelastic Response of Individual GramͲNegative and GramͲPositive Bacterial Cells. J. Bacteriol.2009,191,5518–5525.
(23) Peschel, A.; Langhoff, A.; Johansmann, D. Coupled Resonances Allow to Study the Aging of AdhesiveContactsbetweenaQCMSurfaceandSingle,MicrometerͲSizedParticles.Nanotech. 2015,26,1Ͳ9.
(24) Eliceiri,K.W.;Berthold, M. R.;Goldberg,I. G.;Ibáñez, L.; Manjunath,B.S.; Martone,M.E.; Murphy,R.F.;Peng,H.;Plant,A.L.;Roysam,B.;etal.BiologicalImagingSoftwareTools.Nat. Methods2012,9,697–710.
(25) Meinders,J.M.;VanderMei,H.C.;Busscher,H.J.DepositionEfficiencyandReversibilityof BacterialAdhesionunderFlow.J.ColloidInterfaceSci.1995,176,329–341.
(26) Reviakine, I.; Johannsmann, D.; Richter, R. P. Hearing What You Cannot See and Visualizing WhatYouHear.Anal.Chem.2011,83,8838–8848.
(27) Olsson,A.L.J.;Arun,N.;Kanger,J.S.;Busscher,H.J.;Ivanov,I.E.;Camesano,T.A.;Chen,Y.; Johannsmann, D.; Van der Mei, H. C.; Sharma, P. K. The Influence of Ionic Strength on the Adhesive Bond Stiffness of Oral Streptococci Possessing Different Surface Appendages as ProbedUsingAFMandQCMͲD.SoftMatt.2012,8,9870–9876.