Aggregation liposomes of PEGylated driven by hydrophobic forces Colloids and Surfaces B: Biointerfaces

(1)

ContentslistsavailableatScienceDirect

Colloids and Surfaces B: Biointerfaces

jou rn a l h om ep ag e :w w w . e l s e v i e r . c o m / l o c a t e / c o l s u r f b

Aggregation of PEGylated liposomes driven by hydrophobic forces

Tamás Bozó

^a,∗,1

, Tamás Mészáros

^b,c,1

, Judith Mihály

^d

, Attila Bóta

^d

, Miklós S.Z. Kellermayer

^a,e

, János Szebeni

^b,c

, Benedek Kálmán

^b

aDepartmentofBiophysicsandRadiationBiology,SemmelweisUniversity,1094Budapest,T ˝uzoltóStr.37-47,Hungary

bNanomedicineResearchandEducationCenter,SemmelweisUniversity,1089Budapest,NagyváradSquare4,Hungary

cSeroScienceLtd.,1089Budapest,NagyváradSquare4,Hungary

dBiologicalNanochemistryResearchGroup,InstituteofMaterialsandEnvironmentalChemistry,ResearchCentreforNaturalSciences,HungarianAcademy ofSciences,1117Budapest,Magyartudósokkörútja2,Hungary

eMTA-SEMolecularBiophysicsResearchGroup,SemmelweisUniversity,1094Budapest,T ˝uzoltóStr.37-47,Hungary

a r t i c l e i n f o

Articlehistory:

Received4November2015 Receivedinrevisedform23June2016 Accepted27June2016

Availableonline28June2016

DedicatedtothememoryofBerci(Benedek Kálmán),agreatcolleagueandfriend.

Keywords:

Liposome PEG Aggregation Fusion

Hydrophobiceffect Ammoniumsulfate Kosmotropicsalt

a b s t r a c t

Polyethyleneglycol(PEG)iswidelyusedtostericallystabilizeliposomesandimprovethepharmacoki- neticproﬁleofdrugs,peptidesandnanoparticles.Herewereportthatammoniumsulfate(AS)canevoke theaggregationofPEGylatedvesiclesinaconcentration-dependentmanner.Liposomeswith5mol%

PEGwerecolloidally stableatASconcentrationsupto0.7mM,abovewhichtheyprecipitatedand formedmicron-sizeaggregateswithirregularshape.Whileaggregationwasreversibleupto0.9Mof AS,above1Mfusionoccurred,whichirreversiblydistortedthesizedistribution.Zetapotentialoflipo- somesmarkedlyincreasedfrom−71±2.5mVto2±0.5mVuponraisingtheASconcentrationfrom0 to0.1M,butnoconsiderableincreasewasseenduringfurtherASaddition,showingthattheaggrega- tionisindependentofsurfacecharge.TherewasnoaggregationintheabsenceofthePEGchains,and increasingPEGmolar%shiftedtheaggregationthresholdtolowerASconcentrations.Changesinthe FTIRspectralfeaturesofPEGylatedvesiclessuggestthatASdehydratesPEGchains.Otherkosmotropic saltsalsoledtoaggregation,whilechaotropicsaltsdidnot,whichindicatesageneralkosmotropicphe- nomenon.Thedrivingforcebehindaggregationislikelytobethehydrophobiceffectduetosaltingout thepolymersimilarlytowhathappensduringproteinpuriﬁcationorHydrophobicInteractionChro- matography.Sinceliposomeaggregationandfusionmayresultindifﬁcultiesduringformulationand adversereactionuponapplication,thephenomenadetailedinthispapermayhavebothtechnological andtherapeuticalconsequences.

1. Introduction

Polyethyleneglycol(PEG),alinearpolymerof[–CH2–CH2–O]

units,iscommonlyusedindrugdeliverytomodifypharmacoki- neticproperties ofactiveagents. AttachingPEGchainstosmall molecules, peptides, proteins, oligonucleotides or nanoparticles may effectively reduce their enzymatic degradation and pro- longtheircirculationtimeinblood[1].Animportantexampleis liposomePEGylation,whichprovides“stealth”characteristicsto liposomeshelpingthemtoevadephagocytosisbymacrophages,

∗Correspondingauthor.

E-mailaddresses:bozo.tamas@med.semmelweis-univ.hu(T.Bozó), tmeszaros@seroscience.com(T.Mészáros),mihaly.judith@ttk.mta.hu(J.Mihály), bota.attila@ttk.mta.hu(A.Bóta),miklos.kellermayer@med.semmelweis-univ.hu (M.S.Z.Kellermayer),jszebeni2@gmail.com(J.Szebeni).

1 Theseauthorscontributedequallytothiswork.

which extends their lifetime in the body and results in dose- independentpharmacokinetics(exceptforverylowdoses)[2,3].

Liposometherapymadeitpossibletoincreasethebioavailability ofdrugsthatarepoorlyabsorbed(likeamphotericin-B),toreduce sideeffectsofhighlycytotoxicanti-canceragents(suchasdoxoru- bicin)andtoopennewroutestogeneratesite-selectiveeffect(e.g.:

photodynamictherapyinmaculardegeneration)[4].However,new beneﬁtsmaybeaccompaniedbynewrisks:e.g.,itturnedoutthat liposomes can inducecomplement activation related pseudoal- lergy(CARPA),anewtypeofdrug-inducedacuteimmunetoxicity [5].Itissuspectedthatoneofitscausescouldbethepresenceof liposomalaggregatesintheformulatedproduct[6].

Here we reportthat ammoniumsulfate (AS) andother kos- motropicagentsmayelicittheaggregationandeventhefusionof PEGylatedliposomes.Theaggregatesareformedbyhydrophobic interactionsduetothesolvophobiceffectofincreasingsaltconcen- tration[7].Thisphenomenonissimilartothesalting-outmethod http://dx.doi.org/10.1016/j.colsurfb.2016.06.056

(2)

Table1

Liposomecompositions.

PEGmolar% molarratio(Cholesterol:mPEG:HSPC)

0 38.4:0:61.6

2 38.7:2:59.3

5 38.4:5:56.6

10 38.4:10:51.6

regularlyusedinproteinfractionationand puriﬁcation[8,9].An understandingofthemechanismofaggregateformationcouldpro- videinvaluableinformationforsuccessfuldrugformulationswhere liposomeaggregationcouldbeeitherpreventedorcontrolled.

2. Materialsandmethods 2.1. Materials

Cholesterol, mono PEGylated 1, 2-distearoyl-glycero-3- phophoethanolamine (mPEG-2000-DSPE), and hydrogenated soybeanphosphatidylcholine(HSPC)wereobtainedfromLipoid GmbH(Ludwigshafen,Germany).Ethanol,isopropanol,histidine, sucrose, ammonium sulfate (AS), sodium sulfate, magnesium sulfate,sodiumcitrate, magnesiumchloride,guanidine chloride (GdmCl) were purchased from Sigma Aldrich Kft. (Budapest, Hungary).Salsolinfusion(TEVAHungaryZrt.,Debrecen,Hungary) wasobtainedfromtheUniversityPharmacy,andpuriﬁed water wasproducedbyaMilli-QIntegral3WaterProductionUnit(Merck Millipore,Billerica,MA,USA).

2.2. Liposomepreparation

Liposomesuspensionwithalipidandbuffercompositionsim- ilartotheFDA-approvedandmarketedDoxil^®waspreparedwith theextrusionmethod[10].Thelipidcompositionwascholesterol, mPEGandHSPC(seemolarratiosinTable1).Thelipidsweresolubi- lizedinethanol-isopropanolmixture(50:50),thenthesolutionwas addeddropwiseto0.25MAScontaining0.9%saline(SALSOL)solution.Large,heterogeneouslipidparticleswereextrudedfourtimes through80nm WhatmanNuclepore (Track-EtchedMembranes) membraneﬁlters(Whatman,Maidstone,UK)bymeansofaLipex^TM Extruder(NorthernLipidsInc.,Burnaby,B.C.Canada)at50barand 70^◦Ctoachieveuniformparticlesizedistribution.Theliposomes werethendialyzedagainst10mMhistidinebuffer(pH=7.5)con- taining10w/w%sucrosetoremoveASandorganicsolvents.The totalphospholipidconcentrationwasapproximately15.9mg/ml (cca.21.4mM).Theliposomalstocksolutionswerestoredat4^◦C protectedfromlightandusedwithin2weeks.Thestocksolution wasfurtherdilutedasdictatedbytheexperiments.Thedegreeof dilutionandcorrespondinglipidconcentrationsaregiveninthe textandﬁgurecaptions.

2.3. MixingPEGylatedliposomeswithdifferentsalts

For turbidimetry, light scattering and zeta potential experi- ments20␮lofPEGylatedliposomesweremixedwith980␮lsalt solutionof appropriate concentration(50x dilution). For phase contrastandatomicforcemicroscopyexperimentsPEGylatedlipo- somes were diluted either 200x or 500x with salt solution of appropriate concentration. For 0M concentration physiological salinesolution(Salsol)wasusedfordilution.Theactualsaltand lipidconcentrationsaregiveninthetextandﬁgurecaptions.

2.4. Dilutionofprecipitatesforfusionand aggregation-reversibilitystudies

500␮lofPEGylatedliposomesweremixedwith500␮lofAS solutiontoproduceastockofprecipitatedsamplesofthedesired ASconcentration(from 0.8Mto2.0M).After15minincubation time,20␮lofthesestockswasmixedto980␮lASsolutionsof appropriateconcentrations(down to0.1M).Theactualsalt and lipidconcentrationsaregiveninthetextandﬁgurecaptions.

2.5. Turbidimetry

TheaggregationofPEGylatedliposomeswasfollowedbymea- suring the apparent optical density of the solution. Brieﬂy, a 4␮lsampleofthewell-vortexedsolutionwaspipettedontothe pedestalofaNanoDrop2000UV–visspectrophotometer(Thermo Scientiﬁc Ltd., Wilmington, DE), and the optical density was recordedat250nm.Becausethelipidconcentrationwaskeptcon- stant,anincreaseinopticaldensitycorrespondstoanincreasein lightscattercausedbytheappearanceoflargerparticlesdueto aggregation.Forcomparability,identicallipidconcentrationswere usedinthedifferentsamples.

2.6. Dynamiclightscatteringmeasurement

Thesizedistributionofliposomesandaggregates werechar- acterizedbydynamic lightscattering(DLS)onaZetasizerNano Sinstrument (MalvernInstrumentsLtd,Malvern,UK). Fromthe intensityﬂuctuationsofa633-nmlaserlightscatteredathighangle fromthefreelymovingsuspendedparticlestheirdiffusionconstant wasobtained.SizedistributionwascalculatedbyusingtheStokes- Einsteinequationbythebuilt-in algorithmsoftheinstrument’s software.Light scatteringwas measuredat25±1^◦C. Z-average valuesaredisplayedthroughoutthearticle,whichrepresentthe primaryandmoststableparameterproducedbyDLStechnique[11]

andrecommendedforqualitycontrolreports(ISO22412:2008).Z- averagevaluesrepresentagoodapproximationofhydrodynamic diameterofwelldispersedparticleswithmonomodalsizedistri- bution(indexofpolydispersitytypicallylowerthan0.1)andthus arewellapplicableforPEGylatedvesicles.TheZ-average,however, doesnotreﬂecttherealsizeofprecipitatedsamplesthatareoften heterogeneousinsizeandmaybeirregularlyshaped.Inthelatter caseZ-averagewasusedonlyforroughestimationofparticlesize, whichenabledustofollowliposomeaggregationwithoutexact determinationofaggregatedimensions.Sincedifferentbatchesof PEGylatedliposomeswereusedinthedifferentexperiments,minor variationsareseenintheaveragesizeofcontrolvesicles.

2.7. Zetapotentialmeasurements

PEGylatedliposomesweredilutedwithASsolution,and750␮l ofthismixturewasinjectedcarefullyintofoldedcapillarycells(PCT Kft.,Mosonmagyaróvár,Hungary)toavoidbubbleformation.Zeta potentialmeasurementswereperformedbyusingaZetasizerNano ZSequipment(MalvernInstrumentsLtd.,Worcestershire,UK)in whichparticlevelocityismeasuredaccordingtoalightscattering techniquebasedonDopplereffectevokedbyapairofmutually coherentlaserbeams (4mW,He-Nelaser at633nm).Fromthe autocorrelationfunctionofthescatteredlightintensitytheelec- trophoreticmobilityand,viatheHenryequation,thezetapotential are calculated.Measurements werecarried out in triplicates at 25^◦C.

(3)

2.8. Analysisofzetapotentialdata

Bindingofionstoliposomalsurfaceandtheconcomitantchange ofsurfacepotentialcanbedescribedwellbyLangmuir-Freundlich isotherm[12].

=0+max· (K·c)ⁿ 1+(K·c)ⁿ

where␨isthemeasuredzetapotential,␨0isthezetapotentialat zeroASconcentration,␨maxisthemaximalchangeofzetapoten- tial,Kisthebindingconstant,cistheligandconcentrationandn istheindexofheterogeneitydescribingthecooperativityofion binding.

2.9. Atomicforcemicroscopyandimageanalysis

Atomic forcemicroscope (AFM) images were recorded with a Cypher instrument (Asylum Research, Santa Barbara, CA) by scanningthesamplesinﬂuid witha gold-coatedsilicon nitride cantilever (Olympus Biolever, A lever, typical spring constant:

30pN/nm).100␮lsamplewasappliedona cleanedborosilicate glasscoverslipandincubatedinavaporchamberat23±1^◦C.Non- contact-modeimageswererecordedatalinescanrateof0.5–1Hz.

Allmeasurementswerecarriedoutat28±1^◦C.Imageswereana- lyzedbyusingthebuilt-inalgorithmsoftheAFMdrivingsoftware (IgorPro,WaveMetrics,Inc.,LakeOswego,OR).

2.10. Phasecontrastmicroscopy

MicrographswererecordedwithaNikonEclipseTi-Uinverted microscope(Auro-ScienceKft.,Budapest,Hungary)equippedwith auEyeUI1220LEdigitalcamera(IDSImagingDevelopmentSys- temsGmbH,Obersulm,Germany)usinga40xNikonSPlanﬂuor phasecontrastobjective.

2.11. Infraredspectroscopy

ATR-FTIRspectrawerecollectedwithaVarian2000FTIRScimi- tarSeries(VarianInc.,PaoloAlto,CA)spectrometerequippedwith a‘GoldenGate’(SpecacLtd.,London,UK)singlereﬂectiondiamond ATRaccessory.Themeasurementswereperformedatroomtem- perature:3␮lsamplewasmountedonthetopofthediamondATR crystalandacapwasusedtoavoidsampledrying;128scanswere collectedataresolutionof2cm⁻¹.ATRcorrectionwasexecuted aftereachdatacollection.Allspectralmanipulations,includingsub- tractionsand spectraldeconvolutionswere performedbyusing theGRAMS/32softwarepackage(GalacticIndustriesIncorporation, USA).Bandpositionsforcurveﬁttingweredeterminedusingthe secondderivative.BandshapeswereapproximatedbyLorentzian functions.Theintensitiesandthebandwidthofeachcomponent wereallowedtovaryuntiltheminimal␹²parameterwasreached.

Aftertheﬁttingprocedure,therelativecontributionofa partic- ularcomponentwascalculatedfromtheintegratedareasofthe individualcomponents.

3. Resultsanddiscussion

3.1. PEGylatedliposomescanbeprecipitatedbyammonium sulfate

TheadditionofAStoPEGylatedliposomesinfewmolarcon- centrationinitiated therapidincreaseof opacityofthesample.

Toassessthemagnitudeofliposomeprecipitation,wemeasured theturbidityofliposomalsuspensionsat0–2MASconcentration.

Accordingtothe turbidityvs. ammonium-sulfateconcentration

Fig.1.Turbidity(blackline)andaverageparticlediameter(redline)ofliposomal suspensionsasafunctionofammoniumsulfate(AS)concentration.Average±SD valuesofthreeindependentmeasurementsareshown.Lipidconcentrationwaskept constantthroughoutthemeasurement(50xdilution,0.318mg/ml).(Forinterpre- tationofthereferencestocolourinthisﬁgurelegend,thereaderisreferredtothe webversionofthisarticle.)

curve(Fig.1),noprecipitationoccursbelowanASconcentrationof 0.75M.Above0.75M,turbidityrisesabruptlythenlevelsoffabove 1Mtoavalueanorderofmagnitudegreaterthanintheabsenceof AS.Fromthiscurveweconcludedthatprecipitationbegansome- wherebetween0.75and0.8MASconcentration.Thefallofthe turbidimetrycurveabove1MASconcentrationmaybeexplained byincreasingheterogenityofthesystem,i.e.theformationofa lipid-richprecipitateandalipid-pooraqeousphase.

Wehypothesizedthattheabruptriseinturbiditywasrelatedto asizeincreaseduetotheaggregationofthevesicles.Toassessthe sizeoftheprecipitatesasafunctionofASconcentration,dynamic light scattering measurements werecarried out (Fig.1).Below a concentrationof0.8Mthemeanparticlesize variedbetween 83.2nmand91.5nmindependentlyoftheASconcentration.Upon increasingtheASconcentrationfurther,however,sizeincreased abruptlyto∼2000nmintherangeof0.8–1Mthenmoreslowly above1M.UponreachinganASconcentrationof2Mthemeanpar- ticlesizeexceeded5000nm.Notably,particlesizeisapproximated withtheZ-averagevaluewhichiscalculatedbyassumingspherical geometryandmonomodalsizedistribution.Incaseofaggregation, particleshapeislikelytodeviatefromspherical,which,together withgrowingpolydispersity indices measuredabove0.75MAS (datanotshown)meansthattheZ-averageparametermaycor- respondonlytoanapproximation,withinanorderofmagnitude, oftheaverageparticlediameter.Theresultsoftheturbidimetryand dynamiclightscatteringmeasurementsleadtosimilarconclusion:

precipitationbeginsuponreachinganASconcentrationthreshold (0.7–0.8M),thenprogressivelylargeraggregatesareformedupon incerasingtheASconcentrationfurther.Sinceprecipitationtakes placeinstantenouslyuponmixingthePEGylatedliposomeswith AS,itisratherdifﬁculttofollowaggregationkinetics.Sizeincreases rapidlyandconsiderablyintheﬁrstminuteneededtosetupaDLS measurement.Afterthislagtimeafurthercontinousincreaseof sizewasobservable(Fig.S1),butexactrateandkineticscouldnot beendetermined.

Torevealthemicroscopicdetailsoftheprecipitationprocessand assesswhethervesicleaggregationtakesplaceindeed,morpholog- icalmeasurementswerecarriedout.Phasecontrastmicroscopyof precipitatedsamplesshowedbranchingobjectsofirregularshapes apparently formedofsmallerclusters(Fig.2)which resembled electronmicrographsofliposomalaggregatesseenearlier[13].

Toresolvetheultrastructureofaggregatesandtofollowtheir formation we imaged liposomes withatomic forcemicroscopy (AFM) at various ASconcentrations (Fig.3). At0 and 0.7M AS

(4)

Fig.2.Phasecontrastmicrographofliposomalaggregatesin1MAS.Largebranching objectsandsmalleraggregatesareseen(200xdilution,0.0795mg/mllipidconcentration).

individual, interaction-free liposomes were observed. At an AS concentrationof0.8Mmany liposomeswereobservedinlinear assemblies,pointingattheonsetofaggregation.AtanASconcen- trationof0.9M,largeaggregates wereclearlyseen.In0.8MAS (Fig.3C),besidesthevesiclesﬂatpatcheswithasmoothsurfaceand atopographicalheightof5–7nmwereobserved.Weidentifythem asbilayers,althoughtheyaresomewhatthinnerthanalipidbilayer

coveredwithaPEGpolymerbrushonbothsides(approx.12nm) calculated withadifferent method[14,15]. Thebilayerpatches probablyemergebecauseasosmolalityincreases,vesiclesexhibit agreaterpropensitytoburstonthesubstrate[16,17].Patchforma- tionisageneralphenomenoncharacteristictoliposomalsamples.

Patchesofvarying sizesarefoundin almostallliposomal AFM imagesthroughoutthecorrespondingliterature[18–20]andalsoin Fig.4.Interestingly,patchformationappearstodependontheAS concentrationasevidencedbyourresultsshowninFig.3.While patchformation is only sporadicatlower ASconcentrations, it becomespronouncedabove0.8M(Fig.3C),andat0.9Mmostofthe substrateiscoveredwithaconfluentsupportedlipidbilayer(see thebackgroundofclustersandvesiclesinFig.3D).Liposomeclus- tersobservedin0.9MAS(Fig.3D–F)havediversesizesvaryingfrom fewhundrednmtofew␮m,whichisinanorder-of-magnitudecor- relationwiththeDLSdata(seeFig.1).Notethattheirregularvesicle shapemightbetheresultofimagingartifactsandnotexquisitely ofliposomalshapetransformations.UponraisingtheASconcen- trationabove0.8Mimagingbecamedifficult,whichismostlikely duetothepresenceoflarge,softaggregatesincompletelyimmo- bilizedonthesurface.Severalattemptshavebeenmadetoimage samplesatevenhigherASconcentrationstofindlargeraggregates, butunsuccessfully.

3.2. Aggregationmaypromotevesiclefusion

Totestwhetherliposomeaggregationisfollowedbyfusionand todeterminethethresholdconcentrationoffusion,liposomalsam- pleswereprecipitatedinvariousconcentrationsofAS,then15min laterdilutedto0.1MAS.DLSdatashowednoconsiderablechange ofaveragesizeat0.8and0.9Mprecipitatingconcentration.Aslight increasewasobservedat1.0and1.1Mfollowedbyamoresignif-

Fig.3.Amplitude-contrastAFMimagesofliposomesat(A)0M,(B)0.7M,(C)0.8M,(D)0.9MASconcentration.Sampleswerediluted1to500withtheircorresponding solution(0.0318mg/mllipidconcentration).Figure(E)showstheheight-contrastAFMimageoffigure(D)color-coded,while(F)representstheheightsectionprofiletaken alongsidethethickredlinein(E).Thewhiteandredtrianglesin(E)and(F),respectively,pointatthehighestpointoftheliposomecluster.Thecolorbarinthelowerright cornerdisplaystheheightscaleoffigure(E).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

(5)

Fig.4.(A)Sizeofliposomes(Z-average±SD)dilutedfromvariousprecipitatingASconcentrations(indicatedonxaxis)to0.1M.Reddottedlineshowsthesizeofcontrol liposomes.(B)Amplitude-contrastAFMimageoftheliposomalsuspensioninphysiologicalsalinesolution.(C)and(D)Amplitude-contrastAFMimagesofa1Mand2M AS-aggregatedsampledilutedbackto0.1MASconcentration.WhitearrowheadindicatesalargervesicleinpanelC.InpanelDfewofthelargervesiclesrupturedduring thescantoformbilayerpatchesonthesurface.Lipidconcentrationwasidenticalinthesamples(200xdilution,0.0795mg/mllipidconcentration).(Forinterpretationofthe referencestocolourinthisﬁgurelegend,thereaderisreferredtothewebversionofthisarticle.)

icantandmonotonousincreasefrom1.2M(Fig.4A).AFMimages ofthesesamplesrevealedthattheincreaseofaveragesizemaybe attributedtotheoccurrenceoffusedliposomes.Boththesizeand relativeamountoffusedvesiclesincreasedwithASconcentration:

at1Mliposomeslargerthan theaverageappeared sporadically (Fig.4C),whileat2Mevenlargervesicleswithanirregularshape becamedominant(Fig.4D).Theﬁndingsareinaccordwithsize distributionshowninFig.S2,theaveragesofwhichareshownin Fig.4A.Upto1.1Mamoderate,thenfrom1.2Mamoresigniﬁcant upwardshiftisseenduetotheappearanceoflargerobjects.Inaddi- tion,from1.5Maminorpopulationof several-micron-diameter particlesappearedwhichwasnotobservedinAFMimagespossibly duetotheirsmallnumber.

Takentogether,PEGylatedliposomesmaynotonlyaggregate butalsofuseuponASaddition.Therateoffusionisconcentration dependent,andthethresholdconcentrationis1Mattheemployed incubation time.This alsomeansthat AS-drivenaggregationof thevesiclescanbereverseddependingonASconcentrationand incubationtime.Todisruptaggregatedliposomestheyshouldbe dilutedwellbelowtheASconcentrationatwhichtheystartedto precipitate(Fig.S3).

3.3. Precipitationdoesnotdependonsurface-chargeproperties

Physicalstabilityofcolloidalvesiclesdependslargelyontheir surfacechargedensity,which isapproximatedwiththeelectric potentialdifference(orzetapotential)betweenthebulksolution andtheslippingplaneofionsassociatedtothevesicularsurface.

Itiswidelyacceptedthatazetapotentiallowerthan30mVmakes colloidaldispersionspronetoaggregation[21].Sincethesurface chargepropertiesandthusphysicalstabilityofliposomalvesicles canbelargelyaffectedbyions[22],modiﬁcationoftheliposomes’

zetapotentialbyASadditionmightbeakeyfactorinevokingaggre- gation.Toassessthecharge-modulatingeffectofASonPEGylated vesicles,zetapotentialmeasurementswerecarriedout.

Zetapotentialofthecontrolsamplewas−71.7±2.5mV,which impliesastrongnegativesurfacepotentialandcorrespondstoan extremelyhighcolloidalstability[21].AddingASledtoamassive increaseofthezetapotentialvalueatevenrelativelylowconcen- trations(Fig.5),whichmaybeexplainedbytheassociationofNH₄⁺ ionstotheoriginallynegativevesicularsurface.Coordinationof NH4+ cationstotheetheroxygenswasproposedbyBaileyand Callard[23].Analternativeexplanationisthatstructuralmodiﬁca- tionsofthecharge-alteringPEGchains(discussedlater)mayhave ledtotheobservedchargeincrease.

TheLangmuir-Freundlichisothermﬁtswelltothedatapoints suggestingmonolayerabsorptiontoaheterogeneoussurface.The adsorptionconstantof1001M⁻¹pointsatoutstandingafﬁnityof NH₄⁺ionstothePEG-coveredsurfaceand0.73asindexofhetero- genityindicatesnegativecooperativityoftheions.

Thepronouncedsurface-chargealteringeffectofASlevelsoff ataslowconcentrationasabout0.1M,whichisnearlyanorder ofmagnitudelowerthanthethresholdconcentrationforaggrega- tion.Furthermore,noconsiderablechangeofzetapotentialisseen reachingtheconcentrationregimeofaggregation(above0.75M,

(6)

Fig.5.ZetapotentialofPEGylatedliposomesasafunctionofASconcentration.Red lineshowstheLangmuir-Freundlichisotermﬁttedtothedata.Zetapotentialwas measuredintriplicates,errorbarscorrespondtostandarddeviation.Zetapotential at0ASconcentrationwas−71.7±2.5mV.Insetshowsthedataplottedonaloga- rithmicxaxis.Dilution:50x(0.318mg/mllipidconcentration).(Forinterpretation ofthereferencestocolourinthisﬁgurelegend,thereaderisreferredtotheweb versionofthisarticle.)

Fig.6. EffectofPEGconcentrationontheASconcentrationdependenceofliposomal aggregation.Dilution:50x(0.318mg/mllipidconcentration).Brokenline:0%mPEG;

redline:2%mPEG;blueline:5%mPEG;blackline:10%mPEG.(Forinterpretation ofthereferencestocolourinthisﬁgurelegend,thereaderisreferredtotheweb versionofthisarticle.)

seeFig.1).ThesetogethersuggestthattheASmediatedincreaseof zetapotentialisnotthemechanismbehindliposomeaggregation.

3.4. PrecipitationisPEG-related

ToelucidatetheroleofPEGchainsinliposomeaggregation,lipo- someswithdifferentamountsofPEGchainsontheirsurfacewere producedandmixedwithAS.DLSdatarevealedthatconventional liposomes(i.e.,oneswithoutPEGylation)showednosizeincrease upto2MAS(Fig.6).Thisindicatesthatnoprecipitationtakesplace intheabsenceofthePEGbrushontheliposomalsurfaceandhigh- lightsthattheaggregationevokingeffectofASismediatedviathe PEGpolymerlayer.UponincreasingthePEGcoveragefrom2to 10M%,theprecipitationcurvesshiftedtotheleft,meaningthat lowerconcentrationsofASwereenoughtoelicitaggregation.It againunderpinsthatprecipitationisPEG-related.Consideringthat modiﬁcationofsurfacechargedoesnotaffecttheaggregationof liposomes(seeabove),wehypothesizethatsomestructuraltran- sitionsofthesurfacepolymerchainsmusthaveledtothehigher propensityforaggregation.

3.5. ASdehydratesPEGchains

ConsideringthatprecipitationisrelatedtothepresenceofPEG, itisplausiblethatAS,whichisakosmotropicagent[24,25]dehy- dratedPEGpolymers,leadingtoaggregationofthePEG-covered vesiclesthroughhydrophobicinteractions.Toestimatethehydra- tion level of the PEG layer, attenuated total reflection Fourier transforminfraredspectroscopy(ATR-FTIR)combinedwithcurve fittingprocedurewasapplied.ThemethodwasproposedbyVarga etal.tocharacterizethePEG-layerofstealthliposomesbasedon theratiooftransandgaucheconformationsofC O CgroupsofPEG chain[26].ThecomplexstretchingvibrationC O Cbandaround 1100cm⁻¹ ofthePEGchain canbedecomposed intofive band componentsduetonon-interactingvibrations.Thebandaround 1093cm⁻¹ is related to C O C groups of PEGin trans confor- mation,whiletheonearound1113cm⁻¹ belongs tothegauche conformationrelativetoC Cbond[27].Thetwoextremecom- ponentswithsmallerintensitiesaround1139and1029cm⁻¹can beassignedto␦( CH2 )deformationand␯(C C)stretchingvibra- tions,respectively.Therelativelyintensebandcomponentaround 1068cm⁻¹mightberelatedtothe␯(C OH)bands.Ahigherextent ofintramolecularH-bondingoftheC O CgroupsofthePEGmoi- etytoaneighboringethericoxygenresultsinagreaterproportion ofthe moreconstrained gaucheconformer withan appropriate increaseofrelativeintensity.Thus,theratioofthetransandgauche conformationscanbeamarkertocharacterizethePEGlayerstruc- tureand,indirectly,thehydrationstate[26].

ATR-FTIRspectraofthePEG-liposomeswith1–2wt.%concen- trationaredominatedbythestrongwaterabsorptionbands;so,asa ﬁrststepofspectralevaluationthesubtractionofwaterbackground (ASsolution)wasperformed.Sincethe␯(C O C)vibrationbands overlapwiththephosphatestretchingvibrations(␯PO2−)ofthe

Fig.7. Deconvolutionofthebandaround1100cm-1afterspectralsubtractions:A)PEGylatedliposomesinwater,B)PEGylatedliposomesin1MASsolution,C)PEGylated liposomesin2MASsolution.Emptycirclesdenotethemeasureddatapoints,solidlinesrepresenttheﬁttedspectra,theindividualbandcomponentsandtheresiduals.

(7)

Fig.8.Thetrans/gaucheratioofthe␯(C O C)bandofPEGchainsforSSLsamples inwater,in1Mammoniumsulfateandin1Mguanidinechloride.Averages±SDs ofthreeindependentmeasurementsareshown.Thehighstandarddeviationfor guanidinechloridesamplesmightbecausedbythedifﬁcultyinguanidinechloride backgroundsubtraction.

lipidcomponents,thereferencespectrumofpurehydratedHSPC (hydrogenatedsoyphosphatidylcholine,mainlipidcomponentfor PEGylatedliposomes)wascarefullysubtractedfromthespectraof PEGylatedliposomes.Typicaldeconvolutionsofthebandaround 1100cm⁻¹afterspectralsubtractionarepresentedinFig.7.

Thehighertherelativeintensityoftransconformers,thehigher thehydration levelofthePEGlayer.By addingammoniumsul- fatesalt (1Mconcentration)therelative intensityof ␯(C O C) ingaucheconformationincreases(Fig.8),indicatingthatthekos- motropicsaltreducesthehydrationofthePEGpolymerchains.As tothehigheramountofAS(2Mconcentration),however,anew bandcomponentat1068cm⁻¹dominatesthespectrum.Thisband componentmightbeassignedto␯(C OH)groups.Interestingly,no bandcomponentbelongingto␯(C O C)transcomponentcould bedeconvoluted.Thismayindicateaconformationalchangemore pronouncedthanthetrans−gauchevariation.Similarphenomenon wasobservedalsoformicellesformedbypureDSPE-PEG2000lipid (∼10wt.%)inwater(unpublishedresults).

FTIRdatasupportthenotionthatthemechanismbehindaggre- gationofstealthvesiclesmightbethereductionofhydrationof PEGpolymerchainsduetokosmotropiceffectofAS.Thisisfurther underpinned by the observation that a chaotropic salt, guani- dinechloride(GdmCl)didnotaffectsigniﬁcantlythe␯(C O C) trans/gauche ratio and thus thehydration level of the polymer (Fig.8.,seespectrum,anddeconvolutionofthebandsinFig.S4.

andTableSI.).

3.6. Otherkosmotropicsaltsalsoleadtoprecipitationof PEGylatedliposomes

Westudiedtheeffectofvarioussaltsonstealthliposomesto testthehypothesisthatPEG-relatedprecipitationofliposomesis not speciﬁcfor ASbut ageneralkosmotropic phenomenon. All examinedkosmotropicsalts(sodiumcitrate,sodiumsulfate,mag- nesiumsulfate)ledtotheprecipitationof thevesicles (Fig.S5).

Bycontrast,chaotropicsalts(magnesiumchlorideandguanidine chloride)didnotaggregatetheliposomes(Fig.S6).Theprecipi- tatingeffectofkosmotropicsaltswasfoundtobeconcentration dependent, and theorder of theirthreshold concentrationwas thefollowing: sodiumcitrate<sodiumsulfate<magnesiumsul- fate<ammoniumsulfate(Fig.9).Thisorderisingoodaccordance withtwophaseformingcapacityofionsinPEG-salt-watersystems observedearlier[28].

Thesurfacechargemodifyingeffectofkosmotropicsaltslev- elled off at much lower concentrations than that needed for aggregation(Fig.S7),similarlytowhatwasobservedforAS(Fig.5).

Theadsorptionconstantsoftheionsdonotcorrelatewiththeir precipitatingability.Furthermorenotonlykosmotropic,butalso chaotropicsalts,whichdonotaggregatePEG-liposomes,shiftedthe

Fig.9. Turbidityofliposomalsuspensionsasafunctionofsaltconcentration.Aver- age±SDvaluesofthreeindependentmeasurementsareshown.PEGconcentration was5mol%andlipidconcentrationwaskeptconstantthroughoutthemeasurement (50xdilution,0.318mg/ml).

zetapotentialofliposomesfromstronglynegativevaluestoneutral regime(Fig.S7).Thesetogetherclearlysuggestthataggregation phenomenon isnot connected toionadsorption drivensurface chargealterationofthevesicles,buttokosmotropiceffect.

4. Conclusions

Here we demonstrated that ammonium sulfate and other kosmotropicsaltshavea precipitatingeffectonPEGylatedlipo- somes.Aggregationtakesplaceaboveathresholdconcentration (Figs.1,3,6and9)andleadstotheformationofirregular,micron- sizedaggregates(Figs.2and3D–F).AtcertainASconcentrations theprocessisreversible;aggregatescanbefullydisintegratedby dilution,buthigherASconcentrationsmayevoketheirreversible fusionofvesicles(Fig.4).Reductionofsurfacechargedoesnothave anyeffectonaggregationofvesicles(Fig.5).At0M%PEGcontentno aggregationoccurswhilethepropensityforprecipitationincreases withPEGcoverageintherangeof2–10M%(Fig.6),whichclearly showsthatAS-drivenaggregationofstealthvesiclesisrelatedto thePEGchains.ASleadstothedehydrationofPEGpolymerchains, whileGdmCldoesnotaffectit(Figs.7and8).Otherkosmotropic salts(suchasNa₃citrate,Na₂SO₄,MgSO₄,)alsoprecipitatePEGy- latedliposomes(Fig.9), whilechaotropicsalts(likeGdmCland MgCl₂)do not. We propose thatkosmotropic saltsmay induce saltingout onthepolymerchainspromotingtheirhydrophobic interaction [9] and leading tothe separation of thePEG phase togetherwiththecoupledliposomes.Reversibilityofaggregation isthesimpleconsequenceofthereductionofconcentrationofthe kosmotropicagent,similarlytowhathappens incase ofprotein purificationorduringtheelutionphase ofhydrophobicinterac- tionchromatography.TheobservationthathighASconcentration resultsinrapidfusionofthevesiclesmaybeexplainedbyexces- sivestructuralalterationsofPEGchainsduetotheirhydrophobic modification.ModificationofstructureandhydrationofPEGchains mayleadtostericimbalanceofthevesiclesasitwasshownearlier [15,29].

SinceAS(andotherkosmotropicagents)maybeusedduring formulation of PEGylated nanoparticles(e.g., remote loadingof doxorubicin intoliposomes[14]), great careis neededtoavoid unwantedaggregationorfusionphenomena,whichmaytakeplace during eitherproduction or storage. Thereis a hypothesis that complementactivationrelatedpseudoallergy(CARPA),observed

(8)

intheclinicalpracticewhenPEGylatedliposomesareadministered intravenously,maybecausedbysporadicaggregationorfusionof liposomes[6].Theresultspresentedabovemayopentherouteto produceaggregatesorfusedvesiclesinacontrolledmannerand testtheirroleinpseudoallergicreactions.Reversibleaggregation mayalsobeusedtoseparatePEGylatedliposomes/nanoparticles fromtheirouteraqueousphaseduringformulation.

Acknowledgements

ZetapotentialexperimentswerecarriedoutatDepartmentof Pharmaceutics,SemmelweisUniversitywithsupportofDr.István AntalandtechnicalassistanceofDr.ViktorFülöp,forbothofwhich theauthorsaregrateful.WehighlyappreciateDr.PálGróf’shelpin analysisofzetapotentialdata.

ThisworkwassupportedbygrantsfromtheHungarian Sci- enceFoundation(OTKAK109480).Theresearchleadingtothese results has received funding from the European Union’s Sev- enthFrameworkProgram(FP7/2007-2013)undergrantagreement no HEALTH-F2-2011-278850 (INMiND)and NMP-2012-309820, (NanoAthero).Theauthorsalsoacknowledgetheﬁnancialsupport bytheHungarian GovernmentNationalResearch, Development andInnovationFund(TÉT-13-IL-2-2014-0001).TheFTIRspectro- scopicpartwassupportedbytheJánosBolyaiResearchScholarship oftheHungarianAcademyofSciences(J.M.).

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.colsurfb.2016.06.

056.

References

[1]F.M.Veronese,G.Pasut,PEGylation,successfulapproachtodrugdelivery, DrugDiscov.Today10(2005)1451–1458.

[2]D.Papahadjopoulos,T.M.Allen,A.Gabizon,E.Mayhew,K.Matthay,S.K.

Huang,K.D.Lee,M.C.Woodle,D.D.Lasic,C.Redemann,Stericallystabilized liposomes:improvementsinpharmacokineticsandantitumortherapeutic efﬁcacy,Proc.Natl.Acad.Sci.88(1991)11460–11464.

[3]T.M.Allen,C.Hansen,Pharmacokineticsofstealthversusconventional liposomes:effectofdose,Biochim.Biophys.Acta(BBA)-Biomembr.1068 (1991)133–141.

[4]T.M.Allen,P.R.Cullis,Liposomaldrugdeliverysystems:fromconceptto clinicalapplications,Adv.DrugDeliv.Rev.65(2013)36–48.

[5]J.Szebeni,Complementactivation-relatedpseudoallergy:anewclassof drug-inducedacuteimmunetoxicity,Toxicology216(2005)106–121.

[6]J.Szebeni,P.Bedocs,Z.Rozsnyay,Z.Weiszhar,R.Urbanics,L.Rosivall,R.

Cohen,O.Garbuzenko,G.Bathori,M.Toth,R.Bunger,Y.Barenholz, Liposome-inducedcomplementactivationandrelatedcardiopulmonary distressinpigs:factorspromotingreactogenicityofDoxilandAmBisome, Nanomed.:Nanotechnol.Biol.Med.8(2012)176–184.

[7]P.Cummins,B.O’Connor,Hydrophobicinteractionchromatography,in:D.

Walls,S.T.Loughran(Eds.),ProteinChromatography,vol.681,HumanaPress, 2011,pp.431–437.

[8]Interactionsofnanoparticleswithplasmaproteins:implicationonclearance andtoxicityofdrugdeliverysystems,ExpertOpin.DrugDeliv.8(2011) 343–357.

[9]M.J.Hey,D.P.Jackson,H.Yan,Thesalting-outeffectandphaseseparationin aqueoussolutionsofelectrolytesandpoly(ethyleneglycol),Polymer46 (2005)2567–2572.

[10]R.C.MacDonald,R.I.MacDonald,B.P.Menco,K.Takeshita,N.K.Subbarao,L.R.

Hu,Small-volumeextrusionapparatusforpreparationoflarge,unilamellar vesicles,Biochim.Biophys.Acta1061(1991)297–303.

[11]Dynamiclightscattering−commontermsdeﬁned,in,vol.2015,www.

malvern.com.

[12]X.L.R.Iraolagoitia,M.F.Martini,Ca²⁺adsorptiontolipidmembranesandthe effectofcholesterolintheircomposition,ColloidsSurf.B:Biointerfaces76 (2010)215–220.

[13]G.Drin,V.Morello,J.-F.Casella,P.Gounon,B.Antonny,Asymmetrictethering ofﬂatandcurvedlipidmembranesbyagolgin,Science(NewYork,N.Y.)320 (2008)670–673.

[14]Y.Barenholz,Liposomeapplication:problemsandprospects,Curr.Opin.

ColloidInterfaceSci.6(2001)66–77.

[15]O.Tirosh,Y.Barenholz,J.Katzhendler,A.Priev,Hydrationofpolyethylene glycol-graftedliposomes,Biophys.J.74(1998)1371–1379.

[16]E.Reimhult,F.Höök,B.Kasemo,Intactvesicleadsorptionandsupported biomembraneformationfromvesiclesinsolution:inﬂuenceofsurface chemistry,vesiclesize,temperature,andosmoticpressure,Langmuir19 (2003)1681–1691.

[17]B.Seantier,B.Kasemo,Inﬂuenceofmono-anddivalentionsontheformation ofsupportedphospholipidbilayersviavesicleadsorption,Langmuir25 (2009)5767–5772.

[18]J.Jass,T.Tjarnhage,G.Puu,Fromliposomestosupported,planarbilayer structuresonhydrophilicandhydrophobicsurfaces:anatomicforce microscopystudy,Biophys.J.79(2000)3153–3163.

[19]J.Jass,T.Tjärnhage,G.Puu,Atomicforcemicroscopyimagingofliposomes,in:

D.Nejat(Ed.),MethodsinEnzymology,vol.367,AcademicPress,2003,pp.

199–213.

[20]R.P.Richter,A.R.Brisson,Followingtheformationofsupportedlipidbilayers onmica:astudycombiningAFM,QCM-D,andellipsometry,Biophys.J.88 (2005)3422–3433.

[21]T.M.Riddick,ControlofColloidStabilityThroughZetaPotential,Zeta-Meter Inc.viaLivingstonPublishingCompany,Wynnewood,1968.

[22]L.Wu,J.Zhang,W.Watanabe,Physicalandchemicalstabilityofdrug nanoparticles,Adv.DrugDeliv.Rev.63(2011)456–469.

[23]F.E.Bailey,R.W.Callard,Somepropertiesofpoly(ethyleneoxide)¹inaqueous solution,J.Appl.Polym.Sci.1(1959)56–62.

[24]M.G.Cacace,E.M.Landau,J.J.Ramsden,TheHofmeisterseries:saltand solventeffectsoninterfacialphenomena,Q.Rev.Biophys.30(1997)241–277.

[25]J.A.Queiroz,C.T.Tomaz,J.M.S.Cabral,Hydrophobicinteraction chromatographyofproteins,J.Biotechnol.87(2001)143–159.

[26]Z.Varga,J.Mihály,S.Berényi,A.Bóta,Structuralcharacterizationofthe poly(ethyleneglycol)layerofstericallystabilizedliposomesbymeansofFTIR spectroscopy,Eur.Polym.J.49(2013)2415–2421.

[27]M.Rozenberg,A.Loewenschuss,Y.Marcus,IRspectraandhydrationof short-chainpolyethyleneglycols,Spectrochim.ActaA:Mol.Biomol.Spectrosc.

54(1998)1819–1826.

[28]K.P.Ananthapadmanabhan,E.D.Goddard,Aqueousbiphaseformationin polyethyleneoxide-inorganicsaltsystems,Langmuir3(1987)25–31.

[29]A.Priev,A.Samuni,O.Tirosh,Y.Barenholz,Theroleofhydrationin stabilizationofliposomes:resistancetooxidativedamageofPEG-grafted liposomes,in:G.Gregoriadis,B.McCormack(Eds.),TargetingofDrugs6,vol.

300,Springer,US,1998,pp.147–167.