1

Title:Microfluidic‐controlledmanufactureofliposomesforthesolubilisationof1

apoorlywatersolubledrug.2

3

Authors:ElisabethKastner,VarunVerma,DeborahLowryandYvonnePerrie*4

5

6

Aston Pharmacy School, School of Life and Health Sciences, Aston University,7

Birmingham,UK,B47ET.8

9

10

11

12

*Correspondence: ProfessorYvonnePerrie13

AstonPharmacySchool14

SchoolofLifeandHealthSciences15

AstonUniversity,Birmingham,UK.B47ET.16

Tel:+44(0)121204399117

Fax:+44(0)121359073318

E‐mail:y.perrie@aston.ac.uk19

20

21

Keywords:Liposomes,microfluidics,poorlysolubledrugs,bilayer loading,high22

throughput23

24

25

26

 27

28

29

30

31

32

33

2

Abstract34

Besides their well‐described use as delivery systems for water‐soluble drugs,35

liposomes have the ability to act as a solubilizing agent for drugs with low36

aqueoussolubility.However,akeylimitationinexploitingliposometechnology37

istheavailabilityofscalable,low‐costproductionmethodsforthepreparationof38

liposomes. Here we describe a new method, using microfluidics, to prepare39

liposomal solubilising systems which can incorporate low solubility drugs (in40

thiscasepropofol).Thesetup,basedonachaoticadvectionmicromixer,showed41

highdrug loading (41mol%)of propofol aswell as the ability tomanufacture42

vesicleswithatprescribedsizes(between50to450nm) inahigh‐throughput43

setting.Ourresultsdemonstratetheabilityofmergingliposomemanufacturing44

and drug encapsulation in a single process step, leading to an overall reduced45

processtime.Thesestudiesemphasisetheflexibilityandeaseofapplyinglab‐on‐46

a‐chipmicrofluidicsforthesolubilisationofpoorlywater‐solubledrugs.47

 48

3

1 Introduction 49

The delivery of drugs by liposomes was first described in the 1970s by50

Gregoriadis(GregoriadisandRyman,1971)andthereisnowarangeofclinically51

approved liposome‐based products that improve the therapeutic outcome for52

patients.Whilstliposomesarecommonlyconsideredforthedeliveryofaqueous53

solubledrugs,theyarealsowellplacedtoactassolubilisationagentsfordrugs54

withlowaqueoussolubility.Thisisofconsiderableinterestgiventhatmorethan55

40%ofallnewchemicalentitiesindiscoveryhavelowsolubilityandsubsequent56

issues in bioavailability (Savjani et al., 2012; Williams et al., 2012). The57

encapsulation of low solubility drugs into the bilayer of liposomes allows not58

only for their solubilisation in an aqueous media, but furthermore can offer59

protection from degradation and control over the pharmacokinetic drug60

distributionprofileandimprovedtherapeuticefficacy.61

62

When solubilising drug within the liposomal bilayer, drug incorporation and63

release rates has been shown to depend on the properties of the drug, the64

compositionoftheliposomes,thelipidchoiceandconcentration(Alietal.,2010;65

Alietal.,2013;Mohammedetal.,2004).Forexample, the logPandmolecular66

weight are often considered to impact on bilayer loading, and studies have67

shownthatmolecularweightmayplayadominantrole(Alietal.,2013).When68

consideringthedesignofliposomes,therearearangeofparametersthatimpact69

on bilayer loading efficacy. For example, we have previously shown that70

increasing the bilayer lipophillic volume (by adopting longer alkyl chain lipids71

within the liposomes) increases the loading ability of liposomal systems72

(Mohammed et al., 2004; Ali et al., 2013). Similarly, incorporation of charged73

lipidswithin the liposomal systemmayalso impactonbilayer loading through74

electrostatic repulsion of drugs with like‐charged liposomal bilayers75

(Mohammed et al., 2004). Incorporation of cholesterol, whilst stabilising the76

liposomeswasalsoshowntoinhibitbilayerdrugloading(Alietal.,2010)dueto77

thespace‐fillingactionofcholesterolintheliposomalbilayer.Byincreasingthe78

orientationorderofthephospholipidhydrocarbonchains,cholesteroldecreases79

bilayer permeability. Indeed, the presence of cholesterol in liposomes80

solubilisingpropofolwasshowntoshiftthedrugreleaseprofilefromzero‐order81

4

(when no cholesterol was present) to first order (when 11 to 33 mol% of82

cholesterolwas incorporated). Thismaps to the idea thatwithout cholesterol83

thebilayercanbethoughtofasmore‘porous’innaturecomparedwiththemore84

condensed and less permeable cholesterol‐containing liposomebilayers (Ali et85

al.,2010).86

87

However,whilstawiderangeofstudieshavelookedattheeffectofformulation88

parametersontheapplicationofliposomesassolubilisingagents,morefocusis89

required into making liposomes a cost‐effective solubilising agent. Recent90

advancesinlab‐on‐a‐chipbasedtoolsforprocessdevelopmenthasalreadylead91

tomicrofluidic‐basedmethodologies in drug development (Dittrich andManz,92

2006;Weigletal.,2003;Whitesides,2006).Indeed,microfluidics‐basedmethods93

(whichexploitcontrolledmixingofstreamsinmicro‐sizedchannels)havebeen94

describedforthemanufactureofliposomesandlipidnanoparticles(vanSwaay,95

2013).Liposomeformationbymicrofluidicsprimarilydependsontheprocessof96

controlled alterations in polarities throughout the mixer chamber, which is97

followed by a nanoprecipitation reaction and the self‐assembly of the lipid98

moleculesintoliposomes.Generally,twoormoreinletstreams(lipidsinsolvent99

and an aqueous phase) are rapidly mixed together and flow profiles in the100

chamber itself areof lowReynoldsnumbersandcategorizedas laminar.Using101

microfluidic systems a tight control of the mixing rates and ratio between102

aqueous and solvent streams is achieved,with lower liquid volumes required,103

whichfacilitatesprocessdevelopmentbyreducingtimeanddevelopmentcosts.104

Thesystemsaredesignedwiththeoptionofhigh‐throughputmanufacturingand105

are generally considered as less harsh compared to conventional methods of106

liposome manufacturing that are based on mechanical disruption of large107

vesicles into small and unilamellar ones (Wagner and Vorauer‐Uhl, 2011).108

Within the range of microfluidic mixing devices, we use a chaotic advection109

micromixer,aStaggeredHerringboneMicromixer(SHM).Thefluidstreamsare110

passed through the series of herringbone structures that allow for the111

introductionofachaoticflowprofile,whichenhancesadvectionanddiffusion.A112

chaoticadvectionmicromixer,aswellasflowfocusingmethods,wereshownto113

allowforscalability,associatedwithdefinedvesiclesizes(Belliveauetal.,2012;114

5

Jahn et al., 2007). The method based on chaotic advection was shown to115

reproducibly generate small unilamellar liposomes (SUV)with tight control of116

theresultingliposomesizesatflowratesashighas70mL/mininaparallelized117

mixer‐setup. We have previously shown that microfluidics can be used to118

produce cationic liposomal transfection agents (Kastner et al., 2014), where119

design of experiments and multivariate analysis revealed the ratio between120

aqueousandsolventphasehavingastrongrelevancefortheformationofsize‐121

controlled liposomes. Within this study, we have exploited microfluidics to122

develop a high‐throughput manufacturing process to prepare liposomes123

solubilisingdrugwithintheirbilayer(Figure1).124

125

126

127

Figure 1: Schematic depiction of the liposome formation process based on the128

SHMdesign,achaoticadvectionmicromixer for(A)empty liposomes, (B)drug129

loadedliposomesand(C)chamberlayout.130

131

132

133

134

135

Width200µm

Height78µm

6

136

2 Materials and Methods 137

2.1 Materials 138

EggPhosphatidylcholine(PC)andCholesterolwereobtainedfromSigma‐Aldrich139

Company Ltd., Poole, UK. Ethanol and methanol were obtained from Fisher140

Scientific UK, Loughborough, UK. TRIS Ultra Pure was obtained from ICN141

Biomedicals, Inc., Aurora, Ohio. Propofol (2,6‐Bis(isopropyl)phenol) and 5(6)‐142

Carboxyfluorescein(CF)wasobtainedfromSigma‐AldrichCompanyLtd.,Poole,143

UK.144

145

2.2 Micromixer design and fabrication 146

The micromixer was obtained from Precision NanoSystems Inc., Vancouver,147

Canada. Themixer containedmoulded channelswhichwere 200 µm x 79 µm148

(width x height) with herringbone features of 50 x 31 µm. 1 mL disposable149

syringeswereusedfortheinletstreams,withrespectivefluidconnectorstothe150

chip inlets. Formulations using the micromixer were performed on a151

NanoAssemblr™(PrecisionNanoSystems Inc.,Vancouver,Canada) thatallowed152

for controlof the flowrates (0.5 to6mL/min)and the flowratios (1:1 to1:5,153

ratiobetweensolvent:aqueous)betweentherespectivestreams.154

2.3 Formulation of small unilamellar vesicles using microfluidics 155

Lipids (16:4 molar ratio of PC and Cholesterol, 8:1 w/w) were dissolved in156

ethanol.SUVweremanufacturedbyinjectingthelipidsandaqueousbuffer(TRIS157

10mM, pH 7.2) into separate chamber inlets of themicromixer. The flow rate158

ratio(FRR)(ratiobetweensolventandaqueousstream)aswellasthetotalflow159

rate(TFR)ofbothstreamswerecontrolledbysyringepumps,calibratedtothe160

syringeinnerdiameter.FRRvariedfrom1:1to1:5andTFRvariedfrom0.5to6161

mL/min, extrapolated fromprevious reportedmethodsapplyinga SHMdesign162

withachanneldiameterof200µm(Kastneretal.,2014).TheSUVformulation163

wascollectedfromthechamberoutletanddialysedatroomtemperatureagainst164

TRISbuffer(10mM,pH7.2) forremovalofresidualsolvent.Themodeldrugof165

low aqueous solubility was propofol (2,6‐Bis(isopropyl)phenol), previously166

showntocorrespondtohighencapsulationvalues in liposomalsystemsdueto167

7

its low molecular weight (Ali et al., 2013). To encapsulate propofol, the low168

solubility drug was dissolved with the lipids in ethanol (0.5 to 3mg/mL) and169

thereby liposome formation and encapsulation of the drug was performed170

simultaneouslyusingthemicromixermethod.171

172

2.4 Lipid film hydration and sonication 173

Multilamellar vesicles (MLV) were prepared using the lipid film hydration174

method (Bangham et al., 1965). Basically, lipids were dissolved in175

chloroform/methanol (9:1 v/v) and the organic solvent was subsequently176

removedbyrotaryevaporationundervacuumtoformadrylipidfilmwhichwas177

flushed with N2 to ensure removal of solvent residues. The lipid film was178

hydratedwithTRISbuffer(10mM,pH7.2)toformMLV.SUVwerethenformed179

viaprobesonication(Sonirep150plus,MSE;5minatanamplitudeof5).180

 181

2.5 Measurement of particle characteristics  182

Characterisation of the liposomes included size measurements using dynamic183

light scattering (DLS) (Malvern NanoZS), reported as the z‐average (intensity184

basedmean particle diameter) formonomodal size distributions and the zeta185

potentialusingparticleelectrophoresis(MalvernNanoZS).Polydispersity(PDI)186

measurements(MalvernNanoZS)wereusedtoassessparticledistribution.187

188

2.6 Quantification of drug concentrations 189

QuantificationofpropofolwasperformedbyreversephaseHPLC(Luna5µC18,190

Phenomenex,poresizeof100A,particlesizeof5µm).DetectorwasUV/Vis,at191

268nm.The flowratewasconstantat1.0mL/minthroughoutwithagradient192

elutionfrom5%B(Methanol),95%A(0.1%TrifluoroaceticAcid(TFA)inwater)193

to 100% B over 10 minutes. HPLC‐grade liquids were used, sonicated and194

filtered.Thecolumntemperaturewascontrolledat35ºC.Allanalysiswasmade195

inClarity,DataApexversion4.0.3.876.Quantificationwasachievedbyreference196

to a calibration curve produced from standards (six replicates in ethanol) at197

concentrationsfrom0.01to1mg/mL.ThecalibrationcurvehadalinearityR2≥198

8

0.997, and all measurements were within the level of detection and level of199

quantification.200

201

2.7 Determination of drug loading into liposomes 202

Theamountofdrug loaded into thebilayerwasmeasuredbydeterminationof203

the residual amount of drug in the liposome bilayer after removal of non‐204

entrappeddrugbydialysis(sinkconditions)against1LofTRISbuffer,10mMpH205

7.2 (3500 Da, Medicell Membranes Ltd., London, UK). The drug content was206

measuredbyHPLCasdescribed in section2.6. Thisprotocolwas validatedby207

assessingtherateofpropofolremovalbydialysis.208

 209

2.8 Stability study 210

Forthestabilitystudy,formulationsofpropofol‐loadedSUVwerestoredat4°C,211

25°C and 40°C in pharmaceutical grade stability cabinets over 60 days (time212

pointmeasurementsatday0,7,14,21,28and60).Samplesweretakenatthese213

specifictimepointsformeasurementofparticlecharacteristics(section2.5)and214

drug loading (section 2.6). Sampleswere dialysed against 500mLTRIS buffer215

(10mM, pH7.2, sink conditions) at each time point to remove non‐entrapped216

propofol.Propofolcontentremainingintheliposomeformulationwasassessed217

byHPLCasdescribedinsection2.6.218

219

2.9 Recovery of lipids and propofol   220

Toassesstheoveralllipidandpropofolrecoveryinthemicrofluidicsmethod,the221

amount of lipid and propofol was measured by HPLC and expressed as %222

recovery compared to the initial amount of lipids or propofol available in the223

stock.TheHPLCmethodwasthesameasdescribedsection2.6,andlipidswere224

quantifiedbyanevaporative lightscattering(ELS)detector(Sedere,Sedex90),225

setat52°CandcoupledtotheHPLC.226

227

2.10 Freeze Fracturing Imaging 228

Twomicrolitresof liposomesuspensionwereplacedinaridgedgoldspecimen229

support and frozen rapidly by plunging into a briskly stirred mixture of230

9

propane:isopentane(4:1)cooledinaliquidnitrogenbath.Fracturing,withacold231

knife, and replicationwereperformed inaBalzersBAF400Dapparatusunder232

conditionssimilartothosedescribedpreviouslyforfreeze‐fractureofliposomes233

(Forgeetal.,1978;Forgeetal.,1989).Thereplicasgeneratedwerefloatedoffon234

water,cleanedindomesticbleachdiluted1:1indistilledwater,andthenwashed235

several times in distilled water before mounting on grids for electron236

microscopy.ThereplicaswereviewedinaJEOL1200EXIItransmissionelectron237

microscopeoperatingat80kvanddigitalimagescollectedwithaGatancamera.238

Imagesofthefreeze‐fracturedsamplesarepresentedinreversecontrastsothat239

shadowsappearblack.FracturingimagingwasperformedbyProf.AndrewForge240

atUCLEarInstitute,London,UK.241

242

2.11 Drug release study 243

The in‐vitro release rate of the drug was determined by incubating the drug‐244

loaded liposomes in1LTRISbuffer (10mM,pH7.2)after removalof thenon‐245

incorporated drug, at 37°C in a shaking water bath (150 shakes/min). Three246

independent formulations of drug‐loaded liposomesmadeby themicrofluidics247

method(TFR2mL/min,FRR1:3)andstandardlipidfilmhydrationfollowedby248

sonicationwereincubated(3mLperformulation)andsamplesof200µLwere249

withdrawn at time intervals of 0.5 h, 1 h, 2 h, 4 h, 8 h and 16 h. Drug250

quantificationwas performed as described in section 2.6 and expressed as%251

cumulativereleaserelativetotheinitialamountofdrugencapsulated.252

253

2.12 Incorporation of an aqueous marker within liposomes 254

Tovalidatetheformulationofliposomes,thepresenceofanaqueouscorewithin255

the nanoparticles manufactured was verified by including and imaging of an256

aqueousfluorescentdye.Liposomesweremanufacturedasdescribedinsection257

2.3and2.4with1mMCarboxyfluorescein(CF) included in theaqueousbuffer258

(TRIS,10mM,pH7.2).LiposomeswithentrappedCFwereseparatedfromun‐259

entrapped dye by dialysis over night against 1 L fresh TRIS buffer, pH 7.2.260

Liposomes were imaged under a confocal microscope SP5 TCS II MP, Leica261

10

Microsystems, Leica TCSSP5 II, 63x objective (HCX PLAPO 63x/1.4‐0.6 oil CS).262

ImagesweretakenbyCharlotteBland,AstonUniversity,ARCHAfacility.263

 264

2.13 Statistical tools 265

Ifnotstatedotherwise,resultswerereportedasmean±standarddeviation(SD).266

One‐ or two‐way analysis of variance (ANOVA) was used to assess statistical267

significance, followed by Tukeys multiple comparing test and t‐test was268

performedforpairedcomparisons.Significancewasacknowledgedforpvalues269

less than0.05 (markedwith *).All calculationsweremade inGraphPadPrism270

version6.0(GraphPadSoftwareInc.,LaJolla,CA).271

272

3 Results and discussion 273

3.1 Influence  of  the  flow  rate  ratio  of  aqueous  and  solvent  stream  on 274

liposome size 275

The increase inpolarity throughout thechamberdrives the formationof small276

unilamellar liposomes(SUV) inmillisecondsofmixing.For their formation, the277

rate of mixing as well as the ratio of aqueous to solvent stream has been278

anticipated as crucial factors. The formation of the liposomes is based on a279

nanoprecipitationreaction,wheresupersaturationoccursandtheliposomesare280

formedbyself‐assemblyafteraggregationofthelipidmolecules.Theinitialaim281

ofthisworkwastoassesstheformationofliposomesbymicrofluidicmixingand282

assess the efficacy of this system to act as a solubilising agent. Therefore,283

liposomeswerepreparedfromPCandCholesterol(16:4molarratio,8:1w/w)at284

different total flow rates (TFR) and flow rate ratios (FRR) and the size,285

polydispersityandzetapotentialweremeasured.286

287

Liposomesformedatlowflowrateratio(1:1)showedthelargestsizeofaround288

450nm;increasingtheflowrateratioresultedinsmallerliposomes(around40‐289

50nm)at constant flow ratesof2mL/min (TRIS,10mM,pH7.2) (Figure2A).290

However,increasingtheflowrateratioincreasedpolydispersity(toamaximum291

of0.4;Figure2B).Liposomespreparedata flowrateratioof1:3areshown in292

11

Figure2C,demonstratingtheirsmallnature,withaveragesizesofthevesiclesin293

agreement with average vesicle diameters obtained by particle sizing via294

dynamic light scattering (~40 nm). In contrast, the smallest vesicle size of a295

comparable formulation achievable via probe sonication with this lipid296

formulationwas100nm in size atPDIsof0.3 (datanot shown).Toverify the297

formation of liposomes, rather than micelles, the liposomes made by the298

microfluidicsmethodwerepreparedencapsulatinganaqueousfluorescentdye,299

carboxyfluorescein(CF,1mM),whichwasincludedintheaqueousphaseduring300

liposomemanufacturingbymicrofluidicsandlipidfilmhydration.Afterremoval301

ofthefreeCFbydialysisovernight,theremainingdyeentrappedintheparticles302

wasvisualizedbyconfocalmicroscopy.Brightgreenfluorescentcoresvisiblein303

theparticlesmanufacturedbythemicrofluidicsmethod(Figure2D)wereinline304

withimagesobtainedfromliposomesmanufacturedwiththelipidfilmhydration305

method(imagesnotshown);whichconfirmsthepresenceofaqueouscoresand306

theformationofliposomesinthenovelmicrofluidicsmethod.307

308

12

Figure2: Liposome size (A) andpolydispersity (B)of vesicles formulatedwith309

microfluidicsmethod at increasing flow ratios. ns = not significant (p>0.05), *310

denotes statistical significance (p<0.05) in comparison to FRR 1:1 (C) Freeze311

fracturingelectronmicroscopy imagesforempty liposomesmanufacturedwith312

themicrofluidicsmethod. Bar represents 100 nm. (D) Fluorescentmicroscope313

images of liposomes manufactured with the microfluidics method,314

carboxyfluoresceinwasencapsulatedwithintheaqueouscoreofthevesiclesasa315

controlforthemanufacturingofbilayerliposomes.Barrepresents20µm.316

317

318

319

These impactof flow rate ratioonvesicle size are inagreementwithprevious320

work showing that the increase in FRR reduces the resulting size of the321

liposomes (Jahnet al., 2010;Kastneret al., 2014;ZookandVreeland,2010).A322

correlationbetweenhigher flowrate ratiosandsmaller liposomeparticleshas323

been reported using liposomes composed of 1‐palmitoyl, 2‐oleoyl324

phosphatidylcholine (POPC), cholesterol and the triglyceride triolein, which325

resulted in the production of vesicular structureswith sizes ranging from140326

nmto40nmdependentontheFRRchosenandtriglycerideemulsionsbetween327

20−50nmsizewithnonpolarcores(Zhigaltsevetal.,2012).Theoverall lower328

amount of residual solvent present at higher FRR employed decreases the329

particle fusion (Ostwald ripening), which leads to the formation of smaller330

particles(Zhigaltsevetal.,2012).Theincreaseinpolydispersitymaybearesult331

ofincreaseddilutionathigherFRRreducingtherateofdiffusionalmixingwithin332

themicromixerasnotedinpreviousstudiesapplyingaSHMmixerforliposome333

manufacturing (Kastner et al., 2014).With diffusion being proportional to the334

lipid concentration, increasing FRR is effectively reducing the lipid335

concentration, thus reducing the rate of diffusion, leading topartly incomplete336

nucleation and a lower rate of liposome formation inside the micromixer337

(Balbinoetal.,2013b).Overall, thesefindingsdemonstratethataFRRof1:2to338

1:4resultinliposomesofthesmallestsizeandpolydispersity.Thedilutionfactor339

(due to flow ratios chosen involved in the SHM method) is overall lower340

comparedtoratiosemployedintheflow‐focusingmethod,whichcanreachupto341

13

60(Jahnetal.,2010;Jahnetal.,2007;Jahnetal.,2004).Furthermore,theSHM342

method enhances the diffusionalmixing due to the herringbone structures on343

the channel wall (Stroock et al., 2002), which results in an enhanced mixing344

profilescomparedtotheflow‐focusingtechnique.345

346

3.2 Influence of flow rate on throughput and particle characteristics 347

Toassesstheabilityofthesystemasapotentialhigh‐throughputmanufacturing348

methodforliposomalsolubilisationsystems,weincreasedthetotalflowrate3‐349

foldwhilstmaintainingtheratiobetweenaqueousandsolventstreamconstant.350

Liposome sizewas shown to be independent of the applied flow rate,withno351

significantchangeinvesiclesize(Figure3A),pdi(Figure3B)andzetapotential352

(‐3±2mV;datanotshown).Theseresultssupportthesuitabilityofmicrofluidics353

manufacturingasahighthroughputmethodwithliposomecharacteristicsbeing354

maintained constant whilst increasing the total flow rate in the system. Our355

resultsalsoconfirmthattheflowrateratiousedinthesystemisthemostcrucial356

variableon liposomesize,whichhaspreviouslybeendemonstratedwithother357

systems(Balbinoetal.,2013a;Balbinoetal.,2013b;Jahnetal.,2007;Jahnetal.,358

2004;Kastneretal.,2014).Thescalabilityofthemicrofluidicsmethodhasbeen359

suggested by Belliveau et al. 2013, by parallelization of the mixer chamber.360

Scalability and increase in throughput together demonstrate the industrial361

applicabilitycomparablewithscale‐upoptionsavailable(WagnerandVorauer‐362

Uhl,2011).363

364

365

14

Figure3: Liposome size (A) andpolydispersity (B)of vesicles formulatedwith366

microfluidicsat increasingflowratesandconstantflowratioof1:3,n=3,ns=367

notsignificant(p>0.05).368

369

370

371

As shown, the increase in FRR is the main contributing factor governing372

liposomesize(Figure2A).Nevertheless,anincreaseinFRRwill inevitablylead373

todilutionandlowerliposomeconcentrationsinthefinal liposomesuspension374

produced.Asubsequentconcentrationprocessbasedonfiltration(Pattnaikand375

Ray, 2009), chromatography (Ruysschaert et al., 2005) or centrifugation adds376

additionalprocessingtime.Therefore,tocircumventthisadditionalprocessstep,377

wecounteractedthedilutionofthelipidsathigherFRRbyincreasinginitiallipid378

concentrations introduced to themicromixer at the desired FRR.Through this379

method, liposomes were manufactured at up to 6 fold higher concentrations.380

Increased lipid concentrations at FRR of 1:3 and 1:5 did not significantly381

(p>0.05) influence size and polydispersity compared to the standard lipid382

concentration(Figure4AandB),whereasataFRRof1:1asignificant(p<0.05)383

decreaseinvesiclesizewasobserved(Figure4A).AtthislowerFRR,thehigher384

lipid concentrations may again decreasing particle fusion leading to the385

formationof smallerparticles (Zhigaltsevetal.,2012).Nevertheless, thissetup386

allowstoincreasethefinalliposomeconcentrationaccordingtotheFRRchosen387

without adversely changing resulting vesicle size or polydispersity for the388

smallestvesiclesizesobtainedathigherFRR(Figure4AandBrespectively),due389

tothediffusionalmixingprocessintheSHMdesign.390

391

392

393

394

395

396

397

398

3

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

4

Figu404

the405

den406

stan407

con408

405

406

3.3 408

409

So f416

and417

beu418

load419

stud420

at a421

load422

ran423

417

Usin419

dru420

ure4:Incr

dilution e

notes statis

ndard lip

ncentration

Drug loa

manufac

far,wehav

drapidsyn

usedforah

ding capac

diesshown

a FRR of 1

dingefficie

gingfrom0

ngapropo

ug loading

reaseinth

effect at flo

stical sign

pid concen

ofPCand

ading stud

cturing me

veshownt

thesisoflip

high‐throu

ityof the f

n inFigure

:3 and a T

ency(mol%

0.5to3mg

ofolconcen

(~50 mol%

elipidcon

ow ratios

ificance (p

ntration

Cholestero

ies: The ef

ethod  

that themi

posomes.T

ghputprod

formulation

e2,propofo

TFR of 2m

%;Figure5A

g/mL(effec

ntrationof

%), combin

ncentration

of 1:1, 1:3

p<0.05) in

and (B)

olintheinle

ffect of dru

icrofluidics

Toconsider

ductionofl

nwas con

olwassolu

mL/min. Th

A)wasdet

ctiveconcen

1mg/mL

ned with p

intheeth

3 and 1:5

n comparis

polydispe

etstream,n

ug encaps

smethoda

rtheapplic

liposomes

sidered.Ba

ubilisedwi

he particle

erminedat

ntrationin

in thesolv

particle siz

anolstock

for (A) lip

son to FRR

ersity wit

n=3.

ulation by

allows fors

cabilityoft

assolubilis

asedon the

thin liposo

characteris

tpropofolc

thesolven

ventstream

ze of ~50 n

tocircumv

posome siz

R 1:1 for

th respec

y the liposo

size‐contro

thismetho

singagents

eoptimisa

omesprepa

stics and d

concentrat

ntstream).

mshowedh

nm and a

15

vent

ze, *

the

ctive

ome 

olled

odto

sthe

ation

ared

drug

ions

high

low

16

polydispersity (Figure 5A). Particle size and polydispersity increased notably419

(ca. 600 nm and 0.8 respectively) at the highest propofol concentration (3420

mg/mL in the solvent stream, giving a loading of ~25mol%, Figure 5A),421

suggesting the liposome systemmayhavebecome saturatedordestabilised at422

highpropofolconcentrations(drug‐to‐lipidratio1.72mol/mol). Basedonthis,423

subsequentstudiesadoptedapropofolconcentrationat1mg/mLinthesolvent424

streamforallperformedencapsulationstudies.425

426

The drug encapsulation was further investigated as a function of FRR in the427

microfluidicsmethod.Propofolencapsulation(mol%) in liposomespreparedat428

FRR 1:1, 1:3 and 1:5 remained at approximately 50 mol% with no statistical429

difference.Howeverthiswassignificantlyhigher(p<0.0001)thandrugloading430

in liposomes prepared via sonication (15mol%; Figure 5B). The drug loading431

efficiencyofliposomespreparedbysonicationisinlinewithpreviousreported432

propofol encapsulation (Ali et al., 2013). Furthermore, drug encapsulation did433

notaltervesiclesizeorpolydispersity(Figure5A)andvesiclesizesobtainedby434

dynamic light scatteringwereverifiedby freeze fracturing images (Figure5D).435

Thishigherdrugloadingmaybearesultofthehighlyefficientmixingprocesses436

occurringduringmicrofluidicsthatfavoursincorporationofpropofolwithinthe437

bilayers in the same process as the vesicles form. Indeed, the here presented438

method allows to achieve a propofol encapsulation of ~50 mol%, which439

represents a total propofol amount of ~300 mg/mL in the final liposome440

formulation, representing a 2000‐fold increase to the reported aqueous441

solubilityofpropofol,150µg/mL(Altomareetal.,2003).442

443

444

445

446

447

448

449

450

451

17

452

453

Size

(nm)

40±2 48±2 613±360

PDI 0.03±0.02 0.17 ±

0.01

0.8±0.28

Figure5:(A)Effectofdrugconcentrationsintheethanolinletstream(0.5,1and454

3mg/mL)onencapsulationefficiency(mol%),particlesizeandpolydispersities455

ataflowratioof1:3..(B)Encapsulationefficiency(mol%)ofliposomesformed456

withthemicrofluidicsmethodatflowratiosof1:1,1:3and1:5comparedtothe457

encapsulationefficiencyusingthesonicationmethod.Resultsareaverageoutof458

triplicate formulations and measurements. ns = not significant (p>0.05), *459

denotes statistical significance (p<0.00001) in comparison to microfluidics‐460

basedsamples.(C)Recoveryof lipidsandpropofol inthemicrofluidicsmethod461

atdifferentflowratios.Resultsareexpressedas%comparedtotheinitiallipid462

andpropofolamountpresent(n=3).(D)Freezefracturingelectronmicroscopy463

18

images for liposomes loaded with the low solubility model drug (propofol)464

manufacturedwiththemicrofluidicsmethod.Barrepresents100nm465

466

467

468

Toconsider,drugreleaseprofiles,thein‐vitroreleaseofpropofolencapsulatedin469

liposomesbymicrofluidicswasmonitoredat37°Cover16h.Liposomesformed470

withthemicrofluidicsmethodhadasignificanthigherdrugencapsulatedatthe471

start of the release study (~55mol%) compared to those vesicles formed by472

sonication(20mol%drugencapsulation).However,relativetoinitialloading,an473

initialreleaseofca40%wasobservedat1hforbothformulations,followedbya474

continuous release of 90% of the encapsulated drug was observed over 8 h475

(Figure6).Whereasthefattyalcoholalkylchainlengthwasshowntoaffectthe476

release profile of encapsulated propofol (Ali et al., 2013), here the method of477

liposome manufacturing was shown to mainly affect the amount of drug478

incorporated into the liposomes, without altering the release profile of the479

encapsulated drug against sink conditions. Previous we have shown that480

solubilisation of propofol in phosphatidylcholine liposomes followed a zero‐481

orderreleasekinetics,wheretheincorporationofahigheramountofcholesterol482

shifted the release rates towards a first‐order releasemodel (Ali et al., 2010),483

implying that the release kinetics itself are mainly dominated by the lipid484

composition and physicochemical characteristics rather than the method of485

liposomemanufacturing.Thismayproveadvantageousinthedevelopmentofan486

IVformulation;thepharmacokineticreleaseprofileofpropofolhasbeenstudies487

previouslyinacolloidaldispersionbetween20‐100nm(Caietal.,2012),where488

rapiddistributionofpropofol compared to the commercialproductDiprivan®489

highlighted the need on the development of new techniques for the490

encapsulationoflowsolubilitydrugs.491

492

493

494

495

496

19

497

Figure6: Effect ofmanufacturingmethod to thedrug releaseof propofol from498

liposomes.Resultsshowthecumulativedrugreleaseprofile fromformulations499

manufacturedwith the standard lipid film hydration / sonicationmethod and500

microfluidicsandrepresentpercentagecumulativereleaseofinitiallyentrapped501

propofol,expressedasthemeansofthreeexperiments±SD.502

503

It is important to verify both lipid and drug recovery when using the504

microfluidics method, to ensure cost‐effectiveness and that lipid and drug505

concentrationsremainlockedattheratioinitiallydesignedpriortoformulation.506

Todate,thequantificationoflipidsismainlydominatedbytimeintensiveassays507

likemass spectrometry (Moore et al., 2007). Here,we introduce a simple and508

robustmethodoflipidquantificationbasedonevaporativelightscattering(ELS)509

detectionandHPLCseparation.WecoupledanELSdetectordownstreamaHPLC510

separationmethod,whichallowed forquantificationofanysolids in theeluate511

withalowervolatilitythanthemobilephase.Microfluidicsbasedliposomal‐drug512

formulations showed good recovery of the drug (88 ‐ 92%; Figure 5C),513

independentoftheFRR.Similarly,lipidrecoverywashighatFRRof1:1and1:3514

(97%and89%;forFRR1:1and1:3respectively;Figure5C).Asignificantdrop515

(79%;p<0.01)inlipidrecoverywasnotedataflowratioof1:5,suggestingthat516

higher FRR employed in themicrofluidics methodmay impede lipid recovery517

duetoenhanceddilutioninthechamber.Nevertheless,thesmallestvesiclesize518

(~50nm)canbeobtainedataFRRof1:3(Figure2A)andanyfurtherincreasein519

20

FRRwillnotbenefittheformulation(size,pdianddrugencapsulation).Basedon520

this,wechosetheFRR1:3foralong‐termstabilitystudy.521

522

3.4 The  effect of manufacturing methods on  liposome  stability  and drug 523

encapsulation over 8 weeks  524

TheSHMmethodwaspreviously investigated for theencapsulationofahighly525

solubledrug,withapproximately100%loadingefficienciesbeingreportedusing526

doxorubicinasamodeldrug(Zhigaltsevetal.,2012);theauthorsdemonstrated527

highdrugretentionofencapsulateddrugwithliposomesstoredat4°Coverthe528

courseof eightweeks (Zhigaltsevet al., 2012). Following the assessment that529

liposomesmanufactured by themicrofluidics method yields significant higher530

encapsulationofpropofol,similarlyweperformedaneight‐weekstabilitystudy531

toverifytheintegrityofthevesiclesatdifferentstoragetemperatures.Vesicles532

werepreparedusingmicrofluidicsasdescribedabove,andtheinitialamountof533

propofol encapsulated was determined after removal of free drug by dialysis.534

Vesicles were stored at 4°C, 25°C/60%RH and 40°C/75%RH (standard ICH535

temperatures) in pharmaceutical grade stability cabinets and the formulations536

madebythesonicationmethodwerestoredat25°C/60%RH(Figure7,Table1),537

acting as the control method. The control liposomes formed by sonication538

showed good stability in terms of size retention over the course of the study.539

Similarly, for liposomes prepared using microfluidics, vesicle size remained540

unaffected after storage over 8weeks at 4°C and 25°C. In contrast, liposomes541

stored at 40°C significantly increase in size from initially 55 nm to 120 nm542

(Figure 7A), with no notable affect to polydispersity, suggesting the liposome543

populationasawholehaschanged insizerather thanasub‐setof thevesicles544

(Table1).545

546

547

Table1:Polydispersity atdifferent storage conditions for8weeks.Results are548

meanoutoftriplicateformulationsandmeasurements.549

Day 0 7 14 21 28 60

Microfluidics

4°C 0.403±0.02 0.286±0.01 0.282±0.01 0.295±0.01 0.261±0.01 0.305±0.01

21

25°C 0.403±0.02 0.295±0.01 0.279±0.01 0.301±0.04 0.302±0.03 0.266±0.03

40°C 0.403± 0.02 0.254±0.001 0.121±0.02 0.119±0.001 0.129±0.01 0.221±0.01

Sonication

25°C 0.656±0.02 0.652±0.02 0.522±0.15 0.658±0.049 0.552±0.04 0.505±0.06

550

Figure 7: Size (A) and drug encapsulation (mol%) (B) at different storage551

conditions over 8 weeks. Results are mean of triplicate formulations and552

measurements.553

554

Minor (but not significant) drug loss from the liposomeswas detected for the555

formulationsat4°Cand25°Cafter the first7daysofstorage(Figure7B),after556

whichtheformulationsremainedstablewithfinaldrugencapsulationvaluesof557

41±1 mol% and 41±4 mol% at 4°C and 25°C storage conditions respectively558

(Figure7B).Similarly,withliposomesformulatedusingsonicationshowedand559

initialdruglosswhenstoredat25°C/60%RHwhichthenplateauedout(Figure560

7B). Notabledruglossfromthemicrofluidicsystemswasonlyseenwhenthey561

were stored at elevated temperatures with the formulation stored at 40°C562

showingalmost completedrug lossover the courseof the stability study,with563

only5±1mol%drugremainingencapsulatedafter8weeks,similar to the final564

drug encapsulated in the sonicated liposomes which were stored at565

25°C/60%RH (Figure 7B). Overall, vesicles produced with the microfluidics566

methodwere smallerwith a lowerpolydispersity than thoseobtainedby lipid567

filmhydration/sonication.Thevesiclesmanufacturedbysonicationmaintained568

theirsizearound100±20nmthroughoutthestabilitystudy(storedat25°C)as569

well as their polydispersity (Table 1). Results suggest that the method of570

22

manufacturingmainly impacts thedrugencapsulation rather than thephysical571

properties(size,pdi,zetapotential).Stabilityof the formulations iscrucialand572

these results demonstrate that liposomes formed by themicrofluidicsmethod573

remainovertwomonthsatconditionsof4and25°C.574

575

3.5 Conclusion 576

Here,forthefirsttime,wehavedemonstratedahigh‐throughput,robustmethod577

ofpreparingsize‐controlledliposomesassolubilisingagentsusingmicrofluidics.578

These liposomes have well defined, scalable, process controlled, physico‐579

chemical attributes demonstrating this method is suitable for pre‐clinical and580

clinicalproductionofliposomes.Drugloadingwasshowntobeinanapplicable581

range for clinical application (Biebuyck et al., 1994). Furthermore, using this582

novelmethod,liposomemanufacturinganddrugencapsulationareprocessedin583

a single process step, circumventing an additional drug loading step584

downstream, which notably reduces the time for production of stable drug‐585

loadedvesiclesofspecifiedphysico‐chemicalcharacteristics.586

587

3.6 Acknowledgements 588

Prof. Andrew Forge (UCL Ear Institute, London, UK) is acknowledged for the589

imagingoftheliposomesbyfreezefracturing.CharlotteBland(AstonUniversity,590

ARCHA facility) is acknowledged for the imaging of liposomes by fluorescent591

microscopy. This work was part funded by the EPSRC Centre for Innovative592

ManufacturinginEmergentMacromolecularTherapiesandAstonUniversity.593

594

23

References595

Ali, M.H., Kirby, D.J., Mohammed, A.R., Perrie, Y., 2010. Solubilisation of 596drugs within liposomal bilayers: alternatives to cholesterol as a membrane 597stabilising agent. Journal of pharmacy and pharmacology 62, 1646-1655. 598Ali, M.H., Moghaddam, B., Kirby, D.J., Mohammed, A.R., Perrie, Y., 2013. The 599role of lipid geometry in designing liposomes for the solubilisation of poorly 600water soluble drugs. International journal of pharmaceutics 453, 225-232. 601Altomare, C., Trapani, G., Latrofa, A., Serra, M., Sanna, E., Biggio, G., Liso, 602G., 2003. Highly water-soluble derivatives of the anesthetic agent propofol: in 603vitro and in vivo evaluation of cyclic amino acid esters. European journal of 604pharmaceutical sciences 20, 17-26. 605Balbino, T.A., Aoki, N.T., Gasperini, A.A., Oliveira, C.L., Azzoni, A.R., 606Cavalcanti, L.P., de la Torre, L.G., 2013a. Continuous flow production of 607cationic liposomes at high lipid concentration in microfluidic devices for gene 608delivery applications. Chemical Engineering Journal 226, 423-433. 609Balbino, T.A., Azzoni, A.R., de La Torre, L.G., 2013b. Microfluidic devices for 610continuous production of pDNA/cationic liposome complexes for gene delivery 611and vaccine therapy. Colloids and Surfaces B: Biointerfaces 111, 203-210. 612Bangham, A., Standish, M.M., Watkins, J., 1965. Diffusion of univalent ions 613across the lamellae of swollen phospholipids. Journal of molecular biology 13, 614238-IN227. 615Belliveau, N.M., Huft, J., Lin, P.J., Chen, S., Leung, A.K., Leaver, T.J., Wild, 616A.W., Lee, J.B., Taylor, R.J., Tam, Y.K., 2012. Microfluidic synthesis of highly 617potent limit-size lipid nanoparticles for in vivo delivery of siRNA. Molecular 618Therapy—Nucleic Acids 1, e37. 619Biebuyck, J.F., Smith, I., White, P.F., Nathanson, M., Gouldson, R., 1994. 620Propofol: an update on its clinical use. Anesthesiology 81, 1005-1043. 621Cai, W., Deng, W., Yang, H., Chen, X., Jin, F., 2012. A propofol 622microemulsion with low free propofol in the aqueous phase: Formulation, 623physicochemical characterization, stability and pharmacokinetics. International 624journal of pharmaceutics 436, 536-544. 625Dittrich, P.S., Manz, A., 2006. Lab-on-a-chip: microfluidics in drug discovery. 626Nature Reviews Drug Discovery 5, 210-218. 627Forge, A., Knowles, P., Marsh, D., 1978. Morphology of egg 628phosphatidylcholine-cholesterol single-bilayer vesicles, studied by freeze-etch 629electron microscopy. The Journal of membrane biology 41, 249-263. 630Forge, A., Zajic, G., Davies, S., Weiner, N., Schacht, J., 1989. Gentamicin 631alters membrane structure as shown by freeze-fracture of liposomes. Hearing 632research 37, 129-139. 633Gregoriadis, G., Ryman, B., 1971. Liposomes as carriers of enzymes or drugs: 634a new approach to the treatment of storage diseases. Biochem. J 124. 635Jahn, A., Stavis, S.M., Hong, J.S., Vreeland, W.N., DeVoe, D.L., Gaitan, M., 6362010. Microfluidic mixing and the formation of nanoscale lipid vesicles. Acs 637Nano 4, 2077-2087. 638Jahn, A., Vreeland, W.N., DeVoe, D.L., Locascio, L.E., Gaitan, M., 2007. 639Microfluidic directed formation of liposomes of controlled size. Langmuir 23, 6406289-6293. 641

24

Jahn, A., Vreeland, W.N., Gaitan, M., Locascio, L.E., 2004. Controlled vesicle 642self-assembly in microfluidic channels with hydrodynamic focusing. Journal of 643the American Chemical Society 126, 2674-2675. 644Kastner, E., Kaur, R., Lowry, D., Moghaddam, B., Wilkinson, A., Perrie, Y., 6452014. High-throughput manufacturing of size-tuned liposomes by a new 646microfluidics method using enhanced statistical tools for characterization. 647International Journal of Pharmaceutics. 648Mohammed, A., Weston, N., Coombes, A., Fitzgerald, M., Perrie, Y., 2004. 649Liposome formulation of poorly water soluble drugs: optimisation of drug 650loading and ESEM analysis of stability. International Journal of Pharmaceutics 651285, 23-34. 652Moore, J.D., Caufield, W.V., Shaw, W.A., 2007. Quantitation and 653standardization of lipid internal standards for mass spectroscopy. Methods in 654enzymology 432, 351-367. 655Pattnaik, P., Ray, T., 2009. Improving liposome integrity and easing 656bottlenecks to production. Pharmaceutical Technology Europe 22. 657Ruysschaert, T., Marque, A., Duteyrat, J.-L., Lesieur, S., Winterhalter, M., 658Fournier, D., 2005. Liposome retention in size exclusion chromatography. 659BMC biotechnology 5, 11. 660Savjani, K.T., Gajjar, A.K., Savjani, J.K., 2012. Drug solubility: importance and 661enhancement techniques. ISRN pharmaceutics 2012. 662Stroock, A.D., Dertinger, S.K., Ajdari, A., Mezić, I., Stone, H.A., Whitesides, 663G.M., 2002. Chaotic mixer for microchannels. Science 295, 647-651. 664van Swaay, D., 2013. Microfluidic methods for forming liposomes. Lab on a 665Chip 13, 752-767. 666Wagner, A., Vorauer-Uhl, K., 2011. Liposome technology for industrial 667purposes. J Drug Deliv 2011, 591325. 668Weigl, B.H., Bardell, R.L., Cabrera, C.R., 2003. Lab-on-a-chip for drug 669development. Advanced drug delivery reviews 55, 349-377. 670Whitesides, G.M., 2006. The origins and the future of microfluidics. Nature 671442, 368-373. 672Williams, R.O., Watts, A.B., Miller, D.A., 2012. Formulating poorly water 673soluble drugs. Springer. 674Zhigaltsev, I.V., Belliveau, N., Hafez, I., Leung, A.K., Huft, J., Hansen, C., 675Cullis, P.R., 2012. Bottom-up design and synthesis of limit size lipid 676nanoparticle systems with aqueous and triglyceride cores using millisecond 677microfluidic mixing. Langmuir 28, 3633-3640. 678Zook, J.M., Vreeland, W.N., 2010. Effects of temperature, acyl chain length, 679and flow-rate ratio on liposome formation and size in a microfluidic 680hydrodynamic focusing device. Soft Matter 6, 1352-1360. 681 682 683684

685

686

687

25

688