Asymmetric Supported Lipid Bilayer Formation via Methyl-β-Cyclodextrin Mediated Lipid Exchange: Influence of Asymmetry onLipid Dynamics and Phase BehaviorIlaria Visco,† Salvatore Chiantia,‡ and Petra Schwille*,†

†Department of Cellular and Molecular Biophysics, Max Planck Institute of Biochemistry, Am Klopferspitz 18, 82152 Martinsried,Germany‡Department of Biology, Humboldt University Berlin, Invalidenstrasse 42, 10115 Berlin, Germany

*S Supporting Information

ABSTRACT: Supported lipid bilayers (SLBs) are broadlyused as minimal membrane models and commonly producedby vesicle fusion (VF) on solid supports. Despite itsadvantages, VF does not allow the controlled formation ofbilayers that mimic the leaflet asymmetry in lipid compositionnormally found in biological systems. Here we present asimple, quick, and versatile method to produce SLBs with adesired asymmetric lipid composition which is stable for ca. 4h. We apply methyl-β-cyclodextrin mediated lipid exchange toSLBs formed by VF to enrich the upper leaflet of the bilayerwith sphingomyelin. The bilayer asymmetry is assessed by fluorescence correlation spectroscopy, measuring the lipid mobilityseparately in each leaflet. To check the compatibility of the method with the most common protein reconstitution approaches,we report the production of asymmetric SLBs (aSLBs) in the presence of a glycosylphosphatidylinositol-anchored protein,reconstituted in the bilayer both, via direct protein insertion, and via proteoliposomes fusion. We finally apply aSLBs to studyphase separation and transbilayer lipid movement of raft-mimicking lipid mixtures. The observed differences in terms of phaseseparation in symmetric and asymmetric SLBs with the same overall lipid composition provide further experimental evidence thatthe transversal lipid distribution affects the overall lipid miscibility and allow to temporally investigate leaflet mixing.

■ INTRODUCTION

Supported lipid bilayers (SLBs) are commonly used as aminimal model systems to investigate the properties ofbiological membranes1 and reconstituted protein machi-neries.2,3 SLBs consist of a lipid bilayer adsorbed on thesurface of a solid support and can be produced either byLangmuir−Blodgett (LB) and Langmuir−Schaeffer (LS)deposition, by vesicle fusion (VF) or by a combination ofthese.4 Due to its simplicity, VF5 became a very commonmethod to produce SLBs.The transversal asymmetry in lipid composition is one of the

most challenging features of a biological membrane toreproduce in SLBs. In many eukaryotic cells sphingomyelin(SM) is enriched in the extracellular leaflet of the plasmamembrane, whereas phosphatidylethanolamine and negativecharged lipids, for example, phosphatidylserine (PS) arepredominantly located in the cytoplasmic leaflet.6 To date,the production of asymmetric SLBs (aSLBs) with a specificlipid composition in each leaflet has only been established forthe LB/LS deposition and for the hybrid LB/VF method.7,8 Infew cases, VF has been reported to spontaneously generateasymmetric SLBs depending on lipid mixture, VF conditionsand type of support used.9 For example, electrostatic repulsion/attraction between lipids and support was successfully used to

form SLBs containing asymmetrically distributed charged lipids.In particular, PS was preferentially found in the lower leaflet ofSLBs on titanium dioxide10 or in the upper leaflet of SLBs on asilicon block.11 However, the asymmetry in SLBs produced byVF can be only partially controlled, and a more generalapproach is still missing.A methyl-β-cyclodextrin (MβCD)-mediated lipid exchange

technique was recently devised to prepare small unilamellarvesicles (SUVs) with stable asymmetric lipid compositions12

and later tailored to produce asymmetric giant unilamellarvesicles (GUVs) at high yield.13 This method is based on theability of MβCD to bind phospholipids14 and to exchange themwith a preformed lipid bilayer. Briefly, in this method a lipidbilayer is exposed to a concentrated solution of MβCD loadedwith the desired lipid species. Only the accessible leaflet of thebilayer can directly exchange lipids with the MβCD complexes,that is, can be enriched with the lipid previously in complexwith the MβCD. This confined enrichment generatesasymmetry in the lipid composition of the bilayer. The abilityof MβCD to bind phospholipids or, more generally, lipophilic

Received: February 5, 2014Revised: May 8, 2014Published: June 2, 2014

Article

pubs.acs.org/Langmuir

© 2014 American Chemical Society 7475 dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−7484

compounds (e.g., cholesterol) derives from its chemicalstructure: cyclodextrins are in fact cyclic oligomers of glucose,in which the interior of the circle of glucose units forms anonpolar cavity.15 The approach of the MβCD-mediated lipidexchange method is in principle applicable to every model lipidbilayer, including SLBs.Individual lipids are not only asymmetrically distributed, but

also supposed to be laterally segregated in biologicalmembranes; together with specific proteins they form dynamicnanoscale assemblies also termed rafts, which are responsiblefor membrane subcompartmentalization and functions such astrafficking, endocytosis, and signaling.16 Lipid lateral organ-ization could be addressed in a variety of membrane modelsystems using so-called raft-mimicking lipid mixtures. Althoughtheir phase behavior was extensively investigated, the effect ofcompositional asymmetry on phase separation remains unclear.In a series of seminal articles, Tamm and coworkers7,17,18

investigated the coupling of cholesterol-rich liquid ordered(Lo) domains in asymmetric bilayers of different lipidcomposition, and showed for the first time that Lo domainsin one leaflet can induce phase separation in an opposing leafletwith lipid compositions that would not spontaneously phaseseparate. In the cases in which phase separation is not induced,domains are observed in at least one leaflet. At the same time,Collins and Keller19 showed that a leaflet with a lipidcomposition that does not phase separate can also suppressdomain formation in the leaflet with a lipid composition thatwould phase-separate in a symmetric bilayer. In theseexperiments, phase separation was observed either in bothleaflets or not at all. The results described in the above-mentioned works were obtained in two quite diverse modelmembrane systems (tethered polymer-SLBs and free-standingblack lipid membranes). A third independent approach isneeded to better understand the observed differences in thetwo systems. The biological relevance of studying lipid lateralorganization in asymmetric membrane systems was recentlyunderlined by the work of Hussain and colleagues; they coulddemonstrate that bilayer asymmetry influences the sequesteringof integrins in raft-mimicking lipid mixtures.20

Here we describe how the MβCD-mediated lipid exchangemethod can be applied to SLBs formed by VF to easily produce

flat bilayers with asymmetric lipid composition. Specifically weselectively enriched the upper leaflet of a dioleoylphosphati-dylcholine SLB with SM, incubating the bilayer with MβCD−SM complexes (Figure 1). SM incorporation is known to affectmembrane fluidity.13 Therefore, to assess the asymmetric SMenrichment in the bilayer, we labeled each SLB leaflet with adifferent fluorescent lipid probe and measured their diffusion byfluorescence correlation spectroscopy (FCS). To selectivelylabel each leaflet, the nitrobenzoxadiazole (NBD) moiety of thelipids in the upper leaflet was quenched with sodium dithioniteand lipids conjugated with a spectrally separated dye weresubsequently incorporated in it from the above buffer. Selectivelabeling of each leaflet allows us to probe both leafletsindependently and measure local lipid dynamics. We finallyshow that the produced asymmetric SLBs are stable for at least4 h, fully compatible with the most common proteinreconstitution methods and suitable for studying phaseseparation and transbilayer lipid movement of raft-mimickinglipid mixtures.

■ EXPERIMENTAL SECTIONMaterials. 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC),

cholesterol from ovine wool (Chol), sphingomyelin from porcinebrain extract (bSM) and 1,2-dioleoyl-sn-glycero-3-phosphoethanol-amine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD−DOPE) were pur-chased from Avanti Polar Lipids, Inc. (Alabaster, AL). Alexa Fluor 647carboxylic acid, succinimidyl ester was purchased from LifeTechnologies Corporation (Carlsbad, CA). 3-[(3-Cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 4-nitrophenylphosphate disodium salt hexahydrate (pNPP), alkaline phosphatasefrom human placenta (PLAP), methyl-β-cyclodextrin (MβCD), and(2-hydroxypropyl)-α-cyclodextrin (HPαCD) were purchased fromSigma-Aldrich (St. Louis, MO). Atto647N-acyl chain-labeled sphingo-myelin (Atto647N-SM) was was a gift of E. Sezgin (University ofOxford, UK).

SLBs Formation and Reaction Chamber Assembly. SLBs wereprepared on freshly cleaved mica using the VF method.5 SUVs wereobtained by sonication in an ultrasonic bath (Bransonic UltrasonicCleaners model 2510, Branson Ultrasonics, Danbury, CT). DOPC,SM, and NBD−DOPE were mixed in chloroform at different molarratios; NBD−DOPE concentration was kept at 0.5 %mol for imagingand 0.05 %mol for FCS measurement. After solvent evaporation, thelipid film was rehydrated in SLB buffer (10 mM HEPES pH 7.4, 150

Figure 1. A schematic diagram of aSLB formation via MβCD-mediated lipid exchange and of the leaflet-specific labeling strategy used toquantitatively assess it.

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847476

mM NaCl) at 4 mg/mL lipid concentration and resuspended byvortexing. After sonication, 10 μL of the suspension were diluted inSLB buffer to ca. 0.5 mg/mL final lipid concentration and depositedon freshly cleaved mica, glued on a glass coverslip and sealed attachinga plastic cylinder of 7 mm diameter around it. Adding CaCl2 to a finalconcentration of 3 mM induced SUVs fusion and the formation of alipid bilayer on mica. After 30 min incubation, the sample wasvigorously rinsed with SLB buffer to remove any unfused SUVs. Forlipid mixtures containing SM, SUVs and SLB formation wereperformed at 65 °C; samples were slowly cooled down to roomtemperature (RT) before further usage. SLBs were imaged andfluorescence recovery after photobleaching (FRAP) was recorded toassess the correct formation of the bilayer.PLAP Purification and Labeling. Commercially available PLAP

was further purified and labeled with Alexa Fluor 647 as previouslydescribed21−23 with minor modifications. In particular, Triton X-114was precondensed before usage24 and Superdex 200 10/300 GL (GEHealthcare) columns were used to both remove the detergent andseparate the labeled protein from the free dye. Labeled-PLAP purityand concentration were assessed by SDS-PAGE and absorbance at 280nm, respectively. The degree of labeling was approximately one dyemolecule per PLAP monomer.PLAP Reconstitution in SLBs. PLAP was reconstituted in SLBs

either via a direct protein reconstitution method25 or fromproteoliposomes. In the direct reconstitution method, DOPC SLBslabeled with 0.05 %mol NBD−DOPE formed as described above wereincubated with PLAP at 2 μg/mL final concentration in the presenceof 0.22 mM CHAPS for 2 h at RT or overnight at 4 °C. After PLAPincorporation, SLBs were washed extensively with SLB buffer (FigureS4, Supporting Information (SI)). No preincubation of the membranewith detergent was needed for the incorporation of PLAP into theSLB. In the proteoliposomes method, PLAP was reconstituted intoDOPC large unilamellar vesicles (LUVs) as described earlier.23 PLAP-containing LUVs were incubated 30 min with 0.3 μM NBD−DOPEand used similarly to SUVs to form SLBs as described above. Theincorporation and mobility of the reconstituted PLAP in both methodswas confirmed by confocal microscopy imaging and FRAP; the activityassessed by pNPP assay in the solution above the SLBs. Briefly, theSLB buffer was changed to glycine buffer (Glycine 0.1 M pH 10.4, 1mM MgCl2, 1 mM ZnCl2) and pNPP at 1 mg/mL final concentrationwas added. SLBs were then incubated 1 h at RT. The reaction wasstopped with 3 N NaOH and the absorbance read at 405 nm.Leaflet-Specific Labeling. NBD−DOPE was quenched with 10−

100 ng/mL sodium dithionite. The total number of NBD−DOPEparticles in the laser focus was monitored by point FCS. Aftertreatment with sodium dithionite, particle number decreased of afactor > 2. In these conditions, we can safely assume that almost thetotality of the NBD−DOPE in the upper leaflet (and probably some inthe lower leaflet as well) was quenched. To label the upper leaflet ofthe SLB, Atto647N-SM at 5 pg/μL (for FCS measurement) or 250pg/μL (for imaging) was next added to the buffer above the SLB. Thesamples were then mixed properly and incubated for 2 min beforeextensive washing. When a concentration of 5 pg/μL was used, thenumber of incorporated Atto647N-SM particles was suitable for FCSmeasurement (1−20 particles in average in the focal volume). Whenfluorescently labeled PLAP was reconstituted in SLB, no additionalprobe for the upper leaflet was added. (Figure 1 and SI forexperimental details)Cyclodextrin−Lipid Complexes Preparation. MβCD−SM and

MβCD−Chol complexes were prepared as described elsewhere13 withminor modifications. To prepare MβCD−SM complexes, SMmultilamellar vesicles (MLVs) and MβCD were mixed together at71 and 16 mM final concentration, respectively. The mixture was nextincubated with moderate shaking for 2 h at 65 °C and centrifuged at54 000 g for 15 min at 4 °C. The supernatant contains the MβCD−SM complexes and is deprived of any residual MLVs. To prepareMβCD−Chol complexes, 80 μL of cholesterol 15 mM in isopropanolwere slowly added to 2.2 mL of MβCD 60 mM in H2O preheated in awater bath at 80 °C. The mixture was then incubated with moderateshaking for 1 h at 60 °C and centrifuged at 40 000 g for 15 min at 4

°C. Incubation and centrifugation temperatures were chosen tooptimize the MβCD complexes preparation: a high incubationtemperature allows the binding reaction to faster reach equilibrium,whereas a low centrifugation temperature enhances the pelleting ofany residual SM MLVs. The supernatant was filtered with 0.22 μmfilters. The osmolarity of both MβCD−SM and MβCD−Cholcomplexes was measured (Micro-Sample Osmometer model 210,Advanced Instruments, Norwood, MA). The nominal MβCDconcentration was assumed to still be 71 mM and 60 mM in theMβCD−SM and MβCD−Chol complexes, respectively. This is basedon the assumption that no MβCD is lost during the incubation withthe SM MLVs or cholesterol and in the following centrifugation runs.MβCD−SM and MβCD−Chol complexes were stored at −20 °C untilneeded. MβCD−DOPC and HDαCD−SM/Atto655-DOPE wereprepared as described for MβCD−SM, using DOPC instead of SMor HDαCD instead of MβCD. For HDαCD−SM/Atto655−DOPEcomplexes, 0.05 %mol Atto655−DOPE was added to SM beforeforming MLVs.

MβCD-Mediated Lipid Exchange. SLBs were incubated withdifferent concentrations of MβCD−SM (5−20 mM) or MβCD−Chol(1−10 mM) complexes at RT for 20 and 15 min, respectively. Afterincubation, samples were delicately washed with SLB buffer.Prolonging the SLB incubation with MβCD−SM or MβCD−Cholcomplexes up to 1 h did not influence dramatically the final results.Note that the osmolarity of the MβCD−SM and MβCD−Cholsolutions used to treat the bilayers were matched to the osmolarity ofthe SLB buffer in which the SLBs were formed.

Confocal Fluorescence Microscopy and FluorescenceCorrelation Spectroscopy. Fluorescence confocal imaging andpoint-FCS measurements were performed on LSM 780 ConfoCor3system (Carl Zeiss, Jena, Germany) using a C-Apochromat 40x/1.2 WCorr M27 objective. For both imaging and FCS, the sample wasexcited either by a 488 nm (green channel, e.g., NBD) or a 633 nm(red channel, e.g., Atto647N or Alexa Fluor 647) lasers. Thefluorescence was then collected through a 505−575 nm band-passfilter or a 655 nm long-pass filter with a 40 μm pinhole. For FCSmeasurements, usually 5−9 independent SLBs preparations wereanalyzed. In each preparation, the SLB was typically measured in 6−8randomly chosen membrane spots and for each of them, 10independent fluorescence intensity time tracks of 10 seconds wereacquired. The FCS correlation curves were calculated using the ZeissZEN 2011 Black edition software and fitted using a nonlinear least-squares algorithm to a single-component 2D diffusion model. Typicalnormalized average correlation curves and fitting curves are shown forboth upper and lower leaflet probes in a symmetric and asymmetricSLB (Figure S5, SI). Diffusion times (τD) of both probes at each SLBspot were calculated and averaged for all spots in the same preparation.In order to translate τDs into diffusion coefficients (D), the system wascalibrated daily using a symmetric DOPC SLB with the sameconfiguration of fluorescent probes as used in the asymmetric SLBs.This approach accounts for changes in the confocal volume and allowsus to calculate a relative D (D*), that is, the diffusion coefficient of aprobe in an asymmetric SLB relative to the diffusion coefficient of thesame probe in a symmetric DOPC SLB. D* can be easily calculatedfrom the ratio between the τD of the probe in the symmetric DOPCSLB used for calibration and the τD of the probe in the asymmetricSLB sample.

Image Processing and analysis. Fluorescence confocal imageswere processed and analyzed with Fiji (http://fiji.sc/Fiji). Whenneeded, image contrast was enhanced through normalization. Numberand average size of Lo domains were measured in locally thresholdedimages.

Statistical Analysis. Statistical analysis was performed usingSigmaPlot 12.3 (Systat Software, Inc., San Jose, CA). When the meansof two sets of data were compared (Figures 2B and 4), a t-test or apaired t-test was used for independent or before-and-after samples,respectively. For both t-tests two-tailed P-values are reported.Comparison of more than two sets of data (Figures 2A and 3),were carried out with one-way ANOVA or one-way repeated measuresANOVA for independent samples or samples measured over time,

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847477

respectively. To isolate the group or groups that differ from the others,ANOVA was followed up with an all-pairwise multiple comparisonprocedure (Tukey Test). The level of alpha was kept at 0.05 for allstatistical analyses.

■ RESULTS AND DISCUSSION

The Upper Leaflet of DOPC SLBs Can Be Enrichedwith SM to Form Asymmetric Bilayers. Symmetric DOPCSLBs were prepared on freshly cleaved mica using the VFmethod.5 Mica supports typically had an area of 38 mm2 andwere kept hydrated with 200 μL of SLB buffer at all times afterSLB formation. Assuming the lipid packing constant, the totalamount of lipids in SLB samples (ca. 10−10 mol) is two ordersof magnitude lower than that typically found in vesiclespreparations (e.g., ca. 10−8 mol for GUV samples prepared onindium tin oxide glass), for which high concentrations ofMβCD and SM in solution can be used to produce asymmetricbilayers.12,13 High MβCD−SM:lipid ratios could lead toextensive extraction and solubilization of lipids from the SLB.To determine whether and at which concentration MβCDaffects the integrity of the bilayer, several SLBs labeled with0.05 %mol NBD−DOPE were treated with different concen-trations of MβCD−SM complexes (Figure S1, SI). Treatmentof SLBs with high concentrations of MβCD−SM (20 mM),comparable to those normally used for the preparation ofasymmetric free-standing bilayers, results in extensive damageand eventually complete solubilization of the membrane. Time-lapse confocal imaging shows that dark areas form and enlargein the SLBs. The dark areas are indeed discontinuities of theSLB, as suggested by the absence of NBD−DOPE andconfirmed by the addition of a highly hydrophilic fluorescentprotein with no membrane affinity (i.e. enhanced GFP) to thebuffer above the SLB. The protein colocalizes completely withthe dark areas, confirming that in those regions the mica is notcovered by a bilayer, but exposed to the buffer (data notshown). In contrast, at lower MβCD−SM concentrations (5−16 mM) SLBs showed less or no damage at all, allowing thetreatment of SLBs with MβCD−SM.

Thus, MβCD−SM concentration needs to be low enough topreserve SLBs integrity, but sufficient to enrich the upper leafletof SLBs with SM. The SM enrichment of a membrane leaflet isgenerally expected to reduce membrane fluidity, due to theincreased lipid order.12,13 FCS has been proven to be a veryuseful tool to characterize the fluidity of lipid bilayers; inparticular, the D of fluorescently labeled lipids measured byFCS have been extensively used to evaluate membraneorganization.26 In case of successful enrichment of a membraneleaflet with SM, the D* of a fluorescent probe included thereinshould decrease. We therefore performed FCS on severalDOPC SLBs labeled with Atto647N-SM in the upper leafletbefore and after treatment with different MβCD−SMconcentrations (10−16 mM) to identify the MβCD−SMconcentration needed to minimize the Atto647N−SM D*, thatis, to maximize the SM enrichment (Figure 2A). An MβCD−SM concentration of 12 mM is sufficient to decrease the D* ofthe upper probe to the lowest value obtained in the MβCD−SM concentration range tested, while keeping the damage ofthe SLBs to a minimum. When a nominal MβCD−SMconcentration of 12 mM is used, the MβCD−SM:lipid ratio canbe estimated to be approximately 25000:1.Having optimized the MβCD−SM concentration, we then

asked whether the enrichment in SM after MβCD−SMtreatment is confined to the upper leaflet of the SLB, inother words whether an asymmetric SLB is formed. For thispurpose, several symmetric DOPC SLBs were labeled withNBD−DOPE and Atto647N-SM in the lower and upper leaflet,respectively. We then measured the mobility of the differentprobes by point-FCS before and after treatment with 12 mMMβCD−SM (Figure 2B). The D* of Atto647N-SM decreasesto about 66% of the original value, as expected in case of SMincorporation. From the calibration curve (Figure S2, SI), theSM incorporation can be estimated to be 50−70 %mol of thetotal lipids in the upper leaflet of the membrane. In contrast,the D* value of the NBD−DOPE exhibits a very small decrease(ca. 10%), probably due to limited interleaflet coupling.27 Theobserved decrease in NBD−DOPE diffusion might alternativelybe due to the incorporation of small amounts of SM in the

Figure 2. aSLB formation via MβCD-mediated lipid exchange method. [A] Upper leaflet D*s are plotted before and after incubation with differentconcentrations of MβCD−SM (10−16 mM). Concentrations equal or above 12 mM minimize the upper leaflet D* and make it significantly lowerthan that resulting after treatment with 10 mMMβCD−SM (***p < 0.001). [B] Lower and upper leaflet D*s are plotted before and after incubationwith 12 mM MβCD−SM. Upper leaflet D* significantly decreases to about 67% of the original value (****p < 0.0001), whereas the upper leaflet D*exhibits a meaningful but minor reduction of ca. 10% (**p < 0.01), suggesting that the incorporation of the SM was mostly limited to the upperleaflet and the formation of asymmetric SLBs was achieved.

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847478

lower leaflet during MβCD−SM lipid exchange, although thispossibility seems less likely (SI Figure S4, Ref13.). Altogether,these data suggest that the incorporation of the SM was mostlylimited to the upper leaflet and that the formation of anasymmetric bilayer was achieved. Moreover, it is worth notingthat the influence of the support on the lower leaflet is notsubstantially stronger than that exerted on the upper leaflet, asconfirmed by unsuccessful attempts to measure differences inthe diffusion rate of lipid dyes in each leaflet28 (data notshown). Finally the slower NBD−DOPE dynamics after lipidexchange in the bottom leaflet is not a result of nanoscopicdefects introduced in the bilayer through the MβCD treatment,as suggested by control experiments in which the SLBs weretreated with 12 mM MβCD−DOPC (Figure S3, SI).Leaflet Asymmetry Is Stable for Several Hours.

Experimentally useful aSLBs need to be stable for a reasonableamount of time required to run the desired experiments. Thescrambling of the SLB is supposed to alter the local lipidpacking, and consequently, the diffusion coefficient of the lipidsin each leaflet. As SM flips from the upper to the lower leafletduring equilibration, the Atto647N−SM D* should increase upto the value estimated for symmetric SLB containing 30 %molSM (ca. 80%, Figure S2, SI). Also, the NBD−DOPE D* shoulddecrease until the ratio between lower and upper D is ca. 1 (i.e.,fully scrambled lipids and fluorescent lipid probes). Todetermine how stable aSLBs produced via MβCD-mediatedlipid exchange are, the D* of NBD−DOPE and Atto647N-SMwere measured by point-FCS at different times after MβCD−SM treatment in randomly chosen positions (Figure 3). The

lower/upper leaflet D* ratio does not decrease significantlyuntil at least 4 h after the MβCD-mediated lipid exchange,suggesting that the asymmetry in the SLB is retained for severalhours. However, after 1 day, the asymmetry seems to be lost. Asimilar stability in time was reported for asymmetric POPCbilayers prepared by the LB/VF method.7 Moreover, the timescale of SM flip-flop appears comparable to the previouslyreported DSPC rates.29,30 The major constituent of the usedSM is in fact 18:0 SM (Product specification, http://www.avantilipids.com), and its flip-flop time scale at RT can beestimated to be of the order of hundreds of minutes. Thus, due

to their stability in time, aSLB can be used experimentally asasymmetric membrane models.

The Asymmetry in a SLB Can Also Be Generated inthe Presence of a Reconstituted Membrane Protein. Toassess whether the MβCD-mediated lipid exchange method canbe used to produce asymmetric bilayers containing membraneproteins, the PLAP was reconstituted in SLBs, both via directprotein reconstitution,25 and from proteoliposomes.22,23

In the direct reconstitution method, a symmetric DOPC SLBwas exposed to CHAPS detergent to promote proteinincorporation. In the proteoliposomes method, the PLAP-containing SLBs were prepared using the same VF methodapplied for DOPC SLBs. In both methods, no dye for the upperleaflet was added, since the reconstituted PLAP was previouslyfluorescently labeled allowing its N-terminal amino group toreact with the succinimidyl ester moiety of the Alexa Fluor 647dye; thus labeled-PLAP might be used as a probe for the upperleaflet of the SLB. It is worth noting that only in the case ofdirectly reconstituted PLAP, the protein is expected to beconfined to the upper leaflet.31 In contrast, the reconstitutedPLAP from proteoliposomes is expected to be present also inthe lower leaflet, since the orientation of the protein might berandomized, for example, during proteoliposome fusion andSLB formation. Accordingly, we found that the τD of the PLAPmeasured in symmetric SLB is significantly lower in samples inwhich the PLAP was directly reconstituted than in thoseprepared from proteoliposomes (Figure 4A). The lowerapparent diffusion rate (i.e. higher τD) of proteoliposome-reconstituted proteins in SLB can be explained as aconsequence of the interaction between the large ectodomainof the protein in the lower leaflet and the support,2 in contrastto the limited influence of the support on the mobility of lipidsin the same leaflet.The protein-containing SLBs were finally treated with 12

mM MβCD−SM to produce asymmetric bilayers. Similarly toprotein-free SLBs, the SM enrichment of the upper leafletresults in an increased lipid order, and consequently, a lowerlipid mobility that can be probed at the membrane by pointFCS. If SM incorporation is restricted to the upper leaflet of theSLB, only the D* of the upper probe will experience a severedecrease. The ratio between the D* values of the lower andupper probes will therefore reach values higher than 1. Figure4B confirms the formation of PLAP-containing aSLBs showinghow the lower/upper D* ratio increases after MβCD-mediatedlipid exchange. Interestingly, the D* of the reconstituted PLAPdecreases in both cases after MβCD-mediated lipid exchange(direct reconstitution: D* = 58 ± 6%; proteoliposomes: D* =78 ± 10%, relative to symmetric SLBs with the sameconfiguration of fluorescent probes). The decrease is lower inthe PLAP reconstituted from liposomes, probably as aconsequence of the presence of a significant protein fractionin the lower leaflet. The PLAP in the lower leaflet is notaffected by SM incorporation and its D* should only marginallydecrease, similarly to the D* of the lower leaflet probe (D* =86 ± 7%). The overall PLAP D* (average between upper andlower leaflet) is therefore higher than that measured for thedirect reconstitution method, in which the PLAP is confined tothe upper leaflet. The D* of the lower leaflet probe exhibits asmaller decrease in both cases (direct reconstitution: D* = 87± 16%; proteoliposomes: D* = 86 ± 7%) confirming theasymmetry in the protein-containing SLBs.Altogether, these results show that the MβCD-mediated lipid

exchange method is fully compatible with both the

Figure 3. aSLB stability over time. Lower/upper leaflet D* ratios areplotted before and at different times after incubation with 12 mMMβCD−SM. Before and 22 h after treatment D* ratios aresignificantly lower than those at 0, 2, and 4 h after treatment (***p< 0.001).

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847479

proteoliposomes and the direct reconstitution methods. Usedin conjunction with one of the protein reconstitution methods,aSLBs can provide a more physiological protein environment.By making accessible in the same sample the state before andafter lipid enrichment (i.e. symmetric and asymmetric state),aSLBs are also a handy tool to study how compositional lipidasymmetry can influence protein structure and function. It isnoteworthy that although the proteoliposome reconstitution isthe method of choice for many large transmembrane proteins,32

its application in combination with aSLBs has a limited value.The protein is in fact inserted in in both directions andtherefore interacts with each leaflet differently.SM in the Lower Leaflet of an aSLB is Required for

Phase Separation in Both Leaflets. aSLBs produced asdescribed before and containing up to 60 %mol SM in theupper leaflet do not show macroscopic liquid/gel coexistingdomains, as reported for the corresponding symmetricSLBs.33,34 As already proposed for asymmetric GUVs,27 theupper leaflet of aSLBs might either be in a homogeneous liquidphase or contain submicroscopic gel domains. To address thequestion of whether and how compositional asymmetry affectsphase separation, we prepared SLBs and aSLBs with the sameoverall lipid composition and compared their behavior afteraddition of cholesterol. In fact, although binary mixtures oflipids were crucial to determine the connection between phasebehavior and multiple variables such as lipid composition andtemperature, ternary lipid mixtures containing cholesterol showa richer phase behavior and can better model the animal cellplasma membranes.35

In particular, symmetric DOPC SLBs containing up to 60 %mol of SM and aSLBs containing approximately 60 %mol ofSM in the upper leaflet were produced as described before and

exposed to 5 mM MβCD−Chol complexes. Similarly toMβCD−SM complexes, high concentrations of MβCD−Chol(10 mM) result in a complete destruction of the bilayer (FigureS6G, SI). A concentration of 5 mM was chosen because it wasthe lowest concentration at which phase separation could stillbe induced in symmetric bilayers containing less than 50 %molSM. When aSLBs were treated with 5 mM MβCD−Cholcomplexes, the incorporation of cholesterol in the lower leafletcould be estimated to be up to 20 %mol of the total lipidcontent (Figure S7, SI). This value correlates well with theobserved phase behavior in symmetric and partially asymmetricSLBs, according to a recently improved experimentally-derivedphase diagram of the same mixture.33 Moreover, in absence ofopposing evidence, we assume that the cholesterol equilibratesfast and distributes uniformly throughout the leaflets of theaSLB.36

Only in symmetric SLBs containing more than 30 %mol ofSM and in scrambled aSLBs (aSLBs after 22 h incubationtime), the incorporation of cholesterol results in the formationof Lo domains registered in both lower and upper leaflets(Figure 5, A and C; Figure S6, A−G, SI). Interestingly, in thesymmetric SLBs, preexisting gel domains33 disappeared soonafter cholesterol addition and Lo domains appeared both in thesame and in new positions (Figure 5, A1). Same concentrationsof cholesterol-loaded MβCD failed to induce phase separationin freshly prepared aSLBs, in which SM is still confined to theupper leaflet (Figure 5B).The transition from aSLBs to scrambled aSLB and the

appearance of phase separation can be followed in time (Figure6A). Time-lapse fluorescence imaging allows to record the timeof appearance of Lo domains in partially scrambled aSLBs.Within a single bilayer, Lo domains appear simultaneously all

Figure 4. aSLB formation in the presence of reconstituted PLAP. [A] Labeled-PLAP τD, and [B] lower/upper leaflet D* ratios before and afterincubation with 12 mM MβCD−SM are plotted for both reconstitution methods. The τD of directly reconstituted PLAP is significantly lower (*p <0.05) than the τD of PLAP prepared from proteoliposomes. The lower/upper D* ratio increases in both reconstitution methods after MβCD-mediated lipid exchange (***p < 0.001) and it is significantly lower in the proteoliposomes samples (°°p < 0.01), most probably due to the fractionof PLAP present in the lower leaflet (i.e., not affected by the lipid exchange).

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847480

over the sample, whereas different samples show diverse timing(t1/2 = 372 ± 129 min); this is probably due to experimentalvariation in the SM and cholesterol content of each bilayer.However, once the first domains appear, the time to reach thefinal number of domains is comparable in all measured samples(tf‑0 = 262 ± 35 min). Domains keep growing after the finalnumber of domains is reached, suggesting that at this point theyare not in equilibrium yet (Figure 6, B). As shown before,aSLBs slowly loose their asymmetry through spontaneous flip-flop of lipids causing a gradual and inversely correlated changeof the SM amount in each leaflet. Lo domains appearance ismost probably due to this SM redistribution across the bilayerand thus to lipid flip-flop in general. The onset of the domainappearance can be then used as an indirect measure of lipid flip-flop rate.

As a consequence of lipid flip-flop in aSLBs, the localconcentration of SM in the lower leaflet increases whiledecreasing in the upper one. This change in transversal lipidcomposition in turn induces phase separation. In order tounderstand whether the increase of SM concentration in thelower leaflet (rather than its decrease in the upper leaflet) isdeterminant in promoting phase separation, we producedaSLBs containing 5−10 %mol SM and ca. 60 %mol SM in thelower and upper leaflet, respectively. To do so, we first formedsymmetric DOPC SLBs containing 5−10 %mol SM; thisamount of SM is not sufficient to induce phase separation uponcholesterol incorporation (Figure S6, A−B, SI). We thentreated the same symmetric SLBs with MβCD−SM complexesto obtain a partially aSLB. We assume that these aSLBs have alipid composition of the upper leaflet similar to those discussed

Figure 5. Phase separation and coupling in symmetric [A], scrambled [C], and fully [B] or partially [D] asymmetric SLB after cholesterol addition.[A1] The kymograph shows domain dynamics in symmetric SLBs at the overlaying line depicted in A.

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847481

in the previous paragraphs, that is, ca. 60 %mol SM. Uponcholesterol addition, phase separation could be induced inpartially aSLB containing as low as 5 %mol of SM in the lowerleaflet. This observation reinforces the hypothesis that SM isrequired in the lower leaflet for phase separation in both leaflets(Figure 5D). This is in line with the phase diagram forsymmetric bilayers with analogous composition.33 Dependingon the amount of cholesterol, phase separation requires in fact aminimum amount of SM (ca. 10−30 %mol) and persists up to80−90 %mol SM. Furthermore, it is interesting to notice thatSM, although more abundant in the extracellular leaflet, is alsopresent in the cytoplasmic leaflet of natural membranes inmany cellular types at about 1:10 ratio.37

The observed differences in terms of phase separation insymmetric or aSLBs provide further experimental evidence thatthe transversal lipid distribution affect the overall lipidmiscibility. In accordance with Collins and Keller,19 the lipidcomposition of one leaflet strongly influences the phasebehavior of the other leaflet, suppressing or priming phaseseparation; in particular, our results suggest that a minimalamount of SM is required in the nonspontaneously phaseseparating leaflet for cholesterol to induce phase separation inboth leaflets. As also pointed out in the above-mentioned study,the presence of one leaflet with a composition supporting phaseseparation (in a corresponding symmetric bilayer) is necessarybut not sufficient to induce domain formation in the wholebilayer. Of interest, it is well-known that lipid mixturesmimicking the outer leaflet of the plasma membrane can giverise to raft-like ordered domains. On the other hand, lipidmixtures corresponding to typical inner leaflet compositionsform membranes that are homogeneous.38 Our results addressthis incongruence, suggesting that small changes in thecomposition of the inner leaflet (e.g., even low concentrations

of saturated lipids such as SM, not enough to produce phaseseparation in a corresponding symmetric bilayer) can have aprominent role in the formation of lipid domains spanning thewhole bilayers.Unlike Wan and colleagues,18 we were not able to induce

formation of macroscopic Lo domains in a single SM-containing leaflet of an aSLB. This difference can be explainedtaking into account the individual properties of the two usedsystems. In the LB/VT approach, domains in the LB leaflet arealready present in the original monolayer, which is transferredto a support and stabilized by a tethered polymer cushion. Inour system domains form de novo after addition of cholesterolto the bilayer. These systems produce only apparentlyconflicting results. It might be possible in fact that the LBleaflet resides in a local energy minimum, and the addition ofthe upper layer might not be enough to perturb its state (i.e.,the bilayer system might be kinetically trapped in a non-equilibrium state). The composition of our aSLBs is verysimilar, although not identical to one of the few compositionstested by Wan and colleagues for which no phase separationcould be induced in the upper leaflet (in the presence ofdomains in the lower leaflet). We therefore speculate that thelipid mixtures that in the LB/VT asymmetric system do notshow domain induction from the lower to the upper leaflet arethe same that in our system would suppress phase separation inboth leaflets. The presence of phase separation in the LB leafletmight be due to the specific leaflet-by-leaflet assemblymethodology.Similar considerations might be extended to the LB/LS

system of Garg and colleagues,39 in which both lower andupper leaflets are preformed as monolayers (LB and LS,respectively) with their respectively energetically most favorablephase states and only subsequently assembled on top of eachother.Having observed a different phase behavior in our system, we

decided to investigate asymmetric free-standing bilayers withanalogous lipid compositions as alternative asymmetric bilayermodels. By using asymmetric GUVs, we were not only able tocompletely reproduce the domain registration observed onSLBs (Figure S8, SI), but also to rule out that the domainregistration properties observed in our system are influenced bythe presence of the support.

■ CONCLUSIONSWe have described how the MβCD-mediated lipid exchangemethod can be extended to SLBs formed by VF to producetemporarily stable asymmetric SLBs. This approach is quickand easy to implement, does not require organic solvents, norspecialized equipment or skills, and can be used with a varietyof different lipids to produce bilayers with the desiredcomposition in each leaflet, as already showed for LUVs andGUVs.12,13,40,41 It also provides a more naturally generatedasymmetric system in the way that specific lipids areincorporated when the bilayer already exists. The local lipidcomposition of biological membranes changes continuously,because of membrane recycling and enzymatic activity.35 This isthe case, for example, with SM, being synthesized at theendoplasmic reticulum and Golgi apparatus and eventuallytransported to the plasma membrane, or directly at the outerleaflet of the plasma membrane by the sphingomyelin synthase(SMS2). We also demonstrated that the MβCD-mediated lipidexchange can produce asymmetric bilayers in the presence ofreconstituted proteins. Interestingly, this method offers the

Figure 6. Phase separation dynamics in aSLB. [A] Domain appearancein aSLBs treated with 5 mM MβCD−Chol. [B] Domains number andaverage size were measured and plotted against time.

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847482

advantage of providing in the same sample the state before andafter the establishment of the asymmetry, thus opening thepossibility of directly probing the effect of leaflet-specific lipidcomposition on protein structure and function.In addition, we have shown that aSLBs generated via MβCD-

mediated lipid exchange are a suitable membrane model systemto study phase separation of raft-mimicking lipid mixtures.Using such a system, we showed how small changes in thecomposition of the inner leaflet (e.g., of the plasma membrane),although not sufficient to support phase separation in acorresponding symmetric bilayer, affect the overall lipidmiscibility of the whole bilayer. Our system provides analternative strategy to follow domain dynamics withoutheating/cooling the sample above/below its phase transitiontemperature. In fact, de novo formation of Lo domains isinduced though cholesterol incorporation in the bilayer ratherthan by temperature control. Moreover, in contrast to what wasdescribed for other aSLB systems39 the support does not exertany negative influence on domain registration in our system;when domains can be formed, they are always in registration. Ifthe total cholesterol amount of a single bilayer could beprecisely measured, for example, with mass spectrometry, oursystem could be also used to investigate how cholesteroldistributes between the lower and upper leaflet of SM-containing aSLBs. In fact, even if cholesterol is expected topreferentially interact with SM, experiments performed invarious cells types revealed that cholesterol is enriched in thecytosolic leaflet of biological membranes.42 For example recentquantitative quenching analysis in CHO cells indicated that60−70% of the plasma membrane sterols resides in thecytoplasmic leaflet.43

Finally, our approach has the potential to be combined withother conventional SLB preparation techniques to fine-tune thelipid composition of the bilayer, as suggested by the successfulvesicle-mediated phospholipids exchange recently reported fortethered bilayers.44 The above-mentioned features illustratehow asymmetric SLBs produced via MβCD-mediated lipidexchange can be a flexible and more precise model of thecellular plasma membrane in diverse experimental setups, forexample, those involving the use of surface-sensitive techniques.

■ ASSOCIATED CONTENT

*S Supporting InformationExperimental details, detailed figure captions and supportingfigures (S1−S8). This material is available free of charge via theInternet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: schwille@biochem.mpg.de.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

We thank Dr. Erdinc Sezgin for providing the Atto647N-SM,Dr. Henri G. Franquelim for fruitful discussions, Dr. SabineSuppmann and Elisabeth Weyher-Stingl of the MPI-B corefacility for helping with the PLAP protein purification and forthe preliminary MS measurements. S.C. was supported by theDeutsche Forschungsgemeinschaft (DFG) through the grantCH 1238/3-1.

■ REFERENCES(1) Sackmann, E. Supported membranes: Scientific and practicalapplications. Science 1996, 271, 43−48.(2) Wagner, M. L.; Tamm, L. K. Tethered polymer-supported planarlipid bilayers for reconstitution of integral membrane proteins: Silane-polyethyleneglycol-lipid as a cushion and covalent linker. Biophys. J.2000, 79, 1400−1414.(3) Shen, H.-H.; Lithgow, T.; Martin, L. Reconstitution of membraneproteins into model membranes: Seeking better ways to retain proteinactivities. Int. J. Mol. Sci. 2013, 14, 1589−1607.(4) Sanderson, J. M. Resolving the kinetics of lipid, protein andpeptide diffusion in membranes. Mol. Membr. Biol. 2012, 29, 118−143.(5) Brian, A. A.; McConnell, H. M. Allogeneic stimulation ofcytotoxic T cells by supported planar membranes. Proc. Natl. Acad. Sci.U.S.A. 1984, 81, 6159−6163.(6) Raggers, R. J.; Pomorski, T.; Holthuis, J. C.; Kalin, N.; van Meer,G. Lipid traffic: The ABC of transbilayer movement. Traffic 2000, 1,226−234.(7) Crane, J. M.; Kiessling, V.; Tamm, L. K. Measuring lipidasymmetry in planar supported bilayers by fluorescence interferencecontrast microscopy. Langmuir 2005, 21, 1377−1388.(8) Fabre, R. M.; Talham, D. R. Stable supported lipid bilayers onzirconium phosphonate surfaces. Langmuir 2009, 25, 12644−12652.(9) Wacklin, H. P. Composition and asymmetry in supportedmembranes formed by vesicle fusion. Langmuir 2011, 27, 7698−7707.(10) Rossetti, F. F.; Textor, M.; Reviakine, I. Asymmetric distributionof phosphatidyl serine in supported phospholipid bilayers on titaniumdioxide. Langmuir 2006, 22, 3467−3473.(11) Stanglmaier, S.; Hertrich, S.; Fritz, K.; Moulin, J. F.; Haese-Seiller, M.; Radler, J. O.; Nickel, B. Asymmetric distribution of anionicphospholipids in supported lipid bilayers. Langmuir 2012, 28, 10818−10821.(12) Cheng, H.-T.; Megha; London, E. Preparation and properties ofasymmetric vesicles that mimic cell membranes: Effect upon lipid raftformation and transmembrane helix orientation. J. Biol. Chem. 2009,284, 6079−6092.(13) Chiantia, S.; Schwille, P.; Klymchenko, A. S.; London, E.Asymmetric GUVs prepared by MβCD-mediated lipid exchange: AnFCS study. Biophys. J. 2011, 100, L1−L3.(14) Anderson, T. G.; Tan, A. M.; Ganz, P.; Seelig, J. Calorimetricmeasurement of phospholipid interaction with methyl-[beta]-cyclo-dextrin. Biophys. J. 2004, 86, 174A−174A.(15) Pitha, J.; Irie, T.; Sklar, P. B.; Nye, J. S. Drug solubilizers to aidpharmacologists: Amorphous cyclodextrin derivatives. Life Sci. 1988,43, 493−502.(16) Simons, K.; Sampaio, J. L. Membrane organization and lipidrafts. Cold Spring Harbor Perspect. Biol. 2011, 3, a004697.(17) Kiessling, V.; Crane, J. M.; Tamm, L. K. Transbilayer effects ofraft-like lipid domains in asymmetric planar bilayers measured bysingle molecule tracking. Biophys. J. 2006, 91, 3313−3326.(18) Wan, C.; Kiessling, V.; Tamm, L. K. Coupling of cholesterol-richlipid phases in asymmetric bilayers. Biochemistry 2008, 47, 2190−2198.(19) Collins, M. D.; Keller, S. L. Tuning lipid mixtures to induce orsuppress domain formation across leaflets of unsupported asymmetricbilayers. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 124−128.(20) Hussain, N. F.; Siegel, A. P.; Ge, Y.; Jordan, R.; Naumann, C. A.Bilayer asymmetry influences integrin sequestering in raft-mimickinglipid mixtures. Biophys. J. 2013, 104, 2212−2221.(21) Schroeder, R.; London, E.; Brown, D. Interactions betweensaturated acyl chains confer detergent resistance on lipids andglycosylphosphatidylinositol (GPI)-anchored proteins: GPI-anchoredproteins in liposomes and cells show similar behavior. Proc. Natl. Acad.Sci. U.S.A. 1994, 91, 12130−12134.(22) Kahya, N.; Brown, D. A.; Schwille, P. Raft partitioning anddynamic behavior of human placental alkaline phosphatase in giantunilamellar vesicles. Biochemistry 2005, 44, 7479−7489.(23) Chiantia, S.; Ries, J.; Chwastek, G.; Carrer, D.; Li, Z.; Bittman,R.; Schwille, P. Role of ceramide in membrane protein organization

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847483

investigated by combined AFM and FCS. Biochim. Biophys. Acta,Biomembr. 2008, 1778, 1356−1364.(24) Doering, T. L.; Englund, P. T.; Hart, G. W. Detection ofGlycophospholipid Anchors on Proteins. Curr. Protoc Protein Sci. 2001,Chapter 12, Unit12.5.(25) Milhiet, P.-E.; Gubellini, F.; Berquand, A.; Dosset, P.; Rigaud, J.-L.; Le Grimellec, C.; Levy, D. High-resolution AFM of membraneproteins directly incorporated at high density in planar lipid bilayer.Biophys. J. 2006, 91, 3268−3275.(26) García-Saez, A. J.; Schwille, P. Fluorescence correlationspectroscopy for the study of membrane dynamics and protein/lipidinteractions. Methods 2008, 46, 116−122.(27) Chiantia, S.; London, E. Acyl chain length and saturationmodulate interleaflet coupling in asymmetric bilayers: Effects ondynamics and structural order. Biophys. J. 2012, 103, 2311−2319.(28) Zhang, L.; Granick, S. Lipid Diffusion compared in outer andinner leaflets of planar supported bilayers. J. Chem. Phys. 2005, 123,211104.(29) Liu, J.; Conboy, J. C. 1,2-Diacyl-phosphatidylcholine flip-flopmeasured directly by sum-frequency vibrational spectroscopy. Biophys.J. 2005, 89, 2522−2532.(30) Anglin, T. C.; Cooper, M. P.; Li, H.; Chandler, K.; Conboy, J. C.Free energy and entropy of activation for phospholipid flip-flop inplanar supported lipid bilayers. J. Phys. Chem. B 2010, 114, 1903−1914.(31) Dezi, M.; Di Cicco, A.; Bassereau, P.; Levy, D. Detergent-Mediated Incorporation of transmembrane proteins in giantunilamellar vesicles with controlled physiological contents. Proc.Natl. Acad. Sci. U.S.A. 2013, 110, 7276−7281.(32) Salafsky, J.; Groves, J. T.; Boxer, S. G. Architecture and functionof membrane proteins in planar supported bilayers: A study withphotosynthetic reaction centers. Biochemistry 1996, 35, 14773−14781.(33) Petruzielo, R. S.; Heberle, F. A.; Drazba, P.; Katsaras, J.;Feigenson, G. W. Phase behavior and domain size in sphingomyelin-containing lipid bilayers. Biochim. Biophys. Acta, Biomembr. 2013, 1828,1302−1313.(34) Feigenson, G. W. Phase boundaries and biological membranes.Annu. Rev. Biophys. Biomol. Struct. 2007, 36, 63−77.(35) Feigenson, G. W. Phase diagrams and lipid domains inmulticomponent lipid bilayer mixtures. Biochim. Biophys. Acta 2009,1788, 47−52.(36) Choubey, A.; Kalia, R. K.; Malmstadt, N.; Nakano, A.; Vashishta,P. Cholesterol translocation in a phospholipid membrane. Biophys. J.2013, 104, 2429−2436.(37) Kiessling, V.; Wan, C.; Tamm, L. K. Domain coupling inasymmetric lipid bilayers. Biochim. Biophys. Acta 2009, 1788, 64−71.(38) Wang, T. Y.; Silvius, J. R. Cholesterol does not inducesegregation of liquid-ordered domains in bilayers modeling the innerleaflet of the plasma membrane. Biophys. J. 2001, 81, 2762−2773.(39) Garg, S.; Ruhe, J.; Ludtke, K.; Jordan, R.; Naumann, C. A.Domain registration in raft-mimicking lipid mixtures studied usingpolymer-tethered lipid bilayers. Biophys. J. 2007, 92, 1263−1270.(40) Son, M.; London, E. The dependence of lipid asymmetry uponpolar headgroup structure. J. Lipid Res. 2013, 54, 3385−3393.(41) Lin, Q.; London, E. PLoS One: Preparation of artificial plasmamembrane mimicking vesicles with lipid asymmetry. PLoS One 2014,9, e87903.(42) Gibson Wood, W.; Igbavboa, U.; Muller, W. E.; Eckert, G. P.Cholesterol asymmetry in synaptic plasma membranes. J. Neurochem.2011, 116, 684−689.(43) Mondal, M.; Mesmin, B.; Mukherjee, S.; Maxfield, F. R. Sterolsare mainly in the cytoplasmic leaflet of the plasma membrane and theendocytic recycling compartment in CHO Cells. Mol. Biol. Cell 2008,20, 581−588.(44) Budvytyte, R.; Mickevicius, M.; Vanderah, D. J.; Heinrich, F.;Valincius, G. Modification of tethered bilayers by phospholipidexchange with vesicles. Langmuir 2013, 29, 4320−4327.

Langmuir Article

dx.doi.org/10.1021/la500468r | Langmuir 2014, 30, 7475−74847484