Engineering Asymmetric Lipid Vesicles: Accurate ... - ACS Publications

Article Cite This: Langmuir XXXX, XXX, XXX−XXX

pubs.acs.org/Langmuir

Engineering Asymmetric Lipid Vesicles: Accurate and Convenient Control of the Outer Leaflet Lipid Composition Marie Markones,†,‡ Carina Drechsler,†,‡ Michael Kaiser,† Louma Kalie,† Heiko Heerklotz,*,†,‡,§ and Sebastian Fiedler*,§ †

Institute for Pharmaceutical Sciences, University of Freiburg, Hermann-Herder-Straße 9, Freiburg im Breisgau 79104, Germany BIOSS Centre for Biological Signalling Studies, University of Freiburg, Schänzlestraße 18, Freiburg im Breisgau 79104, Germany § Leslie Dan Faculty of Pharmacy, University of Toronto, 144 College Street, Toronto, Ontario M5S 3M2, Canada ‡

S Supporting Information *

ABSTRACT: The asymmetric distribution of lipids between the two bilayer leaflets represents a typical feature of biological membranes. The loss of this asymmetry, for example the exposure of negatively charged lipids on the extracellular membrane leaflet of mammalian cells, is involved in apoptosis and occurs in tumor cells. Thus, the controlled production of asymmetric liposomes helps to better understand such crucial cellular processes. Here, we present an approach that allows us to design asymmetric model-membrane experiments on a rational basis and predict the fraction of exchanged lipid. In addition, we developed a label-free and nondestructive assay to quantify the asymmetric uptake of negatively charged lipids in terms of the zeta potential. This significantly enhances the applicability, impact, and predictive power of model membranes.

■

INTRODUCTION

Quantitative data on membrane asymmetry mostly originate from studies on planar bilayers16,17 and large unilamellar vesicles (LUVs). The preparation of asymmetric LUVs utilizes a cyclodextrin-assisted (CD) exchange of lipids between two vesicle populations.5,13,15 First, multilamellar-vesicle (MLV) donors are mixed with CD, which extracts lipids without dissolving the MLVs. Then, the MLV−cyclodextrin donor mixture is added to LUV acceptors, such that CD shuttles lipids between the outermost leaflets of donors and acceptors. To separate donor MLVs and acceptor LUVs by centrifugation, either donors9 or acceptors6 are filled with sucrose. The amount of transferred lipid depends on the composition of donor MLVs and acceptor LUVs as well as on the donor-toacceptor ratio. However, it is challenging to estimate the lipid fraction in the donor MLVs accessible to CD (i.e., the exchange-competent fraction). In consequence, the final membrane composition of postexchange LUVs can be difficult to control. Here, we present a lipid-exchange strategy allowing for the precise control of the amount of donor lipid that is incorporated into the outer leaflets of acceptor LUVs. To do so, we treat acceptor LUVs with CD−donor lipid complexes in the absence of donor MLVs. We demonstrate the method by

In most biological membranes, the lipid composition is asymmetric between the membrane leaflets.1 In eukaryotic cellular membranes, for instance, the outer leaflets are enriched in sphingomyelin and phosphatidylcholine (PC), whereas the inner leaflets are rich in phosphatidylethanolamine (PE), phosphatidylserine (PS), and phosphatidylinositol (PI).2 Membranes of Gram-positive bacteria, as a second example, accumulate negatively charged phosphatidylglycerol (PG) in their extracellular membrane leaflets and keep PE and PI mostly on the intracellular side of the membrane, whereas cardiolipin is evenly distributed.3,4 Membrane asymmetry affects membrane potential, order, thermal stability,5,6 dynamics,7 lateral organization,8 and lipid packing.9 The loss of asymmetry, especially the redistribution of negatively charged PS, is associated with apoptosis10,11 and cancer.12 The methods to analyze asymmetric lipid compositions in model membranes comprise steady-state fluorescence anisotropy combined with high-performance thin layer chromatography (HPTLC),5 fluorescence quenching,8 Förster resonance energy transfer,13 and nuclear magnetic resonance spectroscopy combined with gas chromatography.9 PE removal from the outer leaflet has been monitored by trinitrobenzenesulfonate labeling,5,14 and outer leaflet PS contents have been quantified by the binding efficiency of the cationic peptide pL4A18.14,15 © XXXX American Chemical Society

Received: September 10, 2017 Revised: November 30, 2017 Published: January 2, 2018 A

DOI: 10.1021/acs.langmuir.7b03189 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Preparation of Asymmetric LUVs. To generate MβCD−POPG complexes, MβCD was partly saturated with different amounts of POPG from MLVs. To ensure the complete disintegration of MLVs, the mixture was incubated for 20 min at 50 °C with agitation (1000 rpm) in a thermomixer (Eppendorf). After cooling to 28 °C, POPC liposomes were added and quickly mixed by pipetting up and down. The mixture was incubated for 20 min at 28 °C with agitation (1000 rpm). Please refer to Supporting Information for a more detailed protocol. Separation of Postexchange LUVs from Free MβCD and MβCD− POPG Complexes. To separate postexchange LUVs from free MβCD and lipid−MβCD complexes, 500−1000 μL of sample were transferred into a buffer-washed Amicon Ultra 2 mL centrifugal filter (MWCO 100 kDa), concentrated to 0.3−0.5 mL (30−45 min, 3780 × g), and diluted again. Dilution was repeated until MβCD concentration was below 1 mM (typically seven cycles). The asymmetry and integrity of LUVs after the separation process were checked by ζ measurements and DLS (Figures S3 and S5). Long-Term Stability of Membrane Asymmetry. To monitor the long-term stability of the asymmetric lipid distribution, MβCD was removed from postexchange LUVs, and samples were thermostated at 20 °C for 14 days. To record ζ and perform HPTLC, samples were taken on day 0, 4, 7, and 14. Scrambling of Postexchange LUVs. To scramble the lipid composition of postexchange LUVs, we dried 500 μL of 5 mM MβCD-free, asymmetric LUVs, added 500 μL chloroform to homogenize lipids, dried again, rehydrated with water, and prepared LUVs (d = 100 nm) in exactly the same way as done for POPC LUVs. Determination of Lipid Concentration. Lipid concentrations were determined by Bartlett assay.20 High Performance Thin Layer Chromatography (HPTLC). The membrane composition of LUVs was obtained by HPTLC, as described previously.21 Briefly, the mobile phase was composed of 60 mL chloroform, 30 mL methanol, 6.5 mL formic acid, 4.5 mL acetic acid, and 0.1 mL aqueous 0.1% MgCl2 solution. We used a sample applicator, a developing chamber, and a TLS-scanner for densitometry analysis (CAMAG). For HPTLC, MβCD-free, postexchange LUVs were diluted with methanol to the desired concentration, and POPC and POPG concentrations were detected separately. MβCD Density. Density was measured with a DSA 5000 M density and sound velocity meter (Anton Paar) at 28 °C. MβCD Refractive Index. The refractive index was determined with an ABBE refractometer (Fischer) at 28 °C and used to determine the concentration of MβCD solutions. MβCD Viscosity. For viscosity measurements, ζ was obtained of transfer standards (DTS1235) containing MβCD in a concentration range of 30 mM to 70 mM at 28 °C. Dispersant settings in the Zetasizer software contained an experimentally determined refractive index of the DTS1235−MβCD mixture and a dielectric constant of 77.5 (for H2O at 28 °C). Then, the corresponding viscosity setting was adjusted, so that ζ of DTS1235 met the specifications (−42 ± 4.2 mV). The viscosity values (Table S1) were used to adjust dispersant settings in the control software for ζ measurements of samples containing MβCD. Leakage. Leakage was analyzed based on dye efflux from calcein loaded LUVs with the help of time-resolved fluorescence spectroscopy as previously described.18 To do so, MβCD-facilitated lipid exchange was performed on calcein-loaded POPC LUVs. After lipid exchange, samples were diluted to a total lipid concentration of 15 μM and analyzed by time-resolved single photon counting in a FluoTime100 spectromter (PicoQuant). Fluorescence decays were obtained using the FluoFit software (PicoQuant).

preparing LUVs with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoglycerol (POPG)/1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) mixtures in their outer leaflets and pure POPC in their inner leaflets. We then utilize zeta potential (ζ) measurements to quantify POPG uptake. This approach is highly relevant for studies on model membranes in which negatively charged lipids are primarily found on one side of the membrane.

■

EXPERIMENTAL SECTION

Materials. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) was purchased from Avanti Polar Lipids, 1-palmitoyl-2oleoyl-sn-glycero-3-phospho-rac-glycerol ammonium salt (POPGNH4) was kindly contributed by Lipoid. Randomly methylated βcyclodextrin (MβCD) was purchased from Sigma−Aldrich. All other chemicals were purchased from Carl Roth and were of analytical grade. Amicon Ultra 2 mL centrifugal filters (MWCO 100 kDa) and HPTLC-plates were purchased from Merck. Methods. Vesicle Preparation. LUVs of POPC were prepared as described previously.18 Briefly, lipids or lipid mixtures dissolved in chloroform solution were dried to a smooth film and rehydrated with buffer (10 mM Tris pH 7.4 (at 28 °C), 100 mM NaCl, 0.5 mM EDTA). After undergoing five freeze−thaw-cycles, the lipid dispersion was extruded 31 times through an 80-nm nuclepore polycarbonate track-etched membrane (Whatman; GE Healthcare) using a LiposoFast extruder (Avestin). MLVs composed of POPG were prepared in the same way but without extrusion. Isothermal Titration Calorimetry (ITC). ITC was done on a MicroCal ITC 200 (Malvern). LUVs composed of pure POPG, POPC, or POPG−POPC mixtures were injected into MβCD solutions of various concentrations. LUVs at concentrations of 10 mM, 20 mM, and 40 mM were titrated into 20 mM, 30 mM, 40 mM, 50 mM, 60 mM, and 70 mM MβCD solutions. The injection volume was set in a range of 0.4 μL to 1 μL and varied for each experiment depending on MβCD concentrations. The injection interval was set to 300 s in the beginning and end of the titrations, while it was necessary to increase the interval to 600 s in the region of MβCD saturation, at which baseline equilibration required more time. The stirring speed was 750 rpm, filter period 5 s, and the reference power 20.9 μJ/s. All experiments were carried out at 28 °C. Raw thermograms were analyzed with the help of NITPIC.19 Dynamic Light Scattering (DLS). Particle-size distributions of preand postexchange LUVs were measured by DLS on a Zetasizer Nano ZS (Malvern) equipped with a 633 nm He−Ne laser using a detection angle of 173° at 28 °C. Effects on viscosity and refractive index from buffer components were taken into account for the data analysis. Zeta Potential. ζ measurements were performed on the Zetasizer Nano ZS (Malvern) using a flow-through high concentration zeta potential cell (HCC; Malvern). Prior to measurements, the HCC was equilibrated at 28 °C in the instrument for at least 10 min. Also, buffer, water, and transfer standard DTS1235 (Malvern) were incubated at 28 °C. The instrument settings were as follows: temperature 28 °C, incubation time 90 s, number of runs 10−100, and the dispersant setting was water for the transfer standard DTS1235 as well as a customized solvent for each MβCD concentration (Table S1 of the Supporting Information, SI). ζ was measured in triplicate. For each measurement, the lipid dispersion in the tubing of the HCC was pushed a little forward, so that a new part of the sample was transferred into the sample chamber. Values that exceeded a standard deviation of 3 mV were considered of low quality. In such a case, the HCC was refilled and another series of triplicates was obtained. The transfer standard DTS1235 was measured at the beginning of each experiment and about every 30 min in-between samples to assess the quality of the setup. If the standard did not meet its specification (−42 ± 4.2 mV), the HCC was disassembled and cleaned (ddH2O and/or 1% Hellmanex III). Samples that were recorded right before cleaning the cell were remeasured to avoid systematic errors in the data. Between different samples, the assembled HCC was cleaned with preheated water followed by buffer and dried using pressurized air.

■

RESULTS AND DISCUSSION Phase Diagram of MβCD−POPG Complexation Established by ITC. To remove POPG MLVs from exchange mixtures, we identified conditions of complete POPG solubilization by CD. For this purpose, isothermal titration calorimetry (ITC) was used to establish a phase diagram for B


Article

Langmuir

Figure 1. Solubilization of POPG LUVs by MβCD. a) ITC thermogram resulting from injections of 20 mM POPG into 60 mM MβCD at 28 °C. b) Integrated heats of LUV solubilization, Q, as a function of POPG-to-MβCD molar ratio at 28 °C. Lines guide the eye. Arrow indicates saturation for 60 mM MβCD. c) Phase diagram of POPG−MβCD at 28 °C derived from the first break points in (b). Line is described by eq 122 and represents the phase boundary of POPG−MβCD. d) Equilibrium constants, K, of MβCD−lipid complexation as a function of the mole fraction of POPG, XPOPG, in mixed POPG−POPC LUVs. K was obtained assuming a complex stoichiometry, n, of 1:4 (lipid:CD) by nonlinear least-squares fitting of eq 1 to the respective phase diagrams (Figure S2); dotted line guides the eye.

compositions. However, the unit of K in eq 1 is M−(n‑1), leading to values of K that are not comparable among fitting results based on a variable n. To address this issue and with the benefit of improved precision of K, we fixed n at values of 4 in the following NLLS fitting analyses. Interestingly, K of MβCD− lipid complexation from POPG LUVs is about 4 times higher than the value previously reported for POPC LUVs22 (90 M−3) and 5.5 times higher than our own POPC value (Figure 1d). The same trend has been observed in light scattering experiments.13 In addition, increased POPG fractions of up to 60 mol% left K largely unchanged as compared with POPConly LUVs (Figure 1d). This nonlinear behavior suggests that binding affinities are not intrinsically different between the complexes of POPG−MβCD and POPC−MβCD, and that higher values of K in POPG-only LUVs (or LUVs with a large enough fraction of POPG) may result from electrostatic repulsion of POPG molecules in the membrane. This is a key point, because efficient lipid exchange is impaired by large differences in binding affinities of POPG and POPC to MβCD as it causes lipid compositions in the complexes to vary from the composition of the membrane. Our affinity data indicate that this is not the case until 60 mol% POPG in POPC. For POPC-only LUVs, we obtained K = 41 M−3 (Figure S2), a value that is slightly smaller than the 90 M−3 published previously.22 However, the two values are not in conflict with each other taking into account that at a stoichiometry of 4, K depends sensitively on the activity of MβCD. In fact, a

mixtures of POPG and randomly methylated βCD (MβCD). Titration of 20 mM POPG in LUVs into 60 mM MβCD resulted in large endothermic heat signals indicative of complete LUV solubilization followed by smaller exothermic heats typical of dilution of intact LUVs into MβCD (Figure 1a). Similar to the latter, titrations of buffer into solutions of MβCD exhibited exothermic heat signals of comparable magnitude, which were then used to correct for heats of dilution during data analysis (Figure S1). The saturation of MβCD with POPG is marked by the kink in the heat curve that ends the first endothermic plateau (Figure 1b, arrow), as described previously. 22 The saturation points at various MβCD concentrations result in a phase diagram of POPG−MβCD mixtures (Figure 1c). The phase boundary is described by the following:22 c MβCD = nc PG +

⎛ c PG ⎞1/ n ⎜ ⎟ ⎝ K ⎠

(1)

where cMβCD is the total MβCD concentration, n the stoichiometry of the complex, cPG the total POPG concentration, and K the equilibrium constant. To investigate how the lipid affinity of MβCD is affected in mixtures of POPG and POPC, we utilized the same procedure to obtain phase diagrams of mixed LUVs having POPG contents of 17 mol%, 29 mol%, and 60 mol% (Figure S2). Nonlinear least-squares (NLLS) fitting was performed to obtain values of K and n (eq 1). Free fits yielded values of n ≈ 4 for all analyzed lipid C


Article

Langmuir

Figure 2. Use of zeta potential, ζ, to quantify POPG uptake by POPC LUVs. a) Black circles are ζ as a function of the POPG mole fraction in the out outer leaflet of symmetric, POPG−POPC LUVs, Xout PG, with XPG = XPG. Black line is a generic fit by eq 3. b) Star represents postexchange LUVs from 1.7 mM POPG and 60 mM MβCD with 5.3 mM POPC directly after lipid exchange. c) Red, dashed line identifies asymmetry of postexchange LUVs and represents ζ as a function of the total mole fraction of POPG. d) Black line, red line, and filled star same as in previous panels but all plotted against the total mole fraction of POPG. This allows the identification of asymmetry (red, dashed line) as well as the quantification of the POPG mole fraction in the outer leaflet (blue arrow and black line) in a single plot. Empty star is the previously asymmetric sample (filled star) with a scrambled, symmetric lipid composition. a)−d) Error bars are standard deviations (n = 3).

difference in MβCD activity by ±9% alone could account for the difference in the results. Considering that the MβCD has been purchased from different manufacturers in different decades and that even the product used here does not exclude a batch-to-batch variability in the degree of methylation of up to ±35% (1.5−2.1 methyl groups per glucose unit; specification of MβCD used by Anderson et al.22 not known), such a variability has to be expected. We emphasize that the detection of the critical MβCD concentration for complete solubilization of the lipid and the relative comparison of MβCD affinities to different lipid mixtures are the key data for the study here and do not suffer from this problem. Quantification of Asymmetry by ζ Measurements. The zeta potential is an ideal parameter for detecting asymmetric localization of ionic lipids because it depends exclusively on the composition of the outer leaflet. The insulating effect of the low dielectric membrane core renders charges in the inner leaflet “invisible” to the method. Complexes of MβCD with charged lipids do also not affect ζ measurements of liposomes, because the instrument works on the basis of light scattering with its intensity being proportional to the sixth power of size. Hence, small complexes are not detected in the presence of much larger vesicles (Figure S3). To demonstrate the power of the method, we selected a model system in which anionic POPG is taken up into the outer leaflets of zwitterionic POPC LUVs. Before we performed a cyclodextrin-based lipid exchange, we

first recorded ζ of symmetric POPG−POPC mixed LUVs with increasing mole fraction of POPG, XPG, c PG XPG = c PG + c PC (2) obtained from the total, molar concentration of POPG and POPC, cPG and cPC, respectively (Figure 2a). Taking into account the aim of this study, we did not attempt a physical modeling of the curve but did a simple, empirical fit, yielding: out 3 out 2 out ζ = − 93.7 mV(XPG ) + 185 mV(XPG ) − 140 mV(XPG ) − 4.32 mV

(3)

The superscript “out” indicates that ζ depends exclusively on the local POPG content in the outer leaflet of the liposomes. Note that for the symmetrical liposomes used to generate the calibration curve, Xout PG = XPG. Consequently, following cyclodextrin-based lipid exchange, the POPG contents in the outer leaflets of postexchange LUVs can be read directly from the calibration curve (Figure 2b). Notably, ζ measurements also predict if lipid exchange by MβCD successfully generated asymmetric LUVs. Our protocol is based on mixing liposomes of a basis lipid, here POPC, with complexes of POPG and MβCD. It is assumed that (i) there is a full equilibration of the lipids in the complexes and the outer leaflet; the inner leaflet is not accessible, (ii) neither membrane nor complexes have a preference for a certain lipid species so that their POPG content after exchange is the same, and (iii) D


Article

Langmuir

predicted to be 1.0 mM/(1.0 mM + 4.3 mM) = 0.19, and the corresponding ζ is −25 mV (Figure 2d). Generally, we may write the following:

membrane lipids are exchanged 1:1, each inserted molecule is balanced by another one being extracted. The validity of these assumptions will be discussed in the following example. Let us focus on the data depicted in Figure 2b, a sample containing 1.7 mM POPG originally solubilized by 60 mM MβCD and LUVs of POPC with a total concentration of cPC = 5.3 mM. The overall POPG content in the system is XPG = 1.7 mM/(1.7 mM + 5.3 mM) = 0.24. On the basis of the assumptions described above, we can calculate the composition of the outer leaflet after exchange. All POPG equilibrates with half of the POPC (i.e., outer-leaflet POPC) so that Xout asy = 1.7 mM/(1.7 mM + 0.5 × 5.3 mM) = 0.39 with Xout asy being the mole fraction of lipid in the outer leaflet of asymmetric LUVs. Generally, we may write the following: out comp Xasy = XPG =

c PG acc f PC c PC + c PG

scr XPG =

c PG out (1 − Xasy ) c PC

(5)

with quantities on the right-hand side referring to the sample before complex removal (i.e., in agreement with those discussed above). The experimental result after scrambling is ζ = −27 mV, again in agreement with the prediction. This ζ suggests an scr Xscr PG of 0.22; HPTLC analysis yielded a slightly larger XPG = 0.26 (Figure S6). In summary, complete membrane asymmetry can be sensitively monitored in terms of ζ agreeing with the red, dashed “asymmetry curve”; a partial or complete loss of asymmetry implies ζ to approach the black curve (Figure S7). The consistency of the predictions and experimental data supports the applicability of the assumptions and equations presented here. The above findings result in a straightforward protocol for the preparation and quantification of asymmetric vesicles comprising the following steps. First, a target Xout asy needs to be selected. The second step is the selection of a POPC concentration, which is based on how much asymmetric vesicle material is required. The first two choices determine the POPG concentration that is required to reach the selected Xout asy based on eq 4, which can be solved for the POPG concentration:

(4)

with Xcomp being the mole fraction of POPG in MβCD PG complexes and f acc PC being the fraction of POPC accessible for lipid exchange. We assume f acc PC = 0.50 for LUVs as measured by 31 P NMR in a previous study.23 Small positive or negative deviations within 0.50 ± 0.05 can arise from curvature effects and small fractions of oligovesicular systems, respectively. Importantly, the preparation of asymmetric small unilamellar vesicles (SUVs) requires a value of f acc PC = 0.60 because SUVs exhibit a lipid distribution of inner:outer = 40:60 due to their overall smaller diameters (∼30 nm). For Xout asy = 0.39 as predicted, eq 3 yields a corresponding ζ of −36 mV. The experimental value obtained in our example is ζ = −38 mV, which yields a POPG mole fraction in the outer leaflet of Xout asy = 0.43 (Figure 2b). Moreover, for each bulk POPG content, XPG, this procedure predicts a ζ expected for asymmetric, outside-only insertion, and all these points lie on the red, dashed “asymmetry curve” in Figure 2c. The experimental value of ζ is closely positioned to the asymmetry curve and thus indicates asymmetric lipid exchange. Figure 2d presents the calibration curve, the asymmetry curve, and the postexchange data as a function of the total mole fraction of POPG, which allows for the analysis of asymmetry (red asymmetry curve) as well as the quantification of the POPG content of postexchange LUVs in a single plot. In our samples, asymmetry is maintained upon storage at 20 °C for at least 14 days (Figure S4), in agreement with recent observations of slow transbilayer diffusion of 1,2-dipalmitoyl-snglycero-3-phosphocholine (DPPC) in LUVs.24 In addition, the removal of the lipid−MβCD complexes left ζ unchanged as predicted (Figure S3), and LUV integrity is maintained as demonstrated by dynamic light scattering (Figure S5). To further test the protocol, we actively scrambled the lipids between the two leaflets of asymmetric LUVs. In this case, one should expect ζ to be shifted toward the black “symmetry curve”. For a precise positioning of the data in Figure 2d, we need to calculate the POPG content after complex removal and comp scrambling. We have assumed that Xout asy = XPG = 0.39 and that each POPG inserted into the membrane is essentially 1:1 replaced by a POPC being removed, this is, the overall concentration of complexed lipid remains at about cPG. Thus, of the 1.7 mM lipid removed before scrambling, there were 1.7 mM × 0.39 = 0.7 mM POPG and 1.7 mM × (1−0.39) = 1.0 mM of POPC. The lipids remaining in the system are then 1.7 mM − 0.7 mM = 1.0 mM POPG and 5.3 mM − 1.0 mM = 4.3 mM POPC. The POPG content after scrambling, Xscr PG, is thus

c PG =

acc out f PC c PCXasy out 1 − Xasy

(6)

In turn, the POPG concentration determines how much MβCD is required for complexation, which can be decided based on the POPG−MβCD phase diagram or with the help of the speciation curves in Figure S8. Then, the lipid exchange can be performed followed by ζ measurement in the presence of MβCD. Once asymmetry is confirmed and the POPG content of the sample is known, MβCD can be removed with the help of centrifugal ultrafiltration devices. At this stage, the sample is ready for further biophysical analysis. Note that we provide a detailed lipid-exchange protocol as well as an excel sheet to calculate required lipid and MβCD concentrations in the SI. Next, we tested the range of outer leaflet POPG contents that our method was able to generate in POPC LUVs (Figure 3). For this purpose, we targeted Xout asy in a range of ∼0.02 to ∼0.5, which required MβCD concentrations between 30 mM and 60 mM. Xout asy in a range of 0.05−0.45 gave the most reliable results, as indicated by postexchange ζ that coincides with the asymmetry curve. At very low values of Xout asy (0.45), the ζ curve exhibits a rather shallow slope such that ζ is less sensitive to changes in the POPG contents of postexchange LUVs. In consequence, the POPG content obtained by ζ measurements in the shallow-end range of the calibration curve might be less reliable. In addition, the amount of lipid that can be exchanged asymmetrically is limited by the practical challenge of adding highly concentrated MβCD to POPC LUVs without causing membrane defects, which we started to observe at MβCD concentrations of 70 mM (Figures S9 and S10). E


Article

Langmuir

leaflet and vice versa. The consequence is a lipid buffering effect so that the complexed amount of lipid, this is, the effective phase boundary, is adjusted to allow for a largely balanced membrane. This effect has been quantified in a comparable case before, where the membrane partitioning of a membraneimpermeant surfactant, dodecyl maltoside, was shown to be a factor of two higher into underpopulated compared to overpopulated membrane leaflets.23 Thus, performing the exchange assay at any POPG concentration between the POPG phase boundary and the POPC phase boundary (Figure S11) can result in essentially stress-free asymmetric membranes as well as samples free of contamination by symmetric vesicles. In our lab, we typically use medium degrees of lipid saturation (50%−75%) required to reach a certain Xout asy . Further studies will be needed to test whether some subtle differences remain between asymmetric liposomes formed using particularly low or high degrees of saturation of MβCD. As mentioned earlier, the addition of very high absolute concentrations of MβCD to POPC LUVs caused defects in POPC membranes (leakage, Figure S10), most likely as a result of high local MβCD concentrations during mixing. Also, at high degrees of MβCD saturation, care must be taken to avoid contamination by POPG MLVs

Figure 3. Zeta potential, ζ, after lipid exchange at various MβCD concentrations as well as various levels of MβCD saturation with POPG plotted as a function of the total mole fraction of POPG in the exchange mixture, XPG. Refer to legend for colors and symbols. Black line and red, dashed line are symmetric and asymmetric calibration curves from Figure 2, respectively. Inset: Range of MβCD concentrations, cMβCD, required to target various mole fractions of POPG in the outer leaflets of POPC LUVs, Xout asy . Lines for saturation of 25% and 100% based on eq 4 with 5 mM as the total POPC concentration.

Choosing a Suitable MβCD Concentration Profits from a “Lipid Buffer” Effect. With the help of the phase diagram, we can select several MβCD concentrations to complex identical amounts of POPG. As shown in Figure S8, for example, 0.5 mM POPG can be complexed by 70 mM MβCD at 15% saturation, 60 mM MβCD at 25% saturation, 50 mM MβCD at 42% saturation, or 40 mM MβCD at 87% saturation. Consequently, several combinations of POPG, out POPC, and MβCD concentrations lead to the same Xasy (Figure 3, inset). For example, 40 mM MβCD at 90% saturation, 50 mM MβCD at 50% saturation, and 60 mM MβCD at 25% saturation all result in Xout asy of ∼0.18 (Figure 3). At first glance, one should expect only one of these different combinations to yield stable, relaxed, asymmetric membranes, because only one condition removes exactly as many POPC molecules as needed to accommodate the POPG molecules added. If the degree of saturation of MβCD added is too low, then it should extract too much POPC and the outer leaflet gets underpopulated, this is, with lower lateral pressure, compared to the inner one. If the degree of saturation of MβCD is too high, then too little POPC would be extracted and the outer leaflet should get overpopulated. Such pressure imbalances ultimately induce lipid scrambling by transient defects23 that would jeopardize the assay. An ideal MβCD concentration to produce balanced membranes with a certain Xout asy is one that lies on the saturation phase boundary for this particular lipid mixture during exchange. Considering this problem, it may be surprising that we obtained highly asymmetric and fairly stable liposomes over a very broad range of degrees of saturation, all the way from 25% to 90%. This apparent paradox can be resolved taking into account that the “saturation” boundary in our case occurs at as little as one lipid molecule per, for example, 100 MβCD molecules. That means, it is governed by the chemical potential of the lipids in the membrane leaflet that is in equilibrium with the complexes and not by a physical capacity limit of MβCD. In fact, electrostatic repulsion of POPG in the membrane increases its local chemical potential and accounts for a quite dramatic shift of the phase boundary compared to zwitterionic POPC (Figures S2 and S11). The same will happen if the chemical potential is raised by an overpopulation of the outer

■

CONCLUSIONS We present a robust, easy to use, and quantitative method for the preparation of asymmetric lipid membranes. We show that ζ measurements reliably quantify membrane asymmetries that involve negatively charged lipids. Complete solubilization of POPG by MβCD prior to mixing with POPC allows for precise control of the exchanged fraction of POPG. Choosing the degree of MβCD saturation needed for the precise engineering of stress-free, asymmetric LUVs is discussed to profit from a “lipid buffering” effect of MβCD. The presented exchange strategy is transferable to other lipids, improves the general understanding of lipid exchange processes, and puts the preparation of asymmetric vesicles on a more rational basis.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.7b03189. Detailed protocol for the preparation of asymmetric LUVs, 11 supporting figures, and one supporting table (PDF) Excel sheet to calculate concentrations of lipid and MβCD in lipid exchange (XLSX)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: heiko.heerklotz(at)pharmazie.uni-freiburg.de (H.H.). *E-mail: mail(at)sebastianfiedler.net (S.F.). ORCID

Heiko Heerklotz: 0000-0003-4615-7022 Sebastian Fiedler: 0000-0003-0953-4327 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. F


Article

Langmuir Notes

(15) Lin, Q.; London, E. Preparation of Artificial Plasma Membrane Mimicking Vesicles with Lipid Asymmetry. PLoS One 2014, 9 (1), e87903. (16) Crane, J. M.; Kiesling, V.; Tamm, L. K. Measuring Lipid Asymmetry in Planar Supported Bilayers by Fluorescence Interference Contrast Microscopy. Langmuir 2005, 21 (4), 1377−1388. (17) Collins, M. D.; Keller, S. L. Tuning Lipid Mixtures to Induce or Suppress Domain Formation across Leaflets of Unsupported Asymmetric Bilayers. Proc. Natl. Acad. Sci. U. S. A. 2008, 105 (1), 124−128. (18) Fiedler, S.; Heerklotz, H. Vesicle Leakage Reflects the Target Selectivity of Antimicrobial Lipopeptides from Bacillus Subtilis. Biophys. J. 2015, 109 (10), 2079−2089. (19) Keller, S.; Vargas, C.; Zhao, H.; Piszczek, G.; Brautigam, C. A.; Schuck, P. High-Precision Isothermal Titration Calorimetry with Automated Peak-Shape Analysis. Anal. Chem. 2012, 84 (11), 5066− 5073. (20) Bartlett, G. R. Phosphorus Assay in Column Chromatography. J. Biol. Chem. 1959, 234 (3), 466−468. (21) Holzer, M.; Burghardt, A.; Schubert, R. Quantitative HighPerformance Thin-Layer Chromatography Determination of Common Liposome Components and Critical Parameters Influencing the Analysis Results. J. Liposome Res. 2010, 20 (2), 124−133. (22) Anderson, T. G.; Tan, A.; Ganz, P.; Seelig, J. Calorimetric Measurement of Phospholipid Interaction with Methyl-β-Cyclodextrin. Biochemistry 2004, 43 (8), 2251−2261. (23) Heerklotz, H. Membrane Stress and Permeabilization Induced by Asymmetric Incorporation of Compounds. Biophys. J. 2001, 81 (1), 184−195. (24) Marquardt, D.; Heberle, F. A.; Miti, T.; Eicher, B.; London, E.; Katsaras, J.; Pabst, G. 1H NMR Shows Slow Phospholipid Flip-Flop in Gel and Fluid Bilayers. Langmuir 2017, 33 (15), 3731−3741.

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We are indebted to Eva Dengler and Nicole Specht for excellent technical assistance. Moreover, we thank Drs. Jana Broecker, Helen Fan (both Univ. Toronto), and Maria Hoernke (Univ. Freiburg) for comments on the manuscript. H.H. and S.F. were supported by the Natural Sciences and Engineering Research Council of Canada and the Deutsche Forschungsgemeinschaft (FI 2005/1-1), respectively.

■

REFERENCES

(1) Marquardt, D.; Geier, B.; Pabst, G. Asymmetric Lipid Membranes: Towards More Realistic Model Systems. Membranes (Basel, Switz.) 2015, 5 (2), 180−196. (2) Verkleij, A.; Zwaal, R.; Roelofsen, B.; Comfurius, P.; Kastelijn, D.; van Deenen, L. The Asymmetric Distribution of Phospholipids in the Human Red Cell Membrane. A Combined Study Using Phospholipases and Freeze-Etch Electron Microscopy. Biochim. Biophys. Acta, Biomembr. 1973, 323 (2), 178−193. (3) Barsukov, L. I.; Kulikov, V. I.; Bergelson, L. D. Lipid Transfer Proteins as a Tool in the Study of Membrane Structure. Inside-Outside Distribution of the Phospholipids in the Protoplasmic Membrane of Micrococcus Lysodeikticus. Biochem. Biophys. Res. Commun. 1976, 71 (3), 704−711. (4) Rothman, J. E.; Kennedy, E. P. Asymmetrical Distribution of Phospholipids in the Membrane of Bacillus Megaterium. J. Mol. Biol. 1977, 110 (3), 603−618. (5) Cheng, H. T.; Megha; London, E. Preparation and Properties of Asymmetric Vesicles That Mimic Cell Membranes. Effect upon Lipid Raft Formation and Transmembrane Helix Orientation. J. Biol. Chem. 2009, 284 (10), 6079−6092. (6) Cheng, H. T.; London, E. Preparation and Properties of Asymmetric Large Unilamellar Vesicles: Interleaflet Coupling in Asymmetric Vesicles Is Dependent on Temperature but Not Curvature. Biophys. J. 2011, 100 (11), 2671−2678. (7) Chiantia, S.; London, E. Acyl Chain Length and Saturation Modulate Interleaflet Coupling in Asymmetric Bilayers: Effects on Dynamics and Structural Order. Biophys. J. 2012, 103 (11), 2311− 2319. (8) Lin, Q.; London, E. Ordered Raft Domains Induced by Outer Leaflet Sphingomyelin in Cholesterol-Rich Asymmetric Vesicles. Biophys. J. 2015, 108 (9), 2212−2222. (9) Heberle, F. A.; Marquardt, D.; Doktorova, M.; Geier, B.; Standaert, R. F.; Heftberger, P.; Kollmitzer, B.; Nickels, J. D.; Dick, R. A.; Feigenson, G. W.; et al. Subnanometer Structure of an Asymmetric Model Membrane: Interleaflet Coupling Influences Domain Properties. Langmuir 2016, 32 (20), 5195−5200. (10) McEvoy, L.; Williamson, P.; Schlegel, R. A. Membrane Phospholipid Asymmetry as a Determinant of Erythrocyte Recognition by Macrophages. Proc. Natl. Acad. Sci. U. S. A. 1986, 83 (10), 3311−3315. (11) Verhoven, B.; Schlegel, R. A.; Williamson, P. Mechanisms of Phosphatidylserine Exposure, a Phagocyte Recognition Signal, on Apoptotic T Lymphocytes. J. Exp. Med. 1995, 182 (5), 1597−1601. (12) Riedl, S.; Rinner, B.; Asslaber, M.; Schaider, H.; Walzer, S.; Novak, A.; Lohner, K.; Zweytick, D. In Search of a Novel Target  Phosphatidylserine Exposed by Non-Apoptotic Tumor Cells and Metastases of Malignancies with Poor Treatment Efficacy. Biochim. Biophys. Acta, Biomembr. 2011, 1808 (11), 2638−2645. (13) Huang, Z.; London, E. Effect of Cyclodextrin and Membrane Lipid Structure upon Cyclodextrin−Lipid Interaction. Langmuir 2013, 29 (47), 14631−14638. (14) Lin, Q.; London, E. The Influence of Natural Lipid Asymmetry upon the Conformation of a Membrane-Inserted Protein (Perfringolysin O). J. Biol. Chem. 2014, 289 (9), 5467−5478. G
