Article pubs.acs.org/Langmuir
Fusion of Bolaamphiphile Micelles: A Method to Prepare Stable Supported Biomimetic Membranes Y. Kaufman,†,⊥ S. Grinberg,‡ C. Linder,§,⊥ E. Heldman,∥ J. Gilron,⊥ and V. Freger*,# †
Albert Katz International School for Desert Studies and Unit of Environmental Engineering, Ben-Gurion University of the Negev, Sde-Boqer, Israel ‡ Department of Chemistry, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel § Department of Biotechnology Engineering, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel ∥ Department of Clinical Biochemistry, the Faculty of Health Sciences, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel ⊥ Zuckerberg Institute for Water Research, Ben-Gurion University of the Negev, Sde-Boqer, 84990, Israel # Wolfson Department of Chemical Engineering, Technion − Israel Institutes of Technology, Haifa 32000, Israel S Supporting Information *
ABSTRACT: Supported biomimetic membranes (SBMs) on solid substrates have been commonly prepared from vesicleforming double-tail lipids, such as zwitterionic phospholipids, using the method of vesicle fusion. Here we report on the preparation of SBMs on silica surfaces via a similar process of “micelle fusion” from a cationic single-tail bolaamphiphile GLH-20 that forms spherical and elongated thread-like micelles in solution. We demonstrate that, in contrast to zwitterionic phospholipids, GLH-20 self-assembles into a stable contiguous SBM at both low and high ionic strengths. The cationic charge of GLH-20 promotes the formation of a stable SBM through enhanced double-layer interactions with the negatively charged silica surface. It is also shown that spinach aquaporin PM-28 was successfully incorporated within bolaamphiphile SBM in a manner similar to SBMs prepared by vesicle/proteoliposome fusion; thereby the inherent curvature of the micelle surface does not inhibit protein reconstitution. The results suggest that SBMs based on charged bolaamphiphiles might be an attractive platform for applications such as water purification and biosensors, where the stability and low defect rate of SBMs in diverse conditions are crucial for achieving desired performance.
■
rupture, and self-assemble into a contiguous flat membrane.10,11 This mechanism, often called “vesicle fusion”, has apparently been the most common method of preparation of SBMs on solid substrates.10,12,13 The first step in vesicle fusion, vesicle adsorption, is governed by electrostatic-entropic “double-layer” interactions between the vesicle and the surface.14,15 These interactions are strongly affected by the ionic strength of the solution.14,15 As long as double-layer interactions dominate the adhesion, oppositely charged surfaces of the same absolute charge density would attract each other at any ionic strength.16−18 However, in most cases, the charge densities of the surface and vesicles are different, which may result in either attractive or repulsive interactions, depending on the ionic strength.16−18 For instance, Anderson et al. measured repulsive interactions between a negatively charged silica surface and DMPC bilayer, exhibiting zero net-charge behavior (ζ-potential ≈ 0) at neutral
INTRODUCTION Supported biomimetic membranes (SBMs) have attracted much interest as enabling components in promising applications such as aqueous separations and biosensors.1−5 For instance, SBMs with incorporated aquaporin proteins hold potential for creating a high-performance water purification membrane with water permeability and ion rejection largely exceeding those of commercial polymeric membranes.1,6−8 Unfortunately, preparation of stable functioning SBMs and their integration and interfacing with an appropriate robust supporting structure still remains highly challenging. Double-tail zwitterionic phospholipids, common in biological membranes, have been the most studied amphiphilc compounds in the context of SBM preparation. The shape factor (SF) of these lipids, dictating that the dynamic shape of lipid aggregates favors formation of vesicles (1 > SF > 0.5), as lowcurvature aggregates.9 Vesicles may be considered the most natural precursor of SBM, since lipid membranes in both formations have a near-zero curvature. When vesicle curvature is increased by extrusion, phospholipid vesicles tend to adsorb on hydrophilic negatively or positively charged surfaces, © 2013 American Chemical Society
Received: August 5, 2012 Revised: December 29, 2012 Published: January 2, 2013 1152
dx.doi.org/10.1021/la304484p | Langmuir 2013, 29, 1152−1161
Langmuir
Article
layer in the transmembrane configuration (see Figure 1) was assumed to be beneficial for the membrane stability.22 By contrast, single-chain lipids having a lower SF (SF < 0.5) may form more aggregates with a higher surface curvature such as micelles. For instance, Pal and co-workers showed that single-chain bolaamphiphiles self-assembled to rod-like micelles.28 Meister et al. reported that single-chain bolalipids formed thread-like micelles below the gel−liquid transition temperature of the lipid phase but also spherical micelles above it.29 Yuen et al. showed that a single-chain bolalipids formed hexagonal mesophases, which were used as templates for silica formation.30 On the other hand, Popov et al. reported the formation of vesicles from symmetrical single-chained bolaamphiphiles.31 This could be explained by the effects of lipid composition, concentration, ionic strength, and temperature, all of which could profoundly affect the shape and size of bolaamphiphile aggregates. Despite the above potential advantages, preparation of SBM using neutral or charged bolalipids has not been extensively studied. In this paper, we present the characteristics and preparation of an SBM from proprietary positively charged single-chain bolalipid GLH-20 on a silica surface. The SBM characteristics, structure, and effect of preparation conditions, in particular ionic strength, were examined using a number of complementary techniques. Despite the fact that GLH-20 in solution tends to self-assemble into micelles of high surface curvature rather than vesicles, consistent with its low SF, the micelles are shown to self-assemble into a contiguous SBM, as schematically shown in Figure 1. The resulting SBM could also incorporate aquaporin, a water-channel membrane protein. To the best of our knowledge this type of self-assembly, which may be termed “micelle fusion”, has not been reported previously.
pH (see Table 1). They found that in 150 mM NaCl, DMPC vesicles adhered to the surface and fused to form SBMs, Table 1. ζ-Potential of GLH-20 and DMPC Aggregates Measured in Different NaCl Solutions ζ-potential @ 30 °C (mV) amphiphile
1 mM NaCl
25 mM NaCl
150 mM NaCl
GLH-20 DMPC
55 ± 6 0±6
46 ± 3 0±6
21 ± 0.5 0±6
whereas in distilled water formation of SBMs was inhibited.14 It was argued that the surface charge remained nearly constant, which could be explained by the disjoining osmotic pressures of counterions in the interlayer between SBMs that could not be balanced by a low ionic strength solution. It may be expected that by using charged rather than zwitterionic lipids, the surface charge could be partly neutralized by the lipids and thus allow counterions to “escape”. This would reduce the disjoining osmotic pressure and thereby enable SBM formation in a wider range of ionic strengths. Moreover, it is expected that strong charge interactions with the surface may increase SBM stability. Apart from optimized double layer interactions, SBM formation and stability can benefit from the use of bolaamphiphiles.19 Bolaamphiphiles, also called bolalipids, have two hydrophilic headgroups connected by single or double hydrophobic chains.20,21 Bolalipids are commonly present in the membrane of thermophilic and acidophilic archaebacteria and were reported to be a key factor for their stability in extreme conditions.22,23 Bolalipids can self-assemble into diverse dynamic shapes. The SF of double-tail bolalipids is similar to that of phospholipids; therefore they tend to assemble into vesicles and can stabilize and support the function of integral membrane proteins, similar to regular lipid bilayer membranes.19,24 Within vesicles, bolalipids can adopt two generic conformations, namely, either a transmembrane or a hairpin (U-shaped) conformation, or a combination of the two configurations (see Figure 1).25−27 Absence of a preferential fracture point in the middle plane of the lipid
■
MATERIALS AND METHODS
Materials. 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and sodium dodecyl sulfate (SDS) were purchased from SigmaAldrich. 1,2-Dimyristoyl-sn-glycero-phosphoethanolamine-N-(lissamine-Rhodamine B sulfonyl) with ammonium salt (Rh-PE) was purchased from Avanti-Polar Lipids. Preparation and characterization
Figure 1. Schematic illustration of possible curved aggregated structures formed by one-tail bolalipids: spherical or thread-like micelles in solutions that fuse into SBMs on silica. The transmembrane conformation (marked in red) is assumed to contribute to SBM stability. 1153
dx.doi.org/10.1021/la304484p | Langmuir 2013, 29, 1152−1161
Langmuir
Article
Figure 2. Chemical structures of GLH-20, DMTAP, and DMPC. The size of the DMTAP headgroup was assumed to be commensurate with GLH20. DMPC was used as a model zwitterionic phospholipid. of the proprietary bolalipid GLH-20 of molecular weight 1238 g mol−1 possessing two positively charged quaternary amine headgroups is described elsewhere.20 Figure 2 shows the chemical structures of GLH20, dimyristoyltrimethylammonium propane (DMTAP), and DMPC. PM28 aquaporin protein solution, an integral protein of spinach leaf plasma membrane, was kindly supplied by Prof. Per Kjellbom and Dr. Urban Johanson from Lund University (Sweden).32,33 Chloroform was purchased from Fluka. NaCl was purchased from Frutarom. 1-n-Octylβ-D-glucopyranoside (OG) was purchased from A.G. Scientific. All lipids were hydrated in deionized water (Zalion, < 1 μS cm−1). Solutions. DMPC or GLH-20 were hydrated in NaCl solutions for 30 min, and extruded through a polycarbonate track-etched membrane with 100 nm pores (SPI Pore) at 30 °C. The final lipid concentrations were 0.7 mM (∼0.5 mg mL−1) and 1.5 mM (1.9 mg mL−1) for DMPC and GLH-20, respectively. For FRAP experiments, the preparation was identical except that 0.1−1% (mol) of Rh-PE was added to DMPC or GLH-20 by dissolving the lipid mixture in chloroform, followed by chloroform evaporation in vacuum oven at 40 °C for 2 h. Incorporation of PM28 in GLH-20 SBM. A 8.41 mg mL−1 solution of PM28 aquaporin protein was added to a solution of 0.7 mM GLH-20 and 1% (wt) OG (purchased from A.G. Scientific) in 1 mM NaCl to give a lipid-to-protein ratio of 500:1. The solution was incubated at 30 °C for 30 min, and then OG was removed by dialysis using 6−8 kDa molecular weight cutoff dialysis bags (Spectra Por) against 1 mM NaCl solution at 30 °C for 2 h. The sample preparation for AFM imaging and cantilever parameters are described below. Quartz Crystal Microbalance with Dissipation (QCM-D) Experiments. QCM-D experiments were performed using AT-cut quartz crystals mounted in an E4 system (Q-sense AB, Gothenburg, Sweden). Crystals coated with 50 nm silicon dioxide layer, having a fundamental resonant frequency of ∼5 MHz, were purchased from QSense AB. Before each measurement, the crystals were soaked in 2% (wt.) SDS solution for 30 min, thoroughly rinsed with deionized water, dried with N2 gas, and further cleaned in UV/Ozone ProCleaner (BioForce Nanoscience) for 30 min. All QCM-D experiments were performed under flow-through conditions at 0.15 mL min−1; the inlet of the QCM-D sensor chamber was connected to a glass vial containing an appropriate solution and the outlet to the suction line of a digital peristaltic pump (IsmaTec, IDEX). Fluorescence Recovery after Photobleaching (FRAP) Experiments. FRAP experiments were carried out on a spinning disc confocal microscope (UltraView ERS FRET-H-System, Perkin-Elmer), based on an Axiovert-200 M microscope (Zeiss) equipped with a PlanNeofluar 63×/1.4 oil objective and a Hamamatsu interline CCD camera (C9100−50). The samples were prepared as follows: an ∼1 × 1 cm2 silicon wafer with an about 1 nm thick native silica layer was rinsed with ethanol, dried with N2, and cleaned in UV/Ozone
ProCleaner for 30 min. The sample was then covered with an appropriate lipid solution for 10 min, rinsed with 1 mM or 150 mM NaCl according to the ionic concentration of the micelle solution and placed in a closed, temperature controlled chamber, which kept the sample at 30 °C during the experiments. During the FRAP measurements, an area of 100−120 μm was scanned every 1 s, and a selected circular area of ∼10 μm diameter was bleached (200−300 iterations at full laser power at 488 nm, an argon laser source). Before and immediately after bleaching, a sequence of images was acquired using an argon−krypton laser source at 568 nm. Microsoft Excel was used to extract the ROI integrated pixel intensities from raw data files. Topography and Force Profiles. Topography and force profiles were measured using AFM (Nanoscope 3D Multimode, Veeco). The images were acquired using SNL (Bruker) cantilevers with a spring constant of 0.06 N m−1 in contact mode. The sample preparation procedure was identical to that for FRAP measurements; however, in this case, the substrate was QCM-D silica crystal. The NaCl concentration was the same during the scans and the micelle fusion stage. The AFM scans were carried out in lipid-free solutions. Force profiles were acquired using commercially available silica probes (Novascan) of 10 μm diameter attached to a cantilever with spring constant of 0.6 N m−1. The approach and retraction velocities were 1 μm s−1, which induced a negligible drag force on the cantilever (≪0.1 nN) compared to the measured interaction forces (∼1 nN).34 The force profile was converted to interaction energy (W) using Derjaguin approximation,35 W = F/(2πR), where F is the measured force, and R is the radius of the bead. Prior to these experiments, the probes were gently rinsed with chloroform, dried in air, and cleaned in UV/Ozone ProCleaner (BioForce Nanoscience) for 30 min. GLH-20 membrane was prepared on the silica probes by covering it with the appropriate lipid solution for 10 min, and then rinsing with a NaCl solution of the same ionic strength (lipid-free). The temperature of the sample was set at 30 °C during all the measurements. More information about force measurements and their interpretation can be found elsewhere.9,36 ζ-Potential and Dynamic Light Scattering (DLS) Measurements. ζ-Ptential and DLS measurements were carried out using a Zetasizer Nano ZS (Malvern) in a disposable cell at 30 °C. The scattered intensity of the 633 nm laser beam was recorded at an angle of 173° relative to the incident beam. The size of the particles was determined using Zetasizer 7.01 software. Cryo-Transmission Electron Microscopy (TEM). Samples for TEM were prepared by placing a drop of the solution on a 300-mesh Cu grid coated with a porous carbon film (Lacey substrate, Ted Pella Ltd.) in the 30 °C chamber. The grid was blotted to remove excess liquid and plunged into liquid ethane using an automatic plunger (Leica EM GP) to vitrify the solution. Vitrified samples were transferred to a cryo-holder (Gatan 626) and were examined at 1154
dx.doi.org/10.1021/la304484p | Langmuir 2013, 29, 1152−1161
Langmuir
Article
Figure 3. Cryo-TEM images of GLH-20 aggregates in 150 mM (A and B) and in 1 mM NaCl (C) solutions. Images A and B show micrometer-large secondary aggregates of entangled thread-like micelles of GLH-20 formed in 150 mM NaCl at two different magnifications. In image C only spherical micelles, a few nanometers large, can be observed. −178 °C on an FEI Tecnai 12G2 TWIN TEM. The images were recorded on a Gatan 794 charge-coupled device camera at 120 kV in low-dose mode.
about 1.0 nm3 and 4.7 nm, respectively. Thus the skeleton volume (v) and length (lc) per head are half these values, i.e., v = 0.5 nm3 and lc = 2.4 nm. The cryo-TEM images shown in Figure 3 indeed indicate that the diameter of both the spherical micelles in 1 mM NaCl (Figure 3C) and the thread-like micelle in 150 mM NaCl (Figure 3A,B) was about 5 nm, i.e., as expected, twice the lc. The headgroup area (a0) of GLH-20 should be commensurate with that of cationic DMTAP; see Figure 2), which was shown to be about 0.60−0.70 nm2 in NaCl solutions.40,41 Thus the parameter SF (= v/(a0lC)) of GLH-20 could be estimated to be between 0.35 and 0.30.38 Recalling that for SF < 1/2 surfactants are expected to pack into micelles, this estimate seems to agree with the cryo-TEM and DLS results. In fact, the use of the headgroup area of DMTAP for GLH-20 might underestimate a0 for GLH-20, since the latter contains an additional acetyl-ethyl group attached to the amine nitrogen. The actual SF might then be even smaller, and indeed spherical micelles seen in Figure 3C suggest SF < 1/3.38 For comparison, the SF of DMPC and DMTAP in the same solutions is above 1/2, which favors the formation of vesicles and bilayers. Interestingly, Popov et al. observed that in aqueous solutions GLH-20 could also form vesicles and not micelles.31 The difference from the present results could be explained by different concentration of the bolalipid that was about 5 times higher than that used in this study. The vesicles formed in the study of Popov et al. were not stable, and the addition of cholesterol and cholesteryl hemisuccinate was necessary to stabilize the vesicles. These additives, not used in this study, are
■
RESULTS AND DISCUSSION Aggregation of GLH-20 Bolalipid in Bulk Solution. Figure 3 shows cryo-TEM images of aggregates formed from GLH-20 in solutions containing two concentrations of NaCl, 1 mM and 150 mM, at neutral pH. At these ionic strengths GLH20 appears to self-assemble into spherical and thread-like micelles, respectively. The thread-like micelles in 150 mM NaCl are packed in secondary aggregates several micrometers in size (see Figure 3A and B). This is in good agreement with DLS results that revealed the presence of 3 ± 2 nm aggregates in 1 mM NaCl and a highly polydisperse population dominated by several micrometer-sized aggregates in 150 mM NaCl (see Supporting Information, Figure S2). Previous reports have also shown that symmetrical single-chain bolalipids in aqueous solutions can often self-assemble into such micelles (see Introduction).28,29,37 Micelle formation, rather than vesicles, under these conditions is also consistent with the molecular shape of GLH-20 in terms of SF.38 Indeed, the skeleton of GLH-20, i.e., the hydrophobic middle part between the two OH groups (see Figure 2), is a chain of 36 atoms, whose dimensions can be estimated using the known dimensions of the molecular units.39 Assuming the ester groups on the aliphatic tail have a negligible effect on the packing geometry, the total aliphatic volume and fully stretched length of the chain are roughly estimated to be 1155
dx.doi.org/10.1021/la304484p | Langmuir 2013, 29, 1152−1161
Langmuir
Article
headgroup is 0.60−0.70 nm2, the mass of GLH-20 including the ∼0.5 nm thick aqueous interlayer is indeed estimated to be within the range of 350−400 ng cm−2. It was concluded that GLH-20 adsorbed onto the surface in all tested ionic concentrations with the adsorbed mass in the range of a single SBM. The change in dissipation (bottom plot in Figure 4) was insignificant in 1 mM and 25 mM NaCl, suggesting a fairly rigid layer; however, in 150 mM NaCl some increase in dissipation was recorded. The slightly increased dissipation at 150 mM NaCl may be associated with a somewhat thicker aqueous interlayer due to larger concentration of counterions at higher ionic strength, but may also suggest adsorption of some unruptured micelles. To rule out the latter possibility, the flatness and homogeneity of the layer formed on the QCM crystal was examined by AFM. AFM images (see Supporting Information, Figure S1) showed no significant change in topography, and closely resembled the morphology of the clean silica surface. The root-mean-square (RMS) of the surface roughness was 1.12, 1.14, and 0.94 nm for the pristine crystal and the crystals after micelle fusion in 1 mM and 150 mM NaCl, respectively. These values are far below the dimensions of lipid membranes or micelles and indicate no morphological change or presence of unruptured micelles after fusion. The AFM analysis thus suggests that the adsorbed layer was reasonably flat and homogeneous and uniformly coated the silica surface. The presence of a contiguous SBM on the surface could be further verified by FRAP, which examines the lipid lateral diffusivity of lipids. Compared to unruptured vesicles, individual lipids acquire the ability to diffuse laterally over long distances, and diffusivity increases significantly after formation of a contiguous SBM.46,47 Figure 5 displays typical FRAP kinetics
known to stabilize vesicles and promote transition from micelles to larger, more extended aggregates with a smaller surface curvature, such as vesicles.42 They could also reduce repulsion between cationic head groups of GLH-20 (see the Lipid−Substrate Interactions section). The ζ-potentials of GLH-20 and DMPC aggregates in different NaCl solutions are summarized in Table 1. At all examined ionic strengths, GLH-20 micelles are positively charged, in contrast to DMPC vesicles that bear a nearly zero charge in these conditions. The positive charge of GLH-20 should affect its state of aggregation in solution and reduce SF via charge repulsion between the headgroups. More important in the present context is that the positive charge must play an important role in the formation and stabilization of SBM on negatively charged silica and mica surfaces (see below). Self-Assembly of GLH-20 on Silica. Self-assembly of GLH-20 SBM on silica was examined by QCM-D using the same solutions that were employed for cryo-TEM examination. Figure 4 shows the shift in frequency and dissipation after
Figure 4. QCM-D results, frequency (top) and dissipation (bottom) shifts, upon exposure of silica substrate to GLH-20 solutions containing indicated concentrations of NaCl at 30 °C.
exposing quartz crystals coated with a 50 nm silica layer to GLH-20 solutions. The observed QCM response is quite different from vesicle fusion. Previous reports showed that, as intact vesicles are adsorbed, the resonance frequency decreases up to a certain coverage, when vesicles begin to rupture.43,44 Thereafter, the frequency increases due to release of encapsulated water, thus the frequency shift during vesicle fusion goes through a minimum. Since micelles contain no encapsulated water, the frequency during micelle fusion should decrease monotonically until the maximum coverage is obtained, as indeed observed. The mass of the adsorbed layer was estimated from the third overtone using the Sauerbrey equation and was found to be 252 and 358 ng cm−2 for 1 and 150 mM NaCl respectively.45 Anderson et al. showed that the third overtone yields a fairly reasonable estimate of the mass of DMPC bilayer on silica.14 Assuming that the average area per
Figure 5. FRAP results of GLH-20 on silica (30 °C) in 150 mM NaCl. The inset shows a sequence of fluorescence images revealing the lipid mobility on the surface. The fit in the plot corresponds to diffusion coefficients 1.3 ± 0.6 × 10−12 m2 s−1. The kinetics in 1 mM NaCl (not shown) was about the same.
and fluorescence images of silica substrate covered with GLH20 and Rh-PE (
