Preparation of Solvent-Free, Pore-Spanning Lipid Bilayers - American

May 27, 2011 - facturer (Asylum Research). Figure 1. Cellular membranes (a) are supported by an intricate scaf- folding of the cytoskeleton, leading t...
0 downloads 0 Views 5MB Size
ARTICLE pubs.acs.org/Langmuir

Preparation of Solvent-Free, Pore-Spanning Lipid Bilayers: Modeling the Low Tension of Plasma Membranes Marta Kocun,† Thomas D. Lazzara,‡ Claudia Steinem,‡ and Andreas Janshoff*,† †

Institute of Physical Chemistry and ‡Institute of Organic and Biomolecular Chemistry, University of G€ottingen, Tammannstrasse 6, 37077 G€ ottingen, Germany

bS Supporting Information ABSTRACT: Plasma membrane tension, produced by the underlying cytoskeleton, governs many dynamic processes such as fusion, blebbing, exo- and endocytosis, cell migration, and adhesion. Here, a new protocol is introduced to model this intricate and often overlooked aspect of the plasma membrane. Lipid bilayers spanning pores of 600 nm radius were prepared by adsorption and spreading of giant unilamellar vesicles (GUVs) on moderately hydrophilic porous substrates prepared by goldcoating and subsequent self-assembly of a mercaptoethanol monolayer. Rupture of GUVs formed tens of micrometer sized pore-spanning membrane patches displaying low tension of σ e 3.5 mN m1 and lateral diffusion constants of about 8 μm2 s1. Site-specific force indentation experiments were performed to determine membrane tension as a function of lipid composition: for pure DOPC bilayers, a tension of 1.018 ( 0.014 mN m1 was measured, which was increased by the addition of cholesterol to 3.50 ( 0.15 mN m1. Compared to DOPC, POPC bilayers displayed a larger tension of 2.00 ( 0.09 mN m1. Addition and subsequent partitioning of 2-propanol was shown to significantly reduce the membrane tension as a function of its concentration.

’ INTRODUCTION Membrane deformation processes, which include endocytosis, exocytosis, cell motility, spreading, membrane trafficking and repair are regulated by cellular membrane tension.18 The main contribution to the overall tension in the plasma membrane originates from its adhesion to the underlying actin cytoskeleton, while the inherent tension of the lipid bilayer itself is at least 1 order of magnitude smaller and therefore negligible. On a mesoscopic scale, native lipid bilayers connected to the cytoskeleton have been shown to exhibit a moderate lateral tension (σ = 0.010.1 mN m1), high lateral mobility (D = 510 μm2 s1), and mechanical robustness.6,9 The plasma membrane is essentially elastically “decoupled” by the associated cytoskeleton, in particular, the actin cortex with a mesh size ranging between 30 and 300 nm, which confines the mechanical properties to these areas.10 The actin mesh is the one that allows the plasma membrane to act as a locally heterogeneous elastic shell, which performs various functions pertaining to cell migration, blebbing, exocytosis and endocytosis.6,9 The particular architecture of the eukaryotic plasma membrane, composed of a fluid lipid bilayer connected to the actin cortex, has been a challenge to model using artificial membrane systems. An elastic decoupling observable in plasma membranes, i.e., a spatially distinct mechanical response enabled by the architecture of the actin mesh, is not reproduced by membrane mimics such as liposomes that exhibit only global elastic properties.11 Besides liposomes, the most prevailing artificial models currently used for membrane studies are black lipid membranes (BLMs) and solid-supported (tethered) lipid bilayers.1215 While r 2011 American Chemical Society

liposomes and solid-supported bilayers are essentially tension-free, BLMs contain residual solvent and are therefore not well suited to study membrane processes that are strongly influenced by membrane tension.1618 By using structured silicon nitride (Si3N4) supports reminiscent of the actin mesh of the cell cortex (Figure 1), it is in principle conceivable to produce both elastically decoupled free-standing bilayers and membranes with low lateral tension by modifying the adhesion strength of the bilayer with the pore-rims. As of yet, pore-spanning bilayers have been mainly prepared by painting lipids dissolved in an organic solvent onto porous substrates, where the interpore surface is functionalized with a hydrophobic self-assembled monolayer, creating hybrid membranes (nano-BLMs). Similar procedures were also used to prepare free-standing polymer membranes.19 Other methodologies used to generate pore-spanning membranes include vesicle spreading induced by contact with a hydrophobic porous interface or vesicle spreading mediated by strong electrostatic forces.2022 Previously, we studied the local mechanical response of hybrid nano-BLMs to site-specific indentation experiments. These hybrid pore-spanning bilayers showed an extremely high lateral tension (∼20 mN m1) due to the large gain in adhesion energy on the pore-rim which prestressed the bilayer close to rupture. It is, however, much more difficult to prepare defined Received: January 24, 2011 Revised: March 25, 2011 Published: May 27, 2011 7672

dx.doi.org/10.1021/la2003172 | Langmuir 2011, 27, 7672–7680

Langmuir

Figure 1. Cellular membranes (a) are supported by an intricate scaffolding of the cytoskeleton, leading to a laterally tensed membrane depicted in panel b. (c) A pore-spanning membrane patch mimics the stressed membranes by the finite adhesion to the pore-rims.

pore-spanning bilayer structures composed of natural phospholipids such as POPC, with moderate tension (∼1 mN m1) suitable to mimic tension observed in the plasma membrane. As of yet, we have only been able to create pore-spanning bilayers with low tension (0.15 mN m1) on nonfunctionalized porous silicon23 with DPhPC. We were, however, not able to cover the pores with DMPC, POPC, DOPC, etc., all of which possess naturally occurring fatty acids. Along the same lines, we showed that artificial lipids such as N,N-dimethyl-N,Ndioctadecylammonium bromide and 1,2-dioleoyl3-(trimethylammonio)propane chloride can be used to create pore-spanning membranes.22 Again, biologically relevant lipids that are neutral or bear a negative charge could not be used following that procedure. In this present study we were able, for the first time, to create nonhybrid, pore-spanning bilayers consisting of POPC, DOPC, and DOPC/cholesterol mixtures. The prepared bilayers exhibit significantly lower tension (σ e 3.5 mN m1) and high lateral mobility (D = 8 μm2 s1) and are thus particularly interesting for biological applications where plasma membrane mimics are required, i.e., for the functional reconstitution of proteins that “sense” lateral tension such as mechanosensitive channels.

’ EXPERIMENTAL SECTION Materials. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), and 1,2-diphytanoyl-snglycero-3-phosphocholine (DPhPC) were purchased from Avanti Polar Lipids (Alabaster, AL), and cholesterol was purchased from Sigma-Aldrich (Steinheim, Germany). All lipids were dissolved in chloroform (Roth, Karlsruhe, Germany). Membranes were labeled (0.1 mol %) with either 2-(4,4-difluoro-5-methyl-4-bora-3a,4a-diaza-s-indacene-3-dodecanoyl)1-hexadecanoyl-sn-glycero-3-phosphocholine (β-BODIPY 500/510 C12-HPC (Invitrogen, Karlsruhe, Germany) or sulforhodamine-1,2-dihexanoyl-sn-glycero-3-phosphoethanolamine (TexasRed-DHPE, SigmaAldrich). Mercaptoethanol (Sigma-Aldrich), ethanol (Sigma-Aldrich),

ARTICLE

2-propanol (VWR, Fontenay-sous-Bois, France), tetradecanethiol (Fluka, Buchs, Switzerland), and sodium hydrogen phosphate dihydrate (AppliChem, Darmstadt, Germany) were all used without further purification. Water used in preparation of buffers was filtered by a Millipore system (Milli-Q System from Millipore, Molsheim, France; resistance >18 MΩ cm1). Vesicle Preparation. Giant unilamellar vesicles (GUVs) were prepared by electroformation.24,25 Briefly, 50 μL of 1 mg mL1 POPC, DOPC, DOPC/cholesterol, or DPhPC lipid solution was deposited on indium tin oxide (ITO) slides and spread uniformly using a sterile needle. The ITO slides were placed under vacuum for at least 3 h at 64 C to remove any residual solvent. Subsequently, two ITO slides covered with lipid films, conductive copper tape, and a 1 mm thick square Teflon spacer between the slides were used to create a sealed chamber (approximately 1.2 mL volume), which was filled with 0.3 M sucrose solution. The chamber was connected to a waveform generator, and a cycle was carried out at 12 Hz, where voltage increments were performed every 60 s: beginning at 0.05 V, followed by 0.01 V voltage steps until 0.2 V was reached, and finally the voltage was increased by 0.1 V voltage steps until a constant 1.6 V was reached (total duration was 3 h). At the end, a square 5 Hz wave was applied for 5 min. The GUVs were transferred to a plastic vial and stored at 4 C for up to 2 weeks.26 Porous Substrates. Silicon nitride substrates with pore radii of 600 nm were purchased from fluxxion B.V. (Eindhoven, The Netherlands). The porous substrates were coated with a 23 nm thick layer of chromium followed by a 1015 nm thick layer of gold (Bal-Tec MCS610 evaporator equipped with Bal-Tec QSG 100 quartz film thickness monitor). The gold-coated substrates were subsequently oxygen-plasma (1 min) and argon-plasma (1 min) treated and placed in a 20 mM ethanolic mercaptoethanol self-assembly solution for 1 h. The substrate was then rinsed with ethanol prior to use. Coverage of the gold with mercaptoethanol SAM was controlled by impedance spectroscopy (SI/260, Solartron) and contact angle measurements (see Supporting Information, Figures S1 and S2). Pore-Spanning Membranes. Directly after being rinsed with ethanol, the functionalized substrate was placed in a homemade Teflon holder filled with ethanol. Ethanol was then replaced by phosphatebuffered saline (PBS: 20 mM NaH2PO4/Na2HPO4, 100 mM NaCl, pH 7.4), and 20 μL of GUVs were added. GUVs migrated (due to the density difference between 0.3 M sucrose inside and PBS buffer outside the vesicles) to the surface and spread, creating lipid bilayer patches. The resulting patches were observed by fluorescence microscopy, and their location was determined in relation to the rows-and-columns pattern of the porous substrate. The DPhPC hybrid pore-spanning membranes were prepared by placing a gold-coated porous silicon substrate in 10 mM ethanolic tetradecanethiol solution for 1 h. The substrate was rinsed with ethanol, placed in a Teflon holder filled with ethanol, which was subsequently replaced with PBS buffer before DPhPC GUVs were added. Atomic Force Microscopy (AFM). AFM imaging and force indentation curve acquisition was performed in PBS buffer with a topview optics MFP-3D instrument (Asylum Research, Santa Barbara, CA) and silicon nitride AFM probes (MLCT-AU type, 310 μm long C lever) purchased from Bruker AFM Probes (Mannheim, Germany) with spring constants of 0.010.04 N m1 and a tip radius of 1030 nm. The spring constant of each cantilever was calibrated prior to experiment with the thermal noise method according to Hutter and Bechhoefer, refined by Butt and Jaschke.27,28 In brief, the thermal noise method makes use of the equipartition theorem, which relates the thermal noise Æz2æ in the cantilever’s position to the energy of the thermally excited beam. Generally, the thermal noise method is limited by the sensitivity of the deflection signal. The calibration factor (inverted optical lever sensitivity) is obtained from a force curve against a rigid substrate (glass slide). The calibration procedure is integrated in the MFP-3D IGOR software of the manufacturer (Asylum Research). 7673

dx.doi.org/10.1021/la2003172 |Langmuir 2011, 27, 7672–7680

Langmuir

ARTICLE

Fluorescence Microscopy. For the acquisition of fluorescence images, an Olympus BX-51 upright optical microscope (Olympus Germany GmbH, Hamburg, Germany) equipped with filters for TexasRed and BODIPY fluorescence (UMNG2 and UMNB2, Olympus Germany GmbH, Hamburg, Germany) and immersion objectives with 40 (Olympus, LUMPlanFL 40, N.A. = 0.8) and 100 (Olympus, LUMPlanFL 100, N.A. = 1.00) magnifications were used. An upright confocal laser scanning microscope (LSM 710 Axio Examiner, Carl Zeiss MicroImaging GmbH, Jena, Germany) equipped with 493 589 nm and 598690 nm filters and water immersion objectives with 40 (Zeiss, wPlan-APOCHROMAT, N.A. = 1.0) and 63 (Zeiss, wPlan-APOCHROMAT, N.A. = 1.0) magnifications were used for performing z-scans and fluorescence recovery after photobleaching experiments. Fluorescence Recovery after Photobleaching (FRAP). During a FRAP experiment, a small area (approximately 8 μm diameter) of a membrane patch was bleached within 3 s with a high intensity laser pulse. Fluorescence recovery of this area was monitored at the same time as the fluorescence intensity of a reference area on a neighboring, unconnected patch. The reference fluorescence intensity is collected from a patch in close proximity to the bleached patch to overcome the system’s limitation related to the limited lipid reservoir of a given patch. Therefore, the intensity of the entire membrane patch slightly decreases during the bleaching period. Due to the noncontinuous fluorescence of the membrane patch on the porous surface (due to fluorescence quenching on the gold-covered pore-rims), the Gauss radius of the bleached area on the porous substrate was determined by fitting an intensity profile using a MATLAB fitting procedure. Lipid diffusion coefficients were obtained using the theory of Axelrod et al.29 eq 1 was fitted to the experimentally obtained recovery curves:   ¥ 1 ð1Þ ðð  KÞn =n!Þ FK ðtÞ ¼ ðQP0 C0 =AÞ 1 þ nð1 þ 2t=τD Þ n¼0



The fluorescence at time t (FK(t)) is valid for a Gaussian intensity profile with τD = ω2/4D. D is the lateral diffusion constant of the lipids, Q is the product of all quantum efficiencies of light absorption, emission, and detection, A is the attenuation factor of the beam during observation of recovery, P0 is the total laser power, ω is the half-width at e2, C0 is the initial fluorophore concentration, and K is the bleaching induced at time T (K = RTI(0), where RTI(0) is the rate of bleaching in the center).29

’ RESULTS AND DISCUSSION Formation of Pore-Spanning Membranes. The goal of this study was to create solvent-free, pore-spanning lipid bilayers on a hydrophilic support that display considerably reduced lateral tension in order to mimic lipids in a native plasma membrane. As a support for the preparation of pore-spanning membranes, we used silicon nitride (Si3N4) porous substrates displaying a hexagonal pore arrangement with pore radii of 600 nm. First, a thin gold layer was evaporated on the pore-rims and exposed to oxygen plasma, followed by argon plasma in order to clean the gold and the Si3N4 surfaces.30 By evaporating only a thin (1015 nm) gold layer, whose thickness was controlled by quartz crystal microbalance, the gold coverage is restricted predominantly to the pore-rims. Consequently, a continuous self-assembled monolayer (SAM) of mercaptoethanol is formed only atop the porous substrate, producing moderately hydrophilic pore-rims. The integrity of the SAM was monitored by impedance spectroscopy, revealing an average surface coverage of about 90% (see Supporting Information, Figure S2). The hydroxyl-terminated SAM formed on the gold surface is a prerequisite for deformationinduced rupture of GUVs which are subsequently added to the

Figure 2. (a) Chemical functionalization of porous Si3N4 substrates. First, 23 nm of chromium are evaporated, followed by a 1015 nm gold layer (steps 1 and 2). (b) The gold-covered surface is then oxygen and argon plasma treated and placed in a 20 mM self-assembly solution of mercaptoethanol (step 3). (c) Subsequently, the functionalized substrate is placed in PBS, GUVs are immediately added and membrane patches are formed (step 4). Inset: enlargement of a functionalized porerim showing the lipid bilayer drawn to scale. The position of the bilayer on the pores is discussed in more detail in the text.

measuring chamber. When GUVs are added to a clean (nonmodified) Si3N4 porous substrate, the vesicles also spread, but all resulting pore-spanning membranes rupture within a few seconds to a few minutes. The bilayer covers the pore-rim and pore interior instead due to the gain in adhesion energy that disfavors pore-spanning membranes. As the hydrophilic hydroxyl-terminated SAM on Au is chemically distinct from the Si3N4 surface, the phospholipid bilayer interacts in this case preferentially with the pore-rims and avoids contact with the pore interior. Hence, once added onto the functionalized Au-covered porous substrates, the GUVs adsorb onto the surface, flatten, and deform until instability-induced rupture occurs and a pore-spanning bilayer patch is formed (Figure 2). Characterization of Pore-Spanning Membranes. The process of GUV adsorption, deformation, and spreading was observed by means of confocal laser scanning fluorescence microscopy (CLSM) using fluorescently labeled vesicles. Figure 3a shows time-resolved fluorescence images of the adsorption, deformation, and spreading of two POPC GUVs on the mercaptoethanol-functionalized porous substrate. As they burst, the lobes unwind and during the unfolding cover the porous surface. From videomicroscopy (see Supporting Information, movie S1), we envision the process of bilayer formation as follows: the higher interior sucrose density of the vesicles in comparison to the surrounding PBS results in vesicle sedimentation to the porous surface. Similarly to the spreading mechanism on SiO2 planar substrates proposed by Hamai et al.,31 the vesicle initially tumbles along the surface until it is close enough to overcome the activation barrier and attaches firmly to the surface. The increase in adhesion area flattens the sessile liposome until adhesion energy matches the lysis tension, leading to vesicle rupture and planar bilayer formation (Figure 3b). In detail, for strong adhesion, the energetic competition, which determines the conformation of the sessile vesicle, involves the balance between adhesion energy and elastic stretching (Ka: area compressibility modulus, approximately 102 mJ m2) of the membrane. If the elastic tension (σ = Ka ΔA/A0) equals the lysis tension, reached at a relative area change ΔA/A0 = 5%, the vesicle ruptures.32 We typically found 7674

dx.doi.org/10.1021/la2003172 |Langmuir 2011, 27, 7672–7680

Langmuir

Figure 3. (a) Time-resolved fluorescence images showing the spreading of GUVs on a mercaptoethanol-functionalized porous surface. (b) Schematic representation of the vesicle spreading process: (i) vesicle sedimentation on the surface, (ii) adhesion, (iii) flattening and filopodia formation, (iv) vesicle rupture and bilayer patch formation. (c) Fluorescence images showing symmetric and (d) asymmetric bilayer patches. (e) Illustration of the filopodia extending from the sessile giant liposome giving rise to excess lipid material on the pore-rims. (Scale bars: 5 μm) (see Supporting Information, Figure S4).

that the size of the final patch (Figure 3a, 200 ms) is larger than the initial adhesion area of the GUV (Figure 3a, 67 ms). However, the surface area of the pore-spanning bilayer patches is smaller than the total vesicle area. We believe that because of the violence of the rupture process, most of the lipid membrane that is not in proximity to the initial contact area does not deposit as pore-spanning bilayer. We attribute this loss of bilayer material to the uncontrolled nature of the spreading mechanism, which also results in a transformation of excess membrane into smaller vesicles upon rupture and multilayer deposition on the outer pore-rims. In accordance with the work of Hamai et al.,31 we observed two rupture pathways for isolated GUVs yielding symmetric (Figure 3c) and asymmetric patches (Figure 3d), when the rupture initiation site is at the top (symmetric) of the adsorbed vesicle or closer to the side (asymmetric) of a vesicle, respectively. The presence of a brighter noncontinuous halo around the bilayer patch and its overall appearance on the functionalized porous substrate are governed by the thickness-limited quenching of the fluorescence by the underlying gold film (Figure 3e). Only membranes spanning the pore area display detectable fluorescence (Figure 3e, area i). Fluorescence from membranes close to the gold surface is quenched up to a thickness of 1015 nm (Figure 3e, area ii).3335 However,

ARTICLE

Figure 4. (a) Correlation between a CLSM fluorescence image of a pore-spanning membrane labeled with 0.1 mol % Bodipy-DHPE dye and (b) AFM image of the same area. (c) Cross-sections from AFM images show (i) membrane-spanned pores where three out of five pores (from right to left) are covered and (ii) the presence of a single bilayer membrane (approximately 4 nm height difference).

excess lipid material at the outer boundary of the membrane patch produces brightly fluorescent pore-rims attributed to surface-enhanced fluorescence originating from fluorophores that are sufficiently far away from the gold-coated pore-rims (Figure 3e, area iii). We propose that adhesion of the GUV on the pore-rims immediately drives the extension of filopodia along the pore-rims without suspending the pores. This is similar to the spreading of epithelial cells on the same porous substrates, as recently reported.36 After vesicle rupture, the excess lipid material appears as bright fluorescence due to surface-enhanced fluorescence in proximity to the gold surface (also visible in Figure 5). Only those pores that show fluorescence (Figure 4a) display a reduced penetration depth during AFM imaging (Figure 4b and 4c, line profile i) since they are covered with a membrane. The thickness of the membrane is obtained from the line profile acquired on the flat part of the sample (Figure 4b and 4c, line profile ii). A thickness of 45 nm is usually found and is indicative of a single bilayer (see Supporting Information, Figure S3). Close examination of the single plane CLSM fluorescence image (Figure 5a, top image) and the corresponding fluorescence intensity profiles (Figure 5a, bottom image, i and ii) provide further information about the thickness of the lipid membrane spanning the pores that is otherwise not directly accessible by AFM imaging. The image shows a sessile vesicle (Figure 5a, middle) attached and hence partially flattened on the porous substrate, an already formed bilayer patch beside it formed from another adjacent vesicle (Figure 5a, right side) and pore-rim areas that have the highest fluorescence intensities of the image. Because the vesicle that is in contact with the porous surface is still intact, the vesicle area spanning the pores indicates a single bilayer per definition. 7675

dx.doi.org/10.1021/la2003172 |Langmuir 2011, 27, 7672–7680

Langmuir

ARTICLE

Figure 5. (a) Confocal fluorescence laser scanning microscopy images merged to a z-stack were used to obtain a 3D-image of a mercaptoethanol functionalized substrate where three stages of GUV spreading were captured in a single scan: a sessile vesicle, pore-rim with excess lipid material, and pore-spanning membrane (indicated by crossing lines). The sessile vesicle (i) located to the left of a previously ruptured vesicle (ii) exhibits a dome shape confirmed by the orthogonal cut view (top and right side panels in panel a). The highly fluorescent pore-rims surrounding the vesicle result from surfaceenhanced fluorescence originating from filopodia consisting of double-bilayers which possess fluorophores sufficiently far away from the underlying gold film. A patch of pore-spanning membrane is visible on the right side of the vesicle. The fluorescence intensity profile obtained from the line marked with an arrow shows that the fluorescence intensity originating from the lower membrane of the sessile GUV (shaded area i) is identical to that of the porespanning membrane (shaded area ii), which confirms that only a single bilayer spans the pores. (b) Force volume AFM image of a porous surface where only one pore is not covered with a membrane. Comparison between indentation curves performed on the flat surface (b), pore-spanning membrane (), and an empty pore (9). In contrast to the empty pore where the tip must travel about 800 nm before reaching a predetermined set point, the contact point on the membrane covering the pore is identical with that on the adjacent rim indicating that the membrane neither bulges out of the pores nor invaginates into them.

The fluorescence intensity from that sessile GUV (Figure 5a, shaded area i) is identical to the fluorescence intensity originating from the pore-spanning patch located on its right side (Figure 5a, shaded area ii). This observation led us to the conclusion that, indeed, single lipid bilayers were obtained by the preparation procedure. Force volume imaging was performed on the pore-spanning bilayers to determine the vertical position of the lipid bilayer on the porous substrate (Figure 5b). A force volume image is obtained by producing maps of forceindentation curves on the surface. Each pixel of the image represents the distance traveled by the AFM tip before reaching its preset force. In order to reach a given force, the AFM tip travels less when the pore is covered by a membrane in comparison to an uncovered pore. Therefore, covered pores are represented by lighter areas, while darker ones correspond to uncovered pores. According to the contact point of the AFM tip with the membrane, which is identical to the position of the adjacent pore-rim, the bilayer membrane seems to span the pores without lining their insides or forming trapped lumps (Figure 5b, curves with symbols  and b). This behavior contrasts with that of hybrid nano-BLMs deposited on hydrophobic self-assembled monolayers, which line the gold coating inside the pores, thus covering the entire gold surface to gain the maximal adhesion energy regardless of the imposed curvature stress.23 We see this as an indication that

the adhesion of bilayers spread on mercaptoethanol-functionalized porous substrates is significantly weaker than on hydrophobic SAMs. We conclude that the vesicle spreading process of GUVs on hydrophilic OH-terminated mercaptoethanol functionalized porous substrates is very similar to what has been observed previously on flat SiO2 surfaces.37 Lateral Mobility of Pore-Spanning Membranes. An important characteristic of native membranes is the lateral mobility of the phospholipids in the two-dimensional plane of the bilayer. In hybrid membrane systems, where the bottom leaflet of the bilayer is replaced by a hydrophobic monolayer covalently attached to the pore-rims, the membrane is anchored and the diffusion of the lipids is limited to the top lipid leaflet of the bilayer. This anchoring at the pore-rims significantly decreases the overall lipid diffusion constant and alters the appearance of the fluorescence recovery curve of the FRAP experiments. In the preparation method presented here, the pore-rims are hydrophilic and the bilayer lipids are expected to diffuse with diffusion coefficients characteristic of bilayers on a planar hydrophilic surface or even GUVs. From FRAP experiments (Figure 6), an average diffusion coefficient of 8 ( 4 μm2 s1 was obtained, indicating that the observed pore-spanning bilayers are fluid and the lipids are highly mobile. The diffusion constant is in the range reported for giant liposomes (D = 1.16.3 μm2 s1).3840 In contrast, BLMs exhibit very high lateral diffusion coefficients due 7676

dx.doi.org/10.1021/la2003172 |Langmuir 2011, 27, 7672–7680

Langmuir to the presence of solvent (D = 20.6 ( 0.9 μm2 s1),41 while the lateral mobility of lipids in solid-supported membranes on glass (D = 14 μm2 s1),42 as well as in polymer-cushioned membranes (D = 0.881.13 μm2 s1)43 and polymer tethered membranes (D = 0.50.89 μm2 s1)43 is lower. Solvent-containing

Figure 6. Fluorescence recovery after photobleaching (FRAP) was performed on a pore-spanning POPC bilayer labeled with 0.1 mol % Bodipy-DHPE. (a) The fluorescence recovery of a bleached area (i) was recorded simultaneously to the reference area (ii). (b) A fit, according to the theory of Axelrod et al.,29 to the normalized fluorescence recovery curve provides a diffusion constant of 8 ( 4 μm2 s1. (Scale bar: 5 μm).

ARTICLE

pore-spanning hybrid membranes on porous substrates with pore diameters of 7 μm are highly mobile with diffusion constants of D = 14 μm2 s1,44 similar to those of BLMs. In contrast, solventfree, hybrid membranes on porous alumina substrates with pores of 60 nm in diameter exhibit diffusion constants of D = 7 ( 3 μm2 s1,45 close to the values found for GUVs and also found in our case. The immobile fraction of the lipids is