Quantification of nano-scale intermembrane contact areas by using fluorescence resonance energy transfer

Abstract

Nanometer-scale intermembrane contact areas (CAs) formed between single small unilamellar lipid vesicles (SUVs) and planar supported lipid bilayers are quantified by measuring fluorescence resonance energy transfer (FRET) between a homogenous layer of donor fluorophores labeling the supported bilayer and acceptor fluorophores labeling the SUVs. The smallest CAs detected in our setup between biotinylated SUVs and dense monolayers of streptavidin were ≈20 nm in radius. Deformation of SUVs is revealed by comparing the quenching of the donors to calculations of FRET between a perfectly spherical shell and a flat surface containing complementary fluorophores. These results confirmed the theoretical prediction that the degree of deformation scales with the SUV diameter. The size of the CA can be controlled experimentally by conjugating polyethylene glycol polymers to the SUV or the surface and thereby modulating the interfacial energy of adhesion. In this manner, we could achieve secure immobilization of SUVs under conditions of minimal deformation. Finally, we demonstrate that kinetic measurements of CA, at constant adhesion, can be used to record in real-time quantitative changes in the bilayer tension of a nano-scale lipid membrane system.

Intimate contact between 2 apposing lipid bilayer membranes is an essential prerequisite step for many biological processes of critical importance. Architectures like the neurosynaptic junction and the immune synapse that mediate cell–cell signaling, incorporate regions of tight intermembrane contact that span length scales of ≈0.5–10 μm (1, 2). Transient contact areas (CAs) of smaller dimensions (<100 nm) are present in the docking step that precedes exocytosis of vesicles during membrane trafficking or neurotransmitter release (3, 4).

Several studies have examined intermembrane junctions by using reconstituted model systems on solid supports (5, 6). Frequently, the junctions were formed by giant unilamellar vesicles (GUVs) that adhered on supported bilayers through specific or nonspecific interactions (7–9). The formation of CAs between adjacent membranes has been studied by a variety of optical techniques like FRET, fluorescence or reflection interference contrast microscopy that resolved topographical displacements with subnanometer accuracy but had a lateral resolution limited by optical diffraction (5, 6). These experiments provided a wealth of information on the parameters affecting local adhesion and structural rearrangements in the plane in such studies. The limited lateral resolution was not a significant constraint because the dimensions of the junctions were typically several tens of square microns. More recently, significant attention has been put on nano-scale systems that reconstitute membrane docking and fusion between single SUVs (10–12) or viruses (13, 14) and membranes. The CAs formed in such systems are important for characterizing the nature and potency of the adhesion-mediating molecules. However, despite theoretical efforts to describe the adhesion and fusion of SUVs (15, 16) an experimental quantification of the CAs for vesicles below the optical resolution still remains.

In the following, we present a method that allows measuring in real time and in a quantitative manner nano-scale CAs formed by SUVs on functionalized substrates. We measure FRET between fluorophores incorporated in the vesicle and complementary fluorophores conjugated to the substrate molecules inspired by previous work involving intermembrane FRET (5, 6, 17). An adhesive substrate consisting of a monolayer of streptavidin allows immobilization of vesicles and the size of the CA to be quantified by using FRET. We independently measure the linear size of the vesicle and the CA from which we infer the actual shape of a vesicle even if its dimensions are below the optical resolution limit. Accurate measurement of nano-scale CAs is achieved by using a pair of fluorophores having a Förster radius that closely matches the minimal separation between the surfaces thereby ensuring the highest sensitivity to deformation of the vesicle near the substrate. The measured degree of deformation for vesicles ranging in size from ≈70 nm to micrometer-size vesicles is compared with theoretical predictions for FRET between a spherical shell containing acceptors and a donor-labeled substrate.

The method presented here allows the quantification of intermembrane CAs with minimal radius of ≈20 nm. Such information is critical in characterizing intermembrane junctions formed during cell adhesion and vesicular or viral docking and fusion. Knowledge of the CA and the size of a vesicle allows us to infer both its shape and curvature of the bilayer. Nonspherical shape caused by adhesion can cause unwanted strain in the bilayer and increase the porosity of the bilayer, which is important to control in applications where vesicles are used as small reaction containers (18–20). Curvature in bilayers affects both molecular organization (21) and binding of membrane proteins (22). A precise characterization of the curvature of immobilized vesicles permits the use of surface-based assays and single immobilized vesicles to investigate the interaction of membrane curvature-sensitive proteins with bilayers of different curvatures (43). As we demonstrate here, changes of the CA subsequent to immobilization can be used to probe changes in the tension of lipid bilayers. Several classes of proteins and peptides upon binding or inserting to the bilayer exhibit a pronounced influence on membrane tension that often results in membrane deformation (23). CA could thus also be used as a sensitive nano-scale sensor of membrane tension, reporting on the consequences of protein–membrane interactions.

Results and Discussion

We investigate the contact mechanics of lipid vesicles immobilized on a flat substrate as illustrated schematically in Fig. 1A. The glass substrate is coated with a supported lipid bilayer. Next, we add streptavidin molecules, which bind to biotin lipids incorporated in the bilayer. Vesicles suspended in solution spontaneously dock onto the surface and bind to streptavidin through biotin lipids in the vesicle. To achieve FRET between the surface and immobilized vesicles, we use streptavidin labeled with Alexa Fluor 488 and incorporate complementary DiIC₁₈ fluorophores in the vesicle. The monolayer of donor-labeled streptavidin molecules on the surface yields a uniform background fluorescence, F_D, in the absence of acceptor fluorophores. At close contact between a lipid vesicle and the surface, the acceptors in the vesicle will quench the excited donors, resulting in a reduced fluorescence intensity, F_DA, see Fig. 1B and supporting information (SI) Fig. S1A. Direct excitation of the acceptor molecules permits calibration of SUV diameter as outlined in Materials and Methods and in more detail in ref. 24.

Fig. 1.

Measurements and calculations of the rate of energy transfer from a donor-labeled surface to an acceptor-labeled vesicle. (A) Schematic illustration of intersurface FRET within the contact region formed by an immobilized vesicle. (B) Micrograph of small vesicles, R ≈ 100 nm and FRET signatures acquired at the same location. (Upper) Exact determination of the vesicle size is obtained by calibrating the relation between intensity and the surface area of the vesicle. (Lower) A uniform background intensity originating from the donor labeled streptavidin is reduced within the contact region between vesicles and the surface. (Scale bar, 5 μm.) (C) The total transfer of energy from a single donor positioned a distance t away from a vesicle is obtained by summation of the energy transfer to all individual acceptors on the vesicle surface. An additional integration over the plane of donors gives the total transfer of energy from the surface to the vesicle. (D) Radial profiles for the FRET efficiency calculated for 3 vesicle radii, R = 50 nm (blue line), R = 200 nm (black line), and R = 500 nm (red line). (E) A surface plot of the calculated FRET efficiency for a spherical shell having radius R = 50 nm. The total FRET within the recessed region can be quantified from the integration performed in C and can be compared with experimental measurements. (Scale bar, 100 nm.)

FRET Between a Flat Surface and a Spherical Shell.

To obtain information about the CA and the 3-dimensional shape of a vesicle, we quantify the total reduction in donor intensity in a single footprint and relate it to a calculated value for the cumulative rate of energy transfer from a flat substrate to a nondeformed spherical shell. The relationship between the rate of transfer and reduction in donor intensity must be linear because every transfer event corresponds to a photon lost in the donor emission. To ensure that we do not have depletion of the donors in the CA due to electrostatic repulsion with the surface, we performed all experiments close to physiological salt concentration, i.e., 100 mM NaCl (25). Buffering the pH contributed further to the reproducibility of the experiments.

For 2 chromophores separated by a distance z the rate of energy transfer is given by

where R_o is the Förster distance and τ_D is the lifetime of the excited donor. To obtain the rate of energy transfer, k_ts, from a single donor to the all nearby acceptors, we add up the contribution from all acceptors on the spherical shell by integrating Eq. 1 with respect to z(θ) over the entire shell, Fig. 1C

where σ_A is the surface density of acceptors in the lipid vesicle and t is the distance from the donor to the center of the vesicle having size R. However, the curvature of the vesicle bilayer can introduce a spatial dependence on the transition dipole for the fluorophores in the vesicle. DiI fluorophores are known to have transition dipoles nearly parallel to the plane of the bilayer (26) but with significant fluctuations in orientations (27). Fluctuations of the fluid bilayer and in the molecular orientations averages out the curvature dependence of the FRET. Moreover, the random orientations of the donor fluorophores labeling the surface justifies the integration carried out in Eq. 2. The FRET efficiency for a donor on the substrate is given by

A graphical presentation of E_F as function of lateral position on the surface is plotted in Fig. 1D for several vesicle sizes and illustrated as a surface plot in Fig. 1E for a lipid vesicle having R = 50 nm. For the case of an undeformed vesicle, we obtain the cumulative energy transfer rate k_T by integration of Eq. 2 over the entire plane containing donor fluorophores, which gives (see Fig. 1C)

where the second term inside the brackets can be neglected for all vesicle sizes (R > 10 nm). σ_D is the area density of donor molecules distributed across the surface and A is the distance from the center of the vesicle to the nearest point on the surface (see Fig. 1C). To evaluate k_T in Eq. 4 we use R_o = 3.33 nm (28), σ_A = 0.13 nm⁻², τ_D = 4.1 ns (data provided by the SPEX Fluorescence Group, Jobin Yvon Inc., Longjumeau, France) and σ_D = 0.09 nm⁻². The shortest distance between the 2 surfaces, D = A − R, is measured in Fig. S1. The short Förster distance permits us to neglect the contribution from the distal leaflet in Eq. 4, which is located 4–5 nm further away from the donors. The constant relating energy transfer and reduction in donor intensity is measured in a flat geometry with acceptors and donors in apposing planes, see SI Text and Fig. S1. FRET signatures having dimension below the optical diffraction limit appear wider and more shallow relative to the theoretical profiles plotted in Fig. 1 D and E. The total reduction in donor intensity is, however, not affected by the point-spread function, which justifies application of Eq. 4 to subresolution objects. According to Eq. 4, the scaling between the energy transfer and vesicle size should be linear for undeformed vesicles. Information about the degree of deformation of the vesicles can therefore be obtained from the scaling relation between k_T and R. This scaling is independent of uncertainties in the constants entering Eq. 4 like D or R_o.

Nano-Scale Intermembrane CAs Quantified by FRET.

We perform a quantitative analysis of FRET footprints from lipid vesicles immobilized on a supported lipid bilayer. We collect measurements from single vesicles ranging in diameter from ≈70 nm to ≈1 μm as shown in Fig. 2 A and B. Each data point in Fig. 2B represents the integrated reduction in donor intensity, F_D − F_DA, over the entire footprint due to the presence of a single vesicle of a known size.

Fig. 2.

Experimental measurements of energy transfer and intermembrane CA for vesicle sizes ranging from below the optical resolution to micron-sized vesicles. (A) The top row: shows side views of several intensity peaks originating from the membrane dye in vesicles of different sizes. The integrated intensity from each vesicle is linearly proportional to the area of the bilayer containing the dye and can therefore be converted to diameter (24). Below each intensity peak is shown the reduction in donor intensity corresponding to the different vesicle sizes. The reduction in donor intensity can be used to determine the size of the CA assuming that vesicles form flat contact regions as depicted in Fig. 1A. (B) The reduction in donor intensity (F_D − F_DA) is quantified for each vesicle (circles) and plotted versus diameter of the vesicle. Calculations of the reduced donor intensity caused by undeformed vesicles (see Fig. 1C) show a linear scaling with vesicle size (green line). The blue line is a least-square fit of the function F_D − F_DA = cR^p, with p = 2.4 ± 0.3 and c = 2.1 as fitting parameters. The reduction in donor intensity scales faster than the vesicle area (p > 2), which indicates increasing deformation for larger vesicles. (Inset) A deformation parameter is defined by α = r_CA/R_T, where R_T is the distance from the center of the truncated vesicle to the contact line between the vesicle and the substrate, and r_CA is the radius of the contact area. Quantification of vesicle size and the area of the corresponding adhesion disk is sufficient to determine α plotted as the inset graph (see SI Text for details). The black line represents the measured noise from an area corresponding to 1 airy disk with diameter 284 nm.

To estimate the minimum size of the CA that can be detected, we accumulate the random noise across the airy disk for a point-scanning microscope, r_Airy = 0.4·λ/NA, where NA is the numerical aperture of the objective and λ is the wavelength of light. We determine the noise N sampled over n pixels by applying the central limit theorem N = $\sqrt{n}$·σ_n, where σ_n is the standard deviation among individual pixels. The resulting noise N is plotted as the black line in Fig. 2B, and we include data having signals with S/N ≳ 3. Using the value for FRET per area in this system, we estimate the smallest measurable CAs to have a minimum radius of r_CA ≈ 20 nm, which is comparable with CAs formed by viruses and liposomes in biological systems. We compare the measured reduction in donor intensity to calculated values for undeformed spherical vesicles (green line in Fig. 2B) that scale linearly with R (see Eq. 4). Because our vesicles are predominantly spherical, as determined from cryo-TEM experiments (20), we expect them to either become bound to the surface in a “pinned” state where the vesicle maintains its free shape or, if adhesion is strong, adopt the shape of a truncated spherical shell with formation of an adhesion disk of radius r_CA, see Fig. 2B Inset. The data points falling above the green line in Fig. 2B indicate some degree of deformation, which becomes more pronounced for larger vesicles. The CA scales with the area of the vesicle in the case of a constant degree of deformation. In such cases, the reduction in donor intensity will consequently scale with the area of the vesicle (≈R²). Interestingly, a least-square fit of a power-law function, c × R^p, to the data (blue line in Fig. 2B) reveals that the CA scales slightly faster than the area of the vesicle (p = 2.4 ± 0.3). By modeling the deformed vesicles as truncated spherical shells having a truncated radius R_T and a flat contact region, we characterize the extent of deformation with the parameter α = r_CA/R_T, see Fig. 2B Inset. To obtain the CA from the measured donor reduction, we subtract the measured value by the calculated value for an undeformed vesicle of the same size. This gives a good estimate for the CA even for small deformations. Outside the contact zone, the bilayer–substrate distance rapidly increases beyond the critical Förster distance R_o. Knowing the reduction in donor intensity per area (Fig. S1A), we can convert the reduction in donor intensity obtained from the fit in Fig. 2B to CA and, together with the vesicle size, obtain α (see SI Text). The degree of deformation, α, is plotted in Fig. 2B Inset and reveals that small vesicles on average deform only slightly, whereas larger vesicles adopt a more hemispherical shape on the substrate. Severe deformations of vesicles have previously been observed for SUVs immobilized on unprotected glass surfaces having high adhesion potential (29). The equilibrium shape of vesicles on adhesive substrates is determined by the gain in adhesion energy W_A at the cost of bending and stretching the bilayer. Higher tension in highly curved DSPC bilayers has been measured for curvatures ranging from R = 30 nm to R = 320 nm (30). Stretching of the bilayer upon adhesion is therefore more favorable for larger vesicles that are more relaxed. The adhesion energy scales with the area of the vesicle, whereas the bending energy is given by the bending rigidity κ, which is size independent. At a certain vesicle size determined by the adhesion energy and bending rigidity, R_a = $\sqrt{2\kappa /W_{\text{A}}}$, deformation becomes unfavorable (15). The bending rigidity for 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) membranes is κ = 10⁻¹⁹ J (31), and the adhesion energy between a biotinylated membrane and a streptavidin coated substrate at dense coverage has been reported to be 10 μJm⁻² (32). Using these values, we obtain R_a = 140 nm. Our measurements indeed confirm the theoretically predicted behavior; despite a significant heterogeneity, the percentage of deformed vesicles declined sharply for vesicle diameters <150 nm, see Fig. S2.

CA Is Sensitive to Interfacial Energy.

The dense streptavidin layers we used in Figs. 1 and 2 were designed to promote a measurable degree of deformation. Intermembrane attraction, however, can be easily modulated to decrease vesicle deformation if this is desirable. An easy method to reduce deformation is to dilute the density of receptor proteins on the surface (19, 20). However, this would also dilute the layer of donors, making quantification of FRET more challenging. An alternative strategy is to modulate the attraction between vesicles and the substrate by including a low percentage of lipids anchored to polyethylene glycol (PEG) polymers or alternatively decorate the glass substrate with a layer of PEG, which is depicted schematically in Fig. 3 A–D. The PEG polymers provide an entropic barrier against adhesion (7, 33) but can still permit the vesicles to become immobilized through biotin linkers conjugated to the PEG. We examine the effect of adsorbing PEG on the substrate or using head group-linked PEG₂₀₀₀ polymers corresponding to either isolated polymers on the bilayer (1 mol %) or overlapping (5 mol %) (34) forming an extended brush.

Fig. 3.

Adhesion of SUVs to a substrate is reduced in the presence of PEG polymers. (A) Vesicles containing biotinylated lipids are tethered to a streptavidin-coated supported bilayer also containing biotinylated lipids. (B) By introducing 1 mol % PEG₂₀₀₀ lipids, steric repulsion can reduce the deformation of the vesicle. (C) At higher concentration (5 mol %) of PEG₂₀₀₀ lipids, the polymers begin to overlap and extend outwards from the lipid vesicle surface to form a brush layer leading to long-range repulsion. (D) Passivation of the glass substrate with PEG₂₀₀₀ can effectively prevent nonspecific adhesion between the glass and the bilayer. The surface plots show examples of vesicle intensities and corresponding donor intensities in the 4 systems sketched in A–D. When no PEG₂₀₀₀ is present at the interface (A), a relatively large CA is formed that can be detected for vesicles having radius R ≈ 50 nm. Addition of 1 mol % PEGylated lipids to the vesicles (B) reduces the size of the adhesion disk. At 5 mol %, PEGylated lipids (C) the FRET signatures are further reduced even for larger vesicles. FRET signatures on PEGylated glass surfaces (D) were never observed for vesicles below R < 1 μm, indicating a low degree of deformation (see Figs. S6A and S7). The approximate extent of the smallest detectable contact areas, r_CA, is estimated by measuring FRET per area (Fig. S1A) in the 4 systems and is given below each image. (E) Quantification of the donor reduction in the systems (A–C). The measured FRET in the absence of PEG lipids (blue squares) is higher relative to FRET measured for 1% (red circles) or 5% (black asterisk). Vesicles containing 5% PEG lipids initially show very low donor reduction (green triangles), whereas after ≈20 min, the footprints reach a steady state (black asterisk).

Initially, we find the adhesion process to be strongly influenced by the presence of these polymers in all these systems, see Fig. 3. In the absence of PEG polymers, we clearly detect FRET footprints from small vesicles, R ≈ 100 nm (Fig. 3A), whereas in the presence of 1% PEG lipids (Fig. 3B) and 5% PEG lipids (Fig. 3C), larger vesicles show only weak or no FRET signals.

We quantify the reduction in donor intensity for vesicles containing PEG, see Fig. 3E. For comparison, we plot the reduction in donor intensity for vesicles containing no PEG (blue squares), which is notably higher than for vesicles containing 1% (red circles) and 5% (black asterisk) of PEG lipids. Vesicles containing PEG lipids initially show smaller FRET signatures (Fig. 3E, green triangles and Fig. S3 A and B) that become deeper after ≈20 min (Fig. 3E, black asterisk and Fig. S3 C and D). The scaling exponent p is slightly reduced after addition of PEG polymers (Fig. S4A) with p = 2.2 ± 0.2 for 1% PEG and p = 1.8 ± 0.3 for 5% PEG, indicating a constant degree of deformation for different vesicle sizes. Intriguingly, a significant fraction of vesicles containing PEG polymers do not have detectable signatures, Fig. S4B. The size threshold at which we record significant deformation is now shifted compared with non-PEGylated vesicles from 150 nm (Fig. S2) to between 400 and 600 nm for vesicles containing PEG, Fig. S4B. Our data suggest that PEG polymers anchored on mobile lipids can allow secure immobilization of SUVs while significantly suppressing deformation.

Another strategy presented in Fig. 3D is to coat the glass substrate with PEG polymers. Positively charged poly(L-lysine) molecules (PLL) grafted to PEG₂₀₀₀ polymers spontaneously adsorb onto negatively charged glass, providing an effective barrier against nonspecific interactions with the glass (35). We measure dramatic decrease in FRET on this kind of surface. Vesicles <R = 1,000 nm are never observed to give any detectable footprints (see Fig. S5C). Images of GUVs taken at different heights reveal only minimal deformation when vesicles are immobilized on PLL-PEG₂₀₀₀, see Fig. S6A. This should be compared with the severe deformation of GUVs adhering to streptavidin on a supported bilayer as shown in Fig. S6B. We occasionally observe weak footprints associated with GUVs on PEG surfaces, Fig. S7. From the depth of these footprints, we estimate the smallest detectable CAs in this system to have r_CA ≈ 70 nm. The extent of the PEG brush can possibly make it difficult for biotin molecules on vesicles to access the streptavidin molecules randomly oriented within the PEG brush, see Fig. 3D.

Time-Resolved Measurements of Intermembrane Adhesion.

Adhering vesicles expand their surface area and consequently experience lateral stretching. In the limit of strong adhesion of lipid vesicles, an equilibrium between elastic energy stored in the bilayer and adhesion energy is established (36). As reported previously, it is possible to control the tension in the bilayer independently from the surface by exposing the vesicles to higher intensities of laser light (37, 38). The mechanism responsible for tensing the bilayer probably involves oxidation of double bonds in the lipids by free oxygen produced by the fluorophores (38). Indeed, the increase in surface tension in the bilayer caused by illumination reduces the FRET signatures as shown in Fig. 4 A and B. To verify that the change in FRET is caused by a smaller contact zone, we quantify the bleaching of the acceptors (see Fig. 4D). The acceptor intensity reduces by only 11% after the 4 exposures and cannot account for the disappearance of the FRET footprints.

Fig. 4.

Adhesion kinetics of SUVs exposed to strong illumination. (A) Surface plots of FRET signatures after 1 exposure with λ = 543-nm laser on. (B) After 4 exposures, the signatures have vanished. (C) Corresponding image of the SUVs. (D) Bleaching curve of the vesicles versus number of exposures. The signal is reduced by ≈10% after 4 images, which cannot account for the disappearance of the footprints. (E) Time evolution of the tension (Cε, C = 0.01 N/m) within the bilayer for 3 different vesicles having different sizes. Experimental settings are the same as in A. The smallest vesicle becomes highly tensed after a few exposures (circles, R = 181 nm), whereas the tension in the 2 larger vesicles having similar initial tension in the bilayer evolves differently in time (diamonds R = 389 nm and squares R = 636 nm). The vesicles contain 1 mol % of PEG₂₀₀₀ lipids.

We quantitatively explain the deadhesion process by modeling the lipid vesicle as an elastic capsule adhering to a substrate with adhesion energy W_A (39).

where θ is the effective contact angle, ε is the biaxial strain of the capsule surface, and C = 10⁻² N/m is the extensional rigidity (39). Eq. 5 is plotted in Fig. S8A corresponding to an adhesion energy of 10 μJm⁻² for a streptavidin-coated surface (32). ε Is a function of the stretching of the bilayer relative to its original state and therefore depends on the degree of deformation. Under the conditions in Fig. 4, the tension in the bilayer (Cε) increases due to strong illumination resulting in a reduction of the CA and contact angle θ. Measurements of the CA and vesicle size are sufficient to find θ, and thereby the strain can be deduced from Eq. 5.

We perform real-time measurements of the tension in the bilayer in Fig. 4E for 3 vesicles having different sizes. The smaller vesicle having R = 181 nm becomes highly tensed after only a few exposures, whereas the 2 larger vesicles become tensed more gradually.

Several scenarios are possible for the mechanism behind the detachment from the substrate. The binding energy of a single receptor–ligand connection is ≈35K_BT (40), whereas the energy required to pull out a biotin–lipid from a vesicle is ≈30K_BT (41). The strong molecular bonds will cause the receptors and ligands to move laterally inwards from rim of the contact zone as deadhesion proceeds before any molecular detachment between receptors and ligands. Crowding of molecules within the contact zone could eventually result in pulling out of a lipid.

Adhesion sites in biological systems are often dynamic in nature and can involve interfacial molecules that function in cellular communication. Supramolecular structures formed at cellular junctions respond dynamically to changes in their chemical environment and can adopt new configurations. The quantification scheme presented here could help resolve the lateral extent of small-scale structures within these junctions with a relevant temporal resolution for these systems.

Conclusion

The ability to detect nano-scale intermembrane CAs highlights the applicability of this method to biological systems where nano-scale contact zones are frequently observed between adjacent membranes preceding exocytosis or in the process of viral infection.

CAs between SUVs and adhesive substrates are quantified by measuring FRET signatures caused by acceptor-labeled vesicles immobilized on a donor-labeled substrate. Comparing experimental results with analytical calculations of the expected FRET for spherical vesicles reveals that biotinylated vesicles deform on surfaces coated with dense layers of streptavidin. Measurements of vesicle sizes and corresponding FRET signatures allows us to quantify the extent of deformation for vesicle sizes ranging from SUVs to microscopic vesicles. Surface modifications of either the vesicle bilayer or the underlying substrate leads to changes in the adhesive potential, which is reflected in the measured contact area. Therefore, appropriate conditions can be found that lead to secure immobilization while suppressing vesicle deformation.

The analytical and experimental method presented here can easily be customized for investigating different substrates by choosing a FRET pair having a Förster radius that matches the minimal surface separation of the specific system. Moreover, the method can be conveniently extended to adhesion studies of other geometries such as between 2 vesicles containing complementary fluorophores.

Kinetic measurements of contact areas can provide an excellent tool for probing mechanical changes of the membrane caused by membrane–protein interaction or structural changes during thermotropic phase transitions that are associated with significant changes in the mechanical properties of the bilayer.

Materials and Methods

Microscopy.

All fluorescent images were acquired by using a Leica SP5/TCS confocal laser scanning microscope equipped with an oil-immersion objective, N.A. = 1.4. Alexa Fluor 488 and DiIC₁₈ were excited at 488 nm and 543 nm, respectively. Emitted light was collected by using photomultiplier tubes. The acquired spectral ranges for the donor and acceptor emission were 495–540 nm and 550–650 nm, respectively. To avoid complications from cross-talk, only donor emission was used in FRET experiments. DiI molecules were excited and used to determine the size of the lipid vesicle. The axial intensity distribution formed by the objective was measured by operating the microscope in reflection mode and used for calibration of the size of the vesicles. The ambient temperature was 25 ± 1 °C. Tension was induced in the vesicles by increasing the output power from the HeNe laser (543 nm) to accelerate deadhesion but without causing severe bleaching (37, 38).

Image Analysis.

Recorded images were analyzed in Matlab 7.1 (Mathworks). A threshold was applied to the images of the vesicles (see SI Text and ref. 24 for details), and subsequently, all intensities falling above the threshold were added together for obtaining the total integrated intensity for each vesicle. Conversion from intensity to size was achieved by following the procedure described in ref. 24. A similar procedure was used to quantified the FRET signatures (see SI Text).

Preparation of Vesicles.

Vesicles were formed by using a standard rehydration procedure of dried lipid films. Lipids were received in powder form from Avanti Polar Lipids and were dissolved in chloroform. A chloroform solution containing POPC 84 mol % 1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (POPG) 8.5 mol % DiIC₁₈ (Molecular Probes) 6.5 mol %, and biot-DPPE 1 mol % was spread out on a Teflon surface and allowed to evaporate under a flow of nitrogen. A thin lipid film formed on the surface and was hydrated in a sorbitol solution (200 mM, equiosmolar with PBS) at T = 37 °C. A polydisperse distribution, of mainly unilamellar vesicles, was obtained after gentle shaking for 3 h. Smaller unilamellar vesicles were formed after extrusion through 100-nm or 1-μm polycarbonate filters (Avanti's mini extruder; Avanti Polar Lipids Inc.). Vesicles with 1 mol % or 5 mol % DOPE-PEG₂₀₀₀ also contained 0.1 mol % Biot-DSPE-PEG₂₀₀₀.

Surfaces.

Glass coverslips were sonicated in 2% hellmanex and in milliQ water and stored in methanol. Immediately before usage, glass coverslips were plasma etched for 2 min in a plasma cleaner (PDC-32G; Harrick Plasma). To form a supported lipid bilayer (SLB), a solution of small vesicles containing POPC and 1 mol % DPPE-biotin were allowed to fuse to the glass for 30 min. A single bilayer was obtained by thoroughly washing the surface with milliQ water. To form surfaces coated with PLL-PEG₂₀₀₀, a solution containing PLL-PEG₂₀₀₀ and PLL-PEG₂₀₀₀–biotin (ratio 10:1) in 15 mM Hepes buffer was incubated for 30 min and subsequently washed with PBS. The surfaces were incubated for 10 min with 0.018 mg/mL streptavidin labeled with Alexa Fluor 488 dissolved in PBS, and finally the unbound streptavidin was washed out by using PBS. The density of donors on the surface for a monolayer of streptavidin was calculated from the labeling ratio of 4.6 dyes for each streptavidin that occupies 25 nm² (42).

Materials.

Further details about the materials and about image analysis are provided in SI Text.