Real-Time QCM-D Monitoring of Electrostatically Driven Lipid Transfer

J. Phys. Chem. B 2008, 112, 14069–14074

14069

Real-Time QCM-D Monitoring of Electrostatically Driven Lipid Transfer between Two Lipid Bilayer Membranes Angelica Wikstro¨m, Sofia Svedhem, Marc Sivignon, and Bengt Kasemo* Department of Applied Physics, Chalmers UniVersity of Technology, 412 96 Go¨teborg, Sweden ReceiVed: May 05, 2008; ReVised Manuscript ReceiVed: July 08, 2008

The lipid exchange/transfer between lipid membranes is important for many biological functions. To learn more about how the dynamics of such processes can be studied, we have investigated the interaction of positively and negatively charged lipid vesicles with supported lipid bilayers (SLBs) of opposite charge. The vesicle-SLB interaction leads initially to adsorption of lipid vesicles on the SLB, as deduced from the mass uptake kinetics and the concerted increase in dissipation, monitored by the quartz crystal microbalance with dissipation (QCM-D) technique. Eventually, however, vesicles (and possibly other lipid structures) desorb from the SLB surface, as judged from the mass loss and the dissipation decrease. The mass loss is approximately as large as the initial mass increase; i.e., at the end of the process the mass load is that of a SLB. We interpret this interesting kinetics in terms of initial strong electrostatic attraction between the added vesicles and the SLB, forming a structure where lipid transfer between the two bilayers occurs on a time scale of 10-40 min. We suggest that this lipid transfer causes a charge equilibration with an accompanying weakening of the attraction, and eventually repulsion, between the SLB and vesicles, leading to desorption of vesicles from the SLB. The composition of the latter has thus been modified compared to the initial one, although no net mass increase or decrease has occurred. Direct evidence for the lipid exchange was obtained by sequential experiments with alternating positive and negative vesicles, as well as by using fluorescently labeled lipids and FRAP. The above interpretation was further strengthened by combined QCM-D and optical reflectometry measurements. Introduction Interactions at and between lipid membranes is to a large extent governed by the electrostatics at the membrane surface.1 The transfer/exchange of lipids at these interfaces, probably to a large extent dictated by specific lipid transfer proteins but also by nonspecific electrostatics, has been suggested to be important for cell signaling and antimicrobial defense.2 In addition to specific lipid transfer, spontaneous lipid transfer by nonspecific mechanisms plays an important role in a variety of biological processes, such as parasitic invasion of erythrocytes and lipid absorption during digestion.3 The same mechanisms can very efficiently be used as tools to introduce reporter lipids into biological membranes, for example, to follow the fate and metabolism of lipids or to study membrane organization and membrane biogenesis.3 Processes at biological membranes occur in complex contexts, typically involving various membranebound and membrane-associated molecules,4 asymmetric lipid distributions,5 and curvature,6 but it is also of interest and helpful to study the more simple case of two model lipid bilayer membranes interacting with each other, in order to reveal underlying mechanisms and dynamics for lipid membrane interactions and the exchange of matter between them. Attractive model systems for such studies, both from a preparative and analytical point of view, are supported lipid membranes (SLBs) and vesicles. In this study, we address experimentally what happens when lipid vesicles with a net charge (given by their lipid composition) approach a surface with a previously formed SLB composed of lipids of opposite charge. At least three possible outcomes * Corresponding author. Tel: +46 31 772 33 70, Fax: +46 31 772 31 34, E-mail: [email protected].

can be anticipated for this encounter: (i) formation of a second lipid bilayer on top of the first SLB, (ii) adsorption of a (sub)monolayer of intact vesicles on the SLB, and (iii) transfer/ exchange of lipids between the vesicles and the SLB. We present conclusive evidence for case iii, i.e., that transfer/exchange of lipids occurs, and that it is preceded by outcome ii and terminated by desorption of vesicles and/or other lipid structures, leaving an SLB with a new lipid composition. The membrane interactions were monitored in real-time by the quartz crystal microbalance with dissipation monitoring technique (QCM-D).7 Proof of lipid exchange was obtained by sequential positive and negative vesicle exposures to the bilayer as well as by fluorescence microscopy. Reflectometry was used as a complement to these techniques. For the present study, one particularly important capability of the QCM-D technique is its ability to distinguish between lipid vesicles and lipid bilayers on a surface.8,9 This ability originates from the fact that lipid vesicles trap considerable amounts of water, which is measured as part of the mass coupled to the oscillatory motion of the QCM-D sensor surface and independently signaled by the degree of damping of the sensor oscillator: When attached to the surface of the crystal, the vesicles cause a frequency shift (∆f) corresponding to the sum of lipid mass and water mass trapped in and between the vesicles. In addition, the rather viscous adlayer of vesicles gives rise to a high damping and energy dissipation (∆D) value of the oscillator. In contrast, planar lipid bilayers cause a very small dissipation and the frequency shift is then essentially caused by the lipid bilayer mass alone (no or very little trapped water). This much studied behavior led us to believe that QCM-D would be a powerful tool to study vesicle-SLB interactions and, for example, discriminate among

10.1021/jp803938v CCC: $40.75  2008 American Chemical Society Published on Web 10/14/2008

14070 J. Phys. Chem. B, Vol. 112, No. 44, 2008 cases i-iii above. To further strengthen the conclusions we included measurements with a new combined QCM-D/optical reflectometry setup,10 which allows us to obtain a direct measure of the biomolecular (optical) mass (the nonhydrated “dry mass”) on the sensor surface, as opposed to the hydrated mass obtained by QCM-D, in a similar way as previously demonstrated for vesicle f SLB transformation using QCM-D and surface plasmon resonance (SPR) based sensing.11 Materials and Methods Materials. 1-Palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC), 1,2-dioleoyl-sn-glycero-3-[phospho-L-serine] (DOPS), 1,2-dioleoyl-sn-glycero-3-ethylphopshocholine (DOEPC), and 1-palmitoyl-2-[6-[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phosphocholine (NBD-HPPC) were from Avanti Polar Lipids Inc. All other chemicals were from SigmaAldrich. Water was deionized (resistivity >18 MΩ/m) and purified using a MilliQ unit (MilliQ plus, Millipore). Tris buffer (10 mM Tris and 100 mM NaCl) was adjusted to pH 8.0 at room temperature using HCl. For the preparation of negatively charged bilayers, 10 mM CaCl2 was added to the Tris buffer. Buffers were filtered and degassed. Vesicle Preparation. Lyophilized lipids were dissolved in chloroform (stock solutions were stored at -20 °C) and mixed in desired proportions. The solvent was evaporated under a flow of nitrogen while the round-bottomed flask was rotated to form a thin film on the wall of the flask. The lipid film was emulsified in buffer at room temperature, vortexed, and extruded 11 times through a 100 nm polycarbonate membrane, followed by 11 times through a 30 nm membrane (Avanti Polar Lipids Inc.). Vesicles prepared in this way typically measure 80-100 nm, as determined by dynamic light scattering. The vesicle solutions were stored refrigerated under N2 and used within 2 weeks. Charged vesicles were prepared from POPC/DOPS or POPC/ DOEPC mixtures, where the molar fraction of the charged lipid was 10%, 25%, 50%, or 100%. Bulk measurements determining the size and the ζ-potential of the vesicles were performed routinely. QCM-D. The main experimental technique was the quartz crystal microbalance with dissipation monitoring, QCM-D (for details we refer to previous work7 and references therein). The sensing properties of this technique rely on the sharp and stable oscillation resonance when an electric ac field is applied over the electrodes. The surface of the disk-shaped sensor performs a shear oscillation, i.e., with periodic motion back and forth along the sensor surface, with an amplitude of a few nanometers. The frequency of this oscillation is reduced when mass is attached to the surface and increases when mass is detached. Recording of this frequency shift, ∆f, thus gives information about the attached/detached mass to the sensor surface. The second measured quantity in QCM-D is the energy dissipation, D (or damping), of the oscillator. Formally D is defined as the fraction of the total energy stored in the oscillator that is dissipated during one oscillation cycle. Practically, the D factor senses the viscoelastic properties of the mass coupled to the oscillator. For the present study, the most valuable features, in addition to real time monitoring, of the QCM-D technique was the high sensitivity to intact vesicles, due to their high water content, and the ability to identify vesicle attachment/detachment on a SLB through the correlated ∆f and ∆D signals. The sensitivity of QCM-D to the combined contribution of lipid mass and hydration water can be deconvoluted by the use of an optical method,11,12 which in the present case was remedied by reflectometry measurements (see below).10

Svedhem et al. The QCM-D measurements were carried out in batch mode in a Q-sense D 300 cell (Q-Sense AB, Sweden). AT-cut quartz crystals with a fundamental resonance frequency of 5 MHz, coated with SiO2 (Q-Sense AB, Sweden), were used. Prior to the experiments, the crystals were cleaned in 10 mM SDS (>30 min), rinsed with water, dried with N2, and treated in an UV-ozone chamber for 2 × 15 min with rinsing and drying in between. The measurements were carried out at 22 °C. After stabilization of the baseline, 0.5 mL of temperature-stabilized vesicles (125 µg/mL) was injected over the sensor surface. Positively charged vesicles were diluted in Tris buffer and negatively charged vesicles in Tris buffer containing 10 mM CaCl2. After formation of a negatively charged bilayer (using the Ca2+-containing buffer), rinsing with Tris buffer with 1 mM EDTA was performed before addition of positively charged vesicles. Frequency and dissipation shifts were measured at the third overtone. Frequency shifts were normalized to the fundamental frequency of the crystal (5 MHz) by dividing the values by 3. QCM-D/Reflectometry. Combined QCM-D/reflectometry measurements were performed on a prototype instrument from Q-Sense AB.10 The QCM-D (∆f, ∆D) and reflectometry signals were detected simultaneously using one and the same sensor surface, an AT-cut crystal coated with SiO2. The general procedures were the same as for the regular QCM-D experiments, with the exception that continuous flow was used (flow rate 50 µL/min). In order to reduce sample consumption during the exposure to vesicles, the flow outlet was transferred to the sample tube, such that a recirculation loop was formed. The added value of using reflectometry was that hydrated (QCMD) and nonhydrated (reflectometry) mass are measured independently, which is particularly valuable when water-filled vesicles are present on the sensor surface. Fluorescence Microscopy. Fluorescence microscopy measurements were performed on a Zeiss Axioplan 2 fluorescence microscope using a 20× water immersion objective. A nonfluorescent charged bilayer was formed on a QCM-D crystal (one that was also used in the QCM-D measurements), placed in a Petri dish, from negatively charged vesicles (POPC/DOPS 75:25) using the Ca2+-containing buffer or from positively charged vesicles (DOEPC) using the non-Ca2+-containing buffer. After extensive rinsing, without dewetting of the surface, oppositely charged vesicles (DOEPC or POPC/DOPS 75:25) doped with 5% NBD-HPPC (w/w) (λex/em 460/550 nm) were added. The vesicles were allowed to interact with the bilayer for 90 min and then rinsing was repeated. A spot in the bilayer was bleached for 30 s, whereafter the first picture immediately was taken. The bilayer was allowed to recover for 30 min before the second picture was taken. Lipid concentrations and buffers were similar as in the QCM-D experiments. Results SLB Formation by QCM-D. The formation of neutral bilayers on SiO2 substrates from POPC vesicles at high ionic strength (here 10 mM Tris, pH 8, 100 mM NaCl) is well-established.8,13,14 Briefly, initial adsorption of intact vesicles is followed by rupture and fusion of vesicles at a critical coverage, where the rupturing/fusing vesicles form an SLB. The process is conveniently followed in real-time by QCM-D (complementary data have been obtained, e.g., by SPR,11 AFM,14,15 and, as shown here, by reflectometry). Provided the right protocol is followed, a high-quality bilayer is formed over the whole surface, which is manifested by (i) high lateral mobility seen in fluorescence recovery after photobleaching

QCM-D of Electrostatically Driven Lipid Transfer

J. Phys. Chem. B, Vol. 112, No. 44, 2008 14071

Figure 1. Chemical structures of POPC (zwitterionic), DOEPC (cationic), and DOPS (anionic).

(FRAP) measurements,16 (ii) absence of defects during AFM imaging,15 and functionally by (iii) low levels of protein adsorption17 and (iv) high resistance to cell attachment.18 Representative values for the resonant frequency shift (mass change) and the dissipation shift, using a 5 MHz sensor crystal, are ∆f ∼ 25 Hz (at or normalized to the fundamental frequency) and ∆D < 0.5 × 10-6, respectively,8 Here, negatively charged SLBs were formed from mixed POPC/DOPS vesicles (containing 25% or less of DOPS, with 10 mM Ca2+ ions added to the buffer (Ca2+ is necessary for the SLB to form from these vesicles on SiO2), following a similar mechanism as for neutral SLBs (via a critical coverage). The frequency and the dissipation shifts for the POPC/DOPS (75:25) SLB were similar to the values for neutral POPC SLBs. Positively charged SLBs on SiO2 formed by DOEPC vesicles form by a simpler mechanism under similar experimental conditions; individual positively charged vesicles break up and form bilayer patches immediately upon contact with the negatively charged surface; i.e., a critical coverage of adsorbed vesicles is not required for SLB formation [previously shown for dioleoyltrimethylammonium propane (DOTAB)19]. The frequency shift for SLBs composed of DOEPC was typically somewhat smaller (∆f ∼ 22 Hz) than for a POPC SLB. Dissipation was low (∆D < 0.5 × 10-6). The molecular structures of DOPS and DOEPC are depicted in Figure 1. Interaction between Charged SLBs and Oppositely Charged Vesicles. Using the two different kinds of charged SLBs (positively and negatively charged) described above, we explored their interaction with vesicles of opposite charge. We always observed a transient interaction between SLBs of one charge and vesicles of the opposite charge, while no interaction was observed, within the detection limit of the QCM-D set up, between vesicles and SLBs of similar charge, after the bilayer formation had been completed. In Figure 2, two typical examples of the QCM-D results are shown. They are chosen from a larger series of experiments, where different relations between the charge composition of the SLB and the vesicles were investigated. The examples shown demonstrate the general features of our observations for all those cases: Charged vesicles initially adsorb to SLBs of opposite charge, as evidenced by a considerable mass increase (downward frequency shift) and a corresponding increase in dissipation, signaling attachment of intact vesicles (if another bilayer had been formed, much lower frequency and dissipation shifts would have resulted). After some time, which varies with the relative fractions and type of charged lipids (∼17 min in Figure 2A and ∼39 min in Figure 1B), the mass attached to the surface, as well as the dissipation, start to decrease again. The kinetics initially resembles that of SLB formation via a critical surface coverage of vesicles8 and

Figure 2. QCM-D results [frequency (∆) and dissipation (b) shifts] where (i) charged SLBs were formed and (ii) exposed to oppositely charged vesicles. The SLB in A was formed from negatively charged vesicles (POPC/DOPS 75:25) in Ca2+-containing buffer. After rinsing with EDTA-containing buffer, positively charged vesicles (DOEPC) were injected, and the kinetics is characterized by vesicle adsorption (decreasing frequency, increasing dissipation) followed by extremum points after which the frequency and dissipation increase and decrease again, respectively, until the SLB values are obtained. In B, the charge states of SLB and vesicles were the opposite, i.e., the SLB was formed from positively charged vesicles (DOEPC), followed by injection of negatively charged vesicles (POPC/DOPS 75:25). The kinetics is similar as in A but with a different time scale.

could have been mistaken for the formation of a second SLB on top of the first one (case I in the Introduction). However, the mass and dissipation values continue to decrease well beyond those expected for a second SLB on top of the first one (would have been detected as an additional frequency shift of ca. 22-26 Hz) and eventually saturates at values that are the same, or very close to, the values for the initial SLB. Thus, when the whole process is completed, the net mass uptake and dissipation values correspond to those of a single SLB, as if nothing had happened to the initial SLB regarding its structure and mass per unit area. For higher fractions of charged lipids in the SLB, or higher fractions of charged lipids in the added vesicles, the same qualitative behavior has always been observed, but different times were then required for the process to complete. On the basis of these observations, we conclude that there is an initial strong electrostatic attraction between the vesicles and the SLB that, after causing adsorption of the vesicles, is gradually reduced due to transfer of lipid molecules between the adsorbed vesicles and the SLB, until, eventually, the electrostatic interaction becomes too weak, or more likely reversed (see below), due to the lipid/charge exchange, so that vesicles desorb from the surface and leave behind a SLB with the same total mass but with a net change in charge compared to the initial SLB. Sequential Exposures of the SLB to Vesicles of Alternating Charge. To test our conclusion above that transfer of charged lipid molecules really took place, another type of QCM-D experiment was performed. The reasoning was the following: if the vesicle adsorption-desorption sequence had caused transfer of lipids and charge reversal of the SLB, the latter should now interact again with the same type of vesicles that were used to form the initial SLB (when an SLB has been completed with one type of vesicles, it does not interact measurably with those vesicles). After the formation of a negatively charged SLB and subsequent exposure to positively

14072 J. Phys. Chem. B, Vol. 112, No. 44, 2008

Figure 3. QCM-D results [frequency (∆) and dissipation (b) shifts] showing formation of (i) a negatively charged bilayer (POPC/DOPS 90:10), followed by (ii) exposure to positively charged vesicles (DOEPC), and (iii) exposure to negatively charged vesicles (POPC/ DOPS 90:10).

charged vesicles, similar to the experiment in Figure 2, followed by rinsing, we thus expected the modified SLB to have a positive net charge. When this SLB was exposed to negatively charged vesicles (Figure 3iii), the vesicles adsorbed on the SLB in a similar way as in Figure 2, demonstrating that a charge reversal of the SLB had occurred. This result is thus taken as proof that lipid transfer between the SLB and the attached vesicles occurs. We have repeated this type of sequential experiment with different charge (both with respect to the charge sign and the fraction of charged molecules) in the bilayer and vesicles, and also several times in succession, and seen qualitatively similar behavior, but with different time scales and quantitative kinetics. Results similar to the ones described above have been obtained by ATR-IR,20 where the bilayer chemical composition could be followed by infrared spectra. Combined QCM-D/Reflectometry Measurements. The QCM-D technique is sensitive to layers of biomolecular mass being deposited on the sensor crystal, but also to water associated with it.8,9 The associated sensitivity enhancement of the response when dealing with large amounts of trapped water makes the QCM-D more sensitive than optical techniques (e.g., SPR-based techniques, optical wave guides, or reflectrometry) in those situations where large amounts of hydration water is coupled to detected biomolecules. This was, for example, recently demonstrated for a 19-mer peptide, immobilized to supported lipid membranes,21 and it has proven extremely valuable for the unraveling of mechanisms by which lipid vesicles spread on solid supports.8,14 In order to separate how the accumulation of lipid material contributes to the QCM-D signal and the contribution of water in the lipid transfer experiments reported in this study, additional experiments were performed in a new experimental setup, where QCM-D and optical reflectometry data were measured simultaneously on the same surface (i.e., on the QCM-D sensor).10 The reflectometry response is proportional to the amount of biomolecular mass at the surface, excluding the coupled water. The data in Figure 4 fully support the QCM-D findings described above; in the first preparative step, a lipid bilayer with positive lipids is formed (vesicles are injected at step i) and the bilayer formation process goes on between 5 and 12 min). The frequency shift decreases (corresponding to the mass increase of the SLB) and the reflectometry signal follows the frequency signal. The dissipation change is very low, because in this case, vesicles do not accumulate intact on the surface but rupture immediately to bilayer upon surface contact. All these data signal bilayer formation. At step ii vesicles with net negative charge were injected and allowed to interact with the positively charged SLB. The QCM-D results (∆f and ∆D) were discussed above. The reflectometry signal follows the frequency shift, but

Svedhem et al.

Figure 4. Combined QCM-D and reflectometry measurements showing (i) the formation of a positively charged bilayer from DOEPC vesicles, followed by (ii) an injection of negatively charged vesicles (DOPS/ POPC 1:1): (∆) the frequency shift, (b) the dissipation shift, and (0) the reflectometry signal (arbitrary scale).

Figure 5. Fluorescence microscopy images showing bleaching (at t ) 0 min) and recovery of an SLB, after exchange of lipids between an initially nonfluorescent SLB and fluorescently labeled vesicles. First, a nonfluorescent, negatively charged SLB (POPC/DOPS 75:25) was formed, and then the SLB was exposed to positively charged and fluorescently labeled vesicles (DOEPC/NBD-HPPC 95:5).

the relative response to the vesicle adsorption of the optical signal, in terms of total mass, is much smaller. By comparing the initial bilayer formation signals (i.e., between steps i and ii) for ∆f and for reflectometry, with the maxima in these two signals during intact vesicle adsorption (a more detailed discussion can be found elsewhere11), we can estimate that the maximum mass of additional lipids attached to the bilayer (per unit area) at ∼35 min corresponds to a little less than half of a complete bilayer mass and thus to a submonolayer coverage of vesicles on the SLB. These estimates can also be arrived at semiquantitatively by just comparing the change in reflectometry signal for the first SLB formation at ∼10 min, and the amplitude of the maximum at ∼3 min, since both signals correspond to the amount of (“dry”) added lipid mass. These observations also tell us that the main part of the ∆f signal comes from trapped water, which supports that the added lipid mass is in the form of vesicles, or possibly more complex water-filled lipid structures,22 attached to the bilayer. We emphasize the approximate nature of these estimates; ongoing work will provide more precise quantification. However, for the present purpose the reflectometry data is enough to support the QCM-D data with a semiquantitative estimate of how much extra lipid mass that is transiently attached to the SLB during the lipid exchange process. Additional Direct Evidence for Lipid Transfer from Fluorescence Microscopy. Fluorescence microscopy was then employed to provide a final direct test of the concluded transfer of lipid molecules between attached vesicles and the SLB (Figure 5). Nonfluorescent SLBs were exposed to vesicles with fluorescently labeled lipids in the same way as described above. The nonfluorescent SLB became fluorescent after exposure to oppositely charged and fluorescently labeled vesicles. By FRAP, it was demonstrated that the fluorescence was confined to the SLB and that the lipids in the SLB were mobile after the exchange, pointing toward a good quality SLB, in agreement with the QCM-D results. In separate QCM-D experiments,

QCM-D of Electrostatically Driven Lipid Transfer

Figure 6. Schematic illustration showing two different scenarios for the lipid exchange between a charged supported lipid bilayer and electrostatically adsorbed vesicles.

fluorescently labeled charged vesicles were shown to interact similarily, with the oppositely charged SLB, i.e. to show the same kinetics, as nonfluorescent vesicles (not shown). Discussion On the basis of the presented data, we conclude that lipid exchange occurs between vesicles composed of lipids of one charge and SLBs with lipids of opposite charge, i.e. we have demonstrated that electrostatic interaction between two proximal lipid bilayers led to transfer/exchange of lipid molecules, without special promoting molecules in the bilayer(s). In our experiments, we observed qualitatiVely similar kinetics independent of the charge density used, but the kinetics exhibited different time scales, depending on the relative concentrations of charged and noncharged lipids (Figure 1). The advantage of using the QCM-D technique for these measurements is that the interactions of small vesicles can be studied in real-time, e.g., without special added lipids, which are modified dye molecules. Using reflectometry we showed that the amount of transiently attached lipid mass to the SLB, presumably in the form of vesicles, was significantly less than one full monolayer of intact vesicles. Using sequential experiments with vesicles of alternating charge and FRAP experiments, we obtained direct evidence for lipid exchange between the SLB and vesicle membranes. Interactions between oppositely charged vesicles in the liquid bulk phase have been studied previously by other techniques,23 ranging from standard fluorometric vesicle population assays24 to fluorescence video microscopy25 and recording of the whole electrostatically induced fusion between giant vesicles by microfluorescence spectroscopy.26 In the latter experiments, fusion did not occur at low (∼10%) but only at higher charge densities (∼25%). Similar results were obtained in a recent study,27 where video microscopy was used to follow the membrane tension, while small charged vesicles were interacting with oppositely charged giant vesicles. In the latter study, it was shown that localized instabilities were formed on the giant vesicles and that the membrane was strongly destabilized, a nonequilibrium state that was proposed to be an important intermediate state in biological processes. In the following we discuss our results based on two common mechanistic scenarios outlined in Figure 6: (i) a close enough contact between the SLB and adsorbed vesicles to allow for intermembrane flip-flop of lipid molecules or (ii) the transient formation of a fused structure (exemplified by hemifusion). In scenario i, adsorbed vesicles are assumed to retain their integrity as vesicles, while residing on top of the SLB (and the SLB is retaining its integrity as a bilayer), whereas in scenario ii, the lipid membranes melt together, establishing close dehydrated contacts with the SLB, leading to, for example, the formation of a fusion stalk (potentially transforming into a hemifusion diaphragm) connecting the two membranes.28 In case i the lipid exchange would occur via monomer exchange, one after the other, between two intact bilayers, and those monomer “jumps” would then be the rate-limiting steps for the overall kinetics. In case ii exchange would most likely occur by diffusion between

J. Phys. Chem. B, Vol. 112, No. 44, 2008 14073 the upper leaflet of the SLB and the outer leaflet of the vesicle, within the hemifusion structure. In this mechanism, the ratelimiting step would most likely be the formation of the hemifusion structure, since diffusion within a leaflet is fast. It is generally believed that redistribution of lipids between two leaflets of lipid membranes by monomer jumps (flip-flop) is very slow, on the time scale of hours to days.29 In contrast, the exchange of lipids throughout the outer leaflet during a fused state is likely to be very rapid, actually much faster than the measured residence times for vesicle adsorption and desorption on the SLB. Therefore, the time scale of our experiments could be taken as an indication of which of the two scenarios is more probable. However, recent reports suggest that, compared to the less constrained membranes of vesicles in the bulk phase, exchange may be much more rapid in lipid membranes supported on a surface30 or in Langmuir-Blodgett/ Langmuir-Schaefer (LB/LS) films.31,32 The suggested reason for the reported increased flip-flop rates was based on defectmediated exchange in SLBs,30 and in the recent study with the transmembrane peptide gramicidin inserted in LB/LS films, exchange was suggested to be facilitated by the presence of less ordered hydrophobic regions near transmembrane proteins.31 Another mechanism that is likely to speed up mechanism i above (intact vesicles on the SLB) is that the initial electrostatic attraction between the vesicle and the oppositely charged SLB likely leads to significant deformation (flattening) of the vesicle, as has been seen by AFM for intact vesicles on surfaces,15 so that the bending radius becomes much smaller than the average vesicle radius at the outer rim of contact between vesicle and SLB. In these regions of small curvature, lipid flip-flop is likely to be enhanced; i.e., the activation barrier is reduced. Finally, we note that a theoretical estimate of lipid exchange between two adjacent bilayers via an insertion mechanism rather than flip-flop suggests the insertion mechanism to be more probable, with a shorter time scale (lower activation barrier) compared to a flip-flop mechanism.33 Further experiments are needed to determine in detail what mechanisms are in operation during the lipid exchange in the present case. The unraveling of the function of lipids in biological membranes has been slower than for membrane proteins due to the experimental difficulties associated with their study. As demonstrated above, lipid transfer may, for example, be an important mechanism in order to destabilize membranes as they are about to remodel (vesicle budding, exo- and endocytosis, extraction of membrane tubes, etc.). In a recent study, it was also shown how redistribution of charged lipids can result in self-induced haptotactic motion of vesicles on a lipid bilayer.34 We believe the methodology and type of results obtained here and similar results from extended studies to come will elucidate such processes. Conclusion In summary, we have measured, in real-time, interactions between charged SLBs and vesicles of opposite charge, for smaller vesicles (