Langmuir 2009, 25, 3083-3088
3083
Reversible Immobilization of Diffusive Membrane-Associated Proteins Using a Liquid-Gel Bilayer Phase Transition: A Case Study of Annexin V Monomers Jason J. Han and Doo Wan Boo* Department of Chemistry, Yonsei UniVersity, Seoul 120-749, Korea ReceiVed NoVember 25, 2008. ReVised Manuscript ReceiVed January 3, 2009 We report a novel strategy for reversible immobilization of diffusive membrane-associated proteins in a native orientation using a liquid-gel bilayer phase transition, and its application to the single molecule study of Cy3-labeled Annexin V (A5) monomers on supported lipid bilayers containing phosphatidylserine (PS) in a Ca2+-rich environment. Total internal reflection fluorescence single molecule trajectory analysis revealed that, at low membrane occupancy, A5 monomers diffuse randomly on liquid phase bilayers and occasionally collide with other A5 monomers to form short-lived pseudodimers. During the liquid-to-gel bilayer phase transition, diffusive A5 monomers become immobilized mostly as isolated monomers, with some percentage of dimers and trimers. The EDTA-induced unbinding of immobilized Cy3-A5 spots indicates that Ca2+-bridges between A5 and PS lipids are preserved in the immobilized A5 monomers, confirming their native orientation on gel phase bilayers. Furthermore, the persistence of Ca2+-bridges during the liquid-to-gel phase transition, despite negligible A5 binding affinity to gel phase bilayers, strongly suggests the formation of tightly bound A5(Ca2+)m(PS)n complexes that diffuse and become immobilized as single units during the bilayer phase transition.
Introduction Single molecule (SM) techniques have enabled the study of individual biomolecules through a variety of optical- and forcebased instrumental modes. In recent years, SM fluorescence and force spectroscopy combined with real time imaging techniques have been utilized to probe biomolecular phenomena, such as 2-D and 3-D diffusion in vivo,1 protein folding and unfolding,2 and ligand-protein3 and protein-protein interactions4 not resolvable by bulk ensemble methods. Although the basic instrumental platforms for performing such SM optical and force experiments are readily available commercially, custom modifications are often necessary to address specific experimental requirements. In a similar and equally important manner, sample preparation for single molecule experiments often starts from a core set of strategies which could include any combination of, for example, site-specific labeling, immobilization on a substrate, reconstitution into artificial membranes, and so forth. The continued success of single molecule experiments depends upon an inseparable combination of customized instrumental platforms and novel sample preparation techniques. New experimental strategies are continually in need to accommodate the remarkable diversity in the chemical, self-assembly, and overall functional properties within and across different classes of biomolecules (e.g., proteins, DNA and RNA, and lipids). * To whom correspondence should be addressed. E-mail: dwboo@ yonsei.ac.kr. (1) (a) Kusumi, A.; Ike, H.; Nakada, N.; Murase, K.; Fujiwara, T. Semin. Immunol. 2005, 17, 3–21. (b) Kues, T.; Peters, R.; Kubitscheck, U. Biophys. J. 2001, 80, 2954–2967. (2) (a) Weiss, S. Nat. Struct. Biol. 2000, 7, 724–729. (b) Schuler, B. ChemPhysChem 2005, 6, 1206–1220. (c) Zlatanova, J.; Lindsay, S. M.; Leuba, S. H. Prog. Biophys. Mol. Biol. 2000, 74, 37–61. (d) Schlierf, M.; Li, H.; Fernandez, J. M. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 7299–7304. (3) (a) Sako, Y.; Minoghchi, S.; Yanagida, T. Nat. Cell Biol. 2000, 2, 168–172. (b) Neish, C. S.; Martin, I. L.; Henderson, R. M.; Edwardson, J. M. Br. J. Pharmacol. 2002, 135, 1943–1950. (c) Bowen, M. E.; Weninger, K.; Ernst, J.; Chu, S.; Brunger, A. T. Biophys. J. 2005, 89, 690–702. (4) (a) Weisel, J. W.; Shuman, H.; Litvinov, R. I. Curr. Opin. Struct. Biol. 2003, 13, 227–235. (b) Taguchi1, H.; Ueno, T.; Tadakuma, H.; Yoshida, M.; Funatsu, T. Nat. Biotechnol. 2001, 19, 861–865.
In this Article, we present a novel strategy for reversibly immobilizing diffusive membrane-associated proteins in a native orientation on a supported lipid bilayer (SLB). The immobilization technique utilizes reversible temperature control to transition between fluid liquid and solid gel states of a SLB containing saturated lipids,5 and permits the repeated measurements of SM static and dynamic properties on the same samples. The diffusion and immobilization processes of single membrane-bound, dyelabeled proteins can be monitored in real time using total internal reflection fluorescence microscopy (TIRFM), along with the static fluorescence intensity analysis of the immobilized spots. As a case study, we investigated the diffusion and immobilization behavior of Cy3-labeled Annexin V (A5) monomers on phosphatidylserine (PS)-containing SLBs in a Ca2+-rich environment. A5 is a well-studied member of the annexin superfamily of peripheral membrane binding proteins best known for their high binding affinity to negatively charged membranes in the presence of Ca2+.6 The self-assembly of diffusive membrane-bound A5 monomers into tightly bound trimers and subsequent 2-D organization into monolayer crystals have been well documented on model membranes.7-10 Under high levels of A5 membrane occupancy, trimer formation on lipid bilayers has been shown to occur within milliseconds of initial A5 binding.9 Consequently, information regarding the behavior of single membrane-bound A5 monomers has been inferred indirectly from experiments using ensemble measurements. Here, we demonstrate that single Cy3-labeled A5 monomers can be directly observed diffusing and colliding with other A5 monomers across a fluid (5) Charrier, A.; Thibaudau, F. Biophys. J. 2005, 89, 1084–1101. (6) Gerke, V.; Creutz, C. E.; Moss, S. E. Nat. ReV. Mol. Cell Biol. 2005, 6, 449–461. (7) (a) Reviakine, I.; Bergsma-Schutter, W.; Brisson, A. J. Struct. Biol. 1998, 121, 356–361. (b) Oling, F.; Bergsma-Schutter, W.; Brisson, A. J. Struct. Biol. 2001, 133, 55–63. (8) Langen, R.; Isas, J. M.; Lueckei, H.; Haigler, H. T.; Hubbell, W. L. J. Biol. Chem. 1998, 273, 22453–22457. (9) Patel, D. R.; Isas, J. M.; Ladokhin, A. S.; Jao, C. C.; Kim, Y. E.; Kirsch, T.; Langen, R.; Haigler, H. T. Biochemistry 2005, 44, 2833–2844. (10) Richter, R. P.; Him, J. L. K.; Tessier, B.; Tessier, C.; Brisson, A. R. Biophys. J. 2005, 89, 3372–3385.
10.1021/la803903j CCC: $40.75 2009 American Chemical Society Published on Web 02/04/2009
3084 Langmuir, Vol. 25, No. 5, 2009
Han and Boo
bilayer and subsequently immobilized in their native membranebound orientation without chemical modification. The technique can be potentially used for reversible immobilization of other single diffusive peripheral or integral membrane proteins for controlled single molecule force-optical studies.
Materials and Methods Annexin V-Cy3 (A5-Cy3) was purchased from Sigma-Aldrich. According to the supplier, free amines on lysines are targeted as labeling sites, although the exact location of the dye label is unknown. 1,2-Dipentadecanoyl-sn-glycero-3-phosphocholine (15:0PC), 1,2dimyristoyl-sn-glycero-3-[phospho-L-serine] (DMPS), 1,2-Dioleoylsn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3[phospho-L-serine] (DOPS), and 1-palmitoyl-2-[6-[(7-nitro-2-1, 3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phospho-Lserine (NBD-PS) were purchased from Avanti Polar Lipids, Inc. The phase transition temperatures (Tm) for 15:0PC and DMPS are 33 and 35 °C, respectively, whereas both DOPC and DOPS remain in the liquid phase below 0 °C. Liposomes were prepared by mixing 15:0PC/DMPS or DOPC/DOPS at a 9:1 molar ratio in chloroform with or without 0.01% NBD-PS, evaporating to dryness under N2, drying in vacuo for at least 1 h, rehydrating in Ca2+-free buffer (10 mM HEPES, 140 mM NaCl, pH 7.2), and finally subjecting to bath sonication for 5-10 min until the solution became clear. 15:0PC/ DMPS and DOPC/DOPS mixtures were sonicated at 45 °C and room temperature (RT), respectively. Final concentration of all asprepared liposome suspensions was 1 mg/mL. Cover glasses (Fisher, 22 mm squares) were cleaned in piranha solution for 30 min, rinsed thoroughly with ultrapure water, dried with N2 gas, and used immediately. A custom Teflon heating cell containing internal resistance wire for heating was adhered to the cover glass with a thin layer of silicon vacuum grease. To prepare 15:0PC/DMPS lipid bilayers, the heating cell/cover glass assembly and all liposome suspensions and buffers were thermally equilibrated for at least 30 min in a 45 °C oven and maintained at that temperature during bilayer preparation. Lipid bilayers were prepared by diluting liposomes 2× with a Ca2+-rich buffer (10 mM HEPES, 140 mM NaCl, 5 mM Ca2+, pH 7.2) to a final concentration of 0.5 mg/mL lipid and 2.5 mM Ca2+. Dilutions were immediately added to the heating cell buffer reservoir and incubated for 1 h at 45 °C, after which the sample was removed from the oven and cooled naturally to room temperature (RT) and then rinsed 5× with the Ca2+-rich buffer to remove unfused liposomes. DOPC/DOPS bilayers were prepared in a similar manner but at RT. After placing the sample on an Olympus IX70 microscope, an external power supply was connected to the resistance wire in the heating cell. The sample temperature at the lipid bilayer surface was monitored by carefully immersing the tip of a fine gauge thermocouple probe (Omega) until it touched the cover glass surface. Heating and cooling curves were recorded with a temperature meter and RS-232 interface. Sample temperature typically varied by (1 °C over time but was otherwise stable. Stock A5-Cy3 was diluted with Ca2+-rich buffer and added directly to the lipid bilayer buffer to a final concentration of ∼80 nM for saturation studies or ∼14 pM for single molecule experiments. Membrane binding of A5-Cy3 molecules at various temperatures was monitored using 532 nm laser excitation in objective-type TIRFM mode, which selectively excites and images only membrane-bound A5-Cy3 molecules. Fluorescence emission was passed through 532 nm notch and 550 nm long pass filters. Photobleaching experiments were performed using a 488 nm Ar+ laser and appropriate notch and long pass filters. All emission was detected with an EMCCD (Andor iXon).
Results and Discussion SLB Phase Transitioning. Lipid bilayer transitions between the liquid crystal and gel phases were performed using 15:0PC/ DMPS bilayers and the custom heating cell/cover glass assembly shown schematically in Figure 1A. The temperature at the cover glass surface could be precisely controlled and monitored from
Figure 1. (A) Schematic drawing of custom SLB heating cell and TIRFM optical setup. (B) Photobleached region of a gel phase bilayer containing 0.01% NBD-PS lipids after exposure to 488 nm excitation. (C) The same region in (B) after heating the bilayer to ∼40 °C.
RT to ∼45 °C. Bilayers containing 0.01% NBD-PS were prepared, and qualitative fluorescence recovery after photobleaching (FRAP) experiments using a 488 nm laser were performed to evaluate changes in lipid mobility above and below the Tm, which is ∼35 °C for 15:0PC/DMPS lipid mixtures. The photobleached region showed no FRAP at RT (Figure 1B), verifying that lipids were immobilized in the gel phase bilayer. However, when the same bilayer was heated to and equilibrated for several minutes at 40 °C, which is above the specified Tm of both 15:0 PC and DMPS, complete fluorescence recovery was observed (Figure 1C), indicating that lipids were readily diffusing at the elevated temperature. Subsequent repeated photobleaching of the heated bilayers at temperatures between 37 and 40 °C always showed complete FRAP within 30 s, indicating the entire bilayer was in a fluid liquid crystal state at these temperatures. A given bilayer could be repeatedly cycled between liquid and gel states, with liquid crystal bilayers always showing rapid FRAP and gel phase bilayers showing no FRAP. A5 Binding to Gel versus Liquid Crystal Phase SLBs. A recent study demonstrated that A5 shows negligible membrane binding affinity to PS-rich liposomes in the gel phase, but will readily bind when the same liposome/A5 mixture is heated to above the lipid melting temperature where the bilayer is in a fluid liquid crystal phase.9 To verify this observation under our experimental conditions using SLBs, we incubated A5-Cy3 at saturating concentrations (∼80 nM) on RT gel phase 15:0PC/ DMPS (9:1) SLBs and monitored membrane binding by TIRFM and 532 nm excitation. In all cases, we observed less than 1% of the total membrane-bound fluorescence intensity observed for identical experiments using bilayers composed of DOPC/ DOPS lipids, which remain in the liquid crystal phase down to subzero temperatures. However, after heating the same gel phase bilayer/A5-Cy3 mixture to 37-40 °C and reimaging, the bilayers appeared saturated with Cy3 fluorescence, clearly indicating that solution A5-Cy3 molecules became bound to the membrane as a result of the gel-to-liquid phase transition. These observations
DiffusiVe Membrane-Associated Proteins
Figure 2. A5 monomer diffusion trajectories and corresponding MSD plots. (A) Single collision-free trajectory and (B) corresponding MSD plot showing linear fit (dashed line). (C) Two-particle collisional trajectories showing monomer-monomer interactions indicated by circled region. Numbers indicate starting location of each respective trajectory. (D) MSD plots of the trajectories in (C) along with linear fits (dashed lines) to the overall linear trend. Circled region indicates deviations (blue dashed lines) from the overall linear trend as a result of monomer-monomer interactions shown in (C).
are consistent with results of the aforementioned A5 binding to gel phase liposomes study. Diffusion and Collision of A5 Monomers on SLBs. Due to the tendency for A5 to spontaneously form trimers upon membrane binding, it was necessary to verify that membranebound monomers can exist and be observed at low membrane occupancy. Using DOPC/DOPS SLBs and TIRFM imaging at RT, A5-Cy3 monomers could be readily monitored descending from solution, and then binding to and diffusing across the liquid crystal SLBs. At the low concentrations used, diffusing monomers showed both collision-free and two-particle collisional diffusion trajectories. Collision-free trajectories such as that shown in Figure 2A showed linear trends in the corresponding mean square displacement (MSD) plots (Figure 2B). Linear fits to the Einstein equation for random diffusion in two dimensions11 yielded an average diffusion coefficient (D) of 4.0 ( 0.5 × 10-9 cm2/s, as determined from the slopes. This value is in close agreement with the D value reported from a FRAP study using dye-labeled A5 at low membrane occupancy on liquid crystal bilayers.12 For two-particle collisions, two diffusing monomers on separate, noninteracting trajectories were observed randomly colliding and interacting for a brief moment before dissociating and resuming random diffusion along separate, noninteracting trajectories. Figure 2C shows an example of two colliding A5Cy3 monomer trajectories. During the interaction, the two A5 monomers appeared to “flirt” with and dance around each other, sometimes appearing as a single spot, while other times appearing as two resolved spots separated by ∼400 nm, suggesting an interplay of both short and long-range interactions. The corresponding MSD plots (Figure 2D) for each trajectory showed (11) Berg, H. C. Random Walks in Biology; Princeton University Press: Princeton, NJ, 1983. (12) Cezanne, L.; Lopez, A.; Loste, F.; Parnaud, G.; Saurel, O.; Demange, P.; Tocanne, J.-F. Biochemistry 1999, 38, 2779–2786.
Langmuir, Vol. 25, No. 5, 2009 3085
Figure 3. Top: Diffusion trajectory of single A5 monomer during a liquid-to-gel transition. Gray arrow indicates direction of travel from starting point. Bottom: MSD plot of above trajectory with linear fit (dashed line) to the first ∼2 s of data points.
overall linear trends and similar D values to those from collisionfree trajectories. However, the collision and association of two A5-Cy3 monomers into a short-lived interacting pair, or pseudodimer, caused attenuated diffusion of each interacting monomer as evidenced by temporary decreases in both MSD slopes starting at a lag time of ∼1.5 s. These weak monomermonomer interactions typically lasted for ∼500 ms, and from the decreased slopes in these regions a new interaction diffusion coefficient, Dint, could be estimated at ∼1.9 × 10-9 cm2/s, which is approximately half of what was observed for free, random diffusion. Similar Dint values were observed for other interacting pairs, clearly showing that weak monomer-monomer interactions can attenuate diffusion across the membrane. A single diffusing A5-Cy3 monomer could be observed sequentially interacting with two or more other monomers at different points along its trajectory, but in no case did we observe sequential coalescence of monomers into a single multimer. The observations presented here are consistent with a recent study suggesting that under conditions of low membrane occupancy, membrane-bound A5 monomers remain in the diffusive monomeric state even in the presence of protein-protein interactions known to be important for trimer and subsequent monolayer crystal formation.13 Immobilization of Diffusive A5 Monomers on SLBs. The striking difference in A5 membrane binding affinity for gel and liquid crystal phase bilayers was used as a means to immobilize diffusive membrane-bound A5 monomers. This was achieved by incubating A5-Cy3 on 15:0PC/DMPS bilayers preheated to 40 °C and then cooling the bilayer back to RT. Alternatively, A5-Cy3 could be added to a RT gel phase bilayer, heated to 40 °C, and then cooled back to RT to achieve the same effect. A5-Cy3 diffusion on heated liquid crystal bilayers and subsequent immobilization by cooling to the gel phase could be directly monitored and captured in real time using video acquisition on the EMCCD. Figure 3 shows a typical single spot diffusion trajectory obtained immediately after heating was removed from a liquid crystal bilayer. The corresponding MSD plot shows a linear region for ∼2 s after removing heat, during which time the bilayer is still in a fluid state. Fitting these linear regions from (13) Jeppesen, B.; Smith, C.; Gibson, D. F.; Tait, J. F. J. Biol. Chem. 2008, 283, 6126–6135.
3086 Langmuir, Vol. 25, No. 5, 2009
Figure 4. (A) Example TIRFM image of immobilized spots after a liquid-to-gel transition. Image size: 40 × 40 µm. (B) Histogram of spot intensities from several TIRFM images fitted to three Gaussians. (C) Intensity time traces of the numbered spots in (A), showing single-step photobleaching. (D) Decrease in total fluorescence intensity after addition of excess EDTA. Arrow marks time of EDTA addition.
multiple single spot trajectories yielded an average D value of 3.3 ( 0.5 × 10-9 cm2/s. Beyond 2 s, the slope of the MSD plot starts to continually decrease as a result of the liquid-to-gel transition and corresponding decrease in lipid mobility. Finally, the slope reaches zero, at which point the bilayer is in the gel phase and the spot is immobilized. Analysis of Immobilized Fluorescent Spots. After cooling and immobilizing A5-Cy3 on the gel phase bilayers, analysis could be performed on the fluorescent spots of the resulting EMCCD images, an example of which is shown in Figure 4A. Repeated brief snapshots of a given field of view for extended periods of up to 1 h always showed the same fluorescent spot pattern, indicating that the spots were immobilized on the membrane surface and that dissociation back into solution was not occurring. One important question that arose is whether the immobilized spots were A5-Cy3 monomers, dimers, trimers, or a distribution of the three. The single molecule tracking experiment results presented above showed that dynamic monomer-monomer interactions do occur, suggesting that two interacting A5-Cy3 monomers could possibly become immobilized as a pseudodimer. To address these questions, the fluorescence intensities of immobilized spots from several images were quantified and plotted in a histogram, yielding a multimodal distribution that was fitted to three Gaussians (Figure 4B). Analysis of the photobleaching behavior of each fluorescent spot showed that ∼75% of all spots investigated exhibited one-step
Han and Boo
photobleaching (Figure 4C), which is a photophysical signature of single dye molecules and therefore indicates a dye/protein ratio of 1:1 for those spots. Immobilized dimers or trimers of A5-Cy3 with a one-photon photon counting histogram would be expected to show multistep photobleaching behavior due to the presence of at least two or three Cy3 dye molecules. To further verify the 1:1 dye/protein ratio, we analyzed solutions of A5Cy3 with a one-photon counting histogram (PCH)14 to determine the number of species based on molecular brightness (ε): singly labeled A5 has brightness ε, doubly labeled A5 has brightness 2ε, and so forth. For all A5-Cy3 concentrations, only a singlespecies model could fit the resulting PCH, and the resulting molecular brightness was similar to that of standard Cy3 solutions of the same concentration, indicating that A5-Cy3 consisted predominantly of singly labeled A5 (see the Supporting Information). Based on these analyses, we attribute the major peak in the spot intensity histogram to immobilized singly labeled monomeric A5-Cy3. The second peak may have resulted from the immobilization of pairs of interacting A5-Cy3 monomers, while the third peak could possibly be attributed to immobilized trimers. A5-Cy3 Immobilized in Native Orientation. An important requirement of this immobilization technique is that, during the liquid-to-gel phase transition, A5 molecules remain specifically bound in a native orientation, with convex sides toward the membrane and concave N-terminal sides facing the bulk solution. The Ca2+ bridges between the multiple Ca2+ binding loops on the convex side of A5 and the negatively charged PS head groups in the phospholipid membrane are responsible for the strong binding of A5 to fluid membrane surfaces in a very specific and well-known planar orientation.6 Ca2+ chelators such as EDTA and EGTA are known to be able to disrupt these Ca2+ bridges and cause A5 to unbind if the chelator concentration exceeds the Ca2+ concentration. To verify that A5-Cy3 molecules were immobilized in a native, planar orientation, we added an excess of EDTA to samples after immobilization and recorded videos of the overall fluorescence spot density in a given field of view over time. Immediately after EDTA addition, spots began to disappear, resulting in a steady decrease in the total number of spots and, hence, total fluorescence intensity (Figure 4D). Although photobleaching does occur during continuous laser excitation, its contribution to the rate of decrease of fluorescence intensity was found to be negligible compared to the effects of EDTA addition. Hence, we concluded that the Ca2+ bridges responsible for A5 binding to liquid crystal membranes persist through the liquid-to-gel phase transition, and that A5-Cy3 monomers are immobilized in a native, planar orientation on the gel phase bilayers. Proposed Immobilization Mechanism. The overall binding scheme and proposed mechanism for immobilization is summarized in Figure 5. A5 monomers in the bulk solution do not bind to gel phase bilayers. After the bilayer is heated to a fluid liquid crystal phase, A5 monomers attracted to the membrane surface form tightly bound A5(Ca2+)m(PS)n proteolipid complexes, which diffuse as single units across the fluid bilayer. Immobilization occurs as the temperature is lowered and the bilayer experiences a liquid-to-gel phase transition, which is accompanied by an increase in the lipid packing density, and consequently reduced spacing between adjacent phospholipid head groups relative to the liquid crystal phase.9 Despite this, the Ca2+ bridges between A5 and PS lipids are collectively strong enough to persist through the bilayer phase transition, and (14) Huang, B.; Perroud, T. D.; Zare, R. N. ChemPhysChem 2004, 5, 1523– 1531.
DiffusiVe Membrane-Associated Proteins
Figure 5. Proposed mechanism for reversible immobilization of diffusive membrane A5 monomers. A5 does not bind to bilayers in the gel phase (top), but it will bind when the bilayer is heated to a fluid liquid phase (center) to form tightly bound A5(Ca2+)m(PS)n proteolipid complexes (black), which diffuse as single units through the membrane leaflet. Cooling the bilayer back down to the gel phase immobilizes A5(Ca2+)m(PS)n complexes, leaving A5 in a native orientation on the gel phase bilayer surface. The process is reversed by heating the bilayer back to the liquid phase.
A5(Ca2+)m(PS)n complexes become immobilized in the bulk gel phase bilayer. Upon reheating the bilayer to above the liquid crystal phase, lipid diffusion resumes and A5(Ca2+)m(PS)n complexes once again diffuse freely. General Discussion. The work presented addresses the continual need for novel approaches to performing single molecule biophysical studies on both diffusive and immobilized membraneassociated proteins. A5 is an ideal candidate for demonstrating the diffusion and immobilization of single membrane-binding proteins because it binds specifically to membranes in an orientation-specific manner, which precludes the use of conventional irreversible immobilization techniques such as nonspecific adsorption and covalent attachment via molecular tethers. Furthermore, A5 monomers diffuse randomly in the membranebound state and, at high membrane occupancy, are known to rapidly form multimers upon membrane binding, which makes true single molecule analysis difficult. Membrane-associated proteins can interact with each other, with other constituents both in and on the membrane, as well as with free soluble species. In the case of A5, protein-protein interactions between diffusive membrane-bound A5 monomers are important for trimer formation, which precedes 2-D crystallization. However, under conditions of low membrane occupancy, we have directly observed isolated, diffusing monomers as well as monomer-monomer collisions that result in short-lived interacting pairs. The lack of multimer formation from these two-body collisions may simply be a consequence of the fact that, at the lower limit of membrane occupancy, the probability for three monomers to collide and form a trimer is exceedingly low. We note, however, that even at higher nonsaturating levels of membrane occupancy where a crowded sea of discrete, diffusing monomers can be directly observed colliding with high frequency, no obvious multimer formation is observed. These observations, which we are currently investigating in more detail, suggest that trimer formation is neither a spontaneous nor a
Langmuir, Vol. 25, No. 5, 2009 3087
significant process under nonsaturating conditions, and that the biological importance of monomeric A5 in ViVo may require further consideration. The nature of the interactions between each monomer in the observed temporary pseudodimers is unclear but may consist, in part, of membrane-mediated long-range interactions at long distance and similar specific protein-protein interactions known to be important for A5 trimer formation at short distance.15 As suggested from the multimodal spot intensity histogram in Figure 3B, an A5 monomer can become immobilized as a monomer, as part of a pseudodimer, or possibly as part of a trimer. Based on this and the two-particle collision trajectories, we speculate that short-lived, weakly interacting dimers may precede formation of tightly bound trimers under conditions of higher levels of membrane occupancy favorable for trimer formation. The successful immobilization of A5 monomers in a native orientation on gel phase bilayers may shed further light on the nature of A5 interactions with PS lipids. Previous studies have shown that PS lipids become immobilized underneath A5 2-D crystalline monolayer domains on model membranes.12 However, it is unclear whether PS immobilization is a consequence of irreversibly bound A5(Ca2+)m(PS)n complexes coalescing into 2-D monolayer crystals, or PS lipids rapidly exchanging between A5(Ca2+)m(PS)n complexes and the bulk bilayer becoming “trapped” underneath growing monolayer crystals by virtue of their domain size. The presented data demonstrate that A5 monomers can be successfully immobilized in a native orientation on gel phase PS-rich bilayers, indicating that A5-PS interactions are strong enough to maintain A5(Ca2+)m(PS)n complexes under gel phase bilayer conditions that normally inhibit A5 binding. These observations support the irreversible nature of Ca2+mediated A5-PS bonds, versus rapid exchange of PS lipids. Under such a scenario, at higher membrane occupancy, closerange interactions between single A5(Ca2+)m(PS)n complexes drive trimer formation, and coalescence of [A5(Ca2+)m(PS)n]3 trimers into extended, immobile monolayer crystalline domains, in turn, immobilizes bound PS lipids. In summary, using a reversible immobilization strategy, we have demonstrated that A5 monomers can be readily observed diffusing across model fluid bilayers at low membrane occupancy, and subsequently be immobilized in a native orientation using a liquid-to-gel bilayer phase transition. Analysis of single A5 monomer diffusion trajectories on fluid bilayers showed that two colliding monomers can form a short-lived, membrane-bound interacting pair, or pseudodimer, with a common diffusion coefficient value approximately half of those observed for free, randomly diffusing A5 monomers. During the liquid-to-gel bilayer phase transition, diffusive A5 monomers become immobilized mostly as isolated monomers, although a small percentage may become immobilized as dimers, or possibly trimers. The fact that A5 monomers show negligible binding affinity to gel phase bilayers but can be immobilized during a liquid-to-gel phase transition suggests that A5(Ca2+)m(PS)n proteolipid complexes are tightly bound and diffuse as single units across fluid membranes. The immobilization method presented here may be useful for performing controlled single molecule studies on diffusive membrane proteins that bind in an orientation specific manner and are therefore not compatible with conventional immobilization techniques such as nonspecific adsorption or covalent attachment via molecular tethers. For example, interactions of various soluble proteins with the N-termini on the nonmembrane binding side of immobilized A5 monomers could (15) Mo, Y.; Campos, B.; Mealy, T. R.; Commodore, L.; Head, J. F.; Dedman, J. R.; Seaton, B. A. J. Biol. Chem. 2003, 278, 2437–2443.
3088 Langmuir, Vol. 25, No. 5, 2009
Han and Boo
be probed, and it is an area that has not been explored at the single molecule level.
Doctoral Fellowship programs. We thank R.N. Zare for use of PCH analysis software.
Acknowledgment. This work was supported by a Korea Science and Engineering Foundation (KOSEF) grant funded by the Korean government (MEST) (Grant No. R01-2006-00010351-0). J.J.H. acknowledges support from the Yonsei University Post-Doctoral Fellowship and Brain Korea 21 Post-
Supporting Information Available: One-photon photon counting histogram analysis. This material is available free of charge via the Internet at http://pubs.acs.org. LA803903J
