Substrate-Induced Structure and Molecular Dynamics in a Lipid

Dec 13, 2017 - The solid-substrate-dependent structure and dynamics of molecules in a supported lipid bilayer (SLB) were directly investigated via ato...
0 downloads 10 Views 2MB Size
Download PDF
Article Cite This: Langmuir XXXX, XXX, XXX−XXX

pubs.acs.org/Langmuir

Substrate-Induced Structure and Molecular Dynamics in a Lipid Bilayer Membrane Toshinori Motegi,*,† Kenji Yamazaki,§ Toshio Ogino,∥ and Ryugo Tero*,†,‡ †

Electronics-Inspired Interdisciplinary Research Institute and ‡Department of Environmental and Life Sciences, Toyohashi University of Technology, Toyohashi, Aichi 441-8580, Japan § Division of Applied Physics, Graduate School of Engineering, Hokkaido University, Sapporo 060-8628, Japan ∥ Department of Engineering, Yokohama National University, Yokohama 240-8501, Japan S Supporting Information *

ABSTRACT: The solid-substrate-dependent structure and dynamics of molecules in a supported lipid bilayer (SLB) were directly investigated via atomic force microscopy (AFM) and single particle tracking (SPT) measurements. The appearance of either vertical or horizontal heterogeneities in the SLB was found to be strongly dependent on the underlying substrates. SLB has been widely used as a biointerface with incorporated proteins and other biological materials. Both silica and mica are popular substrates for SLB. Using single-molecule dynamics, the fluidity of the upper and lower membrane leaflets was found to depend on the substrate, undergoing coupling and decoupling on the SiO2/Si and mica substrates, respectively. The anisotropic diffusion caused by the locally destabilized structure of the SLB at atomic steps appeared on the Al2O3(0001) substrate because of the strong van der Waals interaction between the SLB and the substrate. Our finding that the well-defined surfaces of mica and sapphire result in asymmetry and anisotropy in the plasma membrane is useful for the design of new plasma-membrane-mimetic systems. The application of well-defined supporting substrates for SLBs should have similar effects as cell membrane scaffolds, which regulate the dynamic structure of the membrane.

■

INTRODUCTION

molecular diffusion barrier, and consequently, signal regulation emerges from the anisotropic diffusion of molecules.11 The other mechanisms underlying the cellular functions of the plasma membrane result from vertical lipid asymmetry. The plasma membranes also act as an asymmetric reaction field both on the outside and inside of the membrane owing to the asymmetric lipid composition and viscosity associated with the diffusion of membrane components.12,13 In eukaryotic cells, several functions of asymmetric plasma membranes have been suggested. The regulatory proteins such as synaptotagmin and several isoforms of protein kinase C or membrane structural proteins, such as spectrin, localize at the intracellular leaflet through their interactions with phosphatidylserine.14,15 In addition, the exposure of particular amino phospholipids, such as phosphatidylethanolamine and phosphatidylserine, toward the extracellular leaflet results in membrane budding and endocytosis or recognition by macrophages.16,17 The lipid raft is also related to the lipid asymmetry in the plasma membrane. Because of the asymmetry of long-chain glycolipids, sphingolipids, and cholesterol, which have been suggested to be enriched in the raft, the lipid raft itself does not appear to maintain a stable symmetry between the two leaflets of the lipid

The lipid bilayer membrane, which is a major component of the plasma membrane, is known to control various biological reactions through the two-dimensional diffusion of molecules such as lipids and proteins. The basic model of the plasma membrane is the fluid mosaic model.1 In more progressive membrane models, such as the picket fence or lipid raft models,2−4 static structural proteins, such as cytoskeleton fences, and other obstacles, such as the aggregation of lipids and proteins embedded in the lipid bilayer, hinder the lateral mobility of membrane components and cause anomalous diffusion. Through molecular diffusion, cells can transiently confine membrane proteins to localized regions within the lipid bilayer, such as the specialized high-viscosity membrane patches composed of cholesterol and sphingolipids known as lipid rafts.5 The curved structures of the plasma membrane, as typified by caveolae, also regulate both the degree and direction of molecular diffusion.6,7 The fact that the cell surface membrane receptors and signaling molecules are enriched in the region of these specific membrane structures,8,9 and that the disruption of these structures leads to abnormal signaling and disease, suggests a central role in cell signaling for these curved lipid membranes.10 For excitatory neurotransmission in hippocampal neurons, it has been reported that the thin necks that connect the dendritic spines to the dendritic shafts act as a © XXXX American Chemical Society

Received: September 12, 2017 Revised: November 13, 2017

A

DOI: 10.1021/acs.langmuir.7b03212 Langmuir XXXX, XXX, XXX−XXX

Langmuir

■

bilayer.18−20 In contrast, in the artificial lipid bilayer systems where the membrane proteins are reconstructed, the reproduction of the aforementioned plasma membrane properties, e.g., anisotropic diffusion of molecules and lipid asymmetry, should lead to the emergence of natural membrane protein functions through modulation of the degree of assembly and orientation. Artificial lipid bilayers, as typified by a solid-substratesupported lipid bilayer (SLB), are suitable for physicochemical measurements, such an atomic force microscopy (AFM) and single particle tracking (SPT), because of their two-dimensional structure. Using an SLB with a well-defined lipid composition, the biological function of proteins can be identified.21−23 However, to date, few aggressive attempts to functionalize SLBs with diverse functions such as anisotropic diffusion or lipid asymmetry, have been reported.24,25 SLBs on hydrophilic substrates are known to exhibit different behaviors depending on the state of the substrate surface.26−29 In one example, the degree of membrane dynamics and lipid diffusion was found to be related to the lubricating properties of the water hydration layer between the membrane and the substrate.30,31 On substrates with a high density of uniformly distributed hydroxyl groups, the formation of a stable hydration layer leads to a situation where water molecules are partially ordered and sheared upon stress.31,32 In contrast, using substrates with a low and heterogeneous density of hydroxyl groups leads to a much thinner and highly ordered water layer, which hinders the diffusion of lipids.32,33 However, the functionalization of the SLB in analogy with the plasma membrane is difficult when only controlling the density of the hydroxyl groups on one kind of substrate, which affects the membrane dynamics and the diffusion of lipids. In addition, the substrate dependence of the physical properties and decoupling of two monolayer leaflets of the SLBs has been investigated via AFM imaging and force spectroscopy.34−37 Thus, the investigation of the molecular diffusion at SLBs formed on various support substrates by the direct observation of single molecules is beneficial. In this study, we chose three kinds of oxide substrates: thermally oxidized silicon (SiO2/Si), mica, and single crystalline sapphire (Al2O3(0001)) with a step-and-terrace structure. These substrates have different optical properties such as transparency and refractive index. SPT measurement is the most appropriate method for quantifying how the local membrane heterogeneity affects the diffusion of incorporated molecules over a range of time scales.38 The direct observation of single molecular dynamics in SLB was achieved by using total internal reflection microscopy (TIRM). For the generation of surface localized evanescent light, which is used as the excitation light in TIRM, the number of supporting substrates is limited. This is because optical aberration occurs because of fluorescence detection upon passing through the underlying substrate. Therefore, other microscopic methods that are independent of the substrate optical properties are required. In our diagonal illumination setup, the samples are flipped vertically on the underlying cover glass.39,40 The excitation light is thus directly irradiated on the SLB surface, and the sensitive detection of single-molecule fluorescence on various substrates is possible. The direct imaging method, combining AFM and SPT of the SLB, could be readily adapted to the other substrates in addition to the cover glass. The results provide new insights into the SLB conformation and will contribute to the construction of new plasma membrane mimetic systems.

Article

EXPERIMENTAL SECTION

Sample Preparation. Chloroform solutions of phosphatidylcholine (from chicken eggs, Avanti Polar Lipids, Inc.) and a dye-labeled lipid (see below) of appropriate volumes were mixed in a glass vial and dried with an N2 stream, followed by overnight evacuation. The ratio of the dye-labeled lipid to egg-PC was 1 × 10−6 mol %. After the addition of a buffer solution (100 mM KCl, 25 mM HEPES/NaOH pH 7.4) to the vacuum-dried lipid film, the lipid vesicle suspension was prepared through agitation, freeze−thaw cycles, extrusion through a 100 nm pore polycarbonate filter, and sonication at 24 kHz for 1 h. The 0.3 mM vesicle suspension was stored in a glass vial purged with Ar at 4 °C to prevent the oxidation of the lipids. For the fluorescence microscopy, β-BODIPY530/550-C12-HPC ((2-(4,4-difluoro-5-methyl-4-bora-3a,4a-diaza-s-indacene-3-dodecanoyl)-1-hexadecanoyl-sn-glycerol-3-phosphocholine; Thermo Fisher Scientific, Inc.) or DiI-C18 (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate; Thermo Fisher Scientific, Inc.) was incorporated as a fluorescent probe. BODIPY-HPC has its dye analogue at the lipid acyl chain. In contrast, DiI-C18 has a dye analogue at the headgroup position. The different positions of the dye labels allow the partitioning of the fluorescent probes into each leaflet of the lipid membrane. The substrates were carefully cleaned and prepared immediately before the SLB formation, as described below. The Al2O3(0001) substrate (single crystal sapphire; Namiki Precision Jewel Co., Ltd.) was annealed in air at 1000 °C for the preparation of a single-stepped surface.41 Thermally oxidized Si(100) (SiO2/Si; the thickness of the SiO2 layer was approximately 90 nm) and the singlestepped Al2O3(0001) surface were cleaned in piranha solution (1:4 v/v solution of 30% H2O2 and sulfuric acid) at 180 °C for 30 min and rinsed twice in boiled Milli-Q water for 10 min. The synthetic mica (KMg3AlSi3O10F2; Ito Kikoh Co., Ltd.) was freshly cleaved using sticky tape, followed immediately by SLB formation. The SLB preparation was achieved by the vesicle fusion method through spontaneous processes, such as vesicle adsorption, rupture, and SLB patch fusion, after the immersion of the substrate in the vesicle suspension.29 The SLB formation conditions were determined for the three substrates by epifluorescence observations, as shown in Figure S1. Monotonous and concentric fluorescence recovery after photobleaching indicated that spatially continuous SLBs had formed on all substrates. Single Particle Tracking Measurements. For the SPT measurements, the fluorescence of the single dye lipids was observed by using an inverted fluorescence microscope (IX-71; Olympus, Inc.) equipped with a 60× oil immersion objective (N.A. 1.45). The system temperature was controlled to be 20 °C. With light excitation by a 532 nm laser, the diffusion of the fluorescent probes in the bilayer was recorded using an EM-CCD camera (iXon DU-897; Andor, Inc.) and an image acquisition software (SOLIS; Andor, Inc.) at a frame rate of 30 frames/s (fps). The pixel size of the SPT recording was 275.86 nm. A schematic of the diagonal illumination setup for SPT is shown in Figure 1a. Superimposed images of typically observed diffusion trajectories over a certain CCD acquisition frame are shown in Figure 1b. The mean-square-displacement (MSD) analysis was applied to the diffusion trajectories obtained from SPT. The MSD was calculated according to previous reports.42,43 The square of displacement (r) was averaged over each time interval (τ) and plotted against τ. The individual MSD plots were averaged for each molecular trajectory. For the normal random diffusion case, the MSD plot shows a linear increase with increasing time interval. According to a common equation based on the Einstein−Stokes equation, r2 = 4Dτ, the diffusion coefficient (D) was obtained from the slope of the linear fit of the averaged MSD. Atomic Force Microscopy. AFM topographies were obtained by using PicoScan2500 (Agilent Technologies) equipped with a closedloop scanner. The observation of the surfaces of the bare substrates and SLB was performed in acoustic-ac mode (conventional tapping mode) in a buffer solution. The cantilever was OMCL-AC240TN (Olympus, Inc.), which typically has a spring constant of 2 N/m. AFM image processing was carried out using a commercial program, SPIP 3.0 (Image Metrology A/S). B

DOI: 10.1021/acs.langmuir.7b03212 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

shown in Figure 3a−c. On each substrate, the diffusion of bright objects in the SLBs was observed for a sufficient time to

Figure 1. (a) Diagonal illumination setup for single-molecule fluorescence observation.40 The samples were flipped vertically on the underlying cover glass. Because the excitation light is directly irradiated onto the SLB surface, the sensitive detection of singlemolecule fluorescence on various substrates is possible. (b) A merged image of typical single-molecule trajectories observed in BODIPYHPC-incorporated egg-PC-SLB on the SiO2/Si substrate and the camera acquisition image. The S/N ratio was sufficient to track the single-molecule diffusion over a period of a few seconds.

■

RESULTS AND DISCUSSION AFM images of the surface structures of the SLBs formed on the SiO2/Si, mica, and Al2O3(0001) substrates in HEPES buffer solution are shown in Figure 2. The AFM images for each bare substrate in HEPES buffer solution are shown in Figure S2 of the Supporting Information. While SiO2/Si had an irregular surface on the subnanometer scale, the surface of the mica was atomically flat. In addition, the Al2O3(0001) shows a unique sawtooth patterned surface with a step-terrace topography composed of Al−O layers.41,44 The roughness of the SLB surface expressed as the root-mean-squared roughness and the mean roughness (Rq [nm] andRa [nm], respectively) on each substrate are (0.19, 0.15) on SiO2/Si (Figure 2a), (0.03, 0.06) on mica (Figure 2b), and (0.09, 0.10) on the terrace region of Al2O3(0001) (Figure 2c). These roughness values reflect those of the underlying substrates, and the values are close to those reported in previous AFM studies.45−47 For the Al2O3(0001) substrate, the streaks of the SLB, which were approximately 0.7 nm lower than the nearby membrane surface on the terrace region, were observed only at the substrate steps (Figure 2c). The SLBs spread over the surfaces of all substrates (Figure S1; also see the Experimental Section for details), and the surface structure of the SLBs are dependent on those of the bare substrates. Subsequently, SPT experiments were performed on the SLBs on each substrate by using the diagonal illumination setup (Figure 1).39,40 The typical diffusion trajectories of a single BODIPY-HPC molecule in the SLB on each substrate are

Figure 3. Single-molecule trajectories in SLB on (a) SiO2/Si, (b) mica, and (c) Al2O3(0001) substrate. Near each trajectory in panels, the tracking time is shown.

enable MSD analyses (>1 s with a CCD acquisition rate of 30 fps). Using our microscopy setup, the observation of singlemolecule diffusion in the SLBs on all the substrates was achieved. With previous single-molecule imaging techniques such as TIRM, it is difficult to observe single-molecule diffusion on substrates that have different optical properties from the glass substrate. In the following section, the results from the MSD analyses for the obtained trajectories are shown. The histograms calculated from D for each trajectory observed in the SLB on SiO2/Si are shown in Figure 4a. The nondiffusive component at D < 0.1 μm2/s originates from the adsorbed lipid vesicles, and the diffusive components were fit to a Gaussian curve whose center was at D = 1.2 μm2/s. The appearance of a single distribution of the diffusive component implies that the SLB where the molecules diffuse is both vertically and horizontally homogeneous and lacks any diffusion

Figure 2. AFM topographic images and the height profiles of SLBs on various substrates. The SLBs were formed on the (a) SiO2/Si, (b) mica, and (c) Al2O3(0001) surfaces. Each image was obtained at the size of (a) 1 μm × 1 μm, (b) 1 μm × 1 μm, and (c) 2 μm × 2 μm. In the profile of the Al2O3(0001) case, the blue triangles mark the positions of step edges. C

DOI: 10.1021/acs.langmuir.7b03212 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

There have also been several suggestive reports of the decoupling of leaflets of SLB on mica. Using AFM force measurements, two-step bilayer piercing events have been observed in SLB on mica.34,35 As another suggestive results for decoupling, the asymmetric melting temperature (Tm) in each leaflet of SLB on mica was obtained by the temperaturecontrolled AFM observation.36,37 Because an atomically flat surface of SLB on mica was also observed by our AFM observation, the decoupling of the leaflets could be caused by the greater interactions between the mica and the SLB than between the SiO2/Si and the SLB. In addition, as shown in Figure S3, the frequency of the distribution of the faster diffusive components is lower than that of the slower diffusive component, reflecting the preferential partitioning of DiI-C18 into the upper leaflet of the SLB on mica. This nonequipartition into the leaflets could originate from the presence of hydrophilic dye groups of DiI-C18 near the thin water layer between the SLB and substrate, which is energetically unfavorable. In comparison with a literature example involving the study of decoupling leaflets of dimyristoylphosphatidylcholine (DMPC) lipid bilayers on mica using fluorescence recovery after photobleaching (FRAP)-based experiments,51 the value of D in the fluid phase is very different from that obtained by the present SPT measurements. In the FRAP experiments, a few micrometer-sized spots are measured, and the obtained ensemble-averaged D is more likely to be affected by defects in the bilayer.52,53 In contrast, in the SPT measurements, a range of D can be found in the homogeneous lipid bilayer membranes because the diffusional speed of each molecule is described by the Maxwell−Boltzmann distribution.52,54 If there is membrane heterogeneity associated with the defects, domains, or decoupling leaflets, the consequent multiple distributions of each diffusive component can be clearly distinguished by an approximate expression. The ratio of the components can also be obtained by SPT measurements (Figure S3). Therefore, the present results related to the direct observation of the diffusion of individual molecules allow the quantification of the minute differences in the fluidity between the leaflets of the SLB on mica. For Al2O3(0001), a single but broadened distribution of diffusive components was obtained in the D histogram (gray, Figure 4c). In the MSD analyses, the two-dimensional diffusion eq 1 can be divided into one-dimensional diffusion eq 2 in each direction, such as along the x- and y-axis.

Figure 4. D histograms determined from the SPT for the SLBs formed on the (a) SiO2/Si, (b) mica, and (c) Al2O3(0001) substrates. In each panel, the number of tracking trajectories (n) is shown. (a) For SiO2/ Si, a single distribution of the diffusive component was obtained. (b) For mica, the two distributions of slow and fast diffusive components with the Gaussian fitting curves (blue, red) are shown. (c) For Al2O3(0001), a histogram was calculated from each one-dimensional D in the direction of parallel or perpendicular to the substrate steps (blue or red, respectively). The histogram obtained from the twodimensional D is also shown (gray color).

obstacles (Figure 3a). The BODIPY-HPC, which has a phosphocholine headgroup and a hydrophobic alkyl chain with the dye analogue, are thought to be equally partitioned into both leaflets because of its small steric hindrance and consequent high affinity for the hydrophobic acyl chain of surrounding lipids.48−50 Therefore, the obtained molecular diffusion, which reflects the membrane fluidity of SLB on SiO2/ Si, indicates that the upper and lower leaflets are coupled. In the case of the mica substrate, from the obtained D histogram (Figure 4b), two distributions of the diffusive components were obtained. After fitting, the values of D at the centers of each Gaussian curve were found to be 0.49 and 1.7 μm2/s, respectively. The frequencies of these distributions are almost comparable. Considering the equipartition of BODIPY-HPC into both leaflets of the SLB, where a homogeneous surface was observed by AFM, the faster and slower diffusive components of the SLB on mica are derived from the decoupled upper and lower leaflets, respectively. In the present experiment, the single crystal mica substrate and one type of lipid (phosphatidylcholine) were used. Thus, the heterogeneous environments causing the difference in the diffusion coefficients of dye lipids are associated with the differences between the leaflet facing toward the bulk solution side (upper leaflet) or substrate side (lower leaflet). The faster component has a value of D close to that in previous SLB studies. Therefore, the slower component is in a leaflet subject to applied stronger hindrance than the other; thus, they were reasonably assigned to the leaflet facing to the solid substrate.

⟨r 2⟩ = ⟨rx 2⟩ + ⟨ry 2⟩ = 4Dτ

(1)

⟨rx 2⟩ = 2Dx τ

(2)

and

⟨ry 2⟩ = 2Dyτ

Therefore, the two-dimensional D was divided into onedimensional D’s in directions parallel and perpendicular to the substrate steps (Figure 4c, inset; blue and red arrows, respectively). The distribution in the parallel direction (blue, Figure 4c) is positioned in a higher D region than that in the perpendicular direction (red, Figure 4c). Again, the streak of the depressed SLB at the step was observed, as shown in Figure 2c. Although the detail of the SLB structure at the step requires clarification, the local SLB structure should have a destabilized structure, such as a curved structure, where the diffusion of the dye lipids is restricted. Thus, the observed anisotropic diffusion of molecules could be caused not just by direct collision with the atomic steps of the Al2O3(0001) substrate but also the low affinity of the local SLB structure at the step edges. Considering D

DOI: 10.1021/acs.langmuir.7b03212 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir the equipartition of BODIPY-HPC into both leaflets of the SLB, the appearance of multiple distributions of the diffusive components in each direction with respect to the substrate steps is expected if there is sufficient interaction between the SLB and substrate, as in the case of the mica substrate. The appearance of a single distribution in each direction with respect to the steps reflects that the leaflets of the SLB on the Al2O3(0001) substrate are almost coupled, as we determined from the result of the temporal resolution in this study. It is possible that the SPT experiments with greater time-resolution (