Packing Density Changes of Supported Lipid Bilayers Observed by

Aug 27, 2014 - Copyright © 2014 American Chemical Society ..... The almost heart-shaped bilayer patches and relative locations of the adsorbed vesicl...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCB

Packing Density Changes of Supported Lipid Bilayers Observed by Fluorescence Microscopy and Quartz Crystal MicrobalanceDissipation Chiho Kataoka-Hamai* and Mahoko Higuchi International Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan S Supporting Information *

ABSTRACT: Various properties of supported lipid bilayers such as diffusion and lipid partitioning are well characterized. However, little attention has been paid to their molecular packing density. In this work, the adsorption of 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) vesicles on glass and silicon dioxide was investigated using fluorescence microscopy, quartz crystal microbalance-dissipation (QCM-D), and atomic force microscopy. Fluorescence recovery after photobleaching data showed that the adsorption of large unilamellar vesicles (LUVs) on glass yielded supported bilayers with full mobility under alkaline (pH 8.3) and acidic (pH 3−4) conditions. These fluid bilayers exhibited quite different diffusion constants; those at alkaline pH were ∼10 times larger than those at acidic pH. The reason for this pH dependence was clarified by investigation of the rupture of giant unilamellar vesicles (GUVs) on glass. Fluorescence data revealed that the area of planar bilayer patches increased at alkaline pH. Thus, we conclude that the rapid diffusion in alkaline solution arises from the decreased molecular density. QCM-D data showed that dissipation increased in a stepwise manner during vesicle fusion on silicon dioxide at alkaline pH. We attribute this behavior to the decrease in packing density of planar bilayers.



have been also studied for drug delivery applications,16 actin polymerization through lipid-charge-induced mechanisms,17 and promotion of neuronal cell adhesion to planar bilayers.18 In this study, we found that the diffusion constants of glasssupported DOTAP bilayers change greatly with pH, which motivated us to determine the underlying mechanism for this dependence. We then found that the molecular packing density of DOTAP SLBs depends on pH, probably because of bilayer− surface interactions. This change in packing density readily explains the pH dependence of the mobility of DOTAP SLBs because the diffusion constant and packing density of bilayers essentially correlate with one another.19 Here we examine the adsorption of DOTAP vesicles on glass and silicon dioxide surfaces. We find that SLBs at alkaline pH have a lower packing density than those at acidic pH from evidence derived from different observables. Fluorescence microscopy data show that expansion of the bilayer area in alkaline solution causes diffusion constants to increase. The change in bilayer area is corroborated by two sets of fluorescence microscopy experiments: direct imaging of vesicle adsorption processes and observation of the change in SLB patch area induced by pH variation. In addition, quartz crystal microbalance-dissipation (QCM-D) measurements show that bilayer expansion at alkaline pH causes dissipation to increase.

INTRODUCTION Supported lipid bilayers (SLBs) have been widely used to model the properties of biological membranes and engineer biofunctional surfaces.1−3 SLBs attach with sufficient stability to a solid surface, and can therefore be characterized by various surface analytical techniques. The good stability of SLBs is achieved through attractive interactions between the surface and the bilayer. A question therefore arises whether bilayer− surface interactions affect bilayer properties. The effects of surfaces on the lipid distribution and mobility have been extensively investigated. Asymmetric interleaflet partitioning of charged lipids sometimes occurs in SLBs formed on a charged surface, which is rationalized by an electrostatic or divalent ion-mediated effect.3−6 The adhesion of a SLB to a surface also markedly slows lipid diffusion.7 Each leaflet in a SLB is in a different environment: one is exposed to the bulk solution, while the other faces the solid surface. Nonetheless, several studies have reported similar diffusion constants for both leaflets, which can be explained by strong frictional coupling between the two lipid monolayers.7−10 Another study has shown that the surface-facing leaflet has lower mobility than the bulk-exposed leaflet.11 In contrast, the influence of the surface on the lipid packing density has received little attention. In the current study, we investigated the lipid packing of SLBs consisting of 1,2-dioleoyl3-trimethylammonium-propane chloride salt (DOTAP). In previous studies, this synthetic cationic lipid has been used to elucidate SLB formation processes12−14 and create bilayer coatings on silica nanoparticles.15 DOTAP-containing SLBs © 2014 American Chemical Society

Received: April 21, 2014 Revised: August 24, 2014 Published: August 27, 2014 10934

dx.doi.org/10.1021/jp503905r | J. Phys. Chem. B 2014, 118, 10934−10944

The Journal of Physical Chemistry B



Article

MATERIALS AND METHODS Materials. DOTAP and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Texas Red 1,2-dihexadecanoyl-snglycero-3-phosphoethanolamine triethylammonium salt (TRDHPE) and 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) were purchased from Life Technologies (Carlsbad, CA, USA). The chemical structures of the lipids used in this study are shown in Figure 1. 7X detergent (No. 76-

yield a total lipid concentration of 2 mM. GUVs prepared in 75, 80, and 81 mM sucrose solutions were diluted 1000-fold with solutions containing 30 mM NaCl and 10 mM citric acid− trisodium citrate, pH 3.0 (0.075 osmolar); 30 mM NaCl and 10 mM NaHCO3−Na2CO3, pH 8.3 (0.080 osmolar); and 30 mM NaCl and 10 mM NaH2PO4−Na2HPO4, pH 6.0 (0.081 osmolar), respectively, before adsorption onto glass. These sucrose concentrations were chosen to prevent osmotic shock of the GUVs upon dilution. Adsorption of LUVs on Glass. Borosilicate glass coverslips (22 × 22 mm, no. 1, Matsunami Glass Ind., Osaka, Japan) were soaked for 20 min in 7× detergent/water (1:3) solution, which was heated to clarity. After rinsing with a copious amount of water, the coverslips were dried under a stream of nitrogen, baked at 440 °C for 5 h, and then cooled to room temperature. LUV solutions (0.7 mM lipid) were deposited on the coverslip surfaces in poly(dimethylsiloxane) (PDMS) chambers (attached to the coverslips with light pressure) and incubated for 10 min at room temperature. Each surface was then rinsed with a large amount of buffer to remove nonadsorbed lipids. Fluorescence Microscopy. Fluorescence recovery after photobleaching (FRAP) measurements21 and fluorescence imaging were performed with a microscope (Nikon Eclipse Ti-E, Tokyo, Japan) equipped with an EMCCD camera (iXonEM + 897, Andor Technology, Belfast, Northern Ireland, UK). The pixel size of the camera was 16 × 16 μm. The filter set used to excite TR-DHPE and DiI was FF01-561/12:Di01R561:FF01-609/54 (Semrock, Rochester, NY, USA). In the FRAP experiments, the dyes were photobleached using a 561 nm laser beam with a Gaussian intensity profile (Sapphire LP, 200 mW, Coherent, Santa Clara, CA, USA). The full width at half-maximum of the bleached spots was ∼4 μm. The laser power at the sample plane was 2.8 mW. Fluorescence recovery was recorded under mercury-arc lamp illumination at 40× magnification. The fraction of mobile dye molecules (F) was obtained from the FRAP data according to the following relationship:

Figure 1. Chemical structures of the lipids used in this study.

670-93) was purchased from MP Biomedicals (Santa Ana, CA, USA). The solutions used in all experiments contained 30 mM NaCl and 10 mM buffer. Citric acid−trisodium citrate (pH 3.0, 4.0), H3PO4−NaH2PO4 (pH 3.0), CH3COOH−CH3COONa (pH 4.0, 4.8), NaH2PO4−Na2HPO4 (pH 6.0, 7.2), and NaHCO3−Na2CO3 (pH 8.3) mixtures were used as buffers to maintain solution pH. All chemicals were used as received. Vesicle Preparation. Large unilamellar vesicles (LUVs) were prepared as follows. Lipids were mixed in chloroform to give the desired compositions. TR-DHPE or DiI (0.5 mol %) was added as a fluorescent marker to the lipid preparations used for fluorescence microscopy measurements. Bulk chloroform was removed under a stream of nitrogen, and any remaining solvent was evaporated under vacuum. Buffer solution was then added to the dried lipids to yield a total lipid concentration of 2 mM. The hydration buffer was the same as that used for adsorption. After hydration, the samples were subjected to 10 freeze−thaw cycles (liquid nitrogen/room temperature), followed by ten extrusions through a polycarbonate membrane filter with 100 nm pores using a mini-extruder (Avanti Polar Lipids). Lipid compositions are expressed as mole ratios. The zeta potential of DOTAP LUVs (0.4 mM lipid) was determined using a light scattering analyzer (ELSZ-1000Z, Otsuka Electronics, Osaka, Japan). Giant unilamellar vesicles (GUVs) were prepared as follows.20 DOTAP and TR-DHPE (0.5 mol %) were mixed in chloroform. Bulk chloroform was removed under a stream of nitrogen, and any remaining solvent was evaporated under vacuum. Sucrose solution was added to the dried lipid film to

F=

I∞ − I0 I i − I0

where Ii, I0, and I∞ are the fluorescence intensities integrated over the bleached spot before the bleaching pulse, immediately after the bleaching pulse, and after recovery (stable signal), respectively. Fully mobile bilayer samples (F ≈ 1) were observed at 100× magnification to ensure the homogeneity of the bilayers (data not shown). A value of F ≈ 1 and homogeneous fluorescence indicated that the glass surface was uniformly coated with pure SLBs. Diffusion constants were determined for the pure SLB samples based on the time at which half of the bleached intensity was recovered.21 The rupture of isolated GUVs on glass was observed under mercury-arc lamp illumination at 100× magnification after deposition of vesicle solutions (2 μM lipid) onto the surface of coverslips in PDMS chambers. The glass coverslips were cleaned prior to use in the same manner as those used for LUV adsorption. Data were collected at 38 and 40 ms/frame using pixel areas of 100 × 100 and 120 × 120, respectively. No photobleaching was observed during the measurements. The GUV solutions contained unilamellar and multilamellar vesicles. The lamellarity of the giant vesicles was determined on the basis of their fluorescence patterns as reported previously.22 Only unilamellar vesicles were analyzed to exclude 10935

dx.doi.org/10.1021/jp503905r | J. Phys. Chem. B 2014, 118, 10934−10944

The Journal of Physical Chemistry B

Article

any possible interbilayer interactions. Fluorescence intensities and surface areas for planar bilayer patches were determined using MetaMorph Software (Molecular Devices, Sunnyvale, CA, USA). Net fluorescence intensities obtained by subtracting background intensities were used for analyses. SLB patch areas were estimated for the bilayer region with fluorescence intensities larger than half of the average intensity calculated over the entire patch. QCM-D Measurements. QCM-D experiments were performed with a Q-Sense E1 system (Q-Sense, Stockholm, Sweden). Quartz crystal sensors with a 50 nm silicon dioxide coating (Q-Sense) were immersed in 2% sodium dodecyl sulfate (SDS) overnight or longer. Before use, the quartz crystals were rinsed thoroughly with water, dried under a stream of nitrogen, and cleaned with UV/ozone for 20 min using a low-pressure mercury lamp (OZU-200, Miyata Elevam, Kanagawa, Japan). The measurements were conducted at 23 °C with a flow rate of 50 μm/min. A baseline was established by flowing buffer over the sensor surface. LUVs (0.1 mM lipid) were then adsorbed for 15 min, and buffer was introduced into the chamber to remove nonadsorbed lipids. The fundamental frequency of the quartz crystals was 5 MHz. The frequency and dissipation shifts for the ninth overtone are reported. For normalization to the fundamental frequency, the frequency shifts were divided by nine. The sensors were cleaned after use by sonication in 2% SDS and ethanol for 5 min each, rinsing with water, and then drying under a stream of nitrogen. The surface was further cleaned with UV/ozone for 20 min, followed by immersion in 2% SDS until use. Data for SLBs were analyzed by the single-layer Voigt-type viscoelastic model using QTools 3 Software (Q-Sense).23 The seventh, ninth, 11th, and 13th overtones were used for the fitting. The density and viscosity of the bulk fluid were fixed to 1000 kg/m3 and 0.001 kg/m·s, respectively. Atomic Force Microscopy (AFM). The silicon dioxide surface of QCM-D sensors was observed in AC mode with an AFM (Asylum Research MFP-3D, Santa Barbara, CA, USA). The sensor crystal was fixed to a Petri dish with epoxy adhesive, and then immersed in 2% SDS overnight or longer. The surface was rinsed thoroughly with water, dried under a stream of nitrogen, and then cleaned with UV/ozone for 20 min before use. When the clean surface was observed in air, Olympus OMCL-AC160TS-C2 cantilevers (∼42 N/m, Tokyo, Japan) were used. When SLB formation was examined, DOTAP LUVs (0.1 mM lipid) were adsorbed for 40 s or 15 min in a solution containing 30 mM NaCl and 10 mM NaHCO3−Na2CO3 (pH 8.3). The sample surface was then rinsed with buffer to remove unbound vesicles, followed by observation using Olympus BLAC40TS-C2 cantilevers (∼0.09 N/m). The scan rate for all measurements was 4.0 μm/s. Experimental Values. Values are quoted as the mean ± standard error (SE).

Figure 2. Mobile fraction (F) obtained after adsorption of DOTAP/ TR-DHPE (99.5:0.5) LUVs on glass in solutions containing 30 mM NaCl and 10 mM buffer. The buffer composition and pH are indicated on the vertical axis. Because the mobile fraction determinations were not normally distributed about a mean (values larger than 1 and less than 0 are theoretically impossible), bars show the maximum and minimum for each measurement.26 Points are the mean of 7−11 values determined for different areas of two or three bilayer samples. Circled points show the buffers that yielded almost full mobility (F ≈ 1) for all FRAP data.

Table 1. Zeta Potential of DOTAP Vesicles and Ionic Strength in Solutions Containing 30 mM NaCl and 10 mM Buffer buffer H3PO4−NaH2PO4 citric acid−trisodium citrate CH3COOH− CH3COONa citric acid−trisodium citrate CH3COOH− CH3COONa NaH2PO4−Na2HPO4 NaH2PO4−Na2HPO4 NaHCO3−Na2CO3

pH

zeta potential (mV)a

ionic strength (mM)b

3.0 3.0

75 ± 1.9 40 ± 2.0

39 35

4.0

67 ± 3.2

31

4.0

35 ± 1.0

42

4.8

71 ± 0.89

35

6.0 7.2 8.3

43 ± 0.41 39 ± 0.26 45 ± 0.48

41 50 40

a

The zeta potential was determined for unlabeled LUVs. Values (±SE) are the average of four or five determinations. bIonic strength was calculated based on the dissociation constants of the solutes.

1);24,25 thus, the TR-labeled vesicles are positively charged in all of the solutions. After adsorption of these cationic vesicles, the mobile fraction (F) varied with buffer type. Phosphate (pH 3.0), citrate (pH 3.0), acetate (pH 4.0), and carbonate (pH 8.3) buffers (Figure 2, circled points) yielded supported bilayers with full mobility (F ≈ 1) for all experiments, while the other buffers (pH 4.0−7.2) did not always result in F ≈ 1. Thus, vesicle fusion seems to occur readily at acidic and alkaline pH, and not near neutral pH. The reason for this pH dependence is unclear. A detailed discussion of these results is provided in the Supporting Information. Dependence of Diffusion Constant on pH. FRAP measurements of TR-DHPE revealed SLB diffusion constants with interesting behavior (Table 2). The diffusion constant at pH 8.3 was an order of magnitude larger than those at pH 3−4.



RESULTS AND DISCUSSION Dependence of Mobile Fraction on pH. We examined the FRAP of TR-DHPE after adsorption of DOTAP LUVs on glass (Figure 2). Adsorption was conducted in solutions containing 30 mM NaCl and 10 mM buffer. The buffer systems used to control solution pH are indicated on the vertical axis. The zeta potential of DOTAP LUVs was positive in all buffers (Table 1) because of the presence of a trimethyl group in the headgroup (Figure 1). TR-DHPE also maintains its net negative charge over the pH range of 2.5−12.3 (Figure 10936

dx.doi.org/10.1021/jp503905r | J. Phys. Chem. B 2014, 118, 10934−10944

The Journal of Physical Chemistry B

Article

Table 2. Diffusion Constants for SLBs on Glassa diffusion constant (μm2/s) buffer

pH

DOTAP/TR-DHPE (99.5:0.5)

DOTAP/DiI (99.5:0.5)

DOPC/TR-DHPE (99.5:0.5)

H3PO4−NaH2PO4 citric acid−trisodium citrate CH3COOH−CH3COONa NaH2PO4−Na2HPO4 NaHCO3−Na2CO3

3.0 3.0 4.0 7.2 8.3

0.35 ± 0.054 0.43 ± 0.12 0.71 ± 0.17 1.9 ± 0.38 4.1 ± 0.39

0.45 ± 0.022 1.1 ± 0.026 1.3 ± 0.10 2.4 ± 0.22 6.2 ± 0.22

1.5 ± 0.023 1.1 ± 0.068 0.97 ± 0.053 1.8 ± 0.034 1.7 ± 0.013

a SLBs were formed by vesicle fusion in solutions containing 30 mM NaCl and 10 mM buffer. Diffusion constants were determined by FRAP. Values (±SE) are the average of 4 to 11 determinations (different areas on one to three bilayer samples). The data in Figure 2 were used to determine the diffusion constants for DOTAP/TR-DHPE bilayers. DOTAP/DiI and DOPC/TR-DHPE bilayers had a mobile fraction of 0.94−1.0.

with a heart-like shape in ∼10−20 ms. In the current study, rupture events were observed at 38−40 ms/frame, which is slower than the time scale of pore expansion. Therefore, expanding pores were not imaged. A rupture event was instead identified by a sequence of two or three images consisting of a fluorescence pattern of a free or adsorbed vesicle, that of a rupturing vesicle (if any), and that of a resultant planar bilayer patch (Figure 3A,B). Previous studies have demonstrated that such fluorescence data can be ascribed to rupture processes.22 Representative data at pH 3.0 (Figure 3A) showed that a GUV in the bulk solution (−4.485 s) adsorbed on the surface (−4.447 s) and then ruptured (−0.038 s) to create a planar bilayer patch (0 s). At pH 8.3 (Figure 3B), a free GUV (−14.732 s) adsorbed on the surface (−14.694 s) and then ruptured (−0.038 s) to form a SLB patch (0 s). A free vesicle produced a blurred fluorescence image. After attachment to glass, the adsorbed area in a deformed vesicle was on the focal plane and therefore clearly imaged. The almost heart-shaped bilayer patches and relative locations of the adsorbed vesicles and final bilayer patches suggest that these rupture events occurred through the asymmetric pathway, as reported previously.22 A time of 0 s indicates the time at which a bilayer patch was completely formed. Because the first frame with a bilayer patch (−0.038 s in Figure 3A,B) may include contributions from both the rupturing vesicle and resulting planar bilayer, the second frame was taken as 0 s. The image at −0.038 s in Figure 3B contains obvious fluorescence emission from the rupturing process. The negative and positive signs for times represent before and after complete rupture, respectively. Localization of TR-DHPE in Adsorbed GUVs. It is noteworthy that the bilayer patches at 0 s exhibited inhomogeneous fluorescence emission. At pH 3.0 (Figure 3A, 0 s), the bilayer area that had initially interacted with the surface before rupture showed higher fluorescence intensity than other areas. At pH 8.3 (Figure 3B, 0 s), the opposite fluorescence intensity was observed. Because photobleaching was not detected, the fluorescence inhomogeneity reflected localization of the TR-DHPE molecules. The fluorescence signals then became uniform over time, probably because of diffusion. The probe distribution at 0 s was analyzed in a quantitative manner (Figure 3C). Iads(0) and Iun(0) denote fluorescence intensities per unit area determined for a given bilayer region within a planar bilayer patch at 0 s (Figure 3C, inset). Iads(0) was determined for the region that had been initially adsorbed on the surface before rupture, while Iun(0) was determined for the region that had not directly contacted with the surface in the original adsorbed vesicle. Thus, a value of Iads(0)/Iun(0) = 1 indicates that TR-DHPE has uniform density over the entire bilayer patch. In contrast, Iads(0)/Iun(0) > 1 (or