ARTICLE pubs.acs.org/Langmuir
Pore Spanning Lipid Bilayers on Mesoporous Silica Having Varying Pore Size Maria Claesson,† Rickard Frost,‡ Sofia Svedhem,‡ and Martin Andersson*,† †
Department of Chemical and Biological Engineering, ‡Department of Applied Physics, Chalmers University of Technology, SE-412 96 G€oteborg, Sweden
bS Supporting Information ABSTRACT: Synthetic lipid bilayers have similar properties as cell membranes and have been shown to be of great use in the development of novel biomimicry devices. In this study, lipid bilayer formation on mesoporous silica of varying pore size, 2, 4, and 6 nm, has been investigated using quartz crystal microbalance with dissipation monitoring (QCM-D), fluorescent recovery after photo bleaching (FRAP), and atomic force microscopy (AFM). The results show that pore-spanning lipid bilayers were successfully formed regardless of pore size. However, the mechanism of the bilayer formation was dependent on the pore size, and lower surface coverages of adsorbed lipid vesicles were required on the surface having the smallest pores. A similar trend was observed for the lateral diffusion coefficient (D) of fluorescently labeled lipid molecules in the membrane, which was lowest on the surface having the smallest pores and increased with the pore size. All of the pore size dependent observations are suggested to be due to the hydrophilicity of the surface, which decreases with increased pore size.
’ INTRODUCTION Synthetic lipid bilayers have similar properties as membranes of living cells. They have high lipid fluidity and mobility in the lateral plane, which are key features for the functionality of transmembrane proteins that regulates the cellular processes, such as the transport of ions through the membrane as well as transduction pathway for signals.1,2 The biomimicking properties of synthetic lipid bilayers are essential in the development of novel techniques for drug delivery, in catalysis and biosensing devices. The use of lipid bilayers as hosts for transmembrane proteins, such as ion channels, in the development of biosensors is an interesting application.3 7 A successful design of such a biosensing device must result in a membrane having a high mechanical stability and at the same time provide a suitable environment for fully functional transmembrane proteins. The two most commonly used designs in the construction of such devices are based either on an aperture spanning membrane (also called black lipid membrane) or on a supported lipid bilayer (SLBs),3,4,8 which are illustrated in Figure 1. In the aperture spanning design, a membrane is “hanging” freely in a gap between two hydrophobic walls, which results in a flexible system when it comes to controlling the environment on both sides of the membrane combined with the fact that the membrane is accessible from both sides. This makes it possible to insert transmembrane proteins with retained functions. However, a drawback with the design is that the membrane has a relatively poor mechanical stability, leading to short life times of r 2011 American Chemical Society
the membranes.9 More stable membranes can be formed on a solid support, either directly onto an inorganic material or via a polymer cushion.7,10 However, the underlying surface gives rise to sterical constraints underneath the membrane, which may change the properties of the inserted proteins due to the lack of space and possible interactions between the protein and the support.6 In order to leave room underneath the membrane, while at the same time keeping the stability of the solid support, we propose to combine the two designs by having mesoporous silica supported lipid bilayers, where the pore-walls are suggested to provide mechanical stability to the membrane and the pore spanning part of the bilayer would provide a desirable environment for inserted transmembrane proteins. Mesoporous silica is advantageous to use since its properties, such as, pore size, pore geometry, and surface chemistry, can be tailored easily, allowing precise design of the support for the desired application. Lipid bilayers on mesoporous silica particles have previously been studied using, for example, cryo transmission electron microscopy (cryo-TEM), small-angle X-ray scattering (SAXS), confocal microscopy, and flow cytometry, with the common conclusion that homogeneous and continuous bilayers were successfully formed on mesoporous silica particles.11 14 Also, the fluidity of lipid bilayers on both nonfunctionalized and Received: April 19, 2011 Revised: June 7, 2011 Published: June 08, 2011 8974
dx.doi.org/10.1021/la201411b | Langmuir 2011, 27, 8974–8982
Langmuir
ARTICLE
Figure 1. Illustration of the two traditional designs of synthetic lipid bilayers as host for transmembrane proteins: (a) aperture spanning membrane and (b) supported lipid bilayer (SLB), followed by (c) the design investigated in this study using a mesoporous support (figures not drawn to scale).
Figure 2. Illustration on the formation of cubic mesoporous silica using P123, BrijS10, and CTAB as template. First the template molecules form a cubic liquid crystal onto which precursor molecules (tetraethyl orthosilicate, TEOS) are self-assembled. Upon calcination, the template agents oxidize and the material shrinks and hardens. The pore size is tuned by the choice of templating molecule; a larger pore size will be obtained using P123 as template compared to BrijS10 and CTAB as illustrated in the schematic cartoon. The illustration shows the formation of an Im3m structure, as was obtained using P123 or BrijS10 as template molecules.
functionalized silica-based materials having larger pores (20 nm to 4 μm) has been investigated using fluorescence recovery after photobleaching (FRAP).15 18 The outcome from these studies was that the lateral diffusion of lipids (D) decreased and that the immobile fraction of lipids increased significantly on the nonfunctionalized porous surfaces compared to on nonporous surfaces, whereas the results were similar for (amine) functionalized porous silica surfaces (pore size 20 nm17) compared to nonporous surfaces. In these studies, it was shown that bilayers tend to follow the curvature into the pores on the nonfunctionalized surfaces. The obtained reduced mobility of lipids was suggested to be an effect caused by the high curvature of the lipid bilayers near the edges of the holes or due to the walls of the pores, which provide a larger surface area for the lipids to diffuse on compared to on a nonporous surface. One study showed the possibility of forming free spanning lipid bilayers on porous surfaces (pore size ∼80 nm) using micro channel devices in which a bilayer front was pushed across the pores by shear forces. However, the results showed that this was only feasible under high pH conditions (pH 9.5).18 Atomic force microscopy (AFM) has been used to study lipid bilayers on oxidized porous silicon thin film reflectors, which showed that a continuous planar lipid bilayer was formed on such a surface.19 Moreover, vesicle
adsorption on chemically nanostructured surfaces having pores with a diameter and depth of 107 and 25 nm, respectively, has been investigated using quartz crystal microbalance with dissipation monitoring (QCM-D) and AFM. The results from this study showed that pore spanning lipid bilayers were possible to form on the surfaces when the pores were modified by biotinamidocaproyl-labeled bovine serum albumin (BBSA). Additionally, the lipid bilayer formation was shown to be promoted by the edges of the pores.20 In a previous study it was shown that lipid bilayers are successfully formed on nonfunctionalized mesoporous silica, having a pore size of 3.9 nm.21 The lipid bilayer was observed to form more rapidly on mesoporous silica compared to on nonporous silica, as observed from the time required to complete the bilayer formation process using quartz crystal microbalance with dissipation monitoring (QCM-D). The diffusivity coefficient of the lipid bilayers was shown to be lower on mesoporous silica using fluorescent recovery after photo bleaching (FRAP). However, the effect that the pore size of pores being less than 10 nm has on the bilayer formation is still unknown and needs to be further investigated. The aim of the present study was to investigate the influence of the pore size of nonfunctionalized mesoporous silica on the formation of pore-spanning lipid bilayers by vesicle fusion using 8975
dx.doi.org/10.1021/la201411b |Langmuir 2011, 27, 8974–8982
Langmuir QCM-D and FRAP. Atomic force microscopy (AFM) was used to confirm that the bilayer formed intact on mesoporous silica as well as to examine how the vesicles deformed on the surface. Mesoporous silica thin films having specific pore sizes of 2, 4, and 6 nm were synthesized and investigated using transmission electron microscopy (TEM), scanning electron microscopy (SEM), nitrogen adsorption and small-angle X-ray scattering (SAXS) using synchrotron radiation. The focus was on mesoporous silica having a bicontinuous cubic structure, which presents pores available in all directions, which insures that pores are accessible on the surface.
’ EXPERIMENTAL SECTION Cubic mesoporous silica thin films with varying pore size were prepared according to the methods described by Alberius et al.,21,22 Coquil et al.,23 and Besson et al.24 The synthesis path for creating such materials is schematically described step by step in Figure 2. The pore size was controlled and adjusted by the choice of template molecule, synthesis temperature, or aging time. The triblock copolymer P123 (poly(ethylene glycol)20-poly(propylene glycol)70-poly(ethylene glycol20, EO20PO70EO20) and the two surfactants, polyoxyethylen(10) stearylether (BrijS10) and cetyl trimethylammonium bromide (CTAB) (all purchased from Aldrich), were used as template molecules with the aim of forming mesoporous silica with pore sizes in the range of 2 to 6 nm. Cubic mesoporous silica with a pore size of 6 nm was prepared by dissolving 10.4 g of tetraethyl orthosilicate (TEOS, 98%, Aldrich) in a solution containing 5.4 g of diluted hydrochloric acid (0.01 M) and 12.0 g of ethanol (200 proof) at room temperature under vigorous stirring for 20 min. The solution was mixed with a solution containing 1.7 g of P123 in 8.0 g of ethanol. In the case of preparing mesoporous silica with a pore size of 4 nm, 1 g of TEOS was mixed with 0.1 g of HCl (0.01 M) and 0.8 g of ethanol. The solution was vigorously stirred for 20 min after which it was mixed with a solution containing 3.6 g of ethanol, 0.34 g of HCl (0.01 M), and 0.17 g of BrijS10. For preparing mesoporous silica with a pore size of 2 nm, 2.08 g of TEOS was mixed with 0.90 g of HCl (0.01 M) and 1.75 g of ethanol. The solution was vigorously stirred for 20 min followed by mixing with a solution containing 0.47 g of CTAB and 10.0 g ethanol. In all three cases the solutions were vigorously stirred for 30 min before they were put in a refrigerator at 4 °C for 10 min. Thin films were then prepared by spin coating, 4000 rpm, onto glass substrates (microscopy glass slides; 26 76 mm and circular glass slides, diameter 25 mm), titanium disks and silica coated AT-cut QCM-D crystals (purchased from Q-Sense AB, Gothenburg, Sweden) using a photo resist spinner (spin150, sps-Europe). The glass slides and the QCM-D crystals were cleaned prior to the spin coating by sonication in detergent (sodium dodecyl sulfate, SDS) for 15 min, deionized water for 5 min, dried using nitrogen gas, and then treated with UV for 15 min. The films were aged for 24 h followed by a heat treatment step, where the temperature was increased from room temperature at a rate of 1 °C/min to 400 °C, where it was left to dwell for 4 h. The formed mesoporous thin films on glass and titanium were characterized using scanning electron microscopy (SEM), transmission electron microscopy (TEM), nitrogen adsorption, and small-angle X-ray scattering (SAXS). Furthermore, contact angle measurements were performed to measure the surface energy. The used SEM was a Leo Ultra 55 FEG SEM (Leo Electron Microscopy, Cambridge, U.K.). Prior to analysis the surfaces were rinsed in detergent (sodium dodecyl sulfate, SDS) for 15 min and deionized water for 5 min, dried using nitrogen gas, and then UV treated for 15 min, and the surfaces that were coated on glass slides were sputtered with gold (20 nm) using an ion sputter JFC-1100E (JEOL, Tokyo, Japan). TEM analysis was carried out on a JEM-1200 EX II TEM
ARTICLE
operated at 120 kV (JEOL, Tokyo, Japan). The TEM specimen preparation included scratching off the mesoporous thin films from the coated glass, dispersing it in ethanol (200 proof), and sonicating for 2 min. A small amount of the dispersion was dropped on the TEM grids (Lacey Formvar/Carbon 300 mesh, Caspilor, Sweden), followed by degassing in air. Nitrogen adsorption was performed using a Micromeritics Tristar (Norcross, GA). Prior to analysis, the samples were collected by scratching off coated glass plates (10 20 cm) and were then degassed in a vacuum oven at 250 °C for 3 h. SAXS measurements were performed on beamline I711 at the Maxlab synchrotron facility (Lund, Sweden).25 The wavelength used was 1.1 Å and the beam size was 0.12 12 mm. The sample was placed 1.5 m from the detector (two-dimensional Mar165 CCD), and the measurement was carried out for 180 360 s with a q-range (q = 4π sin θ/λ) of 0.1 4.0 nm 1. The surface energy (critical surface tension, γc) of each surface was obtained by constructing Zisman plots.26 This was performed by determining the contact angle of a series of liquids with known surface tension (γLV). The liquids used were pentanol (γLV = 25.7 dyns/cm), octanol (γLV = 28.0 dyns/s), and oleyl alcohol (γLV = 31.7 dyns/cm) (all purchased from Aldrich). The contact angles obtained for the tested liquids were plotted as a function of γLV. From the plot, γc was extrapolated by creating a straight line plot of cos(θ) against γLV. At the intercept between the extrapolated line and the horizontal line, cos(θ) = 1, the contact angle equals 0 degrees, which is defined as γc.26 The contact angle measurements were carried out using a dynamic absorption tester (DAT 1100, FIBRO systems AB, Sweden). Prior to analysis the surfaces were cleaned as described above. Vesicles were prepared from chloroform solutions of 1-palmitoyl2-oleoyl-sn-glycero-3-phosphocholine (POPC) (25 mg/mL) (Avanti polar lipids, Alabaster, AL). The chloroform was removed under vacuum using a rotary evaporator resulting in a thin lipid film. The lipid film was hydrated with PBS buffer (0.010 M phosphate buffer with 0.138 M NaCl and 0.0027 M KCl, Aldrich, pH 7.4) to a concentration of 5 mg/mL and sonicated using a bath sonicator for 10 min. The lipid mixture was extruded using a mini extruder (Avanti Mini-Extruder, Avanti polar lipids, Alabaster, AL) utilizing polycarbonate membranes having pore sizes of 100, 50 and 30 nm; 21 times for each membrane. The obtained vesicles were ∼100 nm in diameter as measured by dynamic light scattering (DLS) (Zetasizer nano-ZS Malvern instruments, Worcestershire, U.K.), results not shown.27 Vesicles used for the FRAP measurements were prepared by the addition of 1 wt % of the fluorescent labeled lipid probe, 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine, (rhodamineDHPE), Invitrogen, Carlsbad, CA. Prior to the QCM-D and FRAP investigations, the vesicles were diluted using PBS buffer to a final concentration of 0.05 mg/mL, and in the case of AFM measurements the vesicle solution was diluted to a final concentration of 0.01 mg/mL. QCM-D discs and glass surfaces were rinsed as described above. The QCM-D measurements were performed using a Q-sense E4 (Q-sense AB, Gothenburg, Sweden). QCM-D crystals (14 mm) coated with silica (Q-sense AB, Gothenburg, Sweden) were used and coated with mesoporous silica thin films as described above. The measurements were performed in flow mode, with a flow rate of 50 μL/min. The setup used for FRAP was based on an inverted Nikon Eclipse Ti-E microscope (Nikon Corporation, Tokyo, Japan) equipped with an EMCCD Andor iXon camera (Andor Technology, Belfast, Northern Ireland). Glass surfaces coated with mesoporous silica thin films were used. AFM was performed in liquid using a PicoSPM microscope (Agilent/Molecular Imaging Inc., Palo Alto, CA) and MSCT-AUMN MicroLever silicon nitride tips (Veeco Europe, Dourdan, France) for studying the lipid bilayer and vesicles and NSC36 silica tips (Mikromash, Estonia) for studying the roughness of the surface. The spring constants of the used cantilevers were 0.01 N/nm for studying the interaction between vesicle and the underlying surface, 0.03 N/nm for studying lipid bilayer coverage, and 1.75 N/m for studying the roughness of the surface (according 8976
dx.doi.org/10.1021/la201411b |Langmuir 2011, 27, 8974–8982
Langmuir
ARTICLE
Figure 3. TEM micrographs of cubical mesoporous silica with a pore size of (a) 6 nm (prepared using P123 as template), (b) 4 nm (prepared using BrijS10 as template), and (c) 2 nm (prepared using CTAB as template). In (d) a SEM image clearly showing the presence of pores on the surface of a mesoporous silica thin film prepared using P123 as template. The thin film was deposited on a titanium disk prior to analysis.
Table 1. Characterization of Cubic Mesoporous Silica Using TEM, Nitrogen Adsorption, and SEMa film thickness
template
pore size
pore size
pore size
agent
(nm)N2 abs.
(nm)TEM
(nm)lit
structure
(nm)
P132
5.2
6
5.622
Im3m22
∼200
BrijS10
3.9
4
3 523
Im3m29,30
∼300
Pm3n24
∼200
∼200
nonporous
CTAB
3.4
2
2
24
a
The pore sizes were found to differ in size when TEM and nitrogen adsorption measurements were compared; however, the pore size was largest for mesoporous silica prepared using P123 as template and smallest for mesoporous silica using CTAB as template. The film thicknesses were obtained by measuring the cross section of the films in SEM micrographs. to specifications). The tip radius was
