12734
Langmuir 2008, 24, 12734-12737
Formation and Characterization of Fluid Lipid Bilayers on Alumina Morgan D. Mager, Benjamin Almquist, and Nicholas A. Melosh* Materials Science and Engineering, Stanford UniVersity, Stanford, California 94305 ReceiVed August 20, 2008. ReVised Manuscript ReceiVed October 6, 2008 Fluid lipid bilayers were deposited on alumina substrates with the use of bubble collapse deposition (BCD). Previous studies using vesicle rupture have required the use of charged lipids or surface functionalization to induce bilayer formation on alumina, but these modifications are not necessary with BCD. Photobleaching experiments reveal that the diffusion coefficient of POPC on alumina is 0.6 µm2/s, which is much lower than the 1.4-2.0 µm2/s reported on silica. Systematically accounting for roughness, immobile regions and membrane viscosity shows that pinning sites account for about half of this drop in diffusivity. The remainder of the difference is attributed to a more tightly bound water state on the alumina surface, which induces a larger drag on the bilayer.
As the basic unit of the cell membrane, lipid bilayers mediate a number of important biological phenomena such as raft formation and the incorporation of integral membrane proteins. Supported bilayers on solid substrates allow investigation of these processes with techniques that are difficult to carry out in bulk solution, including atomic force microscopy (AFM),1,2 quartz crystal microbalance,3,4 and neutron scattering.5 However, to preserve biological functionality the bilayer must retain lateral fluidity in its natural liquid-crystalline state. This requirement precludes rigid binding to the substrate surface. Instead, successful support materials retain a 1 to 2 nm water gap between the lipid headgroup and the substrate surface,5,6 which requires a delicate balance of adhesion, repulsion, and hydration forces.6 Silica is one of few materials with this balance, thus most supported bilayer work has been performed on quartz,7 glass,8 mica,9 or other substrates with silica-based surface chemistry. With the growing use of higher refractive index sensors, anodic alumina ion channel devices, and passivated biomaterial interfaces, lipid deposition onto a wider range of materials, in particular alumina, is becoming critical.10,11 Further constraints are often placed on this balance of forces by the fact that vesicle rupture, the most common deposition method, requires relatively strong lipid/substrate adhesion to induce bilayer formation. Although the rupture mechanism is still a subject of active research, the surface adhesion energy is clearly the dominant factor, determined largely by electrostatics12 and van der Waals forces.13 If the adhesion force is too high, then the lipid bilayer can adhere to the surface and lose lateral fluidity. This appears to be the case for chromium and indium * Corresponding author. E-mail:
[email protected]. (1) Egawa, H.; Furusawa, K. Langmuir 1999, 15, 1660–1666. (2) Hui, S. W.; Viswanathan, R.; Zasadzinski, J. A.; Israelachvili, J. N. Biophys. J. 1995, 68, 171–178. (3) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397–1402. (4) Ebara, Y.; Okahata, Y. J. Am. Chem. Soc. 1994, 116, 11209–11212. (5) Koenig, B. W.; Kruger, S.; Orts, W. J.; Majkrzak, C. F.; Berk, N. F.; Silverton, J. V.; Gawrisch, K. Langmuir 1996, 12, 1343–1350. (6) Castellana, E. T.; Cremer, P. S. Surf. Sci. Rep. 2006, 61, 429–444. (7) Johnson, S. J.; Bayerl, T. M.; McDermott, D. C.; Adam, G. W.; Rennie, A. R.; Thomas, R. K.; Sackmann, E. Biophys. J. 1991, 59, 289–294. (8) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554–2559. (9) Radler, J.; Strey, H.; Sackmann, E. Langmuir 1995, 11, 4539–4548. (10) Steltenkamp, S.; Muller, M. M.; Deserno, M.; Hennesthal, C.; Steinem, C.; Janshoff, A. Biophys. J. 2006, 91, 217–226. (11) Wolfrum, B.; Mourzina, Y.; Sommerhage, F.; Offenhausser, A. Nano Lett. 2005, 6, 453–457. (12) Cha, T.; Guo, A.; Zhu, X. Y. Biophys. J. 2006, 90, 1270–1274. (13) Reimhult, E.; Ho¨o¨k, F.; Kasemo, B. J. Chem. Phys. 2002, 117, 7401– 7404.
tin oxide.14 Conversely, other surfaces such as gold, titania, and alumina adsorb vesicles but do not cause rupture.14,15 On these surfaces, the adhesion energy is likely insufficient to tense the membrane to its failure point. Although the interaction energy can be increased to allow rupture by using charged lipids16,17 or surface functionalization,10,12,18 these techniques require extra processing and potentially mask the desired optical or chemical properties of the substrate materials. In general, the adhesive energy required to maintain a fluid bilayer may be significantly less than necessary for vesicle rupture. Here we explore an alternative bilayer formation technique, bubble collapse deposition (BCD),19 that requires lower adhesion energy than vesicle rupture. In contrast to previous results with other methods, BCD creates fluid lipid bilayers on alumina without altering the surface chemistry or requiring charged lipid headgroups. Our results demonstrate that although alumina-lipid adhesion is not sufficient to induce vesicle rupture, alumina can support fluid bilayers if a less stringent deposition method is utilized. We report for the first time the lateral diffusion coefficient of alumina-supported bilayers, which reflects bilayer viscosity, substrate-lipid hydrodynamic drag, and the nature and density of defect (pinning) sites.20,21 The diffusion coefficient for aluminasupported bilayers was 0.6 µm2/s, which is significantly lower than the 1.4-2.0 µm2/s quoted in literature for the same lipids on silica. By controlling for surface roughness and pinning sites, we systematically examine the underlying reasons for this lower diffusion coefficient and conclude that stronger hydrodynamic coupling at the substrate surface is a major factor. These diffusivity effects are thus mediated by lipid/substrate surface interactions, not by the deposition method. However, the methods outlined in this letter have allowed the study of these properties for the first time because BCD is required for initial bilayer formation on alumina. (14) Groves, J. T.; Ulman, N. U.; Cremer, P. S.; Boxer, S. G. Langmuir 1998, 14, 3347–3350. (15) Reimhult, E.; Hook, F.; Kasemo, B. Langmuir 2003, 19, 1681–1691. (16) Rossetti, F. F.; Bally, M.; Michel, R.; Textor, M.; Reviakine, I. Langmuir 2005, 21, 6443–6450. (17) Gritsch, S.; Nollert, P.; Ja¨hnig, F.; Sackmann, E. Langmuir 1998, 14, 3118–3125. (18) Zhang, L. Q.; Longo, M. L.; Stroeve, P. Langmuir 2000, 16, 5093–5099. (19) Mager, M. D.; Melosh, N. A. Langmuir 2007, 23, 9369–9377. (20) Saxton, M. J. Biophys. J. 1982, 39, 165–173. (21) Kuhner, M.; Tampe, R.; Sackmann, E. Biophys. J. 1994, 67, 217–226.
10.1021/la802726u CCC: $40.75 2008 American Chemical Society Published on Web 10/23/2008
Letters
Langmuir, Vol. 24, No. 22, 2008 12735
Materials and Methods Materials. The lipid used throughout was 1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine (POPC) acquired from Avanti Polar Lipids (Alabaster, AL). The fluorescent dye Texas Red dihexadecanoyl-phosphoethanolamine (TR-DHPE) was purchased from Molecular Probes (Eugene, OR) and used at a concentration of 1 mol % in POPC. Unless otherwise noted, all experiments were performed in a buffer solution of 100 mM NaCl + 10 mM Tris adjusted to pH 7.4 with NaOH. All water used was obtained from a Millipore MilliQ system and had a resistivity greater than 18 MΩ · cm. Substrate Preparation. Silicon wafers with 100 nm of thermally grown oxide were obtained from University Wafer (South Boston, MA). These samples were cleaned prior to use with a 60 s exposure to oxygen/argon plasma in a PlasmaPrep II system (SPI Supplies), resulting in a contact angle of 12 for all measurements). The alumina-supported bilayer exhibited a lower mobile fraction of around 86%. This lower mobile fraction accounts for a 30% drop in the diffusion coefficient, which is much less than the observed decrease.
bilayers, confirming its identity as a single bilayer (Figure 1). FRAP measurements on this alumina-supported bilayer revealed that it was laterally fluid. These results show that previous difficulty in forming bilayers on alumina was a result of the deposition technique and was not an intrinsic material limitation. Vesicle rupture requires a large adhesive energy between the lipid and substrate to induce vesicle failure whereas in BCD the bilayer is mechanically driven to be deposited on the surface, so the adhesion requirements for deposition are less stringent. Once the bilayer is on the surface, the adhesive interaction energies among alumina, lipids, and their respective hydration layers are appropriately balanced to allow a fluid bilayer. The bright spots evident in Figure 1 are attributed to individual adsorbed vesicles. After being photobleached, these spots recovered only to the intensity of the surrounding membrane, demonstrating that the vesicles were adsorbed on top of the fluid bilayer rather than incorporated into it. We fit our FRAP data following Soumpasis23 to characterize relative diffusion rates within the bilayer. The lipid diffusivity was significantly lower on alumina than on silica. Two representative data sets along with the fits to the model and residuals are shown in Figure 2. The measured diffusion coefficient for the bilayers on silica was 1.89 ( 0.32 µm2/s with vesicle rupture and 1.57 ( 0.25 µm2/s with BCD. Both values are within the range of 1.4-2.0 µm2/s quoted in the literature for TR-DHPE diffusing in POPC bilayers on silica surfaces.26-29 The difference in the diffusion coefficient between vesicle rupture and BCD deposited bilayers is not presently understood but is within experimental error. The diffusion coefficient on alumina, however, was much lower at 0.62 ( 0.21 µm2/s, more than 2 times less than on silica. Because the diffusion coefficient is considered to be one of the most important functional properties of a supported bilayer, it is crucial to understand the dramatic difference in these two materials. There are three primary factors that can affect supported bilayer mobility: (1) immobile pinning sites that directly block lateral molecular motion, (2) surface roughness, and (3) changes in the hydrodynamic coupling between the bilayer and the underlying hydration layer.20,21 We investigated each of these factors to identify the origins of lower bilayer mobility on alumina. To quantify the fraction of pinning sites, we measured the mobile fraction of the bilayers at very long times (20 min) after photobleaching, which is 3 times longer than the duration of the previous FRAP experiments. The percent difference between the fluorescent intensity at 20 min and the original value was taken to be the immobile fraction of the bilayer. These (26) Starr, T. E.; Thompson, N. L. Langmuir 2000, 16, 10301–10308. (27) Holden, M. A.; Jung, S. Y.; Yang, T. L.; Castellana, E. T.; Cremer, P. S. J. Am. Chem. Soc. 2004, 126, 6512–6513. (28) Schmidt, T.; Schutz, G. J.; Baumgartner, W.; Gruber, H. J.; Schindler, H. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 2926–2929. (29) Yee, C. K.; Amweg, M. L.; Parikh, A. N. J. Am. Chem. Soc. 2004, 126, 13962–13972.
measurements were performed in areas without adsorbed vesicles to avoid the possibility of a spuriously low result. The results of this experiment are shown in Table 1. For both the vesicle rupture and BCD bilayers on silica, the mobile fraction was around 95%, which is consistent with the literature.26,30 The mobile fraction on alumina was slightly lower at 86%. The change in diffusivity attributable to these pinning sites was estimated from the derivation of Saxton,20 which examined diffusion in the presence of both “fast” and “slow” intermixing components. This analysis results in
D * ) (c - 0.5)(1 - r) + √(c - 0.5)2(1 - r)2 + r
(1)
where c is the fraction of the surface exhibiting slow diffusion and r is the ratio of the slow to fast diffusion coefficients. Taking c to be the immobile fraction (14% for alumina) with a diffusion coefficient equal to zero decreases the effective measured diffusion coefficient by 30%. This is less than half of the decrease observed between silica and alumina. Pinning sites likely account for some of the change in the diffusion coefficient, but it is unlikely that immobile defects alone are responsible for the full difference. Surface roughness can also decrease lipid diffusion coefficients. Seu et al.30 found that doubling the surface roughness decreased lipid diffusion coefficients by 20%, even though both the smooth and rough surfaces exhibited sub-nanometer roughness. We first characterized our samples with AFM (Figure 3). These scans revealed rms roughness values of 0.16 and 0.28 nm, respectively, for the silica and alumina surfaces. To determine if this topographical difference is significant, we performed FRAP measurements on BCD bilayers deposited on a quartz coverslip, which is chemically similar to thermal silicon dioxide but had an rms roughness of 0.44 nm. In spite of the larger rms roughness, the diffusion coefficient on the quartz sample was 1.91 ( 0.14 µm2/s, equivalent to that of the smoother sample. Thus we can conclude that the topographical nature of the alumina substrates is not responsible for the lower observed diffusion coefficient. Having found that roughness and pinning sites cannot fully account for the difference in the diffusion coefficient, we examined hydrodynamic coupling between the membrane and substrate. The diffusion coefficient of lipids near a rigid surface is decreased by drag from the intervening fluid21
D)
(
K1(ε) kT 0.25ε2 + ε 4πηm K0(ε)
ε ) ap
ηw ηmh
)
-1
(2)
(3)
where ηm is the membrane viscosity, ηw is the water viscosity in the hydration layer, h is the thickness of the hydration layer, ap is the van der Waals radius of the lipid molecules, and K0 and K1 are the modified Bessel functions of the second kind of zero and first order. With this model, it is possible to analyze independently the effect of strain (ap), lipid-lipid interactions (ηm), and changes in the hydration layer (h, ηw). Previous studies have shown that bilayers will support only a small percentage of strain before rupturing,31 whereas a strain of even 10% would account for less than a 1% decrease in the diffusion coefficient, so the lower coefficient on alumina could not be the result of differences in molecular packing. Changes in the intrinsic membrane viscosity are also unlikely because the temperature, (30) Seu, K. J.; Pandey, A. P.; Haque, F.; Proctor, E. A.; Ribbe, A. E.; Hovis, J. S. Biophys. J. 2007, 92, 2445–2450. (31) Needham, D.; Nunn, R. S. Biophys. J. 1990, 58, 997–1009.
Letters
Langmuir, Vol. 24, No. 22, 2008 12737
Figure 3. Effect of surface roughness on the diffusion coefficient. AFM scans are shown for a thermally oxidized silicon wafer, an ALD-deposited alumina sample, and a quartz coverslip. Although the quartz sample has a higher rms roughness than the alumina sample, it still has a larger diffusion coefficient, indicating that on sub-nanometer roughness scales, the chemical nature of the substrate is more important than the topography.
solution conditions, deposition method, and lipid and probe species were held constant. These observations indicate that changes in the height h and viscosity ηw of the hydration layer are leading candidates for lowered lipid diffusivity on alumina. Supporting this hypothesis is evidence in the literature for a difference in the state of bound water on different oxides. NMR studies performed on nanoparticle slurries have shown that water near an alumina surface is less mobile than that near a silica surface.32,33 This more tightly bound state would result in a higher effective viscosity in the hydration layer, leading to a lower diffusion coefficient in the supported lipid bilayer. To account for the observed decrease in the diffusion coefficient beyond the 30% decrease due to pinning sites, the quantity ηw/h would have to be approximately 3 orders of magnitude larger on alumina than on silica. Although this is a significant effect, viscosity enhancement of up to 7 orders of magnitude over the bulk has been calculated for confined water near surfaces.34 It is not presently clear whether increased interfacial viscosity is solely responsible for the observed decrease in diffusion or whether the more tightly bound water layer would also lead to a smaller hydration gap. In the angstrom-to-nanometer regime, water viscosity is very sensitive to the exact thickness of the layer, making it difficult to decouple the possible contributions of an altered height versus altered viscosity. Future work including fluorescence interference contrast microscopy (FLIC) or neutron reflectivity studies to measure gap height could help determine the exact nature of the lipid/alumina interaction. (32) Gunko, V. M.; Turov, V. V.; Zarko, V. I.; Voronin, E. F.; Tischenko, V. A.; Dudnik, V. V.; Pakhlov, E. M.; Chuiko, A. A. Langmuir 1997, 13, 1529– 1544. (33) Turov, V. V.; Leboda, R. AdV. Colloid Interface Sci. 1999, 79, 173–211. (34) Major, R. C.; Houston, J. E.; McGrath, M. J.; Siepmann, J. I.; Zhu, X. Y. Phys. ReV. Lett. 2006, 96, 177803.
Conclusions We have shown that it is possible to deposit fluid lipid bilayers on ALD alumina by using BCD. This result demonstrates that fluid bilayers can be formed on a wider range of materials using less restrictive deposition techniques than vesicle rupture. Whereas the diffusion coefficient of the silica-supported bilayers was comparable whether VR or BCD was used, the diffusion coefficient of the alumina-supported bilayers is 0.62 ( 0.21 µm2/s, a factor of 2 to 3 lower than that on silica. This result illustrates that whereas the deposition method can determine whether a bilayer is deposited at all, once the bilayer is on the surface its dynamic behavior is mediated by lipid/substrate interactions. The mobile fraction of lipids on alumina is 85%, as compared to 95% on silica. This lower mobile fraction accounts for some but not all of the difference in the diffusion coefficient. We attribute the rest of the difference in mobility to a more tightly bound water layer at the alumina surface, which in turn increases the hydrodynamic drag on the bilayer. Future work with these alumina surfaces including further characterization of the hydration layer will reveal more about the fundamental nature of lipid/oxide interactions. Acknowledgment. We thank M. Preiner for assistance with the deposition of alumina films. ALD of the alumina dielectric was performed using equipment purchased under contracts AFOSR (F49620-02-1-0383) and ONR (YIP, N00014-01-10569). M.D.M. acknowledges the support of the NDSEG Fellowship. Supporting Information Available: XPS analysis of alumina and silica surfaces. Explanation and simulation-based verification of modifications to FRAP fitting analysis. This material is available free of charge via the Internet at http://pubs.acs.org. LA802726U
