Stable Supported Lipid Bilayers on Zirconium Phosphonate Surfaces

Aug 27, 2009 - Supported lipid bilayers that can fully represent biological cell membranes are attractive biomimetic models for biophysical and biomed...
2 downloads 0 Views 3MB Size
pubs.acs.org/Langmuir © 2009 American Chemical Society

Stable Supported Lipid Bilayers on Zirconium Phosphonate Surfaces Roxane M. Fabre and Daniel R. Talham* Department of Chemistry, University of Florida, Gainesville, Florida 32611-7200 Received May 29, 2009. Revised Manuscript Received July 28, 2009 Supported lipid bilayers that can fully represent biological cell membranes are attractive biomimetic models for biophysical and biomedical applications. In this study, we develop a new approach to engineering stable supported lipid membranes and demonstrate their utility for the study of protein-membrane interactions. This system uses a zirconium phosphonate monolayer to modify a substrate and generate a reactive surface that tethers the lipid membrane via a highly covalent bond between surface zirconium ions and divalent phosphate groups in the lipid assembly, for example, from phosphatidic acid. An advantage of the approach is that the zirconium phosphonate modifier can be applied to nearly any surface, allowing the same methods to be used on glass, gold, silicon, or plastic supports. The lipid bilayers are formed by vesicle fusion, either directly on the zirconated surface to form symmetric bilayers or following deposition of a Langmuir-Blodgett lipid layer to generate asymmetric bilayers. The membrane formation was studied by surface plasmon resonance enhanced ellipsometry (SPREE) as the phosphatidic acid composition was varied. We found that 10% of phosphatidic acid generates supported lipid bilayers stable to dehydration. The two-dimensional fluidity of these systems was characterized by fluorescence recovery after photobleaching (FRAP) measurements. Uniform, mobile supported lipid bilayers with lipid diffusion coefficients of ∼4 μm2/s were obtained. SPREE was also used to measure kinetic parameters of the binding of melittin, a bee venom peptide, to asymmetric lipid bilayers with different electrostatic properties. The results are comparable to those obtained by other research groups, confirming that the model membranes behave as expected. Overall, the results of this study prove that supported lipid bilayers on zirconium phosphonate inorganic surfaces make up an attractive biomimetic system that is highly stable, can be used with multiple substrates, and does not require any biomolecule synthetic modifications.

1. Introduction Supported lipid bilayers have emerged as important biomimetic models for investigating many characteristics of cell membranes, ranging from fundamental studies of lipids to the mechanisms of membrane-bound proteins. Several routes to these supported membrane models have now been described, the most commonly used of which are vesicle fusion onto a hydrophilic support to generate a bilayer and transfer of a lipid monolayer onto a hydrophobic support.1-6 The latter approach can be used to form asymmetric bilayer structures. Many of these membrane mimics are viable models for some applications, but most fall short of being generally applicable. Bilayers supported on glass have good lipid mobility due to a thin layer of water between the bilayer and support, but these same systems are poor models for studies of transmembrane processes because the water layer is too thin. Approaches to adding space between the bilayer and the support have been developed, including the use of polymer cushions or other spacers.7-9 These strategies are effective but often involve specialized reagents that are limited to a unique surface material. Another concern is bilayer stability, as they are

typically sensitive to vibration or mechanical perturbation and are generally unstable to air. Recently, a number of strategies for generating air-stable supported bilayers have been developed, including the use of a nanoglassified gold surface,10 lipopolymer membrane,11 hydrogel mesh,12 added cholesterol,13 and modified lipids.14 Each method fulfills its role of mimicking the cell membrane, but many are also limited to specialized conditions. This work describes stable supported lipid bilayers that are formed on zirconium phosphonate-modified surfaces. Zirconium phosphonate surfaces are known to efficiently bind phosphonates and phosphates, and we have previously used this surface to immobilize DNA and as a platform for oligonucleotide arrays.15,16 We showed that the zirconium phosphonate surface selectively binds phosphorylated DNA over nonphosphorylated DNA, which can be attributed to the specific binding of the terminal divalent phosphate in preference to weaker binding of the phosphodiester backbone. We consider several potential advantages to extending this surface to supported bilayers. Strong binding of phosphate to the zirconium phosphonate surface could be used to lend stability to the supported bilayer structure. Via inclusion of

*To whom correspondence should be addressed. E-mail: talham@chem. ufl.edu. Telephone: (352) 392-9016.

(10) Han, J. H.; Taylor, J. D.; Phillips, K. S.; Wang, X. Q.; Feng, P. Y.; Cheng, Q. Langmuir 2008, 24, 8127–8133. (11) Albertorio, F.; Diaz, A. J.; Yang, T. L.; Chapa, V. A.; Kataoka, S.; Castellana, E. T.; Cremer, P. S. Langmuir 2005, 21, 7476–7482. (12) Jeon, T. J.; Malmstadt, N.; Schmidt, J. J. J. Am. Chem. Soc. 2006, 128, 42– 43. (13) Deng, Y.; Wang, Y.; Holtz, B.; Li, J. Y.; Traaseth, N.; Veglia, G.; Stottrup, B. J.; Elde, R.; Pei, D. Q.; Guo, A.; Zhu, X. Y. J. Am. Chem. Soc. 2008, 130, 6267– 6271. (14) Purrucker, O.; Fortig, A.; Jordan, R.; Tanaka, M. ChemPhysChem 2004, 5, 327–335. (15) Nonglaton, G.; Benitez, I. O.; Guisle, I.; Pipelier, M.; Leger, J.; Dubreuil, D.; Tellier, C.; Talham, D. R.; Bujoli, B. J. Am. Chem. Soc. 2004, 126, 1497–1502. (16) Monot, J.; Petit, M.; Lane, S. M.; Guisle, I.; Leger, J.; Tellier, C.; Talham, D. R.; Bujoli, B. J. Am. Chem. Soc. 2008, 130, 6243–6251.

(1) Sackmann, E. Science 1996, 271, 43–48. (2) Tamm, L. K.; McConnell, H. M. Biophys. J. 1985, 47, 105–113. (3) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554–2559. (4) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397–1402. (5) Keller, C. A.; Glasmastar, K.; Zhdanov, V. P.; Kasemo, B. Phys. Rev. Lett. 2000, 84, 5443–5446. (6) Reviakine, I.; Brisson, A. Langmuir 2000, 16, 1806–1815. (7) Kiessling, V.; Crane, J. M.; Tamm, L. K. Biophys. J. 2006, 91, 3313–3326. (8) Naumann, R.; Schiller, S. M.; Giess, F.; Grohe, B.; Hartman, K. B.; Karcher, I.; Koper, I.; Lubben, J.; Vasilev, K.; Knoll, W. Langmuir 2003, 19, 5435–5443. (9) Diaz, A. J.; Albertorio, F.; Daniel, S.; Cremer, P. S. Langmuir 2008, 24, 6820–6826.

12644 DOI: 10.1021/la901920y

Published on Web 08/27/2009

Langmuir 2009, 25(21), 12644–12652

Fabre and Talham

Article

utilized Langmuir-Blodgett (LB) methods in our laboratory and find this process to be efficient and highly reproducible.27,28 Solid supports (gold or glass) are first rendered hydrophobic and then coated with the zirconium phosphonate film.27 Fusion of vesicles to form symmetric bilayers is similar to procedures used on other surfaces.5,29 Asymmetric lipid bilayers are formed by LB transfer of the inner lipid layer onto the zirconium phosphonate surface followed by fusion of vesicles of different lipid composition to add the distal layer. Interactions between the zirconium phosphonate monolayer and lipid layers were studied by ellipsometry and X-ray photoelectron spectroscopy (XPS). Vesicle fusion and the stability of the lipid assemblies under various conditions were investigated by surface plasmon resonance-enhanced ellipsometry (SPREE). In addition, fluorescence recovery after photobleaching (FRAP) was used to study the fluidity of the supported membrane on the zirconium phosphonate film. To validate the model as a membrane mimic, interaction with the membrane protein melittin was studied as the composition of the lipid bilayers was changed, reproducing trends observed with other supported lipid bilayer systems.

2. Experimental Section

Figure 1. Schematic showing symmetric and asymmetric lipid bilayers on zirconium octadecylphosphonate (ODPA)-modified surfaces. (a) Fusion of vesicles to form symmetric bilayers. (b) Langmuir-Blodgett transfer of the inner lipid layer followed by vesicle fusion to form asymmetric lipid bilayers.

a percentage of phosphatidic acid groups in the bilayer, strong covalent linkages can be formed to the surface to anchor the assembly, increasing stability. Furthermore, our methods for preparing zirconium phosphonate-modified surfaces can be extended to nearly any surface material, including glass, gold, or plastic, allowing the same chemistry for preparing the bilayers to be used for different supports. Often, different analytical techniques require specific surfaces, such as gold for SPR, glass or quartz for fluorescence, or silicon for IR spectroscopy, and many procedures for preparing supported bilayers are specific for one surface or another. A technique that can be used for different supports can be helpful. Finally, it can be beneficial to identify supports that can be used to immobilize different biomolecular structures without extensive modification. Zirconium phosphonate has already been used to directly immobilize DNA,17 phosphopeptides,18 and phospholipids,19 and we show here that it can also be used for supported lipid bilayers. Two methods for constructing supported lipid bilayers on zirconium phosphonate-modified surfaces are illustrated in Figure 1. Different procedures are available for modifying surfaces with zirconium phosphonate layers,20-26 but we have long (17) Lane, S. M.; Monot, J.; Petit, M.; Tellier, C.; Bujoli, B.; Talham, D. R. Langmuir 2008, 24, 7394–7399. (18) Zhao, L. A.; Wu, R. A.; Han, G. H.; Zhou, H. J.; Ren, L. B.; Tian, R. J.; Zou, H. F. J. Am. Soc. Mass Spectrom. 2008, 19, 1176–1186. (19) Oberts, B. P.; Blanchard, G. J. Langmuir 2009, 25, 2962–2970. (20) Lee, H.; Kepley, L. J.; Hong, H. G.; Akhter, S.; Mallouk, T. E. J. Phys. Chem. 1988, 92, 2597–2601. (21) Lee, H.; Kepley, L. J.; Hong, H. G.; Mallouk, T. E. J. Am. Chem. Soc. 1988, 110, 618–620. (22) Putvinski, T. M.; Schilling, M. L.; Katz, H. E.; Chidsey, C. E. D.; Mujsce, A. M.; Emerson, A. B. Langmuir 1990, 6, 1567–1571. (23) Vermeulen, L. A.; Thompson, M. E. Chem. Mater. 1994, 6, 77–81. (24) Vermeulen, L. A.; Thompson, M. E. Nature 1992, 358, 656–658. (25) Vermeulen, L. A.; Snover, J. L.; Sapochak, L. S.; Thompson, M. E. J. Am. Chem. Soc. 1993, 115, 11767–11774. (26) Kohli, P.; Rini, M. C.; Major, J. S.; Blanchard, G. J. J. Mater. Chem. 2001, 11, 2996–3001.

Langmuir 2009, 25(21), 12644–12652

2.1. Materials. Reagents were obtained from commercial sources and used as received. The monosodium salt of 1,2-dipalmitoyl-sn-glycero-3-phosphate (DPPA), 1-palmitoyl-2-oleoyl-snglycero-3-phosphocoline (POPC), 1-palmitoyl-2-oleoyl-sn-glycero3-[phospho-rac-(1-glycerol)] (POPG), and 1-oleoyl-2-{6-[(7-nitro2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-phosphocholine (NBD-PC) were purchased in chloroform from Avanti Polar Lipids (Alabaster, AL). The refractive index matching fluid diodomethane and zirconyl chloride (98%) were purchased from Sigma-Aldrich (St. Louis, MO). The buffer component trizma hydrochloride, sodium dodecyl sulfate (SDS), and ethanol were from Sigma. Slides used for SPREE experiments were made of 28.5 nm of gold evaporated on a 4 nm chromium adhesion layer on a clean SF10 glass slide (Schott Glass). Glass microscope slides for FRAP experiments were from Gold Seal (Portsmouth, NH). MilliQ water with a resistivity of 17.9 MΩ cm was used for all experiments. The buffer used throughout the experiments consisted of 10 mM trizma hydrochloride and 100 mM sodium chloride (pH 7.4). The peptide melittin was obtained from G. E. Fanucci and used in the tris buffer. 2.2. Substrate Preparation. Gold slides were cleaned using a solution of 15% ammonium hydroxide, 15% hydrogen peroxide, and 70% water at 60-80 C for 5 min, rinsed with Milli-Q water, and dried with a flow of nitrogen. Gold slides were then plasma cleaned for 10 min, using a Harrick Plasma cleaner/sterilizer. Slides were immersed in a 1 mM octadecylmercaptan (ODM) solution in ethanol for 16 h to be rendered hydrophobic. Then, the substrates were rinsed with ethanol and dried under nitrogen. Glass slides were rendered hydrophobic with an octadecyltrichlorosilane layer. 2.3. Zirconium Phosphonate-Modified Surfaces. Zirconium phosphonate monolayers were prepared using a multiplestep Langmuir-Blodgett deposition technique.28 Monolayers were transferred using a KSV 3000 Teflon-coated LB trough with hydrophobic barriers (KSV Instruments, Stratford, CT). The surface pressure was measured with a filter paper Wilhelmy plate. The aqueous subphase was 2.6 mM CaCl2 adjusted to pH 7.4 with a potassium hydroxide solution. Octadecylphosphonic acid was spread from a 0.30 mg/mL chloroform solution in the LB trough. The solvent was allowed to evaporate for 10 min, and the monolayer was compressed at a rate of 10 mN min-1 to reach (27) Byrd, H.; Whipps, S.; Pike, J. K.; Ma, J. F.; Nagler, S. E.; Talham, D. R. J. Am. Chem. Soc. 1994, 116, 295–301. (28) Byrd, H.; Pike, J. K.; Talham, D. R. Chem. Mater. 1993, 5, 709–715. (29) Richter, R. P.; Berat, R.; Brisson, A. R. Langmuir 2006, 22, 3497–3505.

DOI: 10.1021/la901920y

12645

Article

Fabre and Talham

a surface pressure of 20 mN min-1. Hydrophobic substrates were then dipped down through the monolayer surface at a rate of 8 mm min-1 into a vial sitting in the subphase, transferring the layer. The vial containing the slide was removed from the trough, and 3 mM zirconyl chloride was added to bind a monolayer of Zr4þ ions at the organic template. After 4 days, the slide was rinsed with water and kept in water before being used.

2.4. Lipid Vesicle Solutions and Formation of Supported Lipid Bilayers. Different compositions, given as percentages by mass, of DPPA, POPC, and POPG lipid mixtures were used to form the small unilamellar vesicles.2,30 The chloroform stock solution of the lipid mixture was dried via a nitrogen stream to form a uniform dry lipid film. The film was hydrated in tris buffer to yield a total lipid concentration of 0.5 mg/mL with gentle vortex mixing leading to multilamellar vesicles. Five freeze-thaw cycles were performed on the lipid solution to yield large unilamellar vesicles. The lipid suspension was extruded 11 times through polycarbonate membranes with a pore diameter of 100 nm. Small unilamellar vesicles were also prepared by sonication for 3  10 min. During extrusion, the temperature was kept above the gel to liquid crystalline phase transition temperature. Vesicles were usually used within 1 day of preparation. Symmetric supported lipid bilayers were formed by the adsorption and rupture of phospholipid vesicles directly on the zirconium phosphonate surface in a flow cell. Two steps were used to form asymmetric supported lipid bilayers. After the formation of the zirconium phosphonate layer by LB deposition, the proximal monolayer was also formed using the LB technique. A 1 mM lipid solution in a 3:1 (v/v) chloroform/methanol mixture was spread on pure water. After evaporation of the organic solvent, the monolayer was compressed at a speed of 10 mN min-1. Lipid monolayers were deposited on the hydrophilic surface at a constant surface pressure of 30 mN m-1 by pulling the substrate upward through the air-water interface at a rate of 8 mm min-1. The distal monolayer was formed by vesicle fusion in the flow cell.31

2.5. Ellipsometric and X-ray Photoelectron Spectroscopy (XPS) Measurements. Ellipsometry was used to characterize multilayer thin films and particularly the thickness of the inner layer of the supported lipid bilayer formed by the LB method. Ellipsometric angles and spatially resolved contrast images were acquired using a commercial EP3-SW imaging system (Nanofilm surface analysis). The ellipsometer employed a frequency-doubled Nd:YAG laser (adjustable power up to 20 mW) at 532 nm. Ellipsometry measures two parameters, the reflectance coefficient, Ψ, and phase shift, Δ, that describe the change in light polarization. Equation 1 relates these parameters to the complex Fresnel reflection coefficients for parallel (p) and perpendicular (s) polarized light: F ¼

Rp ¼ tan Ψ expðjΔÞ Rs

w2 4τ1=2

ð2Þ

F¥ -F0 Fi -F¥

ð3Þ

D ¼

M ¼

where w is the width of the beam, τ1/2 is the half-time of the fluorescence recovery, F0 is the fluorescence before bleaching, Fi is the fluorescence right after bleaching, and F¥ is the fluorescence after recovery.

3. Results ð1Þ

For each layer, the angle of incidence varied from 54 to 71 and Δ values were recorded every 0.2. Using Fresnel equations and an appropriate model, layer thicknesses can be determined. XPS was performed using a UHV XPS/ESCA PHI 5100 system. Survey scans and multiplex scans (Au 4f, P 2p3, Zr 3d) were taken with an Al KR X-ray source using a power setting of 300 W and a take-off angle of 45 with respect to the surface. Peak areas were determined using commercial XPS analysis software and Shirley background subtraction. 2.6. SPR Enhanced Ellipsometry (SPREE). SPR provides the means to quantify the equilibrium constants and kinetic constants in very sensitive and label-free biochemical experiments.32 (30) Salafsky, J.; Groves, J. T.; Boxer, S. G. Biochemistry 1996, 35, 14773–14781. (31) Crane, J. M.; Kiessling, V.; Tamm, L. K. Langmuir 2005, 21, 1377–1388. (32) Besenicar, M.; Macek, P.; Lakey, J. H.; Anderluh, G. Chem. Phys. Lipids 2006, 141, 169–178.

12646 DOI: 10.1021/la901920y

Under SPR conditions, ellipsometric parameters give a larger enhancement of detection sensitivity compared to simple SPR techniques.33,34 The minimum detectable signal change of the instrument used in this work was ∼10 mdeg in Ψ which results in a thickness precision of 0.1 nm. The angle of incidence that provides the highest sensitivity in Ψ measurements was chosen and was between 64 and 66 for the kinetic experiments. The ellipsometer, described in an earlier section, was coupled with an SPR cell for SPR experiments in air or in liquid. Linearly p-polarized light was directed through a 60 equilateral prism coupled to a gold-coated glass slide via diodomethane oil as an index matching fluid in the Kretschmann configuration.35 After the vesicles were injected into the flow cell, the flow was stopped for 30 min to 1 h to allow vesicle fusion to occur on the surface. The formation of the lipid bilayer membrane was monitored by recording Ψ data versus time. Buffer was flowing for 30 min to remove free vesicles. The analysis programs AnalysR (Nanofilm) and BIAevaluation (Biacore) for the two-state model were used to further fit the experimental results. 2.7. Fluorescence Recovery after Photobleaching. FRAP experiments were conducted on a confocal laser scanning microscopy system consisting of an Olympus IX-70 inverted microscope with an Olympus Fluoview 500 confocal scanning system. The NBD-POPC was excited with the 458 nm line of an argon ion laser, and emission was detected using a 505 nm long pass filter. A 20 objective lens was used for lipid bilayers, and images were recorded using a CCD camera. The bleaching time was set to 25 s, and background-corrected intensities of the bleaching spot were determined for each image taken until the maximum recovery of fluorescence was reached. A modified glass substrate was fixed at the bottom of the flow cell. Generally, 1 mL suspensions of 0.75 mg/mL lipid vesicles, fluorescently labeled with 2% NBDPC, were applied. After incubation with the vesicle solutions for 30 min, the surface was washed extensively with tris buffer and the FRAP experiment was performed. Diffusion coefficients (D) and mobilities (M) were obtained using the following equations:

3.1. Symmetric Supported Lipid Bilayers. The SPREE technique and a scheme of the different components of the supported bilayer assembly are shown in Figure 2. As shown in Table 1, a seven-layer model was used to fit the experimental data and calculate the thickness of the lipid bilayers assembled onto the zirconium phosphonate surfaces by vesicle fusion. The optical parameters for SF10, Cr, Au, ODM, and buffer have been widely studied in the past.36 The thickness of each of these layers was determined by ellipsometry (Table 1), confirming the fabrication specifications of the Cr and Au layers and expectations from the (33) Westphal, P.; Bornmann, A. Sens. Actuators, B 2002, 84, PII S09254005(02)00037-0. (34) Nabok, A. V.; Tsargorodskaya, A.; Hassan, A. K.; Starodub, N. F. Appl. Surf. Sci. 2005, 246, 381–386. (35) Kretschmann, E.; Raether, H. Z.; Naturforsch, A. Astrophys. Phys. Chem. 1968, 23, 2135–2136. (36) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559–3568.

Langmuir 2009, 25(21), 12644–12652

Fabre and Talham

Article

Figure 3. Monitoring fusion of POPC vesicles with different concentrations of DPPA to form symmetric supported lipid bilayers on the zirconium phosphonate surface. The lipid layer thickness was calculated from conversion of Ψ values (obtained by SPREE) to thickness values. The experiments were conducted at 24 C, and the tris buffer [10 mM trizma hydrochloride and 100 mM sodium chloride (pH 7.4)] was used to rinse the nonfused lipid vesicles.

Figure 2. Schematic of the SPREE experimental setup showing the seven layers that correspond to the optical parameters in Table 1. Table 1. Optical Parameters for the Multilayer Model material medium layer 1 layer 2 layer 3 layer 4 layer 5 medium

SF10 Cr Au ODM Zr-ODPA lipid bilayer buffer

refractive index 1.73 3.05 0.414 1.47 1.49 1.45 1.33

extinction coefficient

thickness (nm)

0 3.33 2.27 0 0 0 0

¥ 4.10 28.43 2.68 2.10 unknown ¥

literature for the ODM and zirconium octadecylphosphonate layers.28,36 The experimental data were fit using AnalysR to the seven-layer model based on the Fresnel equations to yield the bilayer thicknesses. Vesicles of POPC or POPG containing variable amounts of DPPA were used, and they were chosen because the terminal divalent phosphate is expected to covalently bind to the zirconium phosphonate network. An objective was to determine appropriate percentages of DPPA that lend stability to the bilayers but are also sufficiently low that DPPA does not dominate the bilayer characteristics. Because the concentration of DPPA was kept below 50%, the fluidity of the vesicles allowed vesicle fusion. Lipid vesicles with a total lipid concentration of 0.5 mg/mL in tris buffer were introduced into the flow cell, and the ellipsometric Ψ value was recorded over time. SPREE sensorgrams for three different concentrations of DPPA mixed with POPC are shown in Figure 3. The kinetic curves are characterized by a baseline, absorption, and lipid desorption domains. There is an increase in Ψ corresponding to the binding and fusion of vesicles. After the signal stabilized, we assumed complete formation of the supported lipid bilayers and running buffer was passed through the cell, causing a small decrease in Ψ corresponding to desorption of free vesicles in most experiments. Using the Fresnel equations, the seven-layer model was fit to the data to calculate the overall thickness of the lipid bilayers, reported in Table 2. The SPREEdetermined bilayer thicknesses range from 4.23 to 5.18 nm for all compositions studied and are within the range normally reported for supported lipid bilayers.3 Langmuir 2009, 25(21), 12644–12652

Table 2. Thicknesses and Adsorption Times of Symmetric Lipid Bilayers on Zirconium Phosphonate-Modified Gold Slides lipid

d (nm)

adsorption time (min)

POPC POPC/DPPA (10%) POPC/DPPA (50%)

5.18 ( 0.04 4.64 ( 0.03 4.23 ( 0.03

12 15 28

The time constant for vesicle adsorption increases as the concentration of DPPA increases. This result is explained by the higher gel to liquid crystalline transition temperature of saturated DPPA, which increases the rigidity of the lipid assembly, leading to slower kinetics.37 3.2. Bilayer Stability Studied by SPREE. The stability of supported lipid bilayers is an important concern. Confidence in the integrity of the model membranes following various stresses greatly increases their utility. In this system, we thought that occasional covalent linkages between DPPA and the zirconium phosphonate surface would greatly enhance the stability of the lipid bilayers. To test this idea, bilayers containing 10% DPPA were chosen, as this composition demonstrated good miscibility with other lipids at room temperature and was shown in Figure 3 to lead to fast formation of symmetric lipid bilayers. This ratio also complements the composition of mammalian cell plasma membranes, which often consist of 10-20% anionic lipids. Following vesicle fusion and rinsing with buffer, the effects of solvents and dehydration on bilayer stability were explored. Figure 4 shows the influence of ethanol and sodium dodecyl sulfate (SDS), an ionic detergent used for the permeabilization and solubilization of biological and model membranes. A low concentration of ethanol (5%) has no effect on the POPC/DPPA supported lipid bilayer as no membrane is removed from the surface. The same results were obtained for POPC/POPG supported lipid bilayers (data not shown). On the other hand, SDS does remove some of the bilayers. A low concentration of SDS (0.5%) removed 32% of the POPC/DPPA membrane, and 2% SDS removed 58% of the same membrane. Normally, complete removal of the bilayers is expected. The fact that significant portions remain is evidence for a strong interaction with the zirconium phosphonate surface. The stability of the lipid bilayers to dehydration was also investigated within the SPREE flow cell. The 10% DPPA in POPC bilayer was compared to systems composed of only POPC or POPG lipids. Following vesicle fusion and rinsing with buffer, the bilayers were exposed to air for 200 min and rehydrated. (37) Puu, G.; Gustafson, I. Biochim. Biophys. Acta 1997, 1327, 149–161.

DOI: 10.1021/la901920y

12647

Article

Fabre and Talham Table 3. XPS Analysis of the Zirconium Phosphonate and Lipid Monolayer Films calculated relative intensitya (%)

film type

Zr 3d

P 2p3

surface coverage (mol/cm2)

zirconium phosphonate 52 48 4.0  1014 DPPA monolayer on 38 62 2.7  1014 zirconium phosphonate a Relative intensities are determined from experimental peak areas normalized with atomic and instrument sensitivity factors.

Figure 4. SPREE analysis following the effect of different solvent conditions on the stability of POPC supported lipid bilayers containing 10% DPPA. The SPREE measurement used an angle of incidence of 65.1 and was conducted at 24 C. The rinsing tris buffer contained 5% ethanol or SDS, as indicated. Curves for the SDS-containing buffer are shifted by 0.4.

Figure 6. XPS multiplex spectra of the zirconium phosphonate layer before (top) and after (bottom) a DPPA monolayer is deposited. The P:Zr intensity ratio increases with the addition of the lipid layer.

Figure 5. Effect of dehydration on (a) POPC supported lipid bilayers, (b) POPG supported lipid bilayers, and (c) POPC/DPPA supported lipid bilayers followed with SPREE. The angle of incidence was kept at 65.1, and the temperature was 24 C. To rehydrate the surface, trizma hydrochloride buffer was used. The value of Ψ before dehydration is only recovered for the DPPAcontaining bilayer. Plots showing the full range of Ψ are included as Supporting Information.

The SPREE responses for each of the three systems are reported in Figure 5, in which dehydration is seen as a reduction in Ψ of 12648 DOI: 10.1021/la901920y

∼7 (Supporting Information). Upon rehydration, the SPREE signal is recovered only for the DPPA-containing bilayer. According to Figure 5, 81% of the POPC membrane is removed from the surface and 96% of the POPG membrane is removed. In contrast, the bilayer containing 10% DPPA appears to be stable to dehydration, recovering >95% of the membrane. 3.3. Asymmetric Supported Lipid Bilayers. 3.3.1. Inner Lipid Monolayer. Monolayers of DPPA or a mixture of POPC and DPPA are easily transferred to the zirconium phosphonate surface using standard Langmuir-Blodgett deposition methods to form the inner layer of an asymmetric bilayer assembly. The complete process was characterized by XPS and ellipsometry for a DPPA monolayer. The gold XPS signal (Supporting Information) appears to be larger in the starting zirconium phosphonate layer and is attenuated after the lipid layer is added, showing clearly the presence of the lipid monolayer. The coverage of the lipid molecules can be estimated using the ratio of P to Zr (Table 3), based on the known Zr4þ and ODPA coverage in the zirconium phosphonate modifying layer.28 The zirconium phosphonate layer on gold shows a P to Zr ratio of 0.95-1 (Figure 6) which is consistent with the expected 1:1 stoichiometry after considering the attenuation length of the photoelectrons and the depth of each species present in the film.38 After the lipid monolayer had been transferred onto the zirconium phosphonate film, the P:Zr ratio is 1.6:1. This compares well with the expected stoichiometry of a DPPA monolayer on the zirconium phosphonate surface given the cross-sectional area per molecule of ODPA is 24 A˚2 27 and the cross-sectional area of DPPA is 40 A˚2 per molecule. (38) Pike, J. K.; Byrd, H.; Morrone, A. A.; Talham, D. R. Chem. Mater. 1994, 6, 1757–1765.

Langmuir 2009, 25(21), 12644–12652

Fabre and Talham

Article Table 4. FRAP Parameters of POPC Doped with 2% NBD-PC

type of support glass Zr-ODPA film DPPA monolayer on Zr-ODPA film

Figure 7. SPREE analysis of the formation of the outer monolayer by vesicle fusion onto a POPC/DPPA monolayer. The angle of incidence was kept at 65.1 and the temperature at 24 C. The curve for the POPC/POPG monolayer is shifted by 0.1.

Figure 8. Fluorescent images of vesicle fusion to form asymmetric bilayers with a DPPA inner layer. (a) Incomplete vesicle fusion. (b) Homogeneous fluorescence following complete vesicle fusion and rinsing.

The LB deposited lipid monolayer assembly was also characterized with ellipsometry in air. The thicknesses of the ODM, zirconium phosphonate, and DPPA layers were measured independently by ellipsometry, applying a seven-layer model similar to that described in Figure 2. The resulting thickness of the DPPA lipid monolayer, 2.42 nm, is consistent with the thickness of a lipid monolayer.39 3.3.2. Fusion of Lipid Vesicles on the Inner Monolayer Studied by SPREE. Asymmetric lipid bilayers were formed by fusing lipid vesicles onto the hydrophobic surfaces generated after LB deposition of the inner DPPA or POPC/DPPA monolayers. Vesicle solutions formed from POPC or POPC and POPG (70:30) were introduced via the flow cell and the formation of the outer monolayer was followed by SPREE (Figure 7). The kinetics for the two lipid systems are similar. Fitting the data using the model in Figure 2, we find the thickness of the outer POPC layer is 2.63 ( 0.01 nm and that of the outer mixed POPC/POPG layer is 2.66 ( 0.01 nm, consistent with the deposition of a monolayer via vesicle fusion. Fluorescence imaging also supports effective vesicle fusion. A DPPA monolayer on a zirconium phosphonate-modified glass slide was placed in the fluorescence flow cell, and the bilayer was completed by the fusion of POPC vesicles doped with 2% fluorescent lipid NBD-PC. Fluorescence images of free vesicles and of the outer lipid monolayer are shown in Figure 8. The presence of nonfused vesicles is clearly discerned from a uniform layer following vesicle fusion.40 (39) Chunbo, Y.; Desheng, D.; Zuhong, L.; Juzheng, L. Colloid Surf., A 1999, 150, 1–6. (40) Shahal, T.; Melzak, K. A.; Lowe, C. R.; Gizeli, E. Langmuir 2008, 24, 11268–11275.

Langmuir 2009, 25(21), 12644–12652

fusion of POPC vesicles

diffusion coefficient (μm2/s)

mobile fraction (%)

bilayer bilayer monolayer

3.98 ( 0.11 0.72 ( 0.04 2.86 ( 0.16

77.5 ( 0.8 34.4 ( 0.7 76.0 ( 0.9

3.4. FRAP Analysis of Supported Lipid Bilayers. Lateral diffusion FRAP analysis was performed to determine the fluidity of the model membranes. FRAP experiments were performed on zirconium phosphonate-modified glass slides using a confocal microscope with lipid vesicles doped with a small amount of fluorescent lipid, and the diffusion coefficient and mobile fraction of the lipids were determined. The lateral fluidity of POPC lipid bilayers on zirconium phosphonate was compared to that of the same lipid system on a glass surface. For POPC doped with 2% NBD-PC, the average diffusion coefficient for bilayers on glass was determined to be 3.98 ( 0.11 μm2/s with a mobile fraction of 77.5% (Table 4). This value agrees well with previous results reported for lipid bilayers on glass.41 On the other hand, for the same bilayer on the zirconium phosphonate film, the calculated diffusion coefficients were 5-fold lower (0.72 ( 0.04 μm2/s). Moreover, the mobile fraction was also very low, ∼35%. These results indicate that the lipid bilayers have low fluidity on the zirconium phosphonate surface. The strong interaction with the zirconium phosphonate surface will be through the proximal layer, so to study the behavior of the distal layers, we investigated the fluidity of the asymmetric bilayers by including the fluorescent probes in only the outer layer, assuming minimal exchange of lipids between layers.7 The degree of lateral diffusion of the outer layer is significantly higher (2.86 ( 0.16 μm2/s), well within the range associated with fluid lipid bilayers.42 The mobile fraction was also high (76.0%), comparable to other of lipid bilayers with high fluidity. 3.5. Protein Binding to Supported Lipid Bilayers. Insertion and folding of proteins are among many cellular processes involving cell membranes. The ability to study such processes using accurate model systems is an important objective of artificial supported lipid bilayers. Therefore, the ability of the zirconium phosphonate supported lipid bilayers to bind a membrane protein was investigated. Melittin, a major component in honey bee venom, was selected as it is a well-studied small protein and melittin-lipid membrane interactions have already been explored.43-49 Depending on the composition of the lipid membrane, melittin acts in two different ways. It forms transmembrane pores in a zwitterionic phospholipid membrane via the barrel-stave mechanism and acts in a detergent-like manner on negatively charged membranes.45 The membrane lipid composition has a strong influence on the melittin-lipid interaction, so we varied the zwitterionic and anionic lipid composition in (41) Wagner, M. L.; Tamm, L. K. Biophys. J. 2000, 79, 1400–1414. (42) Kalb, E.; Frey, S.; Tamm, L. K. Biochim. Biophys. Acta 1992, 1103, 307– 316. (43) Mozsolits, H.; Wirth, H. J.; Werkmeister, J.; Aguilar, M. I. Biochim. Biophys. Acta 2001, 1512, 64–76. (44) Beschiaschvili, G.; Seelig, J. Biochemistry 1990, 29, 52–58. (45) Papo, N.; Shai, Y. Biochemistry 2003, 42, 458–466. (46) Kleinschmidt, J. H.; Mahaney, J. E.; Thomas, D. D.; Marsh, D. Biophys. J. 1997, 72, 767–778. (47) Chen, X. Y.; Wang, J.; Boughton, A. P.; Kristalyn, C. B.; Chen, Z. J. Am. Chem. Soc. 2007, 129, 1420–1427. (48) Constantinescu, I.; Lafleur, M. Biochim. Biophys. Acta 2004, 1667, 26–37. (49) Wessman, P.; Stromstedt, A. A.; Malmsten, M.; Edwards, K. Biophys. J. 2008, 95, 4324–4336.

DOI: 10.1021/la901920y

12649

Article

Fabre and Talham

Figure 9. Sensorgrams of the binding of melittin to asymmetric lipid bilayers with zwitterionic (POPC), anionic (POPG), and mixed lipid (POPC/POPG) outer layers. The peptide concentration was 10 μM. The data were fit using the two-state reaction model. The angle of incidence was kept at 65.1 and the temperature at 24 C. POPC/POPG and POPG curves are shifted by 0.025 and 0.05, respectively.

asymmetric supported bilayers and monitored the binding of melittin to the lipid membrane. To analyze the binding kinetics, we analyzed the sensorgrams by curve fitting using numerical integration analysis.50 The mechanism of binding of melittin to supported lipid membranes can be represented by a two-step reaction model.43 The first step is the initial binding of the protein to the membrane surface, and the second is the insertion of melittin into the lipid membrane. The two-state reaction model can be represented as ka1

ka2

kd1

kd2

srs PL f srs PL PþL f

ð4Þ

where in the first step, the protein (P) binds to the lipids (L) to give PL, and the complex PL is then changed to PL* in the second step, which cannot dissociate directly to P and L. The corresponding differential rate equations for the two-state reaction model are represented by dΨ1 ¼ ka1 ½PðΨmax -Ψ1 -Ψ2 Þ -kd1 Ψ1 -ka2 Ψ1 þkd2 Ψ2 ð5Þ dt dΨ2 ¼ ka2 Ψ1 -kd2 Ψ2 dt

ð6Þ

where [P] is the protein concentration, Ψ1 and Ψ2 are the response units for the first and second steps, respectively, and Ψmax is the equilibrium binding response. The association and dissociation rates for the first and second steps are ka1, kd1, ka2, and kd2. The association equilibrium or affinity constants for the first and second steps are K1 and K2, respectively, and equal ka/kd. The total affinity constant KA (M-1) is the product K1K2. Asymmetric bilayers were formed of POPC, POPG, and POPC/POPG (70:30, w/w) outer layers with a POPC/10% DPPA LB inner layer. The asymmetric supported lipid bilayers were then exposed to 10 μM melittin, and the peptide-lipid interaction was followed by SPREE in real time. Typical sensorgrams of the binding of melittin to the lipid membrane are shown in Figure 9. The adsorption and desorption steps of the protein binding on the membrane are easily defined, and using the two-step kinetic (50) Morton, T. A.; Myszka, D. G.; Chaiken, I. M. Anal. Biochem. 1995, 227, 176–185.

12650 DOI: 10.1021/la901920y

model, the kinetic and equilibrium parameters were determined. The kinetic analysis was performed at least three times for each lipid composition, and the averaged values for the rate constants and affinity constants are listed in Table 5. The affinity constant for melittin binding to POPG is almost 5-fold higher than the one measured for zwitterionic lipids, Also, the association rate constant for the first step is 4-fold higher for the anionic POPG and 2-fold higher for the mixed membrane than for the zwitterionic POPC layer. This result indicates that electrostatic interactions control the binding affinity of melittin for anionic lipids, consistent with what has been seen before, as melittin in a cationic form is attracted to anionic lipids via direct and fast electrostatic interactions. For the POPG layer, the association rates for the second step are slower, consistent with the idea that the binding between melittin and anionic lipid membrane is controlled by the first step. The interaction between melittin and zwitterionic lipids is normally thought to be driven predominantly by hydrophobic interactions. However, for the zwitterionic (POPC) and mixed lipid (POPC/POPG) layers prepared here, the affinity constants are not very different, suggesting that the previously observed insertion into the membrane is incomplete. This result is similar to work previously observed for lipid monolayer models.43

4. Discussion The goal of the study was to demonstrate that zirconium phosphonate-modified surfaces are practical substrates for supported lipid bilayers. Zirconium phosphonates have previously been shown to selectively bind phosphate-terminated oligonucleotides,15,17 phosphorylated peptides,18 and phospholipids,19 in applications that include bioarrays15,17 and protein pull-down. 18,51 An important biomaterials objective is to develop surfaces and interfaces that can be used for a variety of applications without the need for function-specific pretreatment. We move closer to that objective by extending the utility of the zirconium phosphonate surface to supported lipid bilayers. Vesicle fusion is demonstrated to indeed produce supported bilayers on the zirconium phosphonate surface. As Figure 3 demonstrates, bilayer formation takes place on the same time scale as on other hydrophilic surfaces.5,52,53 The rate of bilayer formation varies with lipid composition, but this observation is attributed to the different stiffness of the vesicles formed from different lipids. Increasing the percentage of the saturated DPPA slows vesicle fusion on the surface. The zirconium phosphonate-modified surface is terminated in oxide and hydroxide groups and therefore provides a hydrophilic surface that interacts with the polar lipid headgroups. At the same time, the Zr4þ ions are known to form strong specific covalent bonds to dibasic phosphate and phosphonate groups, which can displace the oxide or hydroxide groups at the interface. However, less basic groups leave the zirconium ion coordination intact and are restricted to electrostatic and hydrogen binding interactions with the oxide or hydroxide groups. For example, we have previously shown that phosphate-terminated oligonucleotides bind preferentially relative to modified DNA segments containing only phosphodiester groups.16 We therefore decided to incorporate a phosphatidic acid into the lipid assemblies to determine if the stronger covalent interactions expected between the dibasic (51) Zhou, H. J.; Xu, S. Y.; Ye, M. L.; Feng, S.; Pan, C.; Jiang, X. G.; Li, X.; Han, G. H.; Fu, Y.; Zou, H. J. Proteome Res. 2006, 5, 2431–2437. (52) Reimhult, E.; Hook, F.; Kasemo, B. Langmuir 2003, 19, 1681–1691. (53) Zhdanov, V. P.; Keller, C. A.; Glasmastar, K.; Kasemo, B. J. Chem. Phys. 2000, 112, 900–909.

Langmuir 2009, 25(21), 12644–12652

Fabre and Talham

Article

Table 5. Equilibrium and Kinetic Parameters for Adsorption of Melittin to Asymmetric Bilayers Using the Two-State Reaction Model Fit to the SPREE Dataa outer lipid layerb

ka1 (104 M-1 s-1)

kd1 (s-1)

K1 (104 M)

ka2 (s-1)

kd2 (s-1)

K2

POPC 1.07 0.222 4.82 0.0247 0.00503 4.91 POPC/POPG 1.99 0.158 12.6 0.0139 0.00604 2.30 POPG 4.34 0.077 56.4 0.0112 0.00575 1.95 a Data were averaged over three experiments. b All lipid compositions are adsorbed onto a POPC/DPPA (10%) lipid monolayer.

phosphate of the lipid headgroup and the zirconium ions could lead to enhanced stability of the lipid bilayers. As with other hydrophilic surfaces, supported bilayers of POPC or POPG on the zirconium phosphonate surface are unstable to dehydration (Figure 5). On the other hand, including just 10% of DPPA results in supported bilayers that are stable to dehydration (Figure 5). Evidence of the influence of DPPA is seen upon rinsing with the surfactant SDS, which normally removes most lipid assemblies. In the case of the DPPA-containing bilayers, lipid removal is incomplete (Figure 4), suggesting a strong interaction with the surface. However, after SDS rinse, the residual lipid is far more than the few percent DPPA used to form the bilayer; just more than 50% of the bilayer is removed with 2% SDS. By fluorescently labeling the inner layer of an asymmetric bilayer, we see that SDS rinsing leaves the bottom layer attached to the surface (Supporting Information). Even though only 10% DPPA is included in the layer, the whole layer remains, implying that covalent linkage of only a few molecules is enough to stabilize a larger cooperative lipid assembly on the surface. The zirconium phosphonate strategy complements other recent examples of air-stable bilayer systems. Jeon et al. formed a membrane with incorporated proteins that is completely enclosed in a hydrogel mesh.12 Deng and co-workers created supported lipid bilayers supported by cholesterol groups.13 A fluid- and airstable lipopolymer membrane11 was reported by Cremer and coworkers that mechanically strengthens membranes and increases its longevity 3-fold compared to regular lipid bilayers. Cheng et al.10 studied the stability of various supported membranes (PC, DOPC, PE, etc.) on a nanoglassified substrate. A consequence of the stabilizing lipid-zirconium phosphonate interaction is the reduced mobility of the lipids. The FRAPdetermined lipid mobility of the bilayers is significantly smaller than those reported for other supported bilayer models.41 Even for the symmetric POPC bilayer, which does not include the covalent linkages to the support expected with phosphatidic acid, mobility is low. However, the reduced mobility is principally in the inner layer, and the outer leaflet is only minimally impacted by the surface-lipid interaction of the inner layer. By using Langmuir-Blodgett deposition of the inner layer followed by fusion of the vesicle to deposit just the outer layer, the stabilizing effect of the covalent interaction between DPPA and the zirconium surface is retained while the composition of the outer layer is allowed to vary. Indeed, the lipid mobility and mobile fraction of the outer layer in the asymmetric bilayers are comparable to those of other viable model systems.41,54 The viability of using the zirconium phosphonate supported bilayers as membrane mimics was demonstrated using the membrane binding peptide, melittin. With the peptide composed of 26 amino acids, residues 1-20 are predominantly hydrophobic and 20-26 are hydrophilic.55 The latter region has four positive charges that bind to anionic lipids. The association of melittin (54) Phillips, K. S.; Dong, Y.; Carter, D.; Cheng, Q. Anal. Chem. 2005, 77, 2960– 2965. (55) Haberman, E.; Jentsch, J. Hoppe-Seyler’s Z. Physiol. Chem. 1967, 348, 37.

Langmuir 2009, 25(21), 12644–12652

KA (105 M-1) 2.37 2.90 11.0

with phospholipid bilayers has been shown to involve a two-step binding mechanism. The initial binding step, K1, is followed by insertion into the bilayer to form hydrophobic interactions. Electrostatic interactions dominate the binding of melittin to negatively charged membranes via K1, whereas the protein forms subsequent hydrophobic interactions with zwitterionic membranes.56 To perform the membrane binding analysis, asymmetric supported lipid bilayers were formed on the zirconium phosphonate surface with zwitterionic (POPC), anionic (POPG), and mixed (POPC/POPG) lipid outer layers. Previous studies have shown that melittin binding to supported bilayers is poorly described by the Langmuir binding model and that a two-state reaction model is more appropriate. As observed with previous bilayer membrane models, melittin associates faster with a significantly larger K1 with the anionic POPG. Also in line with other model systems, the second (insertion) step is slower than the initial binding step for all membranes. At the same time, the insertion process, K2, is more important for the zwitterionic POPC layer than for the POPG-containing layers, in agreement with previous studies. The melittin binding studies yield results similar to those obtained with other supported bilayer model membrane systems. A feature of the zirconium phosphonate interface is that the chemistry can be transferred to any commonly used analytical substrate. It is now routine to make common surfaces hydrophobic, and the zirconium octadecylphosphonate modifying layer can be deposited onto any hydrophobic surface. Therefore, the same biomolecule attachment chemistry can be used for analytical studies having different substrate requirements, including gold, silicon, glass, and plastic. This transferability lends confidence when comparing results from different experiments that the system has not been perturbed by the different chemical transformations used to prepare different surfaces. A further convenience is that attachment to the zirconium phosphonate surface occurs via the phosphate group, making it unnecessary to synthetically modify the biomolecules to be immobilized. Phosphopeptides, phospholipids, and oligonucleotides as well as many proteins and DNA all naturally contain dibasic phosphate.

5. Conclusions Zirconium phosphonate-modified surfaces are shown to be a versatile platform for supported lipid bilayers. Inclusion of a small percentage of phosphatidic acid into the inner lipid layer to form covalent interactions with the zirconated surface significantly enhances the stability of the bilayer assemblies, retaining their structure following drying in air. As a consequence of the strong surface-lipid interactions, the inner layer shows low fluidity; however, the outer layer proves to be highly fluid, comparable to other supported bilayer systems. The binding of melittin to bilayers containing anionic and zwitterionic lipids demonstrates that the supported lipid bilayers constitute a viable membrane mimic. Attractive features of the zirconium phosphonate-modified surfaces (56) Sharon, M.; Oren, Z.; Shai, Y.; Anglister, J. Biochemistry 1999, 38, 15305– 15316.

DOI: 10.1021/la901920y

12651

Article

include the ability to form symmetric or asymmetric bilayers on multiple substrates with composition control without requiring chemical modifications of the constituent lipids and the compatibility with other phosphorylated biomolecules. Further experiments are underway to improve the fluidity of the inner layer and to enable the study of transmembrane proteins. Acknowledgment. Support from the U.S. National Science Foundation through Grant CHE-0514437 (D.R.T.) is acknowledged. We thank Prof. Weihong Tan for use of the confocal

12652 DOI: 10.1021/la901920y

Fabre and Talham

microscope and Prof. Gail Fanucci for providing the protein and for helpful discussions. Supporting Information Available: XPS survey scans of the zirconium phosphonate surfaces and of a DPPA monolayer on the zirconium phosphonate surface, SPREE experiments showing the effect of dehydration on supported lipid bilayers, FRAP recovery curves of POPC lipid monolayers or bilayers on different supports, and FRAP experiments with the lipid inner layer under different wash conditions. This material is available free of charge via the Internet at http://pubs.acs.org.

Langmuir 2009, 25(21), 12644–12652