Surface Response Methodology for the Study of Supported Membrane

lipid bilayer platforms offering high lateral mobility to incorporated membrane proteins. Hajra Basit , Vinnie Gaul , Sean Maher , Robert J. Forst...
0 downloads 0 Views 569KB Size
Download PDF
J. Phys. Chem. B 2007, 111, 7567-7576

7567

Surface Response Methodology for the Study of Supported Membrane Formation Claire Rossi,†,| Elisabeth Briand,‡ Pierre Parot,§ Michael Odorico,§ and Joe1 l Chopineau*,⊥ UMR 6022 CNRS UniVersite´ de Technologie de Compie` gne, BP 20529, 60205 Compie` gne, France, Laboratoire de Re´ actiVite´ de Surface, UMR CNRS 7609, UniVersite´ Pierre et Marie Curie, 4 Place Jussieu, 75005 Paris, France, CEA Valrho, DSV-iBEB-SBTN, BP 17171, Bagnols sur Ce` ze 30207, France, Centre de Biochimie Structurale, CNRS UMR 5048 INSERM 554 and UniVersite´ de Nıˆmes, Place Gabriel Pe´ ri, 30000 Nıˆmes, France ReceiVed: December 18, 2006; In Final Form: May 2, 2007

We report on the investigations of the formation of the tethered lipid bilayer by vesicle deposition on aminefunctionalized surfaces. The tethered bilayer was created by the deposition of egg-PC vesicles containing 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-poly-(ethyleneglycol)-N-hydroxysuccinimide as anchoring molecules on an amine-coated surface. This approach is an easy route for the formation of a biomimeticsupported membrane. A Doelhert experimental design was applied to determine the conditions leading to the formation of a continuous and defect-free tethered bilayer on different surfaces (gold and glass). Doehlert designs allow modeling of the experimental responses by second-order polynomial equations as a function of experimental factors. Four factors expected to influence bilayer formation were studied: the lipid concentration in the vesicle suspension, the mass percentage of anchoring molecules in the vesicles, the contact time between the vesicles and the surface, and the resting time of the membrane after buffer rinse. The optimization of the membrane preparation parameters was achieved by monitoring lipid assembly formation using surface plasmon resonance spectroscopy on gold and by fluorescence recovery after photobleaching on glass. Three characteristic responses were systematically measured: the bilayer thickness, the lipid diffusion coefficient, and the lipid mobile fraction. The simultaneous inspection of the three characteristics revealed that a restricted experimental domain leads to properties that are in accordance with a bilayer presence. The factors of this domain are a lipid concentration from 0.1 to 1 mg/mL, 4-8% of anchoring molecules in the vesicles, 1-4 h of contact time between vesicles and surface, and 21-24 h of resting time after buffer rinse. Under these conditions, a membrane having a lipid mass per surface between 545 ( 5 and 590 ( 10 ng/cm2, a diffusion coefficient of between 2.5 ( 0.3 × 10-8 and 3.60 ( 0.5 × 10-8 cm2/s, and a mobile fraction between 94 ( 2 and 99 ( 1% was formed. These findings were confirmed by atomic force microscopy observations, which showed the presence of a continuous and homogeneous bilayer in the determined experimental domain. This formation procedure presents many advantages; it provides an easily obtainable biomimetic membrane model for proteins studies and offers a versatile tethered bilayer because it can be adapted easily to various types of supports.

Introduction Since their description,1,2 solid-supported membranes have been used extensively as a model system for cellular membranes. Such artificial structures are interesting for the study of membrane events at the nanoscale. They are stable, and they allow the use of analytical methods that cannot be performed on liposomes or black lipid membrane, such as surface analytical techniques.3,4 Different constructions of supported models have been proposed, such as supported lipid bilayers on hydrophilic surfaces,1,5 a hybrid bilayer,6,7 tethered bilayers3,8 and supported vesicle layers.9,10 Tethered bilayers have been introduced and further modified to overcome problems that occur upon the insertion of integral membrane proteins.11-16 The aqueous reservoir space between the substrate and the bilayer has been * Corresponding author. Phone: 33 466 27 95 63. Fax: 33 466 27 95 54. E-mail: [email protected]. † Universite ´ de Technologie de Compie`gne. ‡ Universite ´ Pierre et Marie Curie. § CEA Valrho. ⊥ CNRS UMR 5048 INSERM 554 and Universite ´ de Nıˆmes. | Present address: Max-Planck Institut fu ¨ r Polymerforschung, Ackermannweg 10, 55128 Mainz, Germany.

improved by the introduction of a spacer between the bilayer and the support to reduce friction and the risk of inactivation for the extra membrane parts of proteins.8,15,17,18 We recently developed a one-step construction of a tethered bilayer model based on egg-PC/DSPE-PEG3400-NHS mixed vesicle deposition onto an amine-grafted surface.15 Introduction of activated lipopolymers into vesicles allows the vesicle/bilayer to anchor on the amine surface via amide bonds. Anchoring and subsequent rupture of the vesicles lead to the formation of a tethered bilayer structure. The presence of polyethylene glycol (PEG) spacer chains impairs contact of the bilayer with the surface and aids the presence of an aqueous reservoir between the substrate and the bilayer. The well-defined geometry of this lipid assembly is convenient to apply surface-sensitive techniques, such as surface plasmon resonance (SPR),19 quartz crystal microbalance with dissipation,20 atomic force microscopy (AFM),21 impedance spectroscopy6,22 or fluorescence recovery after photobleaching (FRAP).2,17 The procedure of the membrane formation was developed on gold and glass surfaces to demonstrate the versatility of this formation procedure for different amine-coated substrates. Both supports were chosen because of the possibility to perform SPR on gold, whereas

10.1021/jp0686792 CCC: $37.00 © 2007 American Chemical Society Published on Web 06/13/2007

7568 J. Phys. Chem. B, Vol. 111, No. 26, 2007

Rossi et al.

Figure 1. Illustration of different lipid assemblies that could be obtained after the deposition of egg-PC liposomes containing DSPE-PEG3400NHS onto amine-coated surfaces (glass or gold). Different lipids organizations could be (1) a tethered bilayer; (2) discontinuous bilayers, aggregates of lipids, or both; (3) multilayers; or (4) vesicles adsorbed onto the amine-coated surface.

optically transparent glass can be used for microscopic observations. Amine-coated gold substrates were appropriate for performing SPR experiments, and optically transparent gold or glass were used for microscopic observations. Systematic studies have been realized for lipid bilayer formation on hydrophilic substrates,4,5,20,23 but were never performed on the construction system we have developed. As schematically illustrated in Figure 1, after vesicle deposition, a homogeneous and fluid lipid bilayer is expected to be formed, but in addition, structures such as vesicles attached to the surface and other intermediate structures could be obtained. Successful construction of a membrane structure on the surface after injection of mixed lipids vesicles results from the complex interaction between multiple chemical and physical parameters such as chemical reactions, interactions, and lipid rearrangements. Such a multiparametric process has to be optimized with the help of an experimental approach, which allows the

screening of the formation conditions. In this work, information on the assembly process was obtained by the second-order experimental design proposed by Doehlert that appears pertinent in the context of this work.24 In this design, the characteristics of the final lipid assembly can be modeled by quadratic equations as a function of the experimental factors and their interactions that intervene in the formation process. In an experimental design, the simultaneous variation of all studied experimental factors diminishes the number of experiments and determines the interacting parameters in contrast to a classical screening approach. The Doehlert design offers a very good efficiency when comparing the number of factors to be studied with the number of needed experiments.25,26 In the case of four factors, this design requests only 21 experiments instead of 256 experiments that are necessary for a classical design (for four levels of values). Another significant advantage of the Doelhert design is the easy

Surface Response Methodology expansion of the experimental domain. Experiments can be reused when the optimal response is outside or at the boundaries of the experimental domain. This particular property allows a better vision of the response tendency around the optimal point. For these reasons, the Doelhert model is a widespread chemometric tool used in many fields, such as analytical chemistry, chemical engineering, and pharmacy, to help with parameter optimization.25,26 The dependence of the vesicle’s binding and membrane formation on four factors was studied.15 They are the lipid concentration in the vesicle suspension, the mass percentage of DSPE-PEG-NHS in the vesicles, the contact time between the vesicles and the surface, and the resting time of the lipid assembly after buffer rinse. The final lipid assembly was characterized by three responses: the optical thickness (converted in lipid mass per surface) measured by SPR experiments, the lateral lipid diffusion coefficient, and the mobile fraction of lipids determined by FRAP experiments. Each response was described by a quadratic equation as a function of the four factors that was adequate to predict responses in all experimental regions. This modelization allows identifying the reasonable parameters that play a major role in the process of bilayer formation and the determination of the best conditions. The homogeneity at the nanoscale of the membrane formed using the optimized set of experimental parameters was verified by AFM, and its anchoring to the amine surface was ascertained using polarization modulation infrared reflection-adsorption spectroscopy. Experimental Section Materials. L-R-Phosphatidylcholine from egg yolk (egg-PC) type XVI-E and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-NBD (7-nitro-2,1,3-benzoxadiazol-4-yl) (DPPENBD, used as fluorescent probe), were purchased from SigmaAldrich (St. Quentin-Fallavier, France). 1,2-Distearoyl-snglycero-3-phosphoethanolamine-poly(ethylene glycol)-Nhydroxysuccinimide(DSPE-PEG3400-NHS)wasfromShearwater Polymers (Huntsville, AL). 2-Mercaptoethylamine (cysteamine hydrochloride, g99%) and Triton X-100 were from SigmaAldrich. Aminopropyldimethylethoxysilane (APDMES, 99%) was from ABCR (Karlsruhe, Germany). Glass microscope slides were from Menzel-Glaser (Braunschweig, Germany). Silicon wafers were purchased from Crystec Technology Trading GmbH (Alto¨tting, Germany). The two-component epoxy glue (EPOTEK 353ND-4) was purchased from Polytec-PI (Pantin, France), and the SuperGlue-3 Loctite was from Henkel Loctite France (Senlis, France). All other chemicals were of analytical grade. All buffers prepared from Milli-Q water (resistivity higher than 18.2 MΩ) were filtered and thoroughly degassed. N-(2Hydroxyethyl)-piperazine-N′-2-ethanesulfonic acid (HEPES) 20 mM, pH 7.8, 150 mM NaCl was the standard buffer. Formation of Self-Assembled Monolayers (SAMs). Selfassembly was performed on different surfaces, which have to be cleaned and prepared properly. Glass microscope slides were prepared as previously described.15 Template stripped gold (TSG) on silicon wafers was obtained according to a published procedure.27,28 Briefly, a gold layer of 47 ( 1 nm was deposited on the wafer by thermal evaporation under vacuum (evaporator, Edwards, model Auto 306, rate 0.01 nm s-1, pressure 2 × 10-6 mbar). For SPR and AFM analysis, the gold surfaces were glued to glass microscope slides with EPO-TEK 353ND-4 (n ) 1.5922) and cured for 60 min at 150 °C. For FRAP observations on gold, the surfaces were glued to glass microscope slides with SuperGlue-3 Loctite. Under analysis conditions, fluorescence

J. Phys. Chem. B, Vol. 111, No. 26, 2007 7569 emission of the SuperGlue-3 is negligible, contrary to the fluorescence of the EPO-TEK glue. The slides were detached from the silicon to expose the TSG ultraflat film and directly immersed overnight in a degassed solution of 2-mercaptoethylamine (5 mM in pure ethanol or in HEPES buffer). The goldamine-coated slides were thoroughly rinsed with ethanol and dried under a nitrogen stream. As glass surfaces, glass microscope cover slips (previously stored in Milli-Q water after cleaning) were dried at 110 °C for 20 min. After cooling, the glass slides were immersed in an APDMES solution (2%, v/v) in toluene for 4 h and thoroughly rinsed with toluene, chloroform, ethanol, and water before being dried in an oven at 110 °C. All functionalized surfaces were used immediately after drying. Preparation of Vesicles and Tethered Membrane Formation. Vesicles were obtained as described previously.15 The hydrodynamic mean diameters of vesicles determined by quasielastic light scattering (Zetasizer 1000/3000, Malvern Instruments, U.K.) were found to be 80 ( 7 nm at 25 °C (average and standard deviation for six independent measurements). Tethered lipid membrane formation was achieved from egg-PC/DSPE-PEG3400-NHS mixed vesicles in buffer. Before vesicles deposition, surfaces were thoroughly rinsed with buffer. The ratio between the suspension volume and the functionalized surface area was kept constant; 390 µL/cm2 of mixed vesicle suspension was injected into the measurement cell. After injection time, the cell was rinsed with buffer at a flow rate of 1 mL/min. FRAP Experiments. For fluorescence microscopy measurements, a fluorescent probe (DPPE-NBD) was added into the vesicles at a 2% molar ratio. The diffusion coefficient and the mobile fraction of the lipids were determined from FRAP experiments according to a method previously published. The diffusion coefficient and the mobile fraction of the lipids were determined according to calculations described previously.15 SPR Measurements. Measurements were performed using a homemade optical set up in Kretschmann configuration7 and using a measurement cell previously described.15 Reflectivity was recorded as a function of incident angle, and the optical thickness was determined according to Fresnel equations using the Winspall program (Max-Planck Institute for Polymer Research, Mainz, Germany). To determine the amount of lipid absorbed onto the surface, optical thicknesses were converted to lipid mass per surface using 1 ng/mm2 for an optical thickness of 10 Å.19,29 All measurements were performed at 25 °C. AFM Experiments. Surface imaging was performed in liquid at room temperature according to a method described by Milhiet et al.30 with a commercial AFM, Multimode and Nanoscope IIIa (Veeco, Santa Barbara, CA), mounted with an EV Scanner and Si3NO4 tips (OTR4, K ) 0.018 pN/nm, radius < 5nm, Veeco). Height images were recorded in contact mode with applied forces below 80 pN, and the scan frequency was between 1 and 4 Hz. The resulting pictures were plane-fitted and flattened with the supplied Nanoscope software V5.012r5. PM-IRRAS Analysis. The reflection absorption infrared spectroscopy (RAIRS) technique is sensitive to molecules adsorbed on metallic surfaces; only vibrational modes with a dipole moment change normal to the surface are observed (metallic surfaces reflection rules). In polarization modulationRAIRS (PM-RAIRS) experiments, the IR beam is quickly modulated between the p and s polarization, and the sum and difference interferograms are processed and Fourier-transformed to yield the differential reflectivity ∆R/R ) (Rp - Rs)/(Rp + Rs). In this technique, the record of a reference spectrum on a

7570 J. Phys. Chem. B, Vol. 111, No. 26, 2007

Rossi et al.

TABLE 1: Doehlert Design for the Four Normalized Factors X1 (Percentage of DSPE-PEG-NHS), X2 (Lipid Concentration), X3 (Contact Time between the Vesicles and the Surface), and X4 (Resting Time after Buffer Rinse)a exptl variables experiment no.

X1

X2

X3

X4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

0 1 0.5 -0.5 -1 -0.5 0.5 0.5 -0.5 0 0.5 -0.5 0 0.5 -0.5 0 0 0.5 -0.5 0 0

0 0 0.866 0.866 0 -0.866 -0.866 0.289 0.289 -0.577 -0.289 -0.289 0.577 0.289 0.289 -0.577 0 -0.289 -0.289 0.577 0

0 0 0 0 0 0 0 0.816 0.816 0.816 -0.816 -0.816 -0.816 0.204 0.204 0.204 -0.612 -0.204 -0.204 -0.204 0.612

0 0 0 0 0 0 0 0 0 0 0 0 0 0.791 0.791 0.791 0.791 -0.791 -0.791 -0.791 -0.791

a

This matrix gives the list of conditions to apply to the membrane formation.

bare sample is useless. The FTIR instrument used is a commercial Nicolet Nexus spectrometer. The external beam was focused on the sample with a mirror at an optimal incident angle of 75°. A ZnSe grid polarizer and a ZnSe photoelastic modulator, modulating the incident beam between p and s polarizations (HINDS Instruments, PEM 90, modulation frequency ) 37 kHz) were placed in front of the sample. The light reflected by the sample was then focused on a nitrogen-cooled MCT detector. All the spectra reported below were recorded at 8 cm-1 resolution by coaddition of 128 scans. Surface Response Methodology. The experimental design methodology is especially adapted for modeling phenomena called “black box”. To apply an experimental design, the identification of the parameters (also called factors) that could influence the process is necessary. The process is evaluated by characteristics of the final result that are called responses (Y). The factors are given in the form of normalized variables (Xi) without units. The transformation of real variables into normalized variables is obtained according to the following relationship:

Xi )

Ui - U0i ∆Ui

where Xi is the normalized value, Ui is the real value of the factor i, Ui0 is the real value of the factor i at the center of the experimental domain, and ∆Ui is the step of variation of the real variable i. The Doehlert matrix and the corresponding experimental conditions are given in Table 1. Each response is finally described by a quadratic equation as a function of the factors, which also takes the interactions between the factors into account: f

Y ) β0 +

f

f-1

f

βiXi + ∑ βiiXi2 + ∑ ∑ βijXiXj ∑ i)1 i)1 i)1 j)i+1

where β0 is the constant coefficient, βi is the linear coefficient, βii is the quadratic coefficient, βij is the second order interaction coefficient, and f is the dimension of the factors space. In this work, f ) 4. For each response, the coefficients of the quadratic equation are given by matrix calculations. They were calculated by the following matrix operation:

β ) (XTX)-1XTY where β is the quadratic equation coefficient matrix, Y is the column matrix of the responses, in the same order as the experiments carried out and X is the matrix resulting from the multiplications between columns Xi of the Doelhert matrix in the following order : 1, X1, X2, ..., X1X2, ..., X1Xf, X2X3, ..., Xf-1Xf, X12, ..., Xf2. An analysis of variance (ANOVA) test was performed to check the adequacy of the model fit. Subsequently, all coefficients were tested to determine the significance of each parameter. The estimate of the coefficients variance is based on the calculation of the variance/covariance matrix and the calculation of residuals variance. The coefficient variance is given by the following formula.

V(βi) ) Vresiduals × Diag (XTX)-1 All calculations were performed using the Excel software (Microsoft Corporation). 3D graphs are plotted using the Maple 9 software (Maplesoft, Waterloo Maple Inc.). Results and Discussion Experimental Design Performing. In Figure 1 (formation way 1) the construction method of the polymer-tethered phospholipid bilayer is illustrated. The substrates were coated with 1-aminoethanethiol (cysteamine) and ADPMES on gold and glass, respectively. These two amine derivatives present the same general organization after self-assembly as shown previously.31,32 When interacting with the surface, the amine group is mainly H-bound with the silanol group on glass or N-bound on gold. In the first part of this work, the experimental parameters for membrane preparation were optimized with the help of an experimental design. Four majors factors were studied that were expected to have a major influence on the vesicle binding and membrane formation. The factors are the lipid concentration in the vesicle suspension, the percentage of DSPE-PEG-NHS in the vesicles, the contact time between the vesicles and the surface, and the resting time of the lipid assembly after buffer rinse. This membrane model was developed with the goal of being used for membrane protein incorporation; in addition, important factors such as ionic strength and temperature were fixed. The buffer was 20 mM HEPES Na, 150 mM NaCl at pH ) 7.8, and the temperature was 25 °C. Explored experimental domains for the lipid concentration, the contact time, and the resting time were restricted from preliminary experiments.15 The experimental domain of the percentage of PEG within the vesicles was selected to avoid having PEG moieties in a “brush” regime. When PEG coils start to repel each other, a “mushroom-to-brush” transition occurs and causes a destabilization of the liposome bilayer, which can induce liposome-to-micelle transition.33 The percentage of DSPE-PEG-NHS below which neighboring coils do not interact laterally can be predicted using scaling laws for

Surface Response Methodology

J. Phys. Chem. B, Vol. 111, No. 26, 2007 7571

TABLE 2: Correspondence between the Normalized and Real Factorsa factor (Ui) % DSPE-PEG-NHS lipid concn contact time of the vesicles resting time after buffer rinse

domain normalized factor (Xi) center (U 0i ) X1 X2 X3 X4

5 wt % 1 mg/mL 12 h 12 h

variation step (∆Ui) 5 wt % 0.9 mg/mL 11 h 12 h

a These data allow the application of the conversion equation (see the Experimental Section).

polymers at interfaces.34 For a “mushroom” regime, the PEG extension length, named Flory radius, is given by

RF ≈ aN3/5 where a is the monomer length (a ) 0.35 nm) and N is the degree of polymerization (for DSPE-PEG3400-NHS, N ) 77). According to this equation, the RF for DSPE-PEG3400-NHS is ∼4.7 nm. The average shape of PEG coils in a “mushroom” regime could be considered as a sphere with the radius RF grafted onto the vesicle surface. The distance, d, separating two grafting sites could be determined from the square root of the area and the mol fraction of anchor molecules in the vesicles (lipids being homogeneously distributed and the polar head area of egg-PC being 62 Å2 in the fluid state).35 When d > RF, the polymers will be separated “mushrooms”, whereas when d < RF, the polymers will interact laterally and a “brush” regime will develop. From these considerations, the percentage of DSPE-PEG3400-NHS for which d is equal to RF can be calculated. At 3%mol DSPE-PEG-NHS, the “mushroom-tobrush” transition will occur. A variation domain below this value was chosen for X1 (0-2.1%mol) to avoid steric hindrance between PEG moieties in the aqueous compartment between the proximal layer and the support. The Doehlert matrix (Table 1) gives the values of the four factors for 21 experiments that must be applied to the model formation. X1, X2, X3, and X4 are the normalized factors that correspond to the percentage of DSPE-PEG-NHS, the lipid concentration, the contact time between the vesicles and the surface, and the resting time after buffer rinse, respectively. In and experimental design, the factors are expressed in a normalized form to compare their effect even if they belong to a different experimental domain with different units. Conversion between normalized and real factors was realized according to the data given in Table 2 (see the Experimental Section). For every listed experimental condition, the lipid assembly process was analyzed systematically by SPR and FRAP. Fluidity and continuity of the lipid assembly on a square centimeter surface were estimated by long-range diffusion measurements using FRAP. Diffusion coefficient (response YD (cm2/s)) and mobile fraction (response YM) of lipid probes incorporated into vesicles were measured on glass supports. Upon fitting the SPR reflectivity curves, the optical thickness of the lipid assembly was evaluated on gold surfaces. This enables us to discriminate multilayer assemblies from monolayers in the case of fluid lipid organizations. To take the variety of final lipid assemblies into account, these results were expressed in lipid mass per surface unit (response Ymass (ng/cm2), see the Experimental Section). Results of SPR and FRAP analysis of the lipid assembly were implemented as the responses of the Doehlert design and are shown in Table 3 (reported data and standard deviation of three independent experiments). Experimental design calculations (see the Experimental Section) lead to three quadratic equations that model the relationships between the four normalized variables

TABLE 3: Average Response and Standard Deviation (in parentheses) for the Lateral Diffusion Coefficient of Lipids YD (cm2/s)a experiment no.

Ymassb (ng/cm2)

YD (10-8 cm2/s)

YMc

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

570 (15) 549 (7) 615 (3) 715 (4) 725 (10) 557 (3) 566 (2) 596 (6) 660 (11) 549 (3) 562 (8) 641 (3) 624 (20) 597 (6) 661 (1) 546 (3) 571 (15) 562 (8) 640 (3) 619 (11) 570 (15)

2.85 (0.05) 2.75 (0.04) 1.00 (0.13) 0.40 (0.03) 0.30 (0.15) 2.85 (0.05) 2.85 (0.05) 1.90 (0.05) 0.60 (0.05) 2.50 (0.05) 2.90 (0.05) 1.80 (0.06) 1.30 (0.04) 1.80 (0.04) 0.60 (0.05) 2.70 (0.04) 3.00 (0.05) 2.00 (0.09) 1.30 (0.04) 0.80 (0.03) 2.00 (0.09)

0.80 (0.02) 0.60 (0.01) 0.42 (0.01) 0.35 (0.02) 0.40 (0.02) 0.83 (0.04) 0.79 (0.01) 0.65 (0.02) 0.40 (0.02) 0.69 (0.05) 0.72 (0.02) 0.67 (0.02) 0.51 (0.02) 0.66 (0.02) 0.46 (0.03) 0.81 (0.01) 0.93 (0.04) 0.53 (0.02) 0.46 (0.02) 0.38 (0.02) 0.67 (0.01)

a The experimental conditions applied for each obtained response are indicated by the same experiment number in Table 1. Each experiment was repeated independently three times according to the experimental data given by the Doehlert matrix (Table 1). b Mass of lipid assembly per surface unit (ng/cm2). c Mobile fraction of lipids.

X1, X2, X3, and X4 and the characterization responses Ymass (lipid mass per surface), YD (lipid diffusion coefficient), YM (lipid mobile fraction) were obtained:

Ymass (ng/cm) ) 570 - 73X1 + 59X2 - 4X3 - 2X4 63X1X2 + 31X1X3 + 24X1X4 + 29X2X3 + 28X2X4 + 41X3X4 + 67X 21 + 35X 22 + 28X 23 + 15 ‚ X 24 YD (cm2/s) ) (2.85 × 10-8) + (0.98 × 10-8)X1 - (1.28 × 10-8)X2 - (0.24 × 10-8)X3 + (0.32 × 10-8)X4 + (0.35 × 10-8)X1X2 + (0.0 × 10-8)X1X3 + (0.19 × 10-8)X1X4 (0.02 × 10-8)X2X3 - (0.35 × 10-8)X2X4 - (1.39 × 10-8)X3X4 - (1.33 × 10-8)X 21 - (0.99 × 10-8)X 22 (0.95 × 10-8)X 23 - (1.07 × 10-8)X 24 YM ) 0.80 - 0.10X1 - 0.23X2 - 0.03X3 + 0.13X4 + 0.07X1X2 + 0.10X1X3 + 0.03X1X4 + 0.06X2X3 - 0.07X2X4 - 0.36X3X4 - 0.30X 21 0.17X 22 - 0.17X 23 - 0.17X 24 An ANOVA was applied to test the relevance of the model. The F-test between the variances of the models (regressions) and errors (residuals) led to the following value of the F ratio: 176, 80.5, and 66.5 for the responses Ymass, YD, and YM, respectively. For each response, the probability, p, that the model variance was equal to the residuals variance (pure error and lack of fit) for the respective degrees of freedom 14 and 48, was below 0.0001. This indicates a fair adjustment, which was also confirmed by the determination of the correlation coefficient (r2) associated with the proposed models. r2 values of 0.981, 0.960, and 0.965 were obtained for Ymass, YD, YM, respectively. Consequently, the performing models can be considered satisfactory for the degrees of freedom under consideration. The plot

7572 J. Phys. Chem. B, Vol. 111, No. 26, 2007

Rossi et al.

Figure 2. Pareto diagrams of (A) the lateral diffusion coefficient of the lipids, (B) the mobile fraction of the lipids, and (C) the mass of the lipid assembly on the surface. Black bars represent the significant coefficients.

of the regression residuals versus estimated values shows a random distribution for variance homogeneity (graphs not shown). Analysis of the Doehlert Modelization. The importance of the factors and their interactions were evaluated when referring to the Pareto diagram. The Pareto diagrams presented in Figure 2 represent and classify the effects of the factors X1 (DSPE-PEG-NHS percentage), X2 (lipid concentration), X3 (contact time of vesicles), and X4 (resting time) by their coefficients β in the quadratic model. This estimated value was calculated from the t-experimental (ratio between the coefficient value and its standard error) divided by the t-Student for a level of confidence of 0.01 and 48 degrees of freedom (68 experiments and 15 coefficients for the quadratic equations). The coefficients were considered statistically significant when their estimated value is g1. Their relative influence level was directly proportional to the estimated value. This analysis allows determining the influence of each factor on the final properties of the lipid assembly. The results shown in Figure 2 (A and B) indicate that both factors influence the fluidity properties (YD and YM) of the lipid construction. In contrast, the mass of the lipid architecture was mainly controlled by the initial lipid concentration and the percentage of anchoring molecules (Figure 2 C). This was confirmed by the kinetic of vesicle adsorption on amine-grafted surfaces followed by SPR measurements shown in Figure 3. Upon addition of the vesicle solution, the SPR reflectivity signal increased and stabilized in less than 1 h. The reflectivity remained constant after rinsing with buffer; thus, X3 and X4 factors do not influence the assembly in the experimental domain chosen. A similar observation was found for all Doehlert value combinations. The estimated value of the interaction coefficient (β12) (Figure 2 C) between the percentage of DSPE-PEG-NHS and the lipid concentration confirm their interdependence. Nevertheless, these two parameters determine the thickness of the lipid architecture, which is reached after 1 h of contact time. After this first step, a rearrangement phenomena depending on the contact and resting time factors influences the fluidity properties of the final lipid architecture. Figure 4 represents the interdependent evolution of the diffusion coefficient (A), the mobile fraction (B) and the lipid mass (C) as a function of the normalized variables X1 and X2.

Figure 3. SPR monitoring of the biomimetic membrane formation on a gold surface. Reflectivity (as a function of time) was recorded after injection of egg-PC/DSPE-PEG-NHS (5%w/w) vesicle suspensions with a lipid concentration of 1 mg/mL on the amine-grafted gold surface. Reflectivity as a function of the incident angle is shown in the inset: (a) SPR spectra of gold covered by cysteamine and (b) after lipid vesicle adsorption on the amine surface.

To represent the response evolution as a function of two parameters, the other factors X3 and X4 were kept constant and therefore equal to the normalized values (-0.7, 0.75) and (-1, 1). Figure 5 represents the response evolution of the diffusion coefficient (A) and the mobile fraction (B) as a function of the normalized factors X3 and X4. To represent the response evolution as a function of two parameters, the other parameters X1 and X2 were kept constant and equal to the normalized values (0.33, -0.69) and (-0.07, -1), respectively. Considering the high influence of the interaction between X3 and X4 (β34), we chose to represent the evolution of YD and YM as a function of X1 and X2 (Figure 4A and B, respectively) and as a function of X3 and X4 (Figure 5A and B, respectively) by setting the two other variables constant. The three response graphs show an identical location for their extrema that correspond to the

Surface Response Methodology

J. Phys. Chem. B, Vol. 111, No. 26, 2007 7573

Figure 5. Z-axis projection of 3D graphs of response evolution as a function of the normalized factors X3 (contact time of the vesicles with the surface) and X4 (resting time after buffer rinse). The values that correspond to the real domain are given in parentheses. The responses are (A) diffusion coefficient of the lipids (YD) and (B) mobile fraction (YM). Color scales for each response are given on the side of the graph; black lines represent isoresponses.

Figure 4. Z-axis projection of 3D graphs of response evolution as a function of the normalized factors: DSPE-PEG-NHS weight percentage, X1, and the total lipid concentration, X2. The values that correspond to the real domain are given in parentheses. The responses are (A) diffusion coefficient (YD), (B) mobile fraction (YM), and (C) the mass per surface unit (Ymass). Color scales for each response are given on the side of the graphs; black lines represent isoresponses.

extreme lipid organization like a bilayer or adsorbed vesicles, as illustrated Figure 1, ways 1 and 4. A high lipid concentration (1.9 mg/mL) and the absence of DSPE-PEG-NHS are conditions that lead to the minimal

diffusion and mobile fraction values (Figure 4). These conditions also give the higher mass of lipid per surface coverage (725 ( 10 ng/cm2) (Table 3, experiment 5) which is not in agreement with a single bilayer formation.20 By AFM imaging, it could be seen that intact vesicles were adsorbed on the surface (data not shown). The amine groups on the surface were mainly nonprotonated and present a neutral global charge. This can explain the adsorption of intact vesicles on the surface (Figure 1, way 4). A similar observation made was by Soufou et al., who reports the adsorption of intact egg-PC vesicles on a protonated carboxylic acid-coated surface.36 Terminally functionalized alkyls that are organized in a densely packed monolayer on the surface undergo a drastic shift of their pKa due to additional intermolecular interactions.37 The pKa of cysteamine determined in previous studies was found to be in the range of 5.0 ( 0.3 for comparable buffer salinity (100250 mM NaCl).38 Moreover, the fusion of lipid vesicles with a surface was recently demonstrated to be critically dependent

7574 J. Phys. Chem. B, Vol. 111, No. 26, 2007

Rossi et al.

TABLE 4: Maximum Values and Locations for Diffusion Coefficient (YD) and Mobile Fraction (YM) Responses. normalized factors diffusion coefficient mobile fraction

real factors

maximum

X1

X2

X3

X4

U1 (%w)

U2 (mg/mL)

U3 (h)

U4 (h)

3.65 10-8 cm2/s 1

0.33 1

-0.69 -0.07

-0.70 -1

0.75 1

6.6 4.6

0.4 0.1

4.3 1

21 24

on the global charge of the surface and the nature of counterions in the buffer.39 Following the above considerations, a simultaneous inspection of the three responses (Ymass, YD, YM) is necessary to gather all information about the organization state of the lipid architecture on the surface and to find the experimental conditions allowing the best formation of the biomimetic membrane. As shown in Figure 4, Ymass and both YD and YM evolved in opposite directions. The possible and different organization state of the lipids on the surface can explain this observation. For example, the experimental conditions under which X1 is between -0.5 and -1 and X2 is between 0 and -1 lead to high values of YD and YM. These conditions led also to high values of Ymass, suggesting the presence of multilayer assemblies, as shown Figure 1, way 3. In the same way, the conditions when X1 is between 0 and 1 and X2 is between 0 and 0.7 lead to low values of Ymass, YD, and YM, suggesting the presence of discontinuous lipid aggregates or interrupted bilayers as, schematically illustrated in Figure 1, way 2. The experimental conditions that give the highest values for YD and YM and the lowest Ymassvalues are expected to lead to a planar and fluid bilayer (Figure 1, way 1).The experimental conditions from which the maximal values for diffusion coefficient and mobile fraction were obtained are reported in Table 4. Determination of the coordinates of the stationary point was resolved by partially differentiating the response function equation system.

{

∂Yi ∂Yi ∂Yi ∂Yi )0 )0 )0 ) 0 where i ) D or M ∂X1 ∂X2 ∂X3 ∂X4

As a consequence, the experimental domain that leads to membrane properties corresponding to the biomimetic requirements was restricted to between 0.1 and 1 mg/mL for the lipid concentration and between 4 and 8% of the mass percentage of DSPE-PEG-NHS. As shown in Figure 5, a short contact time of vesicles with the surface and a long resting time of the lipid assembly after buffer rinse are the best conditions for obtaining fluid bilayer assembly. These factors were restricted to 1-4 h for the contact time and 21-24 h for the resting time after buffer rinse. These conditions lead to a mass per surface area between 545 ( 5 and 590 ( 10 ng/cm2 with a diffusion coefficient

between 2.5 ( 0.3 × 10-8 and 3.60 ( 0.5 × 10-8 cm2/s and a mobile fraction between 94 ( 2 and 99 ( 1%. Characterization of the Tethered Bilayer Model. Under these experimental conditions, the lipid mass is in accordance with previously published data.5,40 The selected values for the lateral diffusion coefficient were comparable to those previously measured in bilayers that were supported directly on silicate substrate (3.6 ( 0.5 × 10-8 cm2/s)41 or in tethered bilayers [(0.8-1.2 × 10-8 cm2/s);17 (3 ( 1 × 10-8 cm2/s)42]. All FRAP experiments discussed above were performed on glass substrates. To check the possibility of the tethered bilayer formation on gold and glass, fluorescence experiments were also performed on the stripped gold surface within the range of the selected experimental domain (data not shown). The obtained values for the diffusion coefficient and the mobile fraction were similar to those found on the glass surface. The lipid diffusion coefficient and the mobile fraction of bilayers formed on both substrates under the same conditions (vesicle solution containing 5% DSPE-PEG-NHS with a lipid concentration of 1 mg/mL, contact time: 2 h 30 min and 24 h resting time after buffer rinsing) were 2.9 ( 0.2 × 10-8 cm2/s and 96 ( 2%, respectively, for the glass and 2.7 ( 0.4 × 10-8 cm2/s and 94 ( 3%, respectively, for the gold support. These data confirm that the results obtained by SPR and FRAP can be compared though the experiments that were performed on different substrates. Moreover, it demonstrates the versatility of bilayer formation on different amine-coated substrates. The continuous evolution of the lipid mass response (Figure 4) showed that many intermediate states (aggregates, vesicles, etc.) exist. The formation of the final lipid assembly from the anchored vesicles needs further investigation. AFM analysis was performed to examine the self-assembly of the lipid structures on the nanometer scale. Measurements were performed under the following conditions: a lipid concentration of 1 mg/mL containing 5% DSPE-PEG-NHS, a contact time of vesicles with the amine surfaces of 2 h 30 min, and a resting time after buffer rinse of 24 h. The stripped gold surface (Figure 6A) was particularly flat with a mean algebraic roughness of 0.31 nm for 500 nm × 500 nm images, as previously shown.27,28 When coating the surface with cysteamine and upon membrane formation, a vanishing of the gold cluster details was observed due to the presence of a

Figure 6. Membrane formation followed by AFM imaging. Height images that were obtained by AFM in contact mode for the same flat gold substrate (A) uncoated, (B) functionalized with cysteamine, and (C) covered with an egg-PC bilayer.

Surface Response Methodology

J. Phys. Chem. B, Vol. 111, No. 26, 2007 7575 is characteristic for the carbonyl stretching vibration in the succinimide ester48 and shows that not all the NHS groups have reacted with the amine surface. Conclusion

Figure 7. PM-IRRAS spectra (a) after formation of the cysteamine monolayer, (b) after deposition of vesicles onto uncoated gold surfaces, and (c) after covalent binding of the vesicles onto the cysteamine selfassembled monolayer.

covering layer of alkyl chains on the hard gold surface. This vanishing effect increased with the thickness of the covering organic layer; therefore, details that were clearly visible on the pure gold surface became less visible on the cysteamine-coated gold surface (Figure 6B) and practically invisible on the gold surface covered with an egg-PC bilayer (Figure 6C). This effect can be measured by the algebraic and the quadratic roughness parameters that were found to decrease while the layer thickness of the coating was increasing (data not shown). In addition, the skewness parameter was calculated for each surface. The skewness parameter is a nondimensional quantity that measures the symmetry of surface data around a mean data profile. The negative value (-1) observed on the bare gold surface suggests deeps that vanish when the surface is covered with egg-PC, resulting in a skewness parameter value of about 0, which suggests an even distribution of data around the mean data plane. All AFM measurements and analyses confirm the presence of a homogeneous surface in the presence of a covering egg-PC bilayer. The efficient anchoring of the lipid architecture was investigated by PM-RAIRS. IR spectra were recorded during the formation procedure (illustrated in Figure 1) (a) after the formation of the cysteamine monolayer, (b) after deposition of egg-PC/DSPE-PEG-NHS vesicles on an uncoated gold substrate, and (c) after deposition of these same vesicles on a cysteamine-coated gold surface (Figure 7). In the first spectrum (a), two main bands are observed, at 1650 and 1525 cm-1. These bands can be attributed to N-H bending and scissoring modes of the NH2 moieties, respectively.43-45 In the second spectrum (b), four bands are characteristic for the mixture of lipid vesicles. The one at 1740 cm-1 can be attributed to the lipid carbonyl stretching mode.46 The large IR absorption band at 1640 cm-1 is most likely composed of different contributions due to the phospholipids and PEG functions. The band at 1470 cm-1 can be attributed to δC-H2 and the last one 1370 cm-1 could be associated with the stretching mode of the PdO group. After deposition of the egg-PC/DSPE-PEG-NHS vesicles onto the amine monolayer, we can observe (Figure 7, spectrum c) the decrease in the band characteristic of δN-H at 1525 cm-1 and the appearance of an IR band at 1595 cm-1, which was absent in the spectrum recorded after the simple adsorption of the vesicles onto the gold surface. This last band can be attributed to the so-called amide II band, which is due to a motion combining both the N-H bending and the C-N stretching vibrations of the group -CO-NH-.47 This group is formed by the covalent binding of the N-hydrosuccinimide moieties of the vesicles to the amine monolayer. The IR band at 1820 cm-1

Our aim was to determine a set of experimental conditions allowing the formation of a tethered bilayer in a reliable and fast way. The construction of the tethered membrane was achieved using vesicle deposition onto an activated surface. The activation of the surface was obtained from self-assembled monolayers bearing amine groups. The bilayer formation step was completed by deposition of vesicles containing a mixture of egg-PC and DSPE-PEG3400-NHS, which acts as both a spacer and an anchor molecule. Different supports, such as glass, gold, and silicon, can be easily amine-coated, which makes the biomimetic membrane construction adaptable to many substrates. In this study, the transposition of the tethered bilayer formation was investigated on glass and gold surfaces. The formation of the lipid assembly was followed systematically by FRAP experiments on glass surfaces and by SPR measurements on gold surfaces. Fluorescence data on both substrates were identical and confirm the pertinence of the comparative FRAP/SPR approach for the experimental design investigation. The influence of the following four factors was studied: lipid concentration in the vesicle suspension, the percentage by mass of DSPE-PEG-NHS in the vesicles, the contact time between vesicles and the surface, and the resting time after buffer rinse. The bilayer formation was studied by means of an original optimization method: the investigation of the effect of four factors was performed using a Doehlert experimental design. This study has led to new qualitative and quantitative information concerning the tethered bilayer formation process. Three response parameters were examined: the lipid mass on the surface, the diffusion coefficient, and the mobile fraction of the final lipid organization. These parameters were described by a second-order model that was adequate to predict responses in all experimental regions. The four factors have a significant influence on the fluidity of the lipid assembly, but only the lipid concentration and the percentage of DSPE-PEG-NHS have a major influence on the lipid mass present on the surface. In all cases, the continuous evolution of the three characteristics (diffusion coefficient, mobile fraction, optical thickness) shows that the lipid assembly undergoes many intermediate organizations steps. The two extremes are the homogeneous bilayer and vesicles adsorbed or linked to the surface at high lipid concentrations and low anchor molecules concentrations. By simultaneous studies of the three characteristics, it appeared that only a restricted experimental domain led to properties in accordance with a bilayer presence. These findings were confirmed by AFM observations that validate the presence of a continuous and homogeneous bilayer in the determined experimental domain. This domain is characterized by a lipid concentration from 0.1 to 1 mg/mL in the vesicle suspension, 4-8% of DSPE-PEG-NHS in the vesicles, 1-4 h for the contact time of the vesicles with the surface, and 21-24 h of resting time after the buffer rinse. The biomimetic membrane formed was presented a lipid mass per surface between 545 ( 5 and 590 ( 10 ng/cm2, a diffusion coefficient of 2.5 ( 0.3 × 10-8 and 3.60 ( 0.5 × 10-8 cm2/s, and a mobile fraction between 94 ( 2 and 99 ( 1%. The membrane model studied here was first envisaged for investigating the calcium-dependent membrane-binding properties of two proteins, the Neurocalcin and the Adenylate cyclase from Bordetella pertussis.15 More recently, the functional

7576 J. Phys. Chem. B, Vol. 111, No. 26, 2007 reconstitution of the voltage-dependent anion channel on solid support was achieved using this biomimetic membrane.49 A similar strategy using a lipid-PEG-thiol that ensures anchorage to a gold surface was described for the functional immobilization of the nicotinic acetylcholine receptor from Torpedo california.50 The experimental approach described in this work is a convenient and fast way for membrane formation. Using mixed vesicles offers the possibility to control the anchor chain density between the bilayer and the substrate in an easier way than by Langmuir-Blodgett transfer.17,51,52 Moreover, this construction allows one to reduce the density of the spacer layer, which is not feasible in the polymer cushion approach13 or with tethered lipid monolayers formed by self-assembly.53 The possibility of dispersing the presence of tethers would reduce steric limitations for the incorporation of large transmembrane proteins. Acknowledgment. We thank CNRS for financial support. We are grateful to Andre Lopez for scientific support and discussions concerning FRAP experiments and experimental design application. We thank Dr. V. Humblot, from the LRS, for helpful discussion concerning the IR interpretation of the spectra. We acknowledge Patrick Cox and Inga Vockenroth for critical reading of the manuscript. References and Notes (1) Brian, A. A.; McConnell, H. M. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 6159-6163. (2) Tamm, L. K.; McConnell, H. M. Biophys. J. 1985, 47, 105-113. (3) Knoll, W.; Frank, C. W.; Heibel, C.; Naumann, R.; Offenhausser, A.; Ruhe, J.; Schmidt, E. K.; Shen, W. W.; Sinner, A. J. Biotechnol. 2000, 74, 137-158. (4) Richter, R. P.; Berat, R.; Brisson, A. R. Langmuir 2006, 22, 34973505. (5) Nollert, P.; Kiefer, H.; Jahnig, F. Biophys. J. 1995, 69, 14471455. (6) Plant, A. L. Langmuir 1993, 9, 2764-2767. (7) Lingler, S.; Rubinstein, I.; Knoll, W.; Offenhausser, A. Langmuir 1997, 13, 7085-7091. (8) Tanaka, M.; Sackmann, E. Nature 2005, 437, 656-663. (9) Masson, L.; Mazza, A.; Brousseau, R. Anal. Biochem. 1994, 218, 405-412. (10) Jung, L. S.; Shumaker-Parry, J. S.; Campbell, C. T.; Yee, S. S.; Gelb, M. H. J. Am. Chem. Soc. 2000, 122, 4177-4184. (11) Terrettaz, S.; Stora, T.; Duschl, C.; Vogel, H. Langmuir 1993, 9, 1361-1369. (12) Cooper, M. A.; Williams, D. H. Anal. Biochem. 1999, 276, 3647. (13) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58-64. (14) Ellson, C. D.; Gobert-Gosse, S.; Anderson, K. E.; Davidson, K.; Erdjument-Bromage, H.; Tempst, P.; Thuring, J. W.; Cooper, M. A.; Lim, Z. Y.; Holmes, A. B.; Gaffney, P. R.; Coadwell, J.; Chilvers, E. R.; Hawkins, P. T.; Stephens, L. R. Nat. Cell Biol. 2001, 3, 679-682. (15) Rossi, C.; Homand, J.; Bauche, C.; Hamdi, H.; Ladant, D.; Chopineau, J. Biochemistry 2003, 42, 15273-15283. (16) Cooper, M. A. J. Mol. Recognit. 2004, 17, 286-315.

Rossi et al. (17) Wagner, M. L.; Tamm, L. K. Biophys. J. 2000, 79, 1400-1414. (18) Sinner, E.-K.; Knoll, W. Curr. Opin. Chem. Biol. 2001, 5, 705711. (19) Knoll, W. Annu. ReV. Phys. Chem. 1998, 49, 569-638. (20) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397-1402. (21) Jass, J.; Tjarnhage, T.; Puu, G. Biophys. J. 2000, 79, 3153-3163. (22) Lang, H.; Duschl, C.; Gra¨tzel, M.; Vogel, H. Thin Solid Films 1992, 210, 818-821. (23) Reimhult, E.; Hook, F.; Kasemo, B. Langmuir 2003, 19, 16811691. (24) Doehlert, D. H. Appl. Stat. 1970, 19, 231-239. (25) Ferreira, S. Talanta 2004, 63, 1061-1067. (26) Gonzalez, G.; Gonzalez-Arjona, D. Talanta 1999, 49, 433-439. (27) 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. (28) Schiller, S. M.; Naumann, R.; Lovejoy, K.; Kunz, H.; Knoll, W. Angew. Chem., Int. Ed. 2003, 42, 208-210. (29) Stenberg, E.; Personn, B.; Ross, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513-526. (30) Milhiet, P. E.; Giocondi, M. C.; Baghdadi, O.; Ronzon, F.; Roux, B.; Le Grimellec, C. EMBO Rep. 2002, 3, 485-490. (31) Kanan, S. M.; Tze, W. T. Y.; Tripp, C. P. Langmuir 2002, 18, 6623-6627. (32) Michota, A.; Kudelski, A.; Bukowska, J. Surf. Sci. 2002, 502503, 214-218. (33) Tirosh, O.; Barenholz, Y.; Katzhendler, J.; Priev, A. Biophys. J. 1998, 74, 1371-1379. (34) deGennes, P. G. AdV. Colloid Interface Sci. 1987, 27, 189-209. (35) Cevc, G. In Phospholipids Handbook, 1st ed.; Marcel Dekker, Inc.: New York, 1993; p 937. (36) Soufou, S.; Thomas, J. L. Biosens. Bioelectron. 2003, 18, 445455. (37) Bain, C. D.; Whitesides, G. M. Langmuir 1989, 5, 1370-1378. (38) Kakiuchi, T.; Iida, M.; Imabayashi, S. I.; Katsumi, N. Langmuir 2000, 16, 5397-5401. (39) Cha, T.; Guo, A.; Zhu, X. Y. Biophys. J. 2006, 90, 1270-1274. (40) Salamon, Z.; Tollin, G. Biophys. J. 1996, 71, 858-867. (41) Kalb, E.; Frey, S.; Tamm, L. K. Biochim. Biophys. Acta 1992, 1103, 307-316. (42) Proux-Delrouyre, V.; Elie, C.; Laval, J.-M.; Moiroux, J.; Bourdillon, C. Langmuir 2002, 18, 3263-3272. (43) Hooper, A. E.; Werho, D.; Hopson, T.; Palmer, O. Surf. Interface Anal. 2001, 31, 809-814. (44) Mougin, K.; Lawrence, M. B.; Fernandez, E. J.; Hillier, A. C. Langmuir 2004, 20, 4302-4305. (45) Liu, Z.; Amiridis, D. J. Phys. Chem. B. 2005, 109, 16866-16872. (46) Wang, L.; Cai, P.; Galla, H. J.; He, H.; Flach, C. R.; Mendelsohn, R. Eur. Biophys. J. 2005, 34, 243-254. (47) Socrates, G. Infrared Characteristic Group Frequencies; John Wiley and Sons Ltd: New York, 1980. (48) Frey, B. L.; Corn, R. M. Anal. Chem. 1996, 68, 3187-3193. (49) Deniaud, A.; Rossi, C.; Berquand, A.; Homand, J.; Campagna, S.; Knoll, W.; Brenner, C.; Chopineau, J. Langmuir 2007, 23, 3898-3905. (50) Sevin-Landais, A.; Rigler, P.; Tzartos, S.; Hucho, F.; Hovius, R.; Vogel, H. Biophys. Chem. 2000, 85, 141-152. (51) Lang, H.; Duschl, C.; Vogel, H. Langmuir 1994, 10, 197-210. (52) Naumann, C. A.; Prucker, O.; Lehmann, T.; Ruhe, J.; Knoll, W.; Frank, C. W. Biomacromolecules 2002, 3, 27-35. (53) Seitz, M.; Ter-Ovanesyan, E.; Hausch, M.; Park, C. K.; Zasadzinski, J. A.; Zentel, R.; Israelachvili, J. N. Langmuir 2000, 16, 6067-6070.