Characterization of the physical properties of model biomembranes at the nanometer scale with the atomic force microscope Yves F. Dufre� ne,a,b,§ Thomas Boland,a James W. Schneider,a William R. Bargera and Gil U. Lee*a a Chemistry Division Code 6177 Naval Research L aboratory W ashington DC 20375-5352 USA. E-mail glee=stm.2.nrl.navy.mil b Center of Marine Biotechnology University of Maryland Columbus Center 701 E. Pratt Street Baltimore MD 21202 USA Receiøed 1st October 1998 Interaction forces and topography of mixed phospholipid»glycolipid bilayers were investigated by atomic force microscopy (AFM) in aqueous conditions with probes functionalized with self-assembled monolayers terminating in hydroxy groups. Short-range repulsive forces were measured between the hydroxy-terminated probe and the surface of the two-dimensional (2-D) solid-like domains of distearoyl-phosphatidylethanolamine (DSPE) and digalactosyldiglyceride (DGDG).The form and range of the short-range repulsive force indicated that repulsive hydration/steric forces dominate the interaction at separation distances of 0.3»1.0 nm after which the probe makes mechanical contact with the bilayers. At loads \5 nN the bilayer was elastically deformed by the probe while at higher loads plastic deformation of the bilayer was observed. Surprisingly a short-range repulsive force was not observed at the surface of the 2-D liquid-like dioleoylphosphatidylethanolamine (DOPE) –lm despite the identical head groups of DOPE and DSPE. This provides direct evidence for the in—uence of the structure and mechanical properties of lipid bilayers on their interaction forces an eÜect which may be of major importance in the control of biological processes such as cell adhesion and membrane fusion.The step height measured between lipid domains in the AFM topographic images was larger than could be accounted for by the thickness and mechanical properties of the molecules. A direct correlation was observed between the repulsive force range over the lipid domains and the topographic contrast which provides direct insight into the fundamental mechanisms of AFM imaging in aqueous solutions. This study demonstrates that chemically modi–ed AFM probes can be used in combination with patterned lipid bilayers as a novel and powerful approach to characterize the nanometer scale chemical and physical properties of heterogeneous biosurfaces such as cell membranes.§ Current address Uniteç de Chimie des Interfaces Universiteç Catholique de Louvain Place Croix du Sud 2/18 B-1348 Louvain-la-Neuve Belgium. 79 Faraday Discuss. 1998 111 79»94 Introduction Many cellular phenomena including cell adhesion and membrane fusion involve speci–c molecular interactions and non-speci–c interactions between bilayer membranes. While the function of (glyco)proteins such as integrins,1 selectins2 and lectins,3 in these interactions has long been recognized there is now growing evidence supporting the involvement of lipids. Gangliosides are charged glycosphingolipids found in most animal membranes which are thought to have important functions in cell adhesion molecular recognition oncogenesis and diÜerentiation.4h6 On the other hand phosphatidylcholine phosphatidylethanolamine and galactolipids the most common lipids in animal and plant membranes prevent membranes and vesicles from coming into contact and reacting adhering or fusing due to strong repulsion between their highly hydrated head groups.7h10 Interaction forces between opposing supported lipid bilayers have been directly measured using the surface forces apparatus (SFA).For uncharged galactolipid bilayers in NaCl solutions two types of forces have been identi–ed i.e. long-range attractive van der Waals forces and shortrange repulsive hydration/steric forces.8 A doubling of the range of the repulsive force was found when comparing monogalactosyldiglyceride with digalactosyldiglyceride.For phosphatidylethanolamine and phosphatidylcholine bilayers in the presence of millimolar levels of Ca2` and Mg2` a third force was detected in agreement with theory i.e. long-range electrostatic double-layer.9 A second approach the osmotic stress (OS) method has been widely used to probe short-range hydration forces between lipid bilayers in multilamellar systems.11h14 SFA and OS studies have provided valuable insight into membrane»membrane interactions but they are not suited for probing local properties of complex biosurfaces i.e. surfaces made of multiple component mixtures and exhibiting lateral organization at the nanometer scale. Since its invention in 1986 atomic force microscopy (AFM Fig.1) has been widely used to image the surface of materials down to molecular resolution.15 AFM has been applied to many biological systems due to its high resolution and ability to operate in aqueous environments. Subnanometer scale resolution has been achieved on proteins,16h18 DNA»protein complexes19,20 and bilayers,21h24 but the resolution of cell surfaces has been limited to 10»30 nm18,19,25,26 due to the complex interplay between topographic mechanical and frictional forces. One solution to this problem is to reconstruct more rigid models for lipid bilayer membranes using the Langmuir» Blodgett (LB) deposition technique.27,28 Such an approach has enabled several research groups to image supported lipid membranes at high resolution under physiological conditions.22,24,29 Besides imaging the very high force sensitivity of the AFM has been used to measure forces with nanometer scale lateral resolution.Quantitative measurements have been made of van der Waals and electrostatic forces,30h32 hydration forces,31,33,34 steric repulsive forces,35,36 speci–c Fig. 1 Schematic of an atomic force microscope. A molecular resolution image of a surface is constructed by rastering a microfabricated probe over a surface under constant low loads. The critical elements of the instrument are the microfabricated cantilever-probe (A) piezoelectric electromechanical actuator (B) optical-lever displacement detector (C) and digital feed-back loop (not shown). The dimensions of the cantilever can be modi–ed to select a spring constant and cantilevers have been microfabricated with spring constants of 100» 0.0001 N m~1.Microfabrication gives the force transducer a high temporal resolution due to the minute mass of the cantilever. Faraday Discuss. 1998 111 79»94 80 intermolecular interactions37,38 and molecular mechanics.39 Of particular interest is the mapping of interaction forces with nanometer scale resolution by recording force»distance curve maps in the (x,y) plane.40h42 A key issue that must be addressed when measuring forces with the AFM is the probe surface chemistry. The most commonly used probes for surface imaging are made of silicon oxynitride. These probes have a complex surface chemistry made of ionizable silanol and amine groups but also of contaminants.43 Reproducible measurements of forces attributable to interactions between speci–c functional groups on a surface require rigorous control of the surface chemistry of the probe.Sharp probes bearing single chemical moieties such as alcohol or aliphatic groups can be produced using self-assembly of functionalized alkanethiols onto Au-coated probes. This method has allowed mapping of the spatial arrangement and interaction forces on surfaces patterned with diÜerent chemical groups.44,45 In this paper we describe the nanometer scale topography mechanical properties and surface forces of lipid bilayers that model biological membranes in aqueous solution. To measure unambiguously short-range forces in the absence of electrostatic double-layer contribution we used lipid bilayers with a net zero surface charge and probes modi–ed by self-assembly of alkanethiol monolayers terminating in hydroxy groups.Bilayers made of phosphatidylethanolamines and galactosyldiglycerides were selected for several reasons ( these lipids are the most common in animal and plant membranes respectively ; (ii) controlled deposition of these bilayers on mica using the LB technique has been described previously ;9 (iii) data on surface forces between such bilayers are available from other techniques for comparison.46 These results represent the –rst systematic analysis of short-range interaction forces of lipid bilayers using chemically de–ned AFM probes and set the stage for the characterization of the physical properties of complex supported bilayers and possibly even living cells with molecular resolution.3)49 and then rinsed 16 S 2p Si 2s O 1s Au 4f C 1s 7@2 Cr 2p3@2 \0.5 3 2 1 11 8 57 37 29 52 \0.5 \0.5 Materials and methods Preparation of thiol monolayers Cleaved [0001] muscovite mica and AFM probes were functionalized with self-assembling alkanethiols using conditions optimized for the formation of organized monolayers.47,48 To this end mica surfaces and probes were coated with a 2 nm thick Cr layer followed by 15 nm of Au using electron beam thermal evaporation at high vacuum (CVC Products Inc. Rochester NY USA). The gold coated surfaces were immediately immersed for 18 h in 0.2 mM ethanolic solutions of 16-hydroxyhexadecane-1-thiol (C -OH) and hexadecane-1-thiol (C -CH 16 15 with ethanol to remove excess alkanethiol.The chemical composition of the C -OH coating was 16 determined by X-ray photoelectron spectroscopy (XPS) using Au/Cr-coated silicon as a standard (Table 1). Sulfur was detected in signi–cant concentrations at the C -OH surface and a dramatic 16 increase in the carbon surface concentration was observed correlating with a decrease in the gold surface concentration. These results are consistent with previous XPS studies of alkanethiol monolayers. 47 Table 1 Surface chemical composition [atomic fraction (%)] determined by XPS for Au/ Cr-coated silicon and Au/Cr-coated silicon after self-assembly of C -OHa Sample Au/Cr-silicon Au/Cr-silicon]C -OH 16 a n-Type S100T silicon wafers (Wafernet San Jose CA USA) cleaved into 1]1 cm2 chips were submitted to the same coating treatments as the cantilevers and directly analyzed by XPS (220iXL Fisons East Grinstead UK).The pressure during analysis was D10~7 Pa. Spectra were collected with monochromatic Al-Ka X-rays with a nominal 150]800 lm spot size. The pass energy was 20 eV and the take-oÜ angle was 90°. The atomic concentrations were calculated using the standard sensitivity factors with a program supplied with the instrument. Results are expressed as atomic fractions (%) with respect to the sum of all elements except hydrogen. 81 Faraday Discuss. 1998 111 79»94 Preparation of supported bilayers The lipids used in the study dioleoylphosphatidylethanolamine (DOPE) distearoylphosphatidylethanolamine (DSPE) and digalactosyldiglyceride (DGDG) were obtained from Sigma Chemical Co.(St. Louis MO USA) and Matreya Inc (Pleasant Gap PA USA) and their purity was con–rmed with electrospray mass spectroscopy. For DGDG the hydrocarbon chains were saturated and mostly made of 18 carbons. All lipids were dissolved at 0.5 mM in chloroform»methanol (4 1). Monolayers were spread at 25 °C at the air/water (triply distilled) interface of a KSV5000 Langmuir»Blodgett system (KSV Instruments Helsinki Finland) compressed at a rate of 1 mN m~1 min~1 and transferred at a constant surface pressure of 25 mN m~1. Mixed monolayers (1 1 molar ratio) of DSPE/DOPE and DGDG/DOPE were transferred onto mica for imaging in air or onto DSPE-coated mica substrates for imaging in water as described previously.50 Water was replaced by 0.15 M NaCl solutions before AFM measurements.Pure lipid monolayers were deposited onto silicon wafers with a 250 Aé thick oxide layer (Transition Technology Interactions Sunnyvale CA USA) for ellipsometry. Atomic force microscopy Measurements were made at room temperature (20»25 °C) using an optical lever AFM equipped with a liquid cell (Nanoscope III Digital Instruments Santa Barbara CA USA). Thermal oxide sharpened microfabricated Si3N4 probe-cantilevers (Park Scienti–c Instruments Mountain View CA USA) with a nominal tip radius of 25 nm were used. Topographic images were taken in the constant-de—ection mode with an applied force kept as low as possible i.e. \1 nN and scan rates of D4 Hz.The sensitivity of the AFM detector was estimated using the slope of the retraction force curve in the region where the probe and sample are in contact. Adhesion maps were obtained by reconstructing 64]64 force»distance curves per image and displaying the pull-oÜ force measured for each force curve. Images were then resampled to 512]512 pixels. For quantitative interpretation of the force»distance curves the non-linear response (displacement/applied voltage) of the piezoceramic actuator of the scanner was corrected using a capacitance-based displacement calibrator (Capacitec Boston MA). The cantilever spring constants ranging from 0.01 to 0.6 N m~1 were determined using micromachined reference cantilevers of precisely controlled spring constants following the procedure of Tortonese and Kirk.51 Ellipsometry Ellipsometric measurements of monolayer properties were made in air at 44 wavelengths between 400 and 800 nm at angles of incidence of 65 70 75 and 85° (M-44 spectroscopic ellipsometry J.A. Woollam Co Inc.). The rotating analyzer ellipsometer uses a broad band light source with a monochromator built into the solid state detector. The calcite Glan Taylor polarizer is mounted in a high precision rotation stage with accurate positioning below 0.01° and a maximum beam deviation of 0°1@. The angle of incidence phase (D) and amplitude (W) were known with an uncertainty of less than 0.005 0.02 and 0.01° respectively. The optical properties of the –lms were –t with a three-layer model.52 The four unknown physical parameters are the –lm thickness –lm coverage and the real and imaginary parts of the permittivity of the –lm.For each wavelength the ellipsometer yields the two re—ection parameters D and W thus measurements at two angles of incidence should in principle be sufficient to solve for all unknowns. In our case W is largely independent of the layer thickness and experiments at additional angles were necessary. The optical constants of the initial bare gold or silicon surface were used as substrate references and the optical constants of the organic layer were –t with a two parameter Cauchy model with no absorptive component (k\0). Film thickness and optical constants of the –lm were then evaluated in an iterative procedure to satisfy variation with wavelength and angle of incidence.Results Knowledge of the physical and chemical properties of lipid bilayer membranes at high spatial resolution is a key step towards a better understanding of complex biological surfaces and cellular 82 Faraday Discuss. 1998 111 79»94 16 15 3 surfaces (Fig. 2A). The C -OH functionalized probes were then used to study the nanometer events such as cell adhesion and membrane fusion. Functionalized AFM probes have been used to investigate the nanometer scale surface properties of mixed lipid bilayers under aqueous conditions. To evaluate the relative contributions of short-range surface forces and mechanics in the measured forces we –rst made force measurements between model hydrophilic and hydrophobic surfaces in water i.e.between C -OH functionalized AFM probes and either C -CH or C16- OH 16 scale surface properties of mixed phospholipid/glycolipid bilayers prepared as follows either DSPE or DGDG (saturated hydrocarbon chains) were mixed with DOPE (unsaturated hydrocarbon chains) and the mixed monolayers were transferred as a second layer onto DSPE-coated surface yielding bilayers with hydrophilic phosphatidylethanolamine and galactoside head groups exposed to the aqueous phase (Fig. 2B). Alkanethiol monolayers 16 and C -CH monolayers was D5° and 110^2° respectively which is consistent with 3 The alkanethiol –lms were characterized with ellipsometry and contact angle measurements. The ellipsometric thickness of the C -CH and C -OH –lms was 2.1^0.1 nm (refractive index 15 3 16 n\3) and 2.4^0.1 nm (n\3) respectively.The thickness of the C -CH –lm is in quantitative 15 3 agreement with earlier ellipsometric studies and con–rms the alkanethiol forms a closed packed monolayer tilted 30° to the surface normal.47 The hydroxy terminated C -OH –lm is 0.3 nm 16 thicker than the methyl terminated C -CH monolayer which we attribute to the formation of a 15 3 water monolayer on the C -OH –lm under ambient conditions. The water contact angle on the C -OH 16 15 close packed monolayers. The forces measured between the C -OH and C -CH surfaces and C -OH probe in water 16 15 3 16 are plotted as a function of surface separation in Fig. 3. The form of the force curve was not aÜected by the maximum force applied to the surface for forces up to 20 nN.Only repulsive forces were observed as the C -OH surface was approached to and retracted from the C -OH probe 16 16 Fig. 3A. The lack of signi–cant adhesive force suggests a strong repulsive force counterbalances the attractive interactions. In contrast to the C -OH/C -OH measurements signi–cant hyster- 16 16 esis was observed in the retracting force curve between the C -CH surface and C -OH probe 15 3 16 Fig. 3B. ìPull-oÜœ forces of D7.5 nN magnitude were observed which are indicative of a strong attractive force when the probe is in contact with the surface. and DSPE were constructed using an energy minimization molecular dynamics algorithm (Chem3D 16 16-OH Fig. 2 Schematic diagrams of forces measurements made between (A) a C -OH modi–ed probe and surface and (B) a C -OH modi–ed probe and a homogeneous DSPE/DSPE lipid bilayer.The space –lling models of 16 C -OH CambridgeSoft Cambridge MA USA). The theoretical length (L ) of the C 16 DSPE and DGDG molecules are 2.4 3.3 and 3.7 nm respectively. The tilts (h) of the DSPE and C -OH in this image were set at 25 and 30° respectively to re—ect the measured thickness (T ) of the monolayers. 16 83 Faraday Discuss. 1998 111 79»94 Fig. 3 Force measurements between a C -OH coated AFM probe and (A) C16-OH and (B) C15-CH3 surfaces in water. The force and separation resolution of the two measurements were varied by an order of 16 magnitude in order to capture the hydration force region of the C16-OH surface and the adhesive force region of the C -CH surface. Data in (A) are shown with circular symbols (1 symbol/5 data points) and the theoreti- 15 cal –t is shown with a solid line.3 Supported lipid bilayers Fig. 4 shows the surface pressure vs. area (n-A) isotherms of DSPE DOPE DSPE/DOPE (1 1) DGDG and DGDG/DOPE (1 1) monolayers at the air/water interface. While DSPE and DGDG monolayers show a two-dimensional ì solid-like œ behavior the DOPE monolayers show a 2-D ì liquid-like œ behavior due to the unsaturated hydrocarbon chains of the molecule. The area per molecule surface compression modulus and theoretical space –lling (SF) thickness of pure lipid –lms were determined from the surface pressure diagrams at the deposition pressure of 25 mN m~1 and are summarized in Table 2. Topographic images made of mixed DSPE/DOPE and DGDG/DOPE monolayers on DSPEcoated mica are presented in Fig.5A and 6A respectively with topographic cross-sections. These images are consistent with previous AFM images of saturated lipid/DOPE monolayers50 in that all the bilayers show phase-separation in the form of elevated microscopic domains surrounded by a continuous matrix. In light of the surface fractions covered by the domains in the diÜerent images as well as the molecular lengths calculated from SF models the higher level domains in the images are ascribed to DSPE and DGDG and the lower level domains to DOPE. Two observations can be made from the topographic images. First defects are observed in the continuous DOPE phase in the form of holes associated with the DOPE layer images.50 These holes appar- Fig.4. Surface pressure»area (n»A) isotherms for pure and mixed (1 1) lipid monolayers at the air/water interface at 25 °C (A) DSPE DOPE and DSPE/DOPE mixture; (B) DGDG DOPE and DGDG/DOPE mixture. Faraday Discuss. 1998 111 79»94 84 Table 2 Physical properties of the LB deposited lipid monolayer and bilayer –lms Ellipsometry AFM step height Cauchy parameters Monolayer thickness/ nm SF thickness/ nmb Bilayer/ nme Monolayer/ nmd Lipid A Bc Surface compression modulus/ mN m~1 a Area molecule~1/ nm2 a » » 1.9 0.67 DOPE 1.9^0.1 79^1 3.0 0.41 DSPE 3.3^0.3 1.8^0.2 3.3^0.1 183^1 2.7 0.47 DGDG 4.9^0.4 2.1^0.1 2.9^0.1 193^1 1.865 [0.0866 1.456 [0.00856 1.6045 [0.0416 a At the deposition pressure of 25 mN m~1 of the –lms.b The theoretical space –lling (SF) thicknesses of the monolayers (T ) were determined from the area per molecule (a) of the lipids using a constant speci–c volume (l) i.e. T \l/a. A speci–c volume of 1.03 cm3 g~1 was chosen; the molecular weights of the DGDG DOPE and DSPE lipids are 778 744 and 748 respectively. c The wavelength (j) dependence of the refractive index (n) of the lipid –lms was –tted with a Cauchy form n\A]B/j. d Mixed DSPE/DOPE and DGDG/DOPE monolayers supported on mica were made by raising freshly cleaved mica vertically through the air/water interface and imaged in air at forces of D1 nN. The step heights between DSPE and DGDG domains and the DOPE matrix were measured from cross-sections in the topographic images.Mean value and standard deviation of height diÜerences of three independent experiments. e Step heights measured between the DSPE and DGDG domains and the DOPE matrix for bilayers made of mixed DSPE/ DOPE and DGDG/DOPE monolayers deposited onto DSPE-coated mica and imaged in 0.15 M NaCl solutions. Fig. 5 (A) Topographic (z-range 20 nm) and (B) adhesion images (5 lm]5 lm) of a mixed DSPE/DOPE monolayer onto DSPE-coated mica in a 0.15 M NaCl solution. Lighter levels in the images correspond to higher height and adhesion. The topographic image was obtained at an applied force of \1 nN and the maximum force applied to the sample for adhesion force mapping was D1 nN. A cross-section (C) taken along a line indicated by the arrow is shown beneath the topographic image while the histogram of the magnitude of the adhesion force (D) (4096 events per image) is shown beneath the adhesion map.85 Faraday Discuss. 1998 111 79»94 Fig. 6 (A) Topographic (z-range 20 nm) and (B) adhesion images (5 lm]5 lm) of a mixed DGDG/DOPE monolayer onto DSPE-coated mica in 0.15 M NaCl solution. A cross-section and adhesion map are shown in Fig. (C) and (D) respectively. ently form during or shortly after deposition and appear to be stable for over 24 h periods. Second the step heights measured between the DSPE and DGDG domains and the surrounding DOPE matrix in the topographic images are 3.3^0.3 and 4.9^0.4 nm respectively clearly larger than predicted from the SF thicknesses of the bilayers. The adhesion force maps shown in Fig.5B and 6B exhibit high contrast the DSPE and DGDG domains presenting systematically low levels of adhesion compared to DOPE. Indeed the corresponding adhesion force histograms show two discrete levels of adhesion one centered at about 0.1 nN for DSPE and DGDG the other centered between 0.5 and 2 nN for DOPE (the magnitude of the adhesive forces can not be directly compared between the diÜerent –lms due to the variations in the radius of curvature of the microfabricated probe). The force measured between the DOPE surface and a hydroxy-terminated probe is plotted as a function of surface separation in Fig. 7A. The form of the force curve was not aÜected by the maximum force applied to the surface for forces up to 20 nN. Upon approach the probe jumpsto-contact at 4.5 nm without experiencing repulsive forces.Upon retraction a hysteresis or pulloÜ force is observed of D5 nN magnitude indicating contact and a strong attractive interaction adhesion between the probe and –lm. A set of typical force»distance curves recorded between a modi–ed AFM probe and the DSPE and DGDG surfaces is given in Fig. 7 B»E. The curves clearly diÜer from those obtained on DOPE and depend on the maximum load applied on the surface. At low maximum loads (\1.5 nN) the approach and retraction curves are similar. The lack of signi–cant jump-to-contact suggests that a strong repulsive force balances the attractive interactions between the lipid surface and the probe. Upon retraction no signi–cant adhesion pull-oÜ forces are observed.These two observations suggest that at low loads molecular contact is not established between the probe and surface of the DSPE and DGDG domains. In contrast when the maximum applied force is increased the probe experiences a steep shortrange repulsion upon approach until it jumps into contact when the force is greater than 5 nN. Faraday Discuss. 1998 111 79»94 86 Fig. 7 Typical force»distance curves recorded over the (A) DOPE (B,C) DSPE (D,E) DGDG surfaces in 0.15 M NaCl solutions ; (B,D) Low maximum applied loads (\1.5 nN) (C,E) high maximum applied forces ([1.5 nN). The occurrence and form of the jump-to-contact were found to be probe dependent which is consistent with other measurements of short-range forces with the AFM.53 Upon retraction adhesion forces of D8 and 4 nN magnitude are found for DSPE and DGDG respectively suggesting that the lipid and alkanethiol head groups are in direct contact.The repulsive portion of the approach curves of the DSPE and DGDG can be –tted to an exponential form at low loads with decay lengths of 0.5^0.2 nm and 0.8^0.2 nm respectively. At high loads the data points no longer –t through an exponential function but through a straight line. (1) Discussion Image contrast mechanisms of supported lipid monolayer We have shown that the topographic contrast of mixed fatty acid and lipid monolayers in air is determined by three factors i.e. length of the molecule (L ) tilt of the molecule (h) and the mechanical properties of the –lm.50,54 The step height (*Z) measured between two lipid phases is *ZM\*T ](dM[dM{) where *T is the diÜerence between the thickness of the two –lms hereafter referred to as relative thickness and d is the deformation of –lm under the load of the AFM probe (M and M@ designate the thinner and thicker –lm respectively).The thickness of the pure monolayers has been independently determined using a simple SF model and ellipsometry Table 2. Two trends are evident in the data. First both techniques 87 Faraday Discuss. 1998 111 79»94 produce similar thicknesses second the solid-like DSPE and DGDG –lms are of similar thickness and are signi–cantly thicker that the more loosely packed liquid-like DOPE –lm. Using contact mechanics the elastic deformation produced by the probe under a known load can be determined if the physical properties of the system are known.Given an applied load P and probe radius R the Hertzian deformation is d\(9P2/16RE*2)1@3 (E Table 2) through the monolayer thickness56,57 (2) E*\[(1[v12)/E1](1[v22)/E2] E where 1 v1 and E2 v2 are Youngœs modulus and and Poissonœs ratio of the probe and sample respectively.55 If volume compressibility is neglected the isothermal thickness compression modulus (ET) of a LB monolayer is related to the surface compression modulus a (3) ET\Ea/T Using ellipsometric thicknesses the E of the DOPE DPSE and DGDG monolayers is 4.3 6.1 T and 7.1]107 N m~2 respectively and for soft materials v\0.5. The elastic modulus of the alkanethiol –lm is about 2]1010 N m~2 58 and thus will make a negligible contribution to E*.If one assumes the mechanical properties of the supported monolayer are dominated by the LB –lm the calculated deformation produced by a 50 nm radius probe at 0.5 nN of load in the DOPE DSPE and DGDG –lms is 1.0 0.8 and 0.7 nm respectively. Using the ellipsometric thickness of the –lms and the Hertzian –lm deformations the theoretical AFM step height of the mixed DSPE/DOPE and DGDG/DOPE monolayers in air (*ZM) is 1.6 and 1.3 nm respectively. The DSPE/DOPE monolayer *Z and measured step height (Table M 2) is in reasonable agreement but *Z of the DGDG/DOPE –lm is 0.8 nm smaller than the M measured step height. We attribute the signi–cant diÜerence between the theoretical and the measured DGDG/DOPE step height to a water –lm between the lipid head groups and mica surface.This nanometer scale water –lm is produced by the repulsive hydration»steric interactions between the lipid head groups and surface which make the water –lmœs thickness highly dependent on the form of the lipid head group. Fig. 8A schematically presents how the water –lmœs thickness in—uences the measured step height. The theoretical AFM step height of the monolayers is (4) *ZM\*T ](dM[dM{)](dL{[dL) where d and dL{ is the thickness of the water –lm associated with the thin and thick lipid –lm L respectively. Interpretation of the force»distance curves (see discussion below) indicates that the range of the hydration»steric interaction (dF) of the DGDG DSPE and DOPE phase is B1.2 0.3 and 0 nm respectively. If we assume that the water –lm thickness is equal to the range of the repulsive hydration»steric force *Z for the DSPE/DOPE and DGDG/DOPE monolayers is 1.9 M and 2.5 nm respectively.The theoretical step heights are in good agreement with their respective measured step heights (Table 2). This discussion indicates that the measured step height of mixed lipid monolayers is determined from molecular length molecular tilt mechanical properties and the nature of the lipid head group»surface interaction. The contribution of all these factors to the AFM imaging mechanism makes step height a sensitive probe of the physical state of a bilayer. Image contrast mechanisms of supported lipid bilayers The step heights measured between the DSPE and DGDG domains and the surrounding DOPE for the bilayers in water are much larger than the step heights measured for the corresponding monolayers in air (Table 2).The increase in the bilayer step height is unexpected as no new factors have been introduced into the image contrast mechanism. However comparison of topographic images of the bilayers (Fig. 5A and 6A) and monolayers50 indicates that deposition of the mixed monolayer in water changes its physical state. Speci–cally the formation of holes in the DOPE phase indicates that DOPE desorbs from the mixed monolayer during or shortly after deposition. A mechanism emerges to explain the image contrast of the mixed bilayers from this observation (Fig. 8B). At the low imaging force used in this study (\1 nN) topographic imaging of the DOPE region is conducted with the probe indenting into the –lm.We expect the deformation of Faraday Discuss. 1998 111 79»94 88 Fig. 8 Schematic diagram of the probe position on the lipid monolayer in air (A) and lipid bilayer in water (B). The unsaturated and saturated lipid»probe interaction is presented on the left and right side of the diagrams respectively. d is the deformation of the lipid monolayers d is the thickness of the water layer *T is M L the relative thickness of the –lms d is the distance of the probe from the surface due to repulsive forces and F *Z is the step height measured by the AFM. the DOPE phase by the probe to be signi–cantly larger in water than it was in air as reorganization will produce a signi–cant decrease in elastic modulus of this phase (Fig. 4). Reorganization of the DOPE phase does not appear to change the DSPE and DGDG internal surface pressure as the surface area occupied by these phases does not increase signi–cantly from air to water images.50 Imaging of the DSPE and DGDG regions appears to take place in a short-range repulsive force mode with the probe some distance oÜ the surface of the lipid –lm.Noncontact imaging has also been reported using repulsive long-range double layer interactions.32,42,59 The theoretical step height of the bilayer resulting from this model is (5) *ZB\*T ]dM]dF{ where *T is the relative thickness of the monolayer. Using the ellipsometric thickness of the –lms (d and assuming the probe penetrates through the entire DOPE –lm MB1.9 nm) *ZB for the mixed DSPE/DOPE and DGDG/DOPE bilayers is 3.6 and 4.1 nm respectively.The theoretical and measured step heights are in reasonable agreement considering the assumptions that have been made in estimating d and dF . Accordingly the image contrast mechanism of lipid bilayers M in water is controlled by the same factors that have been identi–ed for monolayers in air but the physical properties of the mixed lipid –lm is signi–cantly changed by deposition onto a DSPE monolayer on the down stroke. Surface force measurements between alkanethiol monolayers 3) Both imaging and interaction force measurements on the mixed lipid bilayers were made with a C -OH functionalized probe. To gain an understanding of surface forces between alkanethiol 16 surfaces in water we made force measurements between C -OH functionalized probes and model 16 hydrophilic (C -OH) and hydrophobic (C -CH surfaces.16 15 Faraday Discuss. 1998 111 79»94 89 Forces measured between amphiphilic surfaces in aqueous solutions typically result from longrange electrostatic double-layer and van der Waals (vdW) interactions and short-range hydration/ steric interactions. The role of electrostatic double-layer interactions was tested by making force measurements on each surface in water and 0.15 M NaCl. Consistent with our expectation that the surfaces were uncharged no diÜerence in the forces could be detected. As a rule vdW forces between surfaces of identical chemistry are attractive and for a sphere on —at geometry (6) FvdW\[AR/6r2 where A is the Hamaker constant R is the radius of curvature of the sphere and r is the distance between surfaces.60 As both probe and sample surfaces are made of hydrocarbon based monolayers which have similar dielectric constants (eD2) we anticipate that the vdW forces will be attractive with Hamaker constants in the range of 4»7]10~21 J.61,62 Hydration/steric forces are short-ranged and have the form (7) o Fhydration\Fo e~(r@j) where F is the preexponential factor r is the separation between the surfaces and j is the hydration decay length.Repulsive hydration/steric interaction clearly dominate the force measured between the C -OH surface and probe Fig. 3A. These forces are believed to be associated with the reorientation 16 of water molecules at surfaces12h14,63 and entropic —uctuation forces arising from the overlap of thermally excited surface modes of molecular vibrations including long-wavelength undulation forces and molecular scale protrusion forces.10,64,65 A reasonable –t of the total force curve (solid line in Fig.3A) is obtained for decay lengths of 0.3 nm a preexponential factor of 2 nN a vdW radius of 100 nm and Hamaker constant of 6]10~21 J if the vdW origin is set 0.6 nm behind the point of zero separation (PZS) of the measured forces (the AFM measures relative displacement and PZS is de–ned as the position of the probe at maximum force). A vdW radius larger than the speci–ed AFM probe was used for the vdW force calculation which is justi–ed due to the pyramidal shape of the probe. Locating the vdW plane behind the PZS suggests that a water layer is bound to the surfaces at the PZS and a distance of 0.6 nm corresponds to two monolayers of water.These results are similar to the SFA and OS measurements between lipid bilayers in which repulsive hydration forces have an exponential form with 0.3 nm decay length.8,9,11h14 Signi–cant hysteresis is observed in the retracting force curve between the C -CH surface 15 3 and C -OH probe Fig. 3B. The strong adhesion between the C -CH surface and C -OH 16 15 3 16 probe appears to result from a decrease in the magnitude of the hydration force between these surfaces and/or an increase in the vdW forces. These results are similar to force measurements made between identical hydrophobic surfaces in that strong adhesion is observed between the surface but diÜer in that the long-range attractive hydrophobic eÜect is not observed.66 Forces between lipid bilayers and alkanethiol-functionalized probes interplay of hydration/steric forces and mechanical eÜects As in the case of alkanethiol monolayers the interaction forces between the C -OH probe and 16 lipid bilayers is expected to result from a balance between repulsive hydration/steric and attractive vdW interactions.At low loads the surface forces on the solid-like DSPE and DGDG –lms appear to be dominated by repulsive hydration/steric interactions Fig. 7B and D. This seems consistent with measurements of surface forces between symmetric lipid –lms with the SFA and OS but on closer examination there are several important diÜerences. First exponential –ts of the repulsive portion of the DSPE and DGDG forces curve result in decay lengths of 0.5^0.2 and 0.8^0.2 nm respectively which are 2»5 times larger than the decay lengths measured with the SFA and OS.8,9,11h14 Second the probe jumps to contact over a distance of 5»8 nm at loads [5 nN Fig.7C and E. If the jump-to-contact were associated with an attractive vdW force this would place the observed repulsive force beyond the range typically associated with hydration/steric forces. The AFM and SFA instruments are based on the same physical principles and use similar geometries i.e. the equivalent of a sphere on a —at thus they should generate similar results. On closer consideration however there is a signi–cant diÜerence in the pressure ranges over which the AFM and SFA typically make measurements.The theoretical Hertzian pressure generated by Faraday Discuss. 1998 111 79»94 90 a D50 nm radius sphere at 1 nN of load is 2]107 Pa. SFA measurements are typically made at pressures less than 105 Pa due to compliance of an epoxy used in the mechanical loop. Direct comparison to OS measurements is complicated by diÜerences in geometry and the isotropic nature of the osmotic force. The high pressures in the AFM suggests that the mechanical properties of the bilayer may contribute to the measured forces. The role of bilayer mechanics is supported by –tting the repulsive portion of the DSPE and DGDG force curve with models that combined hydration/steric forces with mechanical deformation. A reasonable –t of the repulsive portion of the DSPE force curve Fig.9A is achieved with the combined model (8) d\j ln(P/Fo)](9/16RE*2)1@3P2@3 where j\0.35 nm Fo\0.8 nN R\50 nm and E*\1.7]108 N m~2 if mechanical contact (Hertian plane) takes place 0.3 nm after the hydration/steric force can –rst be detected (hydration/ steric plane). The repulsive portion of the DGDG force curves Fig. 9B is also reasonably well –tted with a combined model for j\0.3 nm Fo\0.15 nN R\50 nm and E*\2.1]108 N m~2 if the Hertzian plane is placed 1.2 nm inside the hydration/steric plane. The distance separating the hydration from the vdW origin re—ects the range of the hydration/steric force. The diÜerence in the range over the DSPE and DGDG –lms is directly correlated with diÜerences in the measured step heights thus con–rming the imaging mechanism discussed above.The diÜerent behaviors observed for DSPE and DGDG surfaces may be understood in terms of diÜerences in hydration and steric properties. The fact that the mechanical component of the force is applied to DSPE after only 0.3 nm may be explained by the relatively low degree of hydration of the phosphatidylethanolamine (PE) head group.12,46 This has been attributed to the fact that PEs form a compact lattice that is stabilized by strong hydrogen bonding between the ammonium and phosphate groups.67 In contrast the fact that the mechanical model only becomes relevant at a separation of D1 nm for DGDG is consistent with the mobility of the digalactoside head group and the large number of hydration water molecules associated with head group.67 As the load is increased mechanical deformation of the bilayers by the probe plays an increasingly important role in the measured forces of the bilayer surfaces.Several observations suggest that the jump-to-contact results from a transition in the mechanical behavior of the bilayer from elastic to plastic deformation representing the penetration of the probe through the bilayer to the Fig. 9 Approaching force curve recorded over (A) DSPE and (B) DGDG domains with theoretical –ts + raw data »»»» Hertzian contribution óó hydration/steric forces and »» combined theoretical force curve. The relative separation axis has been set with its origin at the point at which the probe jumps-tocontact with the surface. 91 Faraday Discuss.1998 111 79»94 mica surface. First the distance over which the jump to contact takes place is in reasonable agreement with the bilayer thickness i.e. TDSPE@DSPE\6.6 nm and TDSPE@DGDG\6.2 nm. Second the strong adhesive forces between the probe and surface after jump-to-contact are consistent with the forces measured between the hydrophilic and hydrophobic surfaces (Fig. 3B). Third nanoindentation studies of thick LB –lms suggest the hardness of the –lm to be on the order of 108 N m~2.68 If we assume the yield pressure of the –lms is equivalent to hardness the transition from elastic to plastic deformation should take place at loads [5 nN. A similar conclusion was reached by Ducker and Clarke69 for self-assembled zwitterionic surfactant monolayers in water.Accordingly these results lead us to conclude that on DSPE and DGDG the probe experiences repulsive forces that are a combination of hydration/steric forces and mechanics. Summary Although the surface properties and interaction forces of lipid bilayers have been investigated for many years direct information at high lateral resolution has been inaccessible up to now. This study demonstrates that chemically functionalized AFM probes can be used to characterize the surface properties (surface morphology surface forces and mechanics) of heterogeneous lipid bilayers on the nanometer scale. Short-range repulsive forces are detected between the hydroxy-terminated probe and the 2-D solid-like DSPE and DGDG domains. The ranges and exponential decay lengths of the forces are larger than those found so far between amphiphilic bilayers using the OS and SFA techniques which may be attributed to the fact that mechanical properties of surfaces are sampled at the high loads that can be achieved with the AFM.This highlights the important role that mechanics may play in biomolecular interactions. The direct correlation between the range of the repulsive force over the two solid lipid domains and the step heights in the topographic images validates the quantitative analysis of the measured forces and sheds new light on the fundamental mechanisms of AFM imaging of lipid bilayers in aqueous solutions. Surprisingly no repulsive force is found over the 2-D liquid-like DOPE region which is related to the low packing density of this phase.Taken together these results provide direct evidence for the in—uence of the structure of lipid bilayers on their interaction forces a behavior that may be of prime importance in the control of cellular interactions. The methodology developed here has promising applications for the nanometer scale mapping of the properties (mechanical properties and molecular interactions) of biosurfaces and the study of surface forces in complex asymmetric systems. Cell surface components such as gangliosides integrins and lectins can be incorporated into a patterned bilayer and characterized with AFM probes functionalized with relevant (bio)chemical moieties. This would greatly improve our understanding of the molecular mechanisms of cellular processes such as cell adhesion membrane fusion molecular recognition and intercellular communication.Acknowledgements This research was supported by the Office of Naval Research (ONR) and by a NATO Research Fellowship (Y.F.D.). We also thank M. Fletcher David L. Allara and R.J. Colton and all members of NRL Code 6177 for valuable discussion M.L. Stevens and J.-B. Green for calibration of the AFM M.D. Porter for providing some of the alkane thiols M. Tortonese for supplying micromachined reference cantilevers N.H. Turner for performing XPS analysis K.Lee for programming associated with the adhesion images and J. Callahan and M. Shahgholi for electrospray mass spectroscopy analysis. References 1 R. O. Hynes Cell 1992 69 11. 2 A. Varki Proc. Natl. Acad. Sci. USA 1994 91 7390.3 N. Sharon and H. Lis Science 1989 246 227. 4 W. Curatolo Biochim. Biophys. Acta. 1987 906 137. Faraday Discuss. 1998 111 79»94 92 5 N. Sharon and H. Lis Sci. Am. 1993 January 74. 6 S-I. Hakomori Prog. Brain Res. 1994 101 241. 7 R. G. Horn Biochim. Biophys. Acta 1984 778 224. 8 J. Marra J. Colloid Interface Sci. 1985 107 446. 9 J. Marra and J. Israelachvili Biochemistry 1985 24 4608. 10 J. Israelachvili and H. Wennerstroé m Nature (L ondon) 1996 379 219. 11 D. M. LeNeveu R. P. Rand and V. A. Parsegian Nature (L ondon) 1976 259 601. 12 R. P. Rand and V. A. Parsegian Biochim. Biophys. Acta 1989 988 351. 13 S. Leikin V. A. Parsegian D. C. Rau and R. P. Rand Annu. Rev. Phys. Chem. 1993 44 369. 14 T. J. McIntosh and S. A. Simon Annu. Rev.Biophys. Biomol. Struct. 1994 23 27. 15 G. Binnig C. F. Quate and Ch. Gerber Phys. Rev. L ett. 1986 56 930. 16 A. L. Weisenhorn B. Drake C. B. Prater S. A. C. Gould P. K. Hansma F. Ohnesorge M. Egger S.-P. Heyn and H. E. Gaub Biophys. J. 1990 58 1251. 17 H. G. Hansma M. Bezanilla F. Zenhausern M. Adrian R. L. Sinsheimer Nucleic Acids Res. 1993 21 505. 18 M. Fritz M. Radmacher and H. E. Gaub Biophys J. 1994 66 1328. 19 M. Radmacher R. W. Tillmann M. Fritz and H. E. Gaub Science 1992 257 1900. 20 H. G. Hansma and J. H. Hoh Annu. Rev. Biophys. Biomol. Struct. 1994 23 115. 21 H-J. Butt K. H. Downing and P. K. Hansma Biophys. J. 1990 58 1473. 22 J. A. N. Zasadzinski C. A. Helm M. L. Longo A. L. Weisenhorn S. A. C. Gould and P. K. Hansma Biophys. J. 1991 59 755.23 J. H. Hoh G. E. Sosinksky J-P. Revel and P. K. Hansma Biophys. J. 1993 65 149. 24 S. W. Hui R. Viswanathan J. A. Zasadzinski and J. N. Israelachvili Biophys. J. 1995 68 171. 25 H-J. Butt E. K. WolÜ S. A. C. Gould B. Dixon-Northern C. M. Peterson and P. K. Hansma J. Struct. Biol. 1990 105 54. 26 J. H. Hoh and C-A. Schoenenberger J. Cell Sci. 1994 107 1105. 27 L. K. Tamm and H. M. McConnell Biophys. J. 1985 47 105. 28 E. Sackmann Science 1996 271 43. 29 J. Mou J. Yang and Z. Shao J. Mol. Biol. 1995 248 507. 30 W. A. Ducker T. J. Senden and R. M. Pashley Nature (L ondon) 1991 353 239. 31 H-J. Butt Biophys. J. 1991 60 1438. 32 T. J. Senden C. J. Drummond and P. Keç kicheÜ L angmuir 1994 10 358. 33 S. J. OœShea M. E. Welland and T. Rayment Appl.Phys. L ett. 1992 60 2356. 34 J. P. Cleveland T. E. Schaé Üer and P. K. Hansma Phys. Rev. B 1995 52 R8692. 35 S. Biggs L angmuir 1995 11 156. 36 R. M. Overney D. P. Leta C. F. Pietroski M. H. Rafailovich Y. Liu J. Quinn J. Sokolov A. Eisenberg and G. Overney Phys. Rev. L ett. 1996 76 1272. 37 G. U. Lee D. A. Kidwell and R. J. Colton L angmuir 1994 10 354. 38 E-L. Florin V. T. Moy and H. E. Gaub Science 1994 264 415. 39 G. U. Lee L. A. Chrisey and R. J. Colton Science 1994 266 771. 40 D. R. Baselt and J. D. Baldeschwieler J. Appl. Phys. 1994 76 33. 41 M. Ludwig W. Dettmann and H. E. Gaub Biophys. J. 1997 72 445. 42 C. Rotsch and M. Radmacher L angmuir 1997 13 2825. 43 G. U. Lee L. A. Chrisey C. E. OœFerrall D. E. PilloÜ N. H. Turner and R. J.Colton Isr. J. Chem. 1996 36 81. 44 C. D. Frisbie L. F. Rozsnyai A. Noy M. S. Wrighton and C. M. Lieber Science 1994 265 2071. 45 J-B. D. Green M. T. McDermott and M. D. Porter J. Phys. Chem. 1995 99 10960. 46 J. Marra J. Colloid Interface Sci. 1986 109 11. 47 R. G. Nuzzo and D. L. Allara J. Am. Chem. Soc. 1983 105 4481. 48 C. D. Bain E. B. Troughton Y-T. Tao J. Evall G. M. Whitesides and R. G. Nuzzo J. Am. Chem. Soc. 1989 111 331. 49 R. G. Nuzzo L. H. Dubois and D. L. Allara J. Am. Chem. Soc. 1990 112 558. 50 Y. F. Dufre� ne W. R. Barger J-B. D. Green and G. U. Lee L angmuir 1997 13 4779. 51 M. Tortonese and M. Kirk SPIE 1997 3009 53. 52 H. G. Tomkins A Userœs Guide to Ellipsometry Academic Press Inc. San Diego 1993. 53 S. J. OœShea M. E. Welland and T.Rayment Appl. Phys. L ett. 1992 61 2240. 54 D. D. Koleske W. R. Barger G. U. Lee and R. J. Colton Mat. Res. Soc. Symp. Proc. 1997 464 377. 55 K. L. Johnson Contact Mechanics Cambridge University Press Cambridge 1985. 56 E. A Evans and R. M. Hochmuth in Current T opics in Membranes and T ransport ed. A. Kleinzeller and F. Bronner Academic Press New York 1978 pp. 1»10. 57 E. A. Evans and P. Kwok Biochemistry 1982 21 4874. 58 K. J. Tupper and D. W. Brenner L angmuir 1994 10 2335. 59 S. Manne J. P. Cleveland H. E. Gaub G. D. Stucky and P. K. Hansma L angmuir 1994 10 4409. 60 S. Nir Prog. Surf. Sci. 1976 8 1. 61 D. Gingell and V. A. Parsegian J. T heor. Biol. 1972 36 41. 93 Faraday Discuss. 1998 111 79»94 62 D. B. Hough and L. R. White Adv. Colloid Interface Sci. 1980 14 3. 63 S. Marcelja and N. Radic Chem. Phys. L ett. 1976 42 129. 64 J. N. Israelachvili and H. Wennerstroé m L angmuir 1990 6 873. 65 J. N. Israelachvili and H. Wennerstroé m J. Phys. Chem. 1992 96 520. 66 J. N. Israelachvili and R. M. Pashley Nature (L ondon) 1982 300 341. 67 J. M. Boggs Biochim. Biophys. Acta 1987 906 353. 68 T. P. Weihs Z. Nawaz S. P. Jarvis and J. B. Pethica Appl. Phys. L ett. 1991 59 3536. 69 W. A. Ducker and D. R. Clarke Colloids Surf. A 1994 93 275. Paper 8/07637G Faraday Discuss. 1998 111 79»94