Repulsion and Attraction by Extracellular Matrix Protein in Cell

of dangling laminin molecules, we estimate a thickness of 110 nm from sedimentation of giant lipid ... We propose that in cell adhesion the laminin cu...
0 downloads 0 Views 213KB Size
1580

Langmuir 2003, 19, 1580-1585

Repulsion and Attraction by Extracellular Matrix Protein in Cell Adhesion Studied with Nerve Cells and Lipid Vesicles on Silicon Chips† Gu¨nther Zeck and Peter Fromherz* Department of Membrane and Neurophysics, Max Planck Institute for Biochemistry, D 82152 Martinsried, Mu¨ nchen, Germany Received July 30, 2002. In Final Form: October 24, 2002 The separation of membrane and substrate in cell adhesion is addressed, in particular the role of extracellular matrix proteins. Using fluorescence interference contrast microscopy, we measure a distance of 90 nm between a neuron membrane and oxidized silicon coated with laminin. For the glycocalix, we obtain 40 nm from the adhesion of neurons on polylysine. We propose that dangling laminin molecules contribute 50 nm to the total cell-solid distance by their repulsive steric force. For an unperturbed cushion of dangling laminin molecules, we estimate a thickness of 110 nm from sedimentation of giant lipid vesicles, after subtraction of membrane undulations. We propose that in cell adhesion the laminin cushion is compressed by the adhesive forces between laminin molecules and integrin receptors.

Introduction The separation of electrically active cells and semiconductors is a crucial parameter for the development of cellular bioelectronic devices. It determines the transductive extracellular potential between cell and semiconductor that arises from current flow along the sheet resistance of the junction and that controls electronic devices in the substrate or voltage-dependent processes in the cell, respectively.1 In previous studies, it was shown that neurons are separated from silica coated with extracellular matrix proteins such as fibronectin and laminin by 50-100 nm.2 A rationalization of this separation in terms of the contributing molecules and forces is a first step toward an optimization of cell-semiconductor hybrids. Cell adhesion is governed by the competition of specific bonding and nonspecific repulsion between cell and substrate.3,4 Steric repulsion may be caused by polymer molecules in the glycocalix of the cell membrane and also in the extracellular matrix as illustrated in Figure 1. Bonding is mediated by the interaction of integrin receptors with extracellular matrix proteins.5-9 The equilibrium of repulsion and attraction determines the separation of cell and substrate.3,10 Vice versa, the distance between cell and substrate reflects the balance of forces. Experimental data on the contribution of extracellular matrix proteins to the cell-substrate distance are not available. In the present paper, we study the role of laminin in the adhesion of neurons from the snail Lymnaea stagnalis * Corresponding author. E-mail: [email protected]. † Part of the Langmuir special issue entitled The Biomolecular Interface. (1) Fromherz, P. ChemPhysChem 2002, 3, 276-284. (2) Braun, D.; Fromherz, P. Phys. Rev. Lett. 1998, 81, 5241-5244. (3) Bell, G. J.; Dembo, M.; Bongrand, P. Biophys. J. 1984, 45, 10511064. (4) Sackmann, E.; Bruinsma, R. ChemPhysChem 2002, 3, 262-269. (5) Timpl, R.; Brown, J. C. BioEssays 1996, 18, 123-132. (6) Yamada, K. M.; Geiger, B. Curr. Opin. Cell Biol. 1997, 9, 76-85. (7) Aplin, A. E.; Howe, A.; Alahari, S. K.; Juliano, R. L. Pharmacol. Rev. 1998, 50, 197-263. (8) Burke, R. D. Int. Rev. Cytol. 1999, 191, 257-284. (9) Giancotti, F. G.; Ruoslahti, E. Science 1999, 285, 1028-1032. (10) Lipowsky, R. Phys. Rev. Lett. 1996, 77, 1652-1655.

Figure 1. Schematic cross section of cell adhesion on a silicon substrate. The blowup illustrates how the lipid core of the cell membrane is separated from the silicon dioxide layer on silicon by the glycocalix and by extracellular matrix protein. The distance dcleft between the lipid bilayer of the cell membrane and the silicon dioxide is measured by FLIC microscopy.

which are used for the assembly of neuroelectronic systems.11-13 We also sediment giant lipid vesicles on laminin and compare the separation of the vesicle membrane and silicon dioxide with the distance of a nerve cell from the same substrate. Additional distance measurements are made with neurons and with giant vesicles on polylysine with neuronal growth factors and with giant lipid vesicles on albumin. From these results, we estimate the contribution of extracellular matrix protein to the total separation between neuron and substrate, and we propose a compression of the laminin cushion by attractive force in cell adhesion. The distance between the substrate and the lipid bilayer of a vesicle and the lipid bilayer of a cell membrane is measured by fluorescence interference contrast (FLIC) microscopy.14-16 That method depends little on ill-known optical properties of the membrane (thickness, refractive index)17 compared to other optical (11) Prinz, A. A.; Fromherz, P. Biol. Cybern. 2000, 82, L1-L5. (12) Jenkner, M.; Mu¨ller, B.; Fromherz, P. Biol. Cybern. 2001, 84, 239-249. (13) Zeck, G.; Fromherz, P. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 10457-10462. (14) Lambacher, A.; Fromherz, P. Appl. Phys. A 1996, 63, 207-216. (15) Braun, D.; Fromherz, P. Appl. Phys. A 1997, 65, 341-348. (16) Lambacher, A.; Fromherz, P. J. Opt. Soc. Am. B 2002, 19, 14351453.

10.1021/la0263209 CCC: $25.00 © 2003 American Chemical Society Published on Web 12/18/2002

Repulsion and Attraction in Cell Adhesion

approaches, such as reflection interference contrast (RIC) microscopy18-20 and total internal reflection fluorescence (TIRF) microscopy.21-23 Materials and Methods Silicon Substrate. We fabricate silicon chips (100 surface) covered by silicon dioxide.14,15 The surface consists of unit cells with a size of 10 µm × 10 µm made of four quadratic oxide terraces as required for FLIC microscopy.15 The height of the terraces is about dox ) 10, 50, 100, and 140 nm measured by ellipsometry on larger reference areas with an accuracy of 0.3 nm. The chips are cleaned for 10 min in a freshly prepared mixture of 30% hydrogen peroxide and 96% sulfuric acid (volume ratio 1:2) and extensively rinsed with Milli-Q water. Coatings. We use three different coatings: (i) Laminin. The chips are made hydrophobic by applying 2% dimethyldichlorosilane in toluene for 15 min, rinsed three times with toluene, and dried with nitrogen. Mouse laminin (Sigma) is adsorbed from a 10 µg/mL solution in phosphate buffer (pH 7) for 12 h at 4 °C and rinsed with buffer. At a temperature of 4 °C, the aggregation of laminin molecules (MW ) 85 kD) in solution is negligible.24 We determine the quantity of the adsorbed laminin on a large hydrophobic wafer (156 cm2) after resolubilization with 1% sodium dodecyl sulfate (SDS) using the Lowry type BCA test25 with respect to a calibration curve with laminin. The resulting mass density of 0.5 ( 0.1 µg/cm2 corresponds to a compact protein film of 3.5 nm thickness, assuming a specific volume of 0.7 cm3/g.26 That value is confirmed by an ellipsometric thickness of 3.5 nm for the dried laminin film using a refractive index of 1.5. (ii) Albumin. The chips are made hydrophobic as for the laminin coating. A solution of bovine serum albumin (BSA) (Sigma) with 2 mg/mL in phosphate buffer (pH 7) is applied for 12 h at 20 °C and rinsed with buffer. The ellipsometric thickness of the dried adsorbed albumin layer is 2.6 nm using a refractive index of 1.5. (iii) Polylysine. Hydrophilic chips are incubated in a 1 mg/mL solution of polylysine (MW ) 10 kD, P6516, Sigma) in TRIS/HCl buffer (pH 8.4) at 20 °C for 4 h. Stained Vesicles. Giant lipid vesicles are prepared by electroswelling.27,28 Palmitoyl-oleoyl-phosphatidyl-choline (Avanti Polar Lipids, Alabaster, AL), cholesterol (Sigma), and the lipoid cyanine dye DiIC18 (Molecular Probes, Eugene, OR) are dissolved at a molar ratio of 80:20:1 in chloroform and methanol (volume ratio, 2:1). One microliter of the 2 mM solution is applied to a glass slide with two ITO (indium-tin oxide) electrodes separated by 0.5 mm and insulated by 100 nm quartz (Kottig and Fromherz, in preparation). After drying, a silicone chamber is attached and 10 mM TRIS/HCl buffer (pH 7.2) with 300 mM saccharose (Merck, Darmstadt, Germany) is added. An ac voltage of 10 Hz is applied to the electrodes for 2 h with an increasing amplitude from 0 to 0.6 V within the first 40 min. The vesicles are dissociated by an ac voltage of 4 Hz and 0.8-1 V. The vesicles are transferred to a Petri dish (Falcon 3001, Becton Dickinson, Plymouth) with a silicon chip containing 300 mM glucose (Merck) with 10 mM TRIS/HCl. They sediment within a few minutes due to the high density of the saccharose solution. The increment of density ∆F (17) Iwanaga, Y.; Braun, D.; Fromherz, P. Eur. Biophys. J. 2001, 30, 17-26. (18) Curtis, A. S. J. Cell Biol. 1964, 20, 199-215. (19) Izzard, C. S.; Lochner, L. R. J. Cell Sci. 1976, 21, 129-159. (20) Lanni, F.; Waggoner, A. S.; Taylor, D. L. J. Cell Biol. 1985, 100, 1091-1102. (21) Bailey, J.; Gingell, D. J. Cell Sci. 1988, 90, 215-224. (22) Truskey, G. A.; Burmeister, J. S.; Grapa, E.; Reichert, W. M. J. Cell Sci. 1992, 103, 491-499. (23) Burmeister, J. S.; Truskey, G. A.; Reichert, W. M. J. Microsc. 1994, 173, 39-51. (24) Yurchenco, P. D.; Tsilibary; E. C.; Charonis, A. S.; Furthmayr, H. J. Biol. Chem. 1985, 260, 7636-7644. (25) Smith, P. K.; et al. Anal. Biochem. 1985, 150, 76-85. (26) Tanford, C. Physical Chemistry of Macromolecules; Wiley: New York, 1961. (27) Angelova, M. I.; Dimitrov, D. S. Faraday Discuss. Chem. Soc. 1986, 81, 303-311. (28) Mathivet, L.; Cribier, S.; Devaux, P. F. Biophys. J. 1996, 70, 1112-1121.

Langmuir, Vol. 19, No. 5, 2003 1581 ) 19.1 kg/m3 is ∆F ) (0.1319cmax - 0.0682cgluc) kg/m3 mM29 with csacch ) 300 mM and cgluc ) 300 mM. Stained Neurons. The preparation and culturing of neurons from the pond snail L. stagnalis follows ref 30. Animals are dissected under sterile conditions. Central ganglia are isolated and treated with trypsin (Sigma, T8253) (0.67 mg/mL) and collagenase (Boehringer Mannheim) (1.33 mg/mL) for 30 min, followed by application of trypsin inhibitor (Sigma, T9003) (0.67 mg/mL) for 15 min. The ganglia are incubated in defined medium (DM) (PAN Systems, Aidenbach, Germany) with enhanced osmolarity (30 mM glucose). The DM consists of (in mM) NaCl 40.0, KCl 1.7, CaCl2 4.1, MgCl2 1.5, Glutamin 1.0, Hepes 10.0 (pH 7.9), and all other ingredients of Leibovitz L15 medium at half of the standard concentrations, with 20 µg/mL gentamycin (Sigma, G3632). The ganglia are pinned on a dissection dish and opened with a tungsten microneedle. Neurons from the A cluster of the pedal ganglia are removed by suction through a firepolished, silanized micropipet. Neurons are placed on two kinds of silicon chips: (i) Hydrophilic Chips Coated with Polylysine. Three complete snail brains are added to 1200 µL of DM to condition the medium with growthpromoting factors that partially adsorb to the surface.31 (ii) Hydrophobic Chips Coated with Laminin. DM (1200 µL) without conditioning factors is used. In both cases, the neurons are incubated at 20 °C for 2-3 days. Then 30 µL of a 2 mM ethanolic solution of the cyanine dye DiIC18 (Molecular Probes) is added. The staining solution is replaced by fresh medium after 15 min. FLIC Microscopy. We use fluorescence interference contrast microscopy to measure the distance dcleft between oxidized silicon and the lipid bilayer of vesicles and the cell membrane in the adhesion region, respectively. The fluorescence of vesicles and cells on the oxide terraces is observed in a microscope through a water immersion objective (100×; numerical aperture, 1.0) (Zeiss Axioskop) using a mercury high-pressure lamp for excitation (546 nm) and a CCD camera for detection (580-640 nm).15 The illumination time is 320 ms for lipid vesicles and 40 ms for nerve cells. The oxide terraces of different heights are identified by illumination with white light on the basis of their interference colors of reflection.15,17 We plot the fluorescence intensity versus the height dox of the terraces and fit the data with a function J ˜ fl according to eq 1 with three parameters, a scaling factor a, a background factor b, and the optical distance ncleftdcleft between membrane and oxide with the refractive index ncleft.

J ˜ fl ) aJfl(dox, dcleft) + b

(1)

The function Jfl(dox, dcleft) is given by the electromagnetic theory of dipole radiation in a layered optical system.16 It takes into account multiple reflection of the illuminating and the emitted light in the cell-silicon system, the finite aperture of the microscope, broad band detection, and nearfield effects. For the distances in the present investigation, however, nearfield effects play no role and are neglected. The transition moment of the dye is in the membrane plane. For the vesicles, the dye molecules are assigned to both membrane surfaces; with neurons only the outer membrane surface is stained. For the cleft, we use the refractive index of water ncleft ≈ nwater ) 1.33. With a volume fraction xprotein of protein and a refractive index nprotein ) 1.5, we estimate ncleft ) xproteinnprotein + (1 - x)nwater. Thus the maximum error of the thickness dcleft would be 10% if the cleft were completely filled with protein. To illustrate the method, the normalized function J ˜ fl(dox) is plotted in Figure 2 as a function of the height of oxide terraces assuming background b ) 0. For a vanishing distance dcleft ) 0 between membrane and chip, the intensity has a minimum at (29) Handbook of Chemistry and Physics; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 1996. (30) Syed, N. I. In Modern Techniques in Neuroscience Research; Windhorst, U., Johansson, H., Eds.; Springer: Berlin, 1999; pp 361377. (31) Wong, R. G.; Hadley, R. D.; Kater, S. B.; Hauser, G. C. J. Neurosci. 1981, 1, 1008-1021.

1582

Langmuir, Vol. 19, No. 5, 2003

Zeck and Fromherz

The checkerboard pattern of fluorescence is surrounded by stationary fringes where the stained membrane bends upward. These fluorescence fringes correspond to Newton fringes observed by RIC microscopy.33 We evaluate the fluorescence profile around the area of adhesion as indicated in Figure 4a and plot it versus the distance in Figure 4b. The maxima and minima of the experimental intensity are identified with the theoretical relation J ˜ fl(d) given by eq 1 where a and b are known from the distance measurement in the adhesion region. The resulting distance profile d(x) is plotted in Figure 4c. It is fitted according to eq 234 with a limiting angle θ ) 0.12 and a length constant λ ) 1.2 µm.

d(x) ) dcleft + θ[x - λ + λ exp(-x/λ)]

Figure 2. Theoretical fluorescence intensity (normalized) of the cyanine dye DiI in a lipid bilayer on oxidized silicon versus the thickness dox of the oxide (ref 16). Four plots for different distances dcleft between membrane and oxide are shown. Illumination at 546 nm, fluorescence detection between 580 and 640 nm, numerical aperture NA ) 1.0. dox ) 0. An increasing width of the cleft with dcleft ) 50, 100, and 150 nm displaces the fluorescence modulation along the dox axis. Adhesion Profiles. We also use FLIC microscopy to determine the profile of giant vesicles in the detached region near the edge of the adhesion area. There a profile of fluorescence intensity is observed with maxima and minima. The experimental data are compared with the theoretical function J ˜ fl(dcleft) with given values of dox, a, and b. An identification of the maxima, of the minima, and of values in between leads to a distance profile d(x).

Results and Discussion In a first step, we probe the repulsive effect of adsorbed laminin on silicon chips by sedimenting lipid vesicles. We begin with a careful check of the contribution of membrane undulation by sedimenting vesicles on adsorbed albumin. In a second step, we estimate the effect of the glycocalix for nerve cells on a substrate coated with polylysine and growth-promoting proteins. The thickness of the protein coat is obtained from the strong attachment of giant vesicles to that substrate. Finally, we measure the distance of neurons grown on a chip coated with laminin. Vesicles on Albumin. Because we use giant vesicles as a probe for the laminin cushion, we must know the role of membrane undulation with respect to a noninteracting solid substrate. For that purpose, we adsorb albumin to oxidized silicon and sediment giant lipid vesicles in analogy to the study by Ra¨dler and Sackmann.32 On a FLIC substrate, we observe a checkerboard pattern of fluorescence as shown in Figure 3a. Fluctuations of the fluorescence intensity are observed due to membrane undulations. Inhomogeneities are visible on some terraces even after averaging for 320 ms. The observed bright stripes mark the transition regions from one terrace to another. The average intensity in the center of the terraces is intermediate on the thinnest oxide no. 1, brightest on no. 2, intermediate again on no. 3, and darkest on no. 4 as plotted in Figure 3c. We fit the data with eq 1 and obtain an average distance dcleft ) 55 ( 1.5 nm. The mean value for nine vesicles is d h cleft ) 51 ( 2 nm. This result is in good agreement with a RIC microscopy study.32 (32) Ra¨dler, J. O.; Feder, T. J.; Strey, H. H.; Sackmann, E. Phys. Rev. E 1995, 51, 4526-4536.

(2)

When we describe the cross section of a vesicle as a spherical cap, the limiting angle is the ratio of the radius of the adhesion area and the radius of the vesicle θ ) Radh/Rves.35 Radh/Rves ) 0.12 is in good agreement with the actual shape of the vesicle as obtained from the FLIC image (Radh) and from the fluorescence image of the vesicle midplane (Rves). The length constant is given by the ratio of the elastic bending modulus and the membrane tension with λ2 ) κel/Σ.34 With λ ) 1.2 µm and κel ) 1 × 10-19 J for lecithin with 20% cholesterol,36,37 the tension is Σ ≈ 8 × 10-8 J/m2, a value that is fairly low32 but compatible with theoretical arguments.35 The average separation dcleft in the adhesion area is determined by the minimum of the interaction potential per unit area V ) Vundu + Vgrav due to membrane undulation and gravity. Van der Waals attraction plays no role: for distances dcleft > 10 nm its static part is shielded by the 10 mM electrolyte with a Debye length κD-1 ) 3 nm, and its dynamic part disappears due to retardation.38 The gravitation potential Vgrav of a vesicle with a volume Vves ) 4πRves3/3 and a contact area Aadh ) 4πRadh2 is given by eq 3 with g ) 9.8 m/s2. The undulation potential Vundu is given by eq 4 (kBT thermal energy) where we neglect the contribution of the tension Σ due to its low value.39,40

Vgrav )

4g∆FRves dcleft 3θ2

Vundu )

3(kBT)2 2π2κeldcleft2

(3)

(4)

From the minimum of V ) Vundu + Vgrav with κel ) 1 × 10-19 J, Rves ) 50 µm, and θ ) 0.1, we obtain a distance dcleft ≈ 50 nm, in good agreement with the experimental result. The experiment on albumin and its comparison with the undulation theory for an inert surface provide the characterization of the giant vesicles to be used as steric probes for adsorbed laminin on a chip. Vesicles on Laminin. When we sediment a giant lipid vesicle on a FLIC chip coated with laminin (3.5 nm thick after drying), we again observe a fluctuating checkerboard pattern of fluorescence. However, the averaged pattern (33) Albersdo¨rfer, A.; Feder, T.; Sackmann, E. Biophys. J. 1997, 73, 245-257. (34) Bruinsma, R. In Physics of Biomaterials; Riste, T., Sherington, D., Eds.; Kluwer: Dordrecht, 1996; p 61. (35) Seifert, U. Phys. Rev. Lett. 1995, 74, 5060-5063. (36) Evans, E.; Rawicz, W. Phys. Rev. Lett. 1990, 64, 2094-2097. (37) Needham, D.; Nunn, R. S. Biophys. J. 1990, 58, 997-1009. (38) Mahanty, J.; Ninham, B. W. Dispersion Forces; Academic Press: London, 1976. (39) Helfrich, W. Z. Naturforsch. C 1978, 33, 305-315. (40) Seifert, U. Adv. Phys. 1997, 46, 13-137.

Repulsion and Attraction in Cell Adhesion

Langmuir, Vol. 19, No. 5, 2003 1583

Figure 3. FLIC microscopy of lipid vesicles on silicon with 5 µm × 5 µm terraces of silicon dioxide. (a) Micrograph of a vesicle on a hydrophobic chip coated with albumin. The illumination time was 320 ms. A unit cell of four terraces (10 µm × 10 µm) is indicated by a frame. The assignment of the terraces with increasing height (about 10, 50, 100, and 140 nm) is shown in the insert. (b) Micrograph of a hydrophobic chip coated with laminin with a frame indicating a unit cell. The illumination time was 320 ms. (c) Fluorescence intensity versus thickness of silicon dioxide for the vesicles on albumin and laminin. The data are fitted by the FLIC theory with a distance of membrane and chip of dcleft ) 55 ( 1.5 nm on albumin and dcleft ) 177 ( 1 nm on laminin.

Figure 4. Vesicle profile on albumin and laminin. (a) Micrograph of a vesicle on albumin with a white arrow indicating the evaluated profile. (b) Experimental fluorescence intensity (drawn line) along profile coordinate x indicated in the lower abscissa. Theoretical fluorescence intensity (dashed line) versus distance dcleft of membrane and chip (upper abscissa). (c) Profile d(x) obtained from a comparison of the experimental and theoretical intensity. A theoretical relation is drawn through the data choosing a limiting angle θ ) 0.12 and a length constant λ ) 1.2 µm. (d), (e), and (f) describe the same kind of data for a vesicle on laminin with a limiting angle θ ) 0.09 and a length constant λ ) 1.2 µm.

as shown in Figure 3b is quite different from that on a chip with albumin. The average fluorescence is intermediate on the thinnest oxide no. 1, lowest on no. 2, intermediate again on no. 3, and brightest on oxide no. 4. Dark stripes mark the transition regions. The average intensity is plotted in Figure 3c versus the thickness of the oxide. From a fit of the data according to eq 1, we obtain an average distance dcleft ) 177 ( 1 nm. The mean

value from 14 vesicles is d h cleft ) 160 ( 12 nm. Thus, with laminin the average distance of bilayer and chip is enhanced by about 110 nm as compared to albumin. On laminin, the checkerboard pattern is surrounded by stationary fringes of fluorescence that are rather similar to those seen on the albumin substrate. We evaluate the profile d(x) of the vesicle as discussed above (Figure 4b,c). We fit it by eq 4 choosing θ ) 0.09 and λ ) 1.2 µm. The limiting angle θ ) Radh/Rves is in good agreement with the shape of the vesicle, and the length constant for the vesicles on laminin is similar to that for vesicles on albumin. Laminin Cushion. The enhancement of the distance by 110 nm with laminin compared to albumin indicates an additional repulsion. It may be due to enhanced membrane undulations induced by molecular interactions of laminin and lipid or to a repulsive effect of laminin. Let us assume that the enhanced distance is due to an enhanced undulation force. When we insert the experimental distance d h cleft ) 160 nm into the potential V ) Vundu + Vgrav, we obtain an elastic modulus κel ≈ 0.05 × 10-19 J. That value is by a factor of 20 lower than usual. Moreover, an unchanged ratio κel/Σ ) λ2 of elastic modulus and tension would indicate that the tension is also lowered by a factor of 20. We see no physical basis for such a concerted effect of laminin on the elastic modulus and on the tension of a lipid bilayer. Instead, we propose that the enhanced distance on laminin is due to a cushion with a thickness of dlam ) 110 nm in 10 mM TRIS/HCl that hinders further sedimentation of the vesicle. We assume that an unchanged undulation potential of the membrane is effective with respect to the surface of that cushion and attribute the enhanced distance dcleft ) 160 nm to a renormalization of the undulation potential with Vundu(dcleft) f Vundu(dcleft dlam).41 Laminin is a filamentous cruciform molecule with a long arm of 77 nm and three short arms of 48, 34, and 34 nm.42 The thickness of the arms is 1-3 nm. In contact with electrolyte, a cushion of dangling molecules may be formed. The fraction of occupied volume herein is about 3% given by the ratio of the 3.5 nm of the dry protein layer and the effective thickness of 110 nm. A similar effect cannot appear with albumin which is a small compact ellipsoidal molecule.43 The sedimentation of a giant lipid vesicle provides a method to probe the effective thickness of the layer of dangling laminin molecules. The detailed conformation of the molecules is unknown. Considering the shape of (41) Castro-Roman, F.; Porte, G.; Ligoure, C. Phys. Rev. Lett. 1999, 82, 109-112. (42) Beck, K.; Hunter, I.; Engel, J. FASEB J. 1990, 4, 148-160. (43) Matsumoto, T.; Inoue, H. Chem. Phys. 1993, 178, 591-598.

1584

Langmuir, Vol. 19, No. 5, 2003

Zeck and Fromherz

Figure 5. FLIC microscopy of a stained giant lipid vesicle and of a snail neuron on a silicon chip coated with polylysine and brain protein on 5 µm × 5 µm terraces of silicon dioxide. (a) Micrograph of the adhesion region of a vesicle with a frame indicating a unit cell of the terraces. The illumination time was 40 ms. (b) Part of the adhesion region of a neuron. The illumination time was 40 ms. (c) Fluorescence intensity versus thickness of silicon dioxide for vesicle and neuron. The data are fitted by the FLIC theory with a distance of membrane and chip of dcleft ) 8 ( 1 nm for the vesicle and dcleft ) 41 ( 2 nm for the neuron.

Figure 6. FLIC microscopy of a snail neuron on a silicon chip coated with laminin on 5 µm × 5 µm terraces of silicon dioxide. The illumination time was 40 ms. (a) Micrograph with a frame indicating a unit cell of the terraces. (b) Fluorescence intensity versus thickness of silicon dioxide. The data are fitted by the FLIC theory with a distance of membrane and chip of dcleft ) 87 ( 2.5 nm.

laminin, we may speculate about an ordered structure with the molecular tetrapodes adsorbed by their short arms with a dangling long arm.5 However, there is no experimental evidence. On the other hand, the mechanical properties of laminin molecules are not sufficiently wellknown to apply theories on the steric force of grafted polymers.44 Vesicle on Polylysine/Brain Protein. Snail neurons are usually cultured on substrates with polylysine and adsorbed proteins from snail brains.30 To determine the thickness of that coating, we again use giant vesicles. They are strongly attached to that substrate without significant undulations. In the contact region, we observe a checkerboard pattern of fluorescence without fluctuations and without fringes in the surround as shown in Figure 5a. The fluorescence is dark on the thinnest oxide no. 1, intermediate on no. 2, bright no. 3, and intermediate on oxide no. 4. The intensity is plotted versus the thickness of the oxide in Figure 5c. From a fit of the FLIC theory, we obtain a distance dcleft ) 8 ( 1 nm. The mean value for eight vesicles is d h cleft ) 6.6 ( 0.5 nm. That separation is due to the coating of polylysine with adsorbed brainderived proteins. It is slightly higher than the distance on polylysine alone.45 Neuron on Polylysine/Brain Protein. When we culture snail neurons on a FLIC chip coated with polylysine and brain protein, we observe a checkerboard pattern of fluorescence as shown in Figure 5b. The fluorescence is homogeneous on each terrace without fluctuations. The pattern, however, is quite different from that of an attached vesicle. The fluorescence is dark on the thinnest oxide no. 1, bright on no. 2 and no. 3, and dark on oxide no. 4. The (44) Milner, S. T. Science 1991, 251, 905-914. (45) Fromherz, P.; Kiessling, V.; Kottig, K.; Zeck, G. Appl. Phys. A 1999, 69, 571-576.

Figure 7. Summary of distance measurements. (a) Increments of cell-chip separation. Left: The distance of a lipid vesicle on laminin arises from the laminin cushion (red) with 110 nm and the undulating membrane (white) as observed on albumin. Center: The separation of a neuron on laminin is assigned to a compressed cushion of laminin (red) and a glycocalix (blue) of about 44 nm. Right: The thickness of the glycocalix (blue) is evaluated from adhesion experiments with neurons on polylysine (PLL) with brain protein, subtracting the thickness of the protein layer as given by the distance of attached vesicles (white). (b) Structural interpretation. Left: Cell with glycocalix (blue) and substrate with laminin cushion (red) before adhesion. Right: Cell adhesion with glycocalix and a compressed cushion of laminin. Some laminin molecules interact with integrin receptors in the cell membrane (black bars) and provide the attractive force of cell adhesion.

data are plotted in Figure 5c versus the height of the terraces. From a fit with the FLIC theory, we obtain dcleft ) 42 ( 2 nm. The mean value of seven cells is d h cleft ) 51 ( 2 nm. There is a significant increase of the distance compared to the thickness of the protein coat of the chip. We assign the effect to the steric repulsion by the glycocalix of the cell membrane with an effective thickness dglyx ≈ 44 nm. Neuron on Laminin. When we culture a nerve cell on a substrate with laminin, we approach a cell membrane that has a coat dglyx ≈ 44 nm to a substrate covered with a coat with dlam ≈ 110 nm. Without interaction, we expect a total distance of 154 nm. A FLIC micrograph of a neuron on laminin is shown in Figure 6a. The fluorescence pattern exhibits stripes with similar high intensities on terraces

Repulsion and Attraction in Cell Adhesion

1 and 2 and similar low intensities on terraces 3 and 4. The fluorescence on each terrace is rather homogeneous. The data are plotted in Figure 6b and fitted by the FLIC theory. We obtain a distance dcleft ) 87 ( 2.5 nm. The mean value of eight cells is d h cleft ) 87.5 ( 1.5 nm. The distance dcleft ≈ 90 nm of a neuron membrane from silicon dioxide coated with laminin is larger than the thickness of the glycocalix dglyx ≈ 44 nm, and it is lower than the sum dglyx + dlam ≈ 154 nm of the glycocalix and an unperturbed laminin cushion. From that comparison, we propose the following: (i) The separation of a cell from a solid substrate is not caused by the glycocalix alone. The extracellular matrix proteins may contribute significantly, here by about 45 nm. (ii) Cell adhesion as it arises by binding of laminin molecules to integrin receptors may significantly compress the cushion of extracellular matrix protein, here from dlam ≈ 110 nm to d′lam ≈ 45 nm. With respect to that proposal, several critical issues have to be considered: (i) The thickness of the unperturbed laminin is measured in electrolyte with low ionic strength, whereas neurons are cultured at a high ionic strength. If salt leads to a contraction of the free laminin cushion, the effect of the attractive forces in cell adhesion on the compression of the laminin is overestimated. (ii) The thickness of the glycocalix is estimated from a neuron attached to polylysine with brain protein. It is assumed to be unchanged for the adhesion on laminin. A modification in the order of (20% as reported for certain monocytes46 cannot be excluded. (iii) Neurons are cultured for 2 days on two different substrates. A differential secretion of extracellular matrix protein cannot be excluded.

Langmuir, Vol. 19, No. 5, 2003 1585

Conclusion The interpretation of the distance measurements is summarized in Figure 7. The separation of a lipid vesicle on laminin (left column in Figure 7a) is determined by the thickness of a free laminin cushion (red) and by the width of the undulating membrane (white) as observed with vesicles on albumin. The separation of a neuron on laminin (central column in Figure 7a) is given by the glycocalix (blue) and by a compressed cushion of laminin (red). The thickness of the glycocalix is indicated by the separation of a neuron on polylysine (right column in Figure 7a) after subtracting the thickness of the protein coat (white), as obtained from the attachment of vesicles. Figure 7b illustrates the structure of laminin and glycocalix before and after adhesion. The attractive forces in cell adhesion that compress the laminin cushion arise from the interaction of integrin receptors protruding from the membrane by about 15 nm47,48 with laminin molecules. Acknowledgment. We thank Armin Lambacher for computing Figure 2 and Daniel Riveline for critical reading of the manuscript. The project was supported by the Deutsche Forschungsgemeinschaft (SFB 266). LA0263209 (46) Sabri, S.; Soler, M.; Foa, C.; Pierre, A.; Benoliel, A. M.; Bongrand, P. J. Cell Sci. 2000, 113, 1589-1600. (47) Erb, E. M.; Tangemann, K.; Bohrmann, B.; Mu¨ller, B.; Engel, J. Biochemistry 1997, 36, 7395-7402. (48) Xiong, J.-P.; Stehle, T.; Diefenbach, B.; Zhang, R.; Dunker, R.; Scott, D. L.; Joachimiak, A.; Goodman, S. L.; Arnaout, M. A. Science 2001, 294, 339-345.