Influence of Substratum Surface Properties on the Organization of

316L Stainless Steel Cell–Substrate Interface Fabricated by Ultrasonic Shot ... Mina Han, Ananthakrishnan Sethuraman, Ravi S. Kane, and Georges ...
1 downloads 0 Views 208KB Size
Langmuir 1999, 15, 2871-2878

2871

Influence of Substratum Surface Properties on the Organization of Adsorbed Collagen Films: In Situ Characterization by Atomic Force Microscopy Yves F. Dufreˆne,* Thibault G. Marchal, and Paul G. Rouxhet Unite´ de chimie des interfaces, Universite´ catholique de Louvain, Place Croix du Sud 2/18, 1348 Louvain-la-Neuve, Belgium Received August 20, 1998. In Final Form: December 22, 1998 Atomic force microscopy (AFM) imaging and force-distance curves have been used to investigate, in situ, the nanoscale organization of collagen adsorbed on polymer substrata covering a wide range of surface roughness and surface hydrophobicity: bisphenol A polycarbonate (PC), poly(ethylene terephthalate) (PET), and poly(vinylidene difluoride) (PVdF) used as such or treated by an oxygen plasma discharge (ox). After collagen adsorption, PC and PCox showed patterned structures under water, the size of which was influenced by substratum surface oxidation. These structures are attributed to aggregated ends of collagen molecules. Extended rupture lengths were observed in the force-distance curves, suggesting that bundles of collagen molecules adhere to the AFM probe and are progressively torn out upon probe retraction. In contrast, on PET, PETox, PVdF, and PVdFox, adsorbed collagen formed a smooth, homogeneous film devoid of any topographic feature, and no extended rupture lengths were observed. After drying, holes in the collagen film were observed on PET and PVdF and not on PETox and PVdFox. The influence of substratum roughness and physicochemical properties is discussed, considering the mobility of collagen molecules at the interface.

Introduction A variety of applications, such as the development of biosensors, bioreactors, and immunoassays, the purification of proteins, and the biocompatibility of polymer implants, rely on protein adsorption. Collagen is a fibrillar protein which has received considerable interest because it is the most abundant structural protein in the animal kingdom1 and is involved in many important biological functions such as tissue structuring and cell attachment.2 Therefore, understanding the mechanisms controlling collagen adsorption at polymer surfaces is of great significance for both fundamental research and applications.3 During the past decades, new insights have been brought into the mechanisms of adsorption of globular proteins. They may involve redistribution of charged groups in the interfacial film, dehydration of the protein and substratum surface, and structural rearrangements in the protein molecules, the importance of these factors depending on the nature of the protein and of the substratum.4-6 While the roles of substratum surface hydrophobicity and surface charge have been extensively studied, little is known about the influence of substratum roughness. Various techniques have been used to investigate the amounts of proteins adsorbed on solids, including solution depletion,7 radiolabeling,8 X-ray photoelectron spectroscopy (XPS),9 total internal reflection fluorescence,10 re* Corresponding author. Telephone: (32) 10 47 35 89. Fax: (32) 10 47 20 05. E-mail: [email protected]. (1) Kadler, K. Protein Profile 1994, 1, 519. (2) Akiyama, S. K.; Nagata, K.; Yamada, K. M. Biochim. Biophys. Acta 1990, 1031, 91. (3) Park, J. B. Biomaterials; Plenum Press: New York, 1979. (4) Norde, W. Adv. Colloid Interface Sci. 1986, 25, 267. (5) Haynes, C. A.; Norde, W. Colloids Surf. B: Biointerfaces 1994, 2, 517. (6) Norde, W. Cells Mater. 1995, 5, 97. (7) Bull, H. B. Biochim. Biophys. Acta 1956, 19, 464. (8) Van Dulm, P.; Norde, W. J. Colloid Interface Sci. 1983, 91, 248.

flection infrared spectroscopy,11 and ellipsometry.12 In addition, the orientation and conformation of proteins on surfaces have been studied using NMR, Raman and IR spectroscopies, fluorescent probes, calorimetry, and circular dichroism.4,13 However, none of these techniques can provide direct information at high spatial resolution. The ability of atomic force microscopy (AFM) to be operated in different environments, including aqueous solutions, makes it an ideal tool to examine biological samples, from biomolecules14-16 to living cells,15,17 under physiological conditions. In particular, a variety of proteins have been examined, including albumin, lysozyme, actin, immunoglobulins, fibronectin, fibrinogen, and collagen.18-28 (9) Paynter R. W.; Ratner, B. D. In Surface and Interfacial Aspects of Biomedical Polymers; Andrade, J. D., Ed.; Plenum Press: New York, 1985; p 189, Vol 2. (10) Hlady, V.; Reinecke, D. R.; Andrade, J. D. J. Colloid Interface Sci. 1986, 111, 555. (11) Jeon, J. S.; Raghavan, S.; Sperline, R. P. Colloids Surf. A: Physicochem. Eng. Aspects 1994, 92, 255. (12) Golander, C.-G.; Kiss, E. J. Colloid Interface Sci. 1988, 121, 240. (13) Andrade, J. D. Surface and Interfacial Aspects of Biomedical Polymers; Plenum Press: New York, 1985; Vol. 2. (14) Weisenhorn, A. L.; Drake, B.; Prater, C. B.; Gould, S. A. C.; Hansma, P. K.; Ohnesorge, F.; Egger, M.; Heyn, S.-P.; Gaub, H. E. Biophys. J. 1990, 58, 1251. (15) Radmacher, M.; Tillmann, R. W.; Fritz, M.; Gaub, H. E. Science 1992, 257, 1900. (16) Hansma, H. G.; Bezanilla, M.; Zenhausern, F.; Adrian, M.; Sinsheimer, R. L. Nucleic Acids Res. 1993, 21, 505. (17) Butt, H.-J.; Wolff, E. K.; Gould, S. A. C.; Dixon Northern, B.; Peterson, C. M.; Hansma, P. K. J. Struct. Biol. 1990, 105, 54. (18) Fritz, M.; Radmacher, M.; Cleveland, J. P.; Allersma, M. W.; Stewart, R. J.; Gieselmann, R.; Janmey, P.; Schmidt, C. F.; Hansma, P. K. Langmuir 1995, 11, 3529. (19) Chernoff, E. A. G.; Chernoff, D. A. J. Vac. Sci. Technol. A 1992, 10, 596. (20) Lin, J. N.; Drake, B.; Lea, A. S.; Hansma, P. K.; Andrade, J. D. Langmuir 1990, 6, 509. (21) Lea, A. S.; Pungor, A.; Hlady, V.; Andrade, J. D.; Herron, J. N.; Voss, E. W., Jr. Langmuir 1992, 8, 68. (22) Radmacher, M.; Fritz, M.; Cleveland, J. P.; Walters, D. A.; Hansma, P. K. Langmuir 1994, 10, 3809. (23) Ta, T. C.; Sykes, M. T.; McDermott, M. T. Langmuir 1998, 14, 2435. (24) Fang, J.; Knobler, C. M. Langmuir 1996, 12, 1368.

10.1021/la981066z CCC: $18.00 © 1999 American Chemical Society Published on Web 03/23/1999

2872 Langmuir, Vol. 15, No. 8, 1999

However, only a limited number of studies have provided information relevant to the molecular mechanisms of protein adsorption. (i) In some cases, emphasis was put on the protein per se, the adsorption on solid substrata being only used as a necessary step of the sample preparation procedure.18,19 (ii) Many images have been obtained on model, molecularly smooth substrata (e.g. mica, graphite);20-23 although they are well-suited for molecular-scale imaging, they may be of limited relevance to surfaces encountered in practical situations. (iii) In some studies, proteins have been examined only in the dried state,19,24-26 which may be not representative of the in vitro, hydrated state. (iv) Proteins were not always adsorbed from solution onto solid substrata but other protocols such as spin-coating25 or spraying27 were used. (v) Imaging proteins under aqueous conditions has often proved to be delicate due to protein rearrangement under the AFM probe.20,21 (vi) Finally, AFM data were generally not combined with quantitative information about the adsorbed amounts.23,25,27,28 Besides topographic imaging, AFM force-distance curves have been used to measure directly surface forces in aqueous environments, such as van der Waals and electrostatic forces,29 solvation forces,30 steric forces,31 and intermolecular forces between complementary biomolecules.32-34 It is of great interest to exploit forcedistance measurements in combination with topographic imaging to detect proteins at interfaces and to probe interaction forces. In this paper, AFM imaging and force-distance measurements are used to characterize, in situ, the nanoscale organization of collagen adsorbed on a variety of polymer substrata, with the aim to understand better the influence of substratum surface characteristics on the adsorbed film. This work opens new possibilities in the field of biomaterial sciences for studying, in situ, the molecular-scale organization and mobility of adsorbed proteins. Materials and Methods Polymer Substrata. Substrata were commercial polymer films of about 15 µm thickness: bisphenol A polycarbonate (PC Lexan 8800 from General Electric, Brussels, Belgium), poly(ethylene terephthalate) (PET Mylar from Dupont de Nemours, Brussels, Belgium), and poly(vinylidene difluoride) (Solef 1008 from Solvay, Brussels, Belgium), denoted, respectively, as PC, PET, and PVdF. Polymer substrata were used as such or after treatment by oxygen plasma discharge (denoted as PCox, PETox and PVdFox) by a procedure described elsewhere.35 Polymer substrata were analyzed by XPS using an SSI X-Probe (SSX-100/206) photoelectron spectrometer from Fisons, interfaced with a Hewlett-Packard 9000/310 computer allowing instrument control, data accumulation, and data treatment. The procedure for XPS analysis was described before.36

Dufreˆ ne et al. Substrata were characterized by the water contact angle using the sessile drop method coupled with an image analysis system. The water (HPLC grade produced by a MilliQ plus system from Millipore, hereafter referred to as MilliQ water) droplet volume was ∼0.3 µL, and each determination was obtained by averaging the results of 10 measurements. Advancing and receeding contact angles were determined using the Wilhelmy plate method (DCA 322 Cahn Dynamic Contact Angle analyzer). Immersion and emersion of the substrata in MilliQ water was accomplished at a rate of 50 µm/s. AFM Measurements. AFM imaging and force-distance measurements were made at room temperature (∼20 °C) using a commercial optical lever microscope equipped with a liquid cell (Nanoscope III, Digital Instruments, Santa Barbara, CA). Contact mode topographic images were taken in the constantdeflection mode, with an imaging force of ∼1 nN and scan rates of 2-4 Hz. Oxide-sharpened microfabricated Si3N4 cantilevers with spring constants ranging from 0.03 to 0.5 N/m and typical radii of curvature of the probes ∼20 nm were obtained from Park Scientific Instruments (Mountain View, CA). The sensitivity of the AFM detector was estimated using the slope of the retraction force curves in the region where probe and sample are in contact. For AFM under water (MilliQ water), the wet samples were carefully mounted in the liquid cell while avoiding dewetting. During the course of an experiment, care was taken to avoid liquid flowing or air bubble formation, which might cause reorganization of the adsorbed proteins. Protein Adsorption. Collagen S from calf skin, containing more than 95% type I collagen, was purchased from Boehringer Mannheim Biochemica (Brussels, Belgium) as a sterile aqueous solution (3 mg/mL; pH 3.0). It was diluted in Dulbecco’s modified Eagle’s medium (DMEM) from Gibco BRL (European Division, Belgium) to a concentration of 30 µg/mL. For AFM under water, polymer substrata (1 cm × 1 cm) were attached to steel sample pucks (Digital Instruments, Santa Barbara, CA) using an epoxy glue (Araldite, Sodiema, France). Substrata were deposited on the bottom of wells of tissue culture plates (Falcon, Becton Dickinson, Belgium) and incubated for 2 h at 37 °C with 2 mL of the collagen solution. Rinsing was accomplished by successive dilutions while preventing dewetting of the samples: after addition of 2 mL of MilliQ water, 2 mL of the solution was aspired; this operation was repeated 10 times. For AFM in air, polymer substrata (2 cm × 0.5 cm) were directly deposited in the wells of the tissue culture plates and incubated for 2 h at 37 °C with 2 mL of the collagen solution. Substrata were withdrawn from the wells, immersed into a beaker containing about 100 mL of Millipore Q water, and slightly agitated for 5 s; this operation was repeated once. The samples were then dried by flushing with a gentle nitrogen flow for about 10 s and stored in a desiccator containing P2O5. For quantifying the adsorbed amounts, collagen was labeled by reductive methylation using [3H] sodium borohydride as described before.36 Substrata (2 cm × 0.5 cm) were incubated for 2 h at 37 °C with 2 mL of a 30 µg/mL solution of labeled collagen and rinsed as for AFM in air, and the radioactivity of labeled collagen was measured by liquid scintillation counting as described elsewhere.35

Results (25) Mertig, M.; Thiele, U.; Bradt, J.; Leibiger, G.; Pompe, W.; Wendrock, H. Surface Interface Anal. 1997, 25, 514. (26) Lestelius, M.; Liedberg, B.; Tengvall, P. Langmuir 1997, 13, 5900. (27) Emch, R.; Zenhausern, F.; Jobin, M.; Taborelli, M.; Descouts, P. Ultramicroscopy 1992, 42-44, 1155. (28) Chen, X.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B.; Williams, P. M. Langmuir 1997, 13, 4106. (29) Ducker, W. A.; Senden, T. J.; Pashley, R. M. Nature 1991, 353, 239. (30) Butt, H.-J. Biophys. J. 1991, 60, 1438. (31) Biggs, S. Langmuir 1995, 11, 156. (32) Lee, G. U.; Kidwell, D. A.; Colton, R. J. Langmuir 1994, 10, 354. (33) Lee, G. U.; Chrisey, L. A.; Colton, R. J. Science 1994, 266, 771. (34) Hinterdorfer, P., Baumgartner, W.; Gruber, H. J.; Schilcher, K.; Schindler, H. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 3477. (35) Marchal, Th. G.; Dufreˆne, Y. F.; Rouxhet, P. G. Manuscript in preparation. (36) Dewez, J.-L.; Berger, V.; Schneider, Y. J.; Rouxhet, P. G. J. Colloid Interface Sci. 1997, 191, 1.

Surface Properties of Bare Substrata. Table 1 presents the water contact angles, the surface chemical composition, and the root-mean-square (rms) surface roughness determined on the polymer substrata PC, PET, and PVdF, as such or after treatment by oxygen plasma discharge (PCox, PETox, PVdFox). After plasma treatment, the oxygen surface concentration was increased, and a significant decrease in contact angles was observed, indicating a decrease in surface hydrophobicity. The surface rms roughness, expressed as the standard deviation of the topographical height, was determined by AFM under water over 5 µm × 5 µm areas. For untreated polymer substrata, it increased from 0.5 to 10.5 nm in the order PC < PET < PVdF. The rms roughness of the plasma treated substrata was not different from that of the

Collagen Adsorption

Langmuir, Vol. 15, No. 8, 1999 2873

Table 1. Water Contact Angles, Surface Chemical Composition and Surface Roughness of Polymer Substrata atomic fraction (%)c substratum PC PCox PET PETox PVdF PVdFox

θsess

(deg)a

89.1 (1.8) 52.2 (2.6) 74.7 (4.4) 30.9 (2.4) 82.8 (1.4) 69.2 (2.2)

θa

(deg)b

96.3 (2.4) 47.9 (1.6) 83.4 (1.8) 27 (5) 90.9 (0.4) 79.6 (3.5)

θr

(deg)b

74.9 (0) 13.9 (4.8) 56.3 (2) 8.2 (16.4) 71.1 (5.6) 36.9 (12.4)

C

O

roughness (nm)d

83.6 (0.5) 72.5 (0.6) 70.4 (0.9) 64.1 (0.6) 47.5 (1.6) 47.2 (1.2)

16.4 (0.5) 26.9 (0.6) 29.6 (0.9) 35.5 (0.7) 0.4 (0.1) 2.8 (0.4)

0.5 (0.1) 0.5 (0.1) 2.4 (0.2) 2.4 (0.6) 10.5 (2.3) 11.4 (0.9)

a Water contact angle measured by the sessile drop technique; mean value of two independent sets of measurements with the difference between duplicates in brackets. b Advancing and receeding water contact angles measured by the Wilhelmy plate method; mean value of two independent sets of measurements with the difference between duplicates given in parentheses. c Determined by XPS; mean value and standard deviation of three independent determinations. d Root-mean-square (rms) surface roughness determined, in water, on 5 µm × 5 µm AFM topographic images; mean value and standard deviation of three measurements obtained from two independent experiments.

Table 2. Amount of Adsorbed Collagen Determined by Radiolabeling and Surface Roughness after Adsorption substratum PC PCox PET PETox PVdF PVdFox

roughness (nm)a in water air-dried 1.2 (0.1) 1.5 (0.2) 3.1 (1.2) 2.7 (0.4) 10.6 (0.5) 10.3 (2.0)

0.8 (0.1) 0.9 (0.4) 2.2 (0.8) 2.6 (0.4) 9.5 (1.3) 11.3 (0.1)

adsorbed amount (µg cm-2)b 0.49 (0.12) 0.54 (0.05) 0.37 (0.06) 0.33 (0.05) 0.46 (0.04) 0.37 (0.03)

a Root-mean-square (rms) surface roughness determined, in water and after air-drying, on 5 µm × 5 µm AFM topographic images; mean value and standard deviation of three measurements obtained from two independent experiments. b Amount adsorbed after conditioning with a 30 µg/mL collagen solution; mean value with the difference between duplicates given in parentheses.

untreated ones, indicating that the plasma treatment did not modify significantly the polymer surface morphology on the nanometer-scale. PC and PCox surfaces were extremely smooth; as a matter of fact their surface roughness was similar to that currently obtained with spin-coated polymer surfaces (data not shown). For all substrata, no significant variation of surface roughness was noted when comparing different zones as revealed by the small standard deviations of the measurements. Differences in substratum surface morphology are further illustrated in Figures 1A,B and 2A,B which present AFM topographic images, together with vertical cross-sections, of PC, PCox, PET, and PVdF. In general, the morphology of real surfaces depends on the length scale of observation. To separate the contributions of different spatial frequencies, or length scales, to the surface relief, statistical analysis of the roughness was performed through 2D power spectral density (PSD) of the fast Fourier transform of the topographic images.37,38 Figure 3 shows the variation of rms roughness as a function of the length scale, as constructed from the mean PSD of 5 µm × 5 µm images. For all substrata, the roughness clearly increased with length scale and reached a value which is the rms roughness of the image. The order of increasing roughness PC < PET < PVdF was found at all length scales (50 nm to 5 µm). The roughness was respectively below 0.5 nm, about 1 nm and about 4 nm at the length scale of 300 nm, which is the length of the collagen molecule. Protein Adsorption. Table 2 (third column) presents, for the six substrata, the amount of adsorbed collagen determined by radiolabeling. This was slightly larger on PC and PCox compared to the other substrata. (37) Biscarini, F.; Samori, P.; Lauria, A.; Ostoja, P.; Zamboni, R.; Taliani, C.; Viville, P.; Lazzaroni, R.; Bre´das, J. L. Thin Solid Films 1996, 284-285, 439. (38) Viville, P.; Lazzaroni, R.; Bre´das, J. L.; Moretti, P.; Samori, P.; Biscarini, F. Adv. Mater. 1998, 10, 57.

The surface topography of polymer substrata after collagen adsorption was investigated under water using contact mode AFM. The topographic images of collagenconditioned PC and PCox, shown in parts C and D of Figure 1, revealed features uniformly distributed accross the surface which were not present on the bare substrata. At imaging forces about 1 nN, images of the same area could be obtained repeatedly without altering the morphology, i.e., without disturbing significantly the adsorbed protein film. Images obtained by forward and backward scanning were identical, indicating no significant contribution of lateral forces to the apparent topographic contrast. The shape of the features observed on the two substrata was strikingly different: PC showed a pattern of small dotlike structures of 5.7 (s.d. 1.3) nm height (10 measurements obtained from two independent sets of experiments) and about 50 nm in diameter, separated by about 0.2-0.4 µm; PCox exhibited a network of elongated structures of 4.7 (0.4) nm thickness and about 0.5 µm length, separated by about 0.5-1.0 µm. The presence of these topographic features was reflected in the roughness: due to collagen adsorption, the surface roughness (rms) of PC and PCox increased from 0.5 nm to 1.2 and 1.5 nm, respectively (Table 2). The small standard deviations indicate that the adsorbed films had constant topographic characteristics in terms of vertical variations. Parts C and D of Figure 2 present topographic images of PET and PVdF after collagen adsorption. Contrary to collagen-conditioned PC, no or very few features attributable to adsorbed proteins could be identified in the images. Similar observations were made with PETox and PVdFox (images not presented). The surface roughness after collagen adsorption (Table 2) was not significantly different from that of the bare substrata. In addition to surface topographic imaging, AFM was also used to determine interaction forces acting between the silicon nitride probe and the surface after collagen adsorption. Parts A and B of Figure 4 present typical force-distance curves recorded in water on collagenconditioned PC. Approach curves recorded at different spots all presented the same features, the probe experiencing a repulsion force at short separation. Upon retraction, two types of behaviors were observed. In about half of the measurements (made with two probe-substrata pairs in at least three different regions of each sample), the approach and retraction curves were similar; i.e., no adhesion events were observed (Figure 4A). In the other half of the measurements (Figure 4B), there were small unbinding events, either single or multiple, at a short separation distance; moreover, at a distance larger than 100 nm, a strong attractive force, referred to as an elongation force, developed gradually until the probe and sample jumped away. The extended rupture length, defined as the distance at which the surfaces jumped away,

2874 Langmuir, Vol. 15, No. 8, 1999

Dufreˆ ne et al.

Figure 1. AFM topographic images (size 5 µm × 5 µm; z-range: 15 nm) in water of bare PC and PCox (A, B) and of PC and PCox after collagen (COLL.) adsorption (C, D). Lighter levels in the images correspond to higher height. Cross-sections taken along the line indicated by the arrows are shown beneath the images. Similar images were obtained in different regions of at least two different samples.

was typically in the range 100-500 nm. Multiple unbinding events and extended rupture lengths were never observed on bare PC (results not shown). PCox gave results similar to PC, including the particularities of retraction curves. In contrast, for PET, PETox, PVdF, and PVdFox, the force-distance curves were all similar to that of Figure 4A; i.e., the approach and retraction curves were always similar. To assess the possible effect of protein adsorption on the AFM probe during force measurements, forcedistance curves were recorded on bare PC with probes that were previously used on collagen-conditioned substrata. The curve obtained, shown in Figure 4C, was clearly different from those obtained on conditioned substrata. AFM Characterization after Protein Adsorption and Drying. Contact mode AFM in air was used to examine the surface morphology of samples dried after conditioning with collagen. A topographic image of col-

lagen-conditioned PC after drying is shown in Figure 5A. A dense and uniform pattern of elongated objects was detected. The same surface morphology was obtained for PCox after drying (result not shown). This change in surface morphology when comparing the dried samples with the hydrated ones was accompanied by a slight decrease in surface roughness (Table 2). After drying, collagen-conditioned PET and PVdF (parts B and C of Figure 5) showed a smooth film into which holes, about 50 nm in diameter, were randomly distributed, the depth of these holes being 4.4 (0.9) nm and 4.7 (0.7) nm, respectively. In contrast, continuous films devoid of holes were observed on PETox and PVdFox (results not shown). For PET, PVdF, PETox, and PVdFox no significant change of rms roughness was noted when comparing the dried conditioned substrata with the hydrated conditioned substrata or with the bare substrata (Table 2).

Collagen Adsorption

Langmuir, Vol. 15, No. 8, 1999 2875

Figure 2. Topographic images (size 5 µm × 5 µm; z-range 25 and 150 nm; for (A, C) and (B, D), respectively) in water of bare PET (A), bare PVdF (B), and of PET and PVdF after collagen adsorption (C, D).

Discussion An overall description of the adsorbed collagen films can be inferred from radiolabeling data. The amount of collagen adsorbed on the different substrata lies in the range of 0.35-0.50 µg cm-2. Considering a specific weight of 1.4 g cm-3, this is equivalent to a thickness of 2.5-3.6 nm in the case of a continuous and homogeneous film. In view of the size of a collagen molecule (stiff triple helix about 300 kDa, 300 nm long, and 1.5 nm in diameter), a closely packed monolayer of molecules lying flat on the surface would represent an adsorbed amount of 0.1 µg cm-2.36 The adsorbed amounts observed here are about four times larger, which is an interesting range to investigate by AFM. Organization of Adsorbed Collagen Films under Water. AFM images of collagen-conditioned PC and PCox show dotlike and elongated structures, respectively. It must be realized that the width of these structures is overestimated due to the finite size of the AFM probe. For a sphere or a cylinder, the apparent width W measured

Figure 3. Surface roughness (rms) as a function of the length scale, constructed from the mean PSD of 5 µm × 5 µm images: PC (triangles), PET (circles), and PVdF (squares); as such (closed symbols) or after oxidation (open symbols).

by AFM can be calculated from W ) 2(Dd)1/2, where D is the probe diameter of curvature and d the actual sphere

2876 Langmuir, Vol. 15, No. 8, 1999

Dufreˆ ne et al.

Figure 4. Typical force-distance curves (A, B) recorded in water on PC after collagen adsorption (PC + COLL.) and forcedistance curve (C) obtained on bare PC in water, using a probe already used to examine PC + COLL. The maximum repulsive force applied to the sample surface was about 2 nN.

or cylinder diameter.39 Considering a D value of 50 nm, the apparent size of the dots observed on PC (20-100 nm) and the apparent width of elongated structures observed on PCox (50-200 nm) are thus compatible with objects of 2-50 nm size and 12.5-200 nm width, respectively. The dotlike structures observed on PC, about 6 nm high, are quite different from collagen fibrils, reported with diameters ranging from 20 to 500 nm. On the other hand, both radiolabeling and XPS40 data indicate that the regions which are not covered by dotlike structures are coated by a continuous layer of collagen molecules lying flat on the surface. These dotlike structures may therefore be attributed to aggregated ends of collagen molecules pointing perpendicular to the surface, as schematically presented in parts A and B of Figure 6 (top). The elongated objects observed on PCox may be fibrils lying down on the surface; however, they may also be due to assemblages of molecular ends pointing perpendicular to the surface, i.e., aggregates of dots similar to those found on PC. The latter model is supported by the fact that similar unbinding events are observed in about half of the retraction curves on both PC and PCox after collagen adsorption (Figure 4), while they are never observed with other substrata. Extended rupture lengths resembling those measured here have been reported for the stretching of flexible biomolecules, including long strands of DNA,33 antibodyantigen complexes,34 and polysaccharide filaments.41 Since (39) Fang, J.; Knobler, C. M.; Gingery, M.; Eiserling, F. A. J. Phys. Chem. B 1997, 101, 8692. (40) Dufreˆne, Y. F.; Marchal, Th. G.; Rouxhet, P. G. Appl. Surf. Sci., in press. (41) Rief, M.; Oesterhelt, F.; Heymann, B.; Gaub, H. E. Science 1997, 275, 1295.

Figure 5. Topographic images (2 µm × 2 µm) of PC (z-range 10 nm), PET (z-range 25 nm), and PVdF (z-range 75 nm) after collagen adsorption and drying.

collagen molecules are fairly stiff filaments, molecular stretching alone does not explain the observed rupture lengths. The curves obtained on bare PC with probes that were previously used on collagen-conditioned substrata never show extended rupture lengths, indicating that the latter are related to the presence of collagen adsorbed on the substratum. The occurrence of extended rupture lengths in the retraction curves would be due to adhesion between the probe and a bundle of collagen molecules and to the progressive tearing off or sliding of collagen molecules as sketched in Figure 6C (top). The nature and properties of the structures observed on PC and PCox could be investigated further using AFM probes functionalized with receptor molecules that specifically recognize collagen. A complementary approach would consist in characterizing the distribution of fluorescent-probe-labeled collagen on the substrata, using near-field scanning optical microscopy. The surface morphology of PET, PETox, PVdF, and PVdFox after collagen adsorption is clearly different from that of PC and PCox in that no features attributable to adsorbed collagen can be visualized in the topographic images. This observation is compatible with three situations: (i) collagen is lacking at the substratum surface;

Collagen Adsorption

Figure 6. Schematic representation of the film of collagen adsorbed on PC (top) and PET or PVdF (bottom): top view (A), cross-section (B), and interaction with the AFM probe upon retraction (C).

(ii) collagen is present as a smooth and continuous film which is not detected by AFM; (iii) collagen structures are present but not detected due to their small size compared to the high substratum roughness. To evaluate the importance of the latter effect, images obtained for PC and PCox after collagen adsorption were superimposed onto those obtained for the other bare substrata. This revealed that PET and PETox after collagen adsorption are clearly devoid of structures as those observed on PC and PCox. The occurrence of such structures on PVdF and PVdFox would not be detected due to the high roughness of the substrata. There are several pieces of evidence indicating that collagen is present at the surface of PET, PETox, PVdF, and PVdFox as a fairly smooth and continuous film (parts A and B of Figure 6 (bottom)). (i) Radiochemical data show that the amount of adsorbed proteins (Table 2) is similar to that found on PC and PCox. (ii) Modeling XPS and radiochemical data indicates that the four substrata, after collagen adsorption and drying, are coated by a collagen film of several nm covering a large fraction of the surface.40 (iii) All the force-distance curves present a deviation from linearity in the contact region similar to that of Figure 4A as opposed to bare substrata. This may originate from hydration repulsion forces between the silicon nitride probe and the adsorbed collagen film or from compression of the adsorbed film. The fact that force curves recorded in different spots were all similar suggests that the protein coverage is continuous on the scale of the probe (∼50 nm). (iv) Drying conditioned PET and PVdF results in the appearance of holes distributed over the surface (see below), which supports the presence of an adsorbed film. The well-resolved images obtained for all conditioned substrata, without apparent alteration of the surface morphology, are in contrast with certain in situ AFM studies on globular proteins. Under imaging forces similar to those of this study, immunoglobulin adsorbed on mica was found to rearrange readily under the scanning probe, the latter behaving as a molecular broom.20,21 On the other hand, immunoglobulin adsorbed on highly oriented pyrolytic graphite (HOPG) could be imaged without obvious distortion.42 In the present study, the ability to image collagen on a variety of substrata without rearrangement is attributed to the filamentous nature of collagen and to its strong adsorption and self-aggregation properties. Networks of monomers or fibrils are indeed expected to provide a better resitance to the lateral and normal forces exerted by the scanning probe compared to small, deformable globular proteins. Effect of Drying on the Adsorbed Film Morphology. The observation of holes after drying confirms the (42) Cullen, D. C.; Lowe, C. R. J. Colloid Interface Sci. 1994, 166, 102.

Langmuir, Vol. 15, No. 8, 1999 2877

presence of smooth adsorbed films on PET and PVdF under water. The occurrence of holes may be either interpreted in terms of dewetting upon drying or attributed to the release of stresses created by shrinkage of a stiff adsorbed layer consecutive to drying. The second explanation is more relevant as the hole size and hole interdistance are typically smaller than the length of the collagen molecule. Interestingly, networks formed by collagen are different from those obtained with thin polymer films and soap foams.25 The latter are known to generate networks which generally decay into rows of drops as a result of dewetting. The lack of hole formation upon drying on PETox and PVdFox points to the influence of substratum hydrophobicity on the stress build-up and release in the drying film. This may be related to the observation that spincoated collagen covered hydrophilic substrata homogeneously, while pore formation took place in the film formed on hydrophobic surfaces.25 Upon drying, the markedly different patterns of the collagen films on PC and PCox transform into the same morphology. This supports the idea that the collagen film organization under water has features which are common to PC and PCox, i.e., assemblages of molecular ends pointing perpendicular to the surface. Influence of Substratum Surface Properties. During the past few decades, studies focusing on the adsorption of globular proteins have shown the importance of substratum surface charge and hydrophobicity,4-6 while the possible influence of substratum roughness has generally been neglected. The above discussion shows that while the amount of collagen adsorbed varies only slightly according to substratum surface properties, as observed for other polymer substrata,36 the nanoscale organization of the adsorbed film and its evolution upon drying differ markedly. The influence of substratum surface properties may be discussed in terms of surface roughness and hydrophobicity by comparing (i) the three untreated materials (PC, PET, and PVdF), which differ markedly by their surface roughness but have fairly close water contact angles, and (ii) the untreated and corresponding oxidized substrata, which show differences in water contact angles at nearly constant roughness. The effect of substratum topographic variations should be considered in the light of the horizontal and vertical dimensions of the collagen molecules. The 2D power spectral density analysis of the topographic images shows that, on a length scale close to the collagen molecular length (300 nm), the roughness of PC, PET, and PVdF is below 0.5 nm, about 1 nm, and about 4 nm, respectively. A mechanism may be suggested in which substratum height variations influence the adsorbed film organization by affecting the mobility of collagen within the adsorbed phase. On PC, which shows height variations smaller than the collagen molecular thickness, collagen molecules would be relatively free to move; this would allow the formation of dotlike structures resulting from the competition between aggregation and affinity with the substratum (Figure 6, top). In contrast, on PET and PVdF, which exhibit height variations either close to or larger than the molecular thickness, collagen molecules would have reduced mobility as a result of their filamentous nature. This would lead to the formation of an adsorbed film devoid of any particular surface structures (Figure 6, bottom). The above data show that structural features of the adsorbed collagen film are related to surface roughness. A surface relief close to the collagen molecule thickness may affect the mobility of the adsorbed molecules and their ability to aggregate. Further experiments are needed

2878 Langmuir, Vol. 15, No. 8, 1999

(i) to determine more accurately how a specific horizontal distribution of topographic features (for instance, steps and terraces of defined dimensions) on surfaces of constant chemical composition affects the organization of the adsorbed collagen molecules and (ii) to assess to what extent the effect of surface roughness is applicable to other proteins. One may anticipate that stiff, filamentous proteins such as actin will behave as collagen, while soft, globular proteins such as albumin will be less (not) sensitive to substratum surface roughness. Substratum physicochemical properties (composition, hydrophobicity, ...) seem to affect the adsorbed film organization in two ways. On smooth substrata, they influence the size of the collagen structures under water: dotlike structures are observed on untreated PC while larger structures are present on PCox. On rough substrata, they influence the film morphology after drying: films with holes are observed on strongly hydrophobic substrata (PET, PVdF) as opposed to less hydrophobic substrata (PETox, PVdFox), suggesting that lowering the substratum hydrophobicity stabilizes the adsorbed film. Several mechanisms may be invoked to explain the influence of substratum physicochemical properties on the film organization in water and after drying. On one hand, the substratum surface may be considered as that of a rigid solid and the mechanisms may be considered in terms of van der Waals forces, electrostatic interactions, hydrophobic interactions and hydration forces. On the other hand, particularly with surface-oxidized substrata, the surface may carry solvated macromolecules protruding into the solution which may interact with collagen. Conclusions This study shows that substratum surface properties influence the nanoscale organization of adsorbed collagen

Dufreˆ ne et al.

films and its evolution upon drying. On substrata exhibiting vertical topographic variations smaller than the collagen molecular thickness, patterned structures attributed to aggregated ends of collagen molecules are observed under water. Extended rupture lengths are detected in the force-distance curves, suggesting the progressive tearing off of collagen molecules from the adsorbed phase. In contrast, on substrata showing vertical topographic variations close to or larger than the collagen molecular thickness, the adsorbed collagen forms a smooth, homogeneous film devoid of any topographic feature and no extended rupture lengths are observed. These observations suggest that a critical substratum height variation close to the collagen molecule thickness may affect the mobility of the adsorbed molecules and their tendency to aggregate. The effect of substratum physicochemical properties is influenced by surface roughness. On smooth substrata, surface oxidation influences the size of aggregated structures. On rough substrata, it influences the film morphology after drying presumably due to stress buildup and release: discontinuous films with holes are observed on strongly hydrophobic substrata while hole formation is prevented on less hydrophobic substrata. Acknowledgment. The support of the National Foundation for Scientific Research (FNRS), of the Foundation for Training in Industrial and Agricultural Research (FRIA), and of the Federal Office for Scientific, Technical, and Cultural Affairs (Interuniversity Poles of Attraction Program) is gratefully acknowledged. The authors thank Prof. P. Grange for the use of the atomic force microscope, and B. Nysten and C. C. Dupont-Gillain for valuable discussions. LA981066Z