Friction Force Spectroscopy As a Tool to Study the Strength and

ARTICLE pubs.acs.org/Langmuir

Friction Force Spectroscopy As a Tool to Study the Strength and Lateral Diffusion of Protein Layers Javier Sotres,* Alejandro Barrantes, and Thomas Arnebrant Biomedical Laboratory Science and Technology, Faculty of Health and Society, Malmoe University, 20506 Malmoe, Sweden

bS Supporting Information ABSTRACT: We present a method to study the strength of layers of biological molecules in liquid medium. The method is based on the Friction Force Spectroscopy operation mode of the Atomic Force Microscope. It works by scratching the sample surface at different applied loads while registering the evolution of the sample topography and of the friction between probe and sample. Results are presented for BSA and β-casein monolayers on hydrophobic surfaces. We show how the simultaneous monitoring of topography and friction allows detecting differences not only between the strength of both types of layers, but also between the lateral diffusion of the proteins within these layers. Specifically, β-casein is shown to form stronger layers than BSA. The yield strengths calculated for both of these systems are in the range 50
70 MPa. Moreover, while no lateral diffusion is observed for BSA, we show that β-casein diffuses along the hydrophobic substrates at a rate higher than the scan velocity of the tip (16 μm s
1 in our case).

1. INTRODUCTION The resistance of protein layers to mechanical damage is one of the characteristics of these systems where less research has been performed, this being a consequence of the lack of experimental techniques. Nevertheless, there is an urge to develop such techniques considering the vast number of natural and artificial systems where protein layers are present and exposed to mechanical damage. In certain bacteria they form the outermost barrier with their environment: the S-layer.1 Proteins are also the main component of conditioning films, which are rapidly formed on surfaces exposed to almost any naturally occurring liquid, from seawater2 to biological fluids such as blood or saliva.3 Multiple types of sensors depend on proteins deployed on their surfaces for their specific sensing properties.4,5 Another example is that of protein-stabilized emulsions/dispersions, where protein layers cover the dispersed phase.6 In their corresponding working environment, stresses of different sources are applied in many of these systems. During industrial processing, protein-stabilized emulsions are exposed to severe mechanical stress, caused for instance by pumps or turbulent eddies. This can damage the emulsifying protein-coverage, leading to coagulation or coalescence of the dispersed phase.7 The protein-covered surfaces of implanted joints constitute another case of surfaces exposed to mechanical stress. This protein-coverage has been proven to reduce the joint wear rate under sliding wear corrosion,8 preventing the release of metal ions which can trigger the inflammation and necrosis of the surrounding tissue. Protein-based sensor devices can also suffer severe mechanical damage that can lower their performance. An example is that of protein layers on electrode surfaces which can be damaged by processes such as electrolysis coupled with stirring.9,10 r 2011 American Chemical Society

Implantable sensors are another example, as they can be mechanically abraded by the extracellular matrix as proved by electron microscopy imaging of explanted sensors.11 In the previous cases, wear of the protein layers constitutes a process to be avoided as it lowers the performance of the systems they are part of. However, the opposite case is also found, i.e., that where an easy mechanical removal of the protein layers is desired. An example is that of proteinaceous deposits on equipment in food processing.12 Thus, the resistance to wear of protein layers adsorbed on surfaces constitutes a critical parameter for the design of the systems where they are incorporated, as well as for understanding their function and performance. This has been the motivation of the present work, where we present and test a novel methodology for studying, at the single asperity level, wear in protein layers adsorbed on solid surfaces. A considerable number of techniques exist for testing the strength with which thin films are adsorbed on substrates. They usually work by monitoring the response to an applied stress. These techniques can be classified regarding the direction in which the stress is applied. One option is to indent the sample within its normal direction with the probe.13 A pure shear stress can also be exerted, for instance by pulling horizontally the adsorbed film with a T-shaped probe.14 The more real-like situation where both normal and shear stresses are simultaneously applied can also be studied. This is usually done by “scratching” the sample; i.e. by sliding a probe along its surface.15 A typical approach to characterize Received: May 5, 2011 Revised: June 23, 2011 Published: June 24, 2011 9439

dx.doi.org/10.1021/la201673r | Langmuir 2011, 27, 9439–9448

Langmuir the resistance/response of the sample to the scratch is to visualize it afterward.16 A more sophisticated approach is to gradually increase the applied force while registering a quantity used to characterize the scratching process.17 Friction between probe and sample is frequently chosen as this quantity.18 This methodology, from now on labeled friction force spectroscopy (FFS), offers the possibility of precisely determining the forces needed to break an adsorbed thin film. This is possible as the onset of the rupture of the film can be detected by a drastic change in the relationship between friction and normal forces.19 The atomic force microscope (AFM)20 is frequently used for studying the mechanical properties of surfaces. It employs a sharp tip attached at the free end of a cantilever to probe the sample. AFM is highly suitable for tribological studies as the relative position between tip and sample can be controlled with a precision below the nm range while controlling and measuring the applied normal and friction forces with pN resolution. Moreover, AFM presents a clear advantage over other tribometer techniques. This advantage originates from the fact that contact between macroscopic surfaces is generally established through a multitude of microscopic asperities. As a consequence, it is extremely difficult to determine magnitudes such as the real contact area between sliding surfaces, and, subsequently, the contact stresses and strains generated between them. In contrast, AFM tips with a final apex size in the order of nanometers can be regarded as single asperities.21 Thus, AFM offers an excellent opportunity for performing mechanical studies in a controlled way. Many of these studies are based on normal indentation measurements, also in the case of thin layers of biological molecules. This ranges from elasticity studies of protein layers22 to measurements of the forces needed to break lipid layers.23 AFM can also be used to apply normal and shear stresses simultaneously for the study of wear. It is for example, possible to perform FFS measurements by monitoring friction while varying the load applied with the sliding tip. Systems studied in this way include again thin layers of biological molecules, from those formed by lipids24 to those formed by proteins.25 AFM wear studies also benefit from the fact that AFM is itself a microscope. The simplest way to visualize a surface is to scan it while maintaining constant the deflection of the cantilever, and thus the applied normal force. AFM imaging gives a high advantage in applications where the worn volume is to be determined, as the scratched area can be visualized without the need of employing a different technique.26 This methodology is also used to study protein monolayers in liquid medium.27 Especially promising is the possibility offered by AFM to follow the evolution of the topography of a sample while scratching it. Even though this possibility is still almost completely unexplored, we recently proved it to be extremely useful in the characterization of the mechanical removal of protein monolayers.25 In this work, we approach the study of the strength of protein monolayers by two-dimensional scratching with an AFM operated in the FFS mode. Specifically, the simultaneous monitoring of friction and topography allowed by this mode is proven to be crucial for determining the strength of the layers, as well as the lateral diffusion of the proteins within these layers. Moreover, we develop a framework which allows characterizing the mechanical stability of the protein layers in terms of a widely used quantity: the yield strength. The technique is tested for bovine serum albumin (BSA) and β-casein monolayers, both adsorbed on hydrophobic substrates. The choice of these systems is not casual. BSA, a serum protein with numerous biochemical applications,28 is considered a representative example of soft globular proteins.29

ARTICLE

In contrast, β-casein, a milk protein with valuable emulsifying properties,30,31 is considered a representative example of unstructured amphiphilic proteins. Different mechanical properties, and therefore resistance to wear, are expected for both types of monolayers. Our results show that the technique is sensitive not only to differences between the yield strengths of both types of layers, but also to differences in their lateral diffusion. Thus, AFM-based friction force spectroscopy is validated as a powerful technique for mechanical studies of protein layers in their natural environment.

2. MATERIALS AND METHODS 2.1. Proteins and Chemicals. Commercially available β-casein (Prod. no. C6905, Sigma-Aldrich) and Bovine Serum Albumin, BSA (Prod. no. A4503, Sigma-Aldrich) were used without further purification. The dry protein samples were dissolved in PBS buffer (Prod. no. P4417, Sigma-Aldrich) to a concentration of 0.2 mg 3 mL
1 for β-casein and 1 mg 3 mL
1 for BSA. β-casein was almost calciumfree as determined by atomic adsorption (OPTIMA 3000, PerkinElmer Co., Norwalk, Connecticut USA), specifically 2  10
6 mol 3 ml
1 when dissolved in water. This corresponds to 0.24 mols of calcium per mole of β-casein. All other chemicals were of analytical grade or higher. Water was treated by a purifying unit (ELGA UHQ PS, Elga Ltd. UK), resistivity >18 MΩ 3 cm. 2.2. Hydrophobization of Silica Surfaces. P-doped (boron) silicon wafers with a resistivity of 10
20 Ω 3 cm (Semiconductor Wafer Inc., Taiwan) were oxidized in an oxygen atmosphere to obtain an oxide layer of approximately 30 nm.32 Prior to hydrophobization, the silica (Si/SiO2) substrates were cleaned as described elsewhere.32 The clean silica surfaces were then hydrophobized by liquid-phase silanization. Specifically, the surfaces were dried in nitrogen gas and immersed in a solution containing 25 μL of dichlorodimethylsilane and 50 mL of trichloroethylene for one hour. After silanization, the surfaces were washed three times in trichloroethylene and three times in ethanol. The water contact angle after hydrophobization was determined to 95 ( 5°. The surfaces were stored in ethanol until use. 2.3. Sample Preparation. Both β-casein and BSA monolayers were prepared by pipetting 50 μL of the respective protein solutions onto a hydrophobized silica surface. After 30 min, samples were gently rinsed with PBS buffer and immediately placed on the microscope not allowing them to dry at any moment. 2.4. Atomic Force Microscopy. A commercial AFM equipped with a liquid cell was employed (Multimode SPM with a Nanoscope IV control unit, Veeco Instruments, Santa Barbara CA). Rectangular silicon nitride levers with a nominal normal spring constant of 0.1 N 3 m
1 were employed (OMLC-RC800PSA, Olympus, Japan). The normal, kN, and torsional, kΦ, spring constants were determined with the AFM Tune IT software (ForceIT, Sweden) for every lever using the methods developed by Sader and coworkers.33,34 The lateral sensitivity of the setup, δΦ, was determined by the pivot method,35 while the normal sensitivity, δN, was calculated from the slope of the deflection of the lever while pressed against a cleaved mica surface. The determination of both normal and lateral sensitivities was carried out in water. 2.5. Friction Force Spectroscopy (FFS). Friction force spectroscopy (FFS) measurements consist of sets of 2D-scans performed on the same area of the sample, each of them at a constant load force, FL, which is varied between scans. Specifically, the scans performed in this work consisted of 128  32 points, and they were performed on 2  2 μm areas, along lines perpendicular to the long axis of the cantilever, and at a tip velocity of 16 μm 3 s
1. 9440

dx.doi.org/10.1021/la201673r |Langmuir 2011, 27, 9439–9448

Langmuir

ARTICLE

For each scan, the topography and the lateral force images are simultaneously recorded both in the trace and retrace directions. From the lateral force images an average friction force, FF, is calculated. Correspondingly, the roughness of the sample, characterized by its height standard deviation, SD, is calculated from the topography images. Thus, a FFS measurement is characterized by representing two different measured parameters as a function of the applied load FL: (i) the roughness of the surface giving rise to the SD vs FL curve, and (ii) the friction force giving rise to the FF vs FL curve. The parameters for describing a FFS measurement were calculated as follows using self-programmed routines written in Matlab (Math Works Inc., Natick, MA). FL is determined by the following: FL ¼ kN 3 δN 3 ðVVertical
VVertical0 Þ

ð1Þ

where VVertical and VVertical 0 are the photodiode vertical signals measured when the tip scans the sample, and when tip is far from the sample, respectively. The total normal force applied on the sample, FN, is the sum of FL and of the adhesion force between tip and sample, Fadh, so that FN = FL + Fadh. Before and after each FFS measurement, force curves were obtained for determining both Fadh and VVertical 0. In our experiments Fadh was, if not negligible, at least much lower than the applied loads. For this reason, our results are presented in terms of FL instead of in terms of FN. FF is calculated for each scan by the following procedure: (i) first an inner square area, with a lateral size half of that scanned, is selected for further calculations in order to avoid border effects, (ii) the difference between the lateral photodiode signals of the trace and of the retrace directions, ΔVLateral, is calculated for each scan line, (iii) ΔVLateral is then averaged over all the scan lines, and (iv) finally FF is obtained by scaling ΔVLateral with the following equation: FF ðΔVl Þ ¼

kΦ 1 ΔVLateral δΦ 3 hef f 3 2

ð2Þ

where heff is the effective height of the probe. As commented, the scan topographies were characterized by their height standard deviation, SD: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uN u u ðhðiÞ
< h >Þ2 t ð3Þ SD ¼ i ¼ 1 N

∑

where N is the number of points in the image, h(i) the height of the ith image point, and the mean height of the whole image. Before each experiment, tips were rubbed against clean mica surfaces in order to clean the tip apex, and also to remove asperities from it. This process has been proved crucial for avoiding dramatic changes in the tip during FFS measurements, leading to reproducible results.36 2.6. Ellipsometry. The adsorption of β-casein and BSA to hydrophobized silica surfaces was monitored in situ by ellipsometry. Theoretical principles can be found elsewhere.37 The experimental setup is based on null ellipsometry according to the principles of Cuypers.38 The instrument used was a Rudolph thin film ellipsometer (type 43603
200E, Rudolph Research, U.S.) automated according to the concept of Landgren and J€onsson.39 A xenon arc lamp was used as the light source, and light was detected at 442.9 nm using an interference filter with UV and

Figure 1. Time evolution of surface excess of BSA and β-casein layers determined by null-ellipsometry. Measurements were performed in PBS buffer. During the initial parts of the experiments, the surfaces were incubated in 1 mg mL
1 and 0.2 mg mL
1 solutions of BSA and β-casein, respectively. After an incubation time of 30 min, the surfaces were rinsed with protein-free buffer solution.

infrared blocking (Melles Griot, The Netherlands). The trapezoid cuvette made of optical glass (Hellma, Germany) was equipped with a magnetic stirrer and thermostatted to 25 °C. Protein stock solutions were added to the cuvette containing the surface immersed in 5 mL of buffer solution. Then, the adsorption was monitored for 30 min followed by rinsing with protein-free buffer. The determination of the silicon complex refractive index, and of the thickness and refractive index of the silicon oxide layer, was performed using air and water as ambient media.39 Four zone measurements were conducted to minimize systematic errors. The refractive index and thickness of the protein film were obtained by means of a four-layer model (Si substrate-SiO2 film-Protein layer-Buffer solution). The adsorbed amount was calculated from the refractive index and thickness of the film assuming a linear increase for the refractive index with the concentration of 0.18 mL 3 g
1 for both proteins.40

3. RESULTS 3.1. Ellipsometry Study of the Formation of BSA and β-Casein Monolayers. The formation of BSA and β-casein

layers on hydrophobized silica surfaces was studied by means of ellipsometry. Figure 1 shows the time evolution of the surface excess, Γ, during the adsorption of both proteins in PBS buffer pH 7.4 followed by a rinsing step. Protein concentration during adsorption was 1 mg 3 mL
1 for BSA and 0.2 mg 3 mL
1 for β-casein. Rinsing with protein-free buffer started after 30 min in both cases. Both proteins show a fast initial adsorption. In the case of β-casein, rinsing induces desorption of ∼20% of the adsorbed amount. In the case of BSA no desorption takes place after the rinsing step, and there is a slight tendency for increase in Γ in agreement with previously published results.41,42 The total amount of adsorbed protein obtained after rinsing, Γp, was 2.24 mg 3 m
2 and 1.58 mg 3 m
2 for β-casein and BSA respectively. 3.2. Friction Force Spectroscopy of BSA Monolayers. A representative FFS measurement on a BSA monolayer in PBS buffer pH 7.4 is shown in Figure 2. Scans were performed at load force intervals of ∼0.5 nN, first by increasing the load and then by decreasing it. Figure 2a shows representative images of the evolution of the topography during the scratch. The corresponding SD vs FL curve (Figure 2b) also illustrates this evolution. The frictional response to the scratch, i.e., the FF vs FL curve, is 9441

dx.doi.org/10.1021/la201673r |Langmuir 2011, 27, 9439–9448

Langmuir

Figure 2. Data corresponding to a FFS measurement on a BSA monolayer on a hydrophobized silica surface in PBS buffer. (a) Representative images of the evolution of the topography of the sample surface during the scratch. (b) and (c) Corresponding SD vs FL curve and FF vs FL curves. In (c) the frictional response obtained with the same tip on a clean substrate is also shown. (d) Topography of a wider area obtained after the scratch.

presented in Figure 2c along with a similar curve obtained with the same tip, and at the same conditions, on a clean hydrophobized silica surface. Finally Figure 2d shows a topography image of a wider area obtained after the scratch at a low applied force (