20572
J. Phys. Chem. B 2006, 110, 20572-20579
Quantification of Surface-Bound Proteins by Fluorometric Assay: Comparison with Quartz Crystal Microbalance and Amido Black Assay Paul Roach,† Neil J. Shirtcliffe,† David Farrar,‡ and Carole C. Perry*,† Interdisciplinary Biomedical Research Centre, School of Biomedical and Natural Sciences, Nottingham Trent UniVersity, Clifton, Nottingham NG11 8NS, U.K., and Smith and Nephew Group Research Centre, Heslington, York YO1 5DF, U.K. ReceiVed: April 6, 2006; In Final Form: August 9, 2006
Protein adsorption is of major and widespread interest, being useful in the fundamental understanding of biological processes at interfaces through to the development of new materials. A number of techniques are commonly used to study protein adhesion, but few are directly quantitative. Here we describe the use of Nano Orange, a fluorometric assay, to quantitatively assess the adsorption of bovine fibrinogen and albumin onto model hydrophilic (OH terminated) and hydrophobic (CH3 terminated) surfaces. Results obtained using this method allowed the calibration of previously unquantifiable data obtained on the same surfaces using quartz crystal microbalance measurements and an amido black protein assay. Both proteins were found to adsorb with higher affinity but with lower saturation levels onto hydrophobic surfaces. All three analytical techniques showed similar trends in binding strength and relative amounts adsorbed over a range of protein concentrations, although the fluorometric analysis was the only method to give absolute quantities of surfacebound protein. The versatility of the fluorometric assay was also probed by analyzing protein adsorption onto porous superhydrophobic and superhydrophilic surfaces. Results obtained using the assay in conjunction with these surfaces were surface chemistry dependent. Imbibition of water into the superhydrophilic coatings provided greater surface area for protein adsorption, although the protein surface density was less than that found on a comparable flat hydrophilic surface. Superhydrophobic surfaces prevented protein solution penetration. This paper demonstrates the potential of a fluorometric assay to be used as an external calibration for other techniques following protein adsorption processes or as a supplemental method to study protein adsorption. Differences in protein adsorption onto hydrophilic vs superhydrophilic and hydrophobic vs superhydrophobic surfaces are highlighted.
1. Introduction Protein adsorption has become a research area of great interest due to its importance in biomedical1 and biosensor applications.2 The adsorption process depends largely on the specific proteinsurface interactions that govern binding and include orientation effects as well as protein conformational change that may occur as the protein molecules adhere.3,4 Water molecules and solvated ions are also incorporated into an adsorbed protein layer, supporting electrostatic and hydrogen-bonding requirements. Advances in staining methodologies have allowed specific identification and quantification of much smaller concentrations of proteins in solution. Fluorescent dyes extend detection limits down to the ng mL-1 range compared to the more traditional methods such as the Bradford assay or ultraviolet (UV) spectroscopy, which only measure down to the µg mL-1 range. These fluorescent assays are a powerful tool for the accurate quantification of low protein concentrations in solution, although the quantification of surface-bound proteins still poses a problem. Investigations of surface-protein interactions commonly use a colloidal solid phase with a large surface area, measuring the remaining protein in solution after incubation and calculating the adsorbed amount by mass balance.5,6 Proteins labeled with a radioactive or fluorescent tag have been used,1,7 although one * To whom correspondence should be addressed. Telephone: 0044 115 8486695. Fax: 0044 115 8486616. E-mail:
[email protected]. † Nottingham Trent University. ‡ Smith and Nephew Group Research Centre.
must be aware that the inclusion of such a label may change the specific protein-surface interactions under investigation, causing unwanted conformational and/or orientational changes, thereby affecting protein adsorption characteristics. Other techniques used to study protein adsorption processes are dependent upon theoretical assumptions that may not be strictly true for such systems. The quartz crystal microbalance (QCM) is a very sensitive technique (ng cm-2) relating the mass of an adsorbing layer to an observed oscillatory frequency shift of a quartz crystal. The problem here is that frequency changes are associated with total mass loading (including water and ions)8 as well as viscoelastic contributions from the adsorbing protein.8-10 Ellipsometric techniques measure small changes in the refractive index at an interface, which can be related to the adsorbing layer thickness using an average density estimate.11 Other methods commonly used to study protein adsorption such as atomic force microscopy (AFM),12-14 infrared (IR) spectroscopy,4,11 circular dichroism,5,15 and X-ray photoelectron spectroscopy (XPS)12,16 give information on conformation and relative amounts adsorbed but do not readily give an absolute quantification of surface-bound protein. Direct assessment of surface-bound proteins using specific antibody binding assays are strongly affected by the conformational and orientational geometry of the adsorbed species.17 If the antibody binding site within the protein is blocked, then attachment may not be possible, leading to serious underestimation of the adsorbed amount of protein.
10.1021/jp0621575 CCC: $33.50 © 2006 American Chemical Society Published on Web 09/12/2006
Quantification of Surface-Bound Proteins The nature of QCM and ellipsometric techniques, along with many of the other methods used to study protein adsorption processes, limits the types of surfaces that can be analyzed, as they are unable to assess rough or porous surfaces.11 However, it has been shown that protein-surface interactions are strongly directed by the topography of the surface.3,5 Development of a label-free technique for the accurate quantification of adsorbed protein in the nanogram range, applicable over a wide range of surface textures, would broaden the possibilities for the investigation of protein adsorption. Here we present a novel assay protocol for the quantification of surface-bound protein using a fluorescent dye, Nano Orange (Molecular Probes), that can be used as a stand-alone method or serve as an external calibration in conjunction with other methods. To show the scope of the technique, we examine the adsorption of two serum proteins (fibrinogen and albumin) onto model hydrophobic and hydrophilic flat surfaces. Results obtained using QCM and amido black staining were compared to fluorescence assay data. Furthermore we investigate adsorption onto porous surfaces using superhydrophobic and superhydrophilic coatings as models to show the feasibility and potential breadth of the technique. 2. Experimental Methods 2.1. Surface Preparation. Gold substrates were prepared by sputter deposition in an Emitech K575, depositing 30 nm gold onto glass microscope slides using a 3 nm titanium bonding layer. Quartz crystal microbalance crystals with Cr/Au electrodes (Testbourne Ltd.) were cleaned before use with piranha etch solution, 1:3 H2O2:H2SO4 (Fisher Chemicals and Sigma, respectively), rinsing thoroughly with distilled, deionized water and ethanol (Haymans) before use. 2.2. Preparation of Thiol Self-Assembled Monolayers. Chemically defined surfaces were prepared using heptanethiol and mercaptoethanol (Sigma) to give hydrophobic and hydrophilic self-assembled monolayers (SAMs), respectively. Goldcoated microscope slides and QCM crystals were incubated in 1 mM ethanolic solutions of the desired thiol, incubating for a minimum of 12 h. After this time the surfaces were rinsed in ethanol to remove any nonbound thiols and dried under nitrogen immediately before use. Surface wettability was analyzed by taking images of 5 µL water droplet equilibrium contact angles on a Kru¨ss DSA10. Droplets were applied to the surface by a microsyringe with a hydrophobized needle, and images were taken immediately to eliminate drying effects. Reported values are averages of at least six contact angle measurements. 2.3. Preparation of Porous Silica Coatings. Porous coatings were prepared as described previously18 by mixing dimethylformamide (DMF, Acros 99%) as solvent (2.5 mL), hydrochloric acid [1.5 mL of 0.12 M diluted from 37% HCl (Analar, Aldrich)], and methyltriethoxysilane (MTEOS, Lancaster, 98%; 2.5 mL, 12.55 mmol) and stirring for 1 h. After this time 1.625 mL of ammonia solution (4.52 M, diluted from Fisher, 35%) was added, the liquid was rapidly mixed, 0.4 mL aliquots were placed onto glass microscope slides (76.2 by 25.4 mm), and a hydrophobic plate was placed on top. These hydrophobic plates were prepared by incubating a microscope slide in 1:20 diluted “Extreme Wash-In Solution” (Grangers) in deionized water for 10 min, rinsing in deionized water, and heating to 100 °C on a hot plate for 5 min. Glass coverslips were used as spacers to ensure reproducibility of film thickness. Gelation was allowed to proceed for at least 4 h, after which time the hydrophobic plate was removed and the gel allowed to air-dry. Films were
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20573 heat-treated at 2.5 °C min-1 to cross-link silane groups and remove residual solvent, resting at a final temperature for 100 min. Heat treatment to 200 and 450 °C gave hydrophobic and hydrophilic films, respectively. Surface area and porosity measurements were performed using a Quantachrome Nova3200 instrument, vacuum degassing samples for a minimum of 12 h at 120 °C immediately prior to analysis. Partial pressures of nitrogen in the range of 0.05-1 were covered using a 56 point adsorption and desorption profile. Multipoint BET analysis was conducted to assess surface area, and the BJH method was used to assess pore size distribution. 2.4. Protein Assay. Bovine serum albumin (BSA, fraction V, lyophilized powder, Aldrich) and bovine fibrinogen (Fg, type I-S, lyophilized powder, Fluka) were diluted in freshly prepared phosphate buffered saline (PBS) using sodium salts: NaH2PO4 and Na2HPO4 (200 mmol phosphate) and NaCl (100 mmol) obtained from Aldrich to give pH 7.4 at 25 °C. Protein solutions with concentrations in the range of 6-2000 µg mL-1 were prepared immediately before use by serial dilution of a 2.0 mg mL-1 stock solution. Protein adsorption isotherms were constructed by incubating a range of protein solution concentrations over a chosen surface. To allow a variety of surfaces to be analyzed, glass rings were attached to surfaces prepared on microscope slides using poly(vinyl acetate) (Loctite), such that each ring defined the boundary of a well in which protein solution could be incubated. In this way multiwell plates were produced specifically for each experiment. 2.5. Fluorimetry: Nano Orange. The Nano Orange protein quantification kit (Molecular probes, N6666) allows accurate determination of protein concentration in the range of 0.1-10 µg mL-1. The Nano Orange reagent is virtually nonfluorescent in aqueous solution, but upon binding to proteins the surfactant dye molecule undergoes a dramatic fluorescence enhancement,19 having a broad excitation peak centered at 470 nm and a broad emission peak centered at 570 nm. A number of studies have examined pH,20 contaminant,21 and temperature effects22 of the assay. Calibration for each protein was conducted separately, each data point being performed in triplicate. The Nano Orange working solution was prepared as per manufacturer’s instructions, using concentrated (500×) Nano Orange protein quantitation reagent and concentrated (10×) Nano Orange protein quantitation diluent. As an example, to prepare 50 mL of Nano Orange working solution, 5 mL of 10× diluent solution was mixed with 45 mL of distilled deionized water. A 100 µL aliquot of the 500× Nano Orange reagent was added and the solution again mixed thoroughly. The vials were protected from light to prevent photodegradation. Aliquots of each protein solution (200 µL) were pipetted into Eppendorf tubes, and the solvent was removed to dryness using a speed-vacuum concentrator (Stratech Scientific London). To each tube, 300 µL of Nano Orange working solution was added and vortex stirred for 20 min. The samples were heated to 96 °C for 10 min, allowed to cool to room temperature, and again vortex stirred to ensure homogeneity. A 250 µL aliquot of each sample was pipetted into a black 96-well plate (Fisher) and analyzed using a Spectra Fluor microplate reader (Tecan), with excitation/emission wavelengths set at 485/595 nm and 3-10 flashes/well with 40 µs integration time. Protein solutions in the range of 6-2000 µg mL-1 were incubated for 1 h over a specific surface (either thiol SAMs or sol-gel coatings) in the multiwell plates prepared as above,
20574 J. Phys. Chem. B, Vol. 110, No. 41, 2006
Roach et al.
after which time the surfaces were rinsed three times with distilled, deionized water to remove any unbound protein. Surface-bound protein was then detached by three sequential rinse cycles of 250 µL undiluted ethanol (∼95% Haymans) followed by 250 µL of distilled, deionized water (six rinses in total). All wash solutions were collected, and the solvent was evaporated to dryness using a speed-vacuum concentrator (Stratech). The three sequential rinse cycles were sufficient to give no further detectable protein on subsequent washings. Furthermore we examined the surfaces after protein incubation and washing and found no difference compared to blank surfaces incubated in buffer, using the amido black staining protocol and Nano Orange fluorescence assay. An independent verification of the sufficiency of the desorption protocol was performed using XPS with untreated, protein-treated, and protein-desorbed samples being examined. Analysis of the N1s region for the protein treated sample identified the presence of a layer of protein. However, both the reference and the washed samples showed barely detectable N1s intensity that did not vary from one another (within measurement limits). These data support the hypothesis that all protein adsorbed on the surface during incubation was removed during the washing procedure described, at least to the detection limits of the techniques used. 2.6. Amide Black Staining. Amido black staining is commonly used for electrophoresis to highlight protein bands in gels, although it has also been used to assess relative amounts of adsorbed protein.23 The dye molecule, naphthol blue-black, interacts with proteins, possibly forming a complex between an amine side chain and the sulfonic acid groups of the dye molecule.24,25 Protein solutions in the range of 6-2000 µg mL-1 were incubated for 1 h over a surface in the multiwell plates prepared as above, after which time the surfaces were rinsed three times with distilled, deionized water to remove any unbound protein. A 200 µL aliquot of the stain solution containing 10% methanol (Haymans), 80% distilled, deionized water, 10% glacial acetic acid (Fisher), and 1 wt % naphthol blue-black (Sigma) was added to each well and left for 5 min. Surfaces were then rinsed to remove any unbound dye using distilled, deionized water followed by three rinse cycles with a wash solution containing 38% methanol (Haymans), 60% distilled, deionized water, and 2% glacial acetic acid (Fisher). The bound dye was then detached from the surface by placing 250 µL of the eluent solution (50% each of ethanol and 50 mM sodium hydroxide containing 0.1 mM EDTA (Fisher)) into each well and leaving for 30 min. A 200 µL aliquot of liquid was then removed from each well into a clear polystyrene 96-well plate (Fisher) and analyzed by UV transmission in a Spectra Fluor microplate reader (Tecan) using a 620 nm incident filter with a 450 nm reference filter. 2.7. Quartz Crystal Microbalance Measurements. QCM technology measures frequency changes of an oscillating quartz crystal that arise due to the addition of an overlayer on the surface of the crystal. If the overlayer is rigid, the frequency change can easily be related to the adsorbed mass by means of the Sauerbrey equation:26
∆f ) -
2f 2
xFq µq
∆m
(1)
where ∆f is the change in frequency observed in Hz, m is the change in mass (g cm-2), and Fq and µq are the density and shear modulus of quartz, respectively. When using QCM for
Figure 1. Schematic of staining of surface bound proteins using specific binding dyes.
liquid studies, one must be aware of deviations from the Sauerbrey equation, due to liquid density and viscosity factors as well as viscoelastic contributions of the overlayer.14 Quartz crystal microbalance measurements were made using a Maxtek PLO10 oscillator. Dual piston liquid pumps were used to smoothly flow solutions at a rate of three volume changes per minute of the FC-550 flow cell over a 25 mm diameter 5 MHz crystal with Cr/Au plate electrodes. PBS served as a background, after which the protein solution was introduced at the same flow rate. 3. Results and Discussion 3.1. Surface Characterization. The surfaces used within this study were chosen for their contrasting chemistries. Flat surfaces treated with heptanethiol and mercaptoethanol to produce thiolated self-assembled monolayer gave water contact angles of 94 and 48°, respectively. The porous surfaces showed contact angles of 155 (superhydrophobic) and 0° (superhydrophilic), respectively. The surface area available for protein adsorption in these porous materials was 36.0 m2 g-1 obtained by calculating the difference between the total surface area measured and the surface area generated by micropores (too small for protein adsorption) within the material. 3.2. Protein Adsorption. The main limitation of standard staining methods for protein adsorption analysis is that they are not directly quantitative. After incubation with protein solution the dye is added and binds to the surface-bound protein. After rinsing to remove any unbound stain, the remaining dye is then analyzed, Figure 1. Unlike solution-phase assays a calibration curve is not easily constructed because the quantity of protein bound to the surface, and hence the corresponding amount of dye, is not known. Only relative measurements can be made.23,27 Calibration methods for the amido black assay have been suggested, which require samples containing proteins to be suspended on cellulose sheets that are air-dried and subsequently stained.28 Such methods are labor-intensive and time-consuming, and because the actual surface of investigation is not used during calibration, accurate quantification is difficult. If the dye binds specifically to certain domains or functional groups within the protein, changes in conformation and/or rearrangement may also cause a change in the amount of dye that can bind. Because it has recently been highlighted that surface topography plays an important role in controlling protein adsorption characteristics,3,4 porous surfaces have also been analyzed. The use of porous materials also highlights the limitations of common methods used to investigate protein adsorption, such as QCM and ellipsometry. 3.3. Flat Surfaces. A Nano Orange fluorescent assay has been used to quantitatively assess the amount of protein adsorbed onto a surface. Results obtained from the Nano Orange assay were compared to QCM measurements and an amido black assay of the same protein-surface systems and used to calibrate them. The flexibility of this fluorescent assay will allow many other surface analysis techniques to be calibrated. 3.3.1. Calibration Using Nano Orange and Comparison with Data Obtained Using QCM and Amido Black Assay. Saturation
Quantification of Surface-Bound Proteins
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20575
Figure 2. Saturation curves of fibrinogen adsorbing onto a model hydrophilic surface.
Figure 3. Linear form of saturation curves for QCM, amido black, and fluorescence assays. Data presented for fibrinogen adsorption onto a flat hydrophilic surface.
curves for specific protein-surface combinations can be constructed over a range of protein concentrations, giving information on protein-surface interactions. By fitting the data to a Langmuir curve, eq 2, relative protein-surface saturation amounts Qm and binding affinities K can be calculated from the shape of the curve.
Q KC ) Qm 1 + KC
(2)
where Q and C are the adsorbate surface concentration and the concentration in the phase adjacent to the surface, respectively. Fibrinogen and albumin were used as model proteins due to their relatively high abundance in plasma, which makes their investigation applicable to biomaterial evaluation. Here we show fibrinogen adsorbing onto a chemically defined hydrophilic surface as example data to illustrate calibration of QCM measurements and the amido black protein assay. Saturation curves constructed using the three techniques are shown in Figure 2. Each of the saturation curves differ slightly due to the mechanisms of the three techniques. Obviously the Qm values will differ because each of the methods gives a number relating to the quantity of protein assessed; e.g. a frequency is given for QCM. Binding affinity values should be the same for all techniques used to assess protein adsorption because this value relates to the equilibrium involving adsorbing and desorbing species. The data show differences that can be explained in part as follows; the higher value found using QCM data is most likely due to the technique measuring water trapped between bound proteins as well as the bound protein molecules themselves. Water molecules adjacent to the microbalance surface will therefore contribute to the initial observed frequency change. Later, when they are replaced by protein molecules, the overall change in frequency will be smaller, as a mass of
water will be replaced by the mass of a protein molecule. Therefore adsorption may appear to be more rapid initially and to slow earlier than it actually does. Amido black staining could be affected by the orientation or conformation of the bound protein, wherein geometric rearrangement of the protein may cause different numbers of amine groups to be accessible to the dye. The Nano Orange fluorescent assay protocol described in this paper removes the protein molecules attached to the surface after incubation, thus allowing the fluorescent dye to bind around the whole outer protein surface. Unlike the amido black assay the Nano Orange assay is less affected by surfaceinduced conformational changes of the protein because in this assay the protein is deliberately denatured. The Lineweaver-Burke linear form of each saturation curve can be obtained by plotting 1/Q vs 1/C over a range of concentrations. Any deviation from an ideal Langmuir curve can then be assessed. The main assumptions within the Langmuir model are that the surface is completely homogeneous with each vacant surface site filled by one adsorbing species and also that there are no lateral interactions between bound species. Although the Langmuir model is generally a good fit for most protein adsorption systems, proteins that spread or denature as they bind to the surface are not accounted for and lateral interactions in protein layers are likely. Deviations from the Langmuir curve may also occur due to inaccuracies in measurement at the lower limits of detection. For this reason only those points lying on a linear region have been used, as shown in Figure 3. The amount of protein adsorbed onto the hydrophilic surface is directly related to the concentration of the incubated protein solution by the equation of the Nano Orange linear Langmuir plot, Figure 3c. Using this information a plot relating the QCM frequency shift and amido black absorbance units to the mass of fibrinogen adsorbed can be constructed, Figure 4.
20576 J. Phys. Chem. B, Vol. 110, No. 41, 2006
Roach et al.
Figure 4. Calibration plot for QCM and amido black. Data obtained for fibrinogen adsorption onto a flat hydrophilic surface.
Figure 5. Saturation curves for fibrinogen and albumin onto “flat” hydrophobic and hydrophilic surfaces obtained using the amido black assay.
TABLE 1: Data Obtained from QCM Measurements3
TABLE 2: Data Obtained from Amido Black Assay
mass of adsorbed protein from Nano Orange assay/(ng cm-2) saturation freq Sauerbrey mass using OH using CH3 value/Hz loading/(ng cm-2) calibration calibration Fg OHa Fg CH3a BSA OH BSA CH3
102.0 96.2 47.2 40.9
1805.4 1702.7 835.4 723.9
377 280 240 218
341 289 337 289
a OH and CH3 represent hydrophilic and hydrophobic surfaces, respectively.
3.3.2. Fibrinogen and Albumin onto Hydrophilic Vs Hydrophobic Surfaces. To further establish the use of the Nano Orange assay method for the quantification of surface-bound protein, the adsorption characteristics of fibrinogen and albumin on hydrophilic and hydrophobic surfaces were compared. It should be noted that protein-protein variability may cause differences in detection efficiency; therefore calibration should be conducted for each protein used. The adsorption profiles and saturation curves for albumin and fibrinogen onto heptanethiol and mercaptoethanol surfaces obtained using QCM have been previously reported.3 It was found that proteins binding strongly to surfaces become deformed, spreading over the surface and giving rise to lower total amounts adsorbed. 3.3.2.1. QCM. QCM results showed that both albumin and fibrinogen adsorbed more rapidly onto hydrophobic surfaces, although this effect was much more pronounced for albumin. A lower saturation level was observed for both proteins adsorbing onto hydrophobic substrates, suggesting surface-induced deformation due to hydrophobic binding. Mass loading calculated using the Sauerbrey equation gave values in the region of 1720 and 780 ng cm-2 for fibrinogen and albumin, respectively, Table 1. The fraction of this mass contribution arising from entrapped water and viscoelastic effects cannot be determined from QCM frequency shifts. Other researchers often use a combination of techniques to gain a more complete picture about the adsorbing species. Here we used the Nano Orange calibration of the QCM to assess the actual protein mass absorbed, finding values similar to those reported by others using surface plasmon resonance29 and radiolabeling30 techniques. Small differences were observed when cross-calibration was checked, i.e., using the calibration plot for fibrinogen onto hydrophilic surfaces to assess the amount of fibrinogen adsorbed onto a hydrophobic surface and
Fg OH Fg CH3 BSA OH BSA CH3
saturation value/102 abs 620 nm
mass of adsorbed protein from Nano Orange assay/(ng cm-2)
7.743 ( 0.224 6.531 ( 0.349 5.224 ( 0.751 2.981 ( 0.597
450 ( 35 285 ( 26 301 ( 1 271 ( 11
vice versa. Results show that fibrinogen calibrations were almost consistent but much larger variation was observed for albumin, Table 1. This is possibly because albumin deforms to a greater extent than fibrinogen.3 3.3.2.2. Amido Black Assay. As with QCM results the amido black assay showed an increase in the amount of protein adsorbed as protein concentration in the incubating solution was increased, Figure 5. The amido black assay, in good agreement with QCM data, showed that fibrinogen and albumin adsorbed in higher amounts to hydrophilic surfaces, Table 2. 3.3.2.3. Nano Orange Assay. Saturation curves constructed using Nano Orange show the same trends as those for QCM3 and amido black, Figure 6. The Nano Orange assay is not affected by entrapped water mass, viscoelasctic contributions, or the geometry of the adsorbing protein molecules and should therefore give more accurate information on surface-protein interactions. The binding affinities of fibrinogen and albumin adsorbing onto hydrophilic and hydrophobic surfaces are compared in Table 3. All three techniques show that both proteins tested have a higher affinity toward hydrophobic surfaces, although they bind with lower saturation values. This is consistent with the hypothesis that proteins may deform to a greater extent upon binding to a surface with a high protein-surface affinity. Examination of the protein binding affinity of the different proteins to the hydrophobic versus hydrophilic surfaces was found to vary in a similar fashion, irrespective of the analysis techniques used. Thus for fibrinogen the affinity to the hydrophobic versus hydrophilic surface was found to be approximately 270% higher from the fluorometric assay, 230% from QCM, and 200% from the amido black assay. Albumin also showed a greater affinity toward hydrophobic surfaces. The protein adsorption trends observed by QCM and amido black assay were verified by the Nano Orange assay, although the differences reported between the techniques highlight the possibility for misinterpretation when using the nonquantitative techniques in isolation.
Quantification of Surface-Bound Proteins
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20577
Figure 6. Saturation curves for (a) fibrinogen and (b) albumin onto “flat” hydrophobic and hydrophilic surfaces obtained from Nano Orange assay. Solid lines show fitted Langmuir curves.
TABLE 3: Data Obtained from Nano Orange Fluorometric Assay
Fg OH Fg CH3 BSA OH BSA CH3
QCM binding affinity
amido black binding affinity
Nano Orange binding affinity
10.9 36.0 5.3 5.4
3.88 ( 0.34 11.58 ( 2.90 1.10 ( 0.33 1.23 ( 0.53
1.43 ( 0.29 5.33 ( 1.01 2.48 ( 1.05 4.00 ( 0.86
3.4. Porous Surfaces. Many techniques used to study protein adhesion processes cannot be employed when examining rough or porous surfaces. Problems associated when using rough surfaces with QCM are well-known. Liquid may become entrapped at the surface, acting as a rigid mass causing a crystal response.31,32 Other mechanisms accounting for this differing frequency response have been proposed, including interfacial slip where the liquid slips across the surface of the crystal due to the shear mode of oscillation.33,34 Ellipsometry also cannot be used to examine rough or porous surfaces because adsorption inside a material cannot be observed.11 With the recent interest in superhydrophobicity and superhydrophilicity18,35,36 and protein adsorption being shown to be controlled by surface topography,4,37 porous materials are of potential interest for biomedical and biosensor applications. In this study Nano Orange and amido black assays have been used to investigate protein adsorption onto these surfaces. Porous silica materials were used that can be produced as superhydrophobic surface coatings and by thermal treatment be switched to become superhydrophilic,18 thereby allowing aqueous phases to readily penetrate into the pore network. The porosity of the material used was characterized by nitrogen gas adsorption, having predominantly micropores that proteins would be too large to enter. The external surface area available for protein adsorption was calculated by the difference between the total and micropore surface area and was found to be 36.0 m2 g-1. Therefore, when coated with a superhydrophilic coating, each well in which protein solution was incubated had approximately 440 times the available surface area compared to flat surfaces tested. Aqueous solutions in contact with superhydrophobic films are suspended on the top of the film by bridging-type wetting,38 Figure 7, and therefore the actual surface area available for protein adsorption (i.e. that fraction of the surface in contact with the aqueous layer) could be considered to be lower than the surface area of a flat surface. The data obtained in these studies have been normalized to the total surface area available because the actual water contact area on the superhydrophobic material is difficult to define. Data from the fluorometric analysis were normalized to the total available surface area per well, although discussion will
Figure 7. Wetting on (a) hydrophobic and (b) superhydrophobic surfaces.
also include comparison to the flat surfaces investigated as described above. Much lower saturation values were observed for both proteins adsorbing onto superhydrophilic surfaces compared to hydrophilic flat surfaces, Figure 8a, Table 4. Again data obtained from amido black assay follow trends similar to those obtained using the fluorescence assay. This difference in the amount of protein adsorbed may be due to differences in protein transport to the surface of the material. Transport of a protein molecule to a flat surface would depend largely on its diffusion through the liquid to the solidliquid interface. When considering porous materials, most of the surface area is within the bulk. and although the protein solution can penetrate into the porous network, a longer period of time will be required for the protein adsorption process to reach equilibrium. The superhydrophobic materials show a much lower amount of protein adsorbed per unit surface area. This is to be expected, due to the bridge-type wetting as described above. When recalculated using the surface area of a flat substrate, the amount of protein adsorbed is greater than would be expected; 495 ng cm-2 compared to 285 ng cm-2 of fibrinogen on superhydrophobic and flat hydrophobic surfaces, respectively. Albumin follows the same trend adsorbing in 397 and 271 ng cm-2 onto superhydrophobic and flat hydrophobic surfaces, respectively. The reason that these values exceed those for flat surfaces is most likely because the surface-bound protein acts as a surfactant, allowing the aqueous phase to penetrate further into the pore network than water would, thus allowing the proteincontaining solution to interact with a larger than initially expected surface area.39 Binding constants show that fibrinogen has a greater affinity toward superhydrophobic compared to superhydrophilic surfaces. Albumin in contrast to fibrinogen is shown to have a higher affinity toward the superhydrophilic compared to superhydrophobic surfaces.
20578 J. Phys. Chem. B, Vol. 110, No. 41, 2006
Roach et al.
Figure 8. Saturation curves for fibrinogen and albumin onto (a) superhydrophilic and (b) superhydrophobic surfaces obtained using the Nano Orange assay. Lines show fitted Langmuir curves.
TABLE 4: Fibrinogen and Albumin Adsorption Parameters onto Superhydrophobic and Superhydrophilic Surfaces by Fluorometric Analysis and Amido Black Assay
Fg OH Fg CH3 BSA OH BSA CH3
binding affinity from Nano Orange
saturation value/ (ng cm-2)
0.76 ( 0.20 3.62 ( 1.32 5.08 ( 0.67 3.34 ( 1.23
4.62 ( 0.63 0.70 ( 0.07 3.09 ( 0.10 0.56 ( 0.06
binding affinity saturation from amido value/ black assay 102 abs 2.47 ( 0.60 4.60 ( 0.98 1.38 ( 0.45 1.04 ( 0.32
54.8 ( 4.9 6.0 ( 3.6 36.5 ( 5.4 10.4 ( 1.5
4. Discussion A fluorescence assay has been used to quantify the amount of protein adsorbed onto surfaces and the results used to calibrate other methods. Values obtained correspond well with those expected from previous reports.29,30 Results showed trends in protein adhesion parameters that were common to all three methods used (Tables 1-3). A greater quantity of fibrinogen compared to albumin adsorbed on surfaces having the same chemistry, with both proteins generally having higher affinity but lower saturation values on hydrophobic surfaces. This may indicate that hydrophobic surfaces induce greater deformation of adsorbing protein molecules. Surface-protein hydrophobic bonding, which is known to play a dominant role in surface-protein interactions, possibly drives the adsorption process and structural change.40 Some differences between the methods were highlighted, which may be due to the nonspecific binding of the fluorescent dye used. Compared to QCM, the Nano Orange assay showed a greater difference between the amount of fibrinogen adsorbed onto hydrophilic compared to hydrophobic surfaces, which may indicate that a large proportion of the QCM frequency shift arises from the viscoelastic properties of the adsorbing protein layer. The levels of albumin adsorbed onto the two surface chemistries examined correlate well for QCM (viscoelastic effects are expected to be smaller than for fibrinogen due to the difference in size) and fluorimetry, although the amido black assay results differ, possibly due to the specific binding requirements of the stain. Binding affinities assessed by all three methods showed good correlation, with fibrinogen having a higher affinity toward hydrophobic compared to hydrophilic surfaces. Albumin also has a higher affinity toward hydrophobic surfaces, with values reported by all three methods being in good agreement. The Nano Orange and amido black assays allowed protein adsorption to be assessed on superhydrophilic and superhydrophobic materials. Results showed that the surface density of protein when normalized to surface area was much less on
superhydrophilic surfaces than would have been expected. This could be due to protein transport through the porous material. A much greater amount of protein was observed on superhydrophobic surfaces than was expected, which may be explained by the surfactant nature of the adsorbing layer, although this requires further investigation. Binding affinities for proteins adsorbing onto porous materials were modified from those found on flat surfaces. Fibrinogen showed a greater affinity toward the superhydrophobic compared to superhydrophilic materials, but values were ∼50 and 70% lower than those on flat hydrophilic and hydrophobic surfaces, respectively. Albumin was found to have a higher affinity toward the superhydrophilic compared to the superhydrophobic surface. 5. Conclusions The key problem with many techniques used to study protein adsorption is that they are not directly quantitative. Other problems often prevent rough or porous surfaces being investigated. We have described a rapid, versatile method for the quantification of adsorbed protein. The method can be used for the investigation of protein adsorption onto any surface type or used to calibrate other surface analysis techniques. A protocol has been developed wherein a standard off-the-shelf fluorometric assay can be used either in collaboration with other methods, acting as an external calibration, or as a stand-alone assay as a novel method for constructing protein saturation curves. The benefits of this method over current techniques are the absolute quantitation of protein loading and the different types of surfaces that can be investigated. Acknowledgment. We gratefully acknowledge Smith and Nephew for funding. Miss E. F. Smith and Dr. M. Alexander at the University of Nottingham are thanked for XPS analysis, and EPSRC is acknowledged for funding of the XPS equipment used. We would like to thank Professor G. McHale and Dr. M. I. Newton from the Physics Division at Nottingham Trent University for access to equipment and Dr. S. V. Patwardhan for useful discussions. References and Notes (1) Shen, M.; Horbett, T. A. J. Biomed. Mater. Res. 2001, 57 (3), 336345. (2) Nam, J. M.; Thaxton, C. S.; Mirkin, C. A. Science 2003, 301, 18841886. (3) Roach, P.; Farrar, D.; Perry, C. C. J. Am. Chem. Soc. 2005, 127, 8168-8173. (4) Barbucci, R.; Lamponi, S.; Magnani, A. Biomacromolecules 2003, 4, 1506-1513.
Quantification of Surface-Bound Proteins (5) Veryegel, A. A.; Siegel, R. W.; Dordick, J. S. Langmuir 2004, 20, 6800-6807. (6) Kondo, A.; Fukuda, H. J. Colloid Interface Sci. 1998, 198, 34-41. (7) Horbett, T. A.; Cheng, C. M.; Ratner, B. D.; Hoffman, A. S. J. Biomed. Mater. Res. 1986, 20, 739-772. (8) Galli Marxer, C.; Collaud Coen, M.; Bissig, H.; Greber, U. F.; Schlapbach, L. Anal. Bioanal. Chem. 2003, 371 (3), 570-577. (9) Lyle, E.-L.; Hayward, G. L.; Thompson, M. Analyst 2002, 127, 1596-1600. (10) McHale, G.; Lucklum, R.; Newton, M. I.; Cowen, J. A. J. Appl. Phys. 2000, 88 (12), 7304-7312. (11) Tengevall, P.; Lundstro¨m, I.; Liedberg, B. Biomaterials 1998, 19, 407-422. (12) Denis, F. A.; Hanarp, P.; Sutherland, D. S.; Gold, J.; Mustin, C.; Rouxhet, P. G.; Dufreˆne, Y. F. Langmuir 2002, 18, 819-828. (13) Sethuraman, A.; Han, M.; Kane, R. S.; Belfort, G. Langmuir 2004, 20, 7779-7788. (14) Zhou, C.; Friedt, J.-M.; Angelova, A.; Choi, K.-H.; Laureyn, W.; Frederix, F.; Francis, L. A.; Campitelli, A.; Engelborghs, Y.; Borghs, G. Langmuir 2004, 20, 5870-5878. (15) Jiang, X.; Jiang, J.; Jin, Y.; Wang, E.; Dong, S. Biomacromolecules 2005, 6, 46-53. (16) Sweryda-Krawiec, B.; Devaraj, H.; Jacob, G.; Hickman, J. J. Langmuir 2004, 20, 2054-2056. (17) Wittmer, C. R.; Van Tassel, P. R. Colloids Surf., B 2005, 41, 103109. (18) Shirtcliffe, N. J.; McHale, G.; Newton, M. I.; Perry, C. C.; Roach, P. Chem. Commun. (Cambridge) 2005, 25, 3135-3137. (19) Criordano, B. C.; Jin, L.; Couch, A. J.; Ferrance, J. P.; Landers, J. P. Anal. Chem. 2004, 76, 4705-4714. (20) Harvey, M. D.; Bablekis, V.; Banks, P. R.; Skinner, C. D. J. Chromatogr. B: Biomed. Sci. Appl. 2001, 754, 345-356. (21) Jones, L. J.; Haugland, R. P.; Singer, V. L. Biotechniques 2003, 34, 850-861.
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20579 (22) Silberblatt, S.; Felder, R. A.; Mifflin, T. E. JALA 2001, 6 (6), 8388. (23) Jones, M. I.; McColl, I. R.; Grant, D. M.; Parker, T. L. J. Biomed. Mater. Res. 2000, 52, 413-421. (24) Tsopelas, C.; Sutton, R. J. Nucl. Med. 2002, 43, 1377-1382. (25) Wilbur, D. Scott J. Nucl. Med. 2003, 44, 1540-1541. (26) Sauerbrey, G. Z. Phys. 1959, 155, 206-222. (27) Whittle, J. D.; Bullett, N. A.; Short, R. D.; Douglas, C. W. I.; Hollander, A. P.; Davies, J. J. Mater. Chem. 2002, 12, 2726-2732. (28) Dieckmann-Schuppert, A.; Schnittler, H.-J. Cell Tissue Res. 1997, 288 (1), 119-126. (29) Sigal, G. B.; Mrksich, M.; Whitesides, G. M. J. Am. Chem. Soc. 1998, 12, 3464-3473. (30) Young, B. R.; Pitt, W. G.; Cooper, S. L. J. Colloid Interface Sci. 1988, 125, 246-260. (31) Martin, S. J.; Frye, G. C.; Ricco, A. J.; Senturia, S. D. Anal. Chem. 1993, 65, 2910-2922. (32) Martin, S. J. Faraday Discuss. 1997, 107, 463-476. (33) Thompson, M.; Kipling, A. L.; Duncan-Hewitt, W. C.; Rajakovic, L. V.; Cavic-Vlasak, B. A. Analyst 1991, 116, 881-890. (34) Yang, M. S.; Thompson, M.; Duncan-Hewitt, W. C. Langmuir 1993, 9, 802-811. (35) Quere, D. Physica A (Amsterdam) 2002, 313 (1-2), 32-46. (36) Quere, D. Nat. Mat. 2003, 2, 457-460. (37) Roach, P.; Farrar, D.; Perry, C. C. J. Am. Chem. Soc. 2006, 128, 3939-3945. (38) Cassie, A. B. D.; Baxter, S. Trans. Faraday Soc. 1944, 40, 546551. (39) Storov, V. M. J. Colloid Interface Sci. 2004, 270, 180-186. (40) Norde, W. In Surface and Interfacial Aspects of Biomedical Polymers, 2nd ed.; Andarde, J. D., Ed.; Plenum Press: New York; 1985.
