1204
Langmuir 2008, 24, 1204-1211
Profiling Protein-Surface Interactions of Multicomponent Suspensions via Flow Cytometry† Darby Kozak, Annie Chen, and Matt Trau* Centre for Nanotechnology and Biomaterials, LeVel 5 East, Australian Institute for Bioengineering and Nanotechnology, The UniVersity of Queensland, St. Lucia, QLD 4072, Australia ReceiVed June 22, 2007. In Final Form: October 22, 2007 This study presents the use of flow cytometry as a high-throughput quantifiable technique to study multicomponent adsorption interactions between proteins and surfaces. Flow cytometry offers the advantage of high-throughput analysis of multiple parameters on a very small sampling scale. This enables flow cytometry to distinguish between individual adsorbent particles and adsorbate components within a suspension. As a proof of concept study, the adsorption of three proteinssbovine serum albumin (BSA), bovine immunoglobulin gamma (IgG) and fibrinogensonto five surfacemodified organosilica microsphere surfaces was used as a model multicomponent system for analysis. By uniquely labeling each protein and solid support type with spectrally distinguishable fluorescent dyes, the adsorption process could be “multiplexed” allowing for simultaneous screening of multiple adsorbate (protein) and adsorbent (particle surface) interactions. Protein adsorption experiments quantified by flow cytometry were found to be comparable to single-component adsorption studies by solution depletion. Quantitative distribution of the simultaneous competitive adsorption of BSA and IgG indicated that, at concentrations below surface saturation, both proteins adsorbed onto the surface. However, at concentrations greater than surface saturation, BSA preferentially adsorbed. Multiplexed particle suspensions of optically encoded particles were modified to produce a positively and negatively charged surface, a grafted 3400 MW poly(ethylene glycol) layer, or a physisorbed BSA or IgG layer. It was observed that adsorption was rapid and irreversible on all of the surfaces, and preadsorbed protein layers were the most effective in preventing further protein adsorption.
Introduction The adsorption of biomolecules through specific and nonspecific interactions is important in industrial, medical, and diagnostic applications. The specific interactions between biomolecules, such as complementary DNA strands1 or antibodyantigen pairs,2 are routinely and increasingly used for medical diagnostics.3 Applications include detecting the presence, susceptibility, or progression of a disease or biological processes within the body.4 However, the accuracy and reliability of these molecular diagnostic devices are currently hampered by their sensitivity or signal-to-noise ratio. By far, the greatest contributing factor to reducing assay sensitivity is the nonspecific adsorption of assay or unassociated biomolecules.5 A variety of methods and “antifouling” layers have been employed to reduce the amount of nonspecifically adsorbed molecules on assay surfaces. These include diluting and purifying the sample to remove potential competitive and strongly adsorbing species or the addition of a known protein to block potential nonspecific adsorption sites prior to analysis. Modification of the assay surface through the grafting of hydrophilic polymer layers has been effective in reducing nonspecific adsorption.6 Poly(ethylene glycol) (PEG),7 †
Part of the Molecular and Surface Forces special issue. * Corresponding author. E-mail:
[email protected]. Telephone: +61 7 334 64173. Fax: +61 7 334 63973. (1) Kononen, J.; Bubendorf, L.; Kallioniemi, A.; Barlund, M.; Schraml, P.; Leighton, S.; Torhorst, J.; Mihatsch, M. J.; Sauter, G.; Kallioniemi, O. P. Nat. Med. 1998, 4 (7), 844-847. (2) Polascik, T. J.; Oesterling, J. E.; Partin, A. W. J. Urol. 1999, 162 (2), 293-306. (3) Srinivas, P. R.; Verma, M.; Zhao, Y. M.; Srivastava, S. Clin. Chem. 2002, 48 (8), 1160-1169. (4) Hartwell, L.; Mankoff, D.; Paulovich, A.; Ramsey, S.; Swisher, E. Nat. Biotechnol. 2006, 24 (8), 905-908. (5) Herrwerth, S.; Rosendahl, T.; Feng, C.; Fick, J.; Eck, W.; Himmelhaus, M.; Dahint, R.; Grunze, M. Langmuir 2003, 19 (5), 1880-1887. (6) Kingshott, P.; Griesser, H. J. Curr. Opin. Solid State Mater. Sci. 1999, 4 (4), 403-412.
polysaccharides8 and polyionic9 layers have all been successfully incorporated as effective antifouling layers. However, the effects of the polymer characteristics such as chemical composition,10 structure,11,12 molecular weight,13 and grafting density14 on the mechanism of steric repulsion is still relatively unknown.15-17 Thus a greater understanding of biomolecule adsorption, both specifically and nonspecifically, is important when considering the development of more sensitive biological diagnostics for the future. Herein, we describe a flow cytometric-based system for the characterization of protein surface interactions. Flow cytometric detection offers the advantages of (i) the small sampling volumes required for biological screening, (ii) multi-parameter analysis on an individual particle basis, which gives rise to quantitative distinction between adsorbate components and individually distinguishable particles within a mixed multicomponent suspension, and (iii) a high-throughput, good mixing, and convenient sampling nature. (7) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; Degennes, P. G. J. Colloid Interface Sci. 1991, 142 (1), 149-158. (8) Bosker, W. T. E.; Patzsch, K.; Stuart, M. A. C.; Norde, W. Soft Matter 2007, 3 (6), 754-762. (9) Lu, J. R.; Murphy, E. F.; Su, T. J.; Lewis, A. L.; Stratford, P. W.; Satija, S. K. Langmuir 2001, 17 (11), 3382-3389. (10) McArthur, S. L.; McLean, K. M.; Kingshott, P.; St. John, H. A. W.; Chatelier, R. C.; Griesser, H. J. Colloids Surf., B: Biointerfaces 2000, 17 (1), 37-48. (11) Ajikumar, P. K.; Kiat, J.; Tang, Y. C.; Lee, J. Y.; Stephanopoulos, G.; Too, H. P. Langmuir 2007, 23 (10), 5670-5677. (12) Sofia, S. J.; Premnath, V.; Merrill, E. W. Macromolecules 1998, 31 (15), 5059-5070. (13) Archambault, J. G.; Brash, J. L. Colloids Surf., B: Biointerfaces 2004, 33 (2), 111-120. (14) Malmsten, M.; Emoto, K.; Van Alstine, J. M. J. Colloid Interface Sci. 1998, 202 (2), 507-517. (15) Carignano, M. A.; Szleifer, I. Colloids Surf., B: Biointerfaces 2000, 18 (3-4), 169-182. (16) Currie, E. P. K.; Norde, W.; Stuart, M. A. C. AdV. Colloid Interface Sci. 2003, 100, 205-265. (17) Morra, M. J. Biomater. Sci., Polym. Ed. 2000, 11 (6), 547-569.
10.1021/la701847x CCC: $40.75 © 2008 American Chemical Society Published on Web 12/08/2007
Protein-Surface Interaction Study Via Flow Cytometry
A wide variety of adsorption systems and characterization methods have been employed to study protein adsorption at interfaces. The greatest focus has been given to the three most common plasma proteins: albumin, immunoglobulin gamma (IgG), and fibrinogen. These proteins are abundant in serum and serve as diagnostic indicators.18,19 Furthermore, there exists a correlation between the adsorption of these proteins and the relative biocompatibility of a surface.20 Previous studies have investigated a range of adsorption conditions, including different media as well as different adsorbate and adsorbent components. The effects of adsorption environment such as pH,21,22 ionic strength,23,24 and temperature25 have been shown to effect the maximum adsorbed amount of protein. At conditions below the protein denaturing temperature, protein adsorption is typically characterized by a Type I “Langmuir-like” isotherm.26 Changing the substrate compositions and conditions such as hydrophobicity, hydrophilicity, and surface charge typically gives rise to changes in the maximum adsorbed amount of protein but does not alter the shape of the isotherm. Protein adsorption has been characterized by a number of direct and indirect techniques, each offering unique advantages for characterizing the adsorption process. Direct measurements such as reflectometry and atomic force microscopy can provide accurate and real-time information about the adsorbed layer formation and thickness and the relative orientation or forces between a protein and surface. Indirect measurements of protein adsorption by the decrease in adsorbate concentration in the solution are typically measured spectroscopically at 270-290 nm (the λmax for most proteins) or through colorimetric tests such as Coomassie Blue (Bradford assay).27 However, all of these methods are limited in their ability to differentiate between the adsorbing species in complex mixtures and generally cannot screen different adsorbent surfaces rapidly and comparatively. Through miniaturization, optical encoding strategies, and highthroughput analysis, these restrictions can be overcome. Toward this end, the application of flow cytometry as a rapid and quantitative technique to characterize single and multicomponent adsorbate and adsorbent surfaces is proposed. Flow cytometry is a high-throughput screening platform commonly used to characterize and segregate cells in biological samples, and is now routinely applied to the rapid analysis of small, multiplexed particle-based assays.28 The strength of flow cytometry as an analysis and diagnostic tool lies in its ability to detect multiple optical parameters on a particle-by-particle basis. With the ability to spectrally differentiate between particles by “optical encoding”, flow cytometry offers a unique ability to simultaneously study multiple competitive or cooperative adsorbate-adsorbent interactions.29 The simultaneous adsorption (18) Himmelfarb, J.; McMonagle, E. Kidney Int. 2001, 60 (1), 358-363. (19) Wick, M. Monoclonal Antibodies in Diagnostic Immunohistochemistry; Marcel Dekker, Inc.: New York, 1988; Vol. 24, p 664. (20) Ratner, B. D.; Bryant, S. J. Annu. ReV. Biomed. Eng. 2004, 6, 41-75. (21) Su, T. J.; Lu, J. R.; Thomas, R. K.; Cui, Z. F. J. Phys. Chem. B 1999, 103 (18), 3727-3736. (22) Norde, W.; Anusiem, A. C. I. Colloids Surf. 1992, 66 (1), 73-80. (23) Giacomelli, C. E.; Avena, M. J.; DePauli, C. P. J. Colloid Interface Sci. 1997, 188 (2), 387-395. (24) Martinrodriguez, A.; Cabrerizovilchez, M. A.; Hidalgoalvarez, R. Colloids Surf., A: Physicochem. Eng. Aspects 1994, 92 (1-2), 113-119. (25) Kiss, E. Colloids Surf., A: Physicochem. Eng. Aspects 1993, 76, 135140. (26) Sing, K. S. W.; Everett, D. H.; Haul, R. A. W.; Moscou, L.; Pierotti, R. A.; Rouquerol, J.; Siemieniewska, T. Pure Appl. Chem. 1985, 57 (4), 603-619. (27) Syrovy, I.; Hodny, Z. J. Chromatogr.: Biomed. Appl. 1991, 569 (1-2), 175-196. (28) Battersby, B. J.; Lawrie, G. A.; Johnston, A. P. R.; Trau, M.Chem. Commun. 2002, (14), 1435-1441. (29) Earley, M. C.; Vogt, R. F.; Shapiro, H. M.; Mandy, F. F.; Kellar, K. L.; Bellisario, R.; Pass, K. A.; Marti, G. E.; Stewart, C. C.; Hannon, W. H. Cytometry 2002, 50 (5), 239-242.
Langmuir, Vol. 24, No. 4, 2008 1205
of an adsorbate onto multiple adsorbents (multiplexed adsorption) can be used to characterize the adsorption process and relative adsorption affinity for each surface. Additionally, multicomponent adsorption is also possible by tagging each of the adsorbate species with spectrally distinguishable fluorophore tags. This enables the quantitative examination of competitive or cooperative adsorption between the adsorbates. As a result, the displacement of one protein species by another or the relative concentration of each type of protein at the surface can be determined. Furthermore, by using both optically encoded particles and proteins with unique fluorescent tags, multicomponent and multiplexed adsorption processes can be simultaneously analyzed. This multicomponent analysis strategy could prove to be a valuable tool in the high-throughput screening and characterization of protein surface interactions for industrial applications, biomedical devices, and sensors. In this study, single-component adsorptions of three common serum proteins were examined by flow cytometry and compared to solution depletion results. This was then extended to quantifying the adsorption distribution of a multicomponent protein solution by using spectrally distinguishable fluorescent tagging of the different protein species. Ultimately, protein adsorption onto five adsorbent surfaces was simultaneously screened in a multiplexed format. Materials and Methods Bovine serum albumin (BSA), bovine IgG, and bovine fibrinogen were purchased from Sigma Aldrich as lyophilized powders. IgG, BSA, and fibrinogen were fluorescently tagged with N-hydroxysuccinimide ester functionalized Atto-tech 390, 488, and 550, respectively. All of the dyes were purchased as pure lyophilized powders from Atto-tech (Germany). Proteins were tagged by making a solution of 2-10 mg/mL of protein in 0.1 M sodium bicarbonate buffer (pH 8.3), which was then mixed with 20 µL of the N-hydroxysuccinimide ester-functionalized dye dissolved in dimethylformamide (DMF; approximate concentration 0.5 mg/mL) for 2 h. The tagged protein solution was then dialyzed against a standard phosphate-buffered saline (PBS) solution (pH: 7.4, 137 mM NaCl) for 3 days to remove any unreacted dye. The degree of fluorophore labeling of the proteins was determined from the proteinto-fluorophore concentration ratio using ultraviolet-visible (UVvis) spectroscopy. 3-Mercaptopropyl trimethoxysilane and 3-aminopropyl trimethoxysilane (APS) were purchased from Lancaster (U.K.) and Sigma Aldrich, respectively, and were used as received to synthesize the organosilica microspheres as per the protocol outlined in the work of Miller et al.33 The microspheres were then covalently labeled with optically distinctive concentrations of Atto-tech 620 to create five optically distinct microsphere populations. The five optically distinct particle populations were then surface-modified with either (i) a blank surface (APS-coated), (ii) physisorbed IgG, (iii) physisorbed BSA, (iv) covalently coupled 3400 MW trifluoroethanesulfonyl-activated PEG (tresylated PEG) in PBS buffer overnight, or (v) grafted 12-hydroxydodecanoic acid by N,N′diisopropylcarbodiimide in DMF solvent (Table 1). N,N′-diisopropylcarbodiimide, trifluoroethanesulfonyl chloride, 3400 MW PEG and 12-hydroxydodecanoic acid were purchased from Sigma Aldrich and used as received. Tresylated PEG was synthesized as per the protocol outlined in the work of Sofia et al.12 APS-coated microsphere samples for solution depletion measurements were dried overnight at 60 °C from an ethanol solution to (30) Riquelme, B. D.; de Isla, N. G.; Valverde, J. R.; Stoltz, J. F. J. Biochem. Biophys. Methods 2006, 68 (1), 31-42. (31) Chatelier, R. C.; Sawyer, W. H. J. Biochem. Biophys. Methods 1987, 15 (1), 49-61. (32) Chatelier, R. C.; Ashcroft, R. G. Cytometry 1987, 8, (6), 632-636. (33) Miller, C. R.; Vogel, R.; Surawski, P. P. T.; Jack, K. S.; Corrie, S. R.; Trau, M. Langmuir 2005, 21 (21), 9733-9740.
1206 Langmuir, Vol. 24, No. 4, 2008
Kozak et al.
Table 1. Surface-Modified Layers of Optically Encoded Particles surface modification (i) (ii) (iii) (iv) (v)
APS-coated preadsorbed IgG-coated preadsorbed BSA-coated 3400 MW PEG 12-hydroxydodecanoic acid
surface layer effect
surface charge/mV
positive surface charge standard capture protein
18 ( 3.6 24 ( 2.4
standard surface blocking protein
-9 ( 6.5
standard antifouling polymer layer negative surface charge
-28 ( 1.2 -34 ( 7.9
form a fine white powder. Solution depletion measurements were conducted by incubating 100 mg of the microspheres in a standard PBS solution containing increasing concentrations of nonfluorescently tagged proteins for 2 h. The particles were separated by centrifugation, and the supernatant concentration was analyzed by the Coomassie Blue (Bradford assay) colorimetric protein detection method at 580 nm using a Perkin-Elmer, Lambda 2 double-beam UV-vis spectrophotometer. Coomassie Blue calibration curves were generated for each protein species used, and adsorption samples were diluted to concentrations within the linear Beer-Lambert portion of the calibration curve. Flow cytometry adsorption measurements were made on a DakoCytomation Mo Flo with three laser excitation lines and six optical parameters, excluding forward and side scatter. The laser specifications were a Coherent Innova 90C Argon ion laser (emission: 366 nm, power: 50 mW) with corresponding 450 ( 32 nm and 530 ( 20 nm detectors, an iCyt Visionary Bioscience, Inc. laser (emission: 488 nm, power: 100 mW) with corresponding 530 ( 20 nm and 580 ( 10 nm detectors, and a 635 nm diode laser at 12 mW power with corresponding 670 ( 15 nm and 710 ( 10 nm detectors. Adsorption measurements by flow cytometry were conducted on 1 mg of microspheres incubated with increasing concentrations of fluorescently tagged protein in a total solution volume of 0.10 mL for 2 h. The number of beads in each (1 mg) sample were measured by a hemocytometer taking an average of three measurements. Fluorescence histograms were composed of at least 10 000 particles and were analyzed using Summit V4.1 software. A four-point calibration curve for each protein adsorbed onto APScoated particles was run each day prior to experimentation. Vertical error bars are calculated from the standard deviation of three sample repetitions. Horizontal error bars are the standard error calculated for each point.
Results and Discussion Solution Depletion Isotherms. Protein adsorption isotherms generated by measuring the amount of protein remaining in solution (solution depletion) were used to characterize the adsorption process on unmodified APS-coated particles. These isotherms were also used as a comparison to the isotherms generated by flow cytometry. The adsorption of the three serum proteinssBSA, IgG, and fibrinogensonto APS coated particles (Figure 1) indicated that all of the proteins adsorbed with the characteristics of a “Langmuir-like” high-affinity Type I isotherm.26,34,35 All three proteins appear to be limited to a monolayer and have a rapid initial adsorption reaching an adsorption plateau well before 0.05 mg/mL. It was noted, however, that IgG appears to have a more gradual initial slope compared to that of BSA and fibrinogen. Kinetic experiments (not shown) indicated that adsorption was rapid for all three proteins, reaching adsorption equilibrium in under an hour. BSA and fibrinogen exhibited a similar rate of adsorption, reaching adsorption equilibrium faster than IgG. The reduced adsorption kinetics and binding affinity of IgG may in part be due to the similar positive (34) Fleer, G. J.; Scheutiens, J. M. H. M.; Cohen Stuart, M. A.; Cosgrove, T.; Vincent, B. Polymers at Interfaces; Chapman & Hall: New York, 1993. (35) Malmsten, M. J. Colloid Interface Sci. 1994, 166, (2), 333-342.
Figure 1. Adsorption isotherms of fibrinogen (O), IgG (4), and BSA (0) onto APS-coated microspheres by solution depletion.
charge of both the IgG- and APS-coated particles at pH 7.4. The ability of a protein to adsorb onto a similarly charged interface is believed to be due to the gradual structural rearrangement of the protein secondary structure to minimize the surface charge effects.36 A rapid high-affinity adsorption for albumin and fibrinogen and a slightly reduced rate for IgG has also been observed on similar hydrophobic silica surfaces for the adsorption of fibrinogen, IgG, and HSA, the human equivalent of BSA.35 A number of theoretical and empirical equations have been used to describe isotherms similar to those observed in this study. Previous studies have proposed that empirical equations such as the Temkin or Freundlich equations and/or extensions to these or the Langmuir equation to include the nonideal effects presented by protein adsorption are good model approximations.37 Regardless of the adsorption model, at sufficiently dilute adsorbate concentrations, the adsorption analysis can be performed in terms of the equilibrium thermodynamics, such that the adsorption is defined by a linear increase with adsorbate concentration and is modeled by Henry’s law, Γ ) KCeq, where Γ is the adsorbed amount, K is the Henry’s law equilibrium constant, and Ceq is the concentration of adsorbate remaining in solution.36,38 From the equilibrium constant (K), an indication of the adsorption process “affinity” via the Gibbs free energy of adsorption ∆Gads ) -RT ln(K) is possible. The adsorption of the three serum proteins in this study all indicated high-affinity Type 1 isotherms, which are believed to be aptly approximated by a high-affinity Langmuir model with a corresponding high-affinity linear Henry’s law region at low adsorbate concentrations. The approximated values of K calculated by fitting the protein isotherm data to the Langmuir equation were 323 ( 100, 21 ( 8, and 670 ( 250 cm, for BSA, IgG, and fibrinogen, respectively. The observed difference between the initial slopes of BSA and fibrinogen to IgG is thus illustrated in the significantly lower K value for IgG. Caution should be used when interpreting these values of K and fitting the protein isotherms to an equilibrium thermodynamic adsorption model such as the Langmuir model. This is due to the often nonideality of protein adsorption such that adsorption equilibrium is difficult to assess because of protein restructuring on the surface, lateral interactions often exist between the adsorbed species at higher concentrations, and the adsorption is often irreversible. However, it does provide a quantitative indication (36) Malmsten, M. Biopolymers at Interfaces, 2nd ed.; Marcel Dekker: New York, 2003; Vol. 110, p 908. (37) Johnson, R. D.; Arnold, F. H. Biochim. Biophys. Acta: Protein Struct. Mol. Enzymol. 1995, 1247 (2), 293-297. (38) Jaroniec, M.; Madey, R. Physical Adsorption on Heterogeneous Solids; Elsevier Science Publishing Company, Inc.: New York, 1988; Vol. 59, p 351.
Protein-Surface Interaction Study Via Flow Cytometry
Figure 2. Flow cytometric isotherm for fluorescently tagged BSA adsorption onto 100 mg/mL of particles.
of protein affinity and an easy starting point for the derivation of the flow cytometric adsorption equation. Flow Cytometric Adsorption Theory and Single-Component Isotherms. Fluorescently tagged proteins have been widely and successfully applied to the characterization of the biological activity of surfaces and biological materials. This includes the study of nonspecific protein adsorption and cell adhesion onto surfaces and the highly sensitive and specific interactions of complementary biomolecule binding used for cell characterization and molecular diagnostics. In all of these cases, it is important, for accurate analysis, to quantify the number or amount of bound fluorescent proteins. To this end, flow cytometry has been extensively employed because of its high-throughput, highly sensitive, semiquantitative analysis of fluorescent intensity.30 An example of a generated semiquantitative flow cytometric isotherm for the adsorption of fluorescently tagged BSA onto APS particles is shown in Figure 2. The isotherm illustrates the general trend of the protein adsorption: strong initial linear adsorption followed by a sharp plateau at approximately 2 × 10-3 mg/mL of added protein. However, this is a semiquantitative isotherm based upon arbitrary/relative fluorescence units that are dependent upon experimental and instrumentation conditions. Therefore, methodologies such as flow cytometric calibration particles or isoperimetric analysis have been used to convert these arbitrary fluorescence units to absolute units for quantitative comparability. Chatelier et al.31 first proposed the direct method of isoparametric analysis to obtain the adsorption parameters of K and Γ by flow cytometry. This direct method of calculating the thermodynamic parameters on a very small sampling scale was seen to hold particular promise in characterizing biological processes where other techniques are often limited. In brief, isoparametric analysis is based upon the conservation of adsorbate mass in the system, Co ) Γ[Α] + Ceq. Titrating a fluorescently tagged adsorbate against at least three different adsorbent particle concentrations [A] results in a family of isotherms of adsorbent fluorescent intensity versus the added adsorbate concentration (Co). Plotting Co versus [A] for particles of equal (isometric) fluorescent intensity results in a straight line with an intercept and slope equal to Ceq and Γ with quantitative units of concentration and adsorbed amount, respectively. In this paper, the adsorbed amount of three different proteins onto five different surfaces was determined using a simplified method of the theory proposed by Chatelier et al.31 This simplified method assumes that, at very low adsorbate concentrations, the protein adsorption is characterized by high-affinity ideal Henry’s
Langmuir, Vol. 24, No. 4, 2008 1207
Figure 3. Flow cytometric calibration plots for fibrinogen (O), IgG (4), and BSA (0) onto APS-coated particles measured at emission wavelengths of 580 ( 10, 450 ( 32, and 530 ( 20 nm, respectively. Horizontal error bars are the standard error calculated for each point. Points represent adsorbed amounts much less than that of saturation.
law conditions, Γ - KCeq. Under these conditions and because of the high adsorption affinity, the value of K can be considered to go to infinity, and thus the amount of adsorbate free in solution will go to zero, Ceqf 0. By substituting this approximation and eliminating the Ceq term from the conservation of mass equation, the Γ can be considered equal to the Co divided by the adsorbent concentration, such that
Γ)
Co [A]
∝ MFI
(1)
In most instances, the Γ of a fluorescently tagged protein is linearly proportional to the mean fluorescence intensity (MFI) of the adsorbate-coated particles, at low adsorbate coverage. Therefore, a linear flow cytometric calibration curve is generated plotting the MFI versus Co divided by the adsorbent particle concentration (59 ( 6.75 × 106 particles per mL, as determined by hemocytometer measurements). The calibration plots for BSA, IgG, and fibrinogen, as shown in Figure 3, all exhibited good linearity with best fit lines being slope: 1.49 × 1013, R2: 0.997; slope: 7.48 × 1012, R2: 0.989; and slope: 3.02 × 1012, R2: 0.947, respectively. This can then be used to convert the semiquantitative flow cytometric isotherm (Figure 2) from relative fluorescent intensity units into absolute units of adsorbed amount. It also allows for the calculation of the equilibrium concentration at higher adsorbate concentrations where adsorption tends toward nonideal or surface saturation conditions. The slope of the calibration curve is dependent upon the instrumentation and experimental conditions: fluorophore efficiency, laser emission wavelength and power, degree of adsorbent tagging, detector sensitivity, and so forth. Nonlinearity at dilute adsorbate concentrations may occur in systems where the fluorescence intensity is affected by the local environment, such as pH change or by energy transfer between closely oriented fluorophores. In this study, all of the labeled proteins exhibited linear calibration plots with R2 values greater than 0.947. The greatest source of error in the flow cytometry calibration curves was in measuring the particle and protein concentrations. The flow cytometry adsorption isotherms (Figure 4) for the fluorescently tagged BSA, IgG, and fibrinogen exhibited adsorption trends similar to those observed by solution depletion. Like the solution depletion results, all isotherms obtained by flow cytometry were high-affinity Type I isotherms. Furthermore, similar trends in adsorption maxima were observed, with
1208 Langmuir, Vol. 24, No. 4, 2008
Figure 4. Solution depletion adsorption isotherms for fibrinogen (O), IgG (4), and BSA (0) with best fit Langmuir curves exhibiting a good correlation to the flow cytometric isotherms for fibrinogen (b), IgG (2), and BSA (9) onto APS-coated microspheres.
fibrinogen adsorbing to the greatest degree followed by IgG and then BSA. However, slight variances were observed between the maximum adsorption values of the flow cytometry and solution depletion results. This is believed to be principally due to the error associated with the adsorption calibration curves generated by the flow cytometric method. Even small errors in the calibration curves can give rise to large differences in the calculated adsorbed amount. Differences between the two adsorption data sets may also arise as a result of the addition of fluorescent tags to adsorbate proteins. Attachment of substituents to proteins can alter their native properties, including hydrophobicity, isoelectric point, and overall electrostatic charge. These tagging effects are mainly dependent upon the degree of protein labeling, but are also a function of the substituent properties and the conjugation chemistry used. Modification of protein amino groups is known to affect the electrostatic properties of proteins and their adsorption behavior. Typically, such changes are associated with a relatively high percentage (15-70%) of amino group modification.39 To minimize these effects in this study, the ratio of fluorophore molecules to free amino groups was intentionally kept low, being 1.2% (0.69 of 60), 1.0% (2.5 of 250) and 5.5% (4.6 of 72-90) for BSA, fibrinogen, and IgG, respectively. As such, we believe the properties and subsequent adsorption behavior of the native and tagged proteins should be indistinguishable. Flow cytometry, unlike many other detection techniques, has the ability to distinguish between particles within a suspension via optical encoding40,41 and also detect and quantify multiple adsorbate species29 on those particles. This enables the characterization of multiple competitive or cooperative adsorbateadsorbent interactions, as illustrated by Figure 5. The analysis and quantification of individual components adsorbed onto a surface from a multicomponent solution (multicomponent adsorption) is possible by tagging each of the different protein components with a spectrally distinguishable fluorophore tag.42 As a result, the displacement of one protein species by another or the relative concentration of each type of protein at the surface can be determined. Additionally, the simultaneous adsorption onto multiple adsorbent particle surfaces (multiplexed adsorption) (39) Wierenga, P. A.; Meinders, M. B. J.; Egmond, M. R.; Voragen, A. G. J.; de Jongh, H. H. J. J. Phys. Chem. B 2005, 109 (35), 16946-16952. (40) Battersby, B. J.; Lawrie, G. A.; Trau, M. Drug DiscoVery Today 2001, 6 (12), S19-S26. (41) Finkel, N. H.; Lou, X. H.; Wang, C. Y.; He, L. Anal. Chem. 2004, 76 (19), 353a-359a. (42) Russell, S. M.; Carta, G. AIChE J. 2005, 51 (9), 2469-2480.
Kozak et al.
Figure 5. Single-component adsorbent (A, B, and C) particles and adsorbate (2, [, and B ) adsorption. Fluorescent tagging of the adsorbents and adsorbates and flow cytometry were used to characterize multicomponent and multiplexed adsorption simultaneously.
via optically encoding the particles can be used to characterize the adsorption process and determine a relative adsorption affinity for each surface. This offers the ability to rapidly and competitively screen protein adsorption onto different surface modifications, be they preventative or conducive to the adsorption process. Furthermore, by using both optically encoded particles and fluorescently tagged proteins, the adsorption process can be both multicomponent and multiplexed. This in turn reduces the number of experiments required for characterizing adsorption interactions and the amount of adsorbent and adsorbate required, which is important when working with biological samples where sample amounts are often very limited. Multicomponent Adsorption Isotherms. Biological media are typically composed of complex mixtures of biomolecules that can interact in a competitive or cooperative fashion with a surface. Understanding these multicomponent interactions is important, as they are often employed as biomolecule-blocking agents or are an indication of an immune response. The competitive adsorption of equal concentrations (mg/mL) of BSA and IgG onto APS-coated particles (illustrated by the box entitled Multicomponent Adsorption in Figure 5) indicated that both proteins were present on the surface (Figure 6). It is believed that these proteins adsorb competitively rather than collaboratively to form a mixed component monolayer on a surface, and this theory is supported by previous findings by Malmsten and Lassen.43 This is indicated by the sum of the adsorbed amounts of the two components at each concentration measured, being less than either of the single-component adsorbed amounts. The ratio of BSA to IgG on the surface, however, indicated preferential BSA adsorption contrary to the findings of Malmsten and Lassen.43 This is most prevalent at higher protein concentrations where the adsorbed amount of BSA (1.65 mg/m2) was almost equal to single-component BSA adsorption results (1.87 mg/m2). The preferential adsorption of BSA over IgG can be explained in part by electrostatic interactions and by the polymer adsorption theory.34 As mentioned previously, BSA exhibited a higher affinity and rate of adsorption than IgG. This was believed to be partially due to the attractive electrostatic forces between (43) Malmsten, M.; Lassen, B. J. Colloid Interface Sci. 1994, 166 (2), 490498.
Protein-Surface Interaction Study Via Flow Cytometry
Figure 6. The competitive adsorption of IgG (2) and BSA (9) onto APS-coated beads by flow cytometry. The single-component adsorptions of IgG (4) and BSA (0) under the same conditions are given as a reference of uncompetitive protein adsorption values.
the positively charged surface and the oppositely charged BSA. The similarly charged IgG and surface would result in an electrostatic repulsive force that gives rise to a reduced rate and affinity for adsorption. Additionally, similar to the adsorption of polydisperse polymers, the smaller molecular weight BSA is more likely to initially adsorb compared to the larger IgG. This is because smaller molecules diffuse more rapidly through a medium to an interface. These smaller adsorbed species are then typically desorbed and replaced by the more favorable adsorption of larger molecular weight species with time. A similar effect has been observed for the adsorption of complex mixtures of proteins of varying molecular weights onto a surface, known as the Vroman effect.44 However, because of the different chemical and structural compositions of proteins, this trend does not always hold true. The displacement of one protein species by another is known to be subjective to the proteins used and the conditions of the system. Similar studies by Malmsten43 and Noh45 on the displacement of HSA by IgG showed contradictory results and depended on how long the HSA was allowed to preadsorb. In this study, the continued preferential adsorption of BSA over IgG is believed to be due to the rapid and irreversible adsorption of BSA. Multiplexed Adsorption Isotherms. Fluorescently doped particles have been used as means of characterizing colloidal systems46 as well as multiplexed high-throughput biomolecular diagnostic supports. However, to the best of our knowledge, flow cytometry combined with optically encoded particles as a means of distinguishing between surface-modified layers for protein adsorption has not been previously demonstrated. This simultaneous quantitative detection method (illustrated by the box entitled Multiplexed Adsorption in Figure 5) has the advantage in that adsorption can be characterized comparatively between the surfaces. Thermodynamically, adsorption occurs onto the surface that will give rise to the greatest decrease in free energy of the system. Modifying a surface to alter its hydrophilic or hydrophobic qualities, or introducing electrostatic or steric forces, etc. changes this adsorption affinity. Multiplexed adsorption gives rise to a mixed yet spectrally distinguishable system with varying surface properties. Theoretically, in a dynamic multiplexed system at equilibrium, the adsorbate should pref(44) Vroman, L.; Adams, A. L. J. Colloid Interface Sci. 1986, 111 (2), 391402. (45) Noh, H.; Vogler, E. A. Biomaterials 2007, 28 (3), 405-422. (46) van Blaaderen, A.; Ruel, R.; Wiltzius, P. Nature 1997, 385 (6614), 321324.
Langmuir, Vol. 24, No. 4, 2008 1209
Figure 7. Two-dimensional flow cytometry histograms of (A) the optical signature of the five encoded particle populationss(i) APScoated, (ii) preadsorbed IgG, (iii) preadsorbed BSA, (iv) 3400 MW PEG layer, and (v) negatively charged beadssbefore the addition of fluorescently tagged protein, and (B) the correlating change in fluorescence intensity of the five populations due to protein adsorption. The greater movement in the X-axis corresponds to a greater amount of protein adsorbed onto the particle surface.
erentially adsorb first onto the particles with a surface that gives rise to the greatest decrease in free energy. When these particles become saturated, the adsorbate will then adsorb onto the surface with the next highest affinity, and so on until all of the particle surfaces are saturated. Five optically distinguishable particle populations were made by incorporating Atto-tech 620 dye into the particles to form five fluorescently distinctive particle populations. As illustrated in Figure 7A, the particles were optically distinguishable in the 670 ( 15 nm emission detector (Y-axis) and exhibited minimal emission in the 530 ( 20 nm emission detector (X-axis). These five particle populations were then modified with five different surface layers, as given in Table 1: (i) a blank surface (APS coated), (ii) physisorbed IgG, (iii) physisorbed BSA, (iv) 3400 MW PEG, or (v) a negative surface. Fluorescent encoding of the particles is believed to have little to negligible effect on the multiplexed adsorption in this study; however, it cannot be ruled out. Fluorescently labeling polystyrene particles has been shown to have varied effects on fibrinogen adsorption.47 These particles, however, are typically saturated with fluorophores, and adsorption variation has been shown to be more dependent upon the particle synthesis. In this study, the maximum fluorophore concentration was approximately 450 000 molecules of dye/µm3, as determined by solution depletion measurements of dye uptake into particles. Furthermore, the dye has been shown to distribute evenly throughout the particle interior.48 This low dye doping concentration in combination with the APS particle modification after dying is believed to limit any surface effects due to the fluorescent dye. Fluorescently tagged BSA, IgG, and fibrinogen were adsorbed onto a multiplexed (mixed) solution of the optically encoded particles. As with the previous flow cytometric measurements, protein adsorption resulted in an increase in the fluorescence intensity channel corresponding to the emission of the protein’s fluorescent tag, e.g., 530 ( 20 nm for the Atto-tech 488-tagged BSA. Figure 7B shows the resulting fluorescent intensity gain in the 530 ( 20 nm emission detector (X-axis) of the five particle populations due to BSA adsorption at the highest protein concentration used, 0.05 mg/mL. The mean and distribution of the fluorescent intensity of each of the five particle populations in the protein adsorption emission channel (X-axis) is then used (47) Muller, R. H.; Ruhl, D.; Luck, M.; Paulke, B. R. Pharm. Res. 1997, 14 (1), 18-24. (48) Miller, C. R.; Vogel, R.; Surawski, P. P. T.; Corrie, S. R.; Ruhmann, A.; Trau, M. Chem. Commun. 2005, (38), 4783-4785.
1210 Langmuir, Vol. 24, No. 4, 2008
Kozak et al.
Figure 8. Multiplexed adsorption of (A) BSA, (B) low-concentration BSA data, (C) IgG, and (D) fibrinogen onto APS-coated (2), negatively charged beads (O), 3400 MW PEG ([), preadsorbed IgG (0), and preadsorbed BSA (4) surface-modified microspheres by flow cytometry. Single-component adsorption data on APS-coated particles are also shown (/).
to generate surface competitive “multiplexed” adsorption isotherms (Figure 8A-D). Multiplexed adsorption indicated similar adsorption trends for each of the three proteins onto the five different surfaces. All of the proteins exhibited the greatest adsorption onto the APScoated particles, followed by the negatively charged surface, then the grafted PEG layer, then the preadsorbed IgG and BSA layers. Protein adsorption onto the APS-coated particles of the multiplexed suspensions was similar to the results for singlecomponent adsorption studies onto APS-coated particles. However, slight deviations between the single and multiplexed adsorption studies were evident. These differences were most pronounced for fibrinogen adsorption onto APS-coated particles (Figure 8D). This greater deviation in the fibrinogen isotherms, not exhibited by the BSA and IgG, is believed to be due to the larger error in the fibrinogen calibration curve compared to the BSA and IgG calibrations. Changing the particle surface charge from positive (18 ( 3.6 mV) to negative (-34 ( 7.9 mV) by grafting 12-hydroxydodecanoic acid was found to decrease the adsorbed amount of all three proteins. Previous findings have shown that altering surface charge and thereby changing the electrostatic interactions between a protein and surface typically affects the adsorption kinetics and maximum adsorbed amount. Interestingly, in this study, with the addition of the 12-hydroxydodecanoic acid layer, even the oppositely charged IgG exhibited a decrease in adsorbed amount, contrary to what was expected for electrostatic interactions alone. This is believed to be due to a combined effect of electrostatic interactions and the differences in surface group functionality.
Ostuni et al.49 illustrated that the terminal group functionality of self-assembled monolayers can play a critical role in protein adsorption interactions; e.g., the addition of hydrophobic or hydrogen bonding donor groups increased protein adsorption. It is therefore, believed that the addition of the 12-hydroxydodecanoic acid layer imparted both electrostatic and surface functional group interactions, which effected protein adsorption. The prevention of protein adsorption (antifouling) is important for biocompatible surfaces and molecular diagnostic sensitivity. Surfaces are often modified with a grafted polymer layer such as PEG to reduce protein adsorption. Particles in this study grafted with a 3400 MW PEG layer decreased the adsorption of all three proteins by approximately half. The mechanism and efficacy of PEG as a steric barrier to prevent protein adsorption is continually being examined and has been observed to be dependent upon the system used. Differences in polymer grafting densities, polymer molecular weights, and proteins studied can all lead to substantially different protein adsorption profiles. Another common method to prevent protein adsorption is the pre-adsorption of a known protein onto the surface. Preadsorbed IgG and preadsorbed BSA were shown to prevent almost any additional protein adsorption, regardless of the adsorbing species. This indicates that an irreversible, nondynamic adsorption exists between the adsorbed proteins and those free in solution. Interestingly, no substantial difference in the relative adsorption affinity of each of the surfaces was observed. This is illustrated in part by the slightly greater increase in protein adsorption for (49) Ostuni, E.; Chapman, R. G.; Holmlin, R. E.; Takayama, S.; Whitesides, G. M. Langmuir 2001, 17 (18), 5605-5620.
Protein-Surface Interaction Study Via Flow Cytometry
the positive APS-coated and negative particle surfaces compared to the other surfaces at low protein concentrations (Figure 8B). However, for an ideal system, one would expect adsorption onto only the surface with the greatest affinity followed then by the next greatest surface affinity population, and so on. The marginal adsorption affinity differences observed in this study are believed to be due in most part to the very rapid and irreversible adsorption of the proteins onto the surfaces. The lack of adsorption equilibrium means that, rather than “probe” the adsorbent surfaces for the surface with greatest change in free energy, the proteins rapidly and irreversibly adsorbed onto the first available surface encountered. This may be overcome in part by reducing the volume fraction of particles and salt concentration used (∼0.03% v/v and 100 mmol, respectively) to increase the length of electrostatic interactions and adsorption kinetics. Increasing the adsorption equilibrium dynamics would allow for the adsorption affinity calculations between different surfaces in the multiplexed solution. Regardless, multiplexed analysis does provide valuable comparative information of the protein-surface adsorption interactions and process.
Conclusions Flow cytometry has been successfully used for the characterization of single-component adsorption on a small sampling
Langmuir, Vol. 24, No. 4, 2008 1211
scale. In addition, the ability to quantitatively determine the adsorption distribution of multicomponent adsorbent and adsorbate systems has been demonstrated. The adsorption of three common serum proteinssBSA, IgG, and fibrinogensonto five modified silica particle surfaces, either individually or in combination, was measured by flow cytometry. The results were comparable to single-component measurements using solution depletion measurements. By using spectrally discriminative fluorophores on each protein species and optically encoding the surface-modified particles, multiple adsorbate-adsorbent interactions can be characterized and directly compared using very small assay volumes. We believe that this high-throughput multicomponent and multiplexed analysis technique will offer convenient and rapid screening and characterization of complex protein surface interactions. This holds promise for the rapid development and improvement of preventive surface coatings for future biological materials and biosensors. Acknowledgment. This work was supported by the Australian Research Council (FF0455861: LP0349397). We would also like to acknowledge Aye Thandar, Chieh-Yu Lu, Christina Lee, and Simon Chen for their assistance with protein adsorption measurements. LA701847X
