Langmuir 2006, 22, 6313-6320
6313
Orientation of a Monoclonal Antibody Adsorbed at the Solid/Solution Interface: A Combined Study Using Atomic Force Microscopy and Neutron Reflectivity Hai Xu, Xiubo Zhao, Colin Grant, and Jian R. Lu* Biological Physics Group, School of Physics and Astronomy, UniVersity of Manchester, SackVille Street Building, SackVille Street, Manchester M60 1QD, UK
David E. Williams Unipath Ltd., Priory Business Park, Bedford MK44 3UP, UK
Jeff Penfold ISIS, Rutherford Appleton Laboratory, CLRC, Chilton, Didcot, OX11, 0QZ, UK ReceiVed NoVember 30, 2005. In Final Form: April 23, 2006 Conformational orientations of a mouse monoclonal antibody to the β unit of human chorionic gonadotrophin (anti-β-hCG) at the hydrophilic silicon oxide/water interface were investigated using atomic force microscopy (AFM) and neutron reflectivity (NR). The surface structural characterization was conducted with the antibody concentration in solution ranging from 2 to 50 mg‚L-1 with the ionic strength kept at 20 mM and pH ) 7.0. It was found that the antibody adopted a predominantly “flat-on” orientation, with the Fc and two Fab fragments lying flat on the surface. The AFM measurement revealed a thickness of 30-33 Å of the layer formed in contact with 2 mg‚L-1 antibody in water, but, interestingly, the flat-on antibody molecules formed small nonuniform clusters equivalent to 2-15 antibody molecules. Parallel AFM scanning in air revealed even larger surface clusters, suggesting that surface drying induced further aggregation. The AFM study thus demonstrated that the interaction between protein and the hydrophilic surface is weak and indicated that surface aggregation can be driven by the attraction between neighboring protein molecules. NR measurements at the solid/water interface confirmed the flat-on layer orientation of adsorbed molecules over the entire concentration range studied. Thus, at 2 mg‚L-1, the adsorbed antibody layer was well represented by a uniform layer with a thickness of 40 Å. This value is thicker than the 30-33 Å observed from AFM, suggesting possible layer compression caused by the tip tapping. An increase in the antibody concentration to 10 mg‚L-1 led to increasing surface adsorption. The corresponding layer structure was well represented by a three-layer model consisting of an inner sublayer of 10 Å, a middle sublayer of 30 Å, and an outer sublayer of 25 Å, with the protein volume fractions in each sublayer being 0.22, 0.42, and 0.10, respectively. The structural transition can be interpreted as a twisting and tilting of segments of the adsorbed molecules, driven by an electrostatic repulsion between them that increases with the surface packing density. Hindrance of antigen access to antibody binding sites, resulting from the change in surface packing, can account for the decrease in antigen binding capacity (AgBC) with increasing surface density of the antibody that is observed.
Introduction The immobilization of antibodies at solid surfaces is important for many applications in biomedical engineering and biotechnology. The performance of immunoassays and biosensors relies critically on immobilized antibodies retaining their biological activity. On the basis of the specific recognition of antigen to antibody, these techniques allow the detection of antigens of interest, with concentrations down to parts per billion or parts per trillion in aqueous environments, including blood, urine, and saliva. These tests are also effective for the medical diagnosis of fatal diseases such as HIV, hepatitis, salmonella, and trichinella.1-5 With the advent of hybridoma technology, many * To whom all correspondence should be addressed. Phone: 44-1612003926; e-mail:
[email protected]. (1) Mantero, G.; Zonaro, A.; Albertini, A.; Bertolo, P.; Primi, D. Clin. Chem. 1991, 37, 422. (2) Gehring, A. G.; Crawford, C. G.; Mazenko R. S.; VanHouten, L. J.; Brewster, J. D. J. Immunol. Methods 1996, 195, 15. (3) Pathirana, S. T.; Barbaree, J.; Chin, B. A.; Hartell, M. G.; Neely, W. C.; Vodyanoy, V. Biosens. Bioelectron. 2000, 15, 135. (4) Gamble, H. R.; Brady, R. C.; Bulaga, L. L.; Berthoud, C. L.; Smith, W. G.; Detweiler, L. A.; Miller, L. E.; Lautner, E. A. Vet. Parasitol. 1999, 82, 59.
pure monoclonal antibodies are now produced that bind specifically to one single antigenic determinant with high affinity.6,7 However, a very common technical obstacle is the low antigen binding capacity (AgBC) when an antibody is immobilized onto a solid substrate.8-11 Because the antigen binding sites are located in the Fab regions and formed by the combination of the hypervariable regions of the heavy and light chains, the decrease in AgBC is generally thought to be due to improper orientations of antibody molecules on the solid surface. It has been proposed12-17 that, if antibodies adopt an improper orientation (5) Abdel-Hamid, I.; Ivnitski, D.; Atanasov, P.; Wilkins, E. Anal. Chim. Acta 1999, 399, 99. (6) Ko¨hler, G.; Milstein, C. Nature 1975, 256, 495. (7) Milstein, C. Sci. Am. 1980, 243, 56. (8) van Erp, R.; Linders, Y. E. M.; van Sommeren, A. P. G.; Gribnan, T. C. J. J. Immunol. Methods 1992, 152, 191. (9) Butler, J. E.; Ni, L.; Brown, W. R.; Joshi, K. S.; Chang, J.; Rosenberg, B.; Voss, E. W. Mol. Immunol. 1993, 30, 1165. (10) Olson, W. C.; Spitznagel, T. M.; Yarmush, M. L. Mol. Immunol. 1989, 26, 129. (11) Spitznagel, T. M.; Clark, D. S. Nat. Biotechnol. 1993, 11, 825. (12) O’Shannessy, D. J.; Dobersen, M. J.; Quarles, R. H. Immunol. Lett. 1984, 8, 273. (13) Hoffman W. L.; O’Shannessy, D. J. J. Immunol. Methods 1988, 112, 113.
10.1021/la0532454 CCC: $33.50 © 2006 American Chemical Society Published on Web 05/28/2006
6314 Langmuir, Vol. 22, No. 14, 2006
in which their active sites are inaccessible for an antigen, their biological activity and function would be reduced because of steric hindrance. As a result, the characterization of the orientation of antibodies on solid surfaces is highly important for the development of immunoassays and biosensors with high efficacy. However, because it has been difficult to determine the structure of protein molecules adsorbed at the solid/water interface, the actual orientation of an antibody adsorbed on a given substrate remains unclear. In a recent study,18 we used spectroscopic ellipsometry to establish the main patterns of antibody adsorption with respect to solution concentration and pH. It was shown that, for the mouse monoclonal antibody to the β unit of human chorionic gonadotrophin (anti-β-hCG), the AgBC at the hydrophilic silica/ water interface was high at low coverage (around 0.7 antigen/ antibody) but decreased markedly with increasing surface packing density. Although the same trend was also observed by Spitznagel et al.11 and Herron et al.,19 it is contrary to the conventional wisdom that the more antibody adsorbed, the higher the AgBC. Our initial study with neutron reflectivity (NR)18 indicated that the antibody adopted a “flat-on” orientation, even when the AgBC was high. In the literature, higher AgBC is generally assumed to be associated with the “end-on” orientation in which the Fab fragments point toward the bulk solution. In contrast, our studies seem to suggest that it is the surface packing density of the antibody that dictates the AgBC. It is therefore important to further explore which factor dominates the AgBC: the orientation or the surface steric hindrance, or a combination of both. A crucial step toward a better understanding of the factors affecting AgBC is to determine the structure of the adsorbed antibody layer. In this study, we report the structural measurement of anti-β-hCG adsorbed at the hydrophilic silicon oxide/water interface using NR, with the particular purpose of revealing the change in the conformational orientations of the adsorbed antibody with respect to increasing surface concentration. Atomic force microscopy (AFM) imaging was also used to reveal the morphological features of anti-β-hCG adsorbed on the silica surface in air and under water. These results show that the antibody adopts a flat-on conformation at the hydrophilic silicon oxide/ water interface, with their Fc and Fab fragments lying flat on the surface. Furthermore, there is a strong tendency for the molecules to form surface clusters. Upon drying in air, the size of the antibody clusters grows. These observations suggest that the interaction between the protein and the hydrophilic substrate is relatively weak in comparison with protein-protein interactions. Experimental Section The NR measurements were carried out on the white beam reflectometer CRISP at the ISIS Neutron Facility, Rutherford Appleton Laboratory, Didcot, UK, using neutron wavelengths from 1 to 6.5 Å. The freshly polished surface of a silicon 〈111〉 block was clamped against a shallow stainless steel trough containing protein solution in D2O. NR was conducted by directing the incoming parallel beam into one end of the silicon block, reflecting it from the solid/ D2O solution interface, and allowing it to exit from the other end of the block. Reflectivity was measured at three different beam incidence angles of 0.35, 0.8, and 1.8°, and the resultant profiles were combined and normalized against unity below the critical angle. (14) Chang, I. N.; Herron, J. N. Langmuir 1995, 11, 2083. (15) Chang, I. N.; Lin, J. N.; Andrade, J. D.; Herron, J. N. J. Colloid Interface Sci. 1995, 174, 10. (16) Buijs, J.; White, D. D.; Norde, W. Colloids Surf., B 1997, 8, 239. (17) Chen, S.; Liu, L.; Zhou, J.; Jiang S. Langmuir 2003, 19, 2859. (18) Xu, H.; Lu, J. R.; Williams, D. E. J. Phys. Chem. B 2006, 110, 1907. (19) Herron, J. N.; Wang, H. K.; Janatova´, V.; Durtschi, J. D.; Christensen, T. A.; Caldwell, K.; Chang, I. N.; Huang, S. J. In Biopolymers at Interfaces, 2nd ed.; Malmsten, M., Ed.; Marcel Dekker: New York, 2003; Chapter 6, p 115.
Xu et al. A flat background was estimated by averaging the reflectivity measured at the wave vector above 0.3 Å-1 and was subtracted from the measured reflectivity. Information about layer thickness and composition was obtained by model fitting based on the optical matrix formula.22 All the AFM imaging was done with a commercial Nanoscope IIIa system (Digital Instruments, Santa Barbara, CA). Because protein surfaces adsorbed onto the silicon wafer would be soft and compliant, the instrument was operated in tapping mode. Standard silicon nitride probes, provided by Olympus with a tip radius of approximately 20 nm, were used at a scan rate of 1 Hz on the protein-covered wafers. The resonant frequencies of the tip varied between ∼8 and 270 kHz in every case. Scans in liquid required the use of a fluid cell, with a rubber O-ring providing the seal between the wafer and the fluid media, and were performed in ultrahigh quality (UHQ) water after 10 min of adsorption. AFM measurements were made at least three times on each sample, and the representative images are shown here. The antibody used in this work was a mouse monoclonal antihCG subtype IgG1, supplied by Unipath, UK (clone no. 3468), and used in phosphate buffer as supplied. This antibody was directed against the β subunit of the hCG. The molecular weight was 150 000, and the isoelectric point was pH 5.5-6.0. D2O was purchased from Sigma-Aldrich and its surface tension was typically over 71 mN‚m-1 at 298 K, indicating the absence of any surface-active impurity. The solution pH was controlled by using phosphate buffer, with the total ionic strength fixed at 0.02 M. The glassware and troughs for the reflection measurements were cleaned using commercial detergent (Decon 90) followed by copious rinsing with UHQ water. All the experiments were performed at 298 K. The procedure for polishing the large 〈111〉 surface of the silicon block has been previously described.20,21 The freshly polished surface was cleaned with 5% Decon solution, and the whole block was copiously rinsed with UHQ water, followed by soaking in acid peroxide solution (600 mL of 98% H2SO4 in 100 mL of 25% H2O2) for 2 min at 90 °C to optimize the surface hydrophilicity.23,24 This procedure was found to produce surfaces that were completely wetted by water, with reproducible thickness and roughness of the oxide layer. All silicon wafer surfaces (8 × 8 mm2) used in the AFM measurements were cut from silicon 〈111〉 wafers (Compact Technology Ltd., UK). The thickness of the native oxide layer was determined using ellipsometry to be around 13 Å. Prior to each experiment, the silicon wafers were freshly cleaned by immersing them in 5% Decon solution, followed by copious rinsing with UHQ water.
Results and Discussion A. AFM Characterization. AFM has been extensively used to image biological systems such as DNA, cells, and protein aggregates. Prior to antibody adsorption in the present study, the 〈111〉 silicon wafer surface was imaged under UHQ water. The scan revealed a highly flat and smooth surface with an average roughness less than 5 Å, indicating that any features shown in images after protein adsorption result from adsorbed antibody molecules. Figure 1a shows images obtained under UHQ water of the adsorbed anti-β-hCG molecules on the native silica surface after incubation in 2 mg‚L-1 anti-β-hCG solution for 10 min with ionic strength kept at 20 mM and pH ) 7.0. It can be seen from Figure 1a that the protein molecules were adsorbed in the form of small clusters instead of discrete molecules. The height of these protein clusters was 30-33 Å, and they varied from 200 to 500 Å in lateral size, indicating the formation of nonuniform surface aggregates. The thickness was consistent with that for the IgG layer adsorbed on a hydrophobic and highly oriented pyrolytic graphite surface reported by Cullen et al.25 from their (20) Su, T. J.; Lu, J. R.; Thomas, R. K.; Cui, Z. F.; Penfold, J. Langmuir 1998, 14, 438. (21) Su, T. J.; Lu, J. R.; Thomas, R. K.; Cui, Z. F. J. Phys. Chem. B 1999, 103, 3727. (22) Born, M.; Wolf, E. Principles of Optics; Pergamon: Oxford, 1970. (23) Brzoska, J. B.; Shahidzadeh, N.; Rondelez, F. Nature 1992, 360, 719. (24) Vig, J. R. Vac. Sci. Technol. 1985, A3, 1027.
Orientation of an Adsorbed Monoclonal Antibody
Figure 1. Tapping-mode AFM images for anti-β-hCG adsorbed onto hydrophilic silicon oxide from a 2 mg‚L-1 anti-β-hCG solution for 10 min, with the ionic strength kept at 20 mM and pH ) 7.0, and then imaged under UHQ water (a) and in air (b). Image size: 1 × 1 µm2.
AFM measurement under liquid solution, and was also close to the short axial length of 38 Å of IgG molecules (the dimensions of an IgG are approximately 142 × 85 × 38 Å3).26-28 This suggests that the antibody molecules adsorb flat-on, with one of their largest surfaces lying on the substrate. The decrease in the height from 38 to 30 Å indicates some extent of deformation upon surface adsorption. The difference could also result from the compression of the AFM tip causing structural deformation of the adsorbed protein. Assuming that each protein molecule has an adsorbed footprint of 142 × 85 Å2, the surface area of the cluster corresponds to 2-15 antibody molecules. Surface protein aggregation suggests an interaction between the protein and the surface that is relatively weak compared to the attraction between protein molecules within the surface. The height measurement suggests the formation of an antibody monolayer, and there is no evidence to indicate bilayer or multilayer formation. Tapping mode is known for its gentle touch on the sample surface and weak drag force within the surface plane.29 However, streaks surrounding the protein clusters are still visible from left to right in Figure 1a. These streaks are parallel with the AFM scan direction, indicating some motion or distortion of antibody molecules on the hydrophilic surface arising from the tip-protein interaction. Such structural disturbance on the adsorbed antibody molecules has also been observed on mica surfaces by others, irrespective of contact mode or tapping mode.30-32 These studies and our own work show that the IgG antibody interacts weakly with this type of substrate. In the course of our study, effort was made to tune the tip-protein interaction to make it weaker than the tip-substrate interaction. This proved to be difficult because down-tuning the tip-protein interaction reduced the resolution of the AFM images in solution. It is, however, interesting to note that, on the hydrophobic surface, AFM tip-induced distortions disappear.25 On the hydrophobic surface, the protein-surface interaction is comparatively strong. Thus, the streaks caused by tip scanning on the hydrophilic surface support the view that the interaction between the protein and the surface is weak. AFM measurements are often performed in air because of the technical convenience and the relatively higher resolution attainable. In this work, the adsorbed protein surfaces were imaged in air immediately after being removed from UHQ water and (25) Cullen, D. C.; Lowe, C. R. J. Colloid Interface Sci. 1994, 166, 102. (26) Silverton, E. W.; Navia, M. A.; Davies, D. R. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 5140. (27) Deisenhofer, J. Biochemistry 1981, 20, 2361. (28) Marquart, M.; Deisenhofer, J.; Huber, R.; Palm, J. J. Mol. Biol. 1980, 141, 369. (29) Putman, C. A. J.; van der Werf, K. O.; de Grooth, B. G.; van Hulst, N. F.; Greve, J. Appl. Phys. Lett. 1994, 64, 2454. (30) Lin, J. N.; Drake, B.; Lea, A. S.; Hansma, P. K.; Andrade, J. D. Langmuir 1990, 6, 509.
Langmuir, Vol. 22, No. 14, 2006 6315
dried in N2 (Figure 1b). It can be seen from Figure 1b that surface drying had a strong effect on the structural topography of the adsorbed anti-β-hCG. The surface aggregates became significantly wider, and their lateral diameters increased from 200-500 Å in water to 500-2500 Å in air. More importantly, the height also increased from 30-33 Å to 60-100 Å. The height of the protein aggregates measured in air would indicate other possible conformations, from flat-on to standing orientations. While the precise molecular orientation inside the deposited surface aggregates is unclear, an important observation here is that surface drying caused structural reorganization of surface adsorbed antibody. The results thus indicate that it is unreliable to use AFM measurements in air to represent protein conformations under water. We note that, on the hydrophobic surface, no discernible feature of aggregation has been observed after drying, again because of the much stronger interaction between the protein and the hydrophobic surface.33 We subsequently used AFM to image antibodies adsorbed from higher bulk concentrations. It was found that, as the concentration increased, the surface protein layers became more compact and then bilayer structural features were observed, with the inner layer being crowded and the outer layer being loose. Under these circumstances, other than revealing the overall structural features, AFM became less powerful for revealing structural details at the molecular level. In comparison, NR was more advantageous for performing in situ measurements at the solid/water interface. B. Neutron Reflectivity. The use of NR for the determination of the structure of biomolecules adsorbed at interfaces is well described in a number of recent papers.20,21,34-43 In this work, a silicon block with 〈111〉 orientation was used, and one of its large faces was polished and was subsequently treated in piranha solution to provide a smooth hydrophilic surface with a freshly formed silicon oxide layer. Because the oxide layer contributed to the neutron signal, it was necessary to determine its thickness and composition prior to protein adsorption. The NR profile measured at the silicon oxide/D2O interface is shown as the dashed line in Figure 2. The fitting of this reflectivity profile gave a thickness of 14 ( 2 Å and a scattering length density (F) of 3.41 × 10-6 Å-2 for the oxide layer. The latter is the same as the value expected for 100% amorphous silicon oxide, indicating a uniform oxide layer. The thickness is consistent with the 12 ( 3 Å obtained from ellipsometric measurements. Furthermore, no roughness was required in the fitting, indicating that the oxide layer was relatively smooth (with surface roughness less than 3 Å). (31) Lea, A. S.; Pungor, A.; Hlady, V.; Andrade, J. D.; Herron, J. N.; Voss, E. W. Langmuir 1992, 8, 68. (32) Thomsom, N. H. J. Microscopy 2005, 217, 193. (33) Xu, H.; Zhao X. B.; Grant, C.; Lu, J. R.; Williams, D. E.; Penfold, J. Langmuir to be submitted for publication, 2006. (34) Fragneto, G.; Thomas, R. K.; Rennie, A. R.; Penfold, J. Science 1995, 267, 657. (35) Liebmann-Vinson, A.; Lander, L. M.; Foster, M. D.; Brittain, W. J.; Vogler, E. A.; Majkrzak, C. F.; Satija, S. Langmuir 1996, 12, 2256. (36) Petrash, S.; Liebmann-Vinson, A.; Foster, M. D.; Lander, L. M.; Brittain, W. J.; Majkrzak, C. F. Biotechnol. Prog. 1997, 13, 635. (37) Su, T. J.; Green, R. J.; Wang, Y.; Murphy, E. F.; Lu, J. R.; Ivkov, R.; Satija, S. K. Langmuir 2000, 16, 4999. (38) Su, T. J.; Lu, J. R.; Thomas, R. K.; Cui, Z. F.; Penfold, J. J. Phys. Chem. B 1998, 102, 8100. (39) Petrash, S.; Cregger, T.; Zhao, B.; Pokidysheva, E.; Foster, M. D.; Brittain, W. J.; Sevastianov, V.; Majkrzak, C. F. Langmuir 2001, 17, 7465. (40) Tiberg, F.; Nylander, T.; Su, T. J.; Lu, J. R.; Thomas, R. K. Biomacromolecules 2001, 2, 844. (41) Choi, E. J.; Foster, M. D.; Daly, S.; Tilton, R.; Przybycien, T.; Majkrzak, C. F.; Witte, P.; Menzel, H. Langmuir 2003, 19, 5464. (42) Forciniti, D.; Hamilton, W. A. J. Colloid Interface Sci. 2005, 285, 458. (43) Cooper, A.; Kennedy, M. W.; Fleming, R. I.; Wilson, E. H.; Videler, H.; Wokosin, D. L.; Su, T. S.; Green, R. J.; Lu, J. R. Biophys. J. 2005, 88, 2114.
6316 Langmuir, Vol. 22, No. 14, 2006
Xu et al.
Figure 2. Plots of NR profiles at the silica/D2O interface with anti-β-hCG concentrations in the solution of 2 (+), 10 (×), and 50 mg‚L-1 (O). The solution pH and ionic strength were fixed at 7 and 20 mM, respectively. The reflectivity profile from the bare silica/ D2O interface is also shown for comparison (dashed line). The solid lines are the best fits. For clarity, error bars are shown for the 50 mg‚L-1 data only. The level of experimental error is similar between the different reflectivity profiles shown in the figure. Table 1. Structural Parameters of an Anti-β-hCG Layer with Bulk Concentration at pH 7.0 and Ionic Strength at 20 mM concn (mg‚L-1)
τ (Å)
2 10
τ1 ) 40 ( 3 τ1 ) 10 ( 3 τ2 ) 30 ( 3 τ3 ) 25 ( 5 τ1 ) 14 ( 3 τ2 ) 30 ( 3 τ3 ) 35 ( 5
50c
(F ( 0.05) × 106 Γa ( 0.2 Γb ( 0.2 (Å-2) φ ( 0.02 (mg‚m-2) (mg‚m-2) F1 ) 5.9 F1 ) 5.7 F2 ) 5.1 F3 ) 6.0 F1 ) 5.4 F2 ) 5.0 F3 ) 6.0
φ1 ) 0.15 φ1 ) 0.22 φ2 ) 0.42 φ3 ) 0.12 φ1 ) 0.3 φ2 ) 0.44 φ3 ) 0.09
0.9 2.5
0.9 2.4
2.9
2.6
a Calculated from three-layer models. b Calculated from single-layer models. c In this case, Fw is 6.25 × 10-6 Å-2 (containing 1.5% H2O in D2O)
To obtain structural information about the adsorbed layer and its variation with bulk antibody concentration, NR measurements were made with the pH and ionic strength of the bulk protein solution fixed at 7.0 and 20 mM, respectively, at antibody concentrations shown in our previous work to give a variation in the AgBC of the adsorbed layer varying from high to low, that is, 2, 10, and 50 mg‚L-1. Figure 2 shows the corresponding NR profiles. Compared to the reflectivity from the bare silica/D2O interface, the adsorption of anti-β-hCG leads to a change in the shape of the reflectivity. There is a broad interference fringe occurring around κ ) 0.05 Å-1, and, apart from this, the overall level and shape of these reflectivity profiles shows only minor variations with increasing protein concentration. Quantitative information about the structure and composition of the adsorbed protein layer can be obtained from model fitting. During such a fitting procedure, a structural model is assumed first, and the corresponding reflectivity is calculated using the optical matrix formula.22 The calculated reflectivity is then compared with the measured one, and the structural parameters are subsequently modified in a least-squares iteration to obtain a good fit. The parameters used in the calculation are the thickness of the layer (τ) and the corresponding scattering length density (F). The solid lines shown in Figure 2 are the best fits to the measured data with the corresponding structural parameters listed in Table 1. Note that we started fitting by using the simplest model, that is, a uniform layer distribution for all adsorbed antibody layers. At the lowest concentration of 2 mg‚L-1, it was found that the reflectivity profile could be best fitted by a uniform
monolayer distribution with τ ) 40 ( 3 Å and F ) 5.90 ( 0.1 × 10-6 Å-2. At a higher adsorbed amount, a three-layer model best described the data. When a protein molecule adsorbs onto a solid substrate, it will form a number of direct contacts with the substrate surface. Because of such direct contacts, partial breakdown of fragments bearing R-helix or β-sheet structures is possible, resulting in interfacial structural rearrangements. However, the degree of interfacial structural alterations upon adsorption would depend on many factors, including protein stability, temperature, interfacial packing density, surface properties, and proteinsurface interactions.44 IgG is a globular molecule, consisting of four polypeptide chains, that is, two light chains and two heavy chains. These four segments are held together by disulfide binds and noncovalent hydrophobic interactions, making the native state of the protein in aqueous solution relatively robust. Upon adsorption onto the hydrophilic silica surface, the internal coherence of the globular protein should prevent it from completely unfolding into lose random structures. The protein molecule on the surface should be largely a globular assembly, although some local deformations might occur because of protein-surface and protein-protein interactions. Because the AgBC at low surface coverage was rather high (about 0.7 antigen/ surface antibody),18 it is reasonable to assume that any severe deformation or degradation of the adsorbed antibody was to a minor extent. This view is supported by the work of Giacomeli et al.,45 who investigated IgG adsorption onto different surfaces using Fourier transform infrared-attenuated total reflection (FTIR-ATR) and found that the secondary structure of adsorbed IgG on the hydrophilic surface resembled that of IgG in solution (approximately 60% β-sheet and almost no R-helix content). Thus, the dimensions of the adsorbed IgG molecule on the silica surface should be approximately comparable with its native dimensions in solution. Measurements of protein layer thicknesses on different substrates can be used to determine their orientations with reference to their solution or crystalline structures.25,44,46,47 Thus, in our case, the thickness of 40 Å for the adsorbed antibody layer at 2 mg‚L-1 is in good agreement with the short axial length of 38 Å in the IgG molecule, indicating the formation of a single monolayer with the protein adsorbed flat-on to the oxide surface. The neutron result is in good agreement with the AFM work under water, and the difference in the thickness of the layer might be attributed to the deformation caused by the tip-protein interaction. Apart from the overall thickness providing information on the orientation adopted by the adsorbed protein, other parameters can also be derived from the fitted τ and F values. For a uniform adsorbed layer, the volume fraction (φp) and the surface excess (Γ) of the protein within the layer can be calculated on the basis of the following equations:
F - Fw Fp - Fw
(1)
Γ ) φp‚τ‚F′p
(2)
φp )
where Fp and Fw are the scattering length density for the protein and water, respectively, F′p is the mass density of the protein, and Fw for D2O is 6.35 × 10-6 Å-2. For calculation of the scattering length density of the protein, complete exchange of (44) Malmsten, M. J. Colloid Interface Sci. 1998, 207, 186. (45) Giacomelli, C. E.; Bremer, M. G. E. G.; Norde, W. J. Colloid Interface Sci. 1999, 220, 13. (46) Malmsten M. Colloids Surf., B 1995, 3, 297. (47) You, H. X.; Lowe, C. R. J. Colloid Interface Sci. 1996, 182, 586.
Orientation of an Adsorbed Monoclonal Antibody
Langmuir, Vol. 22, No. 14, 2006 6317
Table 2. Scattering Length Densities of Different Proteins and Oligosaccharides protein or oligosaccharides
Fp × 106 (Å-2)
mouse monoclonal antibody (Mab61.1.3) mouse monoclonal antibody (Mab231) light chain of Mab231 heavy chain of Mab231 Fab fragment of mouse monoclonal antibody Ctm01 Fv fragment of mouse monoclonal antibody D13 human IgG B12 Fc fragment of human IgG1 Fab fragment of human IgG1 BSA lysozyme chitin starch
3.36 3.33 3.43 3.28 3.35 3.46 3.38 3.30 3.41 3.25 3.65 3.38 3.57
the labile hydrogens in the antibody molecule with D2O is assumed. The following three comments are relevant to the calculation. First, the labile hydrogens are likely to be those on the amide groups in the peptide chains. Because the majority of them are either on the outer surface or accessible to the solvent, they are easily exchanged. As a result, only a relatively small fraction of the labile hydrogens in a globular protein would have difficulty in exchanging with deuterium in D2O. It has been shown that the derived uncertainty in the extent of exchange falls within the experimental error, and has, in fact, a negligible effect on the derived surface excess for a wide range of proteins.20-21,38 Second, although the complete amino acid sequence for anti-β-hCG is not yet known, it is highly reasonable to take its Fp as being the same as other antibodies or their fragments with known amino acid sequences because there is little difference in Fp between different IgG’s. Consequently, Fp for anti-β-hCG was taken to be the average value from the three IgG’s listed in Table 2, that is, 3.36 × 10-6 Å-2. Third, any uncertainty in scattering length density has no effect at all on the fitted layer thickness. IgG has a molecular weight of 150 kDa, and its molecular volume is around 1.76 × 105 Å3, giving a mass density of 1.42 g‚mL-1 for this protein. Thus, at the lowest concentration of 2 mg‚L-1, the φp and Γ of anti-β-hCG in the adsorbed layer have been calculated to be 0.15 ( 0.05 and 0.9 ( 0.3 mg‚m-2, respectively. The surface excess is comparable with that from our previous ellipsometric measurements (see Figure 6). The low protein volume fraction indicates that this protein layer is very loose. It is informative to correlate the AgBC of anti-β-hCG with the layer structural parameters (τ, φp, Γ) at 2 mg‚L-1. We found a high AgBC of about 0.71 in this case.18 The high AgBC corresponds to a flat-on orientation and is contrary to the common wisdom that a high AgBC indicates either an end-on orientation with the two Fab fragments pointing to the bulk solution or a “side-on” orientation with one Fab fragment away from the surface pointing to the solution. To be confident about this somewhat startling conclusion, it is important to interrogate the robustness of the neutron data analysis. According to the IgG molecular dimensions, the two standing orientations would produce two monolayer thicknesses of 85 and 142 Å, respectively. We have attempted to fit the measured reflectivity to these two thicknesses while the scattering length density in each case was floated to obtain the best fit possible, and the results are shown in Figure 3. The disagreement between the calculated profiles (dashed and dotted lines) and the measured profile indicates that the antibody cannot take either of the standing orientations at the hydrophilic silicon oxide/water interface. This adds confidence to the flat-on orientation model depicted in Figure 3 (solid line). Indeed, from a purely geometrical viewpoint, the flat-on orientation is more
Figure 3. Uniform layer fit to the reflectivity profile in the presence of 2 mg‚L-1 of anti-β-hCG in D2O with ionic strength at 20 mM and pH at 7. The solid line was calculated with a layer thickness of 40 ( 3 Å and a surface excess of 0.9 mg‚m-2. The dashed and dotted lines were calculated with thicknesses of 85 Å and 142 Å, respectively, but with their respective scattering length densities varied to obtain the fits closest to the measured data.
stable than other possible options, given that sufficient interfacial space is available. If it is assumed that the antibody is rigid, the limiting surface excess in the flat-on orientation is approximately 3.0 mg‚m-2. Because the protein layer is very loose with φp ) 0.15 at 2 mg‚L-1, the antigen binding sites should be easily accessible from the solution side, giving rise to a higher AgBC. NR assumes uniform layers, so it could be proposed that a minor fraction of the adsorbed antibodies were end-on and that these provided the binding capacity. However, the AgBC measurement showed that a majority of the adsorbed protein molecules were active in this case when the antibody surface coverage was low and the data were well-described by a singlelayer, flat-on orientation. Hence, we need to address whether the binding activity of the flat-on orientation could, contrary to the prevailing assumptions, reasonably be asserted to be high. hCG is not a small molecule (Mr ∼ 38 kDa) and is elongated (75 × 35 × 30 Å). The 3468 clone interacts with one end of the molecule and not with the peripheral glycosides.59 Inspection of a published crystal structure59 of a ternary complex of anti-R and anti-β Fv fragments with hCG, where the anti-β Fv was derived from the same monoclonal antibody as that studied here, indicates that a high AgBC could be obtained from a flat-on orientation of the whole antibody molecule, provided that there was sufficient space around the binding site and that the Fv was able to twist away from the surface somewhat. A more general implication is that only those antibodies that retained a sufficient degree of structural flexibility, or for which the binding site was accessible (48) Lu, J. R.; Su, T. J.; Thirtle, P. N.; Thomas, R. K.; Rennie, A. R.; Cubitt, R. J. Colloid Interface Sci. 1998, 206, 212. (49) Lu, J. R.; Su, T. J.; Thomas, R. K.; Penfold, J.; Richards, R. W. Polymer 1996, 37, 109. (50) Malmsten M. J. Colloid Interface Sci. 1994, 116, 333. (51) Murphy, E. F.; Lu, J. R.; Brewer, J.; Russell, J.; Penfold, J. Langmuir 1999, 15, 1313. (52) Murphy, E. F.; Keddie, J. L.; Lu, J. R.; Brewer, J.; Russell, J. Biomaterials 1999, 20, 1501. (53) Andrade, J. D.; Hlady, V. AdV. Polym. Sci. 1986, 79, 1. (54) Lin, J. N.; Chang, I. N.; Andrade, J. D.; Herron, J. N.; Christensen, D. A. J. Chromatogr. 1991, 542, 41. (55) Hoffman, W. L.; O’Shannessy, D. J. J. Immunol. Methods 1988, 112, 113. (56) Suter, M.; Butler, J. F. Immunol. Lett. 1986, 13, 313. (57) Lin, J. N.; Chang, I. N.; Andrade, J. D.; Herron, J. N.; Christensen, D. A. J. Chromatogr. 1991, 542, 41. (58) Wolfe, C. A. C.; Hage, D. S. Anal. Biochem. 1995, 231, 123. (59) Tegoni, M.; Spinelli, S.; Verhoeyen, M.; Davis, P.; Cambillau, C. J. Mol. Biol. 1999, 289, 1375.
6318 Langmuir, Vol. 22, No. 14, 2006
Figure 4. Different model fittings to the reflectivity profile in the presence of 10 mg‚L-1 of anti-β-hCG in D2O with ionic strength fixed at 20 mM and pH at 7. The solid line was calculated using a three-layer model whose parameters are listed in Table 2. The dashed line was calculated using a one-layer model with a layer thickness of 48 Å and a surface excess of 2.40 mg‚m-2.
to the antigen from the side face, would retain binding activity upon adsorption. The single uniform layer model has also been used to fit the reflectivity profile for the layer formed by adsorption from 10 mg‚L-1 antibody solution. The closest fit was calculated using τ ) 48 Å and F ) 5.3 × 10-6 Å-2 and is shown as a dashed line in Figure 4 together with the measured reflectivity profile. It can be seen from Figure 4 that, below 0.13 Å-1, this single uniform model fits the measured reflectivity well, suggesting that the protein layer becomes not only more packed, but also thicker, and that the increase in layer thickness indicates either some extent of molecular tilting or thickening due to the increased lateral packing density. Above 0.13 Å-1, the calculated reflectivity is obviously lower than the measured one, suggesting that the actual protein layer is most likely nonuniform in its structural distribution, that is, its density profile does not follow a uniform layer distribution. A two-layer model consisting of a mixture of flat-on and tilted orientations was subsequently tried, and was found to be unsuccessful, even though the thickness and the scattering length density in each sublayer were varied over a wide range. This urged us to consider other possible options. Because the isoelectric point of anti-β-hCG is about pH ) 6, its net charge is obviously negative at pH ) 7.0, which would make the adsorbed layer be in electrostatic repulsion, on the whole, with the negatively charged silica surface. Furthermore, with increased packing density, not only would the electrostatic repulsion between the layer and the oxide surface increase, but also the lateral electrostatic repulsion between the charged protein molecules within the layer would increase. These repulsive interactions together would impose strain within the layer. To relieve the strain energy, some local segments bearing net negative charges could be kept away from the oxide surface, thereby giving rise to some degree of deformation and resulting in some depletion of the polypeptide distribution in the proximity of the oxide surface. The globular structural integrity of the antibody molecule could be retained despite these local structural changes. In addition, some IgG molecules could tilt and contribute to the formation of an outer diffuse layer. Consequently, the measured reflectivity was well fitted by a three-layer model consisting of an inner sublayer (next to the silicon oxide) of 10 Å, a middle sublayer of 30 Å, and an outer sublayer of 25 Å. The calculated profile is shown as the solid line in Figure 4. The protein volume fractions in each sublayer were different, and the values were 0.22, 0.42, and 0.10 for the inner, middle, and outer sublayers,
Xu et al.
respectively. It is very important to realize that the three-layer model we have proposed gives no indication of any major structural unfolding for the protein, given that the current approach has limited sensitivity to the structural state of the Fab segments. Protein unfolding upon adsorption would give rise to a gradient of volume fraction of polypeptide with the highest volume fraction next to the surface, as demonstrated previously for lysozyme adsorption on a hydrophobic surface.48 The volume fraction of the inner sublayer is significantly lower than that of the middle sublayer, indicating that this sublayer is not derived from structural denaturation, but from the increased electrostatic repulsion between the layer and the surface forcing some segments of the protein away from the surface. Furthermore, the absence of structural denaturation is implied by the reversibility of antiβ-hCG adsorption with respect to pH, a feature common for proteins retaining their globular framework. The loose outer sublayer was ∼40 Å away from the oxide layer, and any Fab segments within this outer sublayer would not experience much steric hindrance. The minor proportion of segments within this outer sublayer might be responsible for the bioactivity of the adsorbed antibody at high surface coverage. However, Fabs inside the middle and inner sublayers would suffer from steric hindrance due to the high volume fraction of protein in the middle sublayer. Hence, the overall low AgBC can be explained. On the basis of the analysis described above, a possible structural conformation for the adsorption of a 10 mg‚L-1 antiβ-hCG layer is a combination of a flat-on monolayer 40 Å thick, and a tilted flat-on monolayer 65 Å thick, with the fraction of the tilted molecule being about 50%, and with both orientations having a depleted region 5-10 Å thick next to the oxide layer. It should be noted that a tilted layer 65 Å thick could be formed by the flat-on orientation with any of its three fragments being randomly tilted. It is unlikely that the fragments in the outer sublayer arise from the side-on orientation. If this were the case, one would expect a higher concentration of Fab fragments in the outer sublayer and a higher AgBC. When the anti-β-hCG concentration in solution was increased from 2 to 10 mg‚L-1, the surface excess exhibited a significant increase from 0.9 to 2.5 mg‚m-2, and the corresponding structure of the protein layer definitively changed from a uniform layer to a distribution better described by a three-layer model. At 10 mg‚L-1, the surface excess from anti-β-hCG adsorption was close to its saturation limit. Further increase in concentration from 10 to 50 mg‚L-1 led to a further increase in surface excess, from 2.5 to 2.9 mg‚m-2. The best fit for the measured reflectivity at 50 mg‚L-1 was indeed also a three-layer model consisting of a loosely packed inner sublayer of 14 Å, a densely packed middle sublayer of 35 Å, and a diffuse outer sublayer of 35 Å, with corresponding protein volume fractions of 0.3, 0.44, and 0.09, respectively. The calculated profile is shown in Figure 5 as a solid line. Compared with the three-layer structure formed from the 10 mg‚L-1 antibody, increase in the solution concentration to 50 mg‚L-1 caused further deformation and tilting of the adsorbed molecules accompanied by more densely packed protein at the interface, consistent with increased repulsive interactions in a more densely packed layer. The structural conformation of the adsorbed anti-β-hCG layer at 50 mg‚L-1 can be represented by a combination of a slightly tilted flat-on monolayer of 44 Å and a more significantly tilted flat-on monolayer of 79 Å with the fraction of the latter being about 35%. Both orientations had a depleted polypeptide layer of 14 Å next to the oxide layer. We again hypothesize that the 79 Å-thick layer was formed by tilting from the flat-on orientation rather than by flattening from the side-on orientation, even though the dimension is close to the
Orientation of an Adsorbed Monoclonal Antibody
Figure 5. Different model fittings to the reflectivity profile in the presence of 50 mg/L of anti-β-hCG in D2O with ionic strength fixed at 20 mM and pH at 7. The solid line was calculated using a threelayer model whose parameters are listed in Table 2. The dashed line was calculated using a one-layer model with a layer thickness of 52 Å and a surface excess of 2.60 mg/m2.
Figure 6. Surface excesses obtained from the one-layer (O) and the three-layer (0) fits with respect to the anti-β-hCG concentrations. The solution ionic strength was fixed at 20 mM and pH at 7. The surface excesses obtained from the previous ellipsometric measurements18 are also shown (×) for comparison.
projected value from the side-on monolayer, because the AgBC was low. In Figure 5, the reflectivity calculated from the best uniform layer model with τ ) 52 Å and F ) 5.25 × 10-6 Å-2 is also shown for comparison (dashed line). As in the case of 10 mg‚L-1, the single uniform model fitted the measured reflectivity well below κ ) 0.12 Å-1, but, over the high κ range (above 0.12 Å-1), the fit became poor because of the presence of some fine sublayer structure. Despite the obvious deviation over the high κ range, a uniform layer model was adequate for the determination of the adsorbed amount, as has been demonstrated previously.49 The surface excesses calculated from the uniform layer model at different concentrations are given in Figure 6, together with the values obtained from the three-layer models at higher concentrations. The surface excesses obtained from the previous ellipsometric measurements are also shown for comparison. Before comparing these values, it is necessary to interpret how to calculate the surface excess from the above three-layer model. Although NR is sensitive to the protein distribution along the surface normal direction, it cannot reveal how the different segments partition into the interface, and there is little information from other techniques that could help to resolve this situation. Because the scattering length densities in D2O for the different IgG’s and their different parts are very similar (within 2 × 10-7 Å-2), as shown in Table 2 for the surface excess calculation
Langmuir, Vol. 22, No. 14, 2006 6319
based on eqs 1 and 2, it is reasonable to take scattering length densities for all sublayers here to be the same and equal to that for the whole anti-β-hCG molecule (3.36 × 10-6 Å-2). Some of the scattering length densities cited in Table 2 are referenced from the work reported by Cooper et al.,43 which also shows similar values for a range of proteins and oligosaccharides of different origins and biological functions. Figure 6 shows that the surface excesses derived from the single uniform layer and the three-layer models are in excellent agreement, and that both are also close to the results obtained from the standard error (SE) measurements. Su et al.20,21,38 investigated the adsorption of lysozyme and bovine serum albumin (BSA) at the hydrophilic silica/water interface using NR, and found no structural alteration upon adsorption of lysozyme, though some extent of tilting was observed under strong electrostatic interaction. However, for BSA whose three principal domains are loosely joined together, some obvious flattening occurred upon direct contact with the solid surface, suggesting some degree of deformation. In comparison with an antibody, lysozyme is more robust, and BSA is more flexible. Our results indicate that, under electrostatic repulsion, the antibody globular structure was retained upon adsorption. Some negatively charged segments next to the oxide surface were pushed away from the surface, but the extent of the structural deformation was lower than that of BSA because this local effect did not cause the flattening of the IgG molecule. At the lowest concentration of 2 mg/L, the deduced uniform layer distribution with a thickness of 40 Å compares well with the short axial length of 38 Å. Because the surface excess was low, it was difficult to distinguish the three-layer model, with sublayer scattering length densities varying by less than 0.2 × 10-6 Å-2 from the single uniform layer model. If significant deformation existed at the lower surface coverage, the corresponding AgBC should have been very low, which was not the case. We have previously shown that, with increasing concentration of anti-β-hCG, the adsorbed protein amount increases, but the corresponding AgBC decreases substantially.18 On the basis of the above structural analyses for protein layers adsorbed, it seems that the steric hindrance within the layer rather than the orientation adopted by the antibody molecule is the key factor in determining the AgBC. At 2 mg‚L-1, the surface excess was 0.9 mg‚m-2, far less than the saturated value of 3.0 mg‚m-2 estimated for the flat-on orientation. The space between adsorbed antibody molecules would be large, and the antigen molecules could relatively easily access the antibody active sites. There was no detectable deformation that would otherwise reduce the bioactivity of the adsorbed molecule. These structural conditions underpin the observed high AgBC of 0.7 for the loose adsorbed layer formed from a 2 mg/L antibody solution. Although this value is well above those reported in the literature, it only represents 1/3 of the hypothetically available antibody binding sites. The constraint of side-on access of the antigen can account for this effect. With 10 mg‚L-1 antibody in the adsorption solution, the surface excess of anti-β-hCG increased to 2.50 mg‚m-2, still below the maximum value. However, the surface crowding was associated with deformation of the adsorbed molecules near the oxide surface and tilting into the bulk solution. The main contribution to the AgBC would likely be from the outer diffuse layer with only a minor contribution from the two inside sublayers due to steric obstruction and deformation. It can be seen from Table 1 that the middle sublayer was rather closely packed and that it is reasonable to presume that the closer packing of the middle sublayer would prevent access to antigen binding sites both within this sublayer and within the inner sublayer. At 10
6320 Langmuir, Vol. 22, No. 14, 2006
mg‚L-1, the AgBC was only 0.14, showing that the majority of the binding sites were inaccessible. As the surface excess increased at 50 mg‚L-1 to 2.90 mg‚m-2, almost equivalent to complete flat-on surface packing, the antibody layer was even more crowded, explaining the further decrease of AgBC, around 0.10. A number of studies have used ellipsometry to characterize antibody adsorption,14,44,46,50 and it is useful to comment on its technical strength. Although ellipsometry is very sensitive to the adsorbed amount, it is incapable of separating the layer thickness from the refractive index if the layer is ultrathin (less than a few hundredths of an angstrom) because of the strong correlation between the two parameters.51-53 As a consequence, some evident inconsistency about the thickness of the antibody layer has been reported.14,44,50 For instance, it has been reported that the adsorption of IgG on silica and methylated silica surfaces gives layer thicknesses around 170-180 Å, thus suggesting a primary end-on configuration. This result can be compared with another ellipsometrical work showing a layer 40 Å thick from antibody adsorption on methylated silica surface, suggesting a flat-on orientation. Despite these evident differences, the surface adsorbed amount reported in the different studies was comparable. The most common method for the immobilization of antibodies onto biosensors and immunoassays is still via physical adsorption due to its simplicity and cost effectiveness. When physically immobilized, the antibody is usually adsorbed onto the solid surface to the saturation level because it is commonly perceived that the higher the surface coverage of antibodies, the higher the surface activity. Our study has, in fact, revealed an opposite trend. The significance of this study is to reveal the structural conformations of an antibody under a range of concentrations, thereby revealing the molecular mechanistic processes underlying the accessibility of the binding sites. NR measurements show that, over the entire concentration range studied, the antibody molecule adopts a flat-on orientation. As the surface concentration increases, the adsorbed layer is more densely packed, and the electrostatic repulsion within the protein layer and between the protein and surface increases, forcing further structural deformation. The steric hindrance and the strain imposed by the electrostatic repulsion may together constrain the accessibility of the antigen to the binding sites within the adsorbed antibody layer. Although structural orientations such as head-on and sideon are, in principle, much to be preferred, we have shown that, within the surface and solution conditions studied here, it was impossible for the antibody to take orientations other than the flat-on state. Covalent coupling of the antibody through its Fc region has been extensively developed in an attempt to enhance the AgBC of antibodies on a solid surface.12,19,54-58 From the perspective of simple law of physics, it is most unlikely to envisage that a single bond is sufficient to support the large and relatively “soft” antibody molecule and force it to stand and adopt the head-on orientation. Indeed, a similar trend of drastic decrease of AgBC with increased antibody surface coverage has recently been
Xu et al.
reported for antibodies surface-coupled through streptavidinbiotin.60 However, it remains interesting for the future to unravel the conformational orientations when an antibody and its Fab fragments are anchored onto the surface via chemical bonding. It will also be relevant to explore how the effect of surface hydrophobicity and charge affect their orientations using the combined approach of AFM and NR.
Conclusion The main finding of this work was that the antibody adsorbed predominantly flat-on, that is, with the three structural fragments lying flat on the silicon oxide surface. This feature was confirmed by under-water AFM measurement. AFM imaging further revealed nonuniform surface clustering, with the number of antibody molecules in each cluster varying from 2 to 15. Surface drying was found to increase surface clustering. The height of AFM scans for these clusters also increased in air, showing that the aggregation had occurred in three dimensions instead of two dimensions as detected in water. The formation of surface clusters indicated that the interaction between the antibody and the hydrophilic and negatively charged oxide surface is weak, and the attraction between neighboring proteins is relatively strong. An increase in antibody concentration from 2 to 10 mg‚L-1 led to a structural transition from a uniform layer to an inhomogeneous distribution well represented by a three-layer model, characterized by an inner sublayer of 10 Å lean in polypeptide, a middle sublayer of 30 Å rich in polypeptide, and an outer sublayer again lean in polypeptide. This distribution can be explained as the flat-on orientation with some of the fragments tilted toward the outer solution phase. The inner 10 Å proteindepleted region reflected the electrostatic repulsion between the charged protein fragments and the negatively charged surface. Further increase in concentration to 50 mg‚L-1 resulted in a surface excess of 2.9 mg/m2, comparable to the value for fully packed flat-on antibody adsorption. The adsorbed antibody layer clearly showed a three-layer distribution. The steady increase in surface excess was in sharp contrast to the fast decline in antigen binding activity. A reasonable explanation is that the antigen accessibility is constrained by the steric hindrance and the straining caused by electrostatic repulsion within the protein layer and between the protein and the substrate surface. The high volume fraction within the middle sublayer is most likely to hinder the access of antigen from reaching the binding sites within the middle and inner sublayers, and the main contribution to the binding capacity could come from the outer sublayer. Acknowledgment. We thank Unipath Ltd. for funding and for the supply of hCG and antibodies. We thank M. M. Gani and R. Davies (Unipath) for insightful discussion. LA0532454 (60) Vareiro, M. L. M.; Liu, J.; Knoll, W.; Zak, K.; Williams, D.; Jenkins, A. T. A. Anal. Chem. 2005, 77, 2426.
