In Situ Single-Molecule Detection of Antibody−Antigen Binding by

Two antigen/antibody pairs were investigated: chorionic gonadotropin (hCG)/mouse monoclonal anti-hCG and goat IgG (anti-intact hCG)/mouse monoclonal ...
0 downloads 0 Views 345KB Size
Download PDF
Anal. Chem. 2002, 74, 6017-6022

In Situ Single-Molecule Detection of Antibody-Antigen Binding by Tapping-Mode Atomic Force Microscopy Lingyan Li,† Shengfu Chen,† Seajin Oh,‡ and Shaoyi Jiang*,†

Department of Chemical Engineering, University of Washington, Seattle, Washington 98195, and Applied Physical Sciences Laboratory, SRI International, Menlo Park, California 94026

We performed in situ detection of specific and nonspecific binding during immunoreaction on surfaces at the same location before and after analyte was injected using tapping-mode atomic force microscopy (TM-AFM) in liquid and demonstrated the ability of TM-AFM to monitor the occurrence of single-molecule binding events and to distinguish nonspecific from specific binding by examining topographical change. Two antigen/antibody pairs were investigated: chorionic gonadotropin (hCG)/mouse monoclonal anti-hCG and goat IgG (anti-intact hCG)/ mouse monoclonal anti-goat IgG. Antibody (or antigen) molecules were covalently immobilized on uniform mixed self-assembled monolayers (SAMs) terminated with carboxylic acid and hydroxyl groups. Mixed SAMs allow the control of the density of immobilized antibody (or antigen) on surfaces to achieve the detection of individual antigens, antibodies, and antigen/antibody complexes. This in situ TM-AFM-based detection method allows the singlemolecule detection of antigen/antibody binding under near-physiological environment and the distinction of nonspecific from specific binding. It could be extended into a microarray. Ever since its invention, atomic force microscopy (AFM)1 has been widely used in biotechnology and biomedical research, including imaging, force mapping, and sensing applications.2-6 AFM can be used not only to image at the subnanometer scale but also to probe nanoscale forces. Furthermore, AFM can operate in a liquid environment and monitor various processes (e.g., adsorption and reaction) in situ, allowing one to obtain detailed static and dynamic information of biological systems under nearphysiological conditions. Conventional immunoassay techniques, * To whom correspondence should be addressed. E-mail: sjiang@ u.washington.edu. † University of Washington. ‡ SRI International. (1) Binnig, G.; Quate, C. F.; Gerber, C. Phys. Rev. Lett. 1986, 56, 930. (2) Hansma, H.; Hoh, J. Annu. Rev. Biophys. Chem. 1994, 23, 115. (3) Bustamante, C.; Rivetti, C.; Keller, D. Curr. Opin. Struct. Biol. 1997, 7, 709. (4) Takano, H.; Kenseth, J. R.; Wong, S.; O’Brien, J. C.; Porter, M. D. Chem. Rev. 1999, 99, 2845. (5) Tiefenauer, L.; Ros, R. Colloids Surf. B 2002, 23, 95. (6) Reich, Z.; Kapon, R.; Nevo, R.; Pilpel, Y.; Zmora, S.; Scolnik, Y. Biotechnol. Adv. 2001, 19, 451. 10.1021/ac0258148 CCC: $22.00 Published on Web 10/29/2002

© 2002 American Chemical Society

such as radioimmunoassay (RIA) and enzyme immunoassay (EIA), and relatively new techniques, such as surface plasmon resonance (SPR) biosensor, are based on statistical information for a system involving a large number of molecules, for which inhomogeneities are averaged. The ability for AFM to image biological systems at high spatial resolution in liquid in real time makes AFM-based detection methods very attractive to provide local molecular interaction information down to single-molecule resolution and to distinguish nonspecific from specific binding. The feasibility of using AFM as a sensitive tool in analytical biochemical assays and biosensors has been explored recently.7-12 The concept of AFM-based immunoassay detection was proposed by Masai et al.,7 in which scanning tunneling microscope (STM) was used to image gold colloidal particles with attached protein A bound to antibody molecules on conductive surfaces. Dong and Shannon8 covalently attached immunoreagents to two-component self-assembled monolayers (SAMs) of carboxylic acid- and methylterminated thiols and characterized antibody/antigen binding using contact-mode AFM in air. Quist et al.9 distinguished individual human serum albumin (HSA), rabbit anti-HSA, and HSA/anti-HSA complexes adsorbed on mica using tapping-mode AFM (TM-AFM) in air. Distinction was made based on height differences, variation in surface roughness (Rrms), or volume distribution. However, none of these experiments was carried out in situ before and after analyte was injected. Therefore, it is impossible to distinguish nonspecific binding from specific in these experiments. Furthermore, in situ detection of antigen/antibody binding will have much higher sensitivity than those methods using different samples (or different locations on the same sample) before and after immunoreaction. In addition, for sample preparation in many of these previous experiments, proteins were often dried. Upon drying, denaturation or aggregation of proteins could occur. In this work, we performed in situ detection of specific and nonspecific interactions during immunoreaction before and after analyte was injected using TM-AFM in liquid and demonstrated (7) Masai, J.; Sorin, T.; Kondo, S. J. Vac. Sci. Technol. 1990, 8, 713. (8) Dong, Y.; Shannon, C. Anal. Chem. 2000, 72, 2371. (9) Quist, A. P.; Bergman, A. A.; Reimann, C. T.; Oscarsson, S. O.; Sundqvist, U. R. B. Scanning Microsc. 1995, 9, 395. (10) Browning-Kelley, M. E.; Wadu-Mesthrige, K.; Hari, V.; Liu, G. Y. Langmuir 1997, 13, 343. (11) Perrin, A.; Lanet, V.; Theretz, A. Langmuir 1997, 13, 2557. (12) Bergkvist, M.; Carlsson, J.; Oscarsson, S. J. Phys. Chem. B 2001, 105, 2062.

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002 6017

the ability of TM-AFM to monitor the occurrence of singlemolecule binding events and to distinguish nonspecific from specific binding by examining height change in order to discriminate the true signal due to binding events from background noise. To achieve the goal, three aspects are very important. First, AFM should be stable enough so that images can be taken at the same scanning location before and after analyte is injected. The ability to perform AFM at the same scanning location makes it possible to distinguish nonspecific binding from immunoreactions and to detect specific binding in nanometer-scale resolution without labeling. The stability of TM-AFM measurements of immobilized proteins is demonstrated by repeating scanning at the same location for a couple hours. Second, experiments should be performed by TM-AFM in liquid. Unlike contact-mode AFM, the use of TM-AFM can reduce deformation or destruction of biological samples due to significant reduction in lateral forces. Unlike in air, AFM experiments in liquid environment will preserve the structure and functionality of biological samples and prevent denaturation and aggregation of proteins upon drying. Finally, antigens or antibodies should be immobilized on surfaces strong enough for AFM imaging, but without losing their bioactivities, and the density and homogeneity of immobilized antigens or antibodies should be controlled to allow single-molecule binding detection. This can be achieved by covalently immobilizing antigens or antibodies to surfaces covered by molecular-scale uniform mixed SAMs. We demonstrated the control of immobilized protein density on a surface by adjusting the composition of a mixed alkanethiol solution. EXPERIMENTAL SECTION Materials. 11-Mercaptoundecanol (C11OH), and 16-mercaptohexadecanoic acid (C15COOH) were purchased from Aldrich Chemical Co. and used as received. Phosphate-buffered saline (PBS; 138 mM NaCl, 2.7 mM KCl, pH 7.4), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC), and N-hydroxysulfosuccinimide (NHS) were purchased from Sigma Chemical Co. Chorionic gonadotropin (hCG, C0713) and mouse monoclonal anti-hCG (MC097) were purchased from Scripps Laboratories. Goat immunoglobulin (IgG) (anti-intact hCG) (D10500G) and mouse monoclonal anti-goat IgG were purchased from Biodesign. Au(111) Substrate. Gold substrates were prepared by the vapor deposition of gold onto freshly cleaved mica (AshevilleSchoonmaher Mica Co.) in a high-vacuum evaporator (BOC Edwards Auto306) at ∼10-7 Torr. Mica substrates were preheated to 325 °C for 2 h by a radiator heater before deposition. Evaporation rates were 0.1-0.3 nm/s, and the final thickness of gold films was ∼200 nm. Gold-coated substrates were annealed in H2 frame for 1 min before use. SAMs. SAMs were formed by soaking gold-coated substrates (immediately after vacuum deposition or annealing by H2 flame) in the ethanol solution of alkanethiols (1 mM) preheated to 60 °C (in a oven) overnight. SAMs containing carboxylic acid terminal groups were rinsed sequentially with pure ethanol, 10% acetic acid, and ethanol and dried in a N2 stream. Unlike methyl-terminated SAMs, rinsing carboxylic acid-terminated SAMs only with pure ethanol solution will result in unbound thiol molecules present on the surface.13 (13) Li, L. Y.; Chen, S. F.; Jiang, S., submitted to Langmuir.

6018

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

Immunoreagent Immobilization. Carboxylic acid terminal groups of SAMs were activated by 2 mg/mL NHS and 2 mg/mL EDC for 1 h before protein immobilization,14 then rinsed thoroughly with water, and dried in nitrogen. The substrate covered with activated SAMs was then immersed into 5 µg/mL protein solutions (i.e., protein G purified anti-goat IgG or hCG) at 4 °C overnight (Figure 1). Finally, the specimen was thoroughly rinsed with PBS buffer solution and transferred onto the AFM sample stage. TM-AFM. All TM-AFM images were acquired using a multimode Nanoscope III (DI) equipped with a 10-µm E scanner. Commercial Si3N4 cantilevers (NP-S20, DI) with force constants of 0.12 N/m were used. Images were recorded with typical scan rates of 1.0-2.0 Hz. Immobilized proteins (i.e., anti-goat IgG or hCG) were first imaged in PBS buffer solution. Analytes (i.e., goat IgG or anti-hCG) in PBS were then injected into the AFM liquid cell (Figure 2). The change in surface morphology was monitored at the same locations in TM-AFM. One can extract information on single-molecule antigen/antibody binding processes by comparing the height change in AFM images taken before and after analyte is injected. RESULTS AND DISCUSSION Protein Immobilization. Mixed SAMs are promising as platforms for protein adsorption and immobilization15-17 due to the possibility to control chemical and structural properties of a surface by adjusting the abundance, type, and spatial (both normal and lateral) distribution of tail groups. Recently, we developed two methods to prepare molecule-scale uniform mixed SAMs,18,19 in which the uniform distribution of immunoreagents with appropriate surface density can be achieved. In this work, we used mixed SAMs of carboxylic acid-terminated long-chain thiol (C15COOH) and hydroxyl-terminated short-chain thiol (C11OH) to covalently attach antibody or antigen to a substrate. Carboxylic acid-terminated SAMs were chosen to provide a functional group for the attachment of protein molecules, while hydroxyl-terminated SAMs were chosen to mimic protein resistance. The chemical immobilization of immunoreagents on mixed SAMs/Au(111) substrates is illustrated in Figure 1. Mixed SAMs of C15COOH and C11OH are first formed on Au(111) substrates for 24 h at 60 °C. Then, carboxylic acid terminal groups of mixed SAMs are chemically activated by NHS and EDC, followed by the formation of amide bonds between COOH groups on SAMs and exposed lysine residues on proteins. The surface density of antibody (or antigen) can be controlled through varying C15COOH composition in mixed SAMs. Figure 3 shows the topographical images of immobilized antigoat IgG (5 µg/mL) from TM-AFM in PBS buffer solution. Bright spots can be clearly identified in the images, and they are antigoat IgG molecules. As shown in the AFM images, the variation (14) Homola, J.; Dostalek, J.; Chen, S.; Rasooly, A.; Jiang, S.; Yee, S. S. Int. J. Food Microbiol. 2002, 75, 61. (15) Patel, N.; Davies, M. C.; Heaton, R. J.; Tendler, S. J. B.; Williams, P. M. Appl. Phys. A 1998, 66, S569. (16) Higashi, N.; Takahashi, M.; Niwa, M. Langmuir 1999, 15, 111. (17) Lahiri, J.; Isaacs, L.; Grzybowski, B.; Carbeck, J. D.; Whitesides, G. M. Langmuir 1999, 15, 7186. (18) Chen, S. F.; Li, L. Y.; Boozer, C. L.; Jiang, S. Langmuir 2000, 16, 9287. (19) Chen, S. F.; Li, L. Y.; Boozer, C. L.; Jiang, S. J. Phys. Chem. B 2001, 105, 2975.

Figure 1. Chemical immobilization of antibody (or antigen) molecules on molecular-scale uniform mixed self-assembled monolayers/ Au(111).

tration of C15COOH in mixed C15COOH/C11OH solution at low and moderate concentrations. On pure C15COOH SAMs, however, the amount of immobilized anti-goat IgG is lower than that on mixed C15COOH/C11OH SAMs. This is likely caused by the steric packing of terminal carboxylic acid groups, which limits the formation of ester intermediate. Conversion of accessible acid groups up to 80% could be achieved after three repeated reaction cycles, while only ∼50% of carboxylic acid groups are converted to amides during the first cycle of ester formation as reported by Frey and Corn.20 To test the coupling strength of immobilized anti-goat IgG, AFM scanning was repeated several times for 48 min on the same surface area covered by anti-goat IgG. Results show that anti-goat IgG molecules were not displaced (results not shown) for 48 min, verifying that cross-linking proteins on SAMs are stable for repeated AFM scans and are suitable for in situ AFM measurements in liquid. In Situ TM-AFM Detection of Single-Molecule Binding Events. Immunoglobulin (∼150 kDa) consists of three fragments, two separate and identical Fab fragments with active sites for antigen binding and one Fc fragment. The three-dimensional structure is T- or Y-shaped with a dimension of 14.2 × 8.5 × 3.8 nm3 measured by X-ray diffraction.21 Besides tip convolution and tip/sample interactions, viscoelastic properties of soft proteins could affect apparent height in TM-AFM.22 Tip force could compress vertical height but broaden lateral dimensions of an adsorbed protein molecule. Thus, the dimensions of an adsorbed protein molecule measured by TM-AFM are relative. Furthermore, adsorbed protein will spread due to protein/surface interactions. The height of IgG in the TM-AFM images is ∼3 nm while lateral dimensions ∼25-30 nm. In situ detection of the binding of free antigen to immobilized antibody was performed by TM-AFM in solution. An AFM image was first taken in PBS buffer solution for immobilized anti-goat IgG on mixed SAMs (1:99 for C15COOH/C11OH in solution). Then, 1 mL of 5 µg/mL goat IgG was injected into the fluid cell and flowed over the immobilized antibody. After incubation for 30 min, the surface was thoroughly rinsed with flowing PBS buffer solution. Finally, the same area was scanned by TM-AFM in PBS buffer solution. AFM topographical images of immobilized antigoat IgG at the same location before and after goat IgG injection are shown in Figure 4a and c, respectively, while corresponding (off-line) zoomed-in three-dimensional images are shown in Figure 4b and d. Dark features are Au defects, which are used as markers to ensure that the same location is scanned although a slight shift of TM-AFM cannot be totally avoided. On the mixed SAMs, antigoat IgG molecules are separated from each other to provide individual capture sites for single-molecule binding without steric hindrance from neighbor antibodies. At the site indicated by an arrow in Figure 4, peak height increases almost twice, indicating the occurrence of immunoreaction. As discussed previously, height information from TM-AFM is not exactly the height of a molecule.22 Despite this, scanning at the same location in TMAFM allows quantitative identification of different species on surfaces based on their relative difference in height. Many

in anti-goat IgG density results from that in the C15COOH concentration of mixed C15COOH/C11OH solution and the number of immobilized anti-goat IgG increases with the concen-

(20) Frey, B. L.; Corn, R. M. Anal. Chem. 1996, 68, 3187. (21) Silverton, E. W.; Navia, M. A. Davies, D. R. Proc. Natl. Acad, Sci. U.S.A. 1977, 74, 5140. (22) Radmacher, M.; Fritz, M.; Cleveland, J. P.; Walters, D. R.; Hansma, P. K. Langmuir 1994, 10, 3809.

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

6019

Figure 2. Schematic diagram of the experimental setup used for in situ TM-AFM imaging in liquid environment. With this setup, liquid can be injected into or removed from the liquid cell without significant perturbation to the system.

Figure 3. Tapping-mode AFM images of immobilized anti-goat IgG molecules on mixed SAMs. Variation in anti-goat IgG density results from that in the C15COOH concentration of C15COOH/C11OH mixed solution of 1% (a), 10% (b), and 100% (c). Large dark features represent Au defects. Z bar is 20 nm.

biosensing techniques, such as SPR, are based on a full monolayer of immobilized antibody molecules with flowing analytes. Unlike SPR, antigen is in static contact with antibody in TM-AFM. Thus, slow kinetics is expected for immunoreaction. In addition, low surface coverage of the immobilized antibody was chosen in this work in order to provide individual capture sites for singlemolecule detection. Thus, the sensitivity of TM-AFM-based biosensors could be significantly improved if a pump is added for the continuous flowing of analytes, the surface coverage of immobilized antibodies is increased, and better orientation of immobilized antibodies is achieved via adjusting surface chemistry. It should be pointed out that some new peaks were observed in Figure 4c and d. One example is shown by the dashed circle marked in Figure 4. These new peaks are due to nonspecific binding of antigen to hydroxyl or unreacted COOH groups on SAMs. Hydroxyl-terminated SAMs are used to mimic nonfouling surfaces. However, it is not as good as poly(ethylene glycol) (PEG)-terminated SAMs for protein resistance.23-25 6020

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

Increasing surface-immobilized immunoreagents will improve the sensitivity of a biosensor. To determine whether TM-AFM can still recognize individual antibody/antigen complexes when the surface coverage of immunoreagents is near-monolayer, we worked on another antigen/antibody pairschorionic gonadotropin (hCG) and mouse monoclonal ant-hCG. hCG at 5 µg/mL was first immobilized on mixed SAMs (1:9 for C15COOH/C11OH in solution) at 60 °C overnight, and a TM-AFM image was acquired. Then, 100 µg/mL anti-hCG was injected into the fluid cell of TMAFM. After 30 min, the surface was thoroughly rinsed with flowing PBS buffer solution. Finally, the same area was scanned by TMAFM in PBS buffer solution. TM-AFM images of immobilized hCG after exposure to anti-hCG are shown in Figure 5c and d. For comparison, images of the surface at the same location prior to (23) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164. (24) Luk, Y.; Kato, M.; Mrksich, M. Langmuir 2000, 16, 9604. (25) Chapman, R. G.; Ostuni, E.; Takayama, S.; Holmlin, R. E.; Yan, L.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 8303.

Figure 4. In situ tapping-mode AFM images of immobilized anti-goat IgG on mixed C15COOH/C11OH SAMs (1% in solution) before (a, b) and after (c, d) exposure to goat IgG. Images were taken in PBS buffer solution at the same location. Images b and d are corresponding (off-line) zoomed-in three-dimensional images of (a) and (c), respectively. The arrow refers to the specific binding location while the dashed circle refers to the nonspecific binding location. Dark features represent Au defects. Z bar is 20 nm.

antibody exposure are shown in Figure 5a and b. The surface concentration of immunoreagents was increased since the surface composition of C15COOH SAMs was increased. Under such a high surface coverage, we are still able to recognize individual antibody/antigen complexes by comparing TM-AFM images at the same location before and after antibody exposure. We observed many brighter “spots”, which are larger in both lateral and vertical directions after hCG was incubated with anti-hCG, indicating the occurrence of immunoreaction. These complexes are quite stable after repeated rinsing and scanning. From Figure 5c and d, two clearly identifiable types of features are observed. The smaller features are ∼2.9 nm in height, which are also observed prior to exposure to antibody, while the larger features are significantly higher (∼6.5 nm). On the basis of our analysis, we are confident to assign the larger features as antigen/antibody (hCG/anti-hCG) complexes (e.g., the one marked by the arrow in Figure 5c), while the smaller features as antigens (hCG) marked by the circle. Currently, we are extending this work to improve the sensitivity of these measurements and to achieve quantitative measurements

of binding events at different analyte concentrations. A pump has been added to the TM-AFM system for the continuous flow of analytes to increase the binding kinetics between analytes and immobilized proteins. Furthermore, PEG-coated AFM probe has been developed to avoid or reduce tip contamination by protein solution, particularly important for experiments with flowing analytes. Recently, we achieved the control of antibody orientation via an altering microenvironment (e.g., solution and surface properties).26 These improvements will be incorporated in our future TM-AFM experiments. Data analysis would be much more convenient and efficient if a computer program could be developed to process these TM-AFM images before and after antigen/ antibody binding. CONCLUSIONS In this work, we studied antigen/antibody binding in situ for two antigen/antibody pairs based on direct comparison of topographical change in AFM images at the same location before and (26) Chen, S. F.; Liu, L. Y.; Jiang, S., submitted to Langmuir.

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

6021

Figure 5. In situ tapping-mode AFM images of immobilized hCG on mixed C15COOH/C11OH SAMs (10% in solution) before (a, b) and after (c, d) exposure to anti-hCG. Images were taken in PBS buffer solution at the same location. Images b and d are corresponding (off-line) zoomedin three-dimensional images of (a) and (c), respectively. Arrow refers to the specific reaction location while dashed circle refers to the unreacted hCG location. Dark features represent Au defects. Z bar is 20 nm.

after analyte injection using TM-AFM in PBS buffer solution. This in situ TM-AFM-based detection method allows the singlemolecule detection of antigen/antibody binding under nearphysiological environment and the distinction of nonspecific from specific binding. The use of TM-AFM will reduce deformation or destruction of biological samples due to significant reduction in lateral forces while TM-AFM experiments in a liquid environment will preserve the structure and functionality of biological samples. Molecular-scale uniform mixed SAMs enable the stable and homogeneous immobilization of antigens or antibodies on surfaces with controlled surface density to provide individual capture sites for single-molecule detection. Many other immunoassay techniques, such as RIA, EIA, and SPR, are based on statistical results for a system involving a large number of molecules. The ability of an AFM-based immunoassay to monitor single-molecule binding events will improve the sensitivity of a biosensor and provide a wealth of new information about immunoreaction at the molecular level. The ability of an AFM-based immunoassay to distinguish nonspecific interaction from specific will make detection more reliable. Furthermore, this AFM-based method could be extended

6022 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

into a sensor array so that a large number of images acquired will provide quantitative and reliable results while each image will provide single-molecule resolution of antigen/antibody binding events. NOTE ADDED AFTER ASAP POSTING This article was released ASAP on 10/29/2002 with a minor error in the Figure 5 caption. The correct version was posted on 11/04/2002. ACKNOWLEDGMENT L.L. thanks the Center for Nanotechnology at the University of Washington for a graduate research fellowship and for the use of its tapping-mode AFM. The authors gratefully acknowledge the National Science Foundation for financial support under CTS0092699 (CAREER Award). S.O. acknowledges the National Institutes of Health for support, Grant 5 R21 CA86370-02. Received for review May 30, 2002. Accepted September 16, 2002. AC0258148