Surface Modification for Enhancing Antibody Binding on Polymer

Sep 28, 2006 - molecule, hexamethylenediamine (HMD). Consequently, PEI surface modification greatly improved the immunoassay performance of the ...
0 downloads 0 Views 590KB Size
Download PDF
9458

Langmuir 2006, 22, 9458-9467

Surface Modification for Enhancing Antibody Binding on Polymer-Based Microfluidic Device for Enzyme-Linked Immunosorbent Assay Yunling Bai,† Chee Guan Koh,† Megan Boreman,† Yi-Je Juang,‡ I-Ching Tang,‡ L. James Lee,† and Shang-Tian Yang*,† Department of Chemical and Biomolecular Engineering, The Ohio State UniVersity, Columbus, Ohio 43210, and Bioprocessing InnoVatiVe Company, 1275 Kinnear Road, Columbus, Ohio 43212 ReceiVed April 25, 2006. In Final Form: August 6, 2006 A novel surface treatment method using poly(ethyleneimine) (PEI), an amine-bearing polymer, was developed to enhance antibody binding on the poly(methyl methacrylate) (PMMA) microfluidic immunoassay device. By treating the PMMA surface of the microchannel on the microfluidic device with PEI, 10 times more active antibodies can be bound to the microchannel surface as compared to those without treatment or treated with the small amine-bearing molecule, hexamethylenediamine (HMD). Consequently, PEI surface modification greatly improved the immunoassay performance of the microfluidic device, making it more sensitive and reliable in the detection of IgG. The improvement can be attributed to the spacer effect as well as the functional amine groups provided by the polymeric PEI molecules. Due to the smaller dimensions (140 × 125 µm) of the microchannel, the time required for antibody diffusion and adsorption onto the microchannel surface was reduced to only several minutes, which was 10 times faster than the similar process carried out in 96-well plates. The microchip also had a wider detection dynamic range, from 5 to 1000 ng/mL, as compared to that of the microtiter plate (from 2 to 100 ng/mL). With the PEI surface modification, PMMAbased microchips can be effectively used for enzyme linked immunosorbent assays (ELISA) with a similar detection limit, but much less reagent consumption and shorter assay time as compared to the conventional 96-well plate.

Introduction Due to the high selective molecular recognition and high sensitivity, assays via antibody-antigen interactions have drawn a great deal of interest. Enzyme-linked immunosorbent assay (ELISA) is one of the most widely used immunoassays for detection and quantification of chemical and biological molecules in the biotechnology industry. It is also becoming more and more important in clinical diagnostics, food safety testing, and environmental monitoring. However, the conventional ELISA, typically carried out in 96-well microtiter plates, involves a series of mixing, reaction, and washing steps, which not only are laborious but also often lead to large errors and inconsistent results. It usually takes several hours or longer to complete one assay because of the long incubation time required in each step. The long incubation time is a result of inefficient mass transport by molecular diffusion from the solution to the solid surface, although the immuno-reaction itself is a rapid process.1 Recently, microchip-based immunoassays have attracted a lot of attention for their potential advantages of having a high specific surface area, very low reagent consumption, and much reduced assay time because of the device’s microscale.1-7 Several * To whom correspondence should be addressed. Address: 140 West 19th Ave., Columbus, OH 43210. Phone: 614-292-6611. Fax: 614-2923769. E-mail: [email protected]. † The Ohio State University. ‡ Bioprocessing Innovative Company. (1) Rossier, J. S.; Girault, H. H. Lab Chip 2001, 1, 153-157. (2) Mao, H.; Yang, T.; Cremer, P. S. Anal.Chem. 2002, 74, 379-385. (3) Sato, K.; Tokeshi, M.; Odake, T.; Kimura, H.; Ooi, T.; Nakao, M.; Kitamori, T. Anal.Chem. 2000, 72, 1144-1447. (4) Sato, K.; Tokeshi, M.; Kimura, H.; Kitamori, T. Anal.Chem. 2001, 73, 1213-1218. (5) Cheng, S. B.; Skinner, C. D.; Taylor, J.; Attiya, S.; Lee, W. E.; Picelli, G.; Harrison, D. J. Anal.Chem. 2001, 73, 1472-1479. (6) Dodge, A.; Fluri, K.; Verpoorte, E.; de Rooij, N. F. Anal.Chem. 2001, 73, 3400-3409.

microfluidic devices for immunoassays and enzyme assays have been developed and tested.4,8-10 However, because the microfluidic devices have a large surface-to-volume ratio, controlling the surface properties of the microdevices becomes one of the most critical issues in developing microfluidic immunoassays and biosensors. In general, the sensitivity of the biochip is dependent on the total activity of the antibodies or enzymes attached on the surface. How the biochip surface interacts with antibodies and other biomolecules also affects the specificity of the immunoassay. Since the total surface area in a microchip is fixed, it is important to attain a high immobilization activity yield during the immunoassay. The conventional passive adsorption of antibodies onto the surface is mainly driven by hydrophobic interactions, which often cause protein denaturation and reduce protein’s functional sites or activities by even more than 90%.11-13 This problem becomes more serious when there is a large specific surface area in the microdevice. Therefore, developing an efficient surface modification method to enhance the binding efficiency and activity of target protein is critical in biochip development. To date, most of the microfluidic immunoassay systems have been fabricated using silicon, metal, and glass. As the low-cost alternative substrate materials for biochips, polymers can also offer a wide range of physical and chemical properties that afford (7) Jiang, X.; Ng, J. M.; Stroock, A. D.; Dertinger, S. K.; Whitesides, G. M. J. Am. Chem. Soc. 2003, 125, 5294-5295. (8) Eteshola, E.; Leckband, D. Sens. Actuators B 2001, 72, 129-133. (9) Hadd, A. G.; Raymond, D. E.; Halliwell, J. W.; Jacobson, S. C.; Ramsey, J. M. Anal.Chem. 1997, 69, 3407-3412. (10) Cohen, C. B.; Chin-Dixon, E.; Jeong, S.; Nikiforov, T. T. Anal. Biochem. 1999, 273, 89-97. (11) Butler, J. E.; Ni, L.; Nessler, R.; Joshi, K. S.; Suter, M.; Rosenberg, B.; Chang, J.; Brown, W. R.; Cantarero, L. A. J. Immunol. Methods 1992, 150, 77-90. (12) Butler, J. E. J. Immunol. Methods 2000, 22, 4-23. (13) Davies, J.; Dawkes, A. C.; Haymes, A. G. R., C. J.; Sunderland, R. F.; Wilkins, M. J.; Davies, M. C.; Tendler, S. J.; Jackson, D. E.; Edwards, J. C. J. Immunol. Methods 1994, 167, 263-269.

10.1021/la061123l CCC: $33.50 © 2006 American Chemical Society Published on Web 09/28/2006

Enhancing Antibody Binding

good processibility for mass production and recyclability. However, compared to well-established surface modification methods for silicon and glass substrates, the development of efficient surface treatment method for facilitating protein binding on the polymer surface is still in its infancy. Several proteinpolymer surface binding methods have been used in immunoassays, ranging from simple protein passive adsorption on polystyrene beads,3 adsorption via lipid layers grafted on poly(dimethylsiloxane) (PDMS),2,14 to adsorption via protein A bound to PDMS surface.6,8 In developing fluorescence-based biosensors, poly(methyl methacrylate) (PMMA) has many advantages because it is transparent, has a low fluorescence background, and can be easily fabricated. Recently, Lai and co-workers have demonstrated the feasibility of performing enzyme-linked immunosorbent assay on a PMMA-based CD microfluidic platform.15 However, direct adsorption of antibodies to the PMMA surface without any surface modification used in their work yielded a low binding efficiency because the antibody bound poorly onto the untreated PMMA surface. Although various surface modification methods have been developed to introduce functional amine groups on the PMMA surface,16-18 they either consisted of too many steps and yielded a low surface amine density18 or involved unstable intermediates and environmentally unfriendly solvents in their preparation.16 The main objective of this work was to develop a simple, fast, low-cost, and highly efficient surface modification method for PMMA-based microfluidic immunoassay chips. In this study, an amine-bearing polymer, poly(ethyleneimine) (PEI), was used to functionalize and to enhance antibody binding on the PMMA surface. The antibody binding kinetics in the microfluidic channels was studied. X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) were used to study and better understand the effects of PEI surface modification. The performance of an immunoassay for IgG detection using the PEI-modified microchannel was also studied and compared with that of conventional 96-well plates. Experimental Section Fabrication of PMMA Microchannel. A poly(methyl methacrylate) (PMMA) mold for microchannel (140 µm wide, 125 µm deep, and 1.5 cm long) was fabricated using a computer numerically controlled (CNC) machine. Poly(dimethylsiloxane) (PDMS) was prepared from Sylgard 184 silicon elastomer base and curing agent (Dow corning corporation, Midland, MI) at a 10:1 (w/w) ratio. After thorough mixing and degassing under vacuum for 30 min, the PDMS polymer was casted over the PMMA mold and cured in an oven at 70 °C for 1.5 to 2 h. The PDMS daughter mold was then used to produce PMMA microchannels through a microembossing process. Detailed procedures about the fabrication of PMMA mold, PDMS daughter mold, and the PMMA microfluidic chips by microembossing can be found elsewhere.15 Modifications of PMMA Microchannel. To functionalize the PMMA surface, the microchannels were treated with different NH2-bearing chemicals, including poly(ethyleneimine) (PEI; MW ) 75 000), poly(allylamine hydrochloride) (PAH; MW ) 70 000), hexamethylenediamine (HMD), and 1,3-diaminopropane (DAP), all were purchased from Sigma-Aldrich (St. Louis, MO). Surface amination with HMD or DAP followed the method by Fixe et al.17 (14) Yang, T.; Jung, S.; Mao, H.; Cremer, P. S. Anal.Chem. 2001, 73, 165169. (15) Lai, S.; Wang, S.; Luo, J.; Lee, L. J.; Yang, S. T.; Madou, M. J. Anal.Chem. 2004, 76, 1832-1837. (16) Henry, A. C.; Tutt, T. J.; Galloway, M.; Davidson, Y. Y.; McWhorter, C. S.; Soper, S. A.; McCarley, R. L. Anal.Chem. 2000, 72, 5331-5337. (17) Fixe, F.; Dufva, M.; Telleman, P.; Christensen, C. B. Nucleic Acids Res. 2004, 32, e9. (18) Bulmus, V.; Ayhan, H.; Piskin, E. Chem. Eng. J. 1997, 65, 71-76.

Langmuir, Vol. 22, No. 22, 2006 9459

Figure 1. Schematic of the microfluidic ELISA device used in this study. A. The microchannel experimental setup; B. Fluorescence images of the microchannel at different times after the initiation of the enzyme reaction in ELISA. Briefly, the PMMA plates were incubated with 10% HMD or DAP in 100 mM, pH 11.5 borate buffer for 2 h. For PEI or PAH treatment, PMMA plates were first treated in 1 N NaOH solution at 55 °C for 30 min and then immersed in a PEI or PAH solution (0.2%, pH 7) at room temperature for 1 h. The aminated PMMA plates were then placed in the glutaraldehyde solution (1% w/v) at room temperature for 30 min. The PMMA surface was thoroughly rinsed with distilled water after each treatment step. After air-blow drying, the treated microchannels were ready for antibody binding. Sandwich-Type Enzyme Linked Immunosorbent Assay (ELISA). In the ELISA, the rat IgG was the “analyte” or antigen, the polyclonal affinipure goat anti-rat IgG (H+L) was the first antibody, and horseradish peroxidase-conjugated affinipure goat antirat IgG (H+L) was the second antibody. They were purchased from Jackson Immuno Laboratories, Inc. (West Grove, PA), reconstituted in distilled water and stored at -80 °C until use. The substrate solution was prepared by dissolving 3 g/L of 3-p-hydroxyphenylpropionic acid (HPPA) (Sigma-Aldrich, St. Louis, MO) in Tris-HCl buffer (0.15 M, pH 8.5). Prior to use, 1 µL of 30% hydrogen peroxide was added to every 7.5 mL of HPPA solution and mixed thoroughly. PBW washing buffer was prepared from Dulbecco’s phosphate buffered saline, pH 7 (PBS) (Invitrogen Life Technologies, Carlsbad, CA) and contained 0.1% polyoxyethylenesorbitan monolaurate (Tween-20; Bio-Rad Laboratories, Hercules, CA) and 0.5% bovine serum albumin (BSA; Invitrogen Corporation, Grand Island, NY). The blocking buffer contained PBS, 0.1% Tween-20, 1% BSA, and 0.05% sodium azide (Sigma-Aldrich, St. Louis, MO). ELISA procedures were carried out in modified PMMA microchannels and 96-well plates (Nunc Maxisorp). ELISA in 96-Well Plates. To each well, 100 µL each of the following solution was added in sequence: first antibody (10 mg/L), blocking, antigen (0-10 mg/L), second antibody (10 mg/L), and substrate solutions. Between adding different solutions, the individual wells were washed with 200 µL of the washing solution for three times. The first antibody was incubated in the well at 4 °C overnight. The other incubation steps were done at room temperature for 3 h or at 4 °C overnight. The reaction was detected by a CytoFluor 96-well plate fluorescence reader using 360 ( 40/460 ( 40 nm as the excitation and emission filters, respectively. ELISA in Microchannels. The microchannel was first sealed with the Scotch tape (see Figure 1A). Unless otherwise noted, first antibody (10 mg/L), blocking, antigen (1 mg/L), and second antibody (10 mg/L) solutions, 1 µL each, were loaded, using a micropipet, into the microchannel, from the opening at the side of the microchip, in sequence to cover the whole channel area. Incubation was done in a humidified box at room temperature for 1 h. The microchannel was thoroughly washed with 10 µL of PBW solution for three times between each step. The Scotch tape was replaced with a new one prior to the blocking and detection steps to eliminate the undesired binding of antibodies, antigen, and substrate on the Scotch tape. Detection was carried out using an inverted fluorescence microscope (Nikon

9460 Langmuir, Vol. 22, No. 22, 2006 Eclipse TE2000-U). A 100-W mercury light source with a 335/20 nm filter and a dichroic mirror was used as the excitation source. The fluorescence emission signal was obtained through a dichroic mirror and a 405/40 nm filter. Following the reaction, the fluorescent microchannel images (see Figure 1B) were recorded with a 12-bit high-resolution monochrome digital camera system (CoolSnap HQ). The fluorescence intensity was analyzed using the Fryer Metamorph Image Analysis System. The average intensity of independent measurements from at least two different microchannels is reported in this paper. Effects of Surface Modification on Protein Binding Efficiency. First Antibody. Various amounts of the first antibody ranging from 0.1 to 10 mg/L were applied to the pristine (passive adsorption), PEI modified, and HMD modified PMMA microchannels. After 1 h incubation, the microchannels were washed and air-blow dried. The actual amounts of the first antibody bound to the microchannels were determined indirectly by completing the remaining ELISA steps described previously. The enzyme reaction rate (RFU/s) was monitored and used as an indicator of the enzyme activity, which was proportional to the amount of the first antibody bound on the microchannel surface when excessive amounts of antigen (1 mg/L) and second antibody (10 mg/L) were applied in the assay. StreptaVidin. A total of 1 µL of streptavidin solution (10 mg/L) was added to the pristine (passive adsorption), PEI modified, and HMD modified PMMA microchannels. After 1 h incubation, these microchannels were washed with 10 µL of PBW solution for three times, followed by the addition of biotin-HRP solution (20 mg/L). The time response of the enzymatic reaction of HPPA was monitored and reported in RFU/s, indicating the reaction rate or the enzyme activity, which is proportional to the amount of biotin or streptavidin bound to the surface. To study and compare the signal/noise ratio of different chemically modified surfaces, more concentrated streptavidin and biotin-HRP solutions, 200 mg/L, were utilized to increase the background noise to a nonzero level. Surface Blocking. PEI treated microchannels were treated with various blocking reagents (no blocking, 1% ethanolamine, 1% PEGNH2, 1% BSA, 1% ethanolamine + 1% BSA, and 1% PEG-NH2 + 1% BSA) to evaluate their effectiveness of blocking or preventing nonspecific protein binding. After blocking, 20 mg/L of the second antibody was applied to the microchannels, followed with washing and the enzyme reaction. The enzyme reaction was monitored and RFU/s was used to indirectly assess the amount of second antibody nonspecifically bound to the surface, indicating the effectiveness of these different blocking conditions. Adsorption/Binding Rate and Kinetics. The protein adsorption kinetics was studied by loading 1 µL of the first antibody solution (1 mg/L) into each microchannel, which was then placed face down in a Petri dish whose surface had been pretreated with the blocking solution for 1 h. The microchannels were incubated at room temperature for various periods of time. After incubation, the microchannels were washed, air-blow-dried, sealed with Scotch tape, and then followed by the blocking and other remaining ELISA steps described previously. The binding kinetics between the antigen and the first antibody on the microchannel surface was also studied. Each microchannel was loaded with 1 µL of the first antibody solution (10 mg/L) and incubated for 1 h, washed, air-blow-dried, sealed with Scotch tape, and blocked with the blocking solution for 1 h. After blocking, 1 µL of the antigen solution (1 mg/L) was added into each microchannel and incubated for various time periods, followed by completing the remaining ELISA steps described previously. X-ray Photoelectron Spectroscopy (XPS). The surface elementary compositions of the PMMA plates before and after amination with HMD or PEI were analyzed with a Krotos AXIS ultra X-ray photoelectron spectrometer. Measurements were obtained using a monochromatic Al KR X-ray source (240 W) and charge neutralization. The samples were analyzed with a 90° takeoff angle. Deconvolution of spectral peaks was performed using the Kratos software, and the spectra of C(1s) (280-300 eV binding energy), O(1s) (528-542 eV binding energy), and N(1s) (392-410 eV binding energy) were recorded.

Bai et al.

Figure 2. Time courses of the fluorescence signals from enzyme reactions in ELISA carried out in various surface-treated microchannels with various initial first antibody concentrations: 0 mg/L (9); 0.1 mg/L (b); 0.5 mg/L (2); 1 mg/L (0); 5 mg/L (O); 10 mg/L (4). A. PEI-treated PMMA microchannel; B. HMD-treated PMMA microchannel; C. untreated PMMA microchannel. Atomic Force Microscopy (AFM). A 2.5% (w/w) PMMA/toluene solution was prepared by dissolving the PMMA pellets in toluene, which was filtered through a 0.2-µm membrane filter. The PMMA solution was then spin coated onto a cleaned silicon wafer at 1500 rpm for 30 s, and the PMMA-coated silicon wafer was annealed at 150 °C in a vacuum oven for 24 h. After annealing, the PMMA surfaces were treated with PEI or HMD, followed by antibody binding. All the solutions were prepared with 18 MΩ‚cm water. The surface morphology of the modified and protein bound PMMA thin film was investigated in air, using a Nanoscope Ø AFM (Digital Instruments, Santa Barbara, CA) in the tapping mode, with a low spring constant of 0.3 N/m, and at ambient temperature.

Results and Discussion Effects of Surface Treatment on Protein Binding. Figure 2 shows the time responses of the enzymatic reaction of HPPA

Enhancing Antibody Binding

Langmuir, Vol. 22, No. 22, 2006 9461

Figure 3. Effects of the first antibody concentrations (from 0.1 to 10 mg/L) and different surface treatments on the enzyme reaction rates (RFU/s) from the ELISA carried out in the microchannels (b, PEI treated, 2, HMD treated, 9, untreated PMMA surfaces).

as monitored by the fluorescence intensity (RFU) for the PMMA microchannels with PEI, HMD, or no surface treatment. As can be seen in the figure, the enzyme reaction generally was faster when more first antibodies were added or bound to the microchannel surface in the ELISA, suggesting that more antigens and thus second antibody-enzyme conjugates could bind to the surface when there were more active first antibodies. The slope of the time-course plot indicates the initial reaction rate, which should be proportional to the enzyme activity, and thus can be used as an indirect assessment of the amount of active first antibody available on the microchannel surface for the immuno reaction. The effects of the first antibody concentration and different surface treatments on the reaction rate, expressed as RFU/s, are shown in Figure 3. It is noted that the level of RFU/s should be proportional to the actual amount of active first antibody bound to the surface when there are excessive amounts of antigen and second antibody. As can be seen from Figure 3, the reaction rate generally increased with increasing the first antibody concentration applied in the experiments but leveled off at higher concentrations. However, for different surfaces, the reaction rates were very different even though the same amount of the first antibody was applied, indicating that the binding efficiency for the first antibody on these different surfaces was quite different. Passive adsorption yielded the least “active” antibody on the surface at all initial first antibody concentrations studied. For the HMD modified PMMA surface, the fluorescence signal was comparable to that of passive adsorption at low first antibody concentrations but increased substantially as the first antibody concentration increased to 10 mg/L. For the PEI modified PMMA surface, the fluorescence signal was the highest at all initial first antibody concentrations studied. It was about 10 times higher at 1 mg/L of the first antibody concentration as compared to those from the HMD modified and unmodified surfaces. Also, the data from the PEI modified surfaces were more consistent with smaller standard errors than those from the HMD method. It is clear that PEI treatment resulted in the highest binding efficiency for the first antibody. It is also clear that the ELISA performance is strongly dependent on the amount of the first antibody bound on the surface, which greatly affects the subsequent antigenantibody binding and final enzyme reaction rate as seen in this study. Therefore, proper surface treatment to increase the binding efficiency of the first antibody on the PMMA microchannel surface is critically important to the microfluidic ELISA. The effects of different surface treatment methods were further compared with the streptavidin/biotin-HRP system, and the results are shown in Figure 4. As compared with the PMMA surfaces

Figure 4. Effects of surface treatment on streptavidin binding. A. Comparison of the reaction rates as measured by the fluorescence signals from PEI treated, HMD treated, and untreated microchannels (10 mg/L of streptavidin and 20 mg/L of biotin-HRP were applied in this experiment); B. Comparisons of the fluorescence signals and signal/noise ratio (S/N) from PEI and HMD treated microchannels (200 mg/L of streptavidin and 200 mg/L of biotin-HRP were applied in this experiment).

with HMD treatment or without any treatment (passive adsorption), the PEI treated surface also gave ∼10 times higher signal regardless of the level of the background noise, which was zero when the concentrations of streptavidin and biotin-HRP were low but increased significantly with increasing their concentrations in the assay. As shown in Figure 4B, the signal/noise ratio from the PEI treated microchannel was also higher, ∼2 times of that from the HMD treated microchannel. It is thus clear that the PEI treated surface can give the highest fluorescence signal at a faster rate and a better signal/noise ratio; both are critical to ELISA’s sensitivity and detection limit. Figure 5 shows the chemical structures of different aminebearing molecules and how they influence the interactions between the first antibody and the surfaces with different amine groups. As illustrated in Figure 5B, the aminated PMMA surface was functionalized with glutaraldehyde, which allowed the antibody molecules to covalently bind to the PMMA surface through the imine bond between the aldehyde group on the modified surface and the amine group on protein surface. It is noted that the major difference between HMD and PEI is their molecular weight or size; however, the PEI-treated microchannel gave 10 times higher ELISA signal than that from the HMD treated microchannel. The large PEI molecules not only can provide the functional amine groups but also serve as spacers, whereas the small HMD molecules can only provide the functional amine groups. The superior performance of the PEI treated surface thus can be attributed to the spacer function of PEI that allows protein molecules to stay away from the surface and thus to preserve more activity than those directly bound to the surface.

9462 Langmuir, Vol. 22, No. 22, 2006

Bai et al.

Figure 5. Schematics of antibody binding on different PMMA surfaces treated with different amine-bearing chemicals. A. Chemical structures of poly(ethyleneimine) (PEI), poly(allylamine hydrochloride) (PAH), hexamethylenediamine (HMD), and 1,3-diaminopropane (DAP). B. Steps in PMMA surface treatment and subsequent antibody binding: amination, cross-linking with glutaraldehyde (GA), and covalent binding of antibody. C. Illustration of the spatial effects of different surface modifications on antibody binding on the PMMA surface. Left: PEI or PAH coated PMMA surface provides a larger spacer length to allow the antibodies to stay away from the solid surface. Middle: HMD or DAP treated surface provides active binding sites on the PMMA surface. Right: Passive adsorption of antibodies on the PMMA surface that could result in disorientation and activity loss of the antibody upon its binding on surface.

Figure 6. Comparison of enzyme reaction rates (RFU/s) from various immuno surfaces showing the effect of different amine bearing molecules on the first antibody binding (with 1 mg/L first antibody, 1 mg/L antigen, and 10 mg/L second antibody) in the PMMA microchannels.

To further test the prostituted spacer effect, four amine-bearing molecules were investigated: two are amine-bearing polymers, PEI and poly(allylamine hydrochloride) (PAH), and two are small diamine molecules, HMD and 1,3-diaminopropane (DAP). As shown in Figure 6, the PMMA surfaces treated with the two polymers bound ∼10 times more active antibodies than those treated with the two small diamine molecules. Clearly, the amine-

bearing polymers are better than the small diamine molecules for antibody binding because the spacer function of polymers can preserve most of the biological activity of the bound protein molecules. Surface Blocking. As shown in the previous section, PEI treatment greatly increased the binding efficiency of the first antibody on the PMMA surface. However, it also would greatly increase nonspecific binding of antigen and second antibody if the surface were not properly blocked. Therefore, it was necessary to evaluate and optimize the blocking conditions for the PEItreated PMMA surface. Different blocking agents were studied for their effectiveness in preventing the nonspecific protein adsorption. Since first antibody and antigen were not added to the microchannels in the experiment, any enzyme reaction observed would be attributed to the second antibody-enzyme conjugates directly bound on the surface. As shown in Figure 7, ethanolamine and PEG-NH2 were not effective in blocking the surface, whereas BSA at 1% was an efficient blocking agent. Adding ethanolamine or PEG-NH2 to BSA solution did not improve the blocking. Binding Kinetics in the Microchannel. It is of interest to know how fast the antibody would bind to the PMMA surface in the microchannel as this will determine the incubation time for each ELISA step and the total assay time required to complete each assay. The adsorption/binding kinetics of the first antibody on various PMMA surfaces was thus studied, and the results are shown in Figure 8A. In general, the fluorescence signal increased

Enhancing Antibody Binding

Langmuir, Vol. 22, No. 22, 2006 9463

antibody on different surfaces was controlled by the same mechanism, diffusion, which has been reported as the rate-limiting step in the heterogeneous immunosorption reaction on the planar surface19 and microchannels.20 It is noted that the binding kinetics was not affected by the reagent loading as the time to load (pipet and fill) the entire microchannel was less than 1 s. Therefore, any possible effects of convection and mixing on antibody adsorption onto the microchannel surface can be neglected. The adsorption kinetics of antibody can be modeled by Fick’s second law

(

)

∂C ∂2C ∂2C )D 2 + 2 ∂t ∂x ∂y Figure 7. Comparison of the effects of different blocking agents on the nonspecific adsorption of second antibody as monitored by the enzyme reaction rate (RFU/s). E: 1% ethanolamine; P: 1% PEG-NH2; B: 1% BSA; EB: 1% ethanolamine + 1% BSA; PB: 1% PEG-NH2 + 1% BSA. 10 mg/L of the second antibody was added to test the blocking effectiveness. The higher the reaction rate, the less effective the blocking agent.

Figure 8. Kinetics of protein binding on the PMMA surfaces. A. First antibody binding kinetics on various PMMA surfaces (1 mg/L first antibody, 1 mg/L antigen, and 10 mg/L second antibody were added in sequence into the microchannels); B. Antigen binding kinetics on first antibody bound PEI-treated PMMA (10 mg/L first antibody, 1 mg/L antigen, and 10 mg/L second antibody were applied in sequence to the microchannel). Symbols show the experimental data and curves show the model simulations.

rapidly at the beginning and then reached a plateau when all antibodies had been adsorbed to the surface. It was found that the time to reach the plateau was the same (∼10 min) for all three different surfaces studied, although the PEI-treated surface gave much higher signals than those from the HMD treated and untreated surfaces. The results suggest that the adsorption of

(1)

where C stands for the concentration of the first antibody in the solution and x and y are the width and depth of the microchannel, respectively. D is the diffusion coefficient, which can be estimated using the semiempirical equation of Polson21

DAB )

9.40 × 10-15T µ(MA)1/3

(2)

The antibody studied in this work has a molecular weight of 160 000 kg/kg mol, and therefore, D was calculated to be 5.14 × 10-7 cm2 s-1. Because the length of the microchannel is much longer than the cross sectional dimension, the model only considers two dimensions. The initial condition and boundary conditions are as follows At t ) 0, C ) C0 At t > 0, C ) 0 at the three microchannel walls (x ) 0, x ) 140 µm, and y ) 0) ∂C/∂y ) 0 at the surface in contact with the blocked wall of the Petri dish (y ) 125 µm) C ) 0 at microchannel wall can be assumed since the binding reaction of the antibody on the polymer surface is much faster than its diffusion to the wall. The open channel side can be treated as an insulated surface since it was in contact with the blocked surface of the Petri dish. Equation 1 was used to calculate the antibody concentration in the solution and the amount of antibody adsorbed on the surface of the microchannel (dimensions: 140 µm × 125 µm) at any time point. The calculated results were then adjusted with a proportional constant to fit the plateau value shown in Figure 8. In general, the model simulates the data well, confirming that the adsorption of the first antibody in the microchannel was controlled by diffusion. The binding kinetics for the antigen, rat IgG, which also has a molecular weight of 160 000 kg/kg mol, in the microchannel was also studied and simulated using the same mathematical model and diffusion coefficient (5.14 × 10-7 cm2 s-1), and the results are shown in Figure 8B. Again, the model simulation is in good agreement with the experimental data. This result suggests that the reaction between the antigen and antibody is much faster than the diffusion, which is consistent with previous studies.20 Comparison of ELISA in Microchips and in 96-Well Plates. ELISA detection of the rat IgG was carried out in both 96-well plates and microchannels to compare their performance. The same concentrations of first and second antibodies were used in the study, although larger amounts (liquid volumes) were used for the 96-well plates. Figure 9A shows the ELISA assay results (19) Stenberg, M.; Nygren, H. J. Theor. Biol. 1985, 113, 589-97. (20) Rossier, J. S.; Gokulrangan, G.; Girault, H. H.; Svojanovsky, S.; Wilson, G. S. Langmuir 2000, 16, 8489-8494. (21) Geankoplis, C. J. Transport processes and unit operations; Prentice Hall P T R: Englewood Cliffs, NJ, 1993; p 405.

9464 Langmuir, Vol. 22, No. 22, 2006

Figure 9. ELISA detection of rat IgG in various concentrations (0-3 mg/L) carried out in 96-well plates (A) and PEI-treated PMMA microchannels (B).

on 96-well plates. A linear relationship between the fluorescence signal and the antigen concentration was observed from 2 to 100 ng/mL. When the antigen concentration was lower than 2 ng/mL, the signal could not be distinguished from the background noise. On the other hand, at concentrations higher than 100 ng/mL, the fluorescence signal did not increase linearly and reached a plateau at 500 ng/mL. The results for the microchannel are given in Figure 9B, which shows a dynamic linear range from 5 to 500 ng/mL. Compared with the conventional 96-well format, the microchannel provided a wider dynamic range and a similar low detection limit even though a much smaller sample (antigen) volume was used. As shown in Table 1, microchannels also provide additional advantages, including much lower reagent consumption and shorter assay time. Due to the small dimensions of the microchannel, the diffusion time for the molecules to reach the channel surfaces, which is proportional to the square of the diffusion length, is much shorter. Therefore, the incubation times required for the protein binding and enzymatic reaction are much shorter, less than one tenth, in the microchannel than in the 96-well plate. The superior performance in the microchannel was made possible because of the surface modification with the amine-bearing polymer, PEI. XPS and AFM Analyses of Modified PMMA Surfaces. To determine the presence and quantify the atomic composition of nitrogen entities, both the pristine and PEI, HMD modified PMMA surfaces were characterized by XPS, and the results are sown in Figure 10. The X-ray photoelectron survey spectra of these three different surfaces display two obvious peaks with one centered approximately at 285 eV and the other at 532 eV, which indicate the binding energy at C 1s and O 1s core levels, respectively. Also can be seen in the blow-up view of the photoelectron spectra (Figure 10B) is another peak at the binding energy around 400 eV corresponding to the nitrogen 1s core level. It is clear that both treated surfaces had a significantly

Bai et al.

increased nitrogen content, which was negligible or not present on the pristine PMMA surface. The atomic compositions of pristine PMMA and HMD, PEI treated PMMA are summarized in Table 2. Total nitrogen compositions on PEI and HMD treated surfaces are 1.59% and 0.63%, respectively. For PEI molecules, the ratio of primary, secondary, and tertiary amine groups is 1:2:1.22 Only the primary and secondary amines in PEI molecules would be active to react with glutaraldehyde to form the binding sites for the antibody. These amine groups together accounted for 1.19% total nitrogen on the PEI-treated PMMA surface, which was about twice that found on the HMD-treated PMMA surface. It is noted that all nitrogen atoms in HMD are from its primary amine group. It has been reported that the primary amine density on the HMD-treated PMMA surface was 0.28 ( 0.03 nmol NH2/cm2,17 which was in excess of the amount required for binding all antibody molecules on the surface. The average diameter of IgG antibodies, as estimated from AFM images, is between 7 and 10 nm and a densely packed antibody layer will have the surface density of 0.00432 nmol antibody/cm2.17 Hence, the amount of the surface active amine groups on the HMD-treated PMMA surface should be sufficient for binding all first antibodies applied in this study. However, the amount of active antibody bound on the PEItreated PMMA surface was ∼10 times of that on the HMDtreated surface, which cannot be explained by the mere 2-fold difference in their active amine contents. The main difference between HMD and PEI is in their molecular size. The former is a small molecule, whereas the latter is a polymer whose large molecular structure can contribute to the beneficial spacer effect that has been reported to allow the antibody to stay away from the hydrophobic polymer surface and thus avoid its denaturation upon binding on the polymer surface.23,24 It has been reported that the incorporation of a minimum spacer length was imperative for antibody binding on the solid support material and that efficient immobilization of biomolecules required longer spacers.25,26 The small HMD molecules or short spacers would not be sufficient to keep the large antibody molecules away from the surface and overcome steric hindrance from the vicinity of the solid support. To verify this spacer effect, atomic force microscopy (AFM) was used to investigate the morphology of the PMMA surfaces after different treatments. Figure 11 shows the taping mode AFM images of the three different immuno surfaces. The unmodified PMMA surface prepared by spin coating PMMA/toluene solution on a clean silicon substrate displayed a nice and flat surface (Figure 11A) with a root-mean-square (RMS) roughness value of 0.4 nm. With HMD modification, the PMMA surface displayed some small white spots (Figure 11B), and the RMS roughness value increased to 1.4 nm. For the PEI treated PMMA surface, there were many bright spots observed and their dimension ranged from several nm to more than 20 nm (Figure 11C). The RMS roughness value increased to 3.6 nm for the PEI-treated PMMA surface. Similarly, the PAH-treated surface also showed increased RMS roughness (Figure 11D). However, there was an apparent difference in the surface morphology between PEI and PAH treated surfaces. This difference can be attributed to their different molecular linearity: PEI is a highly branched polymer, whereas (22) Juang, R. S.; Ju, C. Y. Ind. Eng. Chem. Res. 1997, 36, 5403-5409. (23) Penzol, G.; Armisen, P.; Fernandez-Lafuente, R.; Rodes, L.; Guisan, J. M. Biotechnol. Bioeng. 1998, 60, 518-23. (24) Andresen, H.; Grotzinger, C.; Zarse, K.; Kreuzer, O. J.; EhrentreichForster, E.; Bier, F. F. Proteomics 2006, 6, 1376-1384. (25) Nouaimi, M.; Moschel, K.; Bisswanger, H. Enzymol. Microb. Technol. 2001, 29, 567-574. (26) Park, K. D.; Okano, T.; Nojiri, C.; Kim, S. W. J. Biomed. Mater. Res. 1988, 22, 977-992.

Enhancing Antibody Binding

Langmuir, Vol. 22, No. 22, 2006 9465

Table 1. Comparisons of Reagent Consumption and Incubation Time for ELISA Performed in 96-Well Plates and Microchannels microchipa 96-well plate

experimental conditions

amount required

procedure

amount (µL)

incubation time (min)

amount (µL)

incubation time (min)

amount (µL)

incubation time (min)

first antibody blocking igg/sample second antibody substrate total

100 100 100 100 100 500

>120 >120 >120 >120 5 >485

1 1 1 1 1 5

60 60 60 60 1 241

0.26 0.26 0.26 0.26 0.26 1.3