On-Demand Patterning of Protein Matrixes Inside a Microfluidic Device

Anal. Chem. 2006, 78, 5469-5473

On-Demand Patterning of Protein Matrixes Inside a Microfluidic Device Hirokazu Kaji, Masahiko Hashimoto, and Matsuhiko Nishizawa*

Department of Bioengineering and Robotics, Graduate School of Engineering, Tohoku University, Sendai 980-8579, Japan

On-demand immobilization of proteins at specific locations in a microfluidic device would advance many types of bioassays. We describe a strategy to create a patterned surface within a microfluidic channel by electrochemical means, which enables site-specific immobilization of protein matrixes and cells under physiological conditions, even after the device is fully assembled. By locally generating hypobromous acid at a microelectrode in the microchannel, the heparin-coated channel surface rapidly switches from antibiofouling to protein-adhering. Since this transformation allows compartmentalizing of multiple types of antibodies into distinct regions throughout the single microchannel, simultaneous assay of two kinds of complementary proteins was possible. This patterning procedure can be applied to conventional microfluidic devices since it requires only some electrodes and a voltage source (1.7 V, DC). Microfluidic systems have been widely used in bioassays because the microspace in a microfluidic channel allows highthroughput analysis with reduced sample volumes.1-3 There have been many efforts to control the fluid flow in a microchannel.4,5 On the other hand, patterned biomolecules within a microfluidic channel are well-suited to many types of bioassays.6-11 On-demand immobilization of proteins at specific locations in a microfluidic device would advance many types of bioassays, since most proteins are unstable to harsh solvents, desiccation, oxidation, and heat.12-14 However, this goal has been elusive because of the limited applicability of conventional surface patterning techniques to miniaturized and semiclosed systems, such as the packaged * To whom correspondence should be addressed. E-mail: nishizawa@ biomems.mech.tohoku.ac.jp. (1) Bange, A.; Halsall, H. B.; Heineman, W. R. Biosens. Bioelectron. 2005, 20, 2488-2503. (2) Mitchell, P. Nat. Biotechnol. 2001, 19, 717-718. (3) Meldrum, D. R.; Holl, M. R. Science 2002, 297, 1197-1198. (4) Beebe, D. J.; Mensing, G. A.; Walker, G. M. Annu. Rev. Biomed. Eng. 2002, 4, 261-286. (5) Sia, S. K.; Whitesides, G. M. Electrophoresis 2003, 24, 3563-3576. (6) Zhan, W.; Seong, G. H.; Crooks, R. M. Anal. Chem. 2002, 74, 4647-4652. (7) Koh, W. G.; Itle, L. J.; Pishko, M. V. Anal. Chem. 2003, 75, 5783-5789. (8) Kaji, H.; Nishizawa, M.; Matsue, T. Lab Chip 2003, 3, 208-211. (9) Khademhosseini, A.; Suh, K. Y.; Jon, S.; Eng, G.; Yeh, J.; Chen, G.-J.; Langer, R. Anal. Chem. 2004, 76, 3675-3681. (10) Chen, H.-Y.; Lahann, J. Anal. Chem. 2005, 77, 6909-6914. (11) Su, J.; Bringer, M. R.; Ismagilov, R. F.; Mrksich, M. J. Am. Chem. Soc. 2005, 127, 7280-7281. (12) Holden, M. A.; Cremer, P. S. J. Am. Chem. Soc. 2003, 125, 8074-8075. (13) Doh, J.; Irvine, D. J. J. Am. Chem. Soc. 2004, 126, 9170-9171. (14) Chilkoti, A.; Hubbell, J. A. MRS Bull. 2005, 30, 175-179. 10.1021/ac060304p CCC: $33.50 Published on Web 06/29/2006

© 2006 American Chemical Society

microfluidic devices. The fluid-based patterning,4,5 which is readily applicable to pattern biomolecules within a microfluidic channel, is limited to generating patterns in the flow geometry. Only a few studies based on photochemistry15,16 or thermally responsive polymer17 have been recently reported to pattern proteins in preassembled devices. Herein, we report a simple method to create a patterned surface within a microchannel by electrochemical means that enables site-specific immobilization of protein matrixes after the device is fully assembled. We prepared a microfluidic device with a microelectrode array at the upper wall of the channel. Scheme 1 shows the essentials of the procedure: (1) introduce a blocking agent (polyethyleneimine (PEI) and heparin) through the microchannel to make the inner wall antibiofouling; (2) generate HBrO local to the microelectrode, which removes a portion of blocking agent, thus making this part of the channel bottom bio-adhesive; and (3) introduce the materials of interest into the microchannel. Since this transformation allows compartmentalizing of multiple types of antibodies into distinct regions throughout the single microchannel, simultaneous assay of two kinds of complementary proteins was possible. The present strategy is based on an approach that is similar to our recent work of micropatterning a cell culture on a glass plate by using HBrO generated at a needle-type microelectrode.18-20 For the cellular patterning, we first prepared a self-assembled monolayer of long-chained alkylsilanes on the substrate surface and then immobilized bovine serum albumin hydrophobically on the substrate as a blocking agent prior to patterning. In contrast, in this study, we have newly adopted the electrostatic assembly of a PEI/heparin multilayer that can be easily formed within a microchannel after assembly and have found that the electrodebased patterning is applicable to the heparin-coated surface. EXPERIMENTAL SECTION Materials. Polyethyleneimine (PEI, average Mw 600, Wako), heparin (sodium salt, Wako), protein A (from Staphylococcus (15) Holden, M. A.; Jung, S.-Y.; Cremer, P. S. Anal. Chem. 2004, 76, 18381843. (16) Balakirev, M. Y.; Porte, S.; Vernaz-Gris, M.; Berger, M.; Arie´, J.; Fouque´, B.; Chatelain, F. Anal. Chem. 2005, 77, 5474-5479. (17) Huber, D. L.; Manginell, R. P.; Samara, M. A.; Kim, B.; Bunker, B. C. Science 2003, 301, 352-354. (18) Kaji, H.; Kanada, M.; Oyamatsu, D.; Matsue, T.; Nishizawa, M. Langmuir 2004, 20, 16-19. (19) Kaji, H.; Tsukidate, K.; Matsue, T.; Nishizawa, M. J. Am. Chem. Soc. 2004, 126, 15026-15027. (20) Kaji, H.; Tsukidate, K.; Hashimoto, M.; Matsue, T.; Nishizawa M. Langmuir 2005, 21, 6966-6969.

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006 5469

Scheme 1. Methodology for Immobilization of Proteins within a Sealed Microchannel

Figure 1. (A) Picture of the microfluidic device. (B) Schematic representation of the side and top views of the device.

aureus, Wako), fibronectin (from human plasma, Wako), bovine serum albumin (BSA, Wako), mouse IgG (Zymed Laboratories), human IgG (Oriental yeast), C3 protein (human, Calibiochem), C4 protein (human, Quidel), mouse monoclonal antibody against C3/C3b (Hycult Biotechnology), mouse monoclonal anti-C4 antibody (Antibody Shop), goat polyclonal anti-C3 antibody (Calibiochem), goat polyclonal anti-C4 antibody (Calibiochem), Cy2 reactive dye (Ab Cy2 labeling kit, Amersham Biosciences), Cy3 reactive dye (Ab Cy3 labeling kit, Amersham Biosciences), 2-methacryloyloxyethyl phosphorylcholine (MPC) polymer (NOF Corporation), and all other chemicals were used as received. Microfluidic Device Fabrication. The microfluidic device consists of three layers (Figure 1): a microelectrode-bearing substrate, silicon rubber, and a glass slide. A Pt microelectrode array (electrode width, 10 µm) and a counter electrode were fabricated on a glass slide by a series of photolithography-based microfabrication technologies. This electrode-bearing substrate was previously coated with 0.5 wt % MPC polymer/ethanol to prevent nonspecific adsorption of proteins. The silicon rubber has a stenciled pattern that forms the channel when sandwiched between the electrode substrate and the glass slide. The inlet for the reservoir and the outlet for aspiration via a syringe pump were provided at either end of the channel. The potential of the electrode array was referred to the Ag/AgCl reference electrode placed at the inlet reservoir. The length of the tube connecting the inlet and the device was made as short as possible to minimize potential drops between the reference electrode and the working electrode. Protein Patterning inside Microchannels. Solutions of PEI (5 mg mL-1) and heparin (2 mg mL-1) were passed through the microchannel, each in turn for 20 min, to coat the inner wall. Subsequently, 0.1 M phosphate buffered saline (PBS) containing 25 mM KBr was introduced into the channel, and a potential pulse of 1.7 V vs Ag/AgCl was applied to the microelectrode array. Then, 5470 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

a solution of Cy3-labeled protein A (protein A-Cy3, 25 µg mL-1) was added, followed by a wash with PBS. Sandwich Immunoassay. A 1 µg mL-1 solution of primary antibody/PBS, a 5 mg mL-1 solution of bovine serum albumin (BSA)/PBS, an appropriate amount of antigen/PBS, and a 25 µg mL-1 solution of fluorescence-labeled secondary antibody/PBS were introduced into the protein A-patterned microchannel in turn and incubated for 20 min at each step. Between steps, PBS was run through the channel for at least 1 min as a rinse. The secondary antibody was conjugated with the Cy dye in accordance with the instructions from Amersham Biosciences. After these procedures, the microchannel was imaged using an ordinary inverted fluorescence microscope. The obtained images were analyzed using Photoshop software (Adobe). Modeling and Simulations. A diffusion model of the oxidizing agent electrogenerated at a microelectrode has been previously described.20 Briefly, the finite element model of the diffusion layer formed at the electrode surface was built using the FEMLAB software package (version 3.1, Comsol) on a PC platform. This model assumes that there is an irreversible two-electron reaction in which 1 mol of HBrO can be directly produced from 2 mol of Br-, and the amount of HBrO generated at the electrode increases in proportion to the duration of electrolysis. We set no condition for the lifetime of the HBrO. To solve the diffusion equations, we used a reasonable coefficient of 1.0 × 10-5 cm2 s-1. The crosssectional dimensions of the domain, including the electrode and the microchannel, were nearly equal to the values in the experimental setup. The entire domain was meshed using triangular finite elements, and all diffusional simulations were conducted with a time step of 0.1 s. As boundary conditions, we ensured that the electrode surface concentration of HBrO during the electrolysis remained high, being independent of the rate of diffusion, and that no mass transfer occurred at any boundary except for the electrode surface.

Figure 3. (A) 2D FEMLAB simulations for the expansion of the diffusion layer of the oxidizing agent from the electrode in the microchannel. (B) Concentration profile of the oxidizing agent at the bottom wall of the channel when the electrolysis period was 10 s. Arrows in the graph depict the edge of the actual protein-patterned area shown in the insert image.

Figure 2. (A) Microelectrode array fabricated at the upper wall of the microchannel. (B) Fluorescence image of the protein A-Cy3 patterned on the bottom wall of the channel. (C) Fluorescence intensity along the white line shown in (B). (D) Plots of the pattern width of protein-immobilized area versus the square root of the electrolysis period of bromide ion. Values represent the mean ( SD of at least three samples.

RESULTS AND DISCUSSION Figure 2 shows the locally immobilized protein within a microfluidic channel prepared according to the strategy shown in Scheme 1. We fabricated a microfluidic device that has the Pt microelectrode array (electrode width, 10 µm) at the upper wall of the channel (Figures 1B and 2A). After the device was fully assembled, solutions of PEI (5 mg mL-1) and heparin (2 mg mL-1) were run through the channel, in turn, to form an electrostatically assembled multilayer with PEI and heparin.21 Heparin and other polysaccharides are known to resist the adhesion of many proteins and cells.21-26 Subsequently, PBS containing 25 mM KBr (pH 7.4) was introduced into the channel, and after the flow was completely (21) Tan, Q.; Ji, J.; Barbosa, M. A.; Fonseca, C.; Shen, J. Biomaterials 2003, 24, 4699-4705. (22) Keuren, J. F. W.; Wielders, S. J. H.; Willems, G. M.; Morra, M.; Cahalan, L.; Cahalan, P.; Lindhout, T. Biomaterials 2003, 24, 1917-1924. (23) Videm, V. Biomaterials 2004, 25, 43-51. (24) Hileman, R. E.; Fromm, J. R.; Weiler, J. M.; Linhardt, R. J. Bioessays 1998, 20, 156-167. (25) Pavesio, A.; Renier, D.; Cassinelli, C. Mora, M. Med. Device Technol. 1997, 8, 24-27. (26) Frazie, R. A.; Matthijs, G.; Davies, M. C.; Roberts, C. J.; Schacht, E.; Tendler, S. J. Biomaterials 2000, 21, 957-966.

stopped, a potential pulse of 1.7 V vs Ag/AgCl was applied to each microelectrode to generate Br2 (subsequently HBrO). By the oxidative action of HBrO, the PEI/heparin multilayer is locally detached from the substrate (see Supporting Information for XPS analysis), allowing the adsorption of proteins. The upper wall of the microchannel coated with the MPC polymer was unaffected by the electrogenerated oxidant. Then, a solution of Cy3-labeled protein A (protein A-Cy3, 25 µg mL-1) was run through the channel, followed by a wash with PBS. As can be clearly seen in Figure 2B, the protein A-Cy3 was locally immobilized on the bottom wall of the channel, corresponding to the pattern of the electrode array at the upper wall (Figure 2A). Although the heparin layer would be unable to completely prevent nonspecific adsorption, the ratio of the fluorescence signal to the background for this protein array was ∼2.5 (Figure 2C). It is worth noting that the fluorescence profile has a sharp edge, although the oxidant generated at the electrode diffuses across the channel and reacts with the polymer layer on the opposite face of the channel. The size of the protein-immobilized area was controlled by the electrolysis period and the channel height. As shown in Figure 2D, the pattern width was proportional to the square root of the electrolysis period at each channel height, indicating that the surface reaction in which the electrogenerated oxidant changes the surface property is fast enough to satisfy the diffusion controlled case. Figure 3A displays the simulated diffusion layer of the electrogenerated oxidant, expanding from the electrode at the upper wall of the channel as the duration of electrolysis increases; the geometry of the layer was taken to be isotropic and centered on the electrode. To estimate the concentration threshold at which the oxidant reacts with the PEI/heparin multilayer, the width of the actual patterned area was fitted to the simulated concentration Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5471

Figure 4. Patterning of multiple types of proteins within a single microchannel using stepwise electrochemical treatment. Fluorescence images show the fluorescence-labeled antibodies anchored to the protein-A patterned area on the floor of the channel. After the first antibody (mouse IgG-Cy3) was patterned (A), the electrochemical treatment was conducted, followed by the immobilization of the second antibody (human IgG-Cy2) (B). Then, the third antibody (mouse IgG-Cy3) was patterned by repeating the process (C).

profile of the oxidant at the bottom wall of the channel shown in Figure 3B. Arrows located at the edge of the protein pattern indicate that the concentration threshold would be on the order of a few tens of picomoles, which is in agreement with out past report for the cellular patterning on a BSA-preadsorbed substrate.20 Figure 4 shows a single microchannel in which multiple types of protein arrays have been immobilized by stepwise patterning. After protein A was site-specifically immobilized on the channel floor using the methodology described above, a solution of the first antibody, Cy3-labeled mouse IgG (mouse IgG-Cy3, 1 µg mL-1), was introduced into the channel, followed by washing with PBS (Figure 4A). Protein A is known to specifically bind the Fc region of IgG and, thus, enables the immobilization of the antibodies with a uniform orientation so as to expose the antigenbiding sites to the liquid phase.27 Subsequently, a solution of BSA (5 mg mL-1) was run through the channel to prevent contamination of the prepatterned protein region with postimmobilized protein. In fact, the BSA treatment satisfactorily reduced the crosscontamination (see Supporting Information). The above-mentioned process was repeated to immobilize the second type antibody, Cy2labeled human IgG (human IgG-Cy2) (Figure 4B), and then repeated a third time for the third type of antibody, mouse IgGCy3 (Figure 4C). Figure 5 demonstrates the simultaneous detection in the microchannel of two kinds of complementary proteins by the antibody sandwich method. The primary antibodies against C3 and C4 (anti-C3 and anti-C4, respectively) were immobilized within the microchannel according to the procedure shown in Figure 4. (27) Kasai, S.; Yokota, A.; Zhou, H.; Nishizawa, M.; Niwa, K.; Onouchi, T.; Matsue, T. Anal. Chem. 2000, 72, 5761-5765.

5472 Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

Figure 5. Simultaneous detection of complementary proteins, C3 and C4. (A) Schematic representation of sandwich immunoassay with multiple types of antibody arrays patterned inside a microchannel. The primary antibodies (anti-C3 and anti-C4) were patterned according to the procedure shown in Figure 4. A solution containing antigens (both C3 and C4) was passed through the channel, and then the fluorescence-labeled secondary antibodies (anti-C3-Cy2 and antiC4-Cy3) were introduced for simultaneous detection. (B) Plots of the fluorescence signals in the sandwich immunoassay. Green and red bars represent the signals obtained from the microchannel in which both anti-C3 and anti-C4 were immobilized. Black bars are the signals obtained from the separate detection of C3 and C4. Fluorescence of backgrounds (nonpatterned regions) was subtracted from each specific signal to exclude the effect of nonspecific adsorption. Values represent the mean ( SD of at least four experiments.

A single solution of a mixture of C3 and C4 was passed through the channel, followed by treatment with fluorescence-labeled antibodies (anti-C3-Cy2 and anti-C4-Cy3) (Figure 5A). Fluorescence measurements show that the signal intensities obtained from the simultaneous detection of C3 and C4 (green and red bars in Figure 5B) are in close agreement with those obtained from their separate detection (black bars in Figure 5B). The results indicate the cross-contamination between the two types of primary antibodies as well as nonspecific adsorption of antigens is satisfactorily low. In this approach, the entire protocol from surface patterning to measurement can be carried out under a mild, aqueous environment within a few hours, suppressing loss of protein activity. The clinical reference values of C3 and C4 (650-1350 ng mL-1 and 130-350 ng mL-1, respectively) indicate that the present system is sensitive enough for diluted samples in practical diagnoses. We describe a technique to spatiotemporally generate proteinadsorptive regions within the microfluidic channel after the device is fully fabricated and enclosed. Even multiple types of proteins can be site-specifically immobilized in a single channel without subjecting proteins to denaturing conditions, such as desiccation, oxidation, and heat. The immobilization of cell-binding proteins leads to the in-channel micropatterning of living cells (see Figure S4 in the Supporting Information). Since the procedure requires only some electrodes and a voltage source (1.7 V, DC), it is readily applicable to a conventional microfluidic device. Some recent

electrochemical techniques28,29 can control protein binding on electrodes and would be attractive for an in-channel bioassay. These modifications use electrodes as scaffolds for protein immobilization. On the other hand, the approach presented here creates a protein-adhering region near the electrode (on the opposite face of the microchannel). This configuration would allow electrochemical bioassays using the electrode as a measuring probe27,30,31 and, therefore, is beneficial for simplification and miniaturization of the overall assay system. (28) Corgier, B. P.; Marquette, C. A.; Blum, L. J. J. Am. Chem. Soc. 2005, 127, 18328-18332. (29) Yousaf, M. N.; Mrksich, M. J. Am. Chem. Soc. 1999, 121, 4286-4287. (30) Mauzeroll, J.; Bard, A. J. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 78627867. (31) Liu, B.; Rotenberg, S. A.; Mirkin, M. V. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9855-9860.

ACKNOWLEDGMENT This study was supported by Industrial Technology Research Grant Program in 05 from NEDO of Japan, and by a Grant-in-Aid for Scientific Research B (No. 17310080) from the Ministry of Education, Science and Culture, Japan. The authors also thank Prof. T. Shoji and Dr. Y. Takeda (Tohoku University) for technical assistance and valuable discussions on XPS analysis. SUPPORTING INFORMATION AVAILABLE Experimental details and additional data (PDF). This material is available free of charge via the Internet at http://pubs.acs.org. Received for review February 18, 2006. Accepted May 26, 2006. AC060304P

Analytical Chemistry, Vol. 78, No. 15, August 1, 2006

5473