Plastic ELISA-on-a-Chip Based on Sequential Cross-Flow

(6) Peplow, M. O.; Correa, P. M.; Stebbins, M. E.; Jones, F.; Davies, P. Appl. Environ. Microbiol. 1999, 65, 1055r1060. (7) Chin, J.; Daniels, J.; Bun...
0 downloads 0 Views 535KB Size
Download PDF
Anal. Chem. 2006, 78, 793-800

Plastic ELISA-on-a-Chip Based on Sequential Cross-Flow Chromatography Joung-Hwan Cho,†,⊥ Seung-Mok Han,‡,⊥ Eui-Hwan Paek,§ Il-Hoon Cho,‡ and Se-Hwan Paek*,†,‡,§,|

Graduate School of Biotechnology, Korea University, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul 136-701, Korea, Program for Bio-Microsystem Technology, Korea University, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul 136-701, Korea, BioDigit Laboratories Corp., Korea University, Biotechnology Building, Technology Incubation Center, 1, 5-ka, Anam-dong, Sungbuk-ku, Seoul 136-701, Korea, and Department of Biotechnology, Korea University, Jochiwon, Choongnam 339-800, Korea

A plastic chip that can perform immunoassays using an enzyme as signal generator, i.e., ELISA-on-a-chip, was developed by incorporating an immunostrip into channels etched on the surfaces of the chip. To utilize an analytical concept of cross-flow chromatography, the chip consisted of two cross-flow channels in the horizontal and vertical directions. In the vertical channel, we placed a 2-mmwide immunostrip for cardiac troponin I (cTnI), which was identical to a conventional rapid test kit except for the utilization of an enzyme, horseradish peroxidase (HRP), as tracer. An enzyme substrate supply channel and a horizontal flow absorption pad compartment were transversely arranged on each lateral side of the signal generation pad of the strip, respectively. Upon application of a sample containing cTnI, it migrated vertically through the membrane strip by capillary action, and antigen-antibody binding occurred. After 15 min, the horizontal flow was initiated by the addition of a chromogenic substrate solution for HRP into the supply channel and by partial superimposition of the horizontal flow absorption pad onto the signal generation pad. A color signal proportional to the analyte concentration was produced on this pad, measured after 5 min as optical densities using a digital camera-based detector, and quantified by integration of the densities under the peak after normalization. Its calibration curve indicated that the detection limit of the chip was ∼0.1 ng/mL and its quantification limit was 0.25 ng/mL. In measuring blindly prepared samples, the chip performance correlated with that of a reference system, Beckman Coulter Access, within 2.5-fold discrepancy at the detection limit. Enzyme-linked immunosorbent assay (ELISA) is an analytical method that utilizes solid-phase immune reactions to detect an analyte in sample via an enzyme labeled to an immunoreagent as * Corresponding author. Phone: +82-2-3290-3438, Fax: +82-2-927-2797. E-mail: [email protected]. † Graduate School of Biotechnology. ‡ Program for Bio-Microsystem Technology. § BioDigit Laboratories Corp. | Department of Biotechnology. ⊥ These authors equally contributed to this study. 10.1021/ac051453v CCC: $33.50 Published on Web 01/06/2006

© 2006 American Chemical Society

a signal generator.1 In this type of assay, a binding reaction partner, antigen or antibody, is typically immobilized on the solid surfaces of microtiter plates consisting of multiple, small-volumecapacity wells made of plastic (e.g., polystyrene).2 Such features of the analytical system not only allowed us an easy separation of the antigen-antibody binding complexes from unbound reagents by washing the surfaces but also allowed us to simultaneously process a number of samples for either qualitative or quantitative measurements.3,4 For these reasons, since its introduction in 1971, it has been widely applied to various fields of analysis, such as medical diagnostics,5 food and environmental monitoring,6 and veterinary examination.7 Compared to other tracers, such as radioisotopes and fluorophore, the enzymes used as signal generators in ELISA are huge, proteinacious molecules, which catalyze each specific substrate.8 The catalytic action amplifies the signal, which, depending on its chemical properties, can be measured using a simple detector based on colorimetry,9 luminometry,10 and electrochemistry,11,12 for example. However, because of their huge molecular sizes, it is difficult to label them to immunoreagents without interferences in antigen-antibody bindings,13,14 which rarely occurs with the small tracers. Enzymes are, moreover, sensitive to environmental variables, including inhibitory substances that may be inadvert(1) Guilbault, G. G. Anal. Chem. 1968, 40, 459-471. (2) Cantarero, L. A.; Butler, J. E.; Osborne, J. W. Anal. Biochem. 1980, 105, 375-382. (3) Engvall, E.; Perlmann, P. Immunochemistry 1971, 8, 871-873. (4) Kasupski, G. J.; Lo, P. L.; Gobin, G.; Leers, W. D. Am. J. Clin. Pathol. 1984, 81, 230-232. (5) Heeschen, C.; Goldmann, B. U.; Langenbrink, L.; Matschuck, G.; Hamm, C. W. Clin. Chem. 1999, 45, 1789-1796. (6) Peplow, M. O.; Correa, P. M.; Stebbins, M. E.; Jones, F.; Davies, P. Appl. Environ. Microbiol. 1999, 65, 1055-1060. (7) Chin, J.; Daniels, J.; Bundesen, P. Vet. Immunol. Immunopathol. 1989, 20, 109-118. (8) Kricka, L. J. Ann Clin. Biochem. 2002, 39, 114-129. (9) Morrin, A.; Guzman, A.; Killard, A. J.; Pingarron, J. M.; Smyth, M. R. Biosens. Bioelectron. 2003, 18, 715-720. (10) Jackson, R. J.; Fujihashi, K.; Kiyono, H. J. Immunol. Methods 1996, 190, 189-197. (11) Ho, W. O.; Athey, D.; McNeil, C. J. Biosens. Bioelectron. 1995, 10, 683691. (12) Zeravik, J.; Ruzgas, T.; Franek, M. Biosens. Bioelectron. 2003, 18, 13211327. (13) Bienjarz, C.; Husain, M.; Barnes, G.; King, C. A.; Welch, C. J. Bioconjugate Chem. 1996, 7, 88-95. (14) Suzawa, T.; Ikariyama, Y.; Aizawa, M. Anal. Chem. 1994, 66, 3889-3894.

Analytical Chemistry, Vol. 78, No. 3, February 1, 2006 793

ently present in samples, and may alter their activities as catalysts.6,7 Nevertheless, such unfavorable factors, although decidedly important, have not significantly restrained their utilization as tracers, and ELISA has been a routine, a standard laboratory method for analyses of complex organic substances for the last two decades.3,12 Despite its popularity, ELISA has rarely been applied to practical analyses conducted outside of the laboratory. This is due to the presence of a repetitive addition and the removal of reagents required during the analytical procedure, even though considerable progress had been made toward automation of the ELISA procedure.15,16 For point-of-care testing (POCT), a method of immunochromatography has been developed that utilizes membrane strips as a solid matrix.17 Tracers used in this format are mostly gold colloids or Latex beads, of which colors, as a result of assays, can be detected by the naked eye.18,19 Although it can offer several advantages in POCT, such as speed and simplicity, the low sensitivity of the assay has been considered a major drawback. Alternatively, other types of signals, fluorescence and magnetic field, for example, have been explored in the efforts to develop high-detection-capability immunosensors.20,21 These sensors have been available for diagnosis of acute cardiac syndrome in the market. However, some limitations in expanding the same technologies to other conventional products are expected because of their high cost and bulky dimensions. We have investigated a POCT version of ELISA that utilizes the concept of cross-flow chromatography22 for widespread application of immunosensors to various analytes with minimal costs and, potentially, dimensions. The concept was developed to use enzymes as tracers in the immunochromatographic assay by sequentially accomplishing antigen-antibody bindings and catalytic reactions to generate signals. In this study, we have constructed an ELISA-on-a-chip (EOC), which enabled us to achieve a semiautomatic switching of the sequential processes for a complete analysis and a miniaturization of the immunosensor. This chip was fabricated by incorporating a conventional immunostrip into a plastic chip with elaborately devised channels on the surfaces. To demonstrate its utilization, the EOC has been applied to the measurement of an analyte, cardiac troponin I as a specific marker of acute myocardial infarction (AMI), with a high sensitivity. EXPERIMENTAL SECTION Materials. A stock of cardiac troponin (cTn) I-T-C complex, cTnI single molecule for immunization, and a monoclonal antibody (clone 19C7) specific to cTnI were supplied by Hytest (Turku, Finland). Human anti-mouse antibody (HAMA) blocker (mouse IgG fraction) and a cardiac marker control were obtained from (15) Johns, M. A.; Rosengarten, L. K.; Jackson, M.; Regnier, F. E. J. Chromatogr., A 1996, 743, 195-206. (16) Mine, T.; Yano, I.; Tatsumi, N.; Terano, Y.; Yamamoto, K.; Goda, T.; Kosaka, Y.; Hishida, K. Osaka City Med. J. 2000, 46, 129-144. (17) Paek, S. H.; Lee, S. H.; Cho, J. H.; Kim, Y. S. Methods 2000, 22, 53-60. (18) Ono, T.; Kawamura, M.; Arao, S.; Nariuchi, H. J. Immunol. Methods 2003, 272, 211-218. (19) Cho, J. H.; Paek, S. H. Biotechnol. Bioeng. 2001, 75, 725-732. (20) Apple, F. S.; Christenson, R. H.; Valdes, R., Jr. Clin. Chem. 1999, 45, 199205. (21) Blake, M. R.; Weimer, B. C. Appl. Environ. Microbiol. 1997, 63, 16431646. (22) Cho, J. H.; Paek, E. H.; Cho, I. H.; Paek, S. H. Anal. Chem. 2005, 77, 40914097.

794

Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

Chemicon International (Temecula, CA) and Cliniqa (Fallbrook, CA), respectively. N-Succinimidyl-3-(2-pyridyldithio) propionate (SPDP), succinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate (SMCC), and dithiothreitol (DTT) were purchased from Pierce (Rockford, IL). Goat anti-mouse antibody, casein (sodium salt type, extracted from milk), human serum (frozen liquid), Triton X-100, and Sephadex G-15 and G-100 were supplied by Sigma (St. Louis, MO). Nitrocellulose (NC) membrane (12-µm pore size) and glass fiber membrane (Ahlstrom 8980) were obtained from Millipore (Bedford, MA). Cellulose membrane (17 CHR chromatography grade) and glass fiber membrane (Rapid 24Q) were purchased from Whatman (Maidstone, England). Horseradish peroxidase (HRP) was supplied by Calbiochem (San Diego, CA), and its substrate containing insoluble 3,3′,5,5′tetramethylbenzidene (TMB) was supplied by Moss (Pasadena, MD). All other reagents used were of analytical grade. Synthesis of HRP-Labeled Antibody. (a) Production of Monoclonal Antibody. A monoclonal antibody specific to cTnI was raised through the adoption of a standard protocol.23 cTnI (30 µg) was emulsified with complete Freund’s adjuvant and injected into the peritoneal cavity of a 6-week-old Balb/c mouse. After 3 weeks, the mouse was immunized with the same amount of cTnI emulsified with incomplete Freund’s adjuvant. An identical procedure was repeated 2 weeks later, and the final immunization was conducted after the same period with cTnI dissolved in 10 mM phosphate buffer, pH 7.4, (PB) containing 140 mM NaCl (PBS). Three days after the final boosting, the mouse splenocytes were collected and fused with murine plasmacytoma (sp2/0 Ag14) as a fusion partner. Fused hybridoma cells were screened based on HAT selection,24 and a cell clone producing antibody specific to cTnI (BD Clone 12) was finally screened by immunoassay using antigen-coated microtiter plates. This antibody was produced as ascitic fluid from a Balb/c mouse and was then purified on a protein G column (5 mL, HiTrap protein G HP; Amersham Biosciences, Piscataway, NJ). The eluted IgG fractions were pooled, concentrated, dialyzed against PBS, and frozen as aliquots until later use. (b) Conjugation between Antibody and HRP. The monoclonal antibody (BD clone 12) was chemically coupled with HRP using cross-linkers as described in a previous paper.22 In brief, the antibody (total 1 mg, 0.5 mL) and HRP (total 1.4 mg, 0.5 mL) dissolved in 100 mM PB containing 5 mM ethylenediaminetetraacetic acid disodium salt were coupled with SMCC and SPDP dissolved in dimethyl sulfoxide, respectively. The coupled SPDP linker was activated using DTT, and both modified proteins were fractionated by means of Sephadex G-15 gel chromatography. The antibody was then immediately combined with the HRP in a 5 molar excess and reacted overnight at 4 °C. This mixture was purified on a Sephadex G-100 gel column (10 × 200 mm). The purified conjugates were quantified by the Bradford method25 and stored as aliquots after snap freezing. Construction of ELISA-on-a-Chip. (a) Preparation of Immunostrip. To accomplish the immunochromatographic assay for cTnI in the vertical direction, different functional membrane (23) Gearhart, P. J.; Sigal, N. H.; Klinman, N. R. Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 1707-1711. (24) Jantscheff, P.; Winkler, L.; Karawajew, L.; Kaiser, G.; Bottger, V.; Micheel, B. J. Immunol. Methods 1993, 163, 91-97. (25) Duhamel, R. C. Colloids Relat. Res. 1983, 3, 195-204.

pads have been employed. Each sample application pad was a glass fiber membrane (2 × 15 mm; Ahlstrom 8980) pretreated with poly(vinyl alcohol) by the manufacturer. A conjugate release pad was fabricated by transferring 8 µL of a conjugate solution onto a glass membrane (2 × 5 mm; Rapid 24Q). The conjugate solution was prepared by diluting the HRP-labeled antibody (2.5 µg/mL) with 100 mM PB containing 0.5% casein (casein-PB), HAMA blocker (150 µg/mL), ascorbic acid (5 mM), Triton X-100 (0.5%, v/v), and trehalose (20%, w/v). A signal generation pad was made by dispensing (1.5 µL/cm) the monoclonal antibody (clone 19C7; 2 mg/mL) in PBS onto a site at 10 mm from the bottom of the NC membrane (2 × 25 mm) using a microdispenser (BioJet 3000, Biodot, Irvine, CA). On the same membrane, goat antimouse antibody (0.2 mg/mL) in PBS was also dispensed onto a site at 17 mm from the bottom. After drying at 37 °C for 1 h, the membrane was kept in a desiccator at room temperature until use. The prepared membrane pads were arranged to be a width of 2 mm, in order from the bottom, sample application pad, conjugate release pad, signal generation pad, and a cellulose membrane (2 × 15 mm) as an absorption pad. Finally, a functional immunostrip was constructed by partially superimposing each contiguous membrane strip and fixing them on a plastic film using doublesided tape. (b) Etching of Plastic Chip. Fluidic channels were made by mechanically engraving the surfaces of a polyacrylamide chip (32 × 76 × 2 mm), essentially enabling us to comprise the immunostrip in the vertical position as a part of fluidic channels and to deliver an aqueous solution crosswise. An immunostrip mounting channel was arranged in the center of the chip by carving the surface to a width of 2 mm, a length of 51 mm, and variable depths, adapting the different thicknesses of each membrane pad of the strip. The bottom of the channel was drilled in an oval shape (5 × 10 mm) to provide a sample application pot with a maximum sample holding capacity of 100 µL. A signal monitoring window was furnished by slitting the chip surface (1 × 18 mm) corresponding to the ceiling of the signal generation pad of the strip. To allow for a flow across this pad, an enzyme substrate supply channel and an absorption pad compartment were installed on each opposing side of the vertical channel. On one side, a substrate supply channel with a depth of 0.8 mm was formed in a shape of a circular triangle expanded to the vertical channel. A substrate supply pot (7-mm diameter) was installed by drilling the surface at an inlet of the channel. Two air ventilation holes (1-mm diameter) were also made at both of the end projection areas near the outlet of the channel. On the other side of the vertical channel, an absorption pad compartment for the flow was built to specific dimensions: a width of 14 mm, length of 12 mm, and a depth of 1 mm. (c) Assembly of EOC. The etched plastic chip was integrated with the immunostrip and a horizontal flow absorption pad by installing them into the vertical channel and the absorption pad compartment, respectively. The absorption pad was prepared by attaching the cellulose membrane (14 × 12 mm) to a plastic film using a double-sided tape. The integrated chip was closed by covering with a laminating film and then bonding an intact plastic chip of the same size using double-sided tape. The chip was finally kept in a desiccator maintained at room temperature until use.

Optimization of Analytical Conditions. (a) Preparation of Standard Samples of cTnI. A stock of cTnI (1 mg/mL; I-T-C complex form) was serially diluted with human serum to prepare samples at predetermined concentrations. The serum itself was determined to be negative for the analyte from the measurement using a clinical analyzer, Beckman Coulter Access, according to the guide of the manufacturer. (b) Immunostrip Width. The effect of the immunostrip width on the sensitivity of analysis was tested without employing the plastic chip as described in the previous paper.22 In brief, each of the standard solutions of cTnI were placed into different microwells, and the immunostrips of selected widths were placed into their respective microwells in an erect position in order to absorb the aqueous solutions into the strips. After a 15-min reaction, horizontally arranged pads22 were positioned to the right and left sides of the signal generation pad, respectively, and the substrate solution for HRP was supplied to allow horizontal flow for 5 min. The signal appeared at the area of the immobilized antibody, and the image was captured using a scanner (HP ScanJet 4670, Hewlett-Packard, Palo Alto, CA). The same procedure was repeated with the strips of different widths. (c) Background Staining. In utilizing the EOC for cTnI analysis, the background staining on the signal generation pad was measured with or without covering the surfaces of the pad with a lamination tape. For each case, the standard solution of cTnI (0.6 ng/mL; total 40 µL) was applied into the sample application pot of the EOC. After 15-min reaction, the substrate solution for HRP (90 µL) was added into the substrate supply pot and the horizontal absorption pad was immediately connected to the signal generation pad of the immunostrip. The visible color intensity of the enzyme signal was confirmed by observation with the naked eye; it was also measured 5 min after the initiation of the substrate supply. The EOC with colored signals was placed under a digital camera (FA185A#ABA, Hewlett-Packard) built within a detector and illuminated from the bottom using a light source (SR0307A-5230, Seho Robot, South Korea). The image of the signal generation pad was captured, and the color densities that appeared on the pad were digitized in the vertical direction using software programmed in C++ language, installed on a personal computer. The data were collected and stored in the Microsoft Excel program. Three repetitions of the same procedure were conducted for each case. (d) Analysis Time. To determine the optimal analysis time of the EOC, the reaction time for antigen-antibody bindings was varied, while the signal generation time was constant, and was maintained at 5 min. The standard samples of cTnI (total 40 µL) were applied into the application pots of distinct EOC, and the vertical flow continued for variable times, from 5 to 15 min. Other analytical conditions not mentioned here were identical to those described above. The color intensity produced on the signal generation pad was measured as optical density against the vertical distance from the top. To quantify the signal proportional to the analyte dose, the measured optical densities were first subtracted from the mean value of the background colors present between the signal and control peaks. The normalized optical densities under the signal peak were then integrated so that a numerical signal value could be assigned. Finally, the signal values corresponding to each analyte dose were plotted for the respective immune reaction time. Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

795

Figure 1. Effect of reduction of the immunostrip width on the sensitivity of an analytical system based on cross-flow chromatography. A shortening of the path of horizontal flow in the assay may alter the flow speed, varying the washing efficiency of residual tracer, as well as the rate of enzymatic conversion, although the vertical flow of a sample (e.g., human serum) remained identical in the assay conditions. This resulted in a change in signal-to-noise ratio on the strip, which indicated a necessity to select a NC membrane as the signal generation pad with an optimal pore size for a given strip width.

Characterization of Analytical Performances (a) Calibration. Under optimal conditions, the responses of the EOC to the analyte concentrations were obtained using the standard samples of cTnI. The samples were added into different EOC, the immune reactions were processed for 15 min, and sequentially, the signal generation was processed for 5 min after the enzyme substrate was supplied. The color image on the signal generation pad was captured within the detector equipped with a digital camera, and the optical densities were measured. The integrated values of signal and control were determined as explained above. The same procedure was repeated three times, and the mean values at each concentration were used to plot a graph of the dose-response curve. For an accurate calibration of the EOC, the curve was linearized by log-logit transformation,26,27 and a regression line was obtained using the least-squares method.28 (b) Correlation Tests. To assess the EOC as a biosensor, it was used for analyzing blindly prepared samples, and the analytical results were compared with those from a clinically accepted analyzer, Beckman Coulter Access, for the same samples. The samples were prepared using cTnI materials supplied by different commercial sources, Hytest and Cliniqa. Analyses using the EOC were performed as described above and by employing the reference system according to the protocol provided by the manufacturer. The analytical data for the identical samples were plotted in the abscissa and the ordinate axes, respectively, to determine their correlations. RESULTS AND DISCUSSION Effect of Scaling-Down. In rapid analytical devices based on chromatography using the lateral flow of aqueous medium, although membrane pads can be cut smaller than 4 mm in width, it would be difficult to hold different pieces of the pads in a precise arrangement. This would cause low analysis reproducibility and inaccuracy in detection. Alternatively, if the small pieces of (26) DeLean, A.; Munson, P.; Rodbard, D. Am. J. Phys. 1978, 235, 97-102. (27) Rockville, M. D. Draft Guidance for Industry and FDA Reviewers. The FDA web site, 12 January 2005. (28) Ott, L. 4th ed.; Wadsworth Publishing: Florence, KY, 1988; pp 870-891.

796

Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

membranes were installed within the channels of a plastic chip, they could be precisely arranged and assembled together for the fabrication of a functional EOC. Such a novel approach would make it possible to use membranes as parts of a complete channel for performing ELISA at any location where specimens were furnished, typically, under nonlaboratory conditions. Since reducing the width of the immunostrip was advantageous, we tested the effect of scaling-down the device on the analytical performances, in particular, the sensitivity of detection (e.g., signal-to-noise ratio as a measure). As the width was reduced while other conditions were kept constant, the analytical results using each strip showed that both signal and background staining gradually decreased (Figure 1). This trend could be the result of an increase in the horizontal flow rate causing a harsh washing of residual tracer components after the completion of the vertical flow. Under the flow conditions of enzyme substrate, the turnover rate of catalysis also seemed to be varied, due to a disturbance in the formation in enzyme-substrate complex. To maintain a maximum signal-to-noise ratio at a determined pad width, we have selected a NC membrane with an appropriate pore size as the signal generation pad. Construction of EOC Sensor System (a) Analytical Plastic Chip. To fabricate a plastic chip installed with membrane pads for ELISA, we have devised fluidic channels by mechanically etching the plastic surfaces. The chip consisted of two distinct flow channels in the vertical and horizontal directions (Figure 2A). The vertical compartment was carved to tightly fit a 2-mm-wide immunostrip, essentially the same as that of a conventional rapid test kit,29,30 except it also used an enzyme tracer (e.g., HRP). A sample application pot and a signal monitoring window were installed by drilling through the chip surfaces. To induce a subsequent horizontal flow, an enzyme substrate supply channel and an absorption pad compartment were horizontally arranged on each lateral side of the signal generation (29) Schwartz, J. G.; Gage, C. L.; Farley, N. J.; Prihoda, T. J. Am. J. Emergency Med. 1997, 15, 303-307. (30) Christenson, R. H.; Fitzgerald, R. L.; Ochs, L.; Rozenberg, M.; Frankel, W. L.; Herold, D. A.; Duh, S. H.; Alonsozana, G. L.; Jacobs, E. Clin. Biochem. 1997, 30, 27-33.

Figure 2. Construction of an analytical chip for ELISA adapting the concept of cross-flow chromatography, i.e., EOC. The chip was fabricated by engraving fluidic channels on the surfaces of a plastic (A) that not only hold the immunostrip (B) used for carrying out antigen-antibody bindings but also provide a place to supply an enzyme substrate solution and to connect the horizontal flow absorption pad (B) at the time of signal generation. Such features of the chip enabled us to realize a total analysis required to conduct diagnosis of a disease (e.g., acute myocardial infarction).

Figure 3. Analytical procedure using the EOC as shown in Figure 2. A sample containing cTnI spiked in a human serum was applied into the sample application pot and the flow was initiated in the vertical direction by capillary action (A). After the completion of antigen-antibody bindings, the substrate solution for enzyme used as tracer (e.g., HRP) was added into the supply pot and the horizontal flow absorption pad was manually connected to the lateral side of the signal generation pad (B). As a consequence, a color signal was produced in proportion to the dose of the analyte, permitting us to accomplish ELISA at the point of care out of laboratory.

portion of the strip, respectively. In the substrate supply channel, a supply pot and two air ventilation holes were located at the inlet and near the outlet, respectively. Two membrane components, the immunostrip and the horizontal flow absorption pad, were prepared for installation into the chip (Figure 2B). The immunostrip was composed of four different, commercially available membranes, furnishing various

functions of sample application, enzyme conjugate release, signal generation, and vertical flow absorption.29,30 They were lengthily disposed in order, partially superimposed on one another, and mounted on a plastic film. This strip was fixed in the vertical channel of the chip. The position of the horizontal flow absorption pad, on the other hand, was variable. If used for analysis, it was placed in a position spatially separate from the immunostrip at Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

797

Figure 4. Effect of lamination on the top of the signal generation pad on the background noise. Covering the surfaces of the pad led the substrate solution through the microporous structure, resulting in a thorough washing of the unbound enzyme tracer and, thus, a uniform, low background (B) relative to that with an uncovered pad (A). Triplicate data from each run were presented.

the beginning, and after the completion of the vertical flow, it was slid onto the lateral side of the signal generation pad to initiate the horizontal flow of an added substrate solution. The channels with such installed membrane components were closed by bonding a flat plastic chip to fabricate a functional EOC that can be used for quantifying cTnI in samples. (b) Analytical Procedure. Using the plastic chip, we performed the cross-flow chromatographic analysis for cTnI from Hytest. The analyte was spiked in a human serum to prepare a standard solution, which was then transferred into the sample application pot of the chip. It was migrated in the vertical direction by capillary action (Figure 3A) and dissolved the detection antibody labeled with HRP, which triggered bindings between this enzyme conjugate and the analyte molecules in the liquid phase. Such binding complexes were carried into the signal generation pad, where the immobilized capture antibody bound them to form a sandwich type of complex. At the time of a complete removal of the excess components, a solution containing a chromogenic substrate for HRP (e.g., insoluble TMB) was supplied into the corresponding pot, and at the same time, the horizontal flow absorption pad was connected to the lateral side of the signal generation pad (Figure 3B). Upon initiation of the flow of substrate, a color analyte signal at the site of the immobilized antibody was produced in proportion to the analyte concentration. A control was also run to monitor the consistency of the assay using a secondary antibody, recognizing the detection antibody, immobilized at a site on the signal generation (see the color signal and control in Figure 3B). (c) Colorimetric Detection. To quantify the color signal, we built a detector based on image capture using a digital camera. After analysis, the chip with colored signals was placed under the camera, and the color densities that appeared on the signal generation pad were digitized in the vertical direction using a software program developed in our group. The data were collected and stored in the Microsoft Excel program installed on a personal computer. Optimization of Analytical Conditions. (a) Background Control. We have revealed in a previous report that the crossflow immunochromatographic analysis using HRP as tracer was 798 Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

Figure 5. Determination of an optimal process time of immune reactions toward the detection limit of cTnI. Under constant conditions for signal generation, the detection capability of the chip was improved in proportion to the time period for antigen-antibody bindings within the selected range. Under the consideration of limited total assay time, the vertical flow was kept for 15 min after which no significant enhancement was found.

30 times more sensitive than a conventional rapid assay employing colloidal gold.22 To reproduce an equal performance on the EOC, we have optimized the analytical conditions, more specifically, toward the horizontal flow for supplying enzyme substrate. Since a thin immunostrip was used, the substrate added in the supply channel tended to flood over the signal generation pad surfaces, which resulted in a high, variable background, probably due to the superficial washing of residual tracers (Figure 4A). Installation of the slit of the signal monitoring window (see Figure 2A) alone was not sufficient to prevent this problem. Further, the surfaces of the signal generation pad were laminated with a cohesive film. This treatment blocked the bypass of the horizontal flow and led the flow through the interstitial spaces of the NC membrane, eventually showing a low, uniform background (Figure 4B). (b) Total Analysis Time. We have optimized the analysis time toward sensitivity under given conditions of reagent concentrations and membrane treatments that were similar to those described in the previous paper.22 The cross-flow chromatographic analysis

Figure 6. Calibration curve of the EOC cartridge and signal detector system for cTnI (left) and a linearized version of the curve by log-logit transformation (right). The analyte signal and control were quantified by integration of the color densities under the respective peaks and then plotted against the analyte concentration (left). The signal was varied in a sigmoidal shape while the control was constant regardless of the analyte concentration. Each standard deviation of replicate measurements is indicated. The sigmoidal signal curve was then linearized to use it for the determination of unknown samples (right). In this plot, a regression equation and correlation coefficient are shown.

was carried out by the two-step protocol, i.e., antigen-antibody bindings and catalytic signal generation. With the selected enzyme label and chromogenic substrate pair, an optimal time for signal generation was 5 min, after which a significant improvement in color intensity was not observed. Under such conditions, the immune reaction time was varied to determine a minimal period offering the lowest detection limit of the analyte (Figure 5). The analyte signal was measured by integrating the normalized optical densities, obtained by subtracting the mean value of the background, under the peak along the vertical position of the signal generation pad, and plotted against the analyte concentration. To detect cTnI at levels below 0.1 ng/mL, a 15-min period of immune reactions was minimal. The reaction time can be shortened by either improving the quality of the enzyme conjugates, e.g., use of a defined enzyme conjugate, or investigating a mass enzyme transport system. Both of these topics are the focus of ongoing research in this laboratory. One of the goals here is to carry out the immune reaction within 10 min without sacrificing the sensitivity. Calibration of the EOC Sensor System. The EOC sensor system has been calibrated for utilization in the quantitative detection of cTnI. Since the future aim was to determine the amount of analyte present in the bloodstream of a potential patient, human serum was used as the medium for spiking a stock to each standard concentration. This protocol not only closely reflected the clinical conditions of the AMI diagnosis but also provided high reproducibility of the analysis. The cTnI materials used for calibration are available from various commercial sources, which, however, may reveal different binding patterns to antibody pairs used in an assay. Indeed, there have been reports that measurements of the analyte were highly variable according to analytical systems employed.31-35 Such lack of standardization has been determined to mainly result from the molecular nature of complex formation (e.g., troponin I-C and I-T-C), proteolytic cleavage after release, and chemical modifications.36,37 (31) Wu, A. H. B. J. Clin. Ligand Assay. 1999, 22, 32-37. (32) Koerbin, G.; Tate, J. R.; Potter, J. M.; Hickman, P. E. Ann. Clin. Biochem. 2005, 42, 19-23. (33) Altinier, S.; Mion, M.; Cappelletti, C.; Zaninotto, M. Clin. Chem. 2000, 46, 991-993. (34) Apple, F. S.; Koplen, B.; Murakami, M. M. Clin. Chem. 2000, 46, 572574. (35) Pagani, F.; Bonetti, G.; Panteghini, M. Clin. Chem. 1999, 45 (Suppl), A144.

A dose-response curve of the sensor was first obtained using standard samples of cTnI, a product of Hytest among commercially available, and was plotted in a semilog graph (Figure 6, left). The cTnI product selected as a calibrator was a form of the cTnI-T-C complex that, according to the data of the manufacturer, corresponded to the real form of the molecule in the serum of AMI patients and was stable after dilution in human serum. As mentioned previously, the signal and control were measured by integrating the optical densities under the respective peaks after normalization. The signal varied in a sigmoidal shape, while the control was kept approximately constant regardless of the dose of analyte. For an accurate calibration, the sigmoidal curve was then converted to a straight line by means of the log-logit transformation (Figure 6, right).26,27 A linear regression equation that highly correlated with the data (correlation coefficient 0.9964) was derived using the least-squares method,28 which can be used for quantifying the analyte in unknown samples. From the calibration curves, the detection limit of the EOC sensor was determined as the analyte concentration corresponding to the analyte signal value, which had been calculated by multiplying the standard deviation of the signal at the zero dose by three and found to be ∼0.1 ng/mL. The quantification limit was also determined using the same procedure except the assurance of 2-fold higher deviation margin, i.e., multiplying the deviation value by six, and found to be 0.25 ng/mL when the Hytest cTnI was used as a calibrator. Comparison of Analytical Performances. The novel EOC sensor system was finally used for measuring concentrations of cTnI in blindly prepared samples, and the analytical results were compared with those determined using a reference system, Beckman Coulter Access, which has been commercially established as a clinical laboratory instrument (Figure 7). The samples were prepared by spiking the materials of cTnI obtained from different sources, Hytest and Cliniqa, with human serum. To test the degrees of correlation, the concentrations measured using the EOC sensor were plotted against those from the reference system. As expected, using the Hytest cTnI, the two groups of data seemed (36) Labugger, R.; Organ, L.; Collier, C.; Atar, D.; Van Eyk, J. E. Circulation 2000, 102, 1221-1226. (37) Katrukha, A. G.; Bereznikova, A. V.; Filatov, V. L.; Esakova, T. V.; kolosova, O. V.; Pettersson, K.; Lovgren, T.; Bulargina, T. V.; Trifonov, I. R.; Gratsiansky, N. A.; Pulkki, K.; Voipio-Pulkki, L. M.; Gusev, N. B. Clin. Chem. 1998, 44, 2433-2440.

Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

799

Figure 7. Correlation of the EOC sensor system with a commercially established system, Beckmann Coulter Access. Samples were blindly prepared to contain cTnI from various sources. The same samples were used to quantitatively determine employing the two analytical systems at the same time, of which results showed an entirely agreeable coincidence. The relative accuracy of the EOC sensor to the reference system is shown in the inset and is expressed as percent overestimation. This deviation increased up to 2.5-fold at a diminished concentration of the analyte in the selected range.

to be well matched, with a slope close to unity, because it was employed as the calibrator of the sensor. The same pattern was also obtained with the Cliniqa product. To evaluate accuracy, we compared the two measured values for each sample and calculated the relative overestimation of the EOC sensor to plot it against the analyte concentration (Figure 7, inset). For both of the cTnI materials, the deviation increased up to 2.5-fold in the selected dose range, 0.05-50 ng/mL, as the cTnI concentration decreased. Such errors may primarily result from the different binding characteristics of antibody pairs to cTnI in the respective systems.31-34 (38) Paek, S. H.; Cho, J. H.; Kim, S. K. U.S. Patent Application No. 10/827,884.

800

Analytical Chemistry, Vol. 78, No. 3, February 1, 2006

In conclusion, a POCT version of ELISA based on the concept of cross-flow immunochromatography was developed by incorporating an immunostrip that employed HRP as tracer into a part of the fluidic channels of a plastic chip, which allowed us to semiautomatically switch the sequential flows from the vertical direction to the horizontal direction. Under optimal conditions, the EOC sensor was able to detect a minimum 0.1 ng/mL cTnI as a model analyte, and the performance was highly correlated with that of a reference system, widely accepted as clinical equipment, in a dynamic range of 0.05-50 ng/mL. Such performances were comparable to those of conventional biosensor systems utilizing fluorophore as a signal generator.20 In addition, the EOC sensor offers an extra advantage, potential production of different types of signals (e.g., photometric, chemiluminometric, and electrochemical),38 depending on the enzyme-substrate pair employed, that can be measured by relatively simple detectors. Moreover, since the novel system adopted a 2-mm-wide immunostrip, it required a minimum 50% smaller sample size than did the conventional sensors. We intend to continue this study to further enhance the detection capability of the sensor or to shorten the analytical time by introducing defined enzyme conjugates. The signal detector is also being supported by supplementing actuators to the detector for the complete automation of the sequential crossflow procedures. ACKNOWLEDGMENT This research was supported by Grant 01-PJ1-PG4-01PT02-0009 from the Ministry of Health & Welfare.

Received for review August 12, 2005. Accepted December 7, 2005. AC051453V