Homogeneous Immunoassay for Detection of TNT and Its

A homogeneous immunoassay for TNT and its analogues is developed using a .... Stock solutions in ethanol of 2,4-dinitrotoluene (DNT) 97% (Aldrich, ...
0 downloads 0 Views 132KB Size
Download PDF
Anal. Chem. 2003, 75, 1188-1195

Homogeneous Immunoassay for Detection of TNT and Its Analogues on a Microfabricated Capillary Electrophoresis Chip Avraham Bromberg† and Richard A. Mathies*

Department of Chemistry, University of California, Berkeley, California 94720

A homogeneous immunoassay for TNT and its analogues is developed using a microfabricated capillary electrophoresis chip. The assay is based on the rapid electrophoretic separation of an equilibrated mixture of an antiTNT antibody, fluorescein-labeled TNT, and unlabeled TNT or its analogue. The band intensities of the free fluorescein-labeled TNT and of the antibody-antigen complex reveal the relative equilibrated concentrations. Titration of the anti-TNT antibody with a fluoresceinlabeled TNT derivative yields a binding constant of (3.9 ( 1.3) × 109 M-1. The dissociation rate constant of the complex is determined by kinetic capillary electrophoresis using a folded channel and a rotary scanner to interrogate the separation at multiple time points. The dissociation rate constant is found to be 0.035 ( 0.005 s-1, and the resulting binding rate constant is (1.4 ( 0.7) × 107 M-1 s-1. Binding constants of TNT and five of its analogues are determined by competitive assays: TNT (4.3 ( 2.6) × 108 M-1; 1,3,5-trinitrobenzene (5.1 ( 3.3) × 107 M-1; picric acid (7.5 ( 4.4) × 106 M-1; 2,4-dinitrotoluene (7.9 ( 4.0) × 106 M-1; 1,3-dinitrobenzene (1.0 ( 0.7) × 106 M-1; and 2,4-dinitrophenol (5.1 ( 3.0) × 104 M-1. TNT and its analogues can be assayed with high sensitivity (LOD 1 ng/mL) and with a wide dynamic range (1-300 ng/mL) using this chip-based method. Since the presentation of the first microfabricated analytical devices1-4 it has been shown that microtechnology can be successfully applied to a wide variety of analytical problems. The micro-total analysis systems (µ-TAS) concept5-11 stimulated the development of a variety of miniature devices that integrate * Corresponding author. Phone: (510) 642-4192. Fax: (510) 642-3599. E-mail: [email protected]. † Permanent address: Department of Physical Chemistry, IIBR, P.O. Box 19, Ness-Ziona 74100, Israel. (1) Terry, S. C., Jerman, J. H., Angell, J. B. IEEE Trans. Electron. Devices 1979, ED-26, 1880-1886. (2) Manz, A.; Miyahara, Y.; Miura, J.; Watanabe, Y.; Miyagi, H.; Sato, K. Sens. Actuators 1990, B1, 249-255. (3) Manz, A.; Harrison, D. J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Ludi, H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253-258. (4) Harrison, D. J.; Manz, A.; Fan, Z. H.; Ludi, H.; Widmer, H. M. Anal. Chem. 1992, 64, 1926-1932. (5) Colyer, C. L.; Tang, T.; Chiem, N.; Harrison, D. J. Electrophoresis 1997, 18, 1733-1741. (6) Kamholz, A. E.; Weigl, B. H.; Finlayson, B. A.; Yager, P. Anal. Chem. 1999, 71, 5340-5347. (7) Dolnik, V.; Liu, S. R.; Jovanovich, S. Electrophoresis 2000, 21, 41-54. (8) Bruin, G. J. M. Electrophoresis 2000, 21, 3931-3951.

1188 Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

elements of chemical analysis to perform fast assays with minimal consumption of reagents and solvents. In our laboratory, microfabricated capillary array electrophoresis systems have been developed and applied, mainly, for high-throughput DNA analysis.12-17 These devices perform analyses roughly 10-fold faster than conventional capillary electrophoresis (CE) and facilitate integrated sample preparation and manipulation.18,19 In principle, such microfluidic arrays should also be applicable to highthroughput electrophoretic immunoassays, but little work has been directed to this end. Chip-based CE is useful in performing immunoassays because the separation is faster than the relaxation time of the components of an equilibrium mixture of antigen (Ag), antibody (Ab), and the formed complex (Ag-Ab). Tagging the Ab or Ag makes it possible to monitor the equilibrium and to assay the studied analyte.20-23 Electrophoretic immunoassays have been carried out with microfabricated systems for serum cortisol,24 for BSA,25,26 for theophylline,26,27 and for serum thyroxine,28 demonstrating the advantage of microcapillary devices over conventional capillaries with respect (9) Hatch, A.; Kamholz, A. E.; Hawkins, K. R.; Munson, M. S.; Schilling, E. A.; Weigl, B. H.; Yager, P. Nat. Biotechnol. 2001, 19, 461-465. (10) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623-2636. (11) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (12) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 1134811352. (13) Simpson, P. C.; Roach, D.; Woolley, A. T.; Thorsen, T.; Johnston, R.; Sensabaugh, G. F.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 2256-2261. (14) Shi, Y. N.; Simpson, P. C.; Scherer, J. R.; Wexler, D.; Skibola, C.; Smith, M. T.; Mathies, R. A. Anal. Chem. 1999, 71, 5354-5361. (15) Liu, S. R.; Shi, Y. N.; Ja, W. W.; Mathies, R. A. Anal. Chem. 1999, 71, 566573. (16) Paegel, B. M.; Emrich, C. A.; Wedemayer, G. J.; Scherer, J. R.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 574-579. (17) Emrich, C. A.; Tian, H.; Medintz, I. L.; Mathies, R. A. Anal. Chem. 2002, 74, 5076-5083. (18) Lagally, E. T.; Medintz, I.; Mathies, R. A. Anal. Chem. 2001, 73, 565-570. (19) Paegel, B. M., Yeung, S. H. I., Mathies, R. A. Anal. Chem. 2002, 74, 50925098. (20) Schmalzing, D.; Nashabeh, W. Electrophoresis 1997, 18, 2184-2193. (21) Bao, J. J. J. Chromatogr., B 1997, 699, 463-480. (22) Heegaard, N.; Nilsson, S.; Guzman, N. J. Chromatogr., B 1998, 715, 2954. (23) Schmalzing, D.; Buonocore, S.; Piggee, C. Electrophoresis 2000, 21, 39193930. (24) Koutny, L. B.; Schmalzing, D.; Taylor, T. A.; Fuchs, M. Anal. Chem. 1996, 68, 18-22. (25) Harrison, D. J.; Fluri, K.; Chiem, N.; Tang, T.; Fan, Z. H. Sens. Actuators, B 1996, 33, 105-109. (26) Chiem, N.; Harrison, D. J. Anal. Chem. 1997, 69, 373-378. (27) Chiem, N. H.; Harrison, D. J. Clin. Chem. 1998, 44, 591-598. 10.1021/ac020599g CCC: $25.00

© 2003 American Chemical Society Published on Web 01/24/2003

to the analysis time and consumption of reagents. Furthermore, such miniature immunoassay devices may be operated in a multichannel mode, as demonstrated by the use of a six-channel chip to simultaneously monitor ovalbumin and anti-estradiol immunoassays.29 There is a large and growing demand in food, health care, and environmental monitoring for inexpensive and sensitive immunoanalytical devices that are reliable, rapid, and capable of highthroughput screening at low cost per assay. For many applications, such devices have to analyze a large number of analytes in parallel to detect, for instance, hormones, tumor markers, pathogens, and metabolic markers in medicine or residues of hazardous materials in the environment. There is also interest in the development of field-portable sensitive analyzers to detect and to monitor explosive contaminated soils,30 groundwater around military bases,31 munitions storage areas, production plants, and other sites. Similar instrumentation might be used in forensic applications to identify residues in soils and on surfaces exposed to explosions32 or to gunshots.33 Several studies have introduced34-37 the possibility of developing a microfabricated capillary chip apparatus to analyze explosives. In one of these, a mixture of 14 explosives was separated by micellar electrokinetic chromatography on a chip and detected by indirect laser-induced fluorescence with a limit of detection (LOD) of 1 µg/mL.37 An electrochemical detector combined with an electrophoresis channel has also been suggested to detect various explosives.34-36 By using a square-wave voltammetry detector, a mixture of four explosives was detected with a LOD of ∼10 µg/mL,36 while by using a screen-printed thick-film electrode a mixture of five explosives was detected with a LOD of 600 ng/mL for 2,4,6-trinitrotoluene (TNT).35 A significant improvement of the detection sensitivity (to 24 ng/mL TNT) was obtained by using a gold working electrode that was formed by electroless deposition onto the capillary outlet.34 Finally, by using fluorescence-based array biosensors, where a patterned array of antibodies specific to TNT was immobilized onto the surface of a planar waveguide, a more sensitive detection of TNT with a LOD of 1 ng/mL with dynamic range of 1-7 ng/mL was obtained.38 The above-mentioned studies have limited applications because of low sensitivity or low dynamic range. Environmental explosive assays require high throughput, high sensitivity, and a wide dynamic range. Combining fluoroimmunoassays with microfabricated capillary array electrophoresis devices that we have developed may result in such capabilities. (28) Schmalzing, D.; Koutny, L. B.; Taylor, T. A.; Nashabeh, W.; Fuchs, M. J. Chromatogr., B 1997, 697, 175-180. (29) Cheng, S. B.; Skinner, C. D.; Taylor, J.; Attiya, S.; Lee, W. E.; Picelli, G.; Harrison, D. J. Anal. Chem. 2001, 73, 1472-1479. (30) Kleibohmer, W.; Cammann, K.; Robert, J.; Mussenbrock, E. J. Chromatogr. 1993, 638, 349-356. (31) Shriver-Lake, L. C.; Breslin, K. A.; Charles, P. T.; Conrad, D. W.; Golden, J. P.; Ligler, F. Ss Anal. Chem. 1995, 67, 2431-2435. (32) Hamels, S., De Bisschop, H. C. Biomed. Chromatogr. 1998, 12, 107-108. (33) Northrop, D. M.; Martire, D. E.; Maccrehan, W. A. Anal. Chem. 1991, 63, 1038-1042. (34) Hilmi, A.; Luong, J. H. T. Anal. Chem. 2000, 72, 4677-4682. (35) Wang, J.; Tian, B. M.; Sahlin, E. Anal. Chem. 1999, 71, 5436-5440. (36) Wang, J.; Polsky, R.; Tian, B. M.; Chatrathi, M. P. Anal. Chem. 2000, 72, 5285-5289. (37) Wallenborg, S. R.; Bailey, C. G. Anal. Chem. 2000, 72, 1872-1878. (38) Sapsford, K. E.; Charles, P. T.; Patterson, C. H.; Ligler, F. S. Anal. Chem. 2002, 74, 1061-1068.

As a first step in the development of immunoassays on microcapillary array devices, we decided to study the detection of TNT and some of its derivatives in order to elaborate the characteristics and limitations of chip-based immunological methods. Here we present immunoassay studies of the reaction between anti-TNT Ab and six poly(nitrobenzenes) on a microcapillary chip. It is shown that binding equilibrium constants and dissociation rate constants can be determined from a combination of equilibrium and kinetic assays. The dissociation constant assays are accomplished with a novel folded channel design that permits interrogation of the same separation at multiple points using a rotary scanner. A sensitive TNT assay having a wide dynamic range was achieved. EXPERIMENTAL SECTION Materials. Monoclonal anti-TNT antibody (Strategic Biosolutions; Newark, DE), 10.9 mg/mL; 5-((aminopentyl)thiouredyl)fluorescein, dihydrobromide salt (fluorescein cadaverine) (FC; Molecular Probes, Eugene, OR); and picrylsulfonic acid solution 5% (w/v) (TNBS; Sigma, St. Louis, MO) were obtained from the indicated suppliers. The following certified solutions (1 mg/mL) were purchased from Restek (Bellefonte PA): picric acid (2,4,6trinitrophenol (TNP) in methanol; TNT in ethanol; 1,3,5-trinitrobenzene (TNB) in ethanol. Stock solutions in ethanol of 2,4dinitrotoluene (DNT) 97% (Aldrich, Milwaukee, WI), 10 mg/mL; 2,4-dinitrophenol (DNP) 99.7% (Supelco, Bellefonte, PA), 20.7 mg/ mL; and 1,3-dinitrobenzene (DNB) 99% (Fluka, Milwaukee, WI), 15.4 mg/mL were prepared and stored in a refrigerator. Phosphate buffers were prepared from Sigma reagents as follows: for immunoassay incubations, 10 mM phosphate, pH 7.4, that contained 10% ethanol and 0.02% Tween 20 (Aldrich); for electrophoresis experiments 10 mM phosphate, pH 7.6; for preparation of 1 µM Ab stock solution, 15 mM phosphate that contained 0.15 M NaCl and 0.02% Tween 20, pH 7.2. Fluorescein-Labeled Trinitrobenzene (TNB-Fl). The TNB moiety was labeled by coupling FC to TNBS according to previous procedures.39 FC solution was prepared by dissolving 6.4 mg of FC in 5 mL of borate buffer (0.17 M Borax and 0.12 M NaCl, pH 8.7), and TNBS solution was prepared by mixing 0.05 mL of 5% TNBS with 0.45 mL of the same borate buffer. The TNBS solution was then added dropwise to the FC solution. The mixed reagents were stored at 4 °C overnight. The product solution was titrated to pH 7.2 with 0.1 N HCl and stored at -20 °C. Before a series of experiments, 100 µL of the crude TNB-Fl was purified by HPLC (Varian model 330), using 0.1 M triethylamine acetate with an acetonitrile gradient over a reversed-phase C18 analytical column. The concentration of the purified TNB-Fl was determined spectroscopically based on  ) 80 000 at 494 nm. Purified aliquots having a concentration of 13 µM (5.3 µL) were stored at -20 °C and defrosted just before use. Stability of TNB-Fl. The purified fluorescein-labeled trinitrobenzene was found to be unstable in the buffer solutions employed. About 40% of the initial electrophoretic signals are decomposed after 40-50 min at room temperature. Significantly improved stability was observed in the presence of 10% ethanol (39) Whelan, J. P., Kusterbeck, A. W., Wemhoff, G. A., Bredehorst, R., Ligler, F. S. Anal. Chem. 1993, 65, 3561-3565.

Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

1189

Table 1. Chip Operating Voltage Parameters

Figure 1. Layout of microfabricated capillaries employed for immunoassays. Left: Straight capillary of 8-cm separation length with a 500-µm offset twin T injector and a narrow 70-µm neck at the detection point. Right: Serpentine capillary with 22-cm separation length. The path of the rotary scanning objective is presented with the six detection points marked.

in the phosphate buffers. Under these conditions, only ∼10% of the labeled TNB is decomposed after ∼1 h at room temperature. Furthermore, for samples kept in ice, only ∼5% decomposition was observed after 80 min and only ∼10% after 7 h. For these reasons, purified TNB-Fl was immediately frozen and samples were kept on ice, except for the immunological incubation and electrophoresis processes. Microcapillary Wafer Fabrication. The microcapillary electrophoresis wafer was fabricated at the UC Berkeley Microfabrication Laboratory according to previously published procedures.13 Their schematic design is depicted in Figure 1. Two kinds of microcapillaries were utilized; one device is based on a straight separation capillary for analytical measurements, while the serpentine channel is used for kinetic measurements. The capillaries were fabricated on a 10 cm diameter × 1.1 mm glass wafer.13 Once produced, 1.5-mm-diameter access holes were drilled for the sample, cathode, waste, and anode reservoirs. The drilled plate was thermally bonded to a clean cover plate of the same dimensions. The depth of the channels in both devices was 27 µm. The length of the straight capillary is 8 cm, and the length of the serpentine capillary is 22 cm. A 500-µm double-T injector was used in the straight capillary with 4-mm arms connecting the sample and the waste reservoirs to the separation channel. The distance from the cathode reservoir to the first cross of the injector is 6 mm, and the distance of the second cross from the detection point is 55 mm. At the detection point, the channel was narrowed from 120 to 70 µm to increase the analyte velocity through the focused detection laser beam. This increased the signals by ∼20% by decreasing photodecomposition.40 The width of all channels in the serpentine capillary device is 100 µm, and the cross injector had 4-mm connecting arms. Immunoassay Procedure. Sample preparation and immunoassay incubations were performed off chip. Samples were prepared in phosphate buffer, pH 7.4, and were of 1-mL volume. During preparation, the sample vials were kept in ice, and after (40) Mathies, R. A., Peck, K., Stryer, L. Anal. Chem. 1990, 62, 1786-1791.

1190 Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

capillary

time (s)

cathode (kV)

waste (kV)

anode (kV)

sample (kV)

straight serpentine

30 10

Injection Mode -0.73 -1.3 -0.3 -0.9

-0.5 -0.5

0.0 0.0

straight serpentine

120 160

Run Mode Grd -0.75 Grd -0.5

-5.0 -13.0

-0.7 -0.5

all the reagents were well mixed, the vials were transferred to a 23 °C water bath for 45-min incubation and then quenched on ice. Fifteen minutes before filling the sample well of the electrophoresis chip, the vials were warmed to room temperature (23 °C). The binding constant of TNB-Fl was evaluated from results obtained from mixtures of 10 nM Ab with concentrations of TNBFl in range of 3-20 nM. Binding constants of the unlabeled nitrocompounds were determined by competitive immunoassays from mixtures of Ab (10 nM), TNB-Fl (5 nM), and various concentrations of the nitrocompound competitors: TNT, 0.5-100 ng/mL; TNB, 1-1000 ng/mL; TNP, 10-1000 ng/mL; DNT, 402000 ng/mL; DNB, 0.3-15 µg/mL; DNP, 1-150 µg/mL. The kinetic dissociation experiments were performed with a mixture of 100 nM Ab and 20 nM TNB-Fl. All samples contained fluorescein in the 0.5-2 nM range for monitoring the electrophoresis process. The binding and the competition experiments were performed on the straight channel, while kinetic dissociation experiments employed the serpentine capillary. Electrophoresis Conditions. Microcapillaries were activated with a 1.5-h incubation with 1 N NaOH followed by rinsing with deionized water and pH 7.6 phosphate buffer, successively. The anode, the cathode, and the waste wells were filled with the same buffer, and the sample well was filled with the studied sample. Injection of a sample plug was done according to the parameters given in Table 1. Capillaries were washed with NaOH, water, and pH 7.6 buffer between experiments, while before every injection; washing was done with buffer only. Rinsing the capillaries with NaOH between injections produced no significant difference compared to rinsing with buffer alone. To increase the volume of the buffer and the sample in the reservoirs on the chip, a 2.5-mm-thick elastomer layer prepared from Sylgard 184 (Dow Corning, Midland, MI) was placed on the surface of the chip. The 40-µL elastomer reservoirs were electrically isolated one from another due to the good seal between the elastomer and the glass surface. Detection and Data Acquisition. The excitation and fluorescence detection systems employed have been described earlier.14,41 The CE separations on the straight capillary were detected using a system with a fixed objective. The fixed objective system is based on argon ion laser excitation (488 nm, 1.7 mW) and confocal detection with a 20 × 0.5 NA achromatic objective (Rolyn Optics, Covina, CA). The objective focused the beam into the constricted part of the separation channel at the detection point indicated in Figure 1. The collected fluorescence was passed (41) Kheterpal, I.; Scherer, J. R.; Clark, S. M.; Radhakrishnan, A.; Ju, J.; Ginther, C. L.; Sensabaugh, G. F.; Mathies, R. A. Electrophoresis 1996, 17, 18521859.

Figure 2. Electropherograms of mixtures of 10 nM Ab, 0.5 nM fluorescein, and TNB-Fl at the indicated concentrations in the range 3-9 nM.

Figure 3. Normalized fluorescence signal of the complex formed between Ab and TNB-Fl (corrected for dissociation) and of the free TNB-Fl as a function the total TNB-Fl added. [Ab] ) 10 nM.

through a dichroic beam splitter and the band-pass filter HQ535/ 60m (Chroma Technology Corp., Brattlboro, VT), focused onto a 200-µm pinhole, and detected by a Hamamatsu photomultiplier. The resulting signal was amplified and filtered using a Stanford Research Systems SR640 (Sunnyvale, CA) and digitized at 4 Hz. A program written in LabVIEW (National Instruments, Austin, TX) and run on a Macintosh Performa 6400 computer controlled the data acquisition and the voltage switching in the CE system. The four-color rotary scanner has been described in detail elsewhere.14 Briefly, an argon laser beam (488 nm 30-175 mW) was focused into the separation channel through a rotating 60 × 0.7 NA objective (Nikon). The emission gathered by the objective is monitored by a four-color detection system based on Hamamatsu photomultipliers. The optimal focus of the scanning objective is achieved by monitoring the water Raman scattering in the 580nm channel. The fluorescein fluorescence was monitored at 520 nm through a 520/30M band-pass filter (Omega Optical). The laser beam was scanned at 5 Hz, and the data were acquired at 5 kHz. The emission signals were amplified and filtered using two SR640s and digitized with a 16-bit A/D converter.

observed: the first peak with the migration time of 20-21 s is due to the Ab-TNB-Fl complex. The second peak with the migration time of 29-31 s is the free TNB-Fl, and the third peak with migration time of 37-41 s is the internal fluorescein standard. The migration times decrease as the concentration of TNB-FL increases; this is probably a result of adsorption of the reagents to the capillary walls, which would change the charge of the glass surface. The original migration times were restored upon flushing the channels with 1 N NaOH. As the concentration of TNB-Fl is increased, the intensity of the complex increases up to an asymptotic value. At the beginning of the titration, weak signals due to the free TNB-Fl are observed; however, they increase significantly after the Ab binding sites are saturated. To develop an accurate evaluation it was necessary to calibrate the relative response of the species. The formed complex is relatively unstable with ∼50% dissociation during the 20-s electrophoretic migration to the detection point (see below). The released TNB-Fl is evidenced by the elevated background between the signals of the complex and the free TNB-Fl especially at elevated TNB-Fl concentrations. From the kinetic study of the dissociation process, which will be described later, the complex signals were corrected according to the evaluated exponential decay parameter koff ) 0.035 s-1. The corrected intensity of the complex signal and the intensity of the observed free TNB-Fl signal were normalized to the observed intensity of the fluorescein signal. Figure 3 presents a typical titration plot of the normalized corrected intensity of the complex signal and normalized intensity of the free TNB-Fl signal as a function of the applied total TNBFl concentration in the full range explored (0-20 nM). This plot reveals that the sensitivity or magnitude of the signal from free TNB-Fl and that from the Ab-TNB-Fl complex are somewhat different and a calibration is required. As expected, the free signal increases very little below 3 nM because all of the added ligand is bound. On the other hand, at concentrations above 15 nM the Ab reached saturation and all added labeled antigen is found as the free form. The response of the complex relative to fluorescein

RESULTS Ab Binding with TNB-Fl. Immunoassay methods are strongly governed by the affinity of the antibody for the antigen. The binding constant of this interaction may be determined by titration of an Ab with labeled antigen. Chip-based electrophoresis then enables the rapid separation of the formed antibody-antigen complex from the labeled antigen and consequently the quantitation of the equilibrium ratio between the formed complex and the labeled antigen. This approach has been used to determine the binding constants of six poly(nitrobenzenes) to the anti-TNT Ab. Figure 2 presents selected electropherograms of several mixtures containing a fixed concentration (10 nM) of anti-TNT Ab and fluorescein (0.5 nM), while the concentration of TNB-Fl is varied in the range of 3-9 nM. Three separated peaks were

Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

1191

Figure 4. Scatchard plots of the binding of TNB-Fl to anti-TNT Ab according to eq 1, where r is the ratio of bound ligand to the Ab and A is the concentration of the free ligand.

was determined from the initial slope of its concentration titration plot, while the response of the free TNB-Fl relative to fluorescein was evaluated from the slope at the end of its titration plot. The ratio of these slopes will yield the ratio of the relative sensitivities of the free TNB-Fl and of the complex. Fifteen titration experiments have been carried out and the ratio of the molar sensitivity of free TNB-Fl to that of the complex is in the range 1.5 ( 0.5. Once the ratio of the molar sensitivities is determined, the binding constant KB of the Ab may be evaluated. We chose to evaluate the results according to Scatchard plots as described by the equation42

r/A ) -r/KD + n/KD

(1)

where KD is the dissociation constant, r is the molar ratio of the bound TNB-Fl to Ab, n is the number of binding sites on the Ab, A is the free TNB-Fl, and the binding constant is obtained from KB ) -1/KD. A typical Scatchard plot is presented in Figure 4. From the evaluation of all 15 titration experiments, a value of KB ) (3.9 ( 1.3) × 109 M-1 was obtained and the intercept on the ordinate yielded the value r ) 1.0 ( 0.1, indicating a 1:1 complex formation. Since the IgG antibodies are bivalent, one would expect a 1:2 ratio for the Ab-Ag complex, respectively. However, the dye with the long binding chain forms a bulky group that may inhibit double occupancy. TNT Assay. Our immunoassay method is based on the competition between the labeled TNB-Fl and the analyte nitro compounds for the Ab binding site. For optimal operation, the molar concentration of TNB-Fl was chosen to be half of the concentration of the Ab, and to ensure an equilibrium state, the time of the immunoassay incubation was 45 min. The ratio of the molecular sensitivities of the free TNB-Fl to the complex in the competition experiments was evaluated from the increase of the signal intensity of the free TNB-Fl to the decrease of the complex (42) Curthoys, N. P.; Rabinowitz, J. C. J. Biol. Chem. 1971, 246, 6942-6952.

1192

Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

Figure 5. Electropherograms of mixtures of 10 nM Ab, 5 nM TNBFl, 0.5 nM fluorescein, and indicated concentration of TNT in the range 0-30 ng/mL.

intensity signal upon addition of the nitro compound competitor. This ratio was derived for each experiment separately, and its value was in the range of 1.1( 0.4. The dissociation constant for the nitro compounds has been evaluated by standard methods for competitive binding.42 If two reagents compete for the same site of the antibody, then the evaluation may be done according to the following equation:42

A(n - r)/r ) KAD + BKAD/KBD

(2)

where n and A are defined as previously, B is the concentration of the free nitro compound, and KAD and KBD are dissociation constants of TNB-Fl and of the nitro compound competitor, respectively. When the ratio KBD/KAD > 10 prevails, then the amount of the bound B is negligible. Therefore, the total concentration of B may be used instead of the free B. From the direct binding experiment, we already saw that r ) 1, then a plot of A(1 - r)/r as a function of r should give a straight line with a slope of KAD/KBD. Figure 5 presents electropherograms of competition experiments with TNT. The complex signal intensity decreases and that of the free TNB-Fl increases as the concentration of TNT increases. Similar experiments with the other nitro compounds have been performed. In Figure 6, competitive binding plots according to eq 2 for the trinitrobenzenes TNT, TNB, and TNP are shown, while in Figure 7 the plots of the dinitro derivatives DNT, DNB, and DNP are presented. According to eq 2 the intercept on the ordinate should yield the dissociation constant KAD. Since this value is very low, it cannot be determined accurately due to the experimental error. Therefore, only the values of the obtained slopes have been used. From the slopes and from the KAD of TNB-Fl, the binding constants of the six nitrobenzenes have been evaluated. These results are presented in Table 2. From an analytical point of view, the ratio between the signal intensities of TNB-Fl to that of the complex may be a sensitive

Table 2. Competitive Binding Properties of TNT Analogs nitro no. of compd expts TNT TNB TNP DNT DNB DNP a

6 5 7 6 6 5

slope

binding const (M-1)

binding const (M-1)a

(1.1 ( 0.3) × 10-1 (1.3 ( 0.4) × 10-2 (1.9 ( 0.5) × 10-3 (2.0 ( 0.4) × 10-3 (2.6 ( 0.9) × 10-4 (1.3 ( 0.3) × 10-5

(4.3 ( 2.6) × 108 (5.1 ( 3.3) × 107 (7.5 ( 4.4) × 106 (7.9 ( 4.0) × 106 (1.0 ( 0.7) × 106 (5.1 ( 3.0) × 104

1.3 × 109 2.1 × 108 3.4 × 107 2.2 × 107 4.3 × 106 nd

Reference 45.

Figure 6. Competitive binding plots of TNB-Fl to the Ab as a function of the concentration of trinitrobenzenes (TNT, TNB, TNP) according to eq 2, where r is the ratio of the bound TNB-Fl to Ab, A is the concentration of free TNB-Fl and B is the concentration of the competitor.

Figure 8. Calibration curve for TNT assay obtained by plotting the ratio of signal intensity of the free TNB-Fl to the signal intensity of the complex Ab‚TNB-Fl as a function of the TNT concentration. Table 3. Limits for Immunodetection of TNT Analogues nitro compd

TNT

TNB

TNP

DNT

DNB

DNP

LOD, 1 3 10 40 90 5000 ng/mL dynamic 1-300 3-1000 10-2000 40-2000 0.09-15 5-150 range ng/mL ng/mL ng/mL ng/mL µg/mL µg/mL

Figure 7. Plots as in Figure 6 for the dinitrobenzenes (DNT, DNB, DNP).

parameter in the immunoassay analysis of TNT and its related derivatives. In Figure 8, a calibration graph for TNT was obtained by plotting such a ratio versus concentration. A linear dependence is obtained with a limit of detection 1 ng/mL. The criterion for detection was that the signal at the LOD should be at least 2 times larger than that the signal observed without any competitor and 3 times larger than the noise level. In Table 3, the LOD of all six nitro compounds studied are summarized. Dissociation Rate Constant Determination. A unique aspect of chip-based electrophoresis is the ability to separate the AbAg complex from its forming reagents faster than the chemical relaxation rate. Therefore, it should be possible to determine the dissociation rate constant koff for the Ab by following the signal intensity as a function of time as the complex migrates along the

separation channel. The analysis of multiple sample injections at different points along the column was attempted, but here was a large spread of data points due to injection variability, mainly because of the pH changes due to the electrolysis process that affects the Ab reactivity. However, several years ago,43 it was shown that a separation could be monitored at multiple time points by using a serpentine-type capillary with a rotary confocal scanning system. The serpentine capillary we used enabled fluorescence detection at six different points located from 2.5 to 17.7 cm after the injector. The dissociation rate constant determination was performed with a sample composed of 100 nM Ab, 20 nM TNBFl, and 2 nM fluorescein. An excess of Ab was used to decrease the free TNB-Fl signals as much as possible. Figure 9 presents the six electropherograms obtained from a single sample injection (43) Paegel, B. M.; Hutt, L. D.; Simpson, P. C.; Mathies, R. A. Anal. Chem. 2000, 72, 3030-3037.

Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

1193

Figure 9. Electropherograms monitoring the dissociation of the AbTNB-Fl complex monitored at six detection points along the separation CE channel by using the rotary scanner.

into the serpentine capillary with the scanner monitoring the electrophoresis process at detection points 1-6. The graphs are displayed in order of signal appearance along the capillary. On a time scale of 10-70 s, the Ab TNB-Fl signals get smaller and the fluorescein peak becomes shorter and broader. Signal due to TNBFl is not observed because an excess of Ab was used in the initial incubation and the TNB-Fl is released continuously and thus cannot form a plug at the different detection points. Even with a single-sample injection, the signals have to be normalized because of inhomogenity of the fabricated capillary dimensions and because the path of scanned laser light within the capillaries is different. We normalized the signals relative to the Gaussian profile fit to the fluorescein signal. The diffusion constant of the complex is estimated to be ∼4 × 10-7 cm2/s,9 while that of fluorescein is 3.3 × 10-6 cm2/s.4 Therefore, monitoring the complex signal height is more correct than monitoring its area because this approach is less sensitive to tails on the bands due to dissociation and adsorption processes. The fluorescein band is more of a Gaussian shape, as expected from diffusion, so for this species we integrated the signal area. Figure 10 presents the results of one of the dissociation experiments and the best-fit single-exponential decay curve. Better fits can be obtained using models that combine exponential decay with a constant background or using biexponential decay. For the constant background model the average best-fit equation is y ) 5.6e-0.062t + 0.67. However, the value 0.67 (about one-third of the signals observed after 20 s from the injection) is unrealistic high to be explained as background emission. From the competition experiments with the polynitro compounds at high concentrations, we know that the background is very low, ∼5% of the signals obtained after 20 s. The best fit to the observed data was obtained using a biexponential decay, which may imply a different affinity of the two binding sites of the Ab. However, since we have only six data points, this fitting process is not well defined and we prefer to report just the single-exponent fit realizing that it is an apparent dissociation constant. The dissociation experiments were per1194 Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

Figure 10. Least-squares fit to an exponential decay of the data points obtained for the dissociation of the Ab-TNB-Fl complex formed from a mixture of 100 nM Ab, 20 nM TNB-Fl, and 2 nM fluorescein. The presented decay curve is given by R ) 5.69e-0.033t.

formed with several laser powers in the range 35-144 mW/cm2 because some photodecomposition may be expected at the detection points where the sample is moving slowly. The average value koff of the seven experiments performed with laser power of 35-52 mW/cm2, where no photodecomposition effect was observed, is 0.035 ( 0.005 s-1. From this value and from the binding experiments, the rate constant of the forward reaction between TNB-Fl and the Ab is found to be kon ) (1.4 ( 0.7) × 107 M-1 s-1. DISCUSSION The results presented here demonstrate that microchip CE can be used to determine equilibrium binding constants and rate constants of an immunoreaction and to carry out sensitive immunoassay analysis. Electrophoretic immunoassays may be performed in one of two formats: a direct assay where the Ab is labeled or by the competition method where the analyte is tagged. In the direct method, the process is monitored by separating the free Ab from the formed complex, while in the competition method, the process is monitored by following the loss of labeled complex and the appearance of free labeled antigen. Since the electrophoretic mobility of typical complexes formed between small molecules and antibodies is similar, we conducted our study using the competition method, exploiting the difference between the electrophoretic mobility of labeled small molecules and the formed complex. Our binding constant KB ) (3.9 ( 1.3) × 109 M-1, dissociation rate constant koff ) 0.035 ( 0.005 s-1, and binding rate constant kon ) (1.4 ( 0.7) × 107 M-1 s-1 results may be compared to those from a biosensor study, where the reaction between a Cy5-labeled TNB and an immobilized Ab on the surface of a fiber-optic probe was examined.44 Their determined forward binding rate constant (44) Vijayendran, R. A.; Ligler, F. S.; Leckband, D. E. Anal. Chem. 1999, 71, 5405-5412.

was kf ) 5.5 × 105 M-1 s-1, and the dissociation rate constant was kr ) 1.5 × 10-2 s-1. From these two values, the binding constant is ∼ 3.7 × 107 M-1. This binding constant is smaller than that obtained from our results by ∼2 orders of magnitude. Since the labeling dyes and the bridge molecules are different in the two systems, this may affect the evaluated kinetic parameters. However, the major reason for this difference is, probably, due to the different state of the Ab in the two systems. In our system, the Ab is completely dissolved in the buffer solution, while in the biosensor system, the Ab is immobilized on a surface, which may significantly influence the binding kinetics. Immobilized Ab might lose its activity, which results in a lower binding rate constant. This interpretation is consistent with the observation that our forward rate constant is higher by a factor of 25. We have determined the dissociation rate constant for the AbAg complex by the unique method of kinetic CE using a folded chip channel and the rotary scanner. Due to the electrophoresis process, the Ab-Ag complex and the free TNB-Fl remain separated along the entire capillary. Hence, it is possible to monitor the dissociation process without the interference of the rebinding reaction. This unique method reveals a new application of the microfabricated capillary technology to study kinetics of chemical reactions. Since the dimensions of the microcapillaries are small, it is possible to apply high voltages to achieve fast electrophoretic separation and therefore to monitor decay of complexes with a relative short lifetime. With such devices it is possible also to study the effect of high voltage on the dissociation rate or to study the influence of saturation of the binding sites on the decay process. By controlling the temperature of the microcapillary device, it should also be possible to determine activation energies of the dissociation process. A quantitative understanding of the dissociation process is also valuable for optimizing electrophoretic assay conditions. For example, in future experiments, higher voltages or shorter distances between the injector and the detection point would be useful to decrease the migration time, and minimize dissociation. The binding constants of the unlabeled poly(nitrobenzenes) that were obtained by the competition experiments are lower by a factor of 3-5 compared to the values determined by competitive ELISA.45 Such differences are not unusual for immunoassays determined by different methods or in different laboratories.46,47 To our opinion, binding constants derived from ELISA-type experiments are overestimated because of nonspecific interactions and because methods that determine the binding constant based on midpoint analysis may be more sensitive to experimental conditions than Scatchard analysis. Nevertheless, it is interesting to note that the ratios among the binding constants obtained in our study are very similar to the ratios of the results reported by Zeck et al.45 Our data support the explanation given there45 for (45) Zeck, A.; Weller, M. G.; Niessner, R. Fresenius J. Anal. Chem. 1999, 364, 113-120. (46) Stanley, C.; Lew, A. M.; Steward, M. W. J. Immunol. Methods 1983, 64, 119-132. (47) Busch, M. H. A.; Boelens, H. F. M.; Kraak, J. C.; Poppe, H.; Meekel, A. A. P.; Resmini, M. J. Chromatogr., A 1996, 744, 195-203. (48) Hilmi, A.; Luong, J. H. T. Environ. Sci. Technol. 2000, 34, 3046-3050.

the molecular structure dependence of the binding constants of the poly(nitrobenzene) molecules, which spans several orders magnitude. The dinitro derivatives are much less tightly bound than the trinitro ones. The Ab we used was raised against the trinitrobenzene moiety, so it is reasonable to assume that the Ab is not optimal for the dinitrobenzene structure. But, even Ab that has been raised against the dinitro moiety has binding constants for the dinitrophenyllysine derivative in the range (1-4) × 107 M-1, which are similar to the results we obtained and to the results of Zeck et al.45 This indicates that the lower values of the dinitro derivatives are mainly caused by the loss of the NO2 binding interaction with the binding site of the Ab. The methyl groups also have a significant effect on the affinity of the nitrobenzene ligands toward the Ab.45 The Ab is typically prepared by coupling the TNB moiety to the carrier protein through position 1 of the benzene ring. Hence, methyl groups at that position will increase the affinity of the methylated nitrobenzenes; the binding constants of TNT and of DNT are larger by a factor of 8 than that of TNB and DNB, respectively. Derivatives with negative charges will decrease the affinity because of the repulsive forces between such groups and the antibody binding site.45 The binding constants of TNP are smaller by a factor of 60 than that of TNT and the binding constant of DNP is smaller by a factor of 150 than that of DNT. Finally, the sensitivity and dynamic range of our microchip TNT analysis method are very good. The LOD of 1 ng/mL demonstrated here is much lower than those obtained by other methods using capillary chip devices.35-37,48 The obtained LODs in previous studies were in the range 24 ng/mL-10 µg/mL. By the assay developed here, TNT may be determined in the range of 1-300 ng/mL. This dynamic range is much broader than that of the biosensor array38 that has similar sensitivity of 1 ng/mL, but with a dynamic range of only 1-7 ng/mL. CONCLUSIONS A homogeneous immunoassay for TNT and some of its derivatives has been developed using a microfabricated capillary electrophoresis chip. The equilibrium binding constant, binding rate constant, and dissociation rate constant of labeled TNT have been determined. A competitive immunoassay has been used to assay unlabeled nitrobenzenes in a range of concentrations. TNT may be assayed with high sensitivity and with a wide dynamic range using the method and apparatus presented here. This work establishes the feasibility of both high-throughput and point-ofanalysis microfabricated immunoassay chip systems. ACKNOWLEDGMENT We thank Stephanie Yeung for expert chip fabrication, Charlie Emrich for valuable assistance with device design, and the Mathies group for help and support. This work was supported by Pluvita Inc. and by Amersham- Biosciences. Received for review September 30, 2002. Accepted December 13, 2002. AC020599G

Analytical Chemistry, Vol. 75, No. 5, March 1, 2003

1195