Competitive Immunoassay for Microliter Protein Samples with

the immunoreaction and fluorescence detection are sepa- rated into two distinct ... is less binding between Ag* and immobilized Ab; there- fore, the a...
0 downloads 0 Views 204KB Size
Download PDF
Anal. Chem. 2004, 76, 1871-1876

Competitive Immunoassay for Microliter Protein Samples with Magnetic Beads and Near-Infrared Fluorescence Detection Xiaoyan Zhao and Scott A. Shippy*

University of Illinois at Chicago, Department of Chemistry (M/C 111), 845 West Taylor Street, Chicago, Illinois 60607

A competitive immunoassay with near-infrared (NIR) fluorescence detection to analyze microliter biological samples with an amol limit of detection (LOD) is described. An important feature about this technique is that the immunoreaction and fluorescence detection are separated into two distinct steps, allowing for independent optimization. In the immunoreaction step, NIR fluorescence-labeled antigen (Ag*) competes with the unlabeled analyte (Ag) for antibodies (Ab) immobilized on the surface of paramagnetic beads. A magnet is then used to separate the bound antigen from the free in the supernatant. As the amount of Ag in the sample increases, there is less binding between Ag* and immobilized Ab; therefore, the amount of Ag* in the supernatant is proportionally related to the amount of Ag in the sample. In the fluorescence detection step, aliquots of the supernatant are concentrated onto a protein binding membrane by a capillary blotting technique with an optimized 33 nL/min flow rate. The fluorescence of the blotted spots is detected with a NIR sensitive photon counting system that is optimized to an instrumental LOD of 30 000 fluorophore molecules. This competitive assay demonstrates a sample LOD of 400 pg/mL of unlabeled rabbit immunoglobulin G spiked into bovine serum. This design features low sample volumes and reagent consumption. The determination of dilute proteins from small-volume biological samples is of particular interest in disease proteomics.1 A rapid clinical application utilizing an immunoassay with reproducible and straightforward quantitation would be ideally suited for the determination of protein biomarkers from volume-limited samples. Since their initial development in the early 1970s,2 solidphase immunoassays have been popular for quantitative detection of low concentrations of proteins and peptides from biological samples. The combination of the selectivity of antibody (Ab)antigen (Ag) interaction and the simplicity of a solid-liquid separation makes these methods powerful and useful for complex biological matrixes, such as blood, plasma, or urine.3-5 Conven* To whom correspondence should be addressed. Tel: 312-355-2426. Fax: 312-996-0431. E-mail: [email protected]. (1) Hanash, S. Nature 2003, 422, 226-32. (2) Catt, K.; Tregear, G. W. Science 1967, 158, 1570. (3) Hage, D. S. Anal. Chem. 1999, 71, 294R-304R. (4) Hunter, W. M.; Corrie, J. E. T. Immunoassays for Clinical Chemistry, 2nd ed.; Churchill Livingstone: New York, 1983. 10.1021/ac035198v CCC: $27.50 Published on Web 03/06/2004

© 2004 American Chemical Society

tionally, immunoassays are carried out in microtiter plates, which suffer from a lack of control over coating solid surfaces and are not applicable for submicroliter sample volumes. For volumelimited samples, there are several reports of immunoassays performed during capillary electrophoresis (CE).6,7 While powerful, the drawback to these CE methods is the requirement for the development of a fast separation (before the Ag/Ab complexes can dissociate) and the well-established phenomena of protein adsorption to the capillary walls.8 Recently, we introduced a capillary blotting approach for use in a dot immunoassay (DIA) format that utilized a microliter sample volume.9 Although this refined DIA method is capable of detecting fmol levels of peptides in microliter biological samples, there are potential limitations in terms of the membrane background generated by nonspecific adsorption and the immune recognition from a small area blot of a highly concentrated matrix protein solution. Importantly, capillary blotting of microspots has been shown to conveniently work for small volume samples. Microscale magnetic beads coated with either immunoglobulins or ligands have proven to be a useful tool for a wide variety of analytes, including viruses,10 proteins,11,12 RNA,13 DNA,14,15 cells,16 and bacteria and toxins.17 Activated beads are commercially available, and molecules such as Abs, Ags, hormones, DNA, and RNA can be covalently bound to the surface of the uniform, superparamagnetic beads. In a magnetic bead-based competitive immunoassay, labeled Ag (Ag*) and Ag molecules in the solution compete for the same bead-immobilized Ab molecules. After (5) Wild, D. The Immunoassay Handbook, 2nd ed.; Nature Publishing Group: New York, 2001. (6) German, I.; Kennedy, R. T. J. Chromatogr., B 2000, 742, 353-62. (7) Pritchett, T.; Evangelista, R. A.; Chen, F. T. Bio/Technology 1995, 13, 144950. (8) Schmalzing, D.; Nashabeh, W. Electrophoresis 1997, 18, 2184-93. (9) Zhao, X. Y.; Kottegoda, S.; Shippy, S. A. Analyst 2003, 128, 357-62. (10) Kassimi, L. B.; Gonzague, M.; Boutrouille, A.; Cruciere, C. J. Virol. Methods 2002, 101, 197-206. (11) Yu, L.; Gaskell, S. J.; Brookman, J. L. J. Am. Soc. Mass Spectrom. 1998, 9, 208-15. (12) Widjojoatmodjo, M. N.; Fluit, A. C.; Torensma, R.; Verhoef, J. J. Immunol. Methods 1993, 165, 11-19. (13) Miyachi, H.; Masukawa, A.; Ohshima, T.; Hirose, T.; Impraim, C.; Ando, Y. J. Clin. Microbiol. 2000, 38, 18-21. (14) Elkin, C.; Kapur, H.; Smith, T.; Humphries, D.; Pollard, M.; Hammon, N.; Hawkins, T. Biotechniques 1302, 32, 1296-300. (15) Cheng, D. W.; Armstrong, K. C. Genome 2002, 45, 977-83. (16) Lopez, J. A.; Bioley, G.; Turtle, C. J.; Pinzon-Charry, A.; Ho, C. S.; Vuckovic, S.; Crosbie, G.; Gilleece, M.; Jackson, D. C.; Munster, D.; Hart, D. N. J. Immunol. Methods 2003, 274, 47-61. (17) Yu, H. Anal. Chim. Acta 1998, 376, 77-81.

Analytical Chemistry, Vol. 76, No. 7, April 1, 2004 1871

incubation and separation of the magnetic beads from the solution, the amount of free Ag* is quantified and related to the sample Ag concentration. Although magnetic beads are a different solid phase from the conventional microtiter or capillary surfaces, assays that are based on the magnetic beads are usually carried out in vials,18,19 and then the fluorescence of the supernatant is measured with a fluorometer. When microtiter plates are used, detection takes place after the beads are collected against the side wall of the microtiter wells.20 In one unique example, CE with laserinduced fluorescence detection of a bead supernatant was used to monitor the amount of anti-fluorescein immunoglobulins with a limit of detection (LOD) of 300 ng/mL.21 A recent and exciting use of magnetic beads is via a capillary flow system.22 The magnetic beads are packed along the capillary, and samples can be passed through the packed bed. Hayes et al. suspects that the immunoreaction mostly occurs on the surface of the packing bed, which results in a faster reaction and easy epifluorescence microscope detection. Volumes as small as several microliters are required for the assay, and detection of proteins at physiologically relevant levels is demonstrated. However, most of the binding surface of the magnetic beads remains intact and buried in the packed bed. The assay was reported to be detection-limited, which is, in part, a result of the detection zone’s being 200-fold smaller than the size of the packed bed. In this study, we explore a new format of fluorescence immunoassay for submicroliter volumes. Magnetic beads are used as the limiting reagent in a competitive immunoassay in which Ag* competes with unlabeled Ag from the sample. By limiting the magnetic beads, the antibody and, hence, the available binding sites are also limited. Assuming fixed volumes of sample and magnetic beads and a constant Ag* concentration, varying the unlabeled Ag concentration results in different partitions between the bound Ag* and Ag. The more unlabeled Ag present in the sample, the less Ag* is bound to the Ab, and therefore, the more Ag* is present in the supernatant. The free Ag* in the supernatant is a function of a given sample concentration of unlabeled Ag. By using samples of known Ag concentrations as standards, the quantity of unknown amount of Ag can be solved. In our assay design, the supernatant is separated from the beads with a rare earth magnet, and our recently developed capillary blotting technique is employed to concentrate unreacted, Ag*-containing supernatant onto a protein binding membrane for detection. Finally, the fluorescence intensity of each blotted spot is related to unlabeled protein Ag in the sample. The immunoreaction and fluorescence detection are separated into two distinct steps, allowing for independent optimization. Because the supernatant spots are concentrated blots of a chemically complex solution that likely fluoresces at visible wavelengths, it is reasonable to move the fluorescence detection to the near-infrared (NIR) region. Most biomolecules do not exhibit fluorescence in the NIR, and immunoassays with NIR fluorescence have been previously reported.9,23 Further, it would be expected that the membrane background (18) Kourilov, V.; Steinitz, M. Anal. Biochem. 2002, 311, 166-70. (19) Nakamura, N.; Hashimoto, K.; Matsunaga, T. Anal. Chem. 1991, 63, 26872. (20) Krylov, S. N.; Dovichi, N. J. Electrophoresis 2000, 21, 767-73. (21) Paquette, D. M.; Banks, P. R. Electrophoresis 2001, 22, 2391-97. (22) Hayes, M. A.; Polson, N. A.; Phayre, A. N.; Garcia, A. A. Anal. Chem. 2001, 73, 5896-902. (23) Patonay, G.; Antoine, M. D. Anal. Chem. 1991, 63, 321A-7A.

1872

Analytical Chemistry, Vol. 76, No. 7, April 1, 2004

caused by the nonspecific protein binding is eliminated in this design, in comparison to the western blot-adapted DIA.9 The detection background is thus limited to membrane autofluorescence and light scattering from intact membrane surfaces. A single microliter sample of serum is used in the assay format, and multiple 170-nL blots are used from each serum sample for detection. The result is an immunoassay with a low LOD to detect proteins from very small volumes of biological samples. EXPERIMENTAL SECTION Chemicals. Unless otherwise stated, reagents were obtained from Sigma Chemical Company (St. Louis, MO). All dilutions were made with pH 7.4 phosphate buffer saline (PBS), and 1 L of PBS contained 8.0 g of NaCl, 0.2 g of KCl, 2.717 g of Na2HPO4‚7H2O, and 0.24 g of KH2PO4. Bovine serum albumin (BSA) at a weight concentration of 0.3% was used to eliminate the analyte loss on the container and tubing walls and block the nonspecific adsorption on the beads. IRDye 800-conjugated affinity-purified rabbit anti-rat immunoglobulin G (IgG) was purchased from Rockland (Gilbertsville, PA), with peak absorption and emission wavelengths at 778 and 806 nm, respectively. Unlabeled rabbit anti-rat IgG was also obtained from Rockland (Gilbertsville, PA). Nitrocellulose membranes with a pore size of 0.2 µm were purchased from Osmonics Inc. (Minnetonka, MN). A solution of erioglaucine disodium salt (FD&C Blue 1) manufactured by McCormick & Company, Inc (Baltimore, MD) was used as received to stain the membrane. Magnetic beads M-280 with surface immobilized sheep anti-rabbit IgG were purchased from Dynal Inc. (Lake Success, NY). The magnetic beads were washed following the manufacturer suggested procedure just before use and prepared at a stock concentration of 1 mg/mL. Competitive Immunoassay. Different concentrations of unlabeled rabbit IgG were used as target Ag in this study. The target Ag was diluted in 0.3% BSA in PBS buffer. A volume of 1 µL of Ag was mixed with 1 µL of Ag* and 1 µL of magnetic beads in 250µL microcentrifuge tubes (Fisher Scientific, Hanover Park, IL). The solutions were vortexed briefly before they were incubated overnight for 12 h at 4 °C. Stirring or shaking was not applied because of the small volume of the reacting solution (3 µL). The tubes were then held to magnets by orthodontic wax for 30 min. The magnetic particles were collected in the bottom of the tube, but the pellet was not visible due to the low concentration of the beads. The supernatant was pipetted out carefully and stored in a new centrifuge tube. A 1-mL insulin syringe (Becton Dickinson and Company, Franklin Lakes, NJ) was used to withdraw the supernatant into a segment of Tygon tubing (∼250-µm i.d.) from which the sample would be blotted by a capillary. Bovine serum spiked with target Ag was used to evaluate the assay for a biological fluid. Capillary Blotting. A capillary blotting method was used to concentrate aliquots of the sample to a small blotting area, as previously described with minor changes discussed below.9 Briefly, a fused-silica capillary (50-µm i.d., 360-µm o.d.) from Biotaq.com Inc (Gaithersburg, MD) was cut to 1.5-cm length. Only tubing that had a smooth cut as seen under a dissecting scope (Nikon USA, Melville, NY) was used as a blotting tip. The samplefilled Tygon tubing was connected between the capillary blotting tip and a 25-µL Hamilton syringe (Reno, NV). An air bubble was used to separate the driving solution in the syringe from the

sample in the Tygon tubing. A rare earth NdFeB magnet (Edmund Scientific, Barrington, NJ) was taped near the junction of the Tygon tubing and capillary blotting tip to retain any magnetic beads that had not been initially separated from solution. The supernatant was delivered to a dry nitrocellulose membrane by a syringe pump (Harvard Apparatus, Holliston, MA) at a flow rate of 33 nL/min for 5 min to deposit a spot of Ag*. The roughly 170-nL deposition allowed efficient sample handling and sufficient fluorescence signal. Although longer blotting times allow more solution to be concentrated into an individual spot, it takes a much longer time to blot a series of different samples. Further, only 1.5 µL of the supernatant was withdrawn from the reaction tube after applying the magnet to avoid disturbing the settled beads, and three spots for each sample are easily and consistently obtained. The size of the blotted spots was measured visually after the blotted membrane was scanned by the diode laser. Briefly, the membrane was soaked in an erioglaucine disodium salt solution (FD&C Blue 1) for 2 s, taken out of the dye solution and blotted with a paper towel. A color difference between a protein spot and the background was evident after 1-2 min. The diameter of the spot was measured under a microscope. Detection System. The instrument used for measuring NIR fluorescence from a membrane was built in-house and has been previously described.9 Briefly, the excitation source is a solid-state laser diode with 780-nm emission. The beam is filtered and focused onto the membrane that was mounted to an x-y positioner. The membrane is scanned by moving the membrane stage programmed through the LabView software (National Instruments, Austin, TX). The fluorescence image was collected with a 4.0× microscope objective (Melles Griot, Irvine, CA), frequencyand spatially filtered before photon counting with a photomultiplier tube system (Hamamatsu, Bridgewater, NJ). The data from the counter was collected with an in-house-designed program in LabView and an E series data acquisition board (National Instruments, TX). The counts for a given peak were taken to be the highest point on the peak. Noise was measured as standard deviation of the baseline. RESULTS AND DISCUSSION Capillary Blotting and NIR Fluorescence Detection. When the sample volume is limited, the measurement of dilute components is challenging. Two notable aspects are the ability to manipulate such diminutive volumes and the low mass available for analysis. This report describes the combination of a low-masssensitive NIR fluorescence detection method with high selectivity of a competitive immunoassay for dilute components in chemically complex solutions. Previously, we reported the modification of a DIA for concentrating neuropeptides from a single microliter volume onto a protein binding membrane, followed by a NIR immunofluorescence western blot.9 This immunoassay technique featured a low-picogram LOD, but was limited by immune recognition of bound neuropeptides and the nonspecific binding of reagent immunoglobulins over the whole membrane. To overcome these challenges, we report a new assay format in which the analyte (Ag) in a microliter volume sample competes with Ag* to bind to Ab-coated magnetic beads in solution. Following an incubation time, the beads with bound Ag and Ag* are separated from solution by exposure to a rare earth magnet. Aliquots of the supernatant solution are concentrated into small

Figure 1. Photomicrograph of capillary blotted spots. The nitrocellulose membrane with 170 nL of 0.3% BSA solution spots blotted on its surface was stained with FD&C BLUE 1 dye solution. The top of the picture shows a ruler with 1/16 in. as the smallest scale.

spots on a membrane via a capillary blotting method. Scanning the blotted membrane with a diode laser allows the direct quantitation of the blotted fluorescence material and, indirectly, the endogenous analyte down to attomole amounts of the protein Ag. For sensitive detection, the Ag* contained in the sample supernatant is concentrated into a small spot on a protein-binding membrane by the use of a capillary blotting method. The flow rate of the supernatant to the membrane was optimized such that the solution flowing to and evaporating from the membrane is roughly equal. Our experiments show that when the flow rate is faster than 60 nL/min, the blot spreads well beyond the outer diameter of the capillary. Spots can be visually observed before they dry. On the other hand, when the flow rate is below 15 nL/ min, not only does the blotting step become very slow and timeconsuming, but also the solution evaporates before it reaches the membrane surface, leaving crystallized salts on the tip. At the optimized flow rate of 33 nL/min, there is a smooth deposition, and the spots appear not to spread beyond the dimension of the capillary tip. By confining the volume to a small membrane area, Ag* is concentrated. The size of the spots generated by capillary blotting contributes to the magnitude of the signal seen by the detector and is important for assay reproducibility. To study the blot size reproducibility, an Ag* solution was deposited on a nitrocellulose membrane over a consistent time interval. Although dried spots are not visible by eye, the spot area was estimated by differential adsorption of a hydrophilic dye. The solid support used in these experiments is hydrophilic, but the blotted spots, covered by a layer of protein, are not as hydrophilic as the blank membrane. Figure 1 is a micrograph of the dye-treated membrane and shows spot diameters of ∼500 µm. The spots are mostly regular in shape and only slightly larger than the outer diameter (360 µm) of the capillary tip. The size of these spots is partially dependent upon the diameter of the capillary used for blotting. Smaller spot areas will likely be possible with smaller-diameter capillaries and flow rates optimized for those capillaries. These results represent a 10-fold decrease in area as compared to the smallest spots generated by a commercial vacuum blotting apparatus (1 × 2 mm).9 The reproducibility of the volume deposition was studied by determining the fluorescence signal of the Ag* from a series of Analytical Chemistry, Vol. 76, No. 7, April 1, 2004

1873

Figure 2. Direct blotting of Ag* onto a nitrocellulose membrane. The linear regression equation is fluorescence counts ) 6.54(counts/ (pg/mL of Ag*)) + 393(counts), (R ) 0.998). The two insets are the cross scannings of one typical spot in the x and y directions. The calibration curve results a LOD of 11.5 pg/mL of Ag* solution. With the blotting volume of 170 nL, the mass LOD is 2 fg.

blots at different concentrations. In Figure 2, the fluorescence signal measured over membrane blots is plotted versus the concentration of Ag*. This calibration curve allows the calculation of a LOD and is found to be 11.5 pg/mL. The error bars in Figure 2 are from three spots from the same 1-µL sample. The volume deposited for each spot is 170 nL, and at this volume, the LOD is 5%) were not helpful because they blocked the capillary tip during deposition (data not shown). Capillary coatings are another alternative for preventing analyte loss with the blotting step.24,25 This is particularly attractive for samples that do not contain a large protein background. Magnetic Bead Immunoassay. In a competitive immunoassay, Ag* competes with unlabeled Ag in the sample for binding to the Ab. The more Ag present in the sample, the less binding there is between Ag* and immobilized Ab; therefore, the more free Ag* is left unbound in the supernatant. Thus, the amount of Ag* in the supernatant is proportionally related to the concentration of Ag in the sample. To monitor the low amounts of Ag*, a highly sensitive detection scheme is required, such as the NIR fluorescence system just described. Additionally, the means to separate bound and free Ag* in small volumes without significant dilution due to a separation is particularly important. In this report, magnetic beads are used for this separation that is applicable to small volumes. Finding the Optimal Magnetic Bead Concentration. Commercially available magnetic beads with immobilized immunoglobulins have been demonstrated in several immunoassay formats.11,18,21,22 In contrast to other reports, the beads in this report not only are used to provide a separation of free and bound Ag* in a magnetic field, but also serve as the limiting reagent. First, the concentration of magnetic beads and the amount of immobilized antibody were optimized for our detection system as the limiting reagent in the competitive immunoassay. The percent binding of Ag* was determined near equilibrium between magnetic beads-Ab and Ag*. A long incubation time, which estimates near equilibrium of the immunoreaction, was chosen for this assay, because this allows estimation of the lowest amount of antibody that can be used in our system. In Figure 3, the percentage of detected free Ag* is plotted versus varying amounts of magnetic beads added to the reaction mixture. When the bead concentration is much higher than Ag* concentration, almost all of the Ag* molecules are collected by the beads, and a low percentage of free Ag* is seen. On the other hand, when the Ag* concentration is much larger than the bead concentration, only a small fraction of Ag* is able to bind to the beads, and the relative amount of free Ag* tends toward 100%. When one reactant is not dominant over the other, the percent free Ag* is dictated by equilibrium and related to reactant concentration. Figure 3 shows a linear equilibrium bead concentration range of 1 µg/mL-1 mg/mL. For maximal dynamic range, the Ab concentration corresponding to the 70% binding ratio, which is a bead concentration of range of 10-20 µg/mL, was selected. The exact value was adjusted in this range following the (24) Horvath, J.; Dolnik, V. Electrophoresis 2001, 22, 644-55. (25) Doherty, E. A.; Meagher, R. J.; Albarghouthi, M. N.; Barron, A. E. Electrophoresis 2003, 24, 34-54.

Figure 3. Binding efficiency vs concentration of magnetic beads. Different concentrations of magnetic beads were mixed with 1 µL of Ag* at 100 ng/mL (- -2- -) and at 10 ng/mL (- -9- -) and incubated overnight. The magnet was applied to the mixture for 30 min to separate the free and bound Ag*. The supernatant with free Ag* was blotted on the membrane, and the fluorescence was measured. A blank was prepared by using the same procedure described above, except that PBS without magnetic beads was used to incubate with Ag*. The percentage of free Ag* was calculated by dividing the result from a concentration of magnetic beads with the result from the blank.

Figure 4. Percent binding vs different concentrations of Ag*. Different concentrations of Ag* were mixed with 1 µL of either magnetic beads at 10-µg/mL or with PBS solution that contained no beads. The incubation conditions were the same as Figure 3.

determination of an optimal concentration of Ag*. In a previous report,26 a similar competitive assay was designed as time-limited, and an excess amount of magnetic beads was used. Because here the magnetic beads are not only a solid-phase support but also a limiting reactant, this extra step yielded a 50-fold improved LOD over the time-limited assay. The advantage of reducing the Ab concentration in an immunoassay has been discussed in detail by others.27 Finding the Optimal Concentration of Ag*. Second, the amount of the Ag* was optimized for low concentration Ag samples. The fluorescent tracer Ag* competes with the analyte for a limited number of binding sites; thus, the ability to detect the difference in tracer fluorescence intensity with or without analyte in the sample determines the dynamic range of the assay. Figure 4 shows a relationship between percent binding of varying concentrations of Ag* incubated with a constant magnetic bead concen(26) Kiselev, M. V.; Gladilin, A. K.; Melik-Nubarov, N. S.; Sveshnikov, P. G.; Miethe, P.; Levashov, A. V. Anal. Biochem. 1999, 269, 393-98. (27) Ekins, R. Nucl. Med. Biol. 1994, 21, 495-521.

tration (determined above). This curve has a peak centered at 1 ng/mL. When the concentration of Ag* is larger than the amount of Ab immobilized on the magnetic beads, only a small number of binding sites are available, so a low percentage of binding is observed. Additionally, low relative binding percent is seen at low Ag* concentrations. When the Ag* concentration is significantly low, the average distance between Ag* and Ab is greatly increased, and the collision rate is reduced. The percent binding drops accordingly, as is shown on the left side of the curve. This suggests that the reaction is still far from equilibrium. The reaction incubation times are not unreasonable given that others have shown that the binding of monoclonal Ab immunoglobulin M to Ag adsorbed to an unshaken microtiter plates required 16-32 h for protein concentration ranging from 3 mg/L to 0.1 mg/L.28 The time for a very dilute solution to reach equilibrium is greatly increased. Furthermore, the reaction at the solid-liquid interface can be diffusion-limited due to depletion of reactants close to the surface, which is several magnitudes slower than reaction-limited kinetics.29 A longer incubation time or higher temperature would likely increase the percent binding at those low bead concentrations, but both modifications would also likely contribute to increased analyte decomposition in a biological sample and were not considered, given low sample availability. Low-intensity sonication may have avoided analyte loss and sped the immune reaction but were not tried in these studies. On the basis of these observations, 1-10 ng/mL levels of Ag* concentration was selected throughout the rest of the experiments. Competitive Immunoassay. The detection limits of the magnetic bead-based immunofluorescent assay were evaluated using various concentrations of rabbit IgG with a volume of 1 µL. The rabbit IgG dose-response curve (data not shown) is based on the competitive assay with 1 µL of Ag* and magnetic beads at the concentration optimized for our detection system. A volume of 1 µL of rabbit IgG was used at different concentrations from 5 pg/mL to 5 mg/mL. The lowest detectable concentration was at 0.5 ng/mL, and the range of quantitation was over nanograms per milliliter to milligrams per milliliter. Below this range, the curve levels off to fit a sigmoidal model. To demonstrate a small-volume competitive immunoassay using a real biological fluid, rabbit IgG was spiked into 10 times diluted bovine serum, corresponding to ∼6 g/L of total protein (the undiluted bovine serum had a total protein concentration at around 60 g/L, which tended to clog the capillary blotting tip). Three assays (with six samples per assay) were carried out in parallel, and the results are shown in Figure 5. A linear regression equation was generated for the linear part of the plot (y ) 113x + 1120, R ) 0.998). The precision of the intraimmunoassay was investigated at three different concentrations: 500 pg/mL, 50 ng/ mL, and 5 µg/mL. The coefficients of variation of the intra-assay were 14.6, 5.9, 5.4% at 500 pg/mL, 50 ng/mL and 5 µg/mL, respectively, and demonstrate acceptable reproducibility. The LOD for the assay was calculated as three times the standard deviation of the signal above the lower range plateau and was found to be 400 pg/mL, corresponding to a mass LOD of 400 fg of analyte per sample. The mass LOD of 5 amol/sample is 30-times better (28) Franz, B.; Stegemann, M. Immunochemistry of Solid-Phase Immunoassay, Butler, J. E., Ed.; CRC Press: Boca Raton, 1991; Chapter 18. (29) Stenberg, M.; Nygren, H. J. Immunol. Methods 1988, 113, 3-15.

Analytical Chemistry, Vol. 76, No. 7, April 1, 2004

1875

immunoassay. Because only small volumes and detectionoptimized, low concentrations were used for expensive assay reagents, costs are greatly reduced. Additionally, multiple samples can be prepared simultaneously. They react under the same conditions for the same length of time, and at the same point, they are all held on the magnets to separate the beads from the solution. Supernatants are stored in separate vials and blotted afterward. The results are not affected by different times elapsed until blotting. Even the blotting speed can be improved equipped with a pump with multiple driving syringes for multiple samples at the same time. All these make automation, sample multiplexing, and high-throughput screening promising future directions. Figure 5. Competitive immunoassay of rabbit anti-rat IgG spiked in 10-times-diluted bovine serum. Different concentrations of 1-µL rabbit IgG in bovine serum (∼6 g/L total protein), 1-µL magnetic beads at a concentration of 20 µg/mL, and 1-µL Ag* at a concentration of 1 ng/mL were mixed well and incubated overnight.

than a flow-based immunoassay using magnetic beads.22 Compared to a LOD of 40 ng/L with an integrated waveguide biosensor,30 our results have a somewhat higher concentration detection limit, although a >1000-fold reduction in sample consumption allows the use of considerably smaller volume samples. Further, the slope at the quantitative portion of Figure 5 is at the same value as the standard rabbit IgG dose-response curve (data not shown), indicating similar sensitivity for both standards and real samples. Importantly, this method functions very well with a chemically complex biological sample, such as serum. This immunoassay method is potentially attractive for clinical applications in which only small-volume samples are generated. The capillary blotting leads to a signal enhancement due to the concentration of fluorophore on the solid phase. The use of the NIR fluorescence provides a low expected background. This allows a lower level of Ag* to be used and improve the LOD from a given (30) Ligler, F. S.; Breimer, M.; Golden, J. P.; Nivens, D. A.; Dodson, J. P.; Green, T. M.; Haders, D. P.; Sadik, O. A. Anal. Chem. 2002, 74, 713-19.

1876

Analytical Chemistry, Vol. 76, No. 7, April 1, 2004

CONCLUSIONS In this new protein detection method, immunoassay and fluorescence detection are separately explored, optimized and evaluated. Near-IR fluorescence detection of capillary blot microspots greatly decreases the detection background, allowing the detection of 30 000 of fluorophore molecules. In the competitive immunoassay, both the beads and the Ag* concentration are optimized to low values for the detection instrumentation. This provides improved detection limits and dramatically decreases reagent consumption. A microliter volume of biological sample is successfully analyzed with a low LOD of 400 pg/mL, corresponding to a mass LOD of 5 amol. More importantly, the competitive assay reported here may also be good for analytes that are not suitable for sandwich-type assays, such as small peptides. ACKNOWLEDGMENT The authors acknowledge the support of the State of Illinois and the University of Illinois Campus Research Board for funding this research.

Received for review October 9, 2003. Accepted February 2, 2004. AC035198V