Accelerated Articles Anal. Chem. 1994,66, 1369-1377 This Research Contribution is in Commemoration of the Life and Science of I. M. Kolthoff (1894- 1993).
Separation-Free Sandwich Enzyme Immunoassays Using Microporous Gold Electrodes and Self-Assembled Monolayer/Immobilized Capture Antibodies Chuanmlng Duan and Mark E. Meyerhoff' Department of Chemistry, The University of Michigan, Ann Arbor, Michigan 48109 A novel separation-free sandwich-type enzyme immunoassay for proteins is performed by designing an electrochemical detection system that enables preferential measurement of surface-bound enzyme-labeled antibody relative to the excess enzyme-labeled reagent in the bulk sample solution. In this initial model system, the assay is carried out using gold-coated microporous nylon membranes (pore size 0.2 fim) which are mounted between two chambers of a diffusion cell. The membrane serves as both a solid phase for the sandwich assay and the working electrode in the three-electrode amperometric detection system. The capture monoclonal antibody is immobilized covalently on the gold side of the membrane via a self-assembled monolayer of thioctic acid. In the separationfree sandwich assay, both model analyte protein (human chorionic gonadotropin; hCG) and alkaline phosphatase labeled anti-hCG (ALP-Ab) are incubated simultaneously with the immobilized captureanti-hCGantibody. Surface-bound ALPAb is spatially resolved from the excess conjugate in the bulk sample solution by introducing the enzyme substrate (4aminophenyl phosphate) through the back side of the porous membrane. The substrate diffuses rapidly through the porous membrane where it first encounters bound ALP-Ab at the gold surface. The enzymatically generated product, aminophenol, is detected immediately by oxidation at the gold electrode (at +0.19 V vs Ag/AgCl), and the magnitude of current is directly proportional to the concentration of hCG in the sample. The response time after substrate addition is less than 1 min, although maximum response toward the analyte protein requires a sample/conjugate preincubation time of 30 min with the porous electrode. The assay is demonstrated to function effectivelyin both buf'fer and whole human blood with a detection 0003-2700/94/036&1369$04.50/0 0 1994 American Chemlcai Society
limit of 2.5 units/L hCG (in blood), which is comparable to most of heterogeneous EIAs that require multiple washing steps.
Enzyme immunoassays (E I A p 9 have become important analytical methods in clinical chemistry laboratories for the selective detection of drugs, hormones, and proteins at trace levels. An EIA method can be heterogeneousor homogeneous, depending on whether or not additional washing steps are required to separate free and bound enzyme label. EIAs can also be competitive or noncompetitive, depending on the availability of antibody binding sites. Of all these different EIA schemes, noncompetitive heterogeneous sandwich assays have the advantage of employing two coexisting determinant sites on the same protein antigen to be detected. As a result, sandwich-type assays generally exhibit better specificity and sensitivity when protein analytes are determined.lJ0 In conventional assays, a given sample or standard containing the analyte protein is allowed to bind to excess solid-phase antibody (capture antibody). The sample matrix is then washed away, and immunobound analyte is then allowed to bind with a second anti-protein antibody which is labeled with ~~
(1) Gosling, J. P. Clin. Chem. 1990, 36, 1408-27. (2) Ishikawa, E.; Hashida, S.;Kohno, T.; Hirota, K.Clin. Chim. Acra 1990,194, 51-72. (3) Ishikawa, E. Clin. Biochem. 1987, 20, 375-85. (4) Monroe, D.A M / . Chem. 1984,56,920A-931A. (5) Price, C. P., Newman, D. J., Eds. Principles and Practice oflmmunoassay; Stockton Press: New York, 1991. ( 6 ) Ngo, T. T., Lenhoff, H. M., Eds.Enyzme-Mediared Immunoassay; Plenum Rcss: New York, 1985. (7) Collins, W.P., Ed. Alrerna?iueImmunoassays; John Wiley & Sons: New York, 1985. (8) Ishikawa, E.; Kawai, T.; Miyai, K.Enzyme Immunoassay; Igaku-Shoiu: New York. 1981. (9) Maggio, E. T., Ed. Enzyme-Immunoassay; CRC Rcss: Boca Raton, FL, 1980. (10) Huna, R. 0. The Clinical Marker hCG Praeger: New York, 1987.
AnaiyticalChemMry, Vol. 66, No. 9, May 1, 1994
1369
an enzyme. After another washing step to eliminate excess enzyme-labeled antibody, the amount of enzyme bound to the solid phase is determined by adding the substrate and measuring the rate of product generation, which is directly proportional to the amount of antigen present. Such methods are now used routinely to detect diagnostically important blood proteins, including creatine kinase-MB (to detect the occurrence of myocardial infarctions), prostate-specific antigen (PSA) (used to screen for prostate cancer), and human chorionic gonadotropin (hCG) (used to confirm pregnancy). Even though sandwich-type enzyme immunoassays are used widely in clinical laboratories throughout the world, such methods require multiple washing steps to implement,’ and this has created the need for rather complex associated instrumentation for high-volume immunoassay analyzers. At the same time, this need for washing steps has limited the adaptation of these assays into more convenient and highly portable test systems that would be valuable for detecting diagnostically important proteins in field locations (e.g., doctors offices, emergency vehicles, etc.). Immunoconcentration!1,12 techniques using capture antibodies immobilized on glass and other filter material have reduced some of the effort associated with implementing sandwich-type EIAs (especially the length of binding incubation times), although discrete separation and washing steps are still required. Further, such immunoconcentration methods are only useful for detecting proteins in serum and plasma samples because the presence of red blood cells will clog such filter-based devices. The development of a rapid, simple, nonseparation method for the determination of proteins has been a long-standing goal since the beginning of immunoassay technology. Gibbons et a1.I3and Armenta et a1.I4used chromogenic and fluorogenic galactoside-Dextran substrates in devising homogeneous enzyme immunoassays for C-reactive protein, ferritin, and immunoglobulins; however, the degree of modulation of enzyme activity in this homogeneous protocol was rather low, rendering the method impractical for real world applications. Chen et al.I5 combined a two-enzyme channeling technique with immunocapillary migration to produce a test-strip format for the detection of hCG. This approach, however, was only a wash-free assay, not a truly separation-free assay, since the strip was removed from the sample before addition of the substrate. Schray and Niedbala16 reported on a separationfree dual solid-phase enzyme immunoassay for macromolecules, but this assay scheme required 24 h to complete. It has been recognized that coupling electrochemical detection for performing enzyme immunoassays would be advantageous, particularly in developing truly homogeneoustype assay^.!^-^^ In such instances, electrodes have the
(14) (15)
(16) (17) (18) (19)
(20)
Valkirs, G.; Chism, D.; Barton, R. Clin.Chem. 1985, 31, 960. Valkirs, G.; Barton, R. Clin.Chem. 1985, 31, 1427-31. Gibbons, I.; Skold, C . ; Rowley, G. L.: Ullman, E. F. Anal. Eiochem. 1980, 102, 167-70. Armenta, R.; Tarn0wski.T.: Gibbons, I.; Ullman, E. F. Anal. Eiochem. 1985, 146, 211-9. Chen, R.; Weng,L.;Sizto,N.C.:Osorio,B.;Hsu,C.J.:Rodgers,R.: Litman. D.J. Clin. Chem. 1984, 30, 1446-51 Schray, K. J.; Niedbala, R. S . Anal. Chem. 1988, 60, 353-6. Ngo, T. T.: Bovaird, J. H.; Lenhoff, H. M. Appl. Eiochem. Eiotechnol. 1985, 11, 63-70. Heineman. W. R.; Halsall. H. B. Anal. Chem. 1985, 57, 1321A-1331A. Ngo, T. T., Ed. EleclrochemicalSensors in Immunological Analysis:Plenum Press: New York, 1987. Brozles, C. A,: Rechnitz, G. A. Anal. Chem. 1986, 58. 1241-5.
1370
Analytical Chemistry, Vol. 66,No. 9,May I, 1994
advantage of being insensitive to the color or turbidity of the test sample, and this would be important in ultimately developing a method that could function effectively in whole blood samples. Heineman et al.22-24have used electrochemical detection to monitor therapeutic drugs in clinical samples (serum) by adapting commercial homogeneous enzyme immunoassay EMIT reagents to a flow injection analysis measurement arrangement or HPLC system (measuring current resulting from oxidation of reduced nicotinamide adenine dinucleotide (NADH) or reduced dichloroindophenol (DCIPH?)). Brown and MeyerhofP5 demonstrated the feasibility of a nonseparation immunoassay using the principles of dual enzyme channeling in conjunction with an ammonium selective membrane electrode detection of the channeled enzyme activities. However, most of the many other reports regarding the use of electrochemical detection to devise enzyme immunoassays or immunosensors have relied on using such sensors as solid phases in more classical heterogeneous assay arrangements in which antibodies are immobilized at the surface of given electrodes, but after incubation steps with sample and reagents, the surface of the electrodes have been washed before adding substrate to measure bound enzyme a c t i ~ i t y . ~ ~Others . ’ ~ have performed more conventionai heterogeneous sandwich assays by physically separating the immobilized capture antibodies (in microwells or immunocolumns, etc.) from the electrochemical detection of the bound enzyme-Ab conjugate a ~ t i v i t y . ~For ~ -example, ~~ Heineman and co-workers pioneered the use of alkaline phosphataseantibody conjugates to perform sandwich immunoasays in which aminophenyl phosphate is used as a substrate (in place of nitrophenyi phosphate), and the aminophenol product is detected anodically with an FIA ~ y s t e m ~ ’ -equipped *~ with a carbon electrode detector. However, despite all of the past and current research activity in this area, there is still no single electrochemical enzyme immunoassay approach that enables the detection of proteins in a sample as complex as whole blood without the need to perform multiple washing steps and other manipulative procedures. The purpose of this report is to introduce a new approach toward achieving the goal of a nonseparation enzyme immunoassay for proteins that could function in samples as complex as whole blood. It is clear that the sandwich-type assay is superior to other types of solid-phase immunoassays with respect to sensitivity, specificity, and kinetics. These advantages arise from the fact that excess of the capture antibody and the enzyme-antibody conjugate are usually employed in the reaction mixture (Le., to drive the equilibrium toward formation of a sandwich even in the presence of low levels of analyte protein). Unfortunately, it is the excess of enzyme-antibody conjugate that makes it difficult to conceive an arrangement that would enable discrimination between ~
~~
(21) Gleria, K. D.; Hill, H. A. 0.;McNeil, C. J. Anal. Chem. 1986, 58, 1203-5. (22) Eggers, H. M.;Halsall, H . B.: Heineman, W . R. Clin. Chem. 1982,28,184851. (23) Wright, D. S . ; Halsall, H. B.; Heineman, W. R. Anal. Chem. 1986,58,29958. (24) Tang, H.; Halsall, H. B.; Heineman, W. R. Clin. Chem. 1991, 37, 245-8. (25) Brown, D.; Meyerhoff, M. E. Eiosens. Eioelecfron. 1991, 6, 615-22. (26) Uditha de Alwis, W.: Wilson, G. S . Anal. Chem. 1985, 57, 2754-6. (27) Xu, Y.; Halsall. H. B.;Heineman. W. R. J. Pharm. Eiomed. Anal. 1989, 7 , 1301-1 1. (28) Gil, E. P.; Tang, H. T.: Halsall, W . R.: Heineman, W. R.: Misiego, A. S. Clin. Chem. 1990, 36, 6625. (29) Xu. Y., Halsall, H. B.; Heineman, W . R. Clin. Chem. 1990, 36, 1 9 4 1 4 .
small amounts of immunobound enzyme label and high background levels of the enzyme-labeled antibody conjugate. It would appear that the key to the development of a successful separation-free sandwich-type EIA lies in the way in which the substrate is introduced to the assay mixture. If the substrate could be delivered only to the enzyme that is bound to the solid phase via the immunological reaction (the sandwich), but not to the excess free enzymeantibody conjugate in the bulk solution, there would be no need to separate the unbound label from the bound. Further, if the product of the enzyme reaction can be detected at the surface of the solid phase immediately after it is formed, the analytical signal originating at the surface of the solid phase relative to the signal originating from the bulk solution (background signal) could be greatly enhanced. Such a principle has already been applied successfully in the area of fluoroimmunoassay, where evanescent wave sensing has been used to spatially resolve fluorescently labeled antibodies bound to the surface of an optical wave guide from excess fluorophore-labeled antibody in the bulk sample pha~e.~OJHowever, due to the innate amplification properties of enzymes, it would seem that use of enzyme rather than fluorophore labels would offer the potential for better detection limits if a convenient method to spatially resolve bound enzyme from the bulksolution could be implemented. In the approach described herein, one side of a microporous membrane is coated with a thin layer of gold (-600 A). This membrane serves as both the solid phase for the sandwich assay and an amperometric electrochemical detector to measure surface-bound enzyme-labeled antibody. Capture antibodies toward a model protein analyte, hCG, are immobilized to the gold-coated membrane via a monolayer of thioctic acid preadsorbed on the gold surface. Such selfassembled layers of thioalkyl derivatives are emerging as a very convenient way to modify gold electrode^,^^ including for the purpose of immobilizing b i ~ r e a g e n t s .As ~ ~illustrated in Figure 1, samples containing analyte protein together with an alkaline phosphatase-anti-hCG conjugate are incubated concurrently with the capture Ab on the sample side of the membrane. After a fixed period of incubation, the sandwich is formed, and surface-bound enzyme-labeled antibody is detected by adding aminophenyl phosphate substrate to the back side of the membrane to initiate the enzymatic reaction. As soon as the substrate diffuses through the membrane, it reacts first with any enzyme near the gold surface (immunobound sandwich) and is converted to product, aminophenol. Since the solid phase on which the sandwich is formed is also the electrochemical detector (see Figure 2), the aminophenol formed can be measured rapidly as it is produced (by oxidation at +0.19 V vs Ag/AgCl, similar to that suggested by H e i ~ ~ e m a).n ~During ~ the short period after addition of substrate from the back side of the membrane, very little substrate diffuses into the bulk solution on the sample side of the membrane; hence, only a relatively small signal (product) (30) Bluestein, B. I.; Craig, M.; Slovacck, R.; Stundtner, L.; Urciuoli, C.; Walczak, I.; Luderer, A. Biosenrors Fiberoptics: Humana: Clifton, NJ, 1991; pp 181223. (31) Bluestein, B. I.; Chen, S.Y. Immunol. Ser. 1990, 53, 145-70. (32) Cheng, Q.;Brajter-Toth, A. Anal. Chem. 1992, 64, 1998-2000. (33) Nakano, K.; Taka, H.; Maeda, M.; Takagi, M. Anal. Sei. 1993, 9, 133-6. (34) Tang, H.; Liemte, C.; Halsall, H.; Heineman, W. R. Anal. Chim. Acra 1988, 214, 187-95.
4
add substrate to back side
to initiate enzymatic reactio
I
B-c
D
product Substrate 0
0
0
0
0
'Q
0
0
0
0
0
0
0
'0' O Q 0
0
' {-e-
Flgurr l.Prlncipleandsequence of steps forthe proposedseparationfree electrochemical sandwich enzyme Immunoassay.
originates from the large excess of conjugate that is present in this solution. During this initial reaction period, the product of immunobound enzyme is concentrated in a small volume adjacent to the surface of the gold electrode. It is estimated that the product concentration in that small volume element immediately adjacent to the gold electrode (i.e., diffusionlayer) may be 2-3 orders higher than in the bulk solution. As will be shown in the case of an assay for hCG, the proposed method allows for the accurate measurement of surface-bound alkaline phosphatase-Ab conjugate without washing away unbound conjugate. This makes it possible to perform a separationfree sandwich-type assay in both buffer and whole blood with relatively short assay times and detection limits comparable to existing heterogeneous EIA methods for hCG.
EXPERIMENTAL SECTION Apparatus. Amperometric measurements were performed with a BAS CV-27 potentiostat in a three-electrode mode at 0.19 V vs Ag/AgCl reference electrode, with a large platinum wire as the auxiliary electrode. The arrangement of the diffusion cell is illustrated in Figure 2A. Data were recorded on a Fisher series 5000 chart recorder. Gold coating of the nylon membranes was accomplished with a Denton Vacuum Desk- 11cold sputter-etch unit. Scanning electron microscopy of the resulting membranes was carried out on a Hitachi S-570 instrument (under Grant BSR-83-14092 from NSF). Analytical Chemisw, Vol. 66, No. 9, May 1, 1994
1371
I
0 n
Goldcoating
U
I
a \
I
II
0 -
0
n U
0 0
Microporous Membrane
C2H5-N=C=N-(CH2)3N(CH3)2 + m(CH2)4C00H 5
gf
03) Figure 2. (A) Diagramof diffusioncellarrangementin which micoporous gold electrode is mounted for nonseparation EIA: (a) addition of substrate to initiate enzymatic reaction: (b) buffer or whole blood in whichanalyteproteinand ALP-Ab conjugate are incubatedwith capture Ab immobilized on the gold surface; (m) gold-coated microporous membrane. (6) Expanded view of microporous membrane configuration: (c) gold coating for electronic connection via clip or solder: (d) microporousnylon membraneof 47-mm outer diameter: (e) disk-shaped gold coating with Ab immobilized as working electrode, 6-mm outer diameter: (f) PVC coating around the gold disk, 30-mm 0.d; (9) open hole between diffusion cells, 9-mm outer diameter; (h) gold coating outlet, 2-mm width.
Reagents. Anti-hCG monoclonal capture antibody and an alkaline phosphatase-anti-hCG antibody conjugate (ALPAb) were gifts from Hybritech Inc. (La Jolla, CA). Human chorionic gonadotropin, heat shock bovine serum albumin, and 4-nitrophenyl phosphate were purchased from Sigma (Saint Louis, MO). (D,L)-thioctic acid and 4-aminophenol were from Aldrich (Milwaukee, WI). A BCA protein assay kit and 1-ethyl-3- [3-(dimethy1amino)propyll carbodiimide (EDC) were from Pierce (Rockford, IL). The enzyme substrate used in this work, aminophenyl phosphate (APP), was synthesized by the reduction of 4-nitrophenyl phosphate according to a previously reported p r o c e d ~ r e . ~The ~ microporous nylon membranes, Nylaflo, with pore size of 0.2 pm, were obtained from Gelman Science (Ann Arbor, MI). Human blood was regenerated by mixing 40% red blood cells with 60% compatible plasma, both gifts from the University of Michigan Hospital blood bank. All other chemicals were reagent grade. Preparationof Microporous Gold Membrane Electrode and Immobilization of Capture Ab. Nylon membranes (circles with outer diameter of 47 mm) were put under a mask which (35) Christie, I.; Treloar, P.; Koochaki, Z.; Vadgama, P. Anal. Chim. Acta 1992, 257, 21-8.
1372 Analytical Chemistry, Vol. 66, No. 9, May 1, 1994
Flgure 3. Sequence of chemical steps used to immobilize anti-hCG antibody on microporous gold electrode.
had a 6-mm hole at the center and a 2-mm-wide outlet strip (for electrical connection purposes). The membranes were coated with gold for 200 s. This yielded a disk shaped gold electrodeat the center of the membrane with an outer diameter of 6 mm and a thickness of -600 A. The gold-coated membrane was put into 2% (w/w) thioctic acid in absolute ethanol for 24 h with shaking. The membranes were then rinsed with ethanol twice and dried. They were then immersed into 1 % (w/w) EDC in anhydrousacetonitrile for 5 h to activate the free carboxyl groups of the thioctic acid (to form an 0-acylurea intermediate). After the membranes were rinsed twice with acetonitrile, a silver wire was fixed onto the outer edge of the narrow (2 mm) gold lead with silver epoxy for electrical connection to the potentiostat (see Figure 2B). A layer of PVC (33% PVC and 67% bis(2-ethylhexyl) sebacate (all w/w %), dissolved in T H F (1:6 w/v), was cast around the center disk electrode, including over the narrow gold lead outlet. This left the 6-mm disk-shaped gold electrode untouched in the center. Monoclonal anti-hCG Ab (30 pL, 2 mg/mL, in 0.1 M borate buffer, pH 8.75) was dropped on the exposed microporous gold for immobilization (see Figure 3 for summary of Ab immobilization chemistry). The membranes were then put into refrigerator for further reaction and storage. After 24 h, the immobilization was complete and the membrane was soaked in 0.1 M borate buffer, pH 8.75, containing 5% (v/v) ethanolamineto block any unreacted but still active 0-acylurea intermediates. The membrane was then rinsed with water and mounted in the diffusion cell to perform nonseparation sandwich assays.
Characterization of the Porous Gold Electrodes and Responsetort-Aminophenol. Figure 2B illustrates an expanded view of the final microporous gold electrode/membrane. After antibody immobilization, the membrane was mounted between the diffusion cells to test its electrochemical response to 4-aminophenol, the product of the alkaline phosphatase reaction when aminophenyl phosphate was used as the substrate. The sample side of the diffusion cell (in contact with the gold electrode) was filled with 2.0 mL of 0.1 M carbonate buffer, pH 10.0. A potential of 0.19 V vs Ag/ AgCl was applied to the gold coating. 4-Aminophenol was dissolved in ethanol, and varying aliquots of this solution were added into thestirred buffer solution. The steady-statecurrent observed following each change in product concentration was recorded. For comparison purposes, the amperometric responses to 4-aminophenol of microporous gold electrodes that were not treated with antibody or thioctic acid were also evaluated to determine how such surface treatment affects the relative electrochemical activity of the gold electrode. Kinetic Study of the Sandwich-Type Immunobinding. The kinetics of the immunobinding of hCG and the ALP-Ab conjugate with capture anti-hCG antibody immobilized on the surface of the microporous gold electrode were studied in both buffer solution and whole blood. A 35 units/L aliquot of hCG (1 unit corresponds to -200 ng of protein) was incubated together with 0.25 mA (milliabsorbance) of ALPAb (note: 0.25 mA is equivalent to 392 units/L ALP activity as determined colorimetrically in 1 M carbonate buffer, pH 10, using p-nitrophenyl phosphate as substrate) in 2 mL of 0.01 M, citrate buffer, pH 7.4, or human blood for different periods of time with mixing before the stirring was stopped and substrate solution was added to the chamber on the back side of the membrane (see below). Separation-Free EIAs. To enhance the spatial resolution of surface-bound ALP-Ab from the bulk unbound reagent, the final separation-free assay was performed using buffers of different pH on the two sides of the membrane: 0.01 M citrate buffer, pH 7.4, with 1 mM MgC12 and 1% heat shock BSA was used on the sample side of the cell, while 2 mL of a 26.5 mM solution of 4-aminophenyl phosphate prepared in 1.O M carbonate buffer (pH 10.0) was added to the back side of the membrane to initiate the measurement of suface-bound ALP-Ab. To obtain an hCG calibration curve, different doses of hCG together with ALP-Ab were added into the sample side cell containing the citrate buffer to make the final volume of liquid on the sample side 2.0 mL, and the final concentration of ALP-Ab conjugate equal to 0.25 mA/2 mL. This sample mixture was incubated in the sample side cell for 27 min with stirring. The stirring was then stopped in the sample solution, and a potential of 0.19 V vs Ag/AgCl was applied to the gold electrode. After another 3 min (to attain a steady-state baseline current), 2 mL of substrate solution was added to the chamber on the opposite side of the membrane. A steadystatecurrent was achieved usually within -50 s for all assays (the actual length of time depends on the hCG concentration; the more hCG present, the faster the response). The current/ time course was monitored on a strip-chart recorder. The current change from baseline to the steady state was recorded vs different doses of hCG and plotted to obtain a calibration curve for hCG. In the case of assays in whole blood, human
Flgure 4. Scanning electron micrograph of the gold-coated side of the microporous nylon membrane.
blood spiked with varying levels of hCG was incubated together with the ALP-Ab conjugate for 27 min in the chamber on the sample side of the membrane. Again, the final volume of the blood was 2.0 mL in which the human blood accounted for no less than 93% (v/v) of the total volume (reagents or hCG standards
