Reduction of the Nonspecific Binding of a Target Antibody and of Its

Anal. Chem. 2005, 77, 7758-7762

Reduction of the Nonspecific Binding of a Target Antibody and of Its Enzyme-Labeled Detection Probe Enabling Electrochemical Immunoassay of an Antibody through the 7 pg/mL-100 ng/mL (40 fM-400 pM) Range Yongchao Zhang and Adam Heller*

Department of Chemical Engineering and Texas Materials Institute, The University of Texas at Austin, Austin, Texas 78712

We describe a simple, potentially low-cost, amperometric, enzyme-amplified, sandwich-type immunoassay, monitoring IgG at a concentration as low as ∼7 pg/mL with a dynamic range of 104. The assay utilizes a screen-printed carbon electrode on which a redox hydrogel and avidin are co-electrodeposited. To neutralize nonspecifically binding positively charged microdomains of the avidin, two polyanions, poly(acrylic acid-co-maleic acid) and poly(acrylic acid), are applied. These polyanions bind to the film not only electrostatically but also by Michael addition reaction to cysteine, lysine, or arginine functions of the avidin. The electrode is then made specific for the analyte, for which rabbit IgG was chosen, by conjugating the filmbound avidin to biotin-labeled anti-rabbit IgG. After exposure to the tested solution and capture of rabbit IgG, the sandwich is completed by conjugation of horseradishperoxidase (HRP)-labeled anti-rabbit IgG. Electrical contact between the HRP and the electrode-bound hydrogel results in the formation of an electrocatalyst for the electroreduction of H2O2 to water. The application of the poly(acrylic acid-co-maleic acid) and the poly(acrylic acid) reduces the nonspecific adsorption-associated noise, lowers the detection limit from 3 ng/mL (∼20 pM analyte antibody concentration) to ∼7 pg/mL (∼40 fM analyte antibody concentration), and also expands the dynamic range to 104. Electrochemical biosensors have several advantages over their photonic counterparts. Unlike the photonic biosensors, they do not require the conversion of electrical power to light and the subsequent reconversion of light to electrical power. Also, their instrumentation is simpler and their cost is correspondingly lower. Hence, the overwhelming majority of the biosensors in actual use, such as those used in glucose monitoring by diabetic people, are no longer photonic as they once were, but electrochemical. Additionally, when the sensing is electrochemical, the assayed samples can be unseparated turbid, or light-absorbing, liquids. * To whom correspondence [email protected].

should

be

addressed.

E-mail:

7758 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Highly sensitive and specific electrochemical nucleic acid sensors have been reported.1-7 Sandwich-type, enzyme-amplified amperometric nucleic acid sensors detect as few as thousands, possibly hundreds, of copies of the acids.8,9 In these assays, a capture sequence is co-electrodeposited with a redox hydrogel on an electrode, the target DNA molecule is captured, and the captured analyte nucleic acid is then enzyme-labeled, by conjugating a detection sequence, which is labeled with a redox enzyme, often horseradish peroxidase (HRP), or bilirubin oxidase. Formation of an electrical contact between the redox enzyme and the redox hydrogel makes the film on the electrode catalytic for the electroreduction of H2O2 or O2 to water. Amperometric enzymeamplified sandwich-type immunoassays, based on the same principle, have also been described.10-18 Unlike in their DNA detecting counterparts, where the background or noise current, caused by nonspecific adsorption of the enzyme-labeled detection sequence, was not significant, the nonspecific adsorption of the enzyme-labeled antibodies posed a severe problem and set the detection limit at a much higher level than that defined by the (1) Palecek, E.; Fojta, M.; Tomschik, M.; Wang, J. Biosens. Bioelectron. 1998, 13, 621-628. (2) Wang, J. Chem.-Eur. J. 1999, 5, 1681-1685. (3) Takenaka, S. Bull. Chem. Soc. Jpn. 2001, 74, 217-224. (4) Fojta, M. Electroanalysis 2002, 14, 1449-1463. (5) Vercoutere, W.; Akeson, M. Curr. Opin. Chem. Biol. 2002, 6, 816-822. (6) Drummond, T. G.; Hill, M. G.; Barton, J. K. Nat. Biotechnol. 2003, 21, 11921199. (7) Lucarelli, F.; Marrazza, G.; Turner, A. P. F.; Mascini, M. Biosens. Bioelectron. 2004, 19, 515-530. (8) Zhang, Y.; Kim, H.-H.; Heller, A. Anal. Chem. 2003, 75, 3267-3269. (9) Zhang, Y.; Pothukuchy, A.; Shin, W.; Kim, Y.; Heller, A. Anal. Chem. 2004, 76, 4093-4097. (10) Athey, D.; McNeil, C. J. J. Immunol. Methods 1994, 176, 153-162. (11) Ho, W. O.; Athey, D.; McNeil, C. J. Biosens. Bioelectron. 1995, 10, 683691. (12) Pemberton, R. M.; Hart, J. P.; Foulkes, J. A. Electrochim. Acta 1998, 43, 3567-3574. (13) Campbell, C. N.; de Lumley-Woodyear, T.; Heller, A. Fresenius’ J. Anal. Chem. 1999, 364, 165-169. (14) Ronkainen-Matsuno, N. J.; Thomas, J. H.; Halsall, H. B.; Heineman, W. R. TrAC, Trends Anal. Chem. 2002, 21, 213-225. (15) Lei, C.-X.; Wu, J.; Wang, H.; Shen, G.-L.; Yu, R.-Q. Talanta 2004, 63, 469474. (16) Wilson, M. S.; Rauh, R. D. Biosens. Bioelectron. 2004, 20, 276-283. (17) Wilson, M. S.; Rauh, R. D. Biosens. Bioelectron. 2004, 19, 693-699. (18) Thomas, J. H.; Kim, S. K.; Hesketh, P. J.; Halsall, H. B.; Heineman, W. R. Anal. Chem. 2004, 76, 2700-2707. 10.1021/ac051218c CCC: $30.25

© 2005 American Chemical Society Published on Web 10/22/2005

two equilibrium constants for the formation of the conjugates.13,19 Most, if not all, researchers of immunoassays historically sought to reduce the nonspecific adsorption.20,21 The detectable signals, whether currents or photon fluxes, were usually much smaller than the nonspecific adsorption associated noise, and companies reducing the noise, for example by application of electrogenerated chemiluminescence, were uniquely successful.22 Here we describe a simple, yet exceptionally effective, method of suppressing the nonspecific adsorption of an immunoreagent on a surface comprising polycationic domains, their treatment with a polyanion. Two polyanionic polymers, poly(acrylic acid-co-maleic acid) and poly(acrylic acid), are used separately to treat the polycationic redox polymer-modified electrodes. Polyanions obviously bind electrostatically to the polycationic redox polymer film. In addition, unlike other polyanions, they can also bind covalently to cysteine, lysine, and arginine residues of avidin through the Michael addition reaction, which proceeds at room temperature and at neutral pH, if their polymerization leaves in some of the macromolecules residual acrylate or maleate functions with reactive double bonds. The reactive polyanions, particularly poly(acrylic acid), reduce 50-fold the ultimate nonspecific adsorption of enzyme-labeled antibody, lowering thereby the detection limit to 40 fM (or 0.4 amol) and also raising the dynamic range to as much as 104. EXPERIMENTAL SECTION Materials and Reagents. ChromPure rabbit IgG (Catalog No. 011-000-003), biotin-SP-conjugated AffiniPure F(ab′)2 fragment of goat anti-rabbit IgG, Fc specific (Catalog No. 111-066-008), and HRP-conjugated F(ab′)2 fragment of goat anti-rabbit IgG, F(ab′)2 specific (Catalog No. 111-036-006) were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). ImmunoPure avidin (Catalog No. 21128) was purchased from Pierce (Rochford, IL). Poly(acrylic acid) (Catalog No. 03326, MW ∼90 000, 25% aqueous solution) was from Polysciences, Inc. (Warrington, PA). Poly(acrylic acid-co-maleic acid) (lot no. 05805CS, 50 wt % solution in water, average Mw 3000) was from Aldrich (Milwaukee, WI). The buffering salts and other chemicals were purchased from Sigma (St. Louis, MO), Aldrich (Milwaukee, WI), or Fisher Scientific (Pittsburgh, PA) and were used as received. The phosphate-buffered saline solution (PBS: 4.3 mM NaH2PO4, 15.1 mM Na2HPO4, 140 mM NaCl, pH 7.2) was prepared using deionized water (Barnstead, Nanopure II, Van Nuys, CA). The electron-conducting redox polymer, polyacrylamide (PAA)-poly(4-vinylpyridine) (PVP)-[Os(bpy)2Cl]+/2+ (Figure 1), a copolymer of PAA and PVP complexed with [Os(2,2′-bipyridine)2Cl2]2+/3+, was synthesized as previously described.8,23 Instruments and Electrodes. Electrochemical measurements were carried out in a Faraday cage with a CH Instruments (Austin, TX) model 832A electrochemical detector, interfaced to a com(19) Qu, Y.; Berghman, L. R.; Vandesande, F. Anal. Biochem. 1998, 259, 167175. (20) Fleminger, G.; Solomon, B.; Wolf, T.; Hadas, E. J. Chromatogr., A 1990, 510, 271-279. (21) Piehler, J.; Brecht, A.; Geckeler, K. E.; Gauglitz, G. Biosens. Bioelectron. 1996, 11, 579-590. (22) Blackburn, G.; Shah, H.; Kenten, J.; Leland, J.; Kamin, R.; Link, J.; Peterman, J.; Powell, M.; Shah, A.; Talley, D.; Tyagi, S. K.; Wilkins, E.; Wu, T.-G.; Massey, R. J. Clin. Chem. 1991, 37, 1534-1539. (23) Zhang, Y.; Kim, H.-H.; Mano, N.; Dequaire, M.; Heller, A. Anal. Bioanal. Chem. 2002, 374, 1050-1055.

Figure 1. Composition of the redox copolymer PAA-PVP-Os. The mers are randomly distributed.

puter (Dell Dimension 8100, Austin, TX). The working electrodes were the disposable 1.0-mm-diameter screen-printed carbon disks (SPE) described previously.23,24 The carbon dots were printed on a flexible polyester film with hydrophilic carbon ink (Electrodag 423SS from Acheson, Port Huron, MI). To avoid the spreading of the 5-20-µL droplets beyond the 1.0-mm-diameter working electrodes, a hydrophobic circle was drawn around each SPE with a felt-tip pen containing hydrophobic ink (Dako Pen, S 2002, Dako Corp., Carpinteria, CA). The measurements were performed at room temperature using a three-electrode electrochemical cell formed by placing a droplet (5-20 µL) of the solutions on the surface of the screen-printed electrode. A 0.5-mm-diameter platinum wire counter electrode and a miniature Ag/AgCl reference electrode (Catalog No. EE008), purchased from Cypress Systems (Lawrence, KS), were used. A Digi-Block Jr, Laboratory Devices block heater, purchased from Aldrich (Milwaukee, WI), was used to perform conjugations at elevated temperatures. Electrodeposition of the Redox Polymer. The redox polymer films were electrodeposited from solutions containing 1 mg/ mL PAA-PVP-Os and 18% (v/v) phosphate buffer. This solution (20 µL) was pipetted onto the SPE, and a potential of -1.0 V (vs Ag/AgCl) was applied for 2 min. The electrode was then rinsed with deionized water and scanned between 0.1 and 0.6 V to confirm the deposition. Attachment of Avidin to the Redox Polymer Film. A drop of 20 µL of phosphate buffer solution containing 1 mg/mL avidin was placed on the electrode on which PAA-PVP-Os was previously electrodeposited. A potential of -1.0 V (vs Ag/AgCl) was applied for 5 min; the electrode was then rinsed with deionized water and refrigerated at 4 °C until use. Application of the Nonspecific Adsorption Preventing Polyanion. Part of the SPEs on which the redox polymer and avidin were electrodeposited were covered with a 10-µL drop of poly(acrylic acid-co-maleic acid) solution (6.5 mg/mL) or poly(acrylic acid) solution (6.5 mg/mL) for 10 min. The electrodes were then rinsed with deionized water and with phosphate buffer solution. Anchoring of the Antigen-Capturing Antibody. The surface of the SPE, containing PAA-PVP-Os and avidin, either with or without the polyanion, was covered with a 10-µL 10 µg/mL PBS droplet of biotin-labeled goat anti-rabbit IgG for 20 min at 40 °C. (24) Dequaire, M.; Heller, A. Anal. Chem. 2002, 74, 4370-4377.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7759

Figure 2. Fabrication of the rabbit IgG-sensing electrode and scheme of the immunoassay. (a) Avidin is coimmobilized by electrodeposition in the redox hydrogel; (b) biotin-labeled anti-rabbit IgG is bound to the hydrogel through avidin-biotin conjugation; (c) the analyte, rabbit IgG, is captured by the hydrogel-bound anti-rabbit IgG, conjugates HRP-labeled anti-rabbit IgG. Electronic contact between the HRP and the redox hydrogel makes the film an electrocatalyst for the reduction of H2O2 to water and causes the flow of a reduction current.

The electrode was rinsed with deionized water and with phosphate buffer and then kept refrigerated at 4 °C until use. Storage at 4 °C for two or more months did not alter the shape, position, or height of the voltammetric peaks. Immunoassays and Measurements. The assays were performed at 40 °C. A 10-µL aliquot of buffer solution, containing the desired amount of the antigen, rabbit IgG, was placed for 20 min on the SPE on which PAA-PVP-OS and avidin/biotin/antirabbit IgG were immobilized by cross-linking and conjugation and then rinsed with deionized water and phosphate buffer. Next, a 10-µL droplet of the buffer solution containing HRP-labeled antirabbit IgG (8 µg/mL) solution was applied for 20 min and the now-ready electrode was rinsed with phosphate buffer. The current-time measurements were performed at room temperature, with 20 µL of PBS on the SPE poised at 0.2 V (vs Ag/AgCl). After ∼100 s, the current stabilized and 2 µL of 10 mM H2O2 was added. RESULTS AND DISCUSSION Preparation of the SPE Electrodes. The SPE electrodes were made antigen (rabbit IgG)-sensing by immobilizing the rabbit IgG-specific goat anti-rabbit IgG to a redox hydrogel-covered SPE (Figure 2). The electrodeposition of redox polymers by their reductive cross-linking is simple, fast, and reproducible.8,23-25 (25) Gao, Z.; Binyamin, G.; Kim, H.-H.; Barton, S. C.; Zhang, Y.; Heller, A. Angew. Chem., Int. Ed. 2002, 41, 810-813.

7760 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

When polymer-bound [Os(bpy)2Cl]+/2+ redox centers, the formal redox potential of which is 0.3 V versus Ag/AgCl, were electroreduced to [Os(bpy)2Cl]+ upon applying a potential of -1.0 V versus Ag/AgCl, the reduced Coulombic binding energy made the innersphere chlorides labile. As a result, Cl- was exchanged by nitrogen of neighboring pyridines, and the polymer was cross-linked. The cross-linking made the polymer insoluble, which then deposited spontaneously on the electrode surface. Conduction of electrons by the redox polymers requires their swelling to hydrogels, where redox centers bound to mobile, flexible polymer segments collide and exchange electrons. Thus, electrodeposition and the formation of the redox hydrogel are readily confirmed either by integrating, at a slow scan rate, the areas of their voltammograms or by measuring the heights of the peaks for the electroreduction/ oxidation of the [Os(bpy)2Cl]+/2+ centers (Figure 3). Earlier, single-stranded DNA modified at one end with an amine was co-electrodeposited8,9,23,24 by exchanging the labile Clof [Os(bpy)2Cl]+ with the amine. Now avidin was co-electrodeposited apparently by exchanging the labile Cl- with histidine, lysine, or arginine residues.25 Incorporation of avidin provided a generic platform electrode, which was made selective for a particular analyte by conjugating its biotinylated antibody. In our case, the analyte rabbit IgG, and biotin-labeled anti-rabbit IgG was conjugated to the electrodeposited avidin. Suppressing Nonspecific Protein Adsorption. In the assay, rabbit IgG, conjugated with the immobilized anti-rabbit IgG on

Figure 3. Electron transferring steps of the HRP-detecting reaction.

Figure 4. Suppression of the nonspecific binding-caused noise current by polyanions having terminal functions forming covalent bonds with amines and with thiols by Michael addition reactions in aqueous solutions at ambient temperature and at neutral pH.

the electrode, and then the rabbit IgG was conjugated with the detection agent, HRP-labeled anti-rabbit IgG. The conjugation brought the labeling HRP into electrical contact with the electrodebound redox hydrogel, making the film a catalyst for the electroreduction of H2O2 to water; the concentration of the antigen was related to the increment in the H2O2 electroreduction current. Unlike the case of nucleic acid monitoring, where the nonspecific adsorption of the HRP-labeled nucleotide was insignificant,8,23 either the antibody or the HRP adsorbed nonspecifically on the avidin-redox polymer film13 and the nonspecific binding raised the detection limit and compressed the dynamic range. Figure 4 shows how substantial the nonspecific adsorption was. In the absence of rabbit-IgG, and despite multiple washings, the H2O2 reduction current was as high as 2.5 × 10-7 A (3.2 × 10-5 A/cm2). Apparently, irrespective of the net surface charge on the HRPlabeled IgG, the HRP-labeled IgG does have polyanionic microdomains, sufficient for its binding to polycationic domains in the redox polymer, containing both avidin and anti-rabbit IgG. The electrostatic interaction of these micro-domains is remarkably neutralized by poly(acrylic acid) (PAA) and also by poly(acrylic acid-co-maleic acid) (PAMAc). The two polyanions are unique in having terminal double bonds, reacting, at ambient temperature and in neutral pH aqueous solutions, by the Michael addition reaction to form covalent bonds with cysteine thiols and with lysine and arginine amines.26 Thus, even isolated surface amines are

Figure 5. Dependence of the H2O2 electroreduction current on the antigen (rabbit IgG) concentration. The 1-mm-diameter screen-printed carbon electrode with PAA-PVP-Os and avidin, treated with poly(acrylic acid), conjugated with biotin-labeled anti-rabbit IgG; 20-min incubation with the 10-µL samples and then 20 min with 10 µL of buffer with 8 µg/mL HRP-labeled anti-rabbit IgG. Electrodes poised at 0.2 V (Ag/AgCl), 1 mM H2O2.

modified, becoming polyanion functions. Overall, the covalent attachment of either PAA or PAMAc reduces the nonspecific binding. Figure 4 shows that the binding of PAMAc to the avidincontaining redox hydrogel, merely by placing the 10-µL poly(acrylic acid-co-maleic acid) solution (6.5 mg/mL) on the electrode and allowing it to react for 10 min, lowers the nonspecific adsorption-caused noise current 4-fold to 7.0 × 10-8 A, corresponding to a current density of 8.9 × 10-6 A/cm2. Treatment with PAA was even more effective. It decreased the noise current caused by the nonspecific adsorption to 5.0 × 10-9 A, corresponding to a 6.4 × 10-7 A/cm2 current density, a 50-fold improvement. Branched versus Linear Polyanions. Both poly(acrylic acid) and poly(acrylic acid-co-maleic acid), are made by free-radical polymerization. If the chains were exclusively linear and their polymerization was ideal, one of the radicals formed of the initiator would react at one chain end and the second at the other end. In this case, the termination reaction would not leave a double bond at the chain’s end and the Michael addition reaction could not take place. The radical polymerization reaction is, however, more complex, the radicals reacting not only in the initiation, propagation, and termination of the chains. They also disproportionate, by one radical abstracting a hydrogen atom from another, so that a saturated end group and an olefinic end group, which can participate in the Michael addition reaction, are formed. The radicals also abstract hydrogen atoms from chains, causing their branching. Chain branching further increases the likelihood of disproportionation, by two radicals at branch ends forming a saturated function and a Michael addition-undergoing olefin. Immunoassays. Figure 5 shows the dependence of the H2O2 electroreduction current on the concentration of the analyte, rabbit IgG. The current increased, with the rabbit IgG concentration through the 1 pg/mL-10 ng/mL range, a range of 4 orders of magnitude. Rabbit IgG was detected at a concentration as low as ∼7 pg/mL, or ∼40 fM, ∼450 times lower than in our previously (26) Hubbell, J. A.; Texto, M.; Elbert, D. L.; Finken, S.; Hofer, R.; Spencer, N. D.; Ruiz-Taylor, L. PCT Int. Appl. (Eidgenossische Technische Hochschule Zurich, Universitat Zurich). U.S. patent 6884628, 2000.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7761

reported amperometric sandwich-type immunoassay,13 utilizing the same immunoreagents. The detectable amount of rabbit IgG in the analyzed 10-µL sample was ∼0.4 amol (4.0 × 10-19 mol, or ∼240 000 molecules), comparable to the most sensitive immunoassays reported.27-33 The time required for the assay was ∼40 min. We project that optimization is likely to reduce the time to