Development of a “Membrane Cloaking” Method for Amperometric

Anal. Chem. 2007, 79, 899-907

Development of a “Membrane Cloaking” Method for Amperometric Enzyme Immunoassay and Surface Plasmon Resonance Analysis of Proteins in Serum Samples K. Scott Phillips, Jong Ho Han, and Quan Cheng*

Department of Chemistry, University of California, Riverside, California 92521

Detection of trace amounts of target proteins in the presence of high concentrations of matrix proteins (e.g., serum samples) without separation steps is of great significance to biomedical research but remains technically challenging. Here we report a “membrane cloaking” method to overcome nonspecific protein adsorption and fouling problems for label-free surface plasmon resonance detection and heterogeneous immunosensing. A thin, hybrid, self-assembled monolayer on gold was formed with 70 mol % mercaptopropanol and 30 mol % cysteamine/propanedithiol to facilitate membrane fusion and covalent attachment of antibodies. After antibody immobilization, the surface was incubated with lipid vesicles, which fused to form a supported membrane. The analyte spiked in serum was introduced for binding, and the membrane and nonspecifically adsorbed proteins on the membrane were subsequently removed using a nonionic surfactant before the final measurement was carried out. Selection of a suitable surfactant can preserve antibody/ antigen binding and selectively remove the membrane, allowing accurate measurement of the captured proteins without interference from nonspecifically adsorbed species. Surface plasmon resonance (SPR) quantification of IgG spiked in undiluted serum (∼75 mg/mL protein) was achieved with the membrane cloaking method, whereas direct measurement without membrane removal resulted in a significantly large error. The cloaking method was also used to develop an enzyme amplified amperometric assay using HRP-conjugated IgG. Detection of concentrations as low as 5 fM proteins was obtained. Finally, a membrane cloaking assay combining SPR and in situ electrochemical measurement was demonstrated on a gold substrate. Similar sensitivity was observed using a continuous flow injection measurement. The method opens new avenues to develop direct assay methods with ultrahigh sensitivity for protein samples using SPR and enzyme-linked amplification mechanisms. A major challenge in protein biosensing is nonspecific adsorption.1-5 In enzyme-linked immunoassay, signal from non* Corresponding author. Tel: 951-827-2702. E-mail: [email protected]. (1) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164-1167. 10.1021/ac0612426 CCC: $37.00 Published on Web 12/20/2006

© 2007 American Chemical Society

specifically adsorbed enzyme-conjugated antibodies interferes with that from specifically bound ones. This background signal often becomes the limiting factor in detection limits. For nonlabeled techniques such as quartz crystal microbalance and surface plasmon resonance (SPR), nonspecific adsorption is especially critical because many real-world samples contain high levels of matrix proteins.6-9 Without purification, direct measurement is ineffective and prone to large errors for analytes that occur at concentrations 1000 to 1 million times less than that of all other proteins in solution. Specific binding signal is easily swamped by response from a miniscule fraction of matrix molecules that adsorb on the sensor interface. The dominant approach to overcoming these challenges has been prevention of nonspecific adsorption. A common strategy used in an immunoassay is to block the surface with a highly concentrated bovine serum albumin (BSA) or casein solution.10-12 In addition to long incubation times and cumbersome procedures, it is well documented that blocking proteins can reduce binding efficiency and interact with other molecules in biological fluids. Another approach is to use supported membranes. In previous reports, we have demonstrated that supported bilayer membranes (SBMs) are an effective biomimetic route to reduce nonspecific adsorption of proteins on PDMS13,14 and silicate15 surfaces. While (2) Feldman, K.; Hahner, G.; Spencer, N. D.; Harder, P.; Grunze, M. J. Am. Chem. Soc. 1999, 121, 10134-10141. (3) Pasche, S.; Voros, J.; Griesser, H. J.; Spencer, N. D.; Textor, M. J. Phys. Chem. B 2005, 109, 17545-17552. (4) Glidle, A.; Yasukawa, T.; Hadyoon, C. S.; Anicet, N.; Matsue, T.; Nomura, M.; Cooper, J. M. Anal. Chem. 2003, 75, 2559-2570. (5) Ta, T. C.; McDermott, M. T. Anal. Chem. 2000, 72, 2627-2634. (6) Stigter, E. C. A.; de Jong, G. J.; van Bennekom, W. P. Biosens. Bioelectron. 2005, 21, 474-482. (7) Barrett, D. A.; Hartshorne, M. S.; Hussain, M. A.; Shaw, P. N.; Davies, M. C. Anal. Chem. 2001, 73, 5232-5239. (8) Masson, J. F.; Battaglia, T. M.; Davidson, M. J.; Kim, Y. C.; Prakash, A. M. C.; Beaudoin, S.; Booksh, K. S. Talanta 2005, 67, 918-925. (9) Reimhult, E.; Larsson, C.; Kasemo, B.; Hook, F. Anal. Chem. 2004, 76, 7211-7220. (10) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H.; Pannu, N. S.; Read, R. J.; Bundle, D. R. Nature 2000, 403, 669-672. (11) Knecht, B. G.; Strasser, A.; Dietrich, R.; Martlbauer, E.; Niessner, R.; Weller, M. G. Anal. Chem. 2004, 76, 646-654. (12) Benkert, A.; Scheller, F.; Schossler, W.; Hentschel, C.; Micheel, B.; Behrsing, O.; Scharte, G.; Stocklein, W.; Warsinke, A. Anal. Chem. 2000, 72, 916921. (13) Phillips, K. S.; Cheng, Q. Anal. Chem. 2005, 77, 327-334. (14) Phillips, K. S.; Dong, Y.; Carter, D.; Cheng, Q. Anal. Chem. 2005, 77, 29602965.

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007 899

Figure 1. Cartoon illustration showing the membrane cloaking and removal process for detection of proteins in serum samples.

the SBMs are suitable for protein concentrations below 1 mg/ mL, highly concentrated biological fluids often exceed their ability to resist adsorption. Recently, buffer additives have also been demonstrated to reduce nonspecific adsorption to the singlemolecule level.16 These additives are continuously injected through the system in the buffer and sample and have been demonstrated to prevent nonspecific adsorption of avidin and BSA at ∼0.2 mg/ mL concentration. Their effectiveness for highly concentrated protein solutions such as blood serum, however, was not reported. In addition to blocking strategies, considerable work has been carried out in the search for chemical groups and surfaces that are “nonstick”.17-23 No perfect surface exists for this purpose, and poly(ethylene glycol) (OEG or PEG) remains one of the most promising structural groups. For SPR-related work, Corn and coworkers have used PEG as a blocking group for both DNA and protein biosensing in SPR imaging.24 Kyo et al. demonstrated dramatic reduction of nonspecific binding using PEG surfaces for SPR biosensing in 11 mg/mL cell lysate.25 Battalglia et al. used another small molecule, ethanolamine, for fiber-optic SPR biosensing.26 Suppression of nonspecific adsorption was achieved for sensing of cytokines in 4 mg/mL cell culture medium. Thus, the strategy of using blocking groups with PEG or ethanolamine shows great promise for intermediate (1-10 mg/mL) concentrations of proteins in solution. Blood serum samples, on the other hand, are known to contain 60-80 mg/mL protein.27 We are not aware of any methods based on preventing nonspecific adsorption that would be satisfactory for SPR analysis in such highly (15) Phillips, K. S.; Han, J.-H.; Martinez, M.; Wang, Z.; Carter, D.; Cheng, Q. Anal. Chem. 2006, 78, 596-603. (16) Huang, B.; Wu, H. K.; Kim, S.; Zare, R. N. Lab Chip 2005, 5, 1005-1007. (17) Chapman, R. G.; Ostuni, E.; Takayama, S.; Holmlin, R. E.; Yan, L.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 8303-8304. (18) Ostuni, E.; Chapman, R. G.; Holmlin, R. E.; Takayama, S.; Whitesides, G. M. Langmuir 2001, 17, 5605-5620. (19) Kane, R. S.; Deschatelets, P.; Whitesides, G. M. Langmuir 2003, 19, 23882391. (20) Kang, S. H.; Yeung, E. S. Anal. Chem. 2002, 74, 6334-6339. (21) (a) Chen, S.; Liu, L.; Jiang, S. Langmuir 2006, 22, 2418-2421. (b) Nolan, C. M.; Reyes, C. D.; Debord, J. D.; Garcia, A. J.; Lyon, L. A. Biomacromolecules 2005, 6, 2032-2039. (22) Roach, L. S.; Song, H.; Ismagilov, R. F. Anal. Chem. 2005, 77, 785-796. (23) Lasseter, T. L.; Clare, B. H.; Abbott, N. L.; Hamers, R. J. J. Am. Chem. Soc. 2004, 126, 10220-10221. (24) Wegner, G. J.; Wark, A. W.; Lee, H. J.; Codner, E.; Saeki, T.; Fang, S.; Corn, R. M. Anal. Chem. 2004, 76, 5677-5684. (25) Kyo, M.; Usui-Aoki, K.; Koga, H. Anal. Chem. 2005, 77, 7115-7121. (26) Battalglia, T. M.; Masson, J.-F.; Sierks, M. R.; Beaudoin, S. P.; Rogers, J.; Foster, K. N.; Holloway, G. A.; Booksh, K. S. Anal. Chem. 2005, 77, 70167023. (27) Porter, W. H.; Haver, V. M.; Bush, B. A. Clin. Chem. 1984, 30, 18261829.

900

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

concentrated solutions without some separation steps. Separation, however, increases analysis time and complexity, while introducing uncertainty due to sample loss during the process. In addition, because of the insoluble nature of many biological molecules when taken out of their host environment, it is possible that even the most protein-resistant surfaces might turn out to be targets for adsorption. This limitation is routinely confronted in the area of biomaterials, where no perfect coating has been found to completely prevent fouling of medical implants. In this report, we develop a fundamentally different approach for direct detection of minute concentrations of target protein in the presence of highly concentrated biological serum without a separation procedure. Instead of trying to block nonspecific adsorption with BSA, ethanolamine, or PEG, we use supported lipid membranes as a temporary and removable “cloaking” layer that eliminates the signal from nonspecific interaction. Figure 1 illustrates the principle and the procedural steps. The capturing entities (in this case, antibodies) are first covalently attached to a SAM on the gold surface. Lipid vesicles are then induced to fuse onto the SAM, generating a supported bilayer surrounding the “antibody islands”. After injection of the target analyte for binding, the supported membrane is removed by a nonionic surfactant, leaving a bare surface with bound proteins for signal transduction. The cloaking process is based on the concept that the surfactant, through careful optimization, selectively disrupts the lipid membrane and removes the nonspecifically adsorbed proteins on the membrane, while leaving the specific antibody-antigen binding undisturbed. It is known that Triton X-100 can disrupt the membrane without denaturing proteins.28 We employ SPR to monitor the change of proteins on the surface before and after the removal process and verify that the concept is experimentally feasible. Two applications are demonstrated to prove the effectiveness of “membrane cloaking”. The first is the study of IgG binding to the immobilized antibodies. Measurement of protein binding by the cloaking method is compared to that without the cloaking membranes, and IgG spiked in undiluted blood serum is tested to examine the potential for direct analysis of clinical samples without purification. In the second application, an enzyme-linked electrochemical immunoassay with horseradish peroxidase (HRP)conjugated IgG is implemented using the parameters optimized in the SPR study. Electrochemical measurement is complementary to SPR and offers high sensitivity, and the membrane method is (28) Makino, S.; Reynolds, J. A.; Tanford, C. J. Biol. Chem. 1973, 248, 49264932.

highly attractive to amplification detection schemesssince the background signal from nonspecifically adsorbed HRP-IgG can be greatly suppressed. The cloaking method also facilitates integration of SPR and electrochemistry: the final step that removes the membrane cloak from the surface allows for both SPR detection and facile electron transfer through the short SAM sublayer for amperometric analysis. Unlike methods based on blocking with BSA, the membranes can be removed and reassembled after the experiment, generating a fresh surface for each measurement. Two formats of electrochemical immunoassay are demonstrated in this work: (1) a downstream detection method with a glassy carbon electrode and (2) in situ electrochemical analysis using the gold SPR substrate as a working electrode. Both formats show predictable response and femtomolar detection limits for assays using HRP-conjugated IgG. EXPERIMENTAL SECTION Materials and Instrumentation. Cysteamine, 1,3-propanedithiol (PDT), 3-mercapto-1-propanol (MPO), bovine serum albumin (BSA), avidin, rabbit anti-avidin IgG, ferrocenecarboxylic acid, Triton X-100, and 3,3′,5,5′-tetramethylbenzidine liquid/hydrogen peroxide substrate system (TMB) were obtained from SigmaAldrich. HRP-conjugated and nanoparticle-conjugated goat antirabbit IgG was from Jackson Immuno Research Inc. Glycine for stripping solution was purchased from Fisher. Phosphatidylcholine (PC) and 1-palmitoyl-2-oeoyl-sn-glycero-3-ethylphosphocholine (DOPC+) were from Avanti, and LC-SMCC heterobifunctional cross-linker and Zeba desalt spin columns were from Pierce. Goat serum was a generous gift from Robert Sargeant (Monoclonal Antibody Production Service, San Diego, CA). SPR gold substrates were fabricated with a 46-nm-thick gold layer deposited by an e-beam evaporator onto cleaned glass slides pretreated with mercaptoalkylsilane. A Biosuplar II instrument (NanoSPR, USA) with a GaAs semiconductor laser source (λ ) 670 nm) was used for SPR experiments, while electrochemical experiments were performed with an Epsilon potentiostat from Bioanalytical Systems. Preparation of IgG Immobilized Chips. To fabricate mixed monolayers on gold, clean substrates were immersed in ethanol solution mixed with 7 mM MPO and either 3 mM cysteamine hydrochloride or 3 mM PDT overnight, followed by extensive rinsing with ethanol and water. Two covalent coupling routes were used and found to be effective for IgG immobilization on the substrates. In the first method, a 50-µL solution of 2.5 mg/mL IgG dissolved in 0.05 M PBS buffer (pH 7.0) containing 5 mM EDTA and 150 mM NaCl was mixed with a 0.76-µL DMSO solution of 44.6 mM LC-SMCC to achieve a 40-fold molar ratio of linker to protein. The solution was stirred 2∼3 times every 10 min at room temperature for 1 h and then desalted. The resulting conjugate solution was incubated with the PDT/MPO substrates overnight at 4 °C, followed by thorough rinsing. In the second method, 1 mg/mL IgG in 0.1 M acetate buffer (pH 5.5) containing 0.15 M NaCl was oxidized by 0.01 M sodium metaperiodate at room temperature for 30 min in the dark with gentle mixing. The oxidized IgG solution was desalted and incubated with the cysteamine/MPO substrates overnight at 4 °C. The chip was rinsed with 0.01 M phosphate buffer (pH 7.3) containing 1 M NaCl and then soaked in 0.01% sodium borohydride in 0.1 M borate buffer (pH 7.3) for 1 h at 37 °C. Finally, it was washed with 0.01 M phosphate buffer containing 0.5 mg/mL Tween 20.

SPR Analysis in Serum. Surface modification was followed and characterized using the tracking mode of SPR angular scanning around the minimum angle. Vesicles prepared at a concentration of 1 mg/mL as previously reported29 and composed of 50/50 egg PC/DOPC+ were injected and fused on the substrates under stopped flow for 1 h. After thorough rinsing to remove free vesicles, 100 µL of a 1:20 dilution of nanoparticleconjugated secondary antibody was injected and incubated for 30 min. After flow was resumed, the supported membrane was removed with a 100-µL injection of 0.5% Triton X-100. The secondary antibody was removed and the surface regenerated with 10 mM glycine (pH 1.7) stripping buffer. Electrochemical Analysis. Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were used to optimize electrochemical parameters for dc potential amperometry. In both studies, Pt was used as the counter electrode and Ag/AgCl as the reference electrode. 1. CV. To examine the blocking of electron transfer by supported membranes and BSA, a SAM-modified Au chip was placed in the cell, and PC/DOPC+ vesicle solution or 1 mg/mL BSA was injected and incubated for 1 h. After rinsing, a cyclic voltammogram was obtained in the range from 0 to 0.5 V at 60 mV/s scan rate. The membrane was removed by rinsing with 0.5% Triton X-100 in the cell, reassembled, and the above process was repeated. 2. DPV. For optimization of TMB dilution and applied potential for amperometry, solutions of HRP and various concentrations of TMB were injected into the electrochemical cell with a SAMmodified Au as the working electrode. DPV was performed with the following settings: 50-ms pulse width, 20-mV step, 55-ms pulse period, and 60-mV pulse amplitude. Signals were obtained after 10-min incubation. 3. Downstream Amperometry. In this measurement, a separate electrochemical cell with a glassy carbon working electrode was attached downstream from the SPR flow cell. The vesicles were fused for 1h on the SAM/IgG-modified gold substrate, followed by rinsing. Then 1 mg/mL BSA solution was injected for 10 min through a sample loop at slow flow rate to prevent adsorption on the flow cell ceiling, junctions, and system tubing. An aliquot of 100 µL of HRP-conjugated anti-rabbit IgG was then injected at a flow rate of 8.3 mL/h and incubated for 20 min, followed by rinsing with 0.5% Triton X-100 to remove the membrane. TMB solution was injected into the cell and allowed to incubate for 10 min. Flow was resumed, and the oxidized TMB substrate was detected downstream at the glassy carbon electrode at a reduction potential of 0.175 V versus Ag/AgCl. Finally, glycine stripping buffer was injected to break the bond between the primary and secondary IgG, thereby regenerating the original surface. 4. In Situ Electrochemistry/SPR (ESPR). An ESPR flow cell (NanoSPR, USA) specifically designed for combined electrochemistry and SPR measurement was used for amperometric detection of HRP oxidized TMB substrate at gold. SPR was used to track the assembly of the membrane, binding of secondary antibody, removal of the membrane, and stripping of IgG as described above. The TMB solution was passed through the cell while 0.175 V (vs Ag/AgCl) was applied to the IgG-modified Au working electrode. (29) Dong, Y.; Phillips, K. S.; Cheng, Q. Lab Chip 2006, 5, 675-681.

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

901

Figure 2. SPR sensorgram for binding of avidin and anti-avidin IgG with the membrane cloaking procedure. (a) Binding of 50 µg/mL avidin, (b) injection and incubation of PC vesicles, (c) injection and incubation of anti-avidin IgG, (d) removal of SBM with Triton X-100, and (e) baseline after SBM removal showing bound avidin and IgG.

RESULTS AND DISCUSSION SPR Characterization of the “Membrane Cloaking” Method. A major advantage of coupling SPR with electrochemical immunoassay is the ability to track mass changes on the gold electrode surface with high accuracy. Fusion of vesicles, binding of analytes, and removal at each step is difficult to understand without the aid of real-time signal transduction. The cloaking method was first tested with substrates on which rabbit anti-avidin IgG had been immobilized using the metaperiodate oxidation method, which has been found to yield high binding efficiencies.30 The 30:70 molar ratio of cysteamine/MPO was chosen after preliminary studies with several ratios in an attempt to create sufficient amine sites for linkage while still allowing for vesicle fusion on the mostly hydroxyl-terminated surface. While the ratio chosen served well for this study, it is important to note that tailoring and surface composition may need to be optimized for applications involving different phospholipids. Figure 2 shows an SPR sensorgram obtained for the surface buildup study using the tracking mode. To determine if significant binding occurred on the surface, 0.7 µM avidin was first injected in step a and incubated for 20 min. After rinsing for 10 min, the signal stabilized, yielding an increase of 0.057°. The PC/DOPC+ vesicles were then injected at step b and allowed to incubate for 1 h. This yields a minimum angle increase of 0.480°, which corresponds to formation of a bilayer membrane, as demonstrated in previous studies.15 After rinsing, anti-avidin IgG was injected at step c and incubated for ∼40 min, followed by rinsing and removal with 0.5% Triton X-100 at (d). The incubation times for both binding steps were based on the minimum time to achieve a stable binding curve in each experiment. This demonstrates a major advantage of SPR as a real-time, label-free technique for optimization of sensor interface fabrication. For other methods, a series of different concentrations and time periods would need to be tested, requiring burdensome use of time and resources. The hypothesis that Triton X-100 removes the membrane but not the bound protein can be verified in two ways from the binding (30) Nisnevitch, M.; Firer, M. A. J. Biochem. Biophys. Methods 2001, 49, 467480.

902

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

study. First, the decrease in the SPR minimum angle associated with injection of Triton X-100 and removal of the SBM is 0.487°, which is nearly identical to the increase seen for vesicle fusion (0.480°) in step b. The difference of 7 mdeg can be accounted for by instrumental drift and noise in this system, which is not temperature regulated. If a significant amount of protein was removed by the surfactant, the decrease in minimum angle would be greater than the value for lipid material only. In another words, it would reflect additional protein mass loss, which was not observed. Other convincing evidence for selective removal by surfactant solution is that the minimum angle after removal at point e minus the starting value before (a) (0.247°) is nearly identical to the sum of the increase (0.250°) in steps a and c, which corresponds to the specific binding of avidin and anti-avidin IgG. If the membrane was not entirely removed, the minimum angle at (e) would be higher than the sum of these two angles. The results clearly show that the use of 0.5% Triton X-100 solution selectively removes the lipid membranes while leaving the specific antibody-antigen interactions intact. SPR Biosensing in Serum with the Cloaking Membrane. A series of experiments were conducted to demonstrate that the cloaking process could be highly useful for SPR biosensing in serum. To closely mimic the reagents used by many investigators and show compatibility, rabbit IgG was immobilized on the substrate and the nanoparticle-conjugated goat anti-rabbit IgG was used as the target analyte. Figure 3a shows an example of the sensorgram data for this system. Vesicles were first injected and incubated for 1h, followed by rinsing and injection of anti-rabbit goat IgG spiked in donkey serum. After 30-min incubation, the excess was rinsed and Triton was injected when the signal stabilized. Before using the cloaking method for quantification of a sample in serum, it is important to verify that results obtained from the cloaking method without any serum are accurate. In other words, the accuracy of binding signal obtained with the cloaking method (indirect measurement) needs to be compared to that obtained in the normal manner by measuring the increase of signal upon binding (direct measurement). The results of these tests are shown in the calibration curve in Figure 3b. The open circles represent the change in minimum angle associated with direct measurement of binding by nanoparticle-conjugated IgG. After binding, the bilayer was removed and rinsed, and the difference (new baseline minus original baseline before membrane fusion) was measured. This “indirect” result represents the signal obtained with the cloaking method. Although a slight deviation is seen at higher concentrations, the curve closely follows the direct binding curve, and the results are within the deviation normally seen in our system for antibody-antibody binding. The cloaking method was then applied to a sample spiked in serum. The open diamond in Figure 3b represents the value obtained for binding of nanoparticle-conjugated anti-rabbit IgG spiked in donkey serum. No dilution was made to the serum, and a protein assay with a Pierce kit indicated that the concentration of protein in the serum was ∼75 mg/mL according to a standard calibration curve. Despite this highly concentrated matrix, the use of the cloaking method generates measurement results comparable to those obtained with the direct method in pure buffer without serum (within ∼10% of corresponding minimum angle). However, when the same concentration of nanoparticle-conjugated antibody

Figure 3. (a) SPR sensorgram showing membrane cloaking process with nanoparticle-conjugated anti-rabbit IgG spiked in serum and (b) calibration curves (without serum) showing binding signal measured directly (open circles) and indirectly (solid circles). The binding signal for a 20-fold dilution of anti-rabbit IgG in serum obtained indirectly through the cloaking method (open diamond) and the direct measurement of the same solution in serum without membrane removal (solid diamond) are included for comparison.

Figure 4. (a) Cyclic voltammograms showing electron transfer on mercaptopropanol/propanedithiol-modified gold surfaces with redox probe of ferrocenecarboxylic acid: original SAM-modified gold (black), after incubation with PC/DOPC+ (green), after membrane removal with Triton and rinsing (blue), and with BSA blocking (pink). (b) DPV curves obtained for optimization showing the current response as a function of the TMB dilutions.

was injected in serum without removal of the membrane (solid diamond in Figure 3b), the increase in minimum angle was ∼1.04°, ∼4 times the correct value and an error of over 300%. The results clearly demonstrate the promise of membrane cloaking for SPR measurement in highly concentrated protein matrixes such as serum. Additional control experiments were performed to verify removal efficiency for pristine PC membranes on cysteamine/MPO mixed layer and serum-treated PC membranes without capture molecules immobilized. In both cases, use of Triton can effectively remove the membrane and return the sensorgram signal to baseline (Supporting Information), indicating that nonspecifically adsorbed proteins from serum do not stick to the surface in the process of membrane removal. Although further optimization of linker length and vesicle fusion process may reduce the deviations between direct and indirect measurements, the feasibility of SPR direct detection of minute protein concentrations in concentrated serum without separation is

undoubtedly established. The surfactant removal process is normally facile, with minor deviations observed on chips with varied use history. Surface contamination and deterioration of the SAM layer may contribute to the observed phenomenon. Application of the Membrane Cloaking Method to Amperometry. In addition to SPR analysis, the cloaking strategy was also characterized and tested for electrochemical measurements. Figure 4a shows cyclic voltammograms of mixed SAM-coated substrates at various stages in the surface fabrication process. Before vesicle assembly, a normal voltammogram was observed for the redox probe ferrocenecarboxylic acid (black curve). After vesicle fusion, the surface was blocked, resulting in greatly reduced electron transfer and a relatively flat voltammogram (green curve). However, after treatment with Triton X-100, the surface could be restored to its previous condition with the same signal as seen for the pristine surface (blue curve). This verifies that the cloaking method is completely reversible. A comparison Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

903

experiment was carried out with the use of BSA to block the surface. Electron transfer for the BSA-blocked surfaces was extremely slow (pink curve), and the blocking process was irreversible, preventing the use of amperometric detection on these surfaces. To demonstrate amperometric analysis with the membrane cloaking method, TMB was chosen as an electrochemical substrate because it is inexpensive, widely available, and commonly used in colorimetric amplification schemes.31 Recently a number of reports have shown its potential as an electrochemically active substrate in enzyme immunoassay.32,33 TMB has several major advantages over other substrates. First, the reduction potential is low, which minimizes the potential for desorption of bound antigens and gold film peeling. Using DPV, it was found that 0.175 V was sufficient to reduce the oxidized TMB on the mixed SAM surface for different dilutions (Figure 4b). It was also found that the background from the unoxidized TMB substrate was very low, and no signal increase was detected for control experiments with TMB without HRP. The reduced TMB has relatively low adsorption on glassy carbon as compared to some other substrates. No fouling was observed in the flow system at normal flow rates. In order to compare the bilayer cloaking method directly to similar IgG detection methods developed for ESPR34 and electrochemistry,35-37 amperometric measurement was first carried out with a separate, downstream electrochemical cell containing a glassy carbon electrode for detection of oxidized TMB. The immunoassay procedure was also similar to others, with the exception of the cloaking method in place of BSA blocking. Similar to the procedure discussed above, vesicles were first incubated for ∼1 h on a rabbit-IgG-modified chip in the SPR flow cell with minimum angle tracking. After fusion was complete, the cell was rinsed and 100 µL of 1 mg/mL BSA solution was injected using a slow flow rate for 10 min in order to cover the Teflon tubing and flow cell parts on which membranes do not assemble. After rinsing, HRP-conjugated goat anti-rabbit IgG was injected and allowed to incubate for 20 min, followed by rinsing with 100 µL of 0.5% Triton X-100 to remove the membrane and nonspecifically adsorbed HRP-IgG. A 100-µL TMB plug was injected, incubated for 10 min, and then moved downstream at a rate of 8.3 mL/h to the glassy carbon electrode where 0.175 V was applied to reduce the oxidized substrate. Figure 5 (top left) shows the amperometric signal recorded for several different concentrations of anti-rabbit IgG that were tested. A flat baseline was observed for TMB samples when no HRP was present. Because of the large size of the SPR sample cell and turbulent forces, the peaks are slightly broadened and show small deviations. As shown in Figure 5 (top right), the response is linear in the 1-100 fM range and shows some indication of saturation starting at ∼0.5 pM concentration. However, a log scale graph (Figure 5, bottom) reveals that signal (31) Volpe, G.; Compagnone, D.; Draisci, R.; Palleschi, G. Analyst 1998, 123, 1303-1307. (32) He, Y.-N.; Cheng, H.-Y.; Zheng, J.-J.; Zhang, G.-Y.; Chen, Z.-L. Talanta 1997, 44, 823. (33) He, Z.; Gao, N.; Jin, W. J. Chromatogr., B 2003, 784, 343-350. (34) Toda, K.; Tsuboi, M.; Sekiya, N.;Ikeda, M.; Yoshioka, K.-I. Anal. Chim. Acta 2002, 463, 219-227. (35) Morier, P.; Vollet, C.; Michel, P. E.; Reymond, F.; Rossier, J. S. Electrophoresis 2004, 25, 3761-3768. (36) Aguilar, Z. P.; Vandaveer, W. R., IV; Fritsch, I. Anal. Chem. 2002, 74, 33213329. (37) Wilson, M. S. Anal. Chem. 2005, 77, 1496.

904

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

remains predictable as high as ∼1 µM. Using the “cloaking method”, detection limit as low as 5.6 fM was achieved, making it highly competitive versus other electrochemical immunoassay methods with detection limits of 53 pM using ESPR,34 8 pM in a microfluidic system,34 and 56 fM using a novel microelectrode design.36 While these reports used p-aminophenyl phosphate instead of TMB, assays using hydroquinone diphosphate have also been reported to have limits of 5.3 pM for IgG.37 We attribute the several orders of magnitude better sensitivity presented here to the substantially decreased background signal with the use of membrane cloaking and better substrate performance. In order to determine the background signal for nonspecifically adsorbed HRP and compare BSA blocking with the cloaking method, several control experiments were performed with SAMtreated substrates that were not modified with rabbit-IgG. The resulting signal levels are shown in the open circle markers in Figure 5 (bottom). Dilutions of 1:1 000 000 and 1:10 000 for HRPIgG did not result in significant signal above the baseline. This is significant because most electrochemical immunoassays with 1:10 000 dilution of HRP-IgG typically report substantial signal from nonspecific adsorption. Using the cloaking system with 1:10 000 dilution, signal from nonspecific adsorbed HRP-conjugated antibody is no longer a limiting factor. Instead, the system noise becomes the major limitation. In addition, incubation of a 1:1000 dilution of HRP-IgG caused a nonspecific signal of ∼10 nA. This is ∼1 order of magnitude below the specific signal at the same concentration. This is in stark comparison to the case where no blocking membrane was used, in which signal from nonspecific adsorption of 1:1000 dilution of HRP-IgG was found to be 4 orders of magnitude higher than that with cloakingsa scenario that would limit detection to the millimolar range. The large jump in signal between 1:1000 and 1:10 000 dilution levels is probably because of the high signal amplification by HRP. Even a minute amount of HRP residue in the cell results in greatly increased background signal. The membrane cloaking method was found to be superior to the traditional BSA-blocking method for preventing buildup of the background signal. The membranes allow for more experiments to be performed on a single chip, thanks to the surface regeneration by simple surfactant removal of cloaking. It should be noted that BSA is still needed to coat other surfaces in the system on which vesicles do not fuse spontaneously, such as the Teflon flow cell, tubing, and the silicon gasket. Without overnight blocking of these surfaces by concentrated BSA, a high background signal was present. The cloaking method works best on the electrode surface, but in flow injection analysis, other surface areas also contribute to overall background signal. An improved flow cell design using new materials on which vesicles can readily fuse might reduce this contribution. Reduction of corners and cracks would be effective to decrease the background signal. Since longer rinsing times were found to have a large effect on the residual signal, it is expected that judicious design of the flow path to minimize stagnant pools would also be beneficial. Implementing Cloaking for SPR with in Situ Electrochemical Detection. In the experiments discussed above, the use of a downstream glassy carbon electrode serves as a convenient comparison for membrane cloaking versus other methods, including BSA blocking. It is noteworthy that the cloaking method not

Figure 5. Top left: flow injection amperometric response for 5.3 nM (green), 0.53 nM (blue), and 5.3 pM (cyan) HRP-labeled goat anti-rabbit IgG and baseline with TMB only (black). Signals were filtered to remove syringe pump noise. Top right: Linear portion of the response curve and the start of saturation for IgG obtained with the downstream amperometry. Bottom: The expanded response range for binding of IgG (closed circles) and signals from nonspecific adsorption of IgG using the cloaking method (open circles) for comparison.

only meets or exceeds the performance of other techniques for sensitivity and reduction of nonspecific adsorption but also leaves an unhindered gold surface free for direct electrochemical detection. To fully capitalize on this feature, a flow injection electrochemical enzyme immunoassay was developed using the gold SPR surface as a working electrode for in situ detection. The protocol was based on the procedure for the glassy carbon-based assay developed above, with slight modifications. The same strategy of membrane assembly, IgG binding, and cloak removal was followed, and the potential applied to the gold surface was also set at 0.175 V. However, since TMB was found to foul the gold surface with extended incubation in static voltammetric studies, the substrate was continuously flowed through, rather than incubated in, the SPR cell. Figure 6 shows the results of electrochemical measurements under varied conditions. A large signal was observed for TMB with immobilized HRP-conjugated IgG, while control experiments with only TMB and no HRP-IgG resulted in only noticeable baseline noise, but no signal could be identified (top left). The immobilized HRP-IgG on the mixed monolayer surface show highly reproducible amperometric signals when TMB was injected in the ESPR flow cell (Figure 6, top right).

The larger working electrode area of the gold SPR substrate as compared to the glassy carbon electrode likely counterbalanced the reduced signal due to the loss of incubation time for the HRP/ TMB oxidation reaction. The cloaking method was further implemented with the rabbit IgG-immobilized substrates and applied to an assay for binding of HRP-conjugated anti-rabbit IgG, as shown in Figure 6 (bottom left). Given the 5-10 fM concentrations detected, it is apparent that the sensitivity obtained using this direct flow detection mode is similar to the incubation/ downstream detection method described above. This further verifies that the membrane cloaking method, when used in conjunction with ESPR, allows facile electron transfer to take place after membrane removal and is capable of rapid and sensitive detection in situ. The whole process is characterized and monitored by SPR, except the final step of detecting bound antigen at very low concentrations, which is carried out by electrochemistry. Routine protein analysis of a large number of samples would gain advantage from this rapid flow injection format and benefit from the reduced complexity and cost of combining SPR and electrochemical analysis in a single enclosure. Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

905

Figure 6. Top left: flow injection response of TMB for in situ ESPR analysis with no HRP (curve a) and immobilized HRP (curve b). Top right: reproducibility of the signal from multiple TMB substrate injections with immobilized HRP in the ESPR flow cell. Bottom: response curve for detection with HRP-conjugated anti-rabbit IgG/TMB using the membrane cloaking method.

CONCLUSIONS While many attempts to overcome the challenge of nonspecific adsorption have been based on better blocking agents or use of PEG groups, we report here a highly effective alternative strategy that takes advantage of the unique transient stability and biomimetic properties of supported membranes as a removable cloak. Using this strategy with SPR, it was shown that a trace amount of protein in ∼75 mg/mL concentration of protein matrix can be analyzed because the nonspecifically adsorbed species are removed simultaneously with the supported membrane by a nonionic surfactant. It was demonstrated that the specific antigenantibody binding interaction was not affected by the levels of surfactant required to remove the cloak. The result is significant in that protein binding can be directly detected even in raw serum conditions where the background signal would result in a >300% nonspecific signal without a cloaking membrane. While it is currently a common practice to use a reference channel in SPR measurements, no reports have demonstrated the ability to clearly detect and quantify the analytical signal from binding of a protein analyte in raw, undiluted blood serum. The difference between the analytical and reference channel, due to immobilization of the 906

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

capture molecules, may contribute to large errors for samples with a high matrix concentration of proteins. Reference channels also require additional instrumental complexity, whereas the cloaking method can be used with an inexpensive single-channel system. Nevertheless, it might be interesting to study whether the cloaking method combined with a reference channel would provide even better S/N performance. As the cloaking method tolerates multiple tests of highly concentrated serum with greatly reduced surface fouling, it can be used for direct detection of clinically relevant analytes in body fluids. In addition, the cloaking method provides a new class of interface for electrochemical enzyme immunoassay. Current approaches rely on separate or downstream detection, or use of porous materials that support binding sites while permitting electron transfer. The cloaking method overcomes the problems by transforming the property of the interface in the process of electrochemical enzyme immunoassay. During the binding phase of the assay, the interface is a solid membrane to suppress nonspecific adsorption, especially that of the enzyme-tagged antibodies. After cloak removal, it is transparent to electron transfer due to the use of a short-chain SAM. Using this strategy

with TMB, detection limits with 1:1000 dilution of HRP-conjugated

ACKNOWLEDGMENT

antibody can be improved by as much as 4 orders of magnitude

This work is supported by Cancer Research Coordinating Committee (CRCC) of University of California and a grant from NSF (BES-0428908).

versus a bare SAM, and concentrations of HRP-IgG as low as 5.3 fM were detected. Finally, it is significant that the combined operation of amperometric electrochemistry and SPR for in situ biosensing has been achieved using the cloaking method. This approach is versatile for method development in the determinations of biologically significant analytes such as protein toxins in food samples or biowarfare agents in complex media because of

SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

the speed, easy validation of assay procedures, and results obtained through two complementary physical techniques for comparison. The method also has great potential in developing

Received for review July 9, 2006. Accepted November 13, 2006.

point-of-care diagnostics due to its low cost, high sensitivity, robustness, and rapid format.

AC0612426

Analytical Chemistry, Vol. 79, No. 3, February 1, 2007

907