Development of a Highly Enantioselective Capacitive Immunosensor

Anal. Chem. 2006, 78, 7592-7596

Correspondence

Development of a Highly Enantioselective Capacitive Immunosensor for the Detection of r-Amino Acids Song Zhang,† Jingjing Ding,† Ying Liu,† Jilie Kong,*,† and Oliver Hofstetter*,‡

Department of Chemistry, Fudan University, Shanghai 200433, People’s Republic of China, and Department of Chemistry and Biochemistry, Northern Illinois University, DeKalb, Illinois 60115

This work describes a highly enantioselective and sensitive immunosensor for the detection of chiral amino acids based on capacitive measurement. The sensor was prepared by first binding mercaptoacetic acid to the surface of a gold electrode, followed by modification with tyramine utilizing carbodiimide activation. The hapten 4-amino-Dphenylalanine was then covalently immobilized onto the electrode by diazotization. Stereoselective binding of an anti-D-amino acid antibody to the hapten-modified sensor surface resulted in capacitance changes that were detected with high sensitivity by a potentiostatic step method. Using capacitance measurement, detection limits of 5 pg of antibody/mL were attained. The exquisite stereoselectivity of the antibody was also utilized in a competitive setup to quantitatively determine the concentration of the analyte D-phenylalanine in nonracemic samples containing both enantiomers of this amino acid. Trace impurities of D-phenylalanine as low as 0.001% could be detected. The R-amino acids represent one of the most important classes of substances in nature that possess a stereogenic center,1 and they are widely used in the food, chemical, and pharmaceutical industries, e.g., as chiral building blocks in drug development.2,3 Since even minor enantiomeric impurities may cause severe pharmacologic and toxicologic side effects, regulatory agencies such as the U.S. Food and Drug Administration (FDA) request the quantitative determination of the individual enantiomers of chiral drugs.4 A variety of methods have been used successfully for the detection of enantiomeric impurities; these include chromatography, nuclear magnetic resonance (NMR), fluorescence * Corresponding authors. Telephone: 0086-21-65642138. 001-815-7536898. E-mail: [email protected]; [email protected]. † Fudan University. ‡ Northern Illinois University. (1) Barrett, G. C. Chemistry and Biochemistry of the Amino Acids; Chapman and Hall: London, 1985. (2) Stinson, S. C. Chem. Eng. News 1997, 75 (42), 38-70. (3) Coppola, G. M.; Schuster, H. F. Asymmetric Synthesis: Construction of Chiral Molecules Using Amino Acids; Wiley: New York, 1987. (4) U.S. Food and Drug Administration, Center for Drug Evaluation and Research. FDA’s Policy Statement for the Development of New Stereoisomeric Drugs, 5/1/92, corrections made 1/3/97; http://www.fda.gov/cder/ guidance/stereo.htm.

7592 Analytical Chemistry, Vol. 78, No. 21, November 1, 2006

measurements, and circular dichroism (CD).5-7 Recently, antibodybased chiral immunosensors were shown to be particularly suited for the sensitive detection of trace impurities at levels typically not attainable with other methods.8-10 However, most chiral immunosensors used thus far combined stereoselective antibodies with rather expensive transduction elements such as SPR9 or NMR,10 thus limiting their practical application. The objective of this study was to develop a simple and inexpensive, yet sensitive chiral immunosensor using electrochemical transduction. Electrochemical methods are especially attractive because they combine sensitivity with convenience, speed, and low costs; their accuracy and precision have been utilized in several chiral recognition studies.11,12 Capacitive measurement is a particularly sensitive electrochemical method, which has been employed extensively for various applications including antibody-based detection of proteins and polysaccharides.13-16 We have previously reported the application of capacitance measurement for sensing the interaction between antibodies and hyaluronan-binding protein.15 While in that study the antibody was covalently immobilized on the electrode, we chose a different approach here to utilize the stereoselective binding capacity of suitable antibodies to detect amino acid enantiomers. This was based on the requirement of capacitive measurement that even though a biorecognition layer formed on the surface of an (5) Schurig, V.; Lindner, W.; Uray, G. In Houben-Weyl, Methods of organic chemistry; Helmchen, G., Hoffmann, R. W., Mulzer, J., Schaumann, E., Eds.; Thieme: Stuttgart, Germany, 1995; Vol. E21a, Part A.3, pp 147-292. (6) Pu, L. Chem. Rev. 2004, 104, 1687-1716. (7) Allenmark, S. Chirality 2003, 15, 409-422. (8) Hofstetter, H.; Hofstetter, O. Trends Anal. Chem. 2005, 24, 869-879. (9) Hofstetter, O.; Hofstetter, H.; Wilchek, M.; Schurig, V.; Green, B. S. Nat. Biotechnol. 1999, 17, 371-374. (10) Tsourkas, A.; Hofstetter, O.; Hofstetter, H.; Weissleder, R.; Josephson, L. Angew. Chem., Int. Ed. 2004, 43, 2395-2399. (11) Attard, G. A. J. Phys. Chem. B 2001, 105, 3158-3167. (12) Stefan, R. I.; van Staden, J. F.; Aboul-Enein, H. Y. Cryst. Eng. 2001, 4, 113-118. (13) Berggren, C.; Bjarnason, B.; Johansson, G. Biosens. Bioelectron. 1998, 13, 1061-1068. (14) Jiang, D. C.; Tang, J.; Liu, B. H.; Yang, P. Y.; Kong, J. L. Anal. Chem. 2003, 75, 4578-4584. (15) Jiang, D. C.; Tang, J.; Liu, B. H.; Yang, P. Y.; Shen, X. R; Kong, J. L. Biosens. Bioelectron. 2003, 18, 1183-1191. (16) Bard, A. J., Faulker, L. R., Eds. Electrochemical Methods-Fundamentals and Application; Wiley: New York, 1980. 10.1021/ac060840h CCC: $33.50

© 2006 American Chemical Society Published on Web 09/28/2006

electrode must be insulating and thin (i.e., in the nanoscale range), binding of an analyte to the sensor must still cause a significantly strong capacitance change to be detected. Thus, we decided to modify the electrode with the low molecular weight hapten 4-amino-D-phenylalanine (4-NH2-D-Phe) and to measure binding of the antibody to the sensor surface utilizing a potentiostatic step method. 4-NH2-D-Phe had previously been employed as hapten in the synthesis of the immunogen for the production of the antiD-amino acid antibody (anti-D-AA) used here17 and was known to be stereoselectively recognized by this antibody. Using a competitive setup, in which suitable binding partners contained in a sample compete with the surface-immobilized hapten for the antibodybinding site, label-free detection of amino acid enantiomers was possible. EXPERIMENTAL SECTION Materials. The monoclonal antibody anti-D-AA, which enantioselectively binds to D-amino acids, was produced as described previously.17 Immunoglobulin G (IgG; from mouse), bovine serum albumin (BSA), mercaptoacetic acid (MAA), N-hydroxysuccinimide (NHS), N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), tyramine (TA), and p-amino-D-phenylalanine hydrochloride (4-NH2-D-Phe) were purchased from Sigma (St. Louis, MO). D-Phenylalanine (D-Phe) and L-phenylalanine (L-Phe) were purchased from Aldrich (St. Louis, MO). 1-Hexadecanethiol was from Fluka (Buchs, Switzerland). A diluted human serum sample (1:10 v/v) was obtained from Fudan University Hospital (Shanghai, China). Triethylamine was purchased from Shanghai Chemical Reagent Co. (Shanghai, China). All other reagents were of reagent grade and used as received. Water was deionized prior to use. Apparatus. Electrochemical experiments were performed with a CHI 660A and a CHI 1030 electrochemical system (CH Instruments, Austin, TX). A single-compartment, three-electrode glass cell was used. The bare and modified gold disk electrodes (with a diameter of 0.22 cm) were used as the working electrodes. Platinum foil and a saturated calomel reference (SCE) were used as a counter electrode and reference electrode, respectively. The immobilization of the hapten 4-amino-D-phenylalanine on the Au surface was investigated in situ with a ChI 420 quartz crystal microbalance (QCM) analyzer (CH Instruments) and quartz crystals (7.995 MHz) sandwiched between two Au electrodes (areas, 0.196 cm2). After immobilization, the gold electrode surface was thoroughly rinsed with water and dried under nitrogen gas. The QCM frequency change (∆F) caused by the increase in mass (∆M) was measured in air and calculated using the Sauerbrey equation: ∆F ) - 2.3 × 10-6F02∆M/A. Here, F0 is the fundamental frequency of QCM and A is the surface of the gold electrode. The surface morphologies of the hapten-modified Au electrode, before and after the specific interaction with anti-D-AA, were measured using a Nanoscope Multimode IV atomic force microscope (AFM; Digital Instruments, Santa Barbara, CA) in tapping mode. Electrode Modification. A schematic representation of the preparation process of the hapten-modified electrode and the (17) Hofstetter, O.; Hofstetter, H.; Wilchek, M.; Schurig, V.; Green, B. S. Int. J. Bio-Chromatogr. 2000, 5, 165-174.

detection principle of the immunosensor is given in Figure 1. The gold electrode was polished carefully with alumina slurries (1.0, 0.3, and 0.05 mm), followed by sonication in distilled water and drying with purified N2. The cleaned Au electrode was then immediately immersed in a solution of 10 mM MAA in absolute ethanol at 4 °C for 12 h, thus forming a self-assembling monolayer with free carboxyl groups exposed on the surface. Afterward, the electrode was thoroughly rinsed with absolute ethanol and dried with purified N2. The carboxyl groups of MAA were activated by immersing the electrode for 60 min in an aqueous solution containing 0.4 M EDC and 0.1 M NHS, followed by rinsing with deionized water. Then, the electrode was immersed overnight in a 0.05 M aqueous solution of TA. Diazotization was then used to couple the hapten 4-NH2-D-Phe via its side chain to the tyraminyl residues, thus preserving both the R-amino and R-carboxyl groups attached to the stereogenic center.9,10 For that, 4-NH2-D-Phe at a concentration of 5.6 mg/mL was dissolved in 0.2 M HCl and mixed with NaNO2 (35 mg/mL) at a ratio of 1:9 (v/v) for 1 h at 0 °C. The solution was then added dropwise to the tyramine-modified electrode immersed in 20 mM sodium borate buffer (pH 9.0). In order to keep the formed monolayer insulating and pin-free, and to ensure a high sensitivity, 1-hexadecanethiol was used to block uncovered pinholes on the gold surface. All fabrication steps were monitored with [Fe(CN)6]3-/4- redox couples, as reported previously.14,18 Electrochemical Measurements. Cyclic voltammetry was performed in 10 mM phosphate buffer, pH 7.0, containing 0.1 M KCl and 5 mM K3[Fe(CN)6]/K4[Fe(CN)6] (1:1 mixture as a redox probe). The scan rate was 100 mV/s, and the potential range was from -0.2 to 0.6 V. Capacitance changes of the hapten-modified immunosensor were evaluated in 10 mM phosphate buffer with a potential-step method. The applied potential pulse was 50 mV, and the current transients following the potential step were recorded. The current values were collected at a frequency of 50 kHz and the first 10 values obtained at 1-s intervals were used for the evaluation of the capacitance. In the competitive setup, solutions of anti-D-AA (at 50 ng/mL) were preincubated with D- or L-Phe at varying concentrations at 37 °C for 30 min prior to capacitance measurement with the hapten-modified electrode. All measurements were performed at room temperature. RESULTS AND DISCUSSION Preparation and Characterization of the Hapten-Modified Gold Electrode. In order to immobilize the hapten 4-NH2-D-Phe on the gold electrode in such a way that the stereogenic center is accessible for interaction with the anti-D-amino acid antibody, a stepwise immobilization technique was applied as shown in Figure 1 and described in detail in the Experimental Section. The fabrication of the hapten-modified immunosensor was followed by cyclic voltammetry in a fairly reversible redox couple of [Fe(CN)6]3-/4-. While the cyclic voltammogram for the bare gold electrode showed clear redox peaks (Figure 2A, line a), selfassembly of the MAA monolayer on the gold electrode resulted in a decrease of the redox peaks (Figure 2A, line b), indicating a considerably reduced electron-transfer rate of [Fe(CN)6]3-/4- on the electrode. Covalent binding of TA to the MAA led to further decrease of the redox peaks (Figure 2A, line c). The disappearance Analytical Chemistry, Vol. 78, No. 21, November 1, 2006

7593

Figure 1. Schematic representations of the fabrication process of the electrode and the principles of antibody-based stereoselective detection of amino acids. The stepwise modification of the Au electrode comprised the following: (a) self-assembly of MAA, (b) binding of TA using carbodiimide activation, (c) immobilization of the hapten 4-NH2-D-Phe by diazotization, and (d) blocking of free spaces on the electrode with 1-hexadodecanethiol. The specific binding of the anti-D-amino acid antibody to the immobilized hapten shown in (e) could be inhibited in the competitive setup by free D-Phe present in test samples.

of the redox peaks and the lower current responses in both the anodic and cathodic processes upon immobilization of 4-NH2-DPhe (Figure 2A, lines d and e, respectively) imply that the hapten formed a covalently linked layer on the gold electrode and that direct access by the redox couple was blocked. The stepwise immobilization of the hapten 4-NH2-D-Phe on the gold electrode was also verified by QCM experiments (see Table 1). The frequency decrease and mass increase in every fabrication step as determined by QCM confirmed the stepwise formation of the MAA, TA, and hapten layers on the sensor. Investigation of the morphology of the hapten-modified electrode by AFM revealed a thin and uniform film with protruding pinnacles; the surface was found to become rougher after binding of antibody (see Figure 1 in Supporting Information). Capacitance Measurement. Capacitance evaluation of a potentiostatic step method was based on the assumption that the current transient after a potential step follows a simple RC model13 (see Figure 2B, inset c′); the capacitance was calculated 7594

Analytical Chemistry, Vol. 78, No. 21, November 1, 2006

according to the following equation:

i(t) )

(

)

u t exp Rs R sC

where i(t) is the current as a function of time t, u is the applied pulse potential, Rs is the electrical resistance of all elements serially connected to the layer, and C is the measured capacitance of the system. Curve a′ in Figure 2B shows a typical current response obtained with the fabricated hapten-modified sensor. A plot of the logarithm of the current values as a function of time (Figure 2B, line b′) gave a straight line (R2 ) 0.999), which allowed determination of both Rs and C from the slope and the y-intercept of the line. These results indicated that the experimental system was in good agreement with the ideal RC model and that the quick potential-step method was suitable for capacitance measurement. Influence of Electrolyte Concentrations and pH. The effect of different detection conditions, such as electrolyte concentration

Figure 2. (A) Cyclic voltammograms recorded in a solution of 0.1 M KCl containing 5 mM [Fe(CN)6]3-/4- with (a) the bare gold electrode, (b) the MAA self-assembled gold electrode, (c) the tyramine-modified gold electrode, (d) the electrode modified with the hapten 4-NH2-D-Phe, and (e) the hapten-modified electrode after treatment with 1-hexadecanethiol. The scan rate was 100 mV/s. The voltage range was from -0.2 to 0.6 V (vs SCE). (B) Current vs time plot obtained with the hapten-modified electrode after applying a potentiostatic step method in 10 mM phosphate buffer, pH 7 (a′). The applied pulse potential was 50 mV, and the sample frequency was 50 kHz. Semilogarithmic representation of log i(t) vs t (b′). Simple RC equivalent circuit (c′). Rs represents the resistance of elements serially connected to the sensor, and C is the capacitance. Table 1. Mass and Capacitance Changes Resulting from the Stepwise Modification of the Sensor Surface as Determined by QCM and Potentiostatic Stepa modified layerb

-∆Fc (Hz)

∆Mc (ng)

amountc (nmol/cm2)

resistanced (Ω/cm2)

capacitanced (nF/cm2)

MAA TA 4-NH2-D-Phe 16-SH

87.67 ( 2.57 44.77 ( 0.55 49.90 ( 2.51 164.73 ( 4.50

116.88 ( 3.56 59.69 ( 0.73 66.50 ( 3.34 219.6 ( 5.99

6.48 ( 0.20 2.24 ( 0.05 1.58 ( 0.10 4.34 ( 0.10

43.89 ( 0.31 48.96 ( 0.29 56.35 ( 0.56 57.24 ( 0.27

23562 ( 40 22233 ( 62 6365 ( 57 5291 ( 13

a Data represent the mean value of three parallel experiments using QCM and the potentiostatic-step method, respectively. b The derivatization of the Au electrode is shown in Figure 1. cThe QCM frequency change (∆F) caused by the increase in mass (∆M) was calculated using the Sauerbrey equation as described in the Experimental Section. dThe resistance and capacitance changes were monitored by a potential-step method and calculated according to the ideal RC model.

and pH, on the capacitance of the hapten-modified sensor system was investigated by a potential-step method. The capacitance of the sensor at different concentrations of phosphate buffer (ranging from 10 to 100 mM) were recorded and calculated (see Table 1 in Supporting Information). The capacitance of the immunosensor clearly increased with increasing ionic strength of the buffer; however, the correlation coefficient of the straight line about log i(t) versus t decreased considerably. While the increased capacitance was caused by the increased density of the diffuse ion cloud surrounding the biorecognition layer on the sensor surface,13 the decreased correlation coefficients indicate that at high ionic strength the experimental system deviates from the idealized simple RC model. In order to ensure stable experimental conditions that are in accordance with a simple RC model, 10 mM phosphate buffer was used in all further investigations. The effect of the buffer pH on the capacitance of the haptenmodified sensor was also investigated by the potential-step method. At pH values between pH 5 and pH 7, the capacitance of the sensor did not vary significantly. However, below pH 4, a clear increase in capacitance was observed. Since low pH values may destroy the insulating layer on the surface of the electrode and cause the formation of pinhole structures, which impair capacitance measurements,15 all following experiments were carried out at pH 7.0. It is noteworthy, furthermore, that it was known from

previous studies19 that the interaction of anti-D-AA with D-amino acids is strongest around neutral pH. Specific Binding of the Anti-D-amino Acid Antibody to the Hapten-Modified Sensor Surface. The ability of the monoclonal antibody anti-D-AA to specifically bind to the hapten 4-NH2-D-Phe, immobilized on the surface of the gold electrode, was investigated using capacitance measurement. It was assumed that the specific interaction between anti-D-AA and the hapten would result in an additional protein layer and would lead to a further decrease of the total capacitance of the sensor. Figure 3 shows a semilogarithmic plot of the capacitance change as a function of antibody concentration. At concentrations between 10 pg/mL and 100 ng/ mL anti-D-AA, a good linear response was obtained that followed the equation -∆C (nF/cm2) ) 854.0 + 192.9 log C (ng/mL); the correlation coefficient was 0.999. The detection limit was 5.0 pg/ mL with a capacitance change of 300.8 ( 11.43 nF/cm2 (n ) 3). It is significantly lower than the detection limits of 0.5 and 10 ng/ mL, respectively, previously reported for a capacitance-based immunosensor for the detection of hyaluronan-binding protein.14,15 (18) Dijksma, M.; Kamp, B.; Hoogvliet, J. C.; van Bennekom, W. P. Anal. Chem. 2001, 73, 901-907. (19) Hofstetter, O.; Lindstrom, H.; Hofstetter, H. J. Chromatogr., A 2004, 1049, 85-95.

Analytical Chemistry, Vol. 78, No. 21, November 1, 2006

7595

Figure 3. Capacitance changes vs the logarithm of antibody concentration obtained with the hapten-modified sensor for (a) antiD-AA or for (b) commercially available immunoglobulin G. The inset (c) shows the linear range of capacitance changes vs the logarithm of anti-D-AA concentration in the range from 10 pg/mL to 100 ng/mL. All measurements were performed with a potential-step method in 10 mM phosphate buffer, pH 7.0. The incubation time was 25 min. Error bars indicate standard deviation of triplicate determinations; missing error bars are obscured by symbols. Other conditions were the same as in Figure 2B.

In order to confirm that the observed capacitance changes were indeed based on the specific interaction between the electrodeimmobilized hapten and the anti-D-AA and were not simply caused by nonspecific adsorption, control experiments were performed with commercially available immunoglobulin G samples. As seen in Figure 3 (line b), virtually no capacitance changes were detected with the commercial samples. Further control experiments with BSA and diluted human serum samples (1:10 v/v) did not cause any significant capacitance changes either (not shown). These results indicated that the capacitance changes observed with antiD-AA were due to specific antibody-hapten interactions and not due to nonspecific adsorption. The reusability of the sensor was investigated by repeatedly monitoring the binding of anti-D-AA samples (at 50.0 ng/mL) to the same hapten-modified electrode under identical experimental conditions. Following each binding, the electrode was regenerated for 20 min in 100 mM triethylamine to remove bound antibody. The relative standard deviation (RSD) of the capacitance change for eight binding events was found to be 1.9%. Furthermore, 60 different hapten-modified electrodes were prepared in a batch and tested under the same conditions; the RSD of the obtained results was ∼3.5%. These results proved that the hapten-modified sensor was stable under the chosen experimental conditions and allowed sensitive, reproducible, and label-free detection of the antibody using capacitance measurement. Detection of D- and L-Phenylalanine. Enantioselective detection of the model analytes D- and L-Phe was achieved in a competitive setup. For that, anti-D-AA at a fixed concentration was first preincubated with samples of the amino acids before binding of the remaining free (i.e., uncomplexed) antibody was determined by capacitance measurement using the hapten-modified gold electrode. As seen in Figure 4, increasing concentrations of the (20) Finn, M. G. Chirality 2002, 14, 534-540.

7596 Analytical Chemistry, Vol. 78, No. 21, November 1, 2006

Figure 4. Inhibition of the binding of anti-D-AA (at 50.0 ng/mL) to the surface-immobilized hapten (4-NH2-D-Phe) by increasing concentrations of D-Phe (O), L-Phe (b), or D-Phe in the presence of 1 mM L-Phe (2). The capacitance changes were measured by a potentiostatic step method. Error bars indicate standard deviation of triplicate determinations; missing error bars are obscured by symbols. Other conditions were the same as in Figure 2B. D-enantiomer

increasingly inhibited binding of the antibody to the electrode-immobilized hapten, and a typical sigmoidal inhibition curve was obtained in a semilogarithmic plot at concentrations ranging from 0.01 µM to 1 mM. In contrast, the L-enantiomer was not bound by the antibody and, therefore, did not cause any significant inhibition. This result is in good agreement with previous studies that reported a very high degree of stereoselectivity for this antibody.10,17,19 Inhibition values obtained in multiple, independent experiments were highly reproducible and deviated by less than 2% (n ) 3). The exquisite stereoselectivity of antiD-AA also allowed detection of the D-enantiomer in the presence of a large excess of the L-enantiomer. Using capacitance measurement, enantiomeric impurities as low as 0.01 µM D-Phe could be detected in nonracemic samples containing L-Phe at a concentration of 1 mM. The detection limit of 0.001% is ∼1 order of magnitude lower than the limit typically attainable with standard techniques such as chiral chromatography.5,10,20 While the experiments reported in this article were conducted with a particular antibody-antigen system, the experimental strategy can easily be applied to other detection problems. Since the described label-free electrochemical immunosensor combines high sensitivity with simplicity and convenience, it should be useful not only for the enantioselective detection of amino acids as described here but also for investigating other targets (e.g., drugs) using appropriate antibodies. ACKNOWLEDGMENT This work was supported by NSFC (20675018, 20335040, 20605005), Huoyindong Foundation (94009), TRAPOYT, and 0452nm003. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review May 8, 2006. Accepted September 8, 2006. AC060840H