Integration of Enzymes and Electrodes: Spectroscopic and

Such optimization studies demonstrated that the most sensitive films for glucose ..... on biosensors, which utilize other enzymes that frequently are ...
0 downloads 0 Views 153KB Size
Anal. Chem. 2002, 74, 5039-5046

Integration of Enzymes and Electrodes: Spectroscopic and Electrochemical Studies of Chitosan-Enzyme Films Xin Wei, Juan Cruz, and Waldemar Gorski*

Department of Chemistry, University of Texas at San Antonio, San Antonio, Texas 78249-0698

A new film-forming solution was developed for the efficient immobilization of enzymes on solid substrates. The solution consisted of a biopolymer, chitosan (CHIT), that was chemically modified with a permeability-controlling agent, Acetyl Yellow 9 (AY9), using glutaric dialdehyde (GDI) as a molecular tether. A model enzyme, glucose oxidase (GOx), was mixed with the CHIT-GDI-AY9 solution and cast on the surface of platinum electrodes to form robust CHIT-GDI-AY9-GOx films for glucose biosensing. UVvisible and infrared spectroscopies were used to determine the composition of the films. The optimized films contained on average 1 molecule of AY9/3 glucosamine units of chitosan and 25 free GDI tethers/1 molecule of GOx. The electrochemical assays of the films indicated both a very high efficiency of enzyme immobilization (∼99%) and large enzyme activity (60 units cm-2). The latter translated into a high sensitivity (42 mA M-1 cm-2) of the Pt/CHIT-GDI-AY9-GOx biosensor toward glucose. The biosensor operated at 0.450 V, had a fast response time (t90% e 3 s), and was free of typical interferences, and its dynamic range covered 3 orders of magnitude of glucose concentrations. The lowest actually detectable concentration was 10 µM glucose. In addition, the biosensor displayed a practical shelf life and excellent operational stability; e.g. its response was stable during 24-h testing under continuous polarization and continuous flow of 5.0 mM glucose solution. The proposed approach to enzyme immobilization is simple, efficient, and cost-effective and should be of importance in the development of biosensors based on other enzymes that are more expensive than glucose oxidase. The design of selective and catalytic molecular systems for bioanalytical applications has become one of the dominant research topics in contemporary electrochemistry. Enzyme electrodes constitute one class of such systems. They integrate the inherent selectivity, or specificity, of enzymatic reactions with the highly efficient electrochemical transduction of the analytical signal. However, the structural design of such biosensors presents several challenges concerning their stability and susceptibility to interferences. An effective integration of electrodes and enzymes must be provided in order to comply with the morphological * Corresponding author: (fax) 210-458-7428; (e-mail) [email protected]. 10.1021/ac020216e CCC: $22.00 Published on Web 08/21/2002

© 2002 American Chemical Society

demands of these fragile biological entities. These requirements have been addressed in the past, with varying results, by immobilizing enzymes using covalent bonding,1-4 bioaffinity attachment,5-9 organic polymers,10-32 entrapment in redox gels,33-40 (1) Bianco, P.; Haladjian, J.; Bourdillon, C. J. Electroanal. Chem. 1990, 293, 151-163. (2) Sasso, S. V.; Pierce, R. J.; Walla, R.; Yacynych, M. Anal. Chem. 1990, 62, 1111-1117. (3) Cass, A. E. G.; Davis, G.; Francis, G. D.; Hill, H. A. O.; Aston, W. J.; Higgins, J. I.; Plotkin, E. V.; Scott, L. D. L.; Turner, A. P. F. Anal. Chem. 1984, 56, 667-671. (4) Kamin, R. A.; Wilson, G. S. Anal. Chem. 1980, 52, 1198-1205. (5) Pentano, P.; Kuhr, W. G. Anal. Chem. 1993, 65, 623-629. (6) Bourdillon, C.; Demaille, C.; Gueris, J.; Moiroux, J.; Saveant, J.-M. J. Am. Chem. Soc. 1993, 115, 12264-12269. (7) Anicet, N.; Bourdillon, C.; Moiroux, J.; Saveant, J.-M. J. Phys. Chem. B 1998, 102, 9844-9849. (8) Katz, E.; Heleg-Shabtai, V.; Bardea, A.; Willner, I.; Rau, H. K.; Haehnel, W. Biosens. Bioelectron. 1998, 13, 741-756. (9) Cosnier, S.; Lepellec, A. Electrochem. Acta 1999, 44, 1833-1836. (10) Chen, X.; Matsumoto, N.; Hu, Y.; Wilson, G. S. Anal. Chem. 2002, 74, 368372. (11) Hodak, J.; Etchenique, R.; Calvo, E. J.; Singhal, K.; Bartlett, P. N. Langmuir 1997, 13, 2708-2716. (12) Yon Hin, B. F. Y.; Smolander, M.; Crompton, T.; Lowe, C. R. Anal. Chem. 1993, 65, 2067-2071. (13) Palmisano, F.; Zambonin, P. G.; Centonze, D. Fresenius J. Anal. Chem. 2000, 366, 586-601. (14) Karyakin, A. A.; Kotel’nikova, E. A.; Lukachova, L. V.; Karyakina, E. E.; Wang, J. Anal. Chem. 2002, 74, 1597-1603. (15) Matsumoto, T.; Ohashi, A.; Ito. N.; Fujiwara, H.; Matsumoto, T. Biosens. Bioelectron. 2001, 16, 271-276. (16) Cosnier, S.; Fologea, D.; Szunerits, S.; Marks, R. S. Electrochem. Commun. 2000, 2, 827-831 and 851-855. (17) Berlin, P.; Klemm, D.; Tiller, J.; Rieseler, R. Macromol. Chem. Phys. 2000, 201, 2070-2082. (18) Zhang, Y.; Hu, Y.; Wilson, G. S.; Moatti-Sirat, D.; Poitout, V.; Reach, G. Anal. Chem. 1994, 66, 1183-1188. (19) Jung, S.-K.; Wilson, G. S. Anal. Chem. 1996, 68, 591-596. (20) Ward, K. W.; Jansen, L. B.; Anderson, E.; Reach, G.; Klein, J.-C.; Wilson, G. S. Biosens. Bioelectron. 2002, 17, 181-189. (21) Sternberg, R.; Bindra, D. S.; Wilson, G. S.; Thevenot, D. Anal. Chem. 1988, 60, 2781-2786. (22) Beh, S. K.; Moody, G. J.; Thomas, J. D. R. Analyst 1989, 114, 1421-1425. (23) Coche-Guerente, L.; Deronzier, A.; Mailley, P.; Moutet, J.-C. Anal. Chim. Acta 1994, 289, 143-153. (24) Situmorang, M.; Gooding, J. J.; Hibbert, D. B. Anal. Chim. Acta 1999, 394, 211-223. (25) Ryan, M. R.; Lowry, J. P.; O’Neill, R. D. Analyst 1997, 122, 1419-1424. (26) Zhang, Z.; Liu, H.; Deng, J. Anal. Chem. 1996, 68, 1632-1638. (27) Bindra, D. S.; Zhang, Y.; Wilson, G. S.; Sternberg, R.; Thevenot, D. R.; Moatti, D.; Reach, G. Anal. Chem. 1991, 63, 1692-1696. (28) Malitesta, C.; Palmisano, F.; Torsi, L.; Zambonin, P. G. Anal. Chem. 1990, 62, 2735-2740. (29) Foulds, N. C.; Lowe, C. R. J. Chem. Soc., Faraday Trans. 1 1986, 82, 12591264.

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002 5039

sol-gel-derived glasses,41,42 carbon pastes,43-45 and carbonpolymer electrodes.46,47 The enzyme immobilization strategy selected in the present investigation relies on using a protective matrix consisting of the biopolymer, chitosan. Chitosan is a linear copolymer of glucosamine and N-acetylglucosamine units. It displays an excellent film-forming ability, good adhesion, biocompatibility, high mechanical strength, and a susceptibility to chemical modifications due to the presence of reactive hydroxyl and amino functional groups.48 It has been used previously as a support for immobilization of enzymes.49-51 However, chitosan’s applications in the development of enzyme-based amperometric biosensors have been very limited.52-56 The goal of this research is to develop single-film coatings for sensitive and interference-free electrochemical biosensors. In this paper, we describe a synthesis of such coatings using film-forming solutions composed of a chitosan modified with a permeabilitycontrolling agent and a redox enzyme. This paper focuses on spectroscopic and electrochemical investigations of the films’ composition, catalytic activity, and analytical performance in amperometric glucose biosensing. EXPERIMENTAL SECTION Reagents. Chitosan (CHIT, MW ∼1 × 106; 75-85% deacetylation), glucose oxidase (GOx) from Aspergillus niger (EC 1.1.3.4; type X-S; 245 900 units g-1), Acetyl Yellow 9 dye (disodium 4-aminoazobenzene-3,4′-disulfonate, AY9), glutaric dialdehyde (50 (30) Schuman, W. Microchim. Acta 1995, 121, 1-29. (31) Trojanowicz, M.; Krawczynski vel Krawczyk, T. Microchim. Acta 1995, 121, 167-181. (32) Gavalas, V. G.; Chaniotakis, N. A.; Gibson, T. D. Biosens. Bioelectron. 1998, 13, 1205-1211. (33) Ohara,-T. J.; Rajagopalan, R.; Heller, A. Anal. Chem. 1994, 66, 2451-2457. (34) Binyamin, G.; Cole, J.; Heller, A. J. Electrochem. Soc. 2000, 147, 27802783. (35) Binyamin, G.; Chen, T.; Heller, A. J. Electroanal. Chem. 2001, 500, 604611. (36) Maidan, R.; Heller, A. Anal. Chem. 1992, 64, 2889-2896. (37) Csoregi, E.; Quinn, C. P.; Schmidtke, D. W.; Lindquist, S.-E.; Pishko, M. V.; Ye, L.; Katakis, I.; Hubbell, J. A.; Heller, A. Anal. Chem. 1994, 66, 31313138. (38) Chen, T.; Friedman, K. A.; Lei, I.; Heller, A. Anal. Chem. 2000, 72, 37573763. (39) Csoregi, E.; Schmidtke; Heller, A. Anal. Chem. 1995, 67, 1240-1244. (40) Bu, H. Z.; English, A. M.; Mikkelsen, S. R. Anal. Chem. 1996, 68, 39513957. (41) Dave, B. C.; Dunn, B.; Valentine, J. S.; Zink, J. I. Anal. Chem. 1994, 66, 1120A-1127A. (42) Sampath, S.; Lev, O. J. Electroanal. Chem. 1997, 426, 131-137. (43) Matuszewski, W.; Trojanowicz, M. Analyst (London) 1988, 113, 735-738. (44) Wang, J.; Lu, F. J. Am. Chem. Soc. 1998, 120, 1048-1050. (45) Schumacher, J. T.; Munch, I.; Richter, T.; Rohm, I.; Bilitewski, U. J. Mol. Catal. 1999, 7, 67-76. (46) Wang, J.; Varughese, K. Anal. Chem. 1990, 62, 318-320. (47) Cespedes, F.; Alegret, S. Trends Anal. Chem. 2000, 19, 276-285. (48) Muzzarelli, R. A. A. Chitin; Pergamon Press: Oxford, U.K., 1977. (49) Krajewska, B. Acta Biotechnol. 1991, 11, 269-277. (50) Spagna, G.; Andreani, F.; Salatelli, E.; Romagnoli, D.; Pifferi, P. G. Process Biochem. 1998, 33, 57-62. (51) Taniai, T.; Sakuragawa, A.; Okutani, T. Anal. Sci. 2000, 16, 517-521. (52) Hikima, S.; Kakizaki, T.; Taga, M.; Hasebe, K. Fresenius J. Anal. Chem. 1993, 345, 607-609. (53) Ng, L.-T.; Yuan, Y. J.; Zhao, H. Electroanalysis 1998, 10, 1119-1124. (54) Miao, Y.; Chia, L. S.; Goh, N. K.; Tan, S. N. Electroanalysis 2001, 13, 347349. (55) Okuma, H.; Watanabe, E. Biosens. Bioelectron. 2002, 17, 367-372. (56) Wang, G.; Xu, J.-J.; Ye, L.-H.; Zhu, J.-J.; Chen, H.-Y. Bioelectrochemistry 2002, 57, 33-38.

5040

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

wt % solution in H2O), ninhydrin reagent solution, sodium cyanoborohydride, hexaammineruthenium(III) chloride, glucose, L-ascorbic acid, uric acid, and acetaminophen were purchased from Sigma-Aldrich. Other chemicals, NaH2PO4‚H2O, NaCl, and NaOH were from Fisher. Glutaric dialdehyde (GDI) was determined to be free of oligomers by UV-visible spectroscopy.57 A 0.10 wt % chitosan solution was prepared by dissolving chitosan flakes in hot 0.05 M HCl. After pH adjustment to ∼5.4, the solution was filtered using a 0.45-µm Millex-HA syringe filter unit (Millipore). A glucose stock solution (1.0 M) was prepared in pH 7.40 phosphate buffer (0.10 M) and was allowed to mutarotate for 24 h at room temperature before use. Both chitosan and glucose solutions were stored in a refrigerator when not in use. All solutions were prepared using deionized water that was purified with a Barnstead NANOpure cartridge system. Electrochemical Measurements. The BAS (Bioanalytical Systems, Inc.) model 100B/W workstation was used to collect electrochemical data. Experiments were performed at room temperature (21 ( 1 °C) in a conventional three-electrode system with 1.6-mm-diameter platinum disk working electrode (BAS), platinum wire as the auxiliary electrode, and the Ag/AgCl/ 3MNaCl (BAS) reference electrode. The oxygen concentration in solutions was monitored using the YSI 5100 dissolved oxygen meter. The platinum disk electrodes were wet-polished prior to use on a sand paper (3M, 2000 grit) and on an Alpha A polishing cloth (Mark V Lab) with successively smaller particles (0.3- and 0.05-µm diameter) of alumina. The slurry accumulated on the electrode surface was removed by ultrasonication for 30 s in deionized water. The pH 7.40 phosphate buffer solution (0.10 M) served as a background electrolyte in all experiments. The experiments were repeated at least three times, and the means of measurements are presented with the standard deviations. Spectroscopic Measurements. The attenuated total reflection Fourier transform infrared (ATR-FT-IR) spectra were collected with a Bruker Vector 22 spectrometer using a Golden Gate unit. Chitosan films were cast on microscope slides and dried in a gel dryer vacuum system (FisherBiotech) for 5 h before analysis. The measurements were performed using pulverized chitosan films that were firmly pressed on a small (2 × 1 mm) diamond window of the Golden Gate unit. The ATR-FT-IR spectra were collected by averaging 16 scans using a resolution of 1 cm-1. The electronic spectra were recorded with a HP-8453 UVvisible diode array spectrophotometer using a quartz cuvette with a path length of 1.0 cm. Scanning electron microphotographs of film cross sections were made with a JEOL 840A scanning electron microscope. Chemical Modifications of Chitosan with GDI and AY9. CHIT was modified with GDI by adding dropwise the 0.10 wt % CHIT solution to a 50 wt % GDI solution until the molar ratio of GDI to chitosan glucosamine units was 200:1. The mixture was stirred at room temperature for 24 h, and the unreacted GDI was separated from the CHIT-GDI product by multiple extractions with ethyl ether. The extractions continued until the UV-visible spectrum of the organic phase showed no presence of the GDI. (57) Rasmussen, K.-E.; Albrechtsen, J. Histochemistry 1974, 38, 19-26.

Figure 1. Chemical structures of chitosan (A) and Acetyl Yellow 9 (B).

The CHIT-GDI product was further modified with AY9 in order to obtain the CHIT-GDI-AY9 film-forming solution for the preparation of biosensors. To this end, the solutions of CHITGDI and AY9 were mixed to have a molar ratio of glucosamine units to AY9 equal to ∼2:1 and allowed to react overnight at room temperature. The CHIT-GDI-AY9 solution was stored in a refrigerator when not used. Ninhydrin Test. The ninhydrin test58 was used to determine the fraction of amino groups that remained in chitosan after its modification with glutaric dialdehyde. Ninhydrin reacts with primary amino groups to form a colored product, diketohydrindylidene-diketohydrindamine, which absorbs at 570 nm. The test was performed by mixing 1.0 mL of either a modified chitosan (0.1 wt %) or the original unmodified chitosan (0.1 wt %) solution with 2.0 mL of ninhydrin reagent solution (ninhydrin, hydrindantin, dimethyl sulfoxide, and lithium acetate at pH 5.2) purchased from Sigma-Aldrich. The mixture was heated at 85 °C for 30 min and cooled to room temperature. A 1.0-mL aliquot of the mixture was diluted with 2.0 mL of water, and the absorbance at 570 nm was measured using similarly treated water as a blank. Preparation of Chitosan-Enzyme Films. The biosensing films were prepared by placing 20-µL aliquots of a casting solution on the surface of platinum electrodes and drying them at 30 °C for 3 h. The resulting films were circular with ∼5-mm diameter. The casting solutions were prepared by mixing equal volumes of the CHIT-GDI-AY9 solution and GOx solution. A casting solution was applied to the electrode within ∼30 s from the moment of mixing. At longer times, the solution became heterogeneous due to enzyme precipitation. The CHIT-GDI-AY9-GOx film electrodes were stored in air at room temperature when not used. Some of the films were treated with a sodium cyanoborohydride in order to reduce the Schiff base bonds formed during film preparation. The reduction was performed overnight, in a refrigerator at 4 °C, by immersing the film electrodes in a buffer solution that contained 10 mM NaBH3CN.59

RESULTS AND DISCUSSION Chemical Modifications of Chitosan. Chitosan chains (Figure 1A) were modified with anionic Acetyl Yellow 9 dye (Figure 1B) in order to introduce a permselectivity against anions. We hypothesized that films made of such a modified chitosan would allow for electrochemical biosensing free of interferences from the redox-active anions, such as ascorbate and urate, which are commonly present in physiological samples. The modification of chitosan with AY9 was accomplished in two steps. First, a solution of CHIT was reacted with GDI. The reaction involved formation of Schiff base structures60 according to

(58) Prochazkova, S.; Varum, K. M.; Ostgaard, K. Carbohydr. Polym. 1999, 38, 115-122.

(59) Lee, R. T.; Lee, Y. C. Biochemistry 1980, 19, 156-162. (60) Roberts, G. A. F.; Taylor, K. E. Makromol. Chem. 1989, 190, 951-958.

CHIT-NH2 + OdCHs(CH2)3sHCdO f CHIT-NdCHs(CH2)3sHCdO + H2O (1)

To react only one aldehyde group of GDI, as shown above, a high molar ratio of GDI to chitosan glucosamine units (200:1) was used. Under such conditions, the cross-linking of chitosan chains with GDI was avoided as indicated by the lack of gel formation. To characterize the product CHIT-GDI, the unreacted GDI was removed from the solution by multiple extractions with ethyl ether, and the ninhydrin test and infrared analysis were performed. The ninhydrin test showed that the absorbance of the CHIT-GDI solution at 570 nm, A570, was equal to 0.06 ( 0.01 AU. The analogous test that was carried out on the solution of original unmodified chitosan yielded A570 ) 0.40 ( 0.05. These absorbances suggest that ∼85% of the chitosan amino groups reacted with GDI. The presence of free aldehyde groups in the CHIT-GDI product was confirmed by the ATR-FT-IR spectroscopy. Figure 2 shows the infrared spectra of the films that were prepared by evaporating water from solutions of the original unmodified CHIT (curve a) and CHIT-GDI product (curve b). A comparison of

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

5041

Figure 2. ATR-FT-IR spectra of (a) CHIT, (b) CHIT-GDI, (c) CHITGDI-AY9, and (d) CHIT-GDI-AY9-GOx films.

curves a and b reveals that the latter spectrum contains three extra absorption bands. The bands at 2940 and 1450 cm-1 are due to the absorption by methylene groups of GDI. The band at 1720 cm-1 is characteristic for the free aldehyde group. In the second modification step, a fraction of aldehyde groups of the CHIT-GDI was reacted with amino groups of AY9 to produce the CHIT-GDI-AY9 structure. The reaction involved formation of Schiff bases in a manner analogous to eq 1. Because this process consumed aldehyde groups of the CHIT-GDI species, the intensity of the band at 1720 cm-1 in the spectrum of the CHIT-GDI-AY9 film decreased (Figure 2, curve c). Simultaneously, a new absorption band at 1600 cm-1 appeared in the spectrum. This band was due to the absorption by phenyl rings of AY9 molecules. The amount of AY9 immobilized in the CHIT-GDI-AY9 material was determined by subtracting the moles of unreacted AY9 from the total number of moles of AY9 used for the reaction. In a typical procedure, 78 µL of a 0.1 wt % CHIT-GDI solution containing 0.18 µmol of AY9 was placed in the cuvette and allowed to evaporate. Considering that on average 80% of chitosan is deacetylated, the resulting film contained ∼0.39 µmol of glucosamine units (FW ) 161 g mol-1). Then, the film was equilibrated with 3.0 mL of water in order to remove unreacted AY9. The equilibration process was monitored for 48 h by recording UV-visible spectra of the solution above the film. The absorbance peak at 385 nm, which is characteristic for AY9, increased in time and stabilized at 0.48 ( 0.05 AU after 24 h in quiescent solution. No absorption was detected for the second aliquot of water that was used to continue the film equilibration. This indicated that all of the unreacted AY9 was removed from the film by the first aliquot. According to the dye calibration plot, A ) 2.3 × 104CAY9 + 1 × 10-4 (R2 ) 0.999); the absorbance A385 ) 0.48 ( 0.05 AU corresponds to the presence of 0.06 µmol of AY9 in 3.0 mL of the solution. Thus, the amount of AY9 that was immobilized in the chitosan film is equal to 0.12 µmol () 0.180.06). A comparison of this number with the amount of chitosan glucosamine units (0.39 µmol) indicates that the molar ratio of AY9 to monomer units in the film is ∼1:3. The foregoing analysis is summarized in Figure 3, which shows a schematic illustration of the composition and structure of the 5042

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

CHIT-GDI-AY9 material. Figure 3 emphasizes that the material contains ∼85% of glucosamine units modified with GDI and ∼30% of glucosamine units modified with GDI-AY9 side functionalities. A large fraction (∼60%) of glucosamine units was intentionally left modified only with GDI. These GDI molecules were used as tethers for enzyme immobilization in the CHIT-GDI-AY9 films. Enzyme Immobilization in CHIT-GDI-AY9 Films. The enzyme immobilization relied on the reaction between the free aldehyde groups of the GDI tethers and amino groups of GOx to form Schiff bases. That the aldehyde groups were used up in the reaction with GOx was supported by the observation that the intensity of the band at 1720 cm-1 in the infrared spectrum of the CHIT-GDI-AY9-GOx film was greatly reduced (Figure 2, curve d). The presence of enzyme in the film gives rise to characteristic bands at 1656 and 1540 cm-1, which can be assigned to the amide I (CdO) stretching and amide II (NsH) deformation of GOx, respectively. These bands are compatible with those caused by the absorption of the carbonyl (1660 cm-1) and the amide (1560 cm-1) groups of chitosan (curve a). The amount of enzyme in the immobilization mixture was varied in order to obtain the most active films toward the oxidation of glucose. Such optimization studies demonstrated that the most sensitive films for glucose determination were formed by (1) mixing 50.0 µL of 0.1 wt % CHIT-GDI-AY9 solution with 50.0 µL of GOx solution (20 mg mL-1), (2) casting 20.0 µL of the mixture on the electrode surface, and (3) evaporating water for 3 h. Such a procedure yielded robust CHIT-GDI-AY9-GOx surface films with an average thickness of 10.0 ( 0.5 µm in a dry state (Figure 4). Some increase in the thickness of wet films can be expected because of the film swelling in a solution. Figure 4 reveals that the films had a smooth surface without any visible pinholes on the micron scale. The CHIT-GDI-AY9-GOx films that were prepared according to the above procedure contained on average 50 nmol of glucosamine units and 1.2 nmol of GOx (MW ) 160 000). Since ∼60% of the glucosamine units contain free GDI tethers, these nanomole numbers show that there are ∼25 free GDI tethers per 1 molecule of GOx in the surface films. Such films were used in all of the subsequent experiments. GOx Assay in the CHIT-GDI-AY9-GOx Electrode Films. Before the assay, the loosely attached GOx was removed from the films by equilibrating them with a stirred 1.0 mL of buffer solution for 5 h. The assay28 relied on the measurement of the rate of Ru(NH3)62+ generation in the following reaction sequence61

GOx(Ox) + glucose f GOx(Red) + gluconolactone (2) GOx(Red) + 2Ru(NH3)63+ f GOx(Ox) + 2H+ + 2Ru(NH3)62+ (3)

To carry out the assay, a bare glassy carbon electrode was placed in 10.0 mL of an assay solution that was composed of a deoxygenated buffer solution containing 10 mM Ru(NH3)63+ and 0.50 M glucose. The electrode was held at the potential where Ru(NH3)62+ ions are easily oxidized (0.100 V), and the currenttime (I-t) trace was recorded continuously while the solution was (61) Chen, L.; Gorski, W. Anal. Chem. 2001, 73, 2862-2868.

Figure 3. Schematic representation of CHIT-GDI-AY9 species present in the film-forming solution.

Figure 5. Cyclic voltammograms recorded at the CHIT-GDI-AY9GOx film electrode in (a) background electrolyte, (b) 50 mM glucose, and (c) 5 mM H2O2 solutions. Background electrolyte, pH 7.40 phosphate buffer. Scan rate, 5 mV s-1. Figure 4. Scanning electron micrograph of the cross section of the CHIT-GDI-AY9-GOx film.

stirred under an argon blanket. After the background current decayed to a constant level, an electrode coated with the CHITGDI-AY9-GOx film was immersed into the solution. This caused an increase in current, which indicated formation of the Ru(NH3)62+ species as expected from eq 3. An initial slope of the I-t trace was used as a measure of GOx activity in the film. The slope was equal to 1.2 ( 0.2 nA s-1 and did not depend on the increase in the rate of stirring of the solution. The latter suggests that the concentration of the glucose and ruthenium complex was sufficiently high to avoid diffusional limitations to the enzymatic reaction in the thin surface film. The film assays were calibrated by adding known amounts of GOx to 10.0-mL aliquots of assay solutions and measuring the slopes of the I-t traces. Such measurements yielded a calibration plot that had a slope of 4.0 ( 0.5 nA s-1 nmol-1, an intercept of 0.02 ( 0.01 nA s-1, and R2 ) 0.990 up to 1.2 nmol of GOx in the solution. The calibration slope 4.0 ( 0.5 nA s-1 nmol-1 indicated that the slope 1.2 ( 0.2 nA s-1, obtained in the film assay, corresponded to 0.3 nmol of active GOx in the surface film. A comparison of this number with the actual amount of GOx incorporated in the film (1.2 nmol) suggested that ∼25% of the enzyme was active in the surface film. Alternatively, one can state

that the activity of the film-bound enzyme constitutes ∼25% of that of the enzyme dissolved in the solution. The lower activity of the film-bound enzyme probably reflects a lower frequency of effective collisions between the enzyme, glucose, and Ru(NH3)63+ species in the film caused by steric hindrances. Either way, the above analysis shows that the surface film contained ∼60 units cm-2 () 0.3 × 10-9 mol × 160 000 g mol-1 × 245 900 units g-1/ 0.196 cm2) of active GOx. That the GOx did not leach from the CHIT-GDI-AY9-GOx film was confirmed by an independent experiment in which 1.0 mL of the deoxygenated buffer solution, which was equilibrated for 5 h with the freshly prepared film, was injected into 9.0 mL of the assay solution. Under such conditions, the slope of the I-t trace was equal to 0.06 ( 0.01 nA s-1, which implied that ∼99% of the GOx added to the casting solution remained in the surface film. Determination of Glucose at the Pt/CHIT-GDI-AY9GOx Film Electrode. Figure 5 shows cyclic voltammograms recorded at the Pt/CHIT-GDI-AY9-GOx film electrode in the buffer solutions containing either glucose (curve b) or H2O2 (curve c). Both voltammograms show well-developed anodic currents in the same potential window (0.20-0.80 V). This confirms the classical mechanism of glucose detection at the film electrode. The detection is accomplished by the electrooxidation of H2O2, Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

5043

Figure 6. Calibration plot for glucose obtained with the CHIT-GDIAY9-GOx film electrode in a stirred solution. Inset: steady-state response to additions of glucose aliquots of (a) 10, (b) 10, (c) 10, and (d) 20 µM. Background electrolyte, pH 7.40 phosphate buffer. Potential, 0.450 V.

which is produced in the reaction between the oxygen and reduced GOx. The latter is generated in the process of glucose oxidation by GOx. The current plateau on curve b indicates that the determination of glucose at the film electrode can be performed at a potential as low as 0.45 V. The calibration curve for glucose was obtained by measuring steady-state currents (E ) 0.450 V) at the film electrode in a stirred buffer solution that was spiked with increasing concentrations of glucose (Figure 6). The dynamic range covered 3 orders of magnitude of glucose concentrations from 10 µM up to 10 mM. A linear least-squares calibration plot for glucose over the range 10 µM-5.0 mM (12 points) had a slope of 0.042 ( 0.002 A M-1 cm-2, intercept 3 × 10-8 A, and R2 ) 0.995. The biosensor’s sensitivity toward glucose (0.042 A M-1 cm-2) was only ∼7 times lower than its sensitivity to H2O2 (0.290 A M-1 cm-2). This indicated a very efficient signal transduction in the Pt/CHIT-GDI-AY9-GOx biosensor. The present sensor is better than our previous chitosan-based biosensor, which displayed a lower sensitivity (0.014 A M-1 cm-2) and a narrower linear dynamic range (2-10 mM) in air-equilibrated glucose solutions.61 The Eadie-Hofstee plot I versus I/Cglucose, based on the calibration points, yielded a straight line from 0.5 to 10 mM glucose with a R2 ) 0.990. The slope of the line yielded an apparent Michaelis-Menten constant equal to 18 mM, while the intercept produced the maximum current density under conditions of enzyme saturation equal to 1.0 mA cm-2. The relative standard deviations for the slope and intercept were below 10%. The inset in Figure 6 shows a typical amperometric trace recorded during the spiking of the stirred buffer solution with micromolar concentrations of glucose. The trace shows that the response time, defined as the time to 90% of the full signal, was below 3 s even in the case of a very diluted (10 µM) glucose spikes. Such a fast response time suggests a rapid charge and mass transfer in the surface film of the biosensor. The actual detection limit of the biosensor was equal to 10 µM. Glucose spikes at cglucose < 10 µM did not generate any current response from the biosensors. Apparently, at such low concentrations, glucose is oxidized mostly at the film/solution interface and not enough H2O2 reaches the platinum surface to be detected. 5044 Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

Figure 7. Amperometric trace recorded at (1) CHIT-GDI-AY9GOx and (2) CHIT-GDI-GOx film electrodes in a stirred solution during the additions of (a) 1.5 mM glucose, (b) 0.10 mM ascorbic acid, (c) 0.10 mM uric acid, and (d) 0.10 mM acetaminophen. Background electrolyte, pH 7.40 phosphate buffer. Potential, 0.450 V.

According to this scenario, thinner surface films should allow for lower detection limits of glucose if required. Selectivity. Figure 7 shows selectivity studies performed with two biosensors Pt/CHIT-GDI-AY9-GOx (curve 1) and Pt/ CHIT-GDI-GOx (curve 2). The determination of glucose at the film containing AY9 dye (curve 1) is practically free of interferences from redox-active species that are commonly present in physiological samples of glucose. Without the AY9, the glucose sensing is not selective (curve 2). A comparison of the two curves indicates that the AY9 dye does not interfere with the determination of glucose and illustrates the efficiency of an anionic AY9 as a permeability-controlling agent. To quantify this efficiency, we calculated a combined percentage error, EΣ%, that was introduced by 0.1 mM ascorbate, 0.1 mM urate, and 0.1 mM acetaminophen to the biosensor’s signal for 5 mM glucose. Based on curve 1 in Figure 7, the error EΣ% for the Pt/CHIT-GDI-AY9-GOx electrode was equal to only 3%. The dye-modified chitosan is an attractive structural material for the development of biosensors because its single film is capable of both immobilizing an enzyme and controlling the access of species to an electrode surface. Typically, additional permselective layers are coated on biosensors in order to improve their selectivity. Materials used for such layers include cellulose acetate,27 Nafion,62 composite of cellulose acetate and Nafion,18 countercharged polyelectrolytes,38 diaminobenzene,2 mercaptosilane,19 and polyethersulfone stabilized with trimethoxysilane.20 The Pt electrodes that were covered with these materials displayed the error EΣ%, as defined above, equal to ∼55, 10, 17, 5, 5, 3, and 2%, respectively, according to the selectivity data reported (2% was for 0.1 mM acetaminophen only). Interestingly, the cellulose acetate, which is structurally similar to chitosan, was the least effective barrier (EΣ% ) 55%) for the interfering species. In a different approach,36,37 the selectivity of a glucose biosensor has been achieved by oxidizing the interfering species in a layer composed of the cross-linked peroxidase and lactate oxidase. Hydrogen peroxide, however, was necessary for such an oxidative elimination of interferences. (62) Vaidya, R.; Atanasov, P.; Wilkins, E. Med. Eng. Phys. 1995, 17, 416-424.

Figure 8. Amperometric trace recorded at the CHIT-GDI-AY9GOx film electrodes in (a) 5.0 mM and (b) 10 µM glucose solutions that were stirred and slowly deoxygenated with argon. The current axis was normalized by assigning 100% to the largest current recorded in each solution. The percentage numbers in the figure represent the amount of oxygen left in the solution (100% t 8.0 mg of O2 L-1). Background electrolyte, pH 7.40 phosphate buffer. Potential, 0.450 V.

Oxygen Demand. The oxygen demand of the Pt/CHITGDI-AY9-GOx biosensor was investigated at two extreme concentrations of glucose (Figure 8). At 5.0 mM glucose (curve a), the response of the biosensor decreased only when the O2 content in the solution dropped to 4.8 mg L-1, which is equal to ∼60% of the O2 concentration present initially in the airequilibrated solution (8.0 mg L-1). At 10 µM glucose (curve b), the biosensor’s response did not depend on the oxygen content in the whole range of oxygen concentrations that could be generated by purging the solution with argon (down to ∼2% of the initial O2 content). Such behavior defines two practical approaches to glucose analysis with the Pt/CHIT-GDI-AY9GOx biosensor. In one mode, the glucose samples should be either oxygenated or diluted with the air-equilibrated buffer solution before analysis in order to avoid false signals due to fluctuations in oxygen concentration. Alternatively, the biosensor can be used in the noninvasive glucose analysis, e.g., in tears where the glucose concentration is ∼100 lower than in blood. Stability. The stability of the Pt/CHIT-GDI-AY9-GOx film electrodes was investigated in a flow system. The studies were performed with electrodes that were soaked overnight in cold (4 °C) buffer solutions in the presence or absence of sodium cyanoborohydride before the experiments. The NaBH3CN was used to reduce the Schiff bases in the original film to more stable secondary amines. Figure 9 shows current traces recorded at the original film electrode and at the reduced film electrode. The original electrode displayed an essentially constant response under continuous polarization and continuous flow of a 5.0 mM glucose solution for 24 h (curve 1). Under the same conditions, the response of the reduced film electrode was lower and could not stabilize (curve 2). Apparently, the NaBH3CN decreased the activity of a film electrode toward the detection of glucose. The electrodes were calibrated on days 1, 2, 4, 7, 10, and 13 using a series of glucose concentrations ranging from 0.50 to 5.0 mM. The original film electrodes retained ∼95% of their initial sensitivity (slope), which demonstrated a good stability of Schiff base bonds in the chitosan matrix. The sensitivity of the reduced film electrodes decreased to 50% on day 7 and to 40% on day 13.

Figure 9. Amperometric trace recorded at the (1) original and (2) reduced CHIT-GDI-AY9-GOx film electrode in a flow system. Arrows indicate the moment of switching from the glucose-free carrier solution to the solution containing 5.0 mM glucose (a) and from 5.0 mM glucose solution back to glucose free solution (b). Inset: a longterm response of the original CHIT-GDI-AY9-GOx film electrode to 5.0 mM glucose in a flow system. Carrier solution, pH 7.40 phosphate buffer. Flow rate, 0.1 mL min-1. Potential, 0.450 V.

Overall, the reduced films required a longer time to reach a stable signal in their first use for glucose determination and, after the stabilization, displayed half to two-thirds of the sensitivity of the original films. The inset in Figure 9 shows more extended studies of longterm stability of the original film electrode. The points are the average currents recorded at the biosensor, which operated continuously (0.45 V) in a 5.0 mM glucose solution for 4-12 h on a given day. The biosensor retained a constant response of 3.32 ( 0.13 µA for the first 9 days when it was used daily. This current is ∼21% smaller than the one recorded in a stirred solution of 5 mM glucose (Figure 6) because of slower convective transport in a flow system. After 9 days, the response decreased gradually to ∼65% of the initial response and remained constant until the study was interrupted on day 47. From days 15 to 47, the biosensor was used only intermittently and its average signal was equal to 2.10 ( 0.14 µA. The latter illustrated good shelf life of a biosensor. A slow decrease in a biosensor signal between day 9 and day 15 is probably caused by a detachment of chitosan chains from the electrode surface. To alleviate this problem, approaches based on cross-linking of chitosan matrix are being explored. The CHIT-GDI-AY9 film-forming solution was very stable when stored at 4 °C. It was usually used for ∼2 months to prepare series of biosensors. The proposed method for the enzyme immobilization was very reproducible. For example, the relative standard deviation of the glucose signal measured using 10 independently prepared Pt/CHIT-GDI-AY9-GOx electrodes was below 5%. Comparison with Other Amperometric Glucose Biosensors. The enzyme electrode described here combines the simplicity and reproducibility of preparation with reliability of operation. After only 3 h of curing, this single-film electrode can work continuously for at least 70 h (pooled over 9 days) while retaining 95% its initial signal. This signal stability can be ascribed to the biocompatibility of the chitosan matrix and the high enzyme activity in the film (60 units cm-2). Other attractive features of the electrode include a quick response time (t90% e 3 s), freedom Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

5045

Table 1. Comparison of Amperometric Glucose Biosensors Based on a Single Polymeric Layer Comprising Glucose Oxidase sensor assembly (approx time)

enzyme activity, units cm-2

sensitivity, mA M-1 cm-2

stabilitya

linear range

interferencesb

ref

Pt/chitosan film (3 h) Pt/nylon mesh (1 day) Pt/cellulose membrane (3 days) Pt/(poly)diaminobenzene) (1 h) Pt/polypyrrole-biotin (1-2 h) Pt/polypyrrole (2-3 h) vitreous C/Os-hydrogel (1 day) C foil/Fc-polycarbonate membrane (2 days) GC/Prussian Blue/Nafion (3 h)

60 52-138 1-3 3 0.18 0.08-0.3 22h ? 0.5-4

42 ? 4-10 9 7 12-34 18 7 50

70 h (95%)/1.5 mo (95%) 48 h (95%)/4 mo (70%) ?/0.3-5 mo (50%) 20 hc(?)/0.3 mo 2 h (87%)/0.5 mo (50%) 18 tests (66%)/1.4 mo (47%) 60 h (50%)/0.7 mo (50%) 50 h (?)/? 7 tests in 4 h/0.5 mo

10 µM-5 mM 1 µM-5 mM 1 µM-3 mM 200 µMd-3 mM 5 µMd-0.7 mM 0.1 µM-2 mM 100 µMd-6 mM 1-30 mM 0.1 µM-1 mM

n ? ? ne ? AA, UA nf ng ?

i 22 21 28 9 23 33 3 14

a Operational stability (hours)/long-term stability (months, mo). Percentages indicate signal retention. b n, below 5% when 0.1 mM ascorbic acid (AA), uric acid (UA), and acetaminophen were used; ?, not reported. c Signal abruptly decreased after 20 h. d Estimated from the data presented. e Only the AA was reported. f Prolonged urate electrooxidation degraded the sensor. g The AA yielded a 4% increase in the signal for 7 mM glucose. h Enzyme loading efficiency. i This work.

from interferences due to the redox of ascorbate, urate, and acetaminophen, wide dynamic range (10 µM-10 mM glucose), and high sensitivity (42 mA M-1 cm-2). These aspects of our electrode compare well with those of the existing glucose biosensors (Table 1). Due to the extensive body of literature that exists for glucose sensors and the format of this article, Table 1 includes only a few representative examples of biosensors that are most closely related to our electrode, i.e., the biosensors relying on a single polymeric layer comprising the glucose oxidase on the electrode surface. Another selection factor was the availability of the relevant data for the comparison. The selected examples represent some of the current trends in sensor development such as a covalent immobilization of an enzyme on the surface of membranes,21,22 enzyme entrapment in electropolymerized nonconductive,28 conductive,23 and biotinylated polymers,9 enzyme “wiring” in redox polymers,33 covalent attachment of an enzyme to the surface of electrode,3 and enzyme immobilization within a cast polymer layer.14 A survey of data in Table 1 reveals that the chitosan-based biosensor displays the best combination of the high enzyme activity with high stability, sensitivity, and selectivity of glucose biosensing. Thus far, single-film enzyme electrodes have tended to exhibit a limited stability and selectivity. Therefore, the multifilm biosensors have been extensively studied. The multilayer design typically results in a slower response time, lower sensitivity, and a complicated biosensor assembly. However, the multifilm biosensors usually have a higher upper limit of a linear range, which is important for in vivo applications. Examples of such biosensors include a three-layered sensor39 that has been prepared by (1) wiring GOx to the electrode through a redox hydrogel, (2) covering with a mixture of permeability controlling hydrogels, and (3) coating with a biocompatible layer of photo-cross-linked tetraacrylated poly(ethylene oxide). The sensor retained ∼96% of its signal after 168 h of continuous operation, was selective against typical interferences, had a response time of 1.2 min, and displayed a linear range up to ∼40 mM glucose with a sensitivity of 2-5

5046 Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

mA M-1 cm-2. In another recent example,10 the glucose oxidase was (1) electrodeposited on a Pt wire, (2) coated with an antiinterference and protective film of polyphenol, (3) covered with a stability-reinforcing membrane of cross-linked (3-aminopropyl)trimethoxysilane, and (4) protected by the polyurethane outer membrane. Such a sensor exhibited a long-term stability for more than 1.7 months, low interference from endogenous species, short response time (