Anal. Chem. 2000, 72, 2963-2968
Operation of a Miniature Redox Hydrogel-Based Pyruvate Sensor in Undiluted Deoxygenated Calf Serum Nenad Gajovic,*,† Gary Binyamin,‡ Axel Warsinke,† Frieder W. Scheller,† and Adam Heller‡
Institute of Biochemistry and Molecular Physiology, Potsdam University, c/o Biotechnologiepark Luckenwalde, D-14943 Luckenwalde, Germany, and Department of Chemical Engineering, The University of Texas at Austin, Austin, Texas
An amperometric sensor for the detection of pyruvate in biological fluids was formed by modifying the tip of a 0.25mm gold wire with a layer of electrically “wired” recombinant pyruvate oxidase (POP). The sensor did not require O2 for its operation. The electroactive area of the tip of the microwire was increased by electrodeposition of platinum black. The POP was adsorbed on the platinum black and then “wired” with the cross-linked, subsequently deposited poly(4-vinylpyridine), part of the pyridine functions of which were complexed with [Os(bpy)2Cl]+/2+ and part quaternized with 2-bromoethylamine. In the resulting thin layer the POP was well “wired”. When the electrode was poised at 0.4 V vs Ag/ AgCl, the sensitivity at pH 6 was 0.26 A cm-2 M-1 and the current increased linearly with the pyruvate concentration through the 2 × 10-6-6 × 10-4 M range. Thiamine diphosphate, flavin adenine dinucleotide, and MgCl2 were not required for the assay, but stabilized the stored enzyme electrode. Placement of a dialysis membrane (MWCO 3500) on the electrode alleviated the severe interference of ascorbate. In calf serum, the detection limit was 30 µM, suggesting that the electrode might be used in the continuous monitoring of pyruvate in hypoxic organs. As the end product of glycolysis, pyruvate is a key diagnostic metabolite. In aerobic eucaryotic cells it is converted to acetylCoA and metabolized through the tricarboxylic acid cycle. Under low oxygen tension, a significant portion of the pyruvate is reduced to L-lactate. The value of L-lactate as an indicator of the status of patients under intensive care has long been recognized. The concentration of lactate increases under hypoxia, poor perfusion of tissue, circulatory shock, and liver failure. A severe (∼10-fold) increase is a prime indicator of lethal risk.1,2 In severe and acute hypoxia, acute liver or renal failure the lactate/pyruvate ratio (L/P) is, however, a better clinical indicator of status than the * Corresponding author: (fax) ++49-3371-681324; (e-mail)
[email protected]. † Potsdam University. ‡ The University of Texas at Austin. (1) Lentner C., uEd. Geigy Scientific Tables: Vol.3 Physical Chemistry, Composition of Blood, Hematology, Somatometric Data; Ciba-Geigy Ltd.: Basle, 1984. (2) Greiling, H., Gressner A. M., Eds. Lehrbuch der klinischen Chemie und Pathobiochemie; Schattauer: Stuttgart, New York, 1989. 10.1021/ac991021i CCC: $19.00 Published on Web 05/23/2000
© 2000 American Chemical Society
concentration of L-lactate.2 The underlying reason for this is the narrow (40-100 µM) range of blood pyruvate in healthy individuals. The normal L/P range is 10-20, while >30 is considered as elevated. Elevated blood pyruvate concentration has been reported in cases of vitamin B1 deficiency, respiratory alkalosis, poisoning by arsenic and mercury, and liver diseases.1 While continuous L-lactate monitoring is already applied in intensive care medicine, in vivo monitoring of pyruvate has not been reported. The in vivo assay of pyruvate is more difficult than that of L-lactate because the clinically relevant range of blood pyruvate is lower, and because the range overlaps, and can be even lower than, the interfering ascorbate concentration range. Consequently, the required assay must be more sensitive and more selective than the sensing of lactate. Native pyruvate oxidase (POP; EC 1.2.3.3) is a thiamine diphosphate- (TPP) and Mg2+-containing flavoenzyme. It catalyzes the three-substrate reaction3 POP
pyruvate + orthophosphate + O2 98 acetyl phosphate + CO2 + H2O2 (1)
As seen from eq 1, a POP-based sensor measures the phosphate concentration,10,11 not the pyruvate concentration, when the concentration of phosphate is low and that of pyruvate is high. The amount of TPP consumed is not stoichiometric, varying with the source organism of the enzyme. For stable operation, flavin adenine dinucleotide (FAD) and Mg2+ are usually added.3,4 Because the active tetrameric holoenzyme dissociates in some solutions to an inactive dimer or to an also inactive monomer,4 the activity of the enzyme can change with the composition of a (3) Sedewitz, B.; Schleifer, K. H.; Go¨tz, F. J. Bacteriol. 1984, 160, 273-278. (4) Risse, B.; G. Stempfer, G.; Rudolph, R.; Mo¨llering, H.; Jaenicke, R. Protein Sci. 1992, 1, 1699-1709. (5) Mizutani, F.; Tsuda, K.; Karube, I.; Suzuki, S. Anal. Chim. Acta 1980, 118, 65-71. (6) Kihara, K.; Yasukawa, E.; Hirose, S. Anal. Chem. 1984, 56, 1876-1880. (7) Zapata-Bacri, A. M.; Burstein, C. Biosensors 1987/1988, 3, 227-232. (8) Spohn, U. Proceedings: The 5th World Congress on Biosensors; 1998; Abstr. p 135, (9) Kulys, J.; Wang, L.; Daugvilaite, N. Anal. Chim. Acta 1992, 265, 15-20. (10) Kubo, I.; Inagawa, M.; Sugawara, T.; Arikawa, Y.; Karube, I. Anal. Lett. 1991, 24, 1711-1727. (11) Ikebukuro, K.; Nishida, R.; Yamamoto H.; Arikawa, Y.; Nakamura, H.; Suzuki, M.; Kubo, I.; Takeuchi, T.; Karube, I. J. Biotechnol. 1996, 48, 67-72.
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000 2963
solution. Pyruvate sensors based on immobilized POP were designed for use under aerobic conditions. The assays were performed with Clark-type, H2O2-electrooxidizing electrodes5-7 and with H2O2-electroreducing peroxidase-modified carbon paste electrodes (CPEs).8 Using a methylene green and POP-modified, TPPcontaining CPE, pyruvate was also assayed in the absence of O2.9 A recombinant POP, requiring only phosphate instead of both phosphate and TPP, is now available. Because phosphate is present in millimolar concentrations in most body fluids, the enzyme opens a route to quasi-reagentless pyruvate sensing in vivo. The 6 unit mg-1 specific activity of this POP is, however, much lower than that of glucose oxidase from Aspergillus species or of lactate oxidase from Pediococcus species, the enzymes used for in vivo sensing of glucose and lactate. A reagentless in vivo sensor also requires the electrical connection (“wiring”) of the reaction centers of POP to the electrode through an electronconducting redox hydrogel.12,13 The hydrogels are applied in reagentless enzyme electrodes, which do not consume O2 and contain no leachable mediator. A non-O2-consuming pyruvate sensor, based on the “wiring” of POP with electropolymerized 2-mercaptohydroquinone, has been reported.20 Here we describe a sensor that is ∼100-fold more sensitive and is less sensitive to interference by ascorbate, both properties being essential in the monitoring of pyruvate in vivo. The particular hydrogel used in this work to “wire” the POP, denoted as POs-EA, was formed by cross-linking of partially bromoethylamine-quaternized and partially [Os(bpy)2Cl]+/2+-complexed poly(4-vinylpyridine) (PVP) with poly(ethylene glycol)diglycidyl ether (PEGDGE). POP was coimmobilized in the gel through a relatively mild cross-linking reaction.14,15 A related implantable miniature (5 × 10-4 cm2 masstransporting area) glucose electrode, based on the “wiring” of glucose oxidase, was described earlier. This sensor did not consume oxygen, had no leachable components, and maintained its apparent sensitivity for several days.16 The forming of a sensitive electrode through the “wiring” POP, an enzyme of low specific activity, poses a unique challenge. When the specific activity of the enzyme is low, its weight fraction in the film modifying the electrode must be increased. However, because the enzyme does not conduct electrons (only the hydrated redox polymer is conductive15) the resistivity increases with the weight fraction of the enzyme. When the resistance of the films increases, the thickness of the layer from which electrons are harvested decreases and electrons are collected only from enzyme molecules in the proximity of the surface of the metallic substrate. It now becomes important to increase the surface area of the base electrode in contact with the “wired” enzyme film and to maintain more of the enzyme in the proximity of the conducting carbon or gold surface. In this study, the sensitivity was increased by (12) Gregg, B. A.; Heller, A. J. Phys. Chem. 1991, 95, 5970-5975. (13) Gregg, B. A.; Heller, A. J. Phys. Chem. 1991, 95, 5976-5980. (14) Katakis, I.; Vreeke, M.; Ye, L.; Aoki, A.; Heller, A. In Advances in Molecular and Cell Biology 15B; JAI Press: Greenwich, CT, 1996; pp 391-409. (15) Rajagopalan, R.; Aoki, A.; Heller, A. J. Phys. Chem. 1996, 100, 3719-3727. (16) Cso ¨regi, 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. (17) Katakis, I.; Ye, L.; Heller, A. J. Am. Chem. Soc. 1994, 116, 3617-3719. (18) Khan, G. F.; Wernet, W. Anal. Chim. Acta 1997, 351 (1-3), 151-158. (19) Kim, C. S.; Oh, S. M. Electrochim. Acta 1996, 41 (15), 2433-2439. (20) Arai, G.; Noma, T.; Habu, H.; Yasumori, I. J. Electroanal. Chem. 1999, 464, 143-148.
2964
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
increasing the specific area of the conductor through deposition of platinum black and by distributing the enzyme nonuniformly in the redox hydrogel, to increase its weight fraction in the proximity of the metal. Through such modification of the electrode, a miniature pyruvate electrode having about a fourth of the sensitivity of the related glucose electrode was built, even though the specific activity of the pyruvate oxidase was only 1/40th of the specific activity of glucose oxidase.16 When shielded by a dialysis membrane, the sensor maintained the sensitivity it had in buffer also in serum, and its selectivity against ascorbate, the dominant electrooxidizable interfering agent, was improved. Chemicals. Pyruvate oxidase (phosphorylating, EC 1.2.3.3, 5.7 unit mg-1) from Toyobo (Osaka, Japan), TPP (>97%), FAD (>95%), pyruvate (>99%), and platinic chloride (H2PtCl6) from Sigma (St. Louis, MO), 2-bromoethylamine hydrobromide from Aldrich (Milwaukee, WI), and PEGDGE 400 and PVP (50 000 Da) from Polysciences (Warrington, PA) were used as received. All other chemicals were of reagent grade. Reagents and Buffers. The stock solution for depositing platinum black was prepared of 0.08 M HCl containing 1% (w/v) H2PtCl6 and 0.03% (w/v) lead(IV) acetate. Unless otherwise stated, So¨rensen-type phosphate buffer (0.067 M KH2PO4 and 0.067 M Na2HPO4), pH 6.0, comprising 2 mM MgCl2, 0.2 mM TPP, and 10 µM FAD was used for preparing the POP stock solutions. The pH was adjusted by changing the relative amounts of the monobasic and dibasic phosphate salts. McIlvaine buffer (0.1 M citrate and 0.1 M Na2HPO4), pH 4.5, was also used to prepare POP stock solutions. Apparatus. An RDE-4 bipotentiostat from Pine Instruments was interfaced to a 386 PC by a low-cost 12-bit AD converter (ADR112) from Ontrak (Ontario, Canada) and used for all cyclic voltammetric (CV) measurements. A home-written DOS program was used for data acquisition and plotting. Currents, down to the picoampere range, were measured with a Biometra EP30 electrochemical detector (Go¨ttingen, Germany). This potentiostat was used for all amperometric experiments with the gold wire microsensor. A water-jacketed (30 °C), stirred electrochemical cell (5-50 mL) from EG&G (Princeton, NJ) with a Ag/AgCl reference electrode (BAS, West Lafayette, IN), a Pt wire counter electrode, and a glass capillary for bubbling N2 was used. The working electrode used for optimizing the coatings was a 3-mm glassy carbon rod in a PTFE Teflon body. The pyruvate microsensor was produced at the 0.25-mm tip of a recessed, polyimide-insulated gold microwire (California Fine Wire). The recess was ∼0.05 mm deep, forming a cavity with a volume of ∼3 nL. Enzyme Electrodes. Preparation of 3-mm-Diameter Electrodes by One-Step Immobilization. The 3-mm-diameter glassy carbon electrodes were polished with 0.3-µm alumina, rinsed, sonicated for 2 min, and rinsed again with deionized water. Enzyme stock solution (50 mg mL-1 in McIlvaine buffer, pH 4.5), POs-EA (10 mg mL-1 in water), and cross-linker PEGDGE (5 mg mL-1 in water) were mixed in various amounts, and 10 µL was pipetted onto the dry electrode. The modified electrode was cured for at least 24 h at room temperature before use. Preparation of 3-mm-Diameter Electrode by Three-Step Immobilization. The polished glassy carbon electrodes were platinized by dipping in the deoxygenated platinic chloride stock solution and poising at -0.3 V (vs Ag/AgCl) for 5 min at room
Table 1: Dependence of the Sensitivity on the Weight Fractions of POP and PEGDGE sensitivity, A cm-2 mol-1 L PEGDGE, %
50% POP
70% POP
80% POP
90% POP
1.5 3 15
0.07 0.07 0.08
0.08 0.14 0.06
0.10 0.20 0.0
0.15 0.14
temperature. The black and sooty electrodes were rinsed with DI water and air-dried. A 10-µL aliquot of a 100 mg mL-1 POP solution (in pH 6 buffer) was pipetted onto the layer and allowed to dry for at least 2 h. The excess nonchemisorbed enzyme was rinsed off with buffer before 5 µL of a freshly prepared POs-EA solution (5 mg mL-1), containing 1 mg mL-1 PEGDGE, was pipetted onto the electrode. The modified electrode was cured for at least 24 h at room temperature. Preparation of 0.25-mm Electrodes. Recessed gold wire electrodes were prepared as reported,16 except that their tip was platinized. POP was adsorbed on the platinum black by filling the recess five times with 100 mg mL-1 POP solution (dissolved in pH 6 buffer with all additives). After each addition, the electrode was allowed to dry for 15 s. The redox polymer was then deposited from the freshly prepared aqueous POs-EA-solution (5 mg mL-1), containing also 1 mg mL-1 PEGDGE, by filling the cavity and allowing the solution to dry (which took less than 10 s). This step was repeated 15 times until the cavity was completely filled, forming a polymer layer of defined thickness. The film was then cured for 48 h before use. The 5 times/15 times filling protocol was applied throughout simply for reproducible results and was not the result of an extensive study. Efforts to speed up the filling protocol by using more concentrated redox polymer stock solutions were hampered by the fact that the polymer easily precipitated inside the microcapillary used to fill the cavity. Optimization of the POP/Redox Polymer Ratio and of the Preparation Procedure. The optimal composition of the threecomponent system formed of POs-EA, cross-linker (PEGDGE), and POP was determined in a two-parameter variation experiment using 3-mm glassy carbon electrodes (see Table 1). In the first batch, the electrodes were made by the one-step immobilization method. The concentration of POs-EA in the coating solution was kept constant at 5 mg mL-1, while the PEGDGE concentration was varied between 0.5 and 5 mg mL-1 and the enzyme loading between 50% and 90% (w/w). In the second batch, POP was immobilized to platinized glassy carbon electrodes and wired using the three-step method. This batch also contained, for comparison, platinized electrodes made by the optimized one-step method. The electrodes were characterized through cyclic voltammetry (-0.05 to +0.45 V vs Ag/AgCl, 20 mV s-1), and their calibration curves were measured through the 0.2-4 mM in deoxygenated phosphate buffer. Preparation and Characterization of the Miniaturized Pyruvate Sensor. The miniaturized pyruvate sensors (Figure 1) were prepared in a three-step procedure similar to that applied in making the 3-mm-diameter platinized carbon electrodes. The pH characteristics and the storage stability (at room temperature) were determined, calibration curves being obtained in argon- and air-saturated, pH 6, 0.07 M phosphate buffer and, additionally, in
Figure 1. Schematic drawing of the microsensor: (a) gold wire (diameter 0.25 mm), (b) polyimide cladding, (c) platinum black (∼5 µm thick) with adsorbed enzyme, (d) redox hydrogel (∼50 µm thick). The cavity at the tip was ∼50-60 µm deep and filled completely with redox polymer.
the same buffer, with 2 mM MgCl2, 0.2 mM TPP, and 10 µM FAD added. Phosphate sensitivity was evaluated in pH 6.0 HEPES buffer in the presence of excess (2 × 10-3 mM) pyruvate. Ascorbate Interference and Its Reduction. Because interference by ascorbate was more severe than by other electrooxidizable interferences, the study of interferring agents was focused on ascorbate. Ascorbate and pyruvate calibration curves were obtained for the “wired” POP sensor and for sensors shielded by a dialysis membrane (Spectrapor 3, MWCO 3500 Spectrum, Houston, TX) on which cellulose acetate and Nafion overlayers were cast in some of the experiments. The dialysis membrane was held in place by an O-ring with the 3-mm PTFE body glassy carbon electrodes. With the 0.25-mm wire electrodes, a homemade “dialysis membrane holder” was made: a blunt syringe needle (with an inner diameter of 0.5 mm) tightly fitted into a piece of Tyvek tubing of equal length with an outer diameter of 3 mm. The tip was covered with a piece of dialysis membrane held in place by a small O-ring. The wire electrode was inserted and fixed with its tip gently pressing against the dialysis membrane. The ratio pyruvate sensitivity/ascorbate sensitivity (P/A) was used as the figure of merit for the selectivity of the sensors. The interference of 1 mM ascorbate and the loss of sensitivity caused by calf serum (5% v/v) were determined. Pyruvate Assay in Serum. Pyruvate, 0.03-0.7 mM, was assayed in deoxygenated undiluted calf serum using the membranecovered microsensor. The same experiment was repeated in undiluted serum spiked with 0.08 mM ascorbate. RESULTS AND DISCUSSION “Wiring” of POP at High Enzyme/Polymer Weight Ratio. Table 1 shows the pyruvate sensitivities of sensors prepared by the one-step method with different POP/redox polymer and crosslinker weight ratios. The highest sensitivity (0.2 A cm-2 M-1) was Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2965
Figure 2. Cyclic voltammograms of electrodes made with increasing POP weight fractions:: solid line, 50% w/w; dashed line, 70% w/w; dotted line, 80% w/w; dash-dotted line:, 90% w/w. 20 mV s-1, pH 6 deoxygenated phosphate buffer.
Figure 3. Calibration curves of pyruvate sensors made by entrapping POP in the redox polymer (open squares) and by its preadsorption on the electrode followed by overcoating the adsorbed enzyme with the “wiring” polymer (solid circles). Deoxygenated phosphate buffer; pH 6, 2 mM Mg2+, 0.2 mM TPP, and 10 µM FAD.
observed for POP 24 mg mL-1 (80% w/w), POs-EA 5 mg mL-1 (17% w/w), and PEGDGE 1 mg mL-1 (3% w/w). Upon further increasing the weight fraction of the enzyme, the sensitivity declined and upper limit of the linear domain of the calibration curve was reduced (not shown). At higher concentrations of crosslinker (5 mg mL-1), pyruvate was detected only above 0.2 mM. Increasing the enzyme loading step by step from 50% to 90% (w/w) reduced the peak oxidation and reduction currents (Figure 2), as expected for dilution of redox centers of the electronconducting hydrogel by an insulating protein. The peak separation and potentials of the reduction and oxidation waves were not significantly affected upon increasing the enzyme’s weight fraction above 50% (w/w), suggesting that the electrons were transported only through a very thin and therefore low-resistance “wired” enzyme film.15 A closer look at the 50% weight fraction data revealed a slightly increased peak separation (as compared to all other voltammograms), indicating an increased resistivity to electron transport. It was not entirely clear if this was the consequence of a higher amount of cross-linking resulting in a lower degree of hydration or just an artifact caused by the apparently inhomogeneous structure of the redox hydrogel: At high enzyme loading (>50% w/w), a precipitated electrostatic adduct was observed, formed of the anionic enzyme (isoelectric point, pH ∼5) and the cationic POs-EA, similar to the precipitate formed of anionic glucose oxidase and POs-EA.17 The precipitation was diminished (but not avoided) when the POP stock solution was prepared in pH 4.5 McIlvaine buffer, in which the protein was less anionic. Khan and Wernet18 have shown that hydrophilic platinum black is a suitable support for sensitive enzymes in amperometric biosensors. Despite its high specific surface, the platinum black electrode has little electrochemical noise or residual current.19 The sensitivity was increased from 0.2 to 0.24 A cm-2 M-1 and a broader linear range was observed when, instead of premixing the redox polymer and POP, the enzyme was first chemisorbed on platinum black, then over-coated, and wired with a solution of POs-EA and cross-linker (Figure 3). Although the weight fraction
of enzyme in the current-producing layer was high, the layer was obviously thin enough for the resistance to remain low and for the enzyme to be electrically well-connected to the electrode. Miniature Pyruvate Sensors. Because the 3-mm-diameter electrodes made by first adsorbing the enzyme on platinum black electrode and coating with the polymeric “wire” were slightly more sensitive, the miniature POP sensors were made by this procedure. The small sensors (Figure 1) were characterized while poised at 0.4 V (vs Ag/AgCl), where their current was at its plateau. The pH dependence of the pyruvate electrooxidation current through the pH 5.9-7.5 range resembled that of H2O2 generation by the dissolved enzyme (Figure 4, circles and dotted line), both having their maximums at pH ∼6.0, far below the reference range of human blood (pH 7.3-7.4). When stored in buffer containing Mg2+, TPP, and FAD at room temperature, the sensor’s half-life was 6.5 days, and the useful lifetime was up to 9 days. The calibration curves did not depend on the presence of TPP, FAD, and Mg2+ (Figure 5). The sensitivity of the miniature electrode in deoxygenated buffer was 0.26 A cm-2 M-1, similar to the 0.24 A cm-2 M-1 sensitivity of the 3-mm electrode, and well above the sensitivities of previously reported POP-based pyruvate electrodes.20 Upon bubbling of nitrogen instead of air through the solution, the sensitivity increased only by ∼8%, showing that electron transport from the redox centers of the enzyme to those of the redox hydrogel was fast relative to electron transport to O2, the natural cosubstrate of POP. The response was linear through the 2 µM-0.6 mM pyruvate concentration range. The 0%-99% response time was 70 s. The sensor’s apparent Michaelis-Menten constant, Kmapp, was 0.8 mM, about twice the 0.34 mM Km of the dissolved enzyme, indicating that the electroxidation of pyruvate was not kinetically but diffusionally limited. The same trend was seen in the calibration curve for phosphate. The apparent electrochemical Km was 3.7 mM phosphate, again about twice the 2.3 mM (phosphate) Km for the dissolved enzyme. Phosphate was detected with a sensitivity of 12 mA cm-2 M-1, the current increasing linearly with phosphate concentration through the 0.2-3 mM range.
2966 Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
Figure 4. Dependence of the current of the miniature sensors without (circles) and with a dialysis membrane (squares) on the pH. The membrane renders the response independent of pH throughout the reference range of human blood (pH 7-7.5). The dotted line shows the pH dependence of dissolved POP. 0.25 mM pyruvate; deoxygenated phosphate buffer; 2 mM Mg2+, 0.2 mM TPP, and 10 µM FAD.
Figure 6. Pyruvate calibration curves in buffer (without additives) and in serum. Open circles, no membrane, no serum; cross-centered circles, no membrane, 5% v/v serum; open squares, with dialysis membrane, no serum; cross-centered squares, with membrane, 5% (v/v) serum; solid squares, with membrane, 100% serum. The membrane eliminates the detrimental effect of serum. Table 2: Pyruvate and Ascorbate Sensitivities of Electrodes without and with Membranes membrane none dialysis, MWCO 3500 + cellulose acetate layer + Nafion layer
Figure 5. Identical pyruvate calibration curves of the miniature POP sensor using phosphate buffer with cofactors added (solid circles) and plain phosphate buffer (open circles). When covered with an additional dialysis membrane, the response drops by a factor of 10 (solid squares).
Interference by Ascorbate and Its Reduction by a Dialysis Membrane. Table 2 shows the ascorbate and pyruvate sensitivities of the miniature electrode without and with membranes separating the miniature sensor from serum. The membrane was mounted right in front of the redox hydrogel (as described in the Experimental Section) forming an “outer” mass-transfer barrier. The best selectivity of pyruvate over ascorbate was obtained with a simple low molecular weight cutoff dialysis membrane. Overcoating of the dialysis membrane with cellulose acetate and Nafion resulted in blockage of the transport of pyruvate. The nonovercoated, plain dialysis membrane reduced the pyruvate sensitivity
pyruvate, A cm-2 mol-1 L
ascorbate, A cm-2 mol-1 L
pyruvate/ ascorbate
0.246 0.043
0.25 0.02
∼1 ∼2
0.005
0.01
∼0.5
∼0
10-fold and raised the detection limit from 2 to 5-10 µM in both buffer and serum. Correspondingly, the linear response range increased from 0.6 to 15 mM pyruvate. The dialysis membrane also diminished the detrimental effect of calf serum. For the sensor without the membrane, the pyruvate (or ascorbate) response in 5% serum (v/v) was only 60% of that in buffer. In contrast, the response of the sensor with the dialysis membrane was nearly the same in buffer and in 100% serum (v/v). (Figure 6, only pyruvate shown). Although the gross current was much higher with 1 mM ascorbate, the presence of ascorbate did not alter the sensitivity to pyruvate either in buffer or in serum, whether or not a dialysis membrane was placed between the sensor and the assayed solution. The pyruvate electrooxidation current of the membrane-separated electrode was practically independent of pH in the 5.7-7.5 range (Figure 4, squares), differing in this respect from the current produced by the sensor without the membrane. Aswithin a given rangespH-independent response is common with enzyme electrodes and occurs if the response is entirely diffusion-limited,21 here accomplished by using an outer masstransfer barrier. This is of particular significance, because the normal blood pH, which is 7.3-7.4, drops in lactic acidosis (to pH ∼7) and this must not have an impact on the sensor’s response. So several beneficial effects of the dialysis membrane were (21) Scheller, F. W.; Schubert, F. Biosensors; Elsevier: New York, 1992; p 62.
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2967
if the ascorbate concentration in the analyte is known or is measured. If it is not known, then two-point calibration is likely to be required, because the ascorbate electrooxidation current, at its typical 80 µM physiological concentration, is similar to that for electrooxidation of pyruvate at 50 µM concentration. Nevertheless, because the incremental current of pyruvate electrooxidation is independent of, and is additive to, the ascorbate electrooxidation current, it should be possible to continuously monitor the pyruvate concentration even in vivo, as long as the ascorbate concentration is also monitored. A target of future research toward in vivo monitoring of pyruvate is the elimination of the residual electrooxidation of ascorbate, which would make possible the precalibration of the sensors in vitro. Such improvement might be based on a suggestion of Gorton,22 who showed that ascorbate electrooxidation is reduced when a mediator’s redox potential is lowered below the potential of the Ag/AgCl electrode and the electrode is poised at about this potential. Figure 7. Pyruvate calibration curves of the dialysis membrane covered miniature sensor in undiluted, deoxygenated calf serum. Squares, without added ascorbate; triangles, with 0.08 mM ascorbate added. The dotted line illustrates the actual response upon increasing the pyruvate concentration by 0.033 mM in each step.
identified: the protection of the sensing layer from detrimental serum ingredients by size exclusion, a mass-transfer-limited response, resulting in an increased linear range and a pHindependent response, and, finally, an improved pyruvate selectivity. This comes at the expense of a slightly increased response time (90 s as compared with 70 s with the plain electrode) and 10-fold reduced sensitivity (but still 1 order of magnitude above the pyruvate electrode described by Arai20). Pyruvate Assay in Calf Serum. The response of the microsensor through the physiological pyruvate concentration range (0.03-0.1 mM) in deoxygenated, undiluted calf serum is shown in Figure 7. After the initially alkaline pH of the serum (pH 8.3) was brought to its physiological value (pH 7.4) by adding solid KH2PO4, the sensor tracked even the lowest physiological pyruvate concentrations. The response dropped by only 10-12% during the 10-h test period. Because the sensor’s calibration curves in calf serum and in phosphate buffer were practically identical, it might be possible to precalibrate the sensor in buffer for use in serum, (22) Gorton, L. Electroanalysis 1995, 7 (1), 23-45.
2968
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
SUMMARY In contrast to previous electrochemical pyruvate biosensors, the “wired” recombinant POP electrode allowed assay of pyruvate under anaerobic solutions, with phosphate as the only essential cosubstrate. The sensitivity of the sensor was higher than that of earlier ones, reaching about a fourth of that of the “wired” glucose oxidase sensor, even though the specific activity of the enzyme was only 1/40th that of glucose oxidase from Aspergillus. The sensitivity was ∼2 orders of magnitude higher than that observed when poly(2-mercaptohydroquinone) was used to wire the POP.20 The characteristics of the 0.25-mm-diameter sensors were reproducible. When covered with a dialysis membrane, they tracked the pyruvate concentration in undiluted calf serum through the 0.1-5 mM range and were not affected by serum ingredients or changes in pH. ACKNOWLEDGMENT The support of the Welch Foundation is gratefully acknowledged.
Received for review September 3, 1999. Accepted March 30, 2000. AC991021I
