Anal. Chem. 1998, 70, 1511-1515
Construction and Characterization of Nitrate Reductase-Based Amperometric Electrode and Nitrate Assay of Fertilizers and Drinking Water Scott A. Glazier*
Biomolecular Materials Group, Biotechnology Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899 Ellen R. Campbell and Wilbur H. Campbell
The Nitrate Elimination Company, Inc., 334 Hecla Street, Lake Linden, Michigan 49945
The construction and characterization of a nitrate reductase-based amperometric electrode for determination of nitrate ion is described. The electrode consisted of nitrate reductase held by dialysis membrane onto a Nafion-coated glassy carbon electrode. Methyl viologen was allowed to absorb into the Nafion layer, which acted as a reservoir for the electron mediator. The utility of the electrode to assay fertilizer and water sample for nitrate was demonstrated. The assays conducted with this electrode compared well with colorimetric and potentiometric assays of the same samples. Nitrate ion is an important analyte in such diverse materials as fertilizers, foods, livestock feeds, wastewater, and drinking water. The methods for determination of nitrate can be divided into three main categories: direct, through a reduced nitrogen species, and indirect.1 Table 1 summarizes a variety of techniques which fall into these categories. Nitrate is a widespread contaminant of groundwater and surface waters worldwide2,3 and is a potential human health threat, especially to infants, causing the condition known as methemoglobinemia, also called “blue baby syndrome”. Chronic consumption of high levels of nitrate may also cause other health problems, including some cancers and teratogenic effects. The data are inconclusive but are cause for concern.4,5 Nitrate concentration is monitored in municipal water supplies worldwide and in foodstuffs to prevent exposure of populations to harmful or toxic levels. The EPA maximum contaminant level is 10 ppm nitratenitrogen (7 × 10-4 mol/L nitrate) for potable water. We are interested in the development of an amperometric biosensor for nitrate determination via the selective reduction of nitrate to nitrite by nitrate reductase enzyme. (1) Sah, R. N. Commun. Soil Sci. Plant Anal. 1994, 25, 2841. (2) Hallberg, G. R. In Nitrogen Management and Groundwater Protection; Follet, R. F., Ed.; Elsevier: Amsterdam, 1993; pp 35-74. (3) Puckett, L. J. Environ. Sci. Technol. 1995, 29, 408A. (4) Kross, B. C.; Hallberg, G. R.; Bruner, D. R.; Cherryholmes, K.; Johnson, J. K. Am. J. Public Health 1993, 83, 270. (5) Bruning-Fann, C. S.; Kaneene, J. B. Vet. Human Toxicol. 1993, 35, 521. S0003-2700(97)01146-3 CCC: $15.00 Published on Web 03/15/1998
© 1998 American Chemical Society
Table 1. Methods for Nitrate Determination direct nitration of phenols and colorimetry oxidation of organics and colorimetry ion-selective electrode detection direct UV-absorbance spectrophotometry gas chromatography after derivatization electrophoretic nitration of salicylic acid ion chromatography reduced nitrogen species reduction to nitrite and spectrophotometry electrochemistry reduction to ammonia and colorimetry potentiometry conductimetry reduction to nitric oxide and chemiluminescence indirect atomic absorption spectrophotomety polarography
Transduction of this reaction involves the monitoring of reduction current of an electron mediator, methyl viologen, which shuttles electrons between the enzyme and an electrode. There have been a few reports of nitrate biosensors in the literature. A patent on enzyme electrodes demonstrated the response of a electrode coated with a viologen-containing polymer to nitrate in the presence of nitrate reductase free in solution.6 Nitrate reduction was achieved with a nitrate reductase-based electrode which utilized a polythiophene bipyridinium film as the electron mediator. The enzymatic reaction could be observed in this system only through prolonged electrolysis, which resulted in an accumulation of nitrite in solution. Catalytic reduction currents could not be distinguished from background, presumably due to low enzymatic activity at the electrode surface.7 A similar study by the same group reported the ability of a bipyridinium thiol bound to a gold electrode to act as an electron mediator for (6) Gregg, B. A.; Heller, A. Patent WO 92/12254, 1992. (7) Willner, I.; Katz, E.; Lapidot, N.; Bauerle, P. Biochem. Bioenerget. 1992, 29, 29.
Analytical Chemistry, Vol. 70, No. 8, April 15, 1998 1511
nitrate reductase.8 A large number of dyes and other redox species were examined for their ability to act as mediators for nitrate reductase in amperometric electrodes from two different bacterial sources.9 An excellent paper described an amperometric nitrate electrode in which nitrate reductase from Escherichia coli was immobilized onto the electrode during electropolymerization of a bipyridinium pyrrole monomer.10 The electrode exhibited good response times to changes in nitrate concentration and provided low detection limits. Through the use of a novel lightactivated quinone mediator, the reduction of nitrate by nitrate reductase could be switched on and off at a thiol-covered gold electrode.11 Finally, an optical biosensor for nitrate was constructed using sol-gel-immobilized nitrate reductase.12 The sensor responded to nitrate in the micromole-per-liter range. Unfortunately, no assays of nitrate in samples were conducted with any of these biosensors. We present here a first-generation amperometric electrode for nitrate based on a soluble nitrate reductase from corn seedlings. In addition to characterizing the electrode, it was used to determine nitrate in fertilizer and drinking water samples. To our knowledge, this is the first demonstration of a nitrate reductasebased electrode being used in the assay of nitrate in real matrixes. EXPERIMENTAL SECTION Note: Reference to commercial supplies and apparatus is for completeness and does not constitute endorsement by the NIST. Chemicals and Apparatus. Nafion solution (mass fraction of 5%), sodium nitrate, and sodium nitrite were purchased from Aldrich Chemical Co. (Milwaukee, WI). 3-(N-Morphlino)propanesulfonic acid (MOPS), sodium chlorate and methyl viologen came from Sigma Chemical Co. (St. Louis, MO). Caution: Methyl viologen is toxic, and direct contact should be avoided. The use of appropriate protective equipment is strongly advised. All other chemicals were reagent grade. Deionized water was used to prepare all solutions. NADH:nitrate reductase (EC 1.6.6.1; 10 units/mL) was purified from corn seedlings using monoclonal antibody-based immunoaffinity chromatography, as described previously.13 The fertilizers were water soluble and obtained locally. An Orion nitrate ion-selective electrode (Orion, Beverly, MA) was used to assay fertilizers. A CV-1B potentiostat (Bioanalytical Systems, West Lafayette, IN) performed all other electrochemical measurements. Nitrate Reductase-Based Electrodes. The enzyme electrodes were constructed from glassy carbon electrodes (3-mm diameter, Bioanalytical Systems). The electrodes were first polished with 1-µm diamond polish on nylon mesh (Bioanalytical Systems), followed by 15-min sonications in water and then ethanol. Next, they were coated with 1.0 × 10-2 mL of Nafion (5 g/L in ethanol) and allowed to dry. Finally, 1.0 × 10-2 mL of nitrate reductase (1 × 10-1 units) was secured behind the dialysis (8) Katz, E.; Itzhak, N.; Willner, I. J. Electroanal. Chem. 1992, 336, 357. (9) Strehlitz, B.; Grundig, B.; Vorlop, K. D.; Bartholmes, P.; Kotte, H.; Stottmeister, U. Fresenius J. Anal. Chem. 1994, 349, 676. (10) Cosnier, S.; Innocent, C.; Jouanneau, Y. Anal. Chem. 1994, 66, 3198. (11) Doron, A.; Portnoy, M.; Lion-Dagan, M.; Katz, E.; Willner, I. J. Am. Chem. Soc. 1996, 118, 8937. (12) Aylott, J. W.; Richardson, D. J.; Russell, D. A. Analyst 1997, 122, 77. (13) Hyde, G. E.; Wilberding, J. A.; Meyer, A. L.; Campbell, E. R.; Campbell, W. R. Plant Mol. Biol. 1989, 13, 233.
1512 Analytical Chemistry, Vol. 70, No. 8, April 15, 1998
membrane (Spectra/Por 2; 12 000-14 000 MWCO; Spectrum, Houston, TX). Measurement Conditions with Nitrate Reductase-Based Electrodes. All electrode work was done in a background buffer composed of 5 × 10-2 mol/L MOPS, 1 × 10-3 mol/L EDTA, 5 × 10-1 mol/L potassium chloride, pH 7.4. Samples and standards were contained in septum-stoppered bottles and delivered to the electrochemical cell in gastight syringes. The electrochemical cell was composed of the enzyme electrode, a platinum wire auxiliary electrode, and a Ag/AgCl reference electrode (Bioanalytical Systems). Solutions in the electrochemical cell were thoroughly purged with nitrogen before a working potential was applied and stirred by nitrogen bubbling during all measurements. Measurement of Responses to Nitrate and Interferences. Before these electrodes could be used, they were always allowed to soak in 1 × 10-3 mol/L methyl viologen (in working buffer) for at least 5 min. This allowed the cation-exchange sites in the Nafion to load with viologen and act as mediator reservoirs. Then, the electrodes were placed into 10-mL volumes of buffer in the electrochemical cell. After a working potential of -0.8 V vs Ag/ AgCl was applied, the background current decayed, and ion standards were added. The current changes from background were noted. The electrodes were allowed to soak in the viologen solution before each subsequent calibration. The electrode, whose response to nitrate was monitored during a 7-day period, was stored in working buffer at 4 °C when not in use. Standard Addition Assay of Fertilizer Samples by Nitrate Reductase-Based Electrodes. Stock solutions of the fertilizer samples were prepared in buffer. The electrode was set up in the electrochemical cell in the same manner as was mentioned for nitrate calibration. The background current was allowed to decay, and then a small volume of fertilizer sample was added, followed by additions of nitrate standard. The samples were effectively diluted by a factor of 20 000 during the analyses. The total changes in nitrate concentration in the cell were intentionally kept small, so that the electrode response was linear. During standard additions, the total cell volume changed less than 0.03%. Once the slope of electrode response had been calculated between the sample and standard additions, it was a simple matter to extrapolate back to background to obtain the nitrate concentration in the samples. Then, a mass fraction of nitrate-nitrogen was calculated for the original granular sample, as is customary for fertilizers. Analyte Addition Assay of Drinking Water Samples by Nitrate Reductase-Based Electrodes. In these assays, the enzyme electrode was set up for nitrate calibration as before. Buffered drinking water samples were prepared using the same salt composition as the working buffer. Then, 1.42 × 10-4 mol/L spikes of nitrate (in the form of sodium nitrate) were added to the buffered samples to serve as internal standards. Several additions of nitrate standard followed by an addition of buffered drinking water were made to the cell. The drinking water samples were effectively diluted by a factor of 111 during the analyses. Again, total changes in nitrate concentration in the cell were intentionally kept small, so that the electrode response was linear. From two of the nitrate additions, the slope of the electrode response was calculated, and the nitrate concentration in the sample was obtained by extrapolation.
Colorimetric Assay of Nitrate in Fertilizer and Drinking Water Samples. This enzyme-based assay is a variation of EPA standard method 40 CFR 141 for determination of nitrate in drinking water and wastewater.14 The fertilizer samples were assayed by construction of a calibration curve of absorbance versus nitrate concentration. For the drinking water samples, a standard addition version of this assay was used. All the drinking water samples were buffered with the MOPS buffer. Here, the samples were spiked with several levels of nitrate and reacted as before to give diazo product. With these data and those of the blank (no sample, buffer only), the concentrations of nitrate in the samples could be obtained by extrapolation. Assay of Fertilizer Samples by Nitrate Ion-Selective Electrode. The electrode was calibrated as per the manufacturer’s instructions. Nitrate concentrations in the samples were determined by comparison with a calibration curve, and, from these, the weight percentage of nitrate-nitrogen was calculated. RESULTS AND DISCUSSION Mediator. In a recent report,15 several electron mediators were tested for their ability to shuttle electrons from an electrode to the corn seedling nitrate reductase and support its activity. Among those tested were azure A, safranin T, neutral red, bromophenol blue, Cibacron blue, and methyl viologen. Through preliminary experiments, it was determined that methyl viologen was an excellent choice for the nitrate reductase-based electrode and -0.8 V vs Ag/AgCl a reasonable working potential. The electrostatic attraction between the 2+ charge on methyl viologen (1+ when reduced) and the ion-exchanger (Nafion), combined with hydrophobic interactions between the viologen and ion-exchanger carbon skeleton, ensure that methyl viologen will be immobilized at some percentage of saturation in the presence of inorganic cations. This method of using Nafion as a mediator reservoir has also served well for glucose16 and urate17 enzyme electrodes. Calibration and Lifetime of the Nitrate Reductase-Based Electrode. Figure 1 shows the response of this electrode to nitrate on the first, second, and seventh days of electrode life. When the working potential was applied, a steady baseline current was reached within 3-4 min. The response time to nitrate additions was less than 1 min. The upper limit of detection of nitrate was approximately 1.2 × 10-2 mol/L, while the lower limit, as determined during the course of drinking water assays to be discussed later, was consistently less than 3.0 × 10-6 mol/L (n ) 8; S/N ) 5.1). The linear range of electrode response, as observed in the drinking water assays, extended from 3.0 to greater than 17.9 µmol/L of nitrate. A recent review1 on nitrate determination listed limits of nitrate detection for a large number of methods; the range was 0.003-13 µmol/L. Hence, the present system compares well with others. Note in Figure 1 that the magnitude of current response decreased as the electrode aged. As with all enzyme biosensors, (14) Campbell, E. R.; Corrigan, J. S.; Campbell, W. H. In Symposium on Field Analytical Methods for Hazardous Wastes and Toxic Chemicals; Air Waste Management Association Meeting, Las Vegas, NV, January 29-31, 1997; AWMA: (15) Mellor, R. B.; Ronnenberg, J.; Campbell, W. H.; Diekmann, S. Nature 1992, 355, 717 (16) Brown, R. S.; Luong, J. H. T. Anal. Chim. Acta 1995, 310, 419. (17) Jin, L.; Ye, J.; Tong, W.; Fang, Y. Mikrochim. Acta 1993, 112, 71.
Figure 1. Calibration of glassy carbon electrode: (9) day 1 (n ) 4), (0) day 2 (n ) 4), and (O) day 7 (n ) 2). Mean ( SE. Working electrode potential, -0.8 V vs Ag/AgCl.
Figure 2. Responses of glassy carbon electrode normalized to the response at the highest nitrate concentration. Mean of all normalized calibrations from day 1, day 2, and day 7 (n ) 10 total). Mean ( SE.
decay of enzyme activity with time is likely the cause. This effect was even seen among different calibration curves obtained on the same day. Experience with the enzyme preparation indicates that total inactivation occurs in 8 h at 1 unit/mL of enzyme free in solution at room temperature. At 10 units/mL (the nominal concentration present under the dialysis membrane of the electrode), inactivation is complete within 2 days at room temperature. The width of the error bars is an indication that the magnitudes of response are not that constant through the course of a day. The reason for larger error bars on the second day, as opposed to the first, is not known at this time. However, if the calibration curves are normalized to the highest current, it is evident that the electrode response maintains the same functional form during its lifetime. Figure 2 illustrates this point clearly. Here, all of the calibrations shown in Figure 1 were normalized and then averaged. The uncertainty bars here are quite small in width. This effect of response “drift” during a given day can likely be minimized by redesign of the electrode, with emphasis placed on enhanced enzyme stability, i.e., through covalent immobilization. Yet, the electrode can be used to perform practical nitrate assays by employing techniques which compensate for drifts in response from one sample to the next. This point Analytical Chemistry, Vol. 70, No. 8, April 15, 1998
1513
Table 2. Comparison of Assays for Nitrate in Fertilizers as Percent Nitrate-Nitrogena electrodes sample
colorimetric
enzyme
ion-selective
A B
6.7 ( 0.3 (n ) 3)
6.5 ( 1.3 (n ) 8) 15.7 ( 0.8 (n ) 4)
6.0 ( 0.2 (n ) 3) 17.3 ( 1.0 (n ) 4)
a
Mean ( SE.
Table 3. Comparison of Determinations of Concentration of Nitrate in Drinking Watera sample
Figure 3. Responses of newly made glassy carbon electrode to high concentrations of (9) chlorate and (O) nitrite.
A B a
colorimetric (mol/L) 10-4
(n ) 3) (1.1 ( 0.1) × (3.1 ( 0.3) × 10-4 (n ) 3)
enzyme electrode (mol/L) (1.0 ( 0.1) × 10-4 (n ) 3) (3.3 ( 0.4) × 10-4 (n ) 4)
Mean ( SE.
will be illustrated later by the standard addition analysis of fertilizers and the analyte addition analysis of drinking water. Interferences. At a working potential of -0.8 V vs Ag/AgCl, the reduction of oxygen at the electrode surface would cause large background currents. The magnitude of this interference was effectively decreased by purging samples and standard thoroughly with nitrogen in septum-stopped vials and transferring them to the electrochemical cell via gastight syringes. In addition, the buffer in the cell was thoroughly purged with nitrogen before potential was applied and stirred by nitrogen bubbling during the measurements. Trace metals ions would have also caused an interference. Metals ions are notorious for their ability to denature enzymes and were controlled by the addition of EDTA. Chelation also prevented the metal ions from being concentrated on the cationexchange sites of Nafion and successively reduced due to the negative working potential. In addition to oxygen and metal ions, interferences from other solution species are possible. The responses of a newly made electrode to chlorate and nitrite are shown in Figure 3. The nitrite response is much smaller than that to nitrate. Hence, the enzyme preparation is apparently low in nitrite reductase activity, which is quite often associated with nitrate reductase in living systems. The electrode response to chlorate was expected on the basis of previous knowledge of the enzyme’s selectivity.18 The halide analogues of nitrate, chlorate, bromate, and iodate are all substrates for nitrate reductase. Higher oxidation state analogues, like perchlorate, are not reduced. The fact that chlorate is a substrate could be significant in the drinking water analysis since it could be present as a result of chlorination. In terms of inhibitors, chloride can cause some inhibition of nitrate reductase. However, the electrode had a lower detection limit for nitrate, of less than 3.0 × 10-6 mol/L. Hence, the electrode responded to nitrate when present at less than 6 × 10-6 times the chloride concentration. Similarly, in a report on an optical approach to nitrate biosensing using a similar nitrate reductase, no interference was found by phosphate, carbonate, hydrogen carbonate, or chloride at concentrations up to 0.5 mol/ L.12
CONCLUSIONS A nitrate reductase-based electrode has been demonstrated which employs nitrate reductase from corn seedlings and a Nafion reservoir to store the electron mediator, methyl viologen. The electrode exhibited a low detection limit for nitrate and rapid response to changing nitrate concentration. Most importantly, it demonstrated the ability to assay real-world samples. The electrode does not require expensive/complex associated instrumentation, unlike techniques such as ion chromatography, polarography, fluorometry, direct spectrophotometry, or chemilu-
(18) Solomonson, J. P.; Vennesland, B. Plant Physiol. 1972, 50, 421.
(19) The Fisher Catalog; Fisher Scientific Co.: Pittsburgh, PA, 1994; p 1359.
1514 Analytical Chemistry, Vol. 70, No. 8, April 15, 1998
In contrast, the commercial nitrate ion-selective electrode used here has a selectivity pattern based on the well-known Hofmeister series. Hence, potential interferences for the ion-selective electrode, depending on their relative concentrations to nitrate, could include perchlorate, iodide, chlorate, cyanide, bromide, nitrite, hydrosulfide, carbonate, chloride, bicarbonate, mono-, di-, and tribasic phosphates, acetate, fluoride, and sulfate.19 Interference by organic anions (e.g., benzoate) are likely as well. Fertilizer Analysis. Table 2 outlines the results of the measurement of nitrate in fertilizer samples. The nitrate reductase-based electrode gave a mean percentage of nitrate-nitrogen which was between that found by the colorimetric and ion-selective electrode techniques for sample A. However, the error of the mean was larger than those for the other two techniques. Refinement of the nitrate reductase-based electrode and the measurement method will likely increase the precision of the results. For sample B, the enzyme electrode gave somewhat lower assay than the ion-selective electrode but had better precision. The values given by both methods were consistent when the confidence intervals were taken into account. Drinking Water. Because of the importance of nitrate monitoring in groundwater and drinking water, a major goal of this work was to demonstrate that our first-generation electrode was capable of measuring nitrate in such samples. Table 3 illustrates the results of colorimetric and electrode assays for nitrate in two drinking water samples. Given the precision of both types of assays, the nitrate values are quite comparable.
minescence, and could be adapted to field analysis. In addition, it has the inherent ability to operate in turbid samples, unlike colorimetric methods. One significant improvement in electrode design could be the inclusion of components to reduce the magnitude of oxygen interference without the need for sample purging.
ACKNOWLEDGMENT This work was done under NIST CRADA CN-1105. Received for review October 17, 1997. Accepted February 15, 1998. AC971146S
Analytical Chemistry, Vol. 70, No. 8, April 15, 1998
1515
