Anal. Chem. 1996, 68, 2946-2950
Linamarin Sensors: Interference-Based Sensing of Linamarin Using Linamarase and Peroxidase Tetsu Tatsuma, Koichiro Tani, and Noboru Oyama*
Department of Applied Chemistry, Faculty of Technology, Tokyo University of Agriculture and Technology, Naka-cho, Koganei, Tokyo 184, Japan Hock-Hin Yeoh
Department of Botany, National University of Singapore, Kent Ridge, Singapore 0511, Republic of Singapore
An interference-based linamarin sensor is developed. Horseradish peroxidase (HRP) is adsorbed on a pyrolytic graphite (PG) electrode, and then linamarase from cassava is cross-linked with glutaraldehyde on the electrode surface. The prepared bienzyme electrode is poised at -300 mV vs Ag/AgCl for 40 s to reduce dissolved O2 to H2O2 at the PG surface. The potential is then stepped to 0 mV, at which point the accumulated H2O2 is reduced, though the O2 reduction does not proceed. Since the H2O2 reduction is catalyzed by HRP, the transient cathodic current is inhibited by cyanide, which is liberated from linamarin by linamarase. Therefore, the transient current is a function of the linamarin concentration. This sensor responds to 1 × 10-5-5 × 10-3 M linamarin and can estimate a linamarin concentration of a cassava extract. Cassava (Manihot esculenta Crantz) is widely cultivated in the tropics for its storage roots, which are the staple food of about 300 million people in the world. Linamarin (phaseolunatin, Figure 1), a cyanogenic glycoside, is found in both the leaves and roots of these plants.1,2 Although linamarin is not toxic in itself, cyanide liberated from linamarin as a result of hydrolysis is highly toxic. The hydrolysis is catalyzed by a Brønsted acid or an enzyme, linamarase, a β-glucosidase (EC 3.2.1.21):
linamarin + H2O f glucose + acetone + HCN Cyanide is detoxicated by rhodanese (EC 2.8.1.1) in the human body as follows:
CN- + S2O32- f SCN- + SO32Although thiocyanate is much less toxic than cyanide, with chronic toxicity it may result in ataxic neuropathy. The ability to determine linamarin content in cassava or cassava-based products is therefore important. Thus, various methods for linamarin determinations have been developed.3-8 (1) Wood, T. J. Sci. Food. Agric. 1966, 17, 85-90. (2) Nartey, F. Phytochemistry 1968, 7, 1307-1312. (3) Cooke, R. D. J. Sci. Food. Agric. 1978, 29, 345-352. (4) Bradbury, J. H.; Egan, S. V. Phytochem. Anal. 1992, 3, 91-94. (5) Yeoh, H. H.; Tan, C. K. C. Biotechnol. Tech. 1994, 8, 337-338. (6) Yeoh, H. H.; Truong, V. D. Food Chem. 1993, 47, 295-298.
2946 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
Figure 1. Molecular structure of linamarin.
Yeoh6,7 has developed potentiometric sensors for linamarin by coupling linamarase from cassava leaf with a cyanide ion-selective electrode. The cyanide generated as a result of the reaction catalyzed by linamarase is detected by the cyanide ion-selective electrode. The sensor can determine 1 × 10-4-2 × 10-2 M linamarin rapidly and reliably, though it is not compact. Since a more sensitive device is needed, Tatsuma et al.8 have devised an amperometric linamarin sensors using linamarase and glucose oxidase (EC 1.1.3.4). Among the amperometric sensors, the linamarase-glucose oxidase-peroxidase (EC 1.11.1.7) trienzyme electrode is so sensitive that it determines >5 × 10-6 M linamarin. In this sensor, linamarin is hydrolyzed by linamarase to liberate glucose, and electrons as well as protons are transferred from glucose to oxygen by glucose oxidase to yield hydrogen peroxide. Thus, generated hydrogen peroxide is reduced by peroxidase with electrons transferred from SnO2 electrode via polypyrrole.9 However, those amperometric sensors based on glucose oxidase activity may suffer from interference from glucose. Therefore, glucose must have been eliminated by something like a precolumn. In view of the interference, cyanide detection-based sensors are better than glucose detection-based sensors. Thus, in the present work, a sensitive and compact linamarin sensor based on the cyanide detection is developed. To date, some interference-based10 cyanide biosensors have been reported. Cyanide is detected on the basis of enzymatic activity, which is reversibly inhibited by cyanide. Albery et al.11 have devised a cyanide sensor utilizing an expensive enzyme, cytochrome oxidase (EC 1.9.3.1), as a catalyst for oxygen reduc(7) Yeoh, H. H. Biotechnol. Tech. 1993, 7, 761-764. (8) Tatsuma, T.; Tani, K.; Oyama, N.; Yeoh, H. H. J. Electroanal. Chem., in press. (9) Tatsuma, T.; Gondaira, M.; Watanabe, T. Anal. Chem. 1992, 64, 11831187. (10) Scheller, F.; Schubert, F. Biosensors; Elsevier: Amsterdam, 1992; pp 260264. (11) Albery, W. J.; Cass, A. E. G.; Mangold, B. P.; Shu, Z. X. Biosens. Bioelectron. 1990, 5, 397-413. S0003-2700(96)00132-1 CCC: $12.00
© 1996 American Chemical Society
Figure 2. Principle of the interference-based linamarin sensing. Reaction 1 proceeds at -300 mV vs Ag/AgCl but not at 0 mV. Although reactions 2-6 occur at both 0 and -300 mV, catalytic current for the H2O2 reduction is much smaller than for the O2 reduction current. Fe, to which imidazole coordinates, stands for the active site of peroxidase.
tion. Smit and Cass12 have developed an alternative cyanide sensor using an inexpensive enzyme, horseradish peroxidase (HRP), as a catalyst for hydrogen peroxide reduction. Smit and Rechnitz13 have used tyrosinase (EC 1.14.8.1) as a catalyst for oxygen reduction. However, these cyanide sensors need the presence of a dissolved mediator for the enzymes. Although Tatsuma and Watanabe14,15 have devised a mediatorless sensor for heme ligands utilizing a heme peptide, their system requires the addition of hydrogen peroxide. Recently, Amine et al.16 have reported a reagentless cyanide sensor that employs cytochrome oxidase and cytochrome c. Tatsuma and Oyama17 have developed inexpensive and reagentless cyanide biosensors using a pyrolytic graphite (PG) electrode, on which horseradish peroxidase (HRP) is adsorbed. The electrode is first poised at -300 mV vs Ag/AgCl to reduce dissolved O2 to H2O2 at the PG surface. The generated H2O2 accumulates in the diffusion layer. The potential is then stepped to 0 mV, at which point the accumulated H2O2 is reduced, though the O2 reduction does not proceed. Thus, the transient H2O2 reduction current is observed at 0 mV. Since the H2O2 reduction is catalyzed by HRP, the activity of which can be inhibited by cyanide,18 the transient cathodic current is suppressed in the presence of cyanide. Therefore, the transient current is a function of the cyanide concentration. The HRP/PG electrode can determine 10-5-10-3 M cyanide. (12) Smit, M. H.; Cass, A. E. G. Anal. Chem. 1990, 62, 2429-2436. (13) Smit, M. H.; Rechnitz, G. A. Anal. Chem. 1993, 65, 380-385. (14) Tatsuma, T.; Watanabe, T. Anal. Chem. 1991, 63, 1580-1585. (15) Tatsuma, T.; Watanabe, T. Anal. Chem. 1992, 64, 143-147. (16) Amine, A.; Alafandy, M.; Kauffmann, J.-M.; Pekli, M. N. Anal. Chem. 1995, 67, 2822-2827. (17) Tatsuma, T.; Oyama, N. Anal. Chem. 1996, 68, 1612-1615. (18) Maehly, A. C. Methods Enzymol. 1955, 2, 801-813.
Here, the HRP/PG electrode is coated with a cross-linked film of linamarase, which catalyzes cyanide liberation from linamarin. The principle of the present sensor is illustrated in Figure 2. That is, H2O2 generated in reaction 1 is electrocatalytically reduced in reactions 2-4, but the reduction is inhibited by cyanide (reaction 6)18 generated in reaction 5. Thus, the reduction current must be a function of the linamarin concentration. EXPERIMENTAL SECTION Materials. Linamarase from cassava leaf was prepared as previously described.19 The stock solution was stored in a freezer. HRP (grade II, RZ number ≈ 3) was purchased from Boehringer. Linamarin and glutaraldehyde, (Sigma, St. Louis, MO) were used as obtained. Linamarin solutions were prepared just before experiments with a phosphate buffer (1/15 M, pH 7.4). Spectroscopic Measurements. Spectroscopic measurements were conducted using a UV/visible spectrophotometer (Ubest-55, Jasco). Preparation of Enzyme Electrodes. The surface of a PG disk electrode (0.2 cm2, Tokai Carbon) was polished with a fine sandpaper and sonicated in pure water. The PG electrode was immersed in a 1/15 M phosphate buffer (pH 7.4) containing HRP (1 mg/mL) for 10 min, followed by thorough rinsing with water, to obtain a HRP/PG electrode. Linamarase cross-linked film was prepared in a manner similar to that described elsewhere.8,20 A 5% glutaraldehyde aqueous solution was mixed with the same volume of linamarase stock solution, which is a sodium citrate buffer (0.05 M, pH 6), containing linamarase (0.22 nkat/µL; 1 nkat of linamarase consumes 1 nmol of linamarin per second at 37 °C). (19) Yeoh, H. H. Phytochemistry 1989, 28, 721-724. (20) Tatsuma, T.; Okawa, Y.; Watanabe, T. Anal. Chem. 1989, 61, 2352-2355.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2947
A 10-µL (unless otherwise noted) aliquot of the mixed soluiton was cast on the HRP/PG electrode surface and left for about 4 h for cross-linking reaction at room temperature. The prepared linamarase-HRP bienzyme electrode was thoroughly rinsed with water. Electrochemical Measurements. Electrochemical measurements were performed at room temperature with a potentiostat LC-4C (BAS, West Lafayette, IN), an Ag/AgCl reference electrode, and a platinum wire as a counter electrode. The electrolyte solution was a 1/15 M phosphate buffer (pH 7.4, 3 mL). Potential-step measurements were performed as follows: First, the linamarase-HRP bienzyme electrode was poised at -300 mV for 40 s (unless otherwise noted). The electrode potential was then stepped again to 0 mV and kept for 240-320 s. Potential was stepped to -300 mV. These potential steps between -300 and 0 mV were repeated until the current behavior became reproducible. After the reproducibility was obtained, a linamarin solution was added to the electrolyte solution and stirred for about 5 s, 30-40 s before the potential step from 0 to -300 mV. A transient response was monitored upon the next potential step from -300 to 0 mV. Preparation of a Cassava Root Extract. Cassava roots (20 g) were homogenized in 0.1 M phosphoric acid solution (100 mL) in a homogenizer. The homogenate was then centrifuged and ultrafiltered (Sartorius Minisart, pore size 0.2 µM). The pH value was then adjusted to 6 with 2 M NaOH and stored in a freezer. The pH value was adjusted to 7.4 with 10 M aqueous NaOH just before use. RESULTS AND DISCUSSION Cyanide Liberation and Complex Formation. Complex formation between HRP and cyanide liberated from linamarin was monitored in the presence of linamarase by using a UV/visible spectrophotometer. The solution was 0.1 M citrate buffer (pH 6), containing 0.017 nkat/µL of linamarase, 3.7 µM HRP, and linamarin. The reference sample was the buffer containing the same amount of linamarase. Figure 3 shows the steady-state spectra obtained at 0, 2, 4, 8, and 16 µM linamarin. The arrows indicate changes with increasing linamarin concentration. The spectrum in the absence of linamarin is typical for HRP without strong ligands. The Soret band at 403 nm and the visible bands at around 500 and 640 nm are characteristic of the free HRP.18 On the other hand, the spectrum in the presence of 16 µM linamarin, which is indicated by the Soret band at 421 nm and the visible band at around 540 nm, is typical of cyanide-coordinated HRP.18 The spectrum of HRP did not change with the addition of linamarin (in a similar concentration range) in the absence of linamarase. Thus, it was verified thant cyanide is liberated from linamarin in the presence of linamarase and that HRP forms a complex with the liberated cyanide. The dissociation constant was calculated to be about 3 µM, on the assumption that the cyanide concentration equals the linamarin concentration. This value is in agreement with the reported value (4 µM) for the HRP-cyanide complex at pH 4-5.18 Therefore, almost all of the linamarin molecules have been decomposed by linamarase. Transient Currents for Hydrogen Peroxide Reduction. As mentioned in the Experimental Section, the potential applied to the linamarase-HRP bienzyme electrode (0.2 cm2) was stepped between -300 and 0 mV vs Ag/AgCl repeatedly. At -300 mV vs Ag/AgCl, a cathodic current of around 1 µA flowed; dissolved O2 was reduced to H2O2, and H2O2 accumulated in the diffusion layer. 2948 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
Figure 3. Steady-state spectra of 3.7 µM HRP in 0.1 M citrate buffer (pH 6), containing 0.017 nkat/µL of linamarase and 0, 2, 4, 8, and 16 µM linamarin.
This current is smaller than that at a PG electrode on which HRP alone is adsorbed (HRP monoenzyme electrode without linamarase film; ∼1 µA/0.05 cm2).17 This can be explained in terms of the O2 diffusion retarded by the linamarase cross-linked film. In this step, a small amount of peroxidase compound III can form, resulting in deterioration of the transient cathodic current in the following step at 0 mV. However, this can be negligible because significant deterioration in the transient current was not observed. Upon the potential step from -300 to 0 mV, the cathodic current decreased, and then the current increased and finally decreased again gradually (Figure 4C,b). The valley (Figure 4C, a) was obtained probably due to the nonfaradaic, charging current. The cathodic peak (Figure 4C) current is ascribed to the reduction of H2O2 which had been accumulated in the diffusion layer.17 Therefore, the cathodic peak current is dealt as a response. The response (around 10 nA, >100 s from the potential step from -300 to 0 mV) was smaller and slower than that of the HRP monoenzyme electrode (∼65 nA/0.05 cm2, 2 µA, respectively. However, response currents upon the potential step to 0 mV were almost independent of the potential at which H2O2 is generated. If the potential for the H2O2 generation is more negative, anodic charging current upon the potential step to 0 mV may be larger. Therefore, the higher H2O2 reduction current (at 0 mV) may be canceled by the higher anodic charging current so that the response current was independent of the potential for the H2O2 generation. Otherwise, reduction of O2 and/or H2O2 to water, which retards H2O2 accumulation, may be faster at a more negative potential. Time for the Hydrogen Peroxide Accumulation. In the above experiments, the time for the H2O2 accumulation was 40 s. We examined longer accumulation times (80 and 120 s). However, response currents upon the potential step from -300 to 0 mV were almost independent of the H2O2 accumulation time. This reflects the fact that the steady state was reached in 40 s for the mass transfer processes, which includes the supply of O2 from the bulk, the reduction of O2 to H2O2, the dissipation of H2O2 to the bulk, and the reduction of H2O2 to water. Therefore, 40 s is sufficient for the H2O2 accumulation in the present system. Responses to Linamarin. In the presence of linamarin (>10-5 M), the transient response is different from that in its absence. The valley was deeper (i.e., the anodic peak was higher) and the cathodic peak lower for the higher linamarin concentration. Finally, the cathodic current vanished at a linamarin concentration above 5 mM. Since the HRP monoenzyme electrode or linamarin monoenzyme electrode exhibited no significant response (anodic or cathodic) to linamarin up to 5 mM, the cathodic current inhibition must be caused by cyanide liberated from linamarin. The lower sensing limit of 10-5 M is an order of magnitude better than that of the previously reported linamarin
sensor which is based on a cyanide ion-selective electrode.6,7 Additionally, the present sensor is much more compact than the previous ion-selective electrode-based sensor. The addition of acetone (e10 mM) or glucose (e10 mM) to the buffer did not change the transient current of the bienzyme electrode. A linamarin sensor with linamarase and glucose oxidase, which we have developed previously,8 should be sensitive not only to linamarin but also to glucose because it responds to linamarin on the basis of oxidation of glucose liberated from linamarin. Therefore, it cannot be used for measurements of a solution containing both linamarin and glucose. On the contrary, the present sensor is based on the response to cyanide liberated from linamarin, so interference from glucose was neglible. The typical responses of the sensor were plotted as a function of the linamarin concentration in Figure 5 (b). The normalized response R was defined as follows: The time (from the potential step to 0 mV) giving the peak current (ip) in the absence of linamarin was defined as T (i.e., ip ) i(T), see Figure 4). The lowest i(T) (obtained at 5-10 mM linamarin) was taken as the background (ib), and all the responses were normalized at the current in the absence of linamarin (ip). That is R ) (i(T)-ib)/ (ip-ib). As can be seen, a sigmoidal curve was obtained, as expected. We have proved that normalized responses to cyanide (R) of the HRP monoenzyme electrode fit to the following equation:17
R)
1 CL/Kapp + 1
(1)
where Kapp is the apparent dissociation constant of the HRPcyanide complex (Kapp g intrinsic dissociation constant) and CL is the cyanide concentration. As for the present system, the responses to linamarin will be dictated by eq 1 (where Kapp is a constant and CL is the linamarin concentration) if the cyanide concentration at the PG electrode surface, on which HRP is immobilized, is proportional to the bulk concentration of linamarin. The dashed curve in Figure 5 is calculated from eq 1 for Kapp ) 2.1 × 10-4 M. Since the curve roughly fits the experimentally obtained responses, we conclude that the response can be analyzed on the basis of eq 1. The mean value of Kapp obtained by the curve fitting for four independently prepared electrodes Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2949
was 2.5 × 10-4 M (standard deviation was 0.6 × 10-4 M). This value is larger than the Kapp value of the HRP monoenzyme electrode for cyanide (6 × 10-5 M).17 The lower cyanide concentration at the PG electrode surface than the bulk concentration of linamarin must be responsible for the higher Kapp values of the present sensor. The partition coefficient of cyanide may be smaller than that of linamarin, and/or the diffusion coefficient of cyanide may be larger than that of linamarin. Linamarase Film Thickness. The above-mentioned film was prepared by casting a 10-µL aliquot of the linamarase-glutaraldehyde mixed solution. Linamarase-HRP bienzyme electrodes with thinner linamarase film (2.5 and 5 µL cast) were also prepared. The thinner linamarase film-coated electrode tended to exhibit larger and faster transient responses. This result is in agreement with the above-mentioned result that the transient responses of the linamarase film-coated electrode were smaller and slower than those of a HRP monoenzyme electrode without a linamarase film. The Kapp value for the electrode with a 5-µL cast film (2 × 10-43 × 10-4 M) was close to that for the electrode with a 10-µL cast film. However, the electrode with a 2.5-µL cast film exhibited a higher Kapp value ∼2 × 10-3 M. This may be because the concentration of generated cyanide is lower in the thinner film. Thus, we conclude that the cast amount should be more than 5 µL for sensitive measurements. Measurement of a Cassava Root Extract. The linamarin concentration of a cassava root extract was estimated using the present linamarase-HRP electrode. The extract was prepared as described in the Experimental Section. Linamarin and glucose concentrations in the extract have been determined by spectrophotometric means using linamarase6 or glucose oxidase21 to be 1.20 × 10-3 and 3.25 × 10-3 M, respectively. After the reproducible peak currents (ip) were obtained, a 300-µL aliquot of the cassava extract solution was added to the buffer (2.7 mL), and a peak current (i(T)) was recorded. Linamarin was then added stepwise until the response was saturated (ib) (i.e., the peak current was completely inhibited). Linamarin concentration CL was calculated from the normalized response R (R ) (i(T) - ib)/ (ip - ib) on the basis of eq 2, which was derived from eq 1: (21) Raabo, E.; Terkildsen, T. C. Scand. J. Clin. Lab. Invest. 1960, 12, 402-407.
2950
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
CL ) Kapp(1/R - 1)
(2)
The linamarin concentrations obtained using two independently prepared electrodes were 0.8 × 10-4 and 1.0 × 10-4 M (2.5 × 10-4 M was used as the Kapp value). These values are roughly in agreement with the linamarin concentration in the buffer, 1.2 × 10-4 M (1.2 × 10-3 M solution was diluted 10-fold). The deviation from the real concentration may be due to the low reproducibility of the Kapp value. Therefore, the fabrication process of the bienzyme electrode should be more sophisticated for more rigorous measurements. Although a small amount of ascorbic acid contained in the extract is a potential interferent, this will lead to overestimation because it gives electrons to the electrode and/or peroxidase compounds I and II; this cannot be a reason for the underestimation. In any event, estimation of the linamarin concentration of a cassava extract is possible with the present sensor. Lifetime of the Sensor. The used sensor was stored in a phosphate buffer (1/15 M, pH 7.4) under refrigeration (below 10 °C) overnight, and then the measurement was conducted again. Cathodic peak currents observed in the absence of linamarin were about half the initial value; HRP was not completely inactivated. However, inhibition of the cathodic current by linamarin was not observed up to 5 mM. Thus, we conclude that linamarase was inactivated during the storage. This was also the case for the linamarase-glucose oxidase electrode that we reported previously.8 The present linamarase-HRP bienzyme electrode will serve as a disposable sensor. For this purpose, the present electrode will be miniaturized to a small sensor tip, though this is beyond the scope of the present work. ACKNOWLEDGMENT This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan (No. 07215216 for N.O. and No. 07750907 for T.T.). Received for review February 7, 1996. Accepted June 10, 1996.X AC9601324 X
Abstract published in Advance ACS Abstracts, July 15, 1996.
