Biosensors & Bioelectronics Vol. 12. No. 11, pp. 1089–1099, 1997 1997 Published by Elsevier Science Limited All rights reserved. Printed in Great Britain 0956–5663/97/$17.00 PII: S0956–5663(97)00073-0
The lipoxygenase sensor, a new approach in essential fatty acid determination in foods Michael Schoemaker,*† Rainer Feldbru¨gge,† Bernd Gru¨ndig† & Friedrich Spener*†‡ *Institut for Biochemie, Universita¨t Mu¨nster, Wilhelm-Klemm-Str. 2, D-48149 Mu¨nster, Germany †Institut fu¨r Chemo- und Biosensorik Mu¨nster, Mendelstr. 11, D-48149 Mu¨nster, Germany (Received 1 August 1996; revised version received 13 May 1997; accepted 16 June 1997)
Abstract: Both an enzyme electrode and enzyme column with immobilized lipoxygenase, respectively, were used for the determination of essential fatty acids. The former was applied in a batch system, the latter was part of a fully automated flow injection analysis (FIA)-system. The oxygen consumption due to the lipoxygenase catalysed oxygenation of essential fatty acids was monitored amperometrically. Both systems were compared with regard to linear ranges of the calibration plots, sensitivities, detection limits, apparent Michaelis–Menten constants and lifetimes. The enzyme electrode showed different sensitivities for linoleic and ␣-linolenic acids, the most common essential fatty acids. The reason for this was not a second oxygenation step by lipoxygenase in case of ␣-linolenic acid, but a different dialytic behaviour of the two substrates. Hence, only the FIA-system was used for the determination of these fatty acids in real matrices such as vegetable oils and margarines. In the presence of detergent the triglycerides of the hydrophobic food samples were converted into water soluble glycerol and free fatty acids by a 15 min incubation with a ready to use lipase/esterase-mix, thus avoiding the use of organic solvents for analysis. Results obtained by the enzymatic FIA-system were in excellent agreement with those obtained by standard gas chromatography. 1997 Published by Elsevier Science Limited Keywords: flow injection analysis (FIA)-system, amperometric enzyme sensor, oxygen electrode, lipoxygenase, essential fatty acids, lipolysis, vegetable oils, margarines
INTRODUCTION Whereas the idea that essential fatty acid (EFA) deficiency is involved in a wide range of clinical disorders emerged in the mid-1950s (Sinclair, 1956), it is only in the last 20 years that epidemi-
‡Author to whom correspondence should be addressed.
ologic studies, clinical investigations and animal experiments provided strong evidence that EFA deficiency in the Western diet is the basis of symptoms which might be called diseases of civilization. Among these symptoms are vascular diseases, diabetic complications, inflammatory disorders such as atopic eczema and arthritis, as well as cancer (Horrobin, 1993; Galli et al., 1993; Simopoulos, 1994). 1089
CAP
Michael Schoemaker et al.
The EFAs are a group of 12 compounds, but linoleic acid (cis, cis-⌬9, ⌬12-octadecadienoic acid) and ␣-linolenic acid (all-cis-⌬9, ⌬12, ⌬15octadecatrienoic acid) are the only EFAs which are found in the diet—particulary in vegetable oils—in some abundance. The other 10 EFAs are either metabolic products of linoleic and ␣linolenic acids or are found in the diet in considerably smaller amounts (Horrobin, 1993; Galli et al., 1993; Simopoulos, 1994). While an absolute EFA deficiency in the diet might be rare, a lowering of EFA concentrations in the body tissues is common because of the increasing levels of other non-EFA fats in the Western diet. These non-EFA fats include saturated fats as well as trans-unsaturated fatty acids which are formed in food processing. The steadily growing knowledge about the correlation between the fatty acid composition of the diet and clinical disorders leads to a growing demand for a rapid and easy to use analytical device for fatty acid determination in foods. However, the determination of water insoluble or poor water soluble analytes such as triglycerides or fatty acids is not only tedious but also time consuming, and practical experience is indispensable. The recent development of organic-phase enzyme electrodes (OPEE) is an attempt to overcome the problems of analysing poor water soluble analytes, even in complex food samples. Hydrophobic food products like butter, margarine, eggs and vegetable oils have been assayed for their cholesterol, phenol, peroxide and moisture content by exploiting the organic-phase activity of cholesterol oxidase, peroxidase and tyrosinase (Hall & Turner, 1991; Wang et al., 1992a; Valencia-Gonza´les & Dı´az-Garcia, 1994; Mannino et al., (1994a, b)). Despite the obvious advantages of enzyme electrodes operating in organic phases (Saini et al., 1991; Wang, 1993), the use of organic solvents comprises the risk of damaging health and environment. Furthermore, not every enzyme remains catalytically active in organic phases. As a matter of fact, up to now only a small number of enzymes was used as the biological component in OPEE, mostly peroxidase and tyrosinase (Hall et al., 1988; Schubert et al., 1991; Wang et al., (1992a, b); Wang & Lin, 1993; Wang & Reviejo, 1993; Wang et al., (1993a, b, c); Mannino et al., (1994a, b)). Up to date, the fatty acid composition of fats and oils is determined mainly by gas chromatography (GC). Although GC is well established in 1090
Biosensors & Bioelectronics
lipid analysis and offers high sensitivities, it is still time consuming and laborious. In this paper we describe the development of enzyme sensors for EFA determination using the isoform I lipoxygenase derived from soybeans. Lipoxygenases in general catalyse the oxygenation of polyunsaturated fatty acids containing a cis, cis-1,4-pentadiene system by molecular oxygen. For the lack of cis, cis-1,4-pentadiene system trans-fatty acids, mono-unsaturated and saturated fatty acids can not serve as lipoxygenase substrates, resulting in a high specificity for EFAs. The isoform I lipoxygenase (LOX) from soybeans differs from the other isoforms by its high pH-optimum (pH 9), high stability and its high turnover number for the non-esterified (free) essential fatty acids (Axelrod et al., 1981).
EXPERIMENTAL SECTION Apparatus Batch and flow amperometric experiments were performed with an electrochemical detector EP 30 (Biometra, Go¨ttingen, Germany). The Clarktype oxygen electrodes were polarized at a constant potential of − 0.6 V versus Ag/AgCl (0.1 M KCl) electrode. Modules of the FIA-system were the buffer reservoir (0.2 M potassium borate buffer, pH 9.0, 0.02% Tween 20), peristaltic pump, autosampler with injection valve, enzyme column and the flow through cell with oxygen electrode, all linked by 0.81 mm i.d. teflon tubing. Details are described by Schoemaker & Spener (1994). Reagents LOX (EC 1.13.11.12), a lyophilized preparation derived from soybean, was obtained from Serva (Heidelberg, Germany, cat. no. 28021.02). One unit will cause an increase in E234 of 0.001 per min at pH 9.0 at 25°C with linoleic acid as substrate. Lipase (EC 3.1.1.3) (Pseudomonas spec.) and esterase (EC 3.1.1.1) (hog liver) were purchased from Boehringer Mannheim (Mannheim, Germany). One unit lipase will liberate 1.0 mol fatty acid from olive oil per min at pH 8.0 at 25°C. One unit esterase will hydrolyse 1.0 mol of ethyl butyrate per min at pH 8.0 at 25°C. The polyethylene membrane, the Spectrapor and Cuprophan dialysis membranes were
Biosensors & Bioelectronics
obtained from PGW Medingen (Freital, Germany), Spectrum Medical Industries (Los Angeles, USA), and Akzo (Wuppertal, Germany), respectively. The Thomapor and Nucleopor dialysis membranes were obtained from Reichelt Chemietechnik (Heidelberg, Germany). Eupergit C 250 L (particle size 200–250 m and pore size 0.2–5.0 m), was purchased from Ro¨hm (Darmstadt, Germany), Brij 96 from ICI (Essen, Germany). Other reagents used were gelatine, linoleic acid (99%), ␣-linolenic acid (98%), triolein (99%), trilinolein (99%), Tween 20, sodium azide, sodium sulfite and sodium cholate, all obtained from Sigma (Deisenhofen, Germany). Real samples were purchased from local supermarkets. All reagents were used as received without further purification, aqueous solutions were prepared with doubly distilled water. Buffers were stirred in an open flask at room temperature for about 0.5 h before use to attain air saturation.
Determination of the apparent Michaelis– Menten coefficients (Kmapp) For the preparation of 10 mM aqueous stock solutions linoleic acid and ␣-linolenic acid were dispersed in Tween 20 as described elsewhere (Axelrod et al., 1981). The same stock solutions were used in other experiments (i.e. calibration, multiple oxygenation, dialysis exchange). The calibration data of the sensors were used to calculate the apparent Michaelis–Menten coefficients, using electrochemical Eadie–Hofstee plots (Gregg & Heller, 1990). To obtain data for the soluble enzyme, 48.0·103 U LOX were dissolved in 200 ml 0.2 M potassium borate buffer (pH 9.0). 2 ml of the enzyme solution was placed into a 2 ml stirred measuring cell with an incorporated Clark-type oxygen electrode. After a steady baseline current was reached, an appropriate volume of the 10 mM fatty acid stock solution was added to obtain the desired concentration. Due to the oxygen consumption in the course of the enzymatic reaction, the reduction current reached a minimum at equilibrium and rose again to its original value as the measuring cell again became air saturated. The height of the signal corresponded to the substrate concentration and was used to calculate the Kmapp of the soluble enzyme.
The lipoxygenase sensor
Batch system LOX was entrapped in gelatine according to Pfeiffer & Wollenberger (1993). To prepare the enzyme electrode, the enzyme layer was sandwiched between an oxygen permeable polyethylene membrane and a dialysis membrane. This membrane sandwich was attached to the tip of the Clark electrode (polyethylene membrane inside). The enzyme electrode was introduced into a stirred measuring cell containing 2 ml 0.2 M potassium borate buffer (pH 9.0). After the baseline current became steady, an appropriate volume of sample was added. The output of the electrochemical detector was displayed on a strip-chart recorder Labograph E 586 (Metrohm, Herisau, Switzerland). Flow system LOX-immobilization on Eupergit C 250 L was done according to the manufacturer’s instructions. The enzyme (100 mg) was dissolved in 6 ml 1.0 M potassium borate buffer (pH 9.0). The solution was added to 1.0 g Eupergit C250 L and after shaking, the resulting suspension was stored in the dark at room temperature. After 72 h incubation, the suspension was filtered on a Bu¨chner funnel and the Eupergit with the immobilized enzyme was washed three times with 20 ml 0.2 M potassium borate buffer (pH 9.0) to remove noncovalently bound enzyme. The immobilized enzyme was stored at 4°C in washing buffer containing 0.01% sodium azide. To build the enzyme reactor, the immobilized LOX was packed into a glass column (0.3 × 5.0 cm) (Omnifit, Cambridge, GB) and incorporated into the FIA-system described above (Schoemaker & Spener, 1994). Sample management and data acquisition were controlled by a personal computer, the communication software (FIABOLO) was developed in-house (Schreiber et al., 1997). Multiple oxygenation The first and second oxygenation activity of LOX were studied spectrophotometrically (UV/VIS Spectrophotometer Lambda 2, Perkin Elmer, ¨ berlingen, Germany). Buffers used were sodium U acetate (pH 4.5–6.0), sodium phosphate (pH 6.0– 8.0) and potassium borate (pH 8.0–10.5), each 0.2 M. At pH 6.0 the enzyme activity was the same in the acetate and phosphate buffer, at pH 1091
Michael Schoemaker et al.
8.0 in the phosphate and borate buffer. Concentrations of 13-hydroperoxyoctadecadienoic acid (13-HPOD) and 13-hydroperoxyoctadecatrienoic acid (13-HPOT), the first oxygenation products of linoleic and ␣-linolenic acid, respectively, and 9,16-di-HPOT, the second oxygenation product of ␣-linolenic acid, were calculated using molar absorption constants of 25,000 at 234 nm (Axelrod et al., 1981), 27,500 at 235 nm (Kim & Sok, 1989), and 40,000 at 267 nm (Sok & Kim, 1990), respectively. Determination of first oxygenation activity Buffer (2.88 ml) and 10 mM fatty acid stock solution (0.09 ml) were pipetted into a 3 ml quartz cuvette. After starting the reaction with the addition of 0.03 ml LOX solution (8.0 · 10−3 U/ml LOX in the corresponding buffer), the increase in absorbance at 234 nm (in the case of linoleic acid) or 235 nm (in the case of ␣-linolenic acid) was determined. Determination of second oxygenation activity Buffer (2.96 ml) and 10 mM ␣-linolenic acid stock solution (0.02 ml) were pipetted into a 3 ml quartz cuvette. After starting the reaction with the addition of 0.02 ml LOX solution (75.0 · 103 U/ml LOX in the corresponding buffer), the increase in absorbance at 267 nm was determined. Dialysis exchange method
Biosensors & Bioelectronics
was divided into 0.1 ml portions in 1 ml tubes and lyophilized (Beta 1-8K, Christ, Osterode, Germany), resulting in 50 ready to use lipolysis vials (Feldbru¨gge et al., 1994). The vials could be stored at 4°C for six months without loss of activity. Hydrolysis of triglycerides Triolein/trilinolein mixtures, oil samples and margarines, respectively, were mixed with an equal amount of Brij 96, which in case of triglycerides was a far better emulsifier than Tween 20. The mixture was vigorously stirred and 5 ml hot water (50°C) was added. After 10 min sonication (Sonorex Super RK 103 H, Bandelin, Berlin, Germany) a further dilution with water up to 100 ml yielded an emulsion of about 0.5 mM triglyceride. An aliquot of 0.5 ml was transferred to a lipolysis vial and incubated for 15 min at room temperature. Then the vial was transferred to the autosampler for analysis. Determination of water content in margarine samples In contrast to vegetable oils, margarines contain considerable amounts of water. Since the content of EFAs relates only to the fatty part of the margarines, the water content must be accounted for in the calculations. Thus, the margarine was kept at 103°C until constant weight was observed, the resulting difference representing the water content.
A dialysis exchange chamber was built to study the dialytic behaviour of linoleic and ␣-linolenic acids across different membranes. A solution of 5 ml 1 mM fatty acid in 0.2 M potassium borate buffer (pH 9.0) was injected on one side of the membrane and LOX (1.5·103 U), dissolved in 5 ml 0.2 M potassium borate buffer (pH 9.0), on the other side. The solution on both sides of the membrane was stirred with magnetic stirring bars. Upon penetration of linoleic and ␣-linolenic acids into the LOX compartment they were immediately oxidized to 13-HPOD and to 13-HPOT, respectively, and were determined spectrophotometrically.
Gas chromatography
Procedures
RESULTS AND DISCUSSION
Lipolysis vials Lipase (375 U), esterase (375 U) and 107.5 mg sodium cholate were dissolved in 5 ml 0.5 M sodium phosphate buffer (pH 8.0). This solution 1092
As a reference, the fatty acid composition of real samples was determined by GC, kindly carried out by the Chemisches Untersuchungsamt, Hagen, Germany. Analyses were done according to American Oil Chemist’s Society official method Ce 1e-89 with the following modifications: A 100 m, 0.25 mm i.d., fused silica-column covered with a 0.2 m film of Supelco SP 2560 was used; Injector PTV 60°C/500°C in 0.1 min, then constantly at 300°C; FID-detector at 270°C; Temperature programme 60°C (1 min), 3°C/min, 240°C (10 min).
Single and multiple oxygenation of the substrates To check whether the soybean LOX had the same affinity to linoleic and ␣-linolenic acid,
Signal to noise ratio = 3.bApparent Michaelis–Menten coefficient, derived from electrochemical Eadie–Hafstee plots.cRelative standard deviation, n = 10.dLipoxygenase (480 U) in 2 ml measuring cell, mean value (n = 2).
0.01–0.20 0.04–2.00 0.02–0.20
linoleic acid 16.3 ± 0.1 17.3 ± 0.2 103 ± 2
␣-linolenic acid 25.9 ± 0.1 17.3 ± 0.2 105 ± 3
linoleic acid 7.7 14.0 4.9
␣-linolenic acid 4.8 14.0 4.9
linoleic acid 0.25 ± 0.03 4.3 ± 0.1 0.70 ± 0.05
␣-linolenic acid 0.60 ± 0.02 4.2 ± 0.4 0.72 ± 0.07
3.6 2.8 — a
Fatty acids have a poor water solubility and, in the deionized state upon reaching the critical micellar concentration, form micelles. This aggregation depends on the chain length of the fatty acid, the number of double bonds present in the
Sensors
Dialytic behaviour of fatty acids
Enzyme electrode Enzymatic FIA Soluble LOX d
KmappbmM Detection limitaM Sensitivity nA mM−1 Linear range mM
Characteristics of the lipoxygenase reaction in the sensors TABLE 1
we determined the apparent Michaelis–Menten coefficients of the dissolved enzyme. The values of (0.70 ± 0.05) mM and (0.72 ± 0.07) mM for linoleic acid and ␣-linolenic acid, respectively, were identical within the margin of error (Table 1). Soybean LOX isoform I has been reported to act upon polyunsaturated fatty acids containing a suitably positioned all-cis-1,4,7-octatriene moiety to generate the dihydroperoxy derivatives while consuming 2 mol O2 per mol of substrate (Bild et al., (1977a, b); van Oz et al., 1981; Kim & Sok, 1989). The enzyme’s preferred site of first oxygenation is the 6 carbon atom of the fatty acid (i.e. the sixth carbon atom from the methyl end). LOX may catalyse a second oxygenation if the first oxygenation product still exhibits a cis, cis-1,4-pentadiene system as do the oxygenation products of arachidonic and ␥-linolenic acids (Bild et al., (1977a, b); Kim & Sok, 1989). It has been thought for a long time that ␣-linolenic acid with the first double bond at the 3 carbon, is not subject to a second oxygenation step, as the product of the first oxygenation, 13-HPOT, lacks a cis, cis-1,4-pentadiene system. Sok & Kim (1990), however, showed recently that soybean LOX is capable of catalysing multiple oxygenation of ␣-linolenic acid, thus generating 9,16-di-HPOT. If such a multiple oxygenation takes place, any LOX based sensor with an oxygen electrode as transducer will show a higher sensitivity for ␣-linolenic acid (3 fatty acid) as compared to linoleic acid (6 fatty acid). Fig. 1 shows the pH-profiles of the oxygenation activity of the enzyme. Optimum pH for the first oxygenation activity was in the range of pH 9.0 for both, linoleic and ␣-linolenic acids. The maximum activity observed for the second oxygenation of ␣-linolenic acid was at pH 6.0. Sok & Kim (1990) found a pH-optimum of 8.5 for the formation of 9,16-di-HPOT using 9-HPOT as substrate instead of ␣-linolenic acid, yet in our experiments at pH 9.0 the second oxygenation activity was negligible. Consequently, all biosensor measurements were carried out at pH 9.0, thus excluding multiple oxygenation.
The lipoxygenase sensor
RSDc
Biosensors & Bioelectronics
1093
Michael Schoemaker et al.
Biosensors & Bioelectronics
linolenic acids across different dialysis membranes, as it was planned to fix the enzyme layer with such a membrane to the tip of the oxygen electrode. In all cases, however, ␣-linolenic acid diffused faster across the membrane than linoleic acid. Enzyme electrode characterization
Fig. 1. pH-dependence of LOX action on linoleic (A) and ␣-linolenic acid (B). First oxygenation activity (쎲), 0.24·103 U LOX, 0.3 mM fatty acid; second oxygenation activity (䊊), 1.5·103 U LOX, 0.07 mM fatty acid; 3 ml cuvettes. Buffers used were sodium acetate from pH 4.5–6.0, sodium phosphate from pH 6.0–8.0, potassium borate from pH 8.0–10.5.
molecule, and the physico-chemical properties of the solvent (Vorum et al., 1992). The increasing number of double bonds affects the critical micellar concentration, which may result in different accessibilities of linoleic acid and ␣-linolenic acid to the enzyme layer in the sensor. Table 2 summarizes the diffusion rates of linoleic and ␣-
The enzyme was immobilized in gelatine, the resulting enzyme membrane was fixed with a dialysis membrane in front of the Clarke type oxygen electrode. In order to optimize the immobilization, the amount of gelatine as well as the enzyme loading were varied. An enzyme membrane consisting of 3% gelatine and a minimum concentration of 1.5·103 U LOX cm−2 membrane was found to be optimal. With increasing gelatine concentrations a decrease in sensitivity was observed due to the more difficult diffusion of the substrate in the enzyme layer (results not shown). Gelatine contents below 3% led to a considerable loss in membrane stability, which made the assemblage of the sensor very difficult and gave unreproducible results. Enzyme loadings of 1.5·103 U LOX cm−2 membrane ensured diffusion controlled conditions (Fig. 2). In subsequent experiments an enzyme loading of 5·103 U LOX cm−2 membrane provided the sensor with a prolonged lifetime. The effect of pH in the range from 4.5 to 10.5 on the electrode response was studied with 0.25 mM linoleic acid as substrate. Two pHruns were performed, each with a new enzyme electrode. Starting from pH 6.5 the sensor response increased rapidly to reach a plateau from pH 8.5 to 10.0 (results not shown) in accordance with the pH-optimum of 9.0 for soluble LOX (Axelrod et al., 1981). No significant change was observed when the pH-profile was determined in
TABLE 2 Fatty acid diffusion across dialysis membranes Membrane
Cuprophan Thomapor Spectrapor Nucleopor a
MWCOa kDa
Thickness m
8–15 10–20 25 100
Molecular weight cut-off of the membrane.
1094
8 20 20 10
Diffusion rate mol min−1 cm−2 linoleic acid
␣-linolenic acid
0.22 0.02 0.06 0.40
0.51 0.08 0.13 0.73
Ratio linoleic acid/␣linolenic acid 0.43 0.25 0.46 0.55
Biosensors & Bioelectronics
The lipoxygenase sensor
mM−1 and (25.9 ± 0.1) nA mM−1, respectively. This difference made the determination of essential fatty acids difficult, because the contents of linoleic and ␣-linolenic acids vary in foods depending on the origin of the oil or fat. Based on our experience described above we could exclude different substrate affinities of the enzyme and multiple oxygenation of ␣-linolenic acid, whereas the different dialytic behaviour of linoleic acid and ␣-linolenic acid gave rise to the different sensitivities observed. Characterization of the enzymatic FIA-system
Fig. 2. Dependence of the enzyme electrode response on enzyme loading. LOX in 3% gelatine, covered with Thomapor dialysis membrane. Substrate 0.1 mM linoleic acid. Mean of experiments with two enzyme membranes.
the opposite direction, which gave evidence for the high pH-stability of the immobilized enzyme. Calibration plots showed different sensitivities of linoleic and ␣-linolenic acids (Fig. 3), the slopes in the linear ranges were (16.3 ± 0.1) nA
Fig. 3. Calibration curves for linoleic (쎲) and ␣-linolenic acids (䊏). 5·103 U LOX cm−2 in 3% gelatine, covered with Cuprophan dialysis membrane. Each point represents the mean of two experiments.
To overcome the problem caused by enzyme immobilization with membranes as supporting materials and to make analysis more effective and convenient, we developed a fully automated FIA-system. The enzyme was covalently bound onto an oxirane-activated methacrylic amide polymer (Eupergit) and the resulting beads were filled into a glass column. Enzymatic turnover after addition of the analyte was monitored with an oxygen electrode placed in a flow through cell at the end of the column. Sensor response and baseline reversion time, i.e. the time that had to pass before the current returned to the baseline, decreased with increasing flow rate (Fig. 4). To check whether this effect was due to the reduced reaction time or to the response time of the oxygen electrode, a sodium sulfite solution was injected into the flow through cell. Sodium sulfite reduced dissolved oxygen and thus decreased the oxygen concentration in the solution. This experiment revealed a dependence of the sensor response on flow rate similar to that of linoleic acid (Fig. 4), giving clear evidence that the loss in sensitivity was due to the slow response of the electrode. Spectrophotometrical product analysis with LOX added to the column effluent showed that even at the highest flow rates tested the residence time of the sample in the enzyme column was sufficient for completion of the enzymatic reaction (data not shown). Higher amounts of substrate in increased sample volumes led to an increase in sensitivity but also to a decrease in the linear range (Fig. 5). Taking into account the sensitivity of analysis, sample throughput, and linear range, a flow rate of 0.45 ml min−1 and a sample loop of 20 l were selected for all subsequent studies. With this flow rate, each analysis could be performed in 5 min, giving a throughput of 12 assays per hour. Due 1095
Michael Schoemaker et al.
Biosensors & Bioelectronics
Fig. 4. Influence of flow rate on FIA-system performance: 2.0 mM linoleic acid (쎲,䊏); 17.9 mM sodium sulfite (䊊). Sample volume 20 l. Sensor response at lowest flow rate was set at 100%. Each point represents the mean of two experiments.
Fig. 5. Dependence of FIA-system response on sample volume: 10 l (쎲); 20 l (왔); 50 l (䊏); 100 l (왖); 200 l (䉬). Flow rate, 0.45 ml min−1. Each point represents the mean of two experiments.
to the completion of the enzymatic reaction in the column, calibration plots of linoleic and ␣linolenic acids were identical within the margin of error (Fig. 6). Comparison of the sensors The characteristics of the enzyme electrode and the enzymatic FIA-system are summarized and 1096
Fig. 6. Calibration curves for linoleic acid (쎲) and ␣linolenic acid (䊏) determined with the enzymatic FIAsystem. Sample volume 20 l; flow rate 0.45 ml min−1. Each point represents the mean of two experiments.
compared with the soluble enzyme in Table 1. It should be pointed out that kinetic measurements in the case of the immobilized enzyme were performed under diffusion limitation. Thus, the data shown do not reflect intrinsic properties of the enzyme itself, but characterize the enzyme electrode and the flow injection system (Kamin & Wilson, 1980; Gregg & Heller, 1990). In accord-
Biosensors & Bioelectronics
ance with the different dialytic behaviours of linoleic and ␣-linolenic acids, data concerning sensor performance such as sensitivity and detection limit, as well as the apparent Km of immobilized LOX for the fatty acids differed in the case of the enzyme electrode. The use of the enzymatic FIA-system widened the linear range, which increased the ratio between the upper and lower limit 2.5 fold with respect to the ratio achieved with the enzyme electrode or soluble enzyme. The operational stabilities of the enzyme electrode and FIA-system are shown in Fig. 7. The enzyme electrode was stored in 0.2 M potassium borate buffer (pH 9.0) at 4°C when not in use. After 19 days a 50% decay of initial sensor response was observed. The enzyme column was stored in 0.2 M potassium borate buffer (pH 9.0) containing 0.01% sodium azide as a preservative at room temperature when not in use. The FIAsystem lost half of its initial response after 37 days. In permanent use, the FIA-system lost 8.4% of its initial sensitivity after 1500 injections of 2.0 mM linoleic acid within six days. The enhanced long term stability of the FIA-system was most likely due to the high enzyme loading in the column. Besides the fact that the FIA-system was fully automated, the advantage of the underlying concept can be summarized as follows: Since in the FIA-system the transducer and enzyme compart-
Fig. 7. Operational stability for the enzyme electrode (䊊) and the enzymatic FIA-system (쎲). Sensor response upon first injection of substrate was set at 100%. Each point represents the mean of two experiments.
The lipoxygenase sensor
ment were separated, the replacement of a spent column by a new one was much easier than the changing of membranes in the enzyme electrode. With appropriate fittings a column replacement was carried out in less than 10 min, unmatched in the case of membrane exchange. Moreover, the immobilization of LOX on Eupergit required less working steps than membrane immobilization and revealed an excellent mechanical stability which made the loading of the columns very convenient. Hydrolysis of triglyceride samples Fatty acids in naturally occurring oils and fats are esterified to yield triglycerides, but only the free fatty acids can serve as substrates for LOX. Thus, prior to analysis triglycerides were hydrolysed with a lyophilized lipase/esterase mix. At first the sample was emulsified with the aid of Brij 96 and an aliquot of the emulsion was transferred to the lipolysis vial containing the lipase/esterase mix. After 15 min incubation at room temperature the emulsion became clear. To prove this point different triolein/trilinolein mixtures of known composition were tested (Fig. 8). The equation of the linear regression was y = 1.02 − 1.22 (r = 0.999, n = 6), where y and
Fig. 8. Comparison of linoleic acid content in different trilinolein/trilein mixtures determined after hydrolysis by the enzymatic FIA-system and calculated from the known composition. Hydrolysis of each mixture and analysis were done in duplicate, each point represents the mean (n = 4). 1097
Michael Schoemaker et al.
Biosensors & Bioelectronics
TABLE 3 Determination of EFA contents in oils and fats for human consumption Sample
Essential fatty acids (wt %)a FIA-system reference
Olive oil I Olive oil II Sunflower oil I Sunflower oil II Sunflower oil III Soybean oil Wheat germ oil Diabetic oil Grape germ oil I
5.3 ± 0.1 13.4 ± 0.6 60.1 ± 0.6 65.9 ± 0.9 63.3 ± 1.9 61.6 ± 0.1 61.3 ± 0.6 62.6 ± 1.5 67.4 ± 0.1
4.7b 8.8c 63.2b 63.0c 64.0c 62.0c 65.1b 70.0c 60.0c
Sample
Essential fatty acids (wt %)a FIA-system reference
grape germ oil II safflower oil I safflower oil II walnut oil
73.6 71.7 75.6 74.1
± ± ± ±
0.6 2.7 0.5 1.0
72.1b 74.9b 75.0c 70.8b
margarine I margarine II margarine III margarine IV
15.5 24.0 27.7 34.1
± ± ± ±
0.4 1.0 1.0 0.8
19.0b 20.0c 26.8b 39.5b
a
wt % of total fatty acids in triglyceride.bGas chromatography.cProvided by manufacturer. Data obtained by the FIA-system are the mean of four determinations.
x are the linoleic acid concentrations determined with the FIA-system and calculated from the known composition, respectively. These data indicated that hydrolysis of triglycerides indeed was complete after 15 min. Due to the presence of the detergent which facilitated the formation of oil–water interfaces necessary for lipase action (Brockman et al., 1988), organic solvents were also not required in this preincubation step. Application of the enzymatic FIA-system Due to the difficulties encountered with the enzyme electrode sensor, we opted for the FIAsystem to test the performance of the lipoxygenase sensor in combination with the lipolysis vials with real samples, i.e. 13 different vegetable oils and four margarines. The values obtained were compared with those provided either by the manufacturer or by GC determination. Table 3 summarizes the results of these analyses. The linear correlation constant calculated from these data is 0.987. Certainly GC offers a much lower detection limit and higher sensitivities than the enzymatic FIA-system presented in this paper, yet essential fatty acids are not trace elements in foods and the sensitivity of the FIA-system was totally sufficient.
CONCLUSION The FIA-system employing lipoxygenase proved its feasibility for the determination of essential fatty acid contents in fats and oils without using organic solvents. The fully automated system, 1098
combining an enzyme column and a flow through cell with an incorporated oxygen electrode, was superior to an enzyme electrode, as it responded with equal sensitivities towards the two most abundant EFAs, linoleic acid and ␣-linolenic acid. The enzymatic FIA-system performed excellently in the determination of EFA contents in vegetable oils and margarines. Vis-a`-vis the established GC analysis after base-catalysed methanolysis of triglycerides, the method elaborated here represents a promising alternative in terms of mild conditions, cost and time saving.
ACKNOWLEDGEMENTS This work was supported in part by the Stiftungsfonds Unilever (TS 022/53.1). We also thank C. Gertz, Chemisches Untersuchungsamt, Hagen, Germany, for GC analyses. The work was part of the Ph. D. thesis of Michael Schoemaker.
REFERENCES Axelrod, B., Cheesbrough, T. M. and Laakso, S. (1981) Lipoxygenase from soybeans. Methods Enzymol. 71, 441–451. Bild, G. S., Ramadoss, C. S. and Axelrod, B. (1977a) Multiple dioxygenation by lipoxygenase of lipids containing all-cis-1,4,7-octatriene moieties. Arch. Biochem. Biophys. 184, 36–41. Bild, G. S., Ramadoss, C. S., Lim, S. and Axelrod, B. (1977b) Double dioxygenation of arachidonic acid by soybean lipoxygenase-1. Biochem. Biophys. Res. Commun. 74, 949–954. Brockman, H. L., Momsen, W. E. and Tsujita, T.
Biosensors & Bioelectronics (1988) Lipid–lipid complexes: properties and effects on lipase binding to surfaces. J. Am. Oil Chem. Soc. 65, 891–896. Feldbru¨gge, R., Renneberg, R. and Spener, F. (1994) Development and practical evaluation of an amperometric triglyceride sensor. Sensors and Actuators B 19, 365–367. Galli, C., Galella, G. & Marangoni, F. (1993) Dietary fatty acids and tumorigenesis. In Antioxidants, Free Radicals and Polyunsaturated Fatty Acids in Biology and Medicine, ed. A. T. Diplock, J. M. C. Gutteridge & V. K. S. Shukla, pp. 199–213. International Food Science Centre, Lystrup. Gregg, B. A. and Heller, A. (1990) Cross-linked redox gels containing glucose oxidase for amperometric biosensor applications. Anal. Chem. 62, 258–263. Hall, G. F., Best, D. J. and Turner, A. P. F. (1988) The determination of p-cresol in chloroform with an enzyme electrode used in organic phase. Anal. Chim. Acta 213, 113–119. Hall, G. F. and Turner, A. P. F. (1991) An organic phase enzyme electrode for cholesterol. Anal. Lett. 24, 1375–1388. Horrobin, D. F. (1993) Medical uses of essential fatty acids. In Antioxidants, Free Radicals and Polyunsaturated Fatty Acids in Biology and Medicine, ed. A. T. Diplock, J. M. C. Gutteridge & V. K. S. Shukla, pp. 181–198. International Food Science Centre, Lystrup. Kamin, R. A. and Wilson, G. S. (1980) Rotating ringdisk enzyme electrode for biocatalysis kinetic studies and characterization of the immobilized enzyme layer. Anal. Chem. 52, 1198–1205. Kim, M. R. and Sok, D. E. (1989) Formation of 6,13dihydroxyoctadecatrienoic acid isomers from ␥linolenic acid. Biochem. Biophys. Res. Commun. 159, 1154–1160. Mannino, S., Cosio, M. S. and Wang, J. (1994a) Determination of peroxide value in vegetable oils by an organic-phase enzyme electrode. Anal. Lett. 27, 299–308. Mannino, S., Cosio, M. S. and Wang, J. (1994b) Organic-phase enzyme biosensor for moisture determination in food products. Analyst 119, 2001–2003. Pfeiffer, D. and Wollenberger, U. (1993) Preparation and characterization of enzyme electrodes using entrapment in gelatin. Biosens. Bioelectron. 8, xix–xx. Saini, S., Hall, G. F., Downs, M. E. A. and Turner, A. P. F. (1991) Organic phase enzyme electrodes. Anal. Chim. Acta 249, 1–15. Schoemaker, M. and Spener, F. (1994) Enzymatic flowinjection analysis for essential fatty acids. Sensors and Actuators B 19, 607–609. Schreiber, A., Feldbru¨gge, R., Key, G., Glatz, J. F. C. & Spener, F. (1997) An immunosensor based
The lipoxygenase sensor on disposable electrodes for rapid estimation of fatty acid-binding protein, an early marker of myocardial infarction. Biosens. Bioelectron., in press. Schubert, F., Saini, S. and Turner, A. P. F. (1991) Mediated amperometric enzyme electrode incorporating peroxidase for the determination of hydrogen peroxide in organic solvents. Anal. Chim. Acta 245, 133–138. Simopoulos, A. P. (1994) Fatty acids. In Functional Foods, ed. I. Goldberg, pp. 355–392. Chapman & Hall, New York. Sinclair, H. M. (1956) Deficiency of essential fatty acids and atheriosclerosis. Lancet 1, 381–383. Sok, D. E. and Kim, M. R. (1990) Enzymatic formation of 9,16-dihydro(per)oxyoctadecatrienoic acid isomers from ␣-linolenic acid. Arch. Biochem. Biophys. 277, 86–93. Valencia-Gonza´les, M. J. and Dı´az-Garcia, M. E. (1994) Enzymatic reactor/room temperature phosphorescence sensor system for cholesterol determination in organic solvents. Anal. Chem. 66, 1716–1731. Van Oz, C. P. A., Rijke-Schilder, G. P. M., Van Halbeek, H., Verhagen, J. and Vliegenthart, J. F. G. (1981) Double dioxygenation of arachidonic acid by soybean lipoxygenase-1. Biochim. Biophys. Acta 663, 177–193. Vorum, H., Brodersen, R., Kragh-Hansen, U. and Pedersen, A. O. (1992) Solubility of long chain fatty acids in phosphate buffer at pH 7.4. Biochim. Biophys. Acta 1126, 135–142. Wang, J. (1993) Organic-phase biosensors—New tools for flow analysis: A short review. Talanta 40, 1905–1909. Wang, J., Dempsey, E., Eremenko, A. and Smyth, M. R. (1993c) Organic-phase biosensing of enzyme inhibitors. Anal. Chim. Acta 279, 203–208. Wang, J. and Lin, Y. (1993) On-line organic-phase enzyme detector. Anal. Chim. Acta 271, 53–58. Wang, J., Lin, Y. and Chen, L. (1993a) Organic-phase biosensors for monitoring phenol and hydrogen peroxide in pharmaceutical antibacterial products. Analyst 118, 277–280. Wang, J., Lin, Y. and Chen, Q. (1993b) Organic-phase biosensors based on the entrapment of enzymes within poly(ester-sulfonic acid) coatings. Electroanalysis 5, 23–28. Wang, J., Naser, N., Kwon, H. S. and Cho, M. Y. (1992b) Tissue bioelectrode for organic-phase enzymatic assays. Anal. Chim. Acta 264, 7–12. Wang, J. and Reviejo, A. J. (1993) Organic-phase enzyme electrode for the determination of trace water in nonaqueous media. Anal. Chem. 65, 845–847. Wang, J., Reviejo, A. J. and Mannino, S. (1992a) Organic-phase enzyme electrode for the determination of phenols in olive oils. Anal. Lett. 25, 1399–1405. 1099