- Email: [email protected]
A needle-type optical enzyme sensor system for determining glucose levels in fish blood
Analytica Chimica Acta 573–574 (2006) 117–124
A needle-type optical enzyme sensor system for determining glucose levels in fish blood Hideaki Endo a,∗ , Yuki Yonemori a , Kazuya Musiya a , Masashi Maita b , Toru Shibuya c , Huifeng Ren a , Tetsuhito Hayashi a , Kohji Mitsubayashi d a Department of Ocean Sciences, Tokyo University of Marine Science and Technology, Konan 4-5-7, Minato, Tokyo 108-8477, Japan Graduate School of Marine Science and Technology, Tokyo University of Marine Science and Technology, Konan 4-5-7, Minato, Tokyo 108-8477, Japan c Toyo Gosei Co. Ltd., 2-1 Wakahagi-4 Inba-mura, Inba-gun, Chiba 270-1609, Japan d Department of Biomedical Devices and Instrumentation, Tokyo Medical and Dental University, 2-3-10 Kanda-Surugadai, Chiyoda, Tokyo 101-0062, Japan b
Received 29 November 2005; received in revised form 21 April 2006; accepted 24 April 2006 Available online 4 May 2006
Abstract A needle-type biosensor system was developed for rapid and simple determination of glucose levels in fish blood. The sensor comprises a needle-type hollow container (18-gauge needle), immobilized enzyme membrane and optic fiber probe with ruthenium complex. The enzyme membrane was prepared from glucose oxidase, azide-unit pendant water-soluble photopolymer and an ultra-thin dialysis membrane. The optic fiber probe was inserted into the rolled enzyme membrane placed in the needle-type hollow container. The calibration curve was linear for glucose levels in fish plasma. One assay was completed within 3 min. A good reproducibility was observed for 60 times without exchange of the enzyme membrane. The sensor was inserted into the caudal vein of fish to measure blood glucose levels. The sensor responded immediately after insertion and glucose levels could be monitored. Good correlations were observed between values determined using the sensor and conventional methods in the range of 48–157 mg dl−l (correlation coefficient, 0.9474). © 2006 Elsevier B.V. All rights reserved. Keywords: Biosensor; Enzyme sensor; Needle; Glucose; Blood; Plasma; Fish
1. Introduction In recent years, Hazard Analysis Critical Control Point (HACCP) systems have been actively adopted by the aquatic food industry. In the field of aquaculture involved in supplying fish, use of HACCP is under close examination for producing healthy fish. Mass outbreaks of pathogenic bacterial infections have killed large quantities of fish in recent years, causing great financial damage to fish farms. In addition, from the perspective of food hygiene for delivering safe fish to markets, farming fish with minimal dependence on antibiotics is important. Periodic checks of fish health and detection of abnormalities as quickly as possible are thus necessary at fish farms. Several studies have shown blood glucose levels closely correlate to the level of stress in fish and represent the state of respiratory or nutri-
∗
Corresponding author. Tel.: +81 3 5463 0616; fax: +81 3 5463 0616. E-mail address: [email protected] (H. Endo).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.04.068
tional disturbance [1–6]. Recent studies have also clarified that decreased levels of blood cholesterol represent a useful indicator of reduced resistance to bacterial infection [7,8]. In this manner, determination of blood glucose and cholesterol is important for ascertaining the health of fish. At present, these tests are conduced using clinical laboratory test kits that have been designed for humans. Furthermore, as each sample needs to be analyzed separately, testing is time- and labor-intensive. When ascertaining the health of fish at fish farms, testing as many fish as possible in a fast and convenient manner is crucial. Recent research has been actively investigating the development of biosensors consisting of an immobilized enzyme membrane and an electrochemical device. We have already developed a biosensor to measure cholesterol levels in fish plasma, in an attempt to establish rapid and convenient methods to ascertain fish health [9]. Since this method is based on flow injection analysis (FIA), continuous measurement is possible and only 5 min is needed to analyze each sample, markedly reducing testing time. However, even this method requires a certain amount of skill
118
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
to draw blood, and since flow injection analysis is employed, the measurement system is complicated and large. Tests can be performed in a laboratory, but convenient testing at fish farms is difficult. Development of a needle-type enzyme sensor system is needed, so a biosensor can be directly inserted into fish to quantify blood components. In the field of clinical medicine, many needle-type enzyme sensors have been developed to measure blood glucose [10–15]. Wang and Zhang [14] developed a miniature needle-type enzyme sensor suitable for the simultaneous amperometric monitoring of glucose and insulin. The sensor was constructed from dual modified carbon-paste working electrodes inserted into a 14-gauge needle. Wu et al. [12] reported an enzyme sensor for whole blood lactate monitoring. The sensor consisted of a stainless steel needle with surface modified by a polymer material for enzyme immobilization. Studies of non-invasive glucose sensors have also been undertaken for application in the field of diabetes [16–18]. Caduff et al. [17] developed a glucose monitoring system based on impedance spectroscopy. Changes in the glucose concentrations can be monitored by varying the frequency in the radio band over a range, optimized to measure the impact of glucose on the impedance pattern. However, most sensor systems are designed for humans or livestock, and such sensors have rarely been used in fish. For application as fish, sensors need to be fairly strong, as fish might resist blood collection. In addition, since testing should be conducted at the fish farm if possible, portability is an important issue. Moreover, with many needle-type enzyme sensors, when a sensor is inserted into a specimen, immobilized enzyme becomes detached when coming into contact with the specimen surface. The method of immobilizing enzyme in a narrow needle-type detector is thus central to the design of a practical needle-type enzyme sensor. The application of fiber optics to the preparation of biosensor systems has been studied. Several fiber optic biosensors using fluorescence of ruthenium complexes for the determination of glucose have been developed [19,20]. The principle of the sensors is based on the monitoring of oxygen consumption due to enzymatic reaction of glucose oxidase. In the present study, to establish a rapid and convenient method for measuring blood components in fish, a needle-type enzyme sensor was developed in which an immobilized enzyme membrane and optic fiber probe using fluorescence of ruthenium complexes were inserted into a needle-type hollow container to measure blood glucose levels. 2. Materials and methods 2.1. Reagents Glucose oxidase (EC 1.1.3.4, Type VII from Aspergillus niger) was obtained from Sigma (MD, USA). Ultra-thin dialysis membrane (Part No. 157-0144-02; thickness; 15 m) was purchased from Technicon Chemical Co. (Orcq, Belgium). Photosensitive polymer (azide-unit pendant water-soluble photopolymer (AWP)) was obtained from Toyo Gosei Kogyo Co. (Tokyo, Japan). All other reagents used for experiments were commercial- or laboratory-grade. Glucose was dissolved in
0.1 M phosphate-buffered solution (PBS; pH 7.0), titrated and stored for ≥12 h to prepare standard glucose solutions. 2.2. Needle-type enzyme sensor and measurement Fig. 1 shows a schematic of the needle-type enzyme sensor and images of the detector region. The sensor consists of a needle-type hollow container, immobilized enzyme membrane and optic fiber probe with ruthenium complex. The sensor assembly is described below. 2.2.1. Needle-type hollow container The hole at the tip of a commercially available 18-gauge needle (diameter: 1.2 mm; Terumo, Tokyo, Japan) was blocked using epoxy resin (Toagosei Co., Tokyo, Japan). Next, 4 × 1.0 mm holes were made on one side of the needle, and 4 × 0.8 mm holes were made on the opposite side (Fig. 1A and E). 2.2.2. Immobilized enzyme membrane Glucose oxidase (1 mg) was dissolved in 200 l of 0.1 M PBS (pH 7.8), and 25 l of the resulting solution was mixed with 80 mg of AWP. Next, a dialysis membrane soaked in distilled water was evenly spread over a glass plate without bubbles and then completely dried. The above-mentioned enzyme and AWP mixture was evenly applied and rubbed into the dialysis membrane. After drying for 1 h in the dark, the membrane was placed under a fluorescent lamp for 1 h to prepare an immobilized enzyme membrane. AWP is a photosensitive polymer, and hardens with time to immobilize the enzyme. The resulting membrane was cut into 8 mm × 3 mm strips, soaked in distilled water again and then dried on filter paper. Due to differences in drying shrinkage between the dialysis membrane and AWP, each strip curled and formed a tube (Fig. 1A (3)). 2.2.3. Needle-type enzyme sensor A tube-shaped immobilized enzyme membrane was inserted into the tip of the needle-type hollow container (Fig. 1C). The container was soaked in 0.1 M PBS (pH 7.0) so that the immobilized enzyme membrane would swell and adhere to the needletype hollow container. The container was also lightly tapped to eliminate air bubbles. Next, the optic fiber probe including a ruthenium complex (FOXY-AL300: Ocean Optics, FL, USA) was inserted inside the immobilized enzyme membrane (Fig. 1D) to set up the needle-type glucose sensor. 2.2.4. Oxygen measurement system An excitation light-emitting diode (LED) light source (USBLS-450; Ocean Optics, FL, USA), UBS2000 spectrometer (Ocean Optics) and VT900 personal computer (NEC, Tokyo, Japan) are connected to the needle-type enzyme sensor to set up a glucose determination system. The optic fiber probe includes the ruthenium complex at the tip. Since the distal end of the probe tip comprises a thin layer of hydrophobic sol–gel material, ruthenium complex is trapped in the sol–gel matrix. When light (475 nm) from the LED excites the ruthenium complex at the probe tip, the excited ruthenium complex fluoresces, emitting energy at 600 nm. If the excited ruthenium complex encounters
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
119
Fig. 1. Schematic of the needle-type enzyme sensor system (A, B) and images of the detector region (C–E). (1) Needle-type hollow container, (2) round-shaped holes, (3) immobilized enzyme membrane, (4) optic fiber probe, (5) ruthenium complex, (6) excitation LED light source, (7) spectrometer, (8) personal computer.
an oxygen molecule, excess energy is transferred to the oxygen molecule in a non-radioactive transfer, decreasing or quenching the fluorescence signal. Energy is collected by the probe and carried through the optical fiber to the spectrometer. An
A/D converter is used to convert analog data to digital data, which are then recorded using computer software. The amount of dissolved oxygen in the needle-type hollow container can be measured based on changes in fluorescence intensity.
120
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
2.3. Measurement of standard glucose solutions using the sensor First, the sensor was soaked in 100 ml of 0.1 M phosphate buffer solution (pH 7.0) in a state of oxygen saturation to fill the needle-type hollow container with buffer solution. Once sensor output had stabilized, the sensor was placed in a test tube containing one of the standard glucose solutions (3.6–18.0 mg dl−1 ) to measure changes in concentration of dissolved oxygen inside the hollow container due to oxidization by glucose oxidase. The glucose oxidase-sensing scheme involves the employment of glucose oxidase according to: D-glucose + O2
glucose oxidase
−→
D-gluconolactone + H2 O2
To clean the hollow container, the sensor was again placed in 100 ml of buffer solution in a state of oxygen saturation, and the sensor was placed in another test tube once output had stabilized. The above procedures were repeated to measure levels of glucose in each standard glucose solution. 2.4. Measurement of blood glucose by the sensor Blood glucose levels were measured in tirapia (Oreochromis niloticus), as this African fish is farmed in various areas of Japan. Tirapias (n = 20; weight, 300–1000 g) were raised in an 800 l tank and used as needed. In the present study, blood glucose levels were measured by collecting blood samples or direct insertion of the sensor.
2.4.1. In vitro measurement Each tirapia (weight, 800–1000 g) was captured using a net and transferred to a bowl containing 200 mg l−1 of 2phenoxyethanol. Next, a syringe that had been cleaned using 3000 units ml−1 of heparin sodium solution was inserted into the caudal vein of the fish to collect 6–8 ml of whole blood, which was transferred to a test tube. Each blood sample was centrifuged at 3000 rpm for 10 min to prepare a plasma sample. As with the standard glucose solutions, the sensor was soaked in 0.1 M PBS, and once sensor output had stabilized, the sensor was soaked in 200 l of plasma or whole blood sample in the micro-test tube (volume, 400 l) to measure blood glucose level. The above procedures were repeated to measure glucose levels in each sample. 2.4.2. In vivo measurement After anesthetizing each fish using the same method, the sensor tip was inserted into the caudal vein. Blood glucose levels were measured by immobilizing each fish for about 3 min. Once the test was completed, the fish was returned to another tank, and the sensor was cleaned using buffer solution. The above procedures were repeated to measure glucose levels in each fish. 2.5. Conventional blood glucose determination To measure blood glucose levels using conventional method for references, a commercially available colorimetric assay kit using Trinder’s reaction (C-II glucose test; Wako Pure Chemi-
Fig. 2. Response curves for standard glucose solutions. (A) Response curve at various glucose concentrations, (B) response curve at a glucose concentration of 10.8 mg dl−1 . r, enlarged view of the initial response in B. Assay conditions were as follows: temperature 25 ◦ C; pH 7.0.
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
121
cal Industries, Osaka, Japan) was employed. When each plasma sample was mixed with reagent from the kit containing enzyme and color-fixing agent, glucose was oxidized and then yielded hydrogen peroxide via glucose oxidase. The hydrogen peroxide formed created red pigmentation by quantitative oxidation condensation with phenol and 4-aminoantipyrine. Glucose level was thus determined by measuring absorbance of the red color. 3. Results and discussion 3.1. Typical response curves of the sensor system Concentrations of each standard glucose solution were measured using the enzyme sensor, and Fig. 2A shows the results. When the sensor was transferred from buffer solution to glucose solution, sensor response (oxygen concentration) increased transiently from the baseline, then gradually decreased. This was attributed to dissolved oxygen in the buffer solution in the hollow container being oxidized by the glucose oxidase, and this reduction in dissolved oxygen was detected by the sensor. In addition, the transient increase in response was caused when the sensor touched the air during transfer to the buffer or solution. Sensor response decreased with increasing glucose concentration due to the fact that oxygen is consumed by the action of glucose oxidase (Fig. 2A). Fig. 2B shows the response curve at a glucose concentration of 10.8 mg dl−1 . In the present study, based on oxygen decrease 1 or 3 min after oxygen concentration returned to baseline, slope of the regression line was calculated using the least squares method to determine oxygen reduction per unit time. With the above-mentioned methods, 1 and 3 min sensor outputs were compared using standard glucose solutions (3.6–18.0 mg dl−1 ). Little difference was observed between the two values. As the objective of the present study was to measure blood glucose levels by directly inserting the sensor into fish, testing time should be as short as possible. In subsequent experiments, 1 min oxygen reduction was utilized as analytical signal (mg l−1 min−1 ) of the sensor. Fig. 3 shows the relationship between analytical signal and concentration in standard glucose solutions. At glucose concentrations of 1.8–19.3 mg dl−1 , a
Fig. 4. Effect of temperature on analytical signal. Experimental conditions were the same as in Fig. 2, except for temperature.
Fig. 5. Effect of pH on analytical signal. Experimental conditions were the same as in Fig. 2, except for pH of the buffer solution.
favorable linear relationship was noted (correlation coefficient: 0.9947, y = 0.3124x − 0.6368). The upper limit for standard glucose solutions was around 20.0 mg dl−1 . 3.2. Effects of measurement conditions on analytical signal
Fig. 3. Relationship between analytical signal and glucose concentration in standard glucose solutions.
As the sensor comprises a biological catalyst (enzyme), response can vary depending on temperature and pH of the solutions in which the immobilized enzyme membrane is placed. We therefore investigated the effects of these factors on analytical signal. Fig. 4 shows the effects of temperature on analytical signal. Output increased as temperature increased above 15 ◦ C, then began to decrease above 35 ◦ C. Since the body temperature of fish (tirapia) farmed in temperate areas is 20–30 ◦ C, matching the ambient water temperature, the present sensor is suitable in such fish. Fig. 5 shows the effects of pH on analytical signal. Glucose levels were measured by adjusting the pH of buffer solution in the needle-type hollow container from 5.6 to 8.2, and maximum response was obtained at pH 6.5. Since the bodily fluids of fish are mostly neutral, no problems are likely to occur using the sensor in fish.
122
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
Fig. 6. Reproducibility of sensor system measurements. Measurement was performed using a 0.5 mg dl−1 glucose standard solution. Experimental conditions were the same as in Fig. 2, except for glucose concentration. Scale of the vertical axis was made equal to the vertical axis of Fig. 3. Table 1 Changes in correlation coefficient of calibration curves Days
Correlation coefficient
1 3 5 8 11 16 21 33 41 50
0.9926 0.9963 0.9973 0.9897 0.9973 0.9985 0.9863 0.9886 0.9944 0.9909
Average
0.9932
Standard deviation
0.0040
Experimental conditions such as pH and temperature were the same as in Fig. 2.
3.3. Reproducibility and stability of the sensor system Fig. 6 shows reproducibility of the sensor system. Measurement was performed at 25 ◦ C using a 0.5 mg dl−1 glucose standard solution. As shown by the results, good reproducibility was observed for ≥60 times without exchange of the enzyme membrane (standard deviation: 0.047). We also investigated stability of the sensor to gain an appreciation of likely practical operational lifetime. After glucose determination, the sensor was stored at 5 ◦ C for up to 50 days, and each determination was performed at different time points. At each time point, a glucose calibration curve was drawn and a correlation coefficient for the linear relationship was calculated. Table 1 shows changes in correlation coefficient of calibration curves over 50 days. Correlation coefficient remained extremely high, indicating that the sensor can be used for extended periods of time.
Fig. 7. Relationship between analytical signal and glucose level in fish plasma. Experimental conditions such as pH and temperature were the same as in Fig. 2.
tional assay. A fine linear relationship existed at blood glucose levels of 37–105 mg dl−1 (correlation coefficient: 0.9741). When compared to standard glucose solutions (Fig. 3), marked changes were noted in the upper limit of glucose determination. This might be attributable to enzyme inhibitors in blood and delays in substrate diffusion inside the hollow container due to blood viscosity. Since different calibrations were observed between plasma samples and standard glucose solution, a standard sample for glucose calibration to measure blood glucose levels should be prepared. Moreover, to measure blood glucose levels by inserting the needle-type sensor into fish, glucose levels need to be quantified in whole blood, not plasma. A whole blood sample was thus collected from the same tirapias, and the sensor was placed in each whole blood sample to measure glucose. Fig. 8 shows the relationship for analytical signal of the sensor between whole blood and plasma samples in each fish. A good correlation was observed (correlation coefficient: 0.9745) and a regression line of y = 1.0667x + 0.1399 was defined. Mea-
3.4. Measurement of blood glucose level by the sensor 3.4.1. In vitro measurement A plasma sample was collected from several tirapias, and the sensor was placed in each sample to measure glucose. Fig. 7 shows the relationship between analytical signal and conven-
Fig. 8. Relationship for analytical signal of the sensor between whole blood and plasma samples from the same fish. Experimental conditions such as pH and temperature were the same as in Fig. 2.
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
Fig. 9. Response curves obtained by inserting the sensor into fish. “a” represents the point in time when the sensor was placed in the fish, while “b” represents the point of time when the sensor was removed. Sample (a1 and b1 ): 50 mg dl−1 for glucose level in fish; sample (a2 and b2 ): 73 mg dl−1 for glucose level in fish. Experimental conditions such as pH and temperature were the same as in Fig. 2.
123
a calibration curve. A good correlation was observed between values determined by the sensor and conventional methods in the range of 48–157 mg dl−1 (correlation coefficient: 09474, y = 0.8452x − 3.4018). The blood glucose level in tirapias is generally around 50 mg dl−1 [1], but greater individual differences were noted in the present study. This was probably due to stress. In the present study, individual differences were encountered in the amount of time required to capture fish using a net and collect blood samples. Fish were thus subjected to differing levels of stress [1,4]. Our proposed method requires significantly less time than conventional methods. Blood oxygen levels are known to fluctuate depending on fish condition, and since the present sensor was used to detect oxygen dissolved in buffer solution inside the needle-type hollow container, this relationship was obtained without being affected by blood oxygen. 4. Conclusions The present findings confirm that blood glucose levels in fish can be rapidly and conveniently measured using the sensor. Since the sensor was housed in a needle-type hollow container, blood glucose level could be measured without causing enzyme detachment from the sensor. Even when sensor response was unstable, subsequent measurements could be taken because the immobilized enzyme membrane in the hollow container could be easily exchanged. Furthermore, as the sensor, spectrometer and personal computer are small, this test system is easily transportable for use at fish farms. However, the reproducibility of measurements can be reduced if the sensor is inserted in different manners, and if the fish does not stay still, sensor response is unstable. Future improvements should address these problems and result in the development of a practical sensor system.
Fig. 10. Correlation between the values determined using the proposed sensor and the conventional methods. Experimental conditions such as pH and temperature were the same as in Fig. 2.
surement values for plasma determined by the sensor seemed basically the same as values for whole blood. Since conversion of measurement values between plasma and whole blood samples can be provided using the above formula, each plasma sample could be used as a standard sample for blood glucose calibration. In this study, various glucose levels of unused fish plasma samples stored at −80 ◦ C were employed as standard samples for calibration at each measurement. 3.4.2. In vivo measurement The sensor was inserted into the caudal vein of tirapias to measure blood glucose levels. Fig. 9 shows response curves for the sensor. The sensor responded immediately after insertion and level of oxygen decreased with time. In this manner, even when the sensor was directly inserted into fish, level of blood glucose could be measured. Fig. 10 shows the correlation between values determined using the proposed sensor and conventional methods. Blood glucose levels determined using the sensor were calculated using the result in Figs. 7 and 8 as
Acknowledgments This research was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology. We wish to thank Mr. Syoji Yanagisawa (Yanagisawa Factory, Tokyo, Japan) for making the needle-type hollow container. References [1] D.T. Nolan, R.L.J.M. Op’t Veld, P.H.M. Balm, S.E.W. Bonga, Aquaculture 177 (1999) 297. [2] C.E. Trenzado, T.R. Carrick, T.G. Pottinger, Gen. Comp. Endocrinol. 133 (2003) 332. [3] R.D. Van Anholt, F.A. Spanings, W.M. Koven, S.E. Bonga, Gen. Comp. Endocrinol. 139 (2004) 215–226. [4] M.J. Chowdhury, E.F. Pane, C.M. Wood, Comp. Biochem. Physiol. C: Toxicol. Pharmacol. 139 (2004) 163. [5] S. Jentoft, A.H. Aastveit, P.A. Torjesen, O. Andersen, Comp. Biochem. Physiol. A: Mol. Integr. Physiol. 141 (2005) 353. [6] M. Carballo, J.A. Jimenez, A. de la Torre, J. Roset, M.J. Munoz, Environ. Toxicol. 20 (2005) 119. [7] M. Maita, H. Aoki, Y. Yamagata, S. Satoh, N. Okamoto, T. Watanabe, Fish Pathol. 33 (1998) 59. [8] M. Maita, K. Satoh, Y. Fukuda, H. Lee, J.R. Winton, N. Okamoto, Fish Pathol. 33 (1998) 129.
124
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
[9] H. Endo, M. Maita, M. Takikawa, H. Ren, T. Hayashi, N. Urano, K. Mitsubayashi, Fish. Sci. 69 (2003) 1194. [10] M. Ishikawa, D.W. Schmidtke, P. Raskin, C.A.P. Quinn, J. Diabetes Complicat. 12 (1998) 295. [11] M. Quinto, I. Losito, F. Palmisano, C.G. Zambonin, Fresen. J. Anal. Chem. 367 (2000) 692. [12] M. Wu, Z. Lin, Y. Li, S. Ren, Sens. Actuators B 66 (2000) 269. [13] A. Hiratsuka, K. Kojima, H. Suzuki, H. Muguruma, K. Ikebukuro, I. Karube, Analyst 126 (2001) 658. [14] J. Wang, X. Zhang, Anal. Chem. 73 (2001) 844.
[15] R.A. Jeong, J.Y. Hwang, S. Joo, T.D. Chung, S. Park, S.K. Kang, W.Y. Lee, Kim.F H.C., Biosens. Bioelectron. 19 (2003) 313. [16] M. Gourzi, A. Rouane, R. Guelaz, M. Nadi, F. Jaspard, J. Med. Eng. Technol. 27 (2003) 276. [17] A. Caduff, E. Hirt, Y. Feldman, Z. Ali, L. Heinemann, Biosens. Bioelectron. 19 (2003) 209. [18] M. Gourzi, A. Rouane, R. Guelaz, M.S. Alavi, M.B. McHugh, M. Nadi, P. Roth, J. Med. Eng. Technol. 29 (2005) 22. [19] L. Li, D.R. Walt, Anal. Chem. 67 (1995) 3746. [20] Z. Rosenzweig, R. Kopelman, Anal. Chem. 68 (1996) 1408.
A needle-type optical enzyme sensor system for determining glucose levels in fish blood Hideaki Endo a,∗ , Yuki Yonemori a , Kazuya Musiya a , Masashi Maita b , Toru Shibuya c , Huifeng Ren a , Tetsuhito Hayashi a , Kohji Mitsubayashi d a Department of Ocean Sciences, Tokyo University of Marine Science and Technology, Konan 4-5-7, Minato, Tokyo 108-8477, Japan Graduate School of Marine Science and Technology, Tokyo University of Marine Science and Technology, Konan 4-5-7, Minato, Tokyo 108-8477, Japan c Toyo Gosei Co. Ltd., 2-1 Wakahagi-4 Inba-mura, Inba-gun, Chiba 270-1609, Japan d Department of Biomedical Devices and Instrumentation, Tokyo Medical and Dental University, 2-3-10 Kanda-Surugadai, Chiyoda, Tokyo 101-0062, Japan b
Received 29 November 2005; received in revised form 21 April 2006; accepted 24 April 2006 Available online 4 May 2006
Abstract A needle-type biosensor system was developed for rapid and simple determination of glucose levels in fish blood. The sensor comprises a needle-type hollow container (18-gauge needle), immobilized enzyme membrane and optic fiber probe with ruthenium complex. The enzyme membrane was prepared from glucose oxidase, azide-unit pendant water-soluble photopolymer and an ultra-thin dialysis membrane. The optic fiber probe was inserted into the rolled enzyme membrane placed in the needle-type hollow container. The calibration curve was linear for glucose levels in fish plasma. One assay was completed within 3 min. A good reproducibility was observed for 60 times without exchange of the enzyme membrane. The sensor was inserted into the caudal vein of fish to measure blood glucose levels. The sensor responded immediately after insertion and glucose levels could be monitored. Good correlations were observed between values determined using the sensor and conventional methods in the range of 48–157 mg dl−l (correlation coefficient, 0.9474). © 2006 Elsevier B.V. All rights reserved. Keywords: Biosensor; Enzyme sensor; Needle; Glucose; Blood; Plasma; Fish
1. Introduction In recent years, Hazard Analysis Critical Control Point (HACCP) systems have been actively adopted by the aquatic food industry. In the field of aquaculture involved in supplying fish, use of HACCP is under close examination for producing healthy fish. Mass outbreaks of pathogenic bacterial infections have killed large quantities of fish in recent years, causing great financial damage to fish farms. In addition, from the perspective of food hygiene for delivering safe fish to markets, farming fish with minimal dependence on antibiotics is important. Periodic checks of fish health and detection of abnormalities as quickly as possible are thus necessary at fish farms. Several studies have shown blood glucose levels closely correlate to the level of stress in fish and represent the state of respiratory or nutri-
∗
Corresponding author. Tel.: +81 3 5463 0616; fax: +81 3 5463 0616. E-mail address: [email protected] (H. Endo).
0003-2670/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2006.04.068
tional disturbance [1–6]. Recent studies have also clarified that decreased levels of blood cholesterol represent a useful indicator of reduced resistance to bacterial infection [7,8]. In this manner, determination of blood glucose and cholesterol is important for ascertaining the health of fish. At present, these tests are conduced using clinical laboratory test kits that have been designed for humans. Furthermore, as each sample needs to be analyzed separately, testing is time- and labor-intensive. When ascertaining the health of fish at fish farms, testing as many fish as possible in a fast and convenient manner is crucial. Recent research has been actively investigating the development of biosensors consisting of an immobilized enzyme membrane and an electrochemical device. We have already developed a biosensor to measure cholesterol levels in fish plasma, in an attempt to establish rapid and convenient methods to ascertain fish health [9]. Since this method is based on flow injection analysis (FIA), continuous measurement is possible and only 5 min is needed to analyze each sample, markedly reducing testing time. However, even this method requires a certain amount of skill
118
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
to draw blood, and since flow injection analysis is employed, the measurement system is complicated and large. Tests can be performed in a laboratory, but convenient testing at fish farms is difficult. Development of a needle-type enzyme sensor system is needed, so a biosensor can be directly inserted into fish to quantify blood components. In the field of clinical medicine, many needle-type enzyme sensors have been developed to measure blood glucose [10–15]. Wang and Zhang [14] developed a miniature needle-type enzyme sensor suitable for the simultaneous amperometric monitoring of glucose and insulin. The sensor was constructed from dual modified carbon-paste working electrodes inserted into a 14-gauge needle. Wu et al. [12] reported an enzyme sensor for whole blood lactate monitoring. The sensor consisted of a stainless steel needle with surface modified by a polymer material for enzyme immobilization. Studies of non-invasive glucose sensors have also been undertaken for application in the field of diabetes [16–18]. Caduff et al. [17] developed a glucose monitoring system based on impedance spectroscopy. Changes in the glucose concentrations can be monitored by varying the frequency in the radio band over a range, optimized to measure the impact of glucose on the impedance pattern. However, most sensor systems are designed for humans or livestock, and such sensors have rarely been used in fish. For application as fish, sensors need to be fairly strong, as fish might resist blood collection. In addition, since testing should be conducted at the fish farm if possible, portability is an important issue. Moreover, with many needle-type enzyme sensors, when a sensor is inserted into a specimen, immobilized enzyme becomes detached when coming into contact with the specimen surface. The method of immobilizing enzyme in a narrow needle-type detector is thus central to the design of a practical needle-type enzyme sensor. The application of fiber optics to the preparation of biosensor systems has been studied. Several fiber optic biosensors using fluorescence of ruthenium complexes for the determination of glucose have been developed [19,20]. The principle of the sensors is based on the monitoring of oxygen consumption due to enzymatic reaction of glucose oxidase. In the present study, to establish a rapid and convenient method for measuring blood components in fish, a needle-type enzyme sensor was developed in which an immobilized enzyme membrane and optic fiber probe using fluorescence of ruthenium complexes were inserted into a needle-type hollow container to measure blood glucose levels. 2. Materials and methods 2.1. Reagents Glucose oxidase (EC 1.1.3.4, Type VII from Aspergillus niger) was obtained from Sigma (MD, USA). Ultra-thin dialysis membrane (Part No. 157-0144-02; thickness; 15 m) was purchased from Technicon Chemical Co. (Orcq, Belgium). Photosensitive polymer (azide-unit pendant water-soluble photopolymer (AWP)) was obtained from Toyo Gosei Kogyo Co. (Tokyo, Japan). All other reagents used for experiments were commercial- or laboratory-grade. Glucose was dissolved in
0.1 M phosphate-buffered solution (PBS; pH 7.0), titrated and stored for ≥12 h to prepare standard glucose solutions. 2.2. Needle-type enzyme sensor and measurement Fig. 1 shows a schematic of the needle-type enzyme sensor and images of the detector region. The sensor consists of a needle-type hollow container, immobilized enzyme membrane and optic fiber probe with ruthenium complex. The sensor assembly is described below. 2.2.1. Needle-type hollow container The hole at the tip of a commercially available 18-gauge needle (diameter: 1.2 mm; Terumo, Tokyo, Japan) was blocked using epoxy resin (Toagosei Co., Tokyo, Japan). Next, 4 × 1.0 mm holes were made on one side of the needle, and 4 × 0.8 mm holes were made on the opposite side (Fig. 1A and E). 2.2.2. Immobilized enzyme membrane Glucose oxidase (1 mg) was dissolved in 200 l of 0.1 M PBS (pH 7.8), and 25 l of the resulting solution was mixed with 80 mg of AWP. Next, a dialysis membrane soaked in distilled water was evenly spread over a glass plate without bubbles and then completely dried. The above-mentioned enzyme and AWP mixture was evenly applied and rubbed into the dialysis membrane. After drying for 1 h in the dark, the membrane was placed under a fluorescent lamp for 1 h to prepare an immobilized enzyme membrane. AWP is a photosensitive polymer, and hardens with time to immobilize the enzyme. The resulting membrane was cut into 8 mm × 3 mm strips, soaked in distilled water again and then dried on filter paper. Due to differences in drying shrinkage between the dialysis membrane and AWP, each strip curled and formed a tube (Fig. 1A (3)). 2.2.3. Needle-type enzyme sensor A tube-shaped immobilized enzyme membrane was inserted into the tip of the needle-type hollow container (Fig. 1C). The container was soaked in 0.1 M PBS (pH 7.0) so that the immobilized enzyme membrane would swell and adhere to the needletype hollow container. The container was also lightly tapped to eliminate air bubbles. Next, the optic fiber probe including a ruthenium complex (FOXY-AL300: Ocean Optics, FL, USA) was inserted inside the immobilized enzyme membrane (Fig. 1D) to set up the needle-type glucose sensor. 2.2.4. Oxygen measurement system An excitation light-emitting diode (LED) light source (USBLS-450; Ocean Optics, FL, USA), UBS2000 spectrometer (Ocean Optics) and VT900 personal computer (NEC, Tokyo, Japan) are connected to the needle-type enzyme sensor to set up a glucose determination system. The optic fiber probe includes the ruthenium complex at the tip. Since the distal end of the probe tip comprises a thin layer of hydrophobic sol–gel material, ruthenium complex is trapped in the sol–gel matrix. When light (475 nm) from the LED excites the ruthenium complex at the probe tip, the excited ruthenium complex fluoresces, emitting energy at 600 nm. If the excited ruthenium complex encounters
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
119
Fig. 1. Schematic of the needle-type enzyme sensor system (A, B) and images of the detector region (C–E). (1) Needle-type hollow container, (2) round-shaped holes, (3) immobilized enzyme membrane, (4) optic fiber probe, (5) ruthenium complex, (6) excitation LED light source, (7) spectrometer, (8) personal computer.
an oxygen molecule, excess energy is transferred to the oxygen molecule in a non-radioactive transfer, decreasing or quenching the fluorescence signal. Energy is collected by the probe and carried through the optical fiber to the spectrometer. An
A/D converter is used to convert analog data to digital data, which are then recorded using computer software. The amount of dissolved oxygen in the needle-type hollow container can be measured based on changes in fluorescence intensity.
120
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
2.3. Measurement of standard glucose solutions using the sensor First, the sensor was soaked in 100 ml of 0.1 M phosphate buffer solution (pH 7.0) in a state of oxygen saturation to fill the needle-type hollow container with buffer solution. Once sensor output had stabilized, the sensor was placed in a test tube containing one of the standard glucose solutions (3.6–18.0 mg dl−1 ) to measure changes in concentration of dissolved oxygen inside the hollow container due to oxidization by glucose oxidase. The glucose oxidase-sensing scheme involves the employment of glucose oxidase according to: D-glucose + O2
glucose oxidase
−→
D-gluconolactone + H2 O2
To clean the hollow container, the sensor was again placed in 100 ml of buffer solution in a state of oxygen saturation, and the sensor was placed in another test tube once output had stabilized. The above procedures were repeated to measure levels of glucose in each standard glucose solution. 2.4. Measurement of blood glucose by the sensor Blood glucose levels were measured in tirapia (Oreochromis niloticus), as this African fish is farmed in various areas of Japan. Tirapias (n = 20; weight, 300–1000 g) were raised in an 800 l tank and used as needed. In the present study, blood glucose levels were measured by collecting blood samples or direct insertion of the sensor.
2.4.1. In vitro measurement Each tirapia (weight, 800–1000 g) was captured using a net and transferred to a bowl containing 200 mg l−1 of 2phenoxyethanol. Next, a syringe that had been cleaned using 3000 units ml−1 of heparin sodium solution was inserted into the caudal vein of the fish to collect 6–8 ml of whole blood, which was transferred to a test tube. Each blood sample was centrifuged at 3000 rpm for 10 min to prepare a plasma sample. As with the standard glucose solutions, the sensor was soaked in 0.1 M PBS, and once sensor output had stabilized, the sensor was soaked in 200 l of plasma or whole blood sample in the micro-test tube (volume, 400 l) to measure blood glucose level. The above procedures were repeated to measure glucose levels in each sample. 2.4.2. In vivo measurement After anesthetizing each fish using the same method, the sensor tip was inserted into the caudal vein. Blood glucose levels were measured by immobilizing each fish for about 3 min. Once the test was completed, the fish was returned to another tank, and the sensor was cleaned using buffer solution. The above procedures were repeated to measure glucose levels in each fish. 2.5. Conventional blood glucose determination To measure blood glucose levels using conventional method for references, a commercially available colorimetric assay kit using Trinder’s reaction (C-II glucose test; Wako Pure Chemi-
Fig. 2. Response curves for standard glucose solutions. (A) Response curve at various glucose concentrations, (B) response curve at a glucose concentration of 10.8 mg dl−1 . r, enlarged view of the initial response in B. Assay conditions were as follows: temperature 25 ◦ C; pH 7.0.
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
121
cal Industries, Osaka, Japan) was employed. When each plasma sample was mixed with reagent from the kit containing enzyme and color-fixing agent, glucose was oxidized and then yielded hydrogen peroxide via glucose oxidase. The hydrogen peroxide formed created red pigmentation by quantitative oxidation condensation with phenol and 4-aminoantipyrine. Glucose level was thus determined by measuring absorbance of the red color. 3. Results and discussion 3.1. Typical response curves of the sensor system Concentrations of each standard glucose solution were measured using the enzyme sensor, and Fig. 2A shows the results. When the sensor was transferred from buffer solution to glucose solution, sensor response (oxygen concentration) increased transiently from the baseline, then gradually decreased. This was attributed to dissolved oxygen in the buffer solution in the hollow container being oxidized by the glucose oxidase, and this reduction in dissolved oxygen was detected by the sensor. In addition, the transient increase in response was caused when the sensor touched the air during transfer to the buffer or solution. Sensor response decreased with increasing glucose concentration due to the fact that oxygen is consumed by the action of glucose oxidase (Fig. 2A). Fig. 2B shows the response curve at a glucose concentration of 10.8 mg dl−1 . In the present study, based on oxygen decrease 1 or 3 min after oxygen concentration returned to baseline, slope of the regression line was calculated using the least squares method to determine oxygen reduction per unit time. With the above-mentioned methods, 1 and 3 min sensor outputs were compared using standard glucose solutions (3.6–18.0 mg dl−1 ). Little difference was observed between the two values. As the objective of the present study was to measure blood glucose levels by directly inserting the sensor into fish, testing time should be as short as possible. In subsequent experiments, 1 min oxygen reduction was utilized as analytical signal (mg l−1 min−1 ) of the sensor. Fig. 3 shows the relationship between analytical signal and concentration in standard glucose solutions. At glucose concentrations of 1.8–19.3 mg dl−1 , a
Fig. 4. Effect of temperature on analytical signal. Experimental conditions were the same as in Fig. 2, except for temperature.
Fig. 5. Effect of pH on analytical signal. Experimental conditions were the same as in Fig. 2, except for pH of the buffer solution.
favorable linear relationship was noted (correlation coefficient: 0.9947, y = 0.3124x − 0.6368). The upper limit for standard glucose solutions was around 20.0 mg dl−1 . 3.2. Effects of measurement conditions on analytical signal
Fig. 3. Relationship between analytical signal and glucose concentration in standard glucose solutions.
As the sensor comprises a biological catalyst (enzyme), response can vary depending on temperature and pH of the solutions in which the immobilized enzyme membrane is placed. We therefore investigated the effects of these factors on analytical signal. Fig. 4 shows the effects of temperature on analytical signal. Output increased as temperature increased above 15 ◦ C, then began to decrease above 35 ◦ C. Since the body temperature of fish (tirapia) farmed in temperate areas is 20–30 ◦ C, matching the ambient water temperature, the present sensor is suitable in such fish. Fig. 5 shows the effects of pH on analytical signal. Glucose levels were measured by adjusting the pH of buffer solution in the needle-type hollow container from 5.6 to 8.2, and maximum response was obtained at pH 6.5. Since the bodily fluids of fish are mostly neutral, no problems are likely to occur using the sensor in fish.
122
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
Fig. 6. Reproducibility of sensor system measurements. Measurement was performed using a 0.5 mg dl−1 glucose standard solution. Experimental conditions were the same as in Fig. 2, except for glucose concentration. Scale of the vertical axis was made equal to the vertical axis of Fig. 3. Table 1 Changes in correlation coefficient of calibration curves Days
Correlation coefficient
1 3 5 8 11 16 21 33 41 50
0.9926 0.9963 0.9973 0.9897 0.9973 0.9985 0.9863 0.9886 0.9944 0.9909
Average
0.9932
Standard deviation
0.0040
Experimental conditions such as pH and temperature were the same as in Fig. 2.
3.3. Reproducibility and stability of the sensor system Fig. 6 shows reproducibility of the sensor system. Measurement was performed at 25 ◦ C using a 0.5 mg dl−1 glucose standard solution. As shown by the results, good reproducibility was observed for ≥60 times without exchange of the enzyme membrane (standard deviation: 0.047). We also investigated stability of the sensor to gain an appreciation of likely practical operational lifetime. After glucose determination, the sensor was stored at 5 ◦ C for up to 50 days, and each determination was performed at different time points. At each time point, a glucose calibration curve was drawn and a correlation coefficient for the linear relationship was calculated. Table 1 shows changes in correlation coefficient of calibration curves over 50 days. Correlation coefficient remained extremely high, indicating that the sensor can be used for extended periods of time.
Fig. 7. Relationship between analytical signal and glucose level in fish plasma. Experimental conditions such as pH and temperature were the same as in Fig. 2.
tional assay. A fine linear relationship existed at blood glucose levels of 37–105 mg dl−1 (correlation coefficient: 0.9741). When compared to standard glucose solutions (Fig. 3), marked changes were noted in the upper limit of glucose determination. This might be attributable to enzyme inhibitors in blood and delays in substrate diffusion inside the hollow container due to blood viscosity. Since different calibrations were observed between plasma samples and standard glucose solution, a standard sample for glucose calibration to measure blood glucose levels should be prepared. Moreover, to measure blood glucose levels by inserting the needle-type sensor into fish, glucose levels need to be quantified in whole blood, not plasma. A whole blood sample was thus collected from the same tirapias, and the sensor was placed in each whole blood sample to measure glucose. Fig. 8 shows the relationship for analytical signal of the sensor between whole blood and plasma samples in each fish. A good correlation was observed (correlation coefficient: 0.9745) and a regression line of y = 1.0667x + 0.1399 was defined. Mea-
3.4. Measurement of blood glucose level by the sensor 3.4.1. In vitro measurement A plasma sample was collected from several tirapias, and the sensor was placed in each sample to measure glucose. Fig. 7 shows the relationship between analytical signal and conven-
Fig. 8. Relationship for analytical signal of the sensor between whole blood and plasma samples from the same fish. Experimental conditions such as pH and temperature were the same as in Fig. 2.
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
Fig. 9. Response curves obtained by inserting the sensor into fish. “a” represents the point in time when the sensor was placed in the fish, while “b” represents the point of time when the sensor was removed. Sample (a1 and b1 ): 50 mg dl−1 for glucose level in fish; sample (a2 and b2 ): 73 mg dl−1 for glucose level in fish. Experimental conditions such as pH and temperature were the same as in Fig. 2.
123
a calibration curve. A good correlation was observed between values determined by the sensor and conventional methods in the range of 48–157 mg dl−1 (correlation coefficient: 09474, y = 0.8452x − 3.4018). The blood glucose level in tirapias is generally around 50 mg dl−1 [1], but greater individual differences were noted in the present study. This was probably due to stress. In the present study, individual differences were encountered in the amount of time required to capture fish using a net and collect blood samples. Fish were thus subjected to differing levels of stress [1,4]. Our proposed method requires significantly less time than conventional methods. Blood oxygen levels are known to fluctuate depending on fish condition, and since the present sensor was used to detect oxygen dissolved in buffer solution inside the needle-type hollow container, this relationship was obtained without being affected by blood oxygen. 4. Conclusions The present findings confirm that blood glucose levels in fish can be rapidly and conveniently measured using the sensor. Since the sensor was housed in a needle-type hollow container, blood glucose level could be measured without causing enzyme detachment from the sensor. Even when sensor response was unstable, subsequent measurements could be taken because the immobilized enzyme membrane in the hollow container could be easily exchanged. Furthermore, as the sensor, spectrometer and personal computer are small, this test system is easily transportable for use at fish farms. However, the reproducibility of measurements can be reduced if the sensor is inserted in different manners, and if the fish does not stay still, sensor response is unstable. Future improvements should address these problems and result in the development of a practical sensor system.
Fig. 10. Correlation between the values determined using the proposed sensor and the conventional methods. Experimental conditions such as pH and temperature were the same as in Fig. 2.
surement values for plasma determined by the sensor seemed basically the same as values for whole blood. Since conversion of measurement values between plasma and whole blood samples can be provided using the above formula, each plasma sample could be used as a standard sample for blood glucose calibration. In this study, various glucose levels of unused fish plasma samples stored at −80 ◦ C were employed as standard samples for calibration at each measurement. 3.4.2. In vivo measurement The sensor was inserted into the caudal vein of tirapias to measure blood glucose levels. Fig. 9 shows response curves for the sensor. The sensor responded immediately after insertion and level of oxygen decreased with time. In this manner, even when the sensor was directly inserted into fish, level of blood glucose could be measured. Fig. 10 shows the correlation between values determined using the proposed sensor and conventional methods. Blood glucose levels determined using the sensor were calculated using the result in Figs. 7 and 8 as
Acknowledgments This research was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology. We wish to thank Mr. Syoji Yanagisawa (Yanagisawa Factory, Tokyo, Japan) for making the needle-type hollow container. References [1] D.T. Nolan, R.L.J.M. Op’t Veld, P.H.M. Balm, S.E.W. Bonga, Aquaculture 177 (1999) 297. [2] C.E. Trenzado, T.R. Carrick, T.G. Pottinger, Gen. Comp. Endocrinol. 133 (2003) 332. [3] R.D. Van Anholt, F.A. Spanings, W.M. Koven, S.E. Bonga, Gen. Comp. Endocrinol. 139 (2004) 215–226. [4] M.J. Chowdhury, E.F. Pane, C.M. Wood, Comp. Biochem. Physiol. C: Toxicol. Pharmacol. 139 (2004) 163. [5] S. Jentoft, A.H. Aastveit, P.A. Torjesen, O. Andersen, Comp. Biochem. Physiol. A: Mol. Integr. Physiol. 141 (2005) 353. [6] M. Carballo, J.A. Jimenez, A. de la Torre, J. Roset, M.J. Munoz, Environ. Toxicol. 20 (2005) 119. [7] M. Maita, H. Aoki, Y. Yamagata, S. Satoh, N. Okamoto, T. Watanabe, Fish Pathol. 33 (1998) 59. [8] M. Maita, K. Satoh, Y. Fukuda, H. Lee, J.R. Winton, N. Okamoto, Fish Pathol. 33 (1998) 129.
124
H. Endo et al. / Analytica Chimica Acta 573–574 (2006) 117–124
[9] H. Endo, M. Maita, M. Takikawa, H. Ren, T. Hayashi, N. Urano, K. Mitsubayashi, Fish. Sci. 69 (2003) 1194. [10] M. Ishikawa, D.W. Schmidtke, P. Raskin, C.A.P. Quinn, J. Diabetes Complicat. 12 (1998) 295. [11] M. Quinto, I. Losito, F. Palmisano, C.G. Zambonin, Fresen. J. Anal. Chem. 367 (2000) 692. [12] M. Wu, Z. Lin, Y. Li, S. Ren, Sens. Actuators B 66 (2000) 269. [13] A. Hiratsuka, K. Kojima, H. Suzuki, H. Muguruma, K. Ikebukuro, I. Karube, Analyst 126 (2001) 658. [14] J. Wang, X. Zhang, Anal. Chem. 73 (2001) 844.
[15] R.A. Jeong, J.Y. Hwang, S. Joo, T.D. Chung, S. Park, S.K. Kang, W.Y. Lee, Kim.F H.C., Biosens. Bioelectron. 19 (2003) 313. [16] M. Gourzi, A. Rouane, R. Guelaz, M. Nadi, F. Jaspard, J. Med. Eng. Technol. 27 (2003) 276. [17] A. Caduff, E. Hirt, Y. Feldman, Z. Ali, L. Heinemann, Biosens. Bioelectron. 19 (2003) 209. [18] M. Gourzi, A. Rouane, R. Guelaz, M.S. Alavi, M.B. McHugh, M. Nadi, P. Roth, J. Med. Eng. Technol. 29 (2005) 22. [19] L. Li, D.R. Walt, Anal. Chem. 67 (1995) 3746. [20] Z. Rosenzweig, R. Kopelman, Anal. Chem. 68 (1996) 1408.