J . Chem. Tech. Biotechnol. 1990, 49, 255-265

Amperometric Bi-enzyme Based Biosensor for the Detection of Lactose-Characterization and Application Dorothea Pfeiffer," Edmundas V. Ralis,b Alexander Makower" & Frieder W. Scheller" "Academy of Sciences of the GDR, Central Institute of Molecular Biology, 1115 Berlin-Buch, Robert-Rossle-Stral3e 10, GDR *Laboratory 'Ferment', 232028 Vilnius, Fermentu 8, USSR (Received 8 November 1989; accepted 18 January 1990)

ABSTRACT Based on the glucose oxidase-8-galactosidase sequence an enzyme probe for the specific determination of lactose has been developed. 8-Galactosidases from different sources have been compared, the sensor containing 8galactosidase from Curvularia inaequalis has been characterized in respect of optimal p H , enzyme loading, apparent activity and functional stability. The response of the bi-enzyme probe depends linearly on lactose concentration between 0.02 and 3.00 mmol dm- '. The application to different milk and foodstuff samples resulted in good correlations toward enzymatic - 1.67) mmol dm- ') and infrared detection photometric (y = (0.956~ (y = (1.0772~-0.3909)%). Using a measuring frequency of 100 h-' the serial imprecision is about 2 % for diluted milk, urine, or foodstuff samples. Key words: bi-enzyme probe, 8-galactosidase, glucose oxidase, lactose analysis, milk, foodstuff, urine.

1 INTRODUCTION

In addition its well-known importance for the quality of milk and dairy products, the disaccharide lactose is a basic parameter in waste water control and veterinary medicine. The methods of lactose determination presently used are characterized by long sample preparation times, including filtration and dilution and by high systematic error and insufficient specificity of the non-enzymatic methods. Thus, it is important to have a method that is fast, simple, and specific for lactose. Both galactose oxidase (EC 1.1.3.9)' and the 8-galactosidase (EC 3.2.1.23kglucose 255 J . Chem. Tech. Biotechnol.0268-2575/90/%03.5001990 Society of Chemical Industry. Printed in Great Britain

256

D. PJ$er,

E . V. Ralis, A . Makower, F . W. Scheller

glucose

02

lactose

Fig. 1. Scheme of the enzyme sequence realized in the lactose sensor. GOD, Glucose oxidase; j-GAL, 8galactosidase.

oxidase (EC 1.1.3.4) coupled system have been used for enzymatic monitoring combined with photometrical, thermometrical or electrochemical detection. The glucose oxidase-p-galactosidase system connected with a thermistor has been used for calorimetric measurements of lactose in whey.2 An increased sensitivity was achieved by co-immobilization of c a t a l a ~ e .The ~ enzyme sequence of glucose oxidase and a-galactosidase has been used for the determination of lactose in milk and powdered milk by monitoring oxygen consumption, which demanded a timeconsuming sample re treatment.^ The same bi-enzyme system has been applied to the registration of the steady-state current of the hydrogen peroxide produced in the glucose oxidation, resulting in a measuring frequency of about 10 h-’.’ This paper describes the development of a fast amperometric lactose membrane sensor having high accuracy based on the combination of p-galactosidase and glucose oxidase. The enzyme sequence is schematically represented in Fig. 1. The enzymes catalyse the following reactions: lactose + H 2 0 fl-D-glucose+ O2

p-galaclosidase

D-galactose +p-D-glucose

glucose oxidase

D-glucono-h-lactone

+H , 0 2

(1)

(2)

The reduction of oxygen consumed in reaction (2) or the oxidation of hydrogen peroxide is indicated using a Pt-Ag/AgCl oxygen electrode. P-Galactosidases of different sources are compared, the bi-enzyme membrane is biochemically characterized and the application of the optimized sensor to milk, dairy products and urine is discussed. 2 EXPERIMENTAL 2.1 Materials

Glucose oxidase from Penicillium notaturn (VEB Arzneimittelwerk Dresden, GDR) (46 U mg- ’) immobilized in polyurethane (commercial product of the Central Institute of Molecular Biology, Academy of Sciences, GDR)6 was used as the base

Bi-enzyme biosensor for lactose detection

257

of bi-enzyme membranes. Four different fl-galactosidases were used: two from Escherichia coli (Sigma, USA, 20 U mg-’; Boehringer Mannheim, FRG, 30 U mg- I ) , one from Bifidobacterium adolescentes (isolated noncommercially at the Central Institute of Nutrition, Potsdam-Rehbrucke, GDR, 32 U ~ m - ~ ) and ,’ one from Curuularia inaequalis (Company Ferment, Vilnius, USSR, 0.8 U mg-I). All of the activity data are related to the substrate lactose at 25°C. Raw and stabilized milk and whey samples from VEB Milchhof Prenzlau (GDR) and Molkereigenossenschaft Pasewalk (GDR)were used. Dairy products used were bought from local supermarkets; bovine urine samples were obtained from animals at VEG Birkholz (GDR) and VEG Heinersdorf (GDR). The polyurethane SYSPUR was obtained from VEB Synthesewerk Schwarzheide, Schwarzheide (GDR), photogelatin was purchased from Gelatinewerk Calbe, Calbe (GDR). Hydrophilic cellulose dialysis membrane (thickness 20 pm) was obtained from VEB Filmfabrik Wolfen, Wolfen (GDR). The hydrophobic polyethylene membrane used has a thickness of 15 pm (VEB Metra Radebeul, Radebeul, GDR). All other chemicals used were of analytical grade.

2.2 Preparation of the enzyme membrane 2.2.1 Immobilization in gelatin fl-Galactosidase membranes were prepared by mixing of 50pl 5 % (w/v) gelatin solution with the following amounts of the respective fl-galactosidases:

fl-galactosidase from Escherichia coli, 10.0 U; fl-galactosidase from Bifidobacterium adolescentes, 4-0 U; fl-galactosidase from Curuularia inaequalis, 10.0 U. The mixtures were spread on an area of 1 cmz of an even polyethylene support as described in detail previously.’ Additionally, cross-linking using polyvinylisocyanate (40 mm3 of a 5% (w/v) solution cm-’) was used.6 The bi-enzyme sandwich membrane was obtained by combining the commercial glucose oxidase layer with a j?-galactosidase layer between two cellulose membranes.

2.3 Preparation of samples Raw milk, stabilized milk and whey samples were diluted with buffer in the ratios 1 :50, 1 :lo0 and 1 :200 without any other pretreatment. Skimmed milk powder and baby food were analysed after dissolution in buffer, cottage cheese was suspended in water and analysed without extraction or filtering. Bovine urine samples, taken by catheterizing the urinary bladder, were diluted 1:50 with buffer solution before analysis. 2.4 Apparatus and procedure Modified Clark-type electrodes (0.5 mm platinum) (VEB Metra Radebeul, Radebeul, GDR) were covered with the bi-enzyme sandwich membrane and brought into a measuring cell. The indicator electrode was polarized at + 600 mV versus Ag/AgCl for monitoring the hydrogen peroxide produced during the glucose oxidase catalysed reaction (eqn (2)).

D. PfkiJier, E . V. Ralis, A . Makower, F .

3 ELECTRONICS

ILJ

W.Scheller

II

Fig. 2. Schematic diagram of the flow system for lactose determination. FTC, Flow-through cell containing the enzyme sensor; P, pump; TS, thermosensor; S, sample, SD, sample dispenser; V, valve; SU, burner supply; W, waste; PR, printer.

In the case of discrete measurements the manual analyser Glukometer GKM or the polarograph GWP 673 (Center of Scientific Equipment, Academy of Sciences, GDR) were used with a stirred measuring cell (batch system). Both the current-time behaviour of the enzyme sensor and the derivative current signal after addition of lactose containing samples to 2 cm3 stirred buffer solution were measured. Segmented continuous flow analysis with a derivative current signal is used in the Enzyme Chemical Analyser ECA 20 (VEB Prufgerate-Werk Medingen, GDR) with a flow-through cell containing the bi-enzyme membrane, schematically represented in Fig. 2. Prediluted lactose containing samples were pumped through the cell. In the case of lactose analysis in urine the applied potential of the indicator electrode was - 600 mV, indicating oxygen reduction. The Clark-type electrode was covered first with a hydrophobic polyethylene membrane carrying the bienzyme sandwich membrane. All samples used were air-saturated. The Milkoscan 104 (AN VOSS Electric, Denmark) was used for the infrared measurements and the Specol 11 (VEB Carl-Zeiss-Jena, Jena, GDR) for the enzymatic photometric analysis by the p-galactosidase-glucose oxidase-peroxidase method.

3 RESULTS AND DISCUSSION 3.1 Membrane composition and enzyme loading

Using glucose oxidase membranes with a constant activity of 46 U cm-’ the pgalactosidase activity was varied between 0.2 and 20.0 U cm-’ in a gelatin matrix. The lactose signal increased almost linearly between 0.2 and 4.0 U p-galactosidase cm-’. Above 6.0 U cm-’ no significant increase of the differentiated current signal was found at pH 6.0 in the presence of 0.75 mmol dm-3 lactose (see Fig. 3). Thus, the transition to a diffusioncontrolled overall reaction can be assumed to occur at this enzyme loading. With regard to a sufficient functional stability, all further experiments were carried out with enzyme membranes containing 10 U ern-'. The apparent activity of the B-galactosidase membrane was determined as follows. Both a glucose oxidase electrode polarized at +600 mV and a p-

Bi-enzyme biosensor for lactose detection

259

Enzyme loading (U crri')

Fig. 3. Sensitivity of the amperometric bi-enzyme lactose sensor as a function of the ~-galactosidase loading at constant glucose oxidase activity. Activity of glucose oxidase, 46 U cm-'; concentration of lactose: 1 mmol dm-'; phosphate buffer, 0.066 mol dm-'; pH 6.0.

galactosidase membrane with 10 U cm-* fixed in a holder were dipped into the same stirred saturated (5 mmol d m - 7 lactose solution. From the timedependent current increase ofthe glucose oxidase electrode an apparent activity of 1 U cm-2 of the P-galactosidase membrane was calculated, indicating that only 10% of the activity was utilized. On the other hand, for an unmodified gelatin layer containing the enzyme, an apparent activity of 90 % was found. Obviously, the substrate influx, and alternatively the enzyme activity, are decreased significantly by cross-linking of the enzyme layer using polyvinylisocyanate.

3.2 pH dependence The effect of pH on the response of the bi-enzyme sensor was studied from pH 4.0 to 8.0 using 0.66 mmol dm-3 phosphate and 0.1 mmol dm-3 citric acid-phosphate buffer. In solution P-galactosidase from E. coli shows a pH optimum at 7.5 (Ref. 9) whereas the bi-enzyme system has an optimal pH between 8.0 and 8.5.'' This optimum was shifted towards the acidic region for gelatinentrapped enzymes. As shown in Fig. 4 the optimal pH was 6.0 in both buffer systems. These results agree ' with observations by some other

3.3 Functional stability The functional stability of the lactose sensor prepared from P-galactosidases from three different sources was characterized by the measurement to 2.0 mmol dm-3 lactose solution in 0.66 mmol dm-3 phosphate buffer solution of pH 6.0. Between analyses, the sensor was stored at 25°C. As presented in Fig. 5, the initial sensitivity decreased to 50% after three days or 150measurements,when using P-galactosidase from Escherichia coli from Boehringer (Mannheim, FRG). The application of the

260

D . Pjeiffer, E . V. Ralis, A . Makower, F . W. Scheller

0

Fig. 4. Influence of pH on the sensitivity of the /?-galactosidaselglucoseoxidase sensor. A, Citric acid/ phosphate buNer, 01 mol (McIlvain); B, phosphate buffer, 0.66 mol drn-’ (Soerensen); concentration of lactose, 2 rnmol drn-3.

Time (days)

Fig. 5. Functional stability of the lactose sensor using P-galactosidases of different sources. A, Bifdohacteriurn adolescentes, 4 U cm-’; B, Escherichia coli (Sigma), 10 U cm-’; C, Curuularia inaequalis, 10 U cm-*; D, Escherichia coli (Boehringer), 15 U crn-’; concentration of lactose, 2 mmol drn-’; phosphate buffer, @66 mol dm-’, pH 6.0.

cnzyme from Bifidobacterium adolescentes or Escherichia coli (Sigma, USA) resulted in a more stable sensor; 50% of the initial sensitivity remained after 10and 15 days, respectively. The most stable enzyme was found to be that from Curuularia inaequalis. After an increase of sensitivity during the first five days corresponding to swelling of the gelatin, this sensor was stable in response over more than 15 days. Since half of the initial sensitivity was retained after 50 days, or more than 3000 measurements, this enzyme layer was used for further investigations and applied to milk, foodstuff and urine analysis.

26 1

Bi-enzyme biosensor for lactose detection

/ /*

Phosphate buffer.0.06 rnol dm-3

pH 6.0 t ~ 2 5 . C activity P-galactosidase ; 1 0 Ucm-* activity glucose o x i d a u : 4 6 U ern-'

Y

0

1 -

2.5

5.0

Concentration lactose (rnrnol/drn-3)

Fig. 6. Calibration graph for aqueous lactose standard solutions.

3.4 Electrode response using the batch and the flow analysis systems It is well known that Mg2+ ions are activators of 8-galactosidase. Addition of 1 mmol dm-3 Mg2+ to the background solution increased the sensitivity of the bienzyme sensor by 100% in the case of 8-galactosidase from Escherichia coli. Using the enzyme from Curuularia inaequalis, addition of Mg2 caused no significant increase in sensitivity. Thus, phosphate buffer only was used for this sensor. The total thickness of the bi-enzyme membranes used in lactose measurements is about 100 pm, whereas for mono-enzyme membranes typically 60-pm thick layers are used. This higher thickness is probably the reason for the lower sensitivity of this diffusioncontrolled bi-enzyme sensor' as compared with mono-enzyme electrodes. In the case of both batch and flow systems the detection limit was about 2 . 0 ~lo-' mol dm-3 and the linear concentration range extended over two decades, as shown by the calibration graph in Fig. 6. For the determination of glucose,13 lactate,14 lysine," etc., detection limits of 1-5 x mol dm-3 are achieved with the respective mono-enzyme electrodes. The application of the 100-pm thick membrane in the batch system resulted in a response time of 8 s and a measuring frequency of about 60 h- '. Using the flowthrough system of the Enzyme Chemical Analyser ECA 20, up to 120 samples h - ' were possible in the linear range up to 3 mmol dm-3. It is shown in Fig. 7 that the increase of the measuring frequency from 40 h- to 120 h- resulted in a decrease of the response of only 20%. Independent of the measuring system and frequency used, the serial precision for 20 successive measurements of 0.1 mmol dm-3 and 0.3 mmol dm-3 lactose calibration solutions was below 1.5 %. +

'

3.5 Application of the lactose sensor to milk and foodstuff analysis In a series of 20 diluted milk samples the coefficient of variation was below 2 %, not significantly different from those of pure lactose solution (,< 1.5 %). Forty successive

4

262

D. Pfeiffer. E . V. Ralis, A . Makower, F . W . Scheller

+ I

a

1,

'

1

40

I

120

80

Measuring frequency (sample h-')

Fig. 7. Dependence of the relative sensitivity of the bienzyme lactose sensor on the measuring frequency. Concentration of lactose, 1 mmol dxK3; phosphate buffer, 066 mol dm-3, pH 6.0.

yr1.0772 x -0,3909 (g 100cm?) n = 30 r-0.9701

.,A

/ /* 3.5

4.0

4.5

50

Concentration lactose/ infrared analysis (g i o o ~ m - ~ )

Fig. 8. Correlation between bi-enzymeelectrode and infrared analysis applying raw and pasteurized milk samples.

measurements of diluted milk samples gave a signal decrease of only 0.5 %. Thus, the broad variety of components of this biological solution did not markedly affect the performance of the bound enzymes. The application of 20 different raw and pasteurized milk samples to a pure glucose oxidase probe using a batch system gave no measurable current signals. Hence, glucose interference is not a factor that needs to be taken into consideration for milk analysis. The suitability of the bi-enzyme lactose sensor was established by comparing results obtained with it for pasteurized whole and raw milk with those obtained by infrared spectroscopy. Figure 8 shows the good correlation between the two met hods.

263

Bi-enzyme biosemor for lactose detection

TABLE 1 Comparison of Lactose Analysis Using the Bi-enzyme Lactose Sensor and the Enzymatic Photometric Detection Sample

Concentration of lactose (mmol dm- 3,

fl- Galactosidase + glucose oxidaselperoxidase (photometric detection)

Milk I Milk I1 Powdered milk ‘Babysan’ I ‘Babysan’ I1 ‘Milasan’ ‘Manasan’ Cottage cheese I Cottage cheese I1

Bi-enzyme lactose sensor

125.70 61.40

123.30 61.96

105.90 93.60 96.60 128.60 105.30 87.70

105.50 93.40 86.44 13040 94.50 83.00

X = 100.60 j = 97.26 y=(1.0021x- 3.54) mmol d m - 3 r = 0.9761

The developed sensor was compared with enzymatic photometric detection for a wide variety of foodstuffs. The glucose produced during the fl-galactosidase catalysed reaction (soluble enzymes) was detected using the glucose oxidaseperoxidase indicating system. Table 1 represents the lactose contents obtained by both methods. The somewhat lower concentrations found by the biosensor should be based on the more specific principle compared with the photometric analysis.

3.6 Analysis of urine Seven to nine weeks before parturition, lactation in the milk gland is interrupted by gradual extension of the milking interval. Thus, microorganisms may have more time in which to colonize the udder. Constituents of milk are degradated or secreted by blood or urine. Thus, lactose can be used as an indicator of resorption processes within the udder.16 As seen in Fig. 9, the normal urine lactose value of udderhealthy animals within peripartale time is below 3.00 mmol dmW3.In the case of udder disease, the abnormal high lactose value of more than 15 mmol dm-3 is caused by an increased permeability. These concentrations are in accordance with those described elsewhere.”

4 CONCLUSIONS The comparison between the described bi-enzyme lactose sensor and conventional methods resulted in excellent correlations for milk and foodstuff analysis. The highly stable, precise and fast sensor offers a convenient alternative to available

264

D. P’iffer,

E . V. Ralis, A . Makower, F . W.Scheller

__

~

006 mol d ~ n ’phosphate ~ buffer

pH 6.0

tr25.C

I

0

~-

5

10

Time (days)

Fig. 9. Timecourse of lactose concentration in urine of cows. A, Animal without mastitis; B, increasing mastitis during gradual decrease of lactation; C, subclinical state of mastitis.

lactose analysis methods. Determination of urine lactose concentrations seems to be a helpful method for mastitis diagnosis in cows.

ACKNOWLEDGEMENTS The authors wish to thank Dr J. Schultz and Dr G. Zunft (Central Institute of Nutrition, Academy of Sciences, Potsdam, GDR) for placing the /3-galactosidase from Bijdobacterium adolescentes at their disposal and Miss H. Beyer (VEB Uckermarkischer Milchhof, Prenzlau, GDR) and Mr J. Raatz (Humboldt University, Berlin, Germany) for providing milk and urine samples, respectively.

REFERENCES 1. Yellow Springs Instruments, Ohio, Specification Sheet Lactose Analysis Model 27, 1988.

2. Danielsson, B., Mattiasson, B., Karlsson, R. & Winquist, F., Use of an enzyme thermistor in continuous measurements and enzyme reactor control. Biotechnol. Bioeng., 21 (1979) 174946.

Ei-enzyme biosensor for lactose detection

265

3. Mattiasson, B. & Danielsson, B., Calorimetric analysis of sugars and sugar derivatives with aid of an enzyme thermistor. Carbohydrate Res., 102 (1982) 273-82. 4. Cheng, F. S. & Christian, G. D., Rapid enzymatic determination of lactose in food products using amperometric measurement of the rate of depletion of oxygen. Analyst, 102 (1977) 124-31. 5. Pilloton, R., Mascini, M., Casella, I . G., Festa, M. R. & Bottari, E., Lactose determination in raw milk with a two-enzyme based electrochemical sensor. Analyt. Lett., 20(11) (1987) 1803-14. 6. Nentwig, J., Scheller,F., Weise, H., Heinrich, G., Kirstein, D., Becker, D., Herrmann, P. & Pfeiffer, D., DD Patent 277 888 4, 1985. 7. Schulze, J., Herzog, R. & Zunft, H.-J., D D Patent 255 8012, 1985. 8. Scheller, F., Pfeiffer, D., Janchen, M., Seyer, I., Siepe, M. & Pittelkow, R., DD Patent 150500, 1979. 9. Manjon, A,, Klorca, F. I., Bonete, M. J., Bastida, J. & Iborra, J. L., Properties of pgalactosidase covalently immobilized to glycophasecoated porous glass. Process Biochem., (1985) 17-22. 10. Matsumoto, K . , Hamada, O., Ukeda, H. & Osajima, Y., .Flow injection analysis of lactose in milk using a chemically modified lactose electrode. Agric. Biol. Chem., 49(7) (1985) 2131-5. 1 1 . Batsalova, K., Kunchev, K., Popova, Y., Kozhukharova, A. & Kirova, N., Hydrolysis of lactose by p-galactosidase immobilized in polyvinyl alcohol. Appl. Microbiol. Biotechnol., 26 (1987) 227-30. 12. Carr, P. W. & Bowers, L. D. (eds), In Immobilized enzymes in analytical and clinical chemistry. John Wiley & Sons, New York, 1980, pp. 233-5. 13. Scheller, F., Pfeiffer, D., Kiihn, M., Hundertmark, J., Quade, A., Janchen, M., Lange, G., Holesch, H. & Dittmer, H., Glucosemessung in verdiinntem Vollblut mit einer Enzymelektrode. Acta Biol. Med. Germ., 39 (1980) 671-9. 14. Schubert, F., Pfeiffer, D., Wollenberger, U., Kiihnel, S., Hanke, G., Hauptmann, B. & Scheller, F., Enzyme-Chemical Analyzer ECA 2O/ESAT 6600. In Biosensors: Applications in Medicine, Environmental Monitoring and Process Control, ed. R. D. Schmidt & F. W. Scheller. Verlag Chemie, Weinheim, 1989, pp. 11-15. 15. Pfeiffer, D., Risinger, L., Wollenberger, U., Scheller, F. & Johansson, G., Amperometric amino acid electrodes. In Biosensors: Applications in Medicine, Environmental Monitoring and Process Control, ed. R. D. Schmid & F. W. Scheller. Verlag Chemie, Weinheim, 1989, pp. 27-31. 16. Wendt, D., Mielke, H. & Fuchs, H.-W., In Euterkrankheiten. VEB Gustav-FischerVerlag, Jena, 1986, pp. 127-9. 17. Sissoko, S., Der Laktosegehalt in Harn und Milch des Rindes bei subklinischer BStreptokokkenmastitis. PhD Thesis, Hannover, 1985.