Jounrol 0 1991 ADONIS
o/Biotechnology, 18 (1991) Elsevier Science Publishers 016816569100080M
BIOTEC
161-112 B.V. 0168-1656/91/$03.50
161
00597
On-line determination of glucose concentration throughout animal cell cultures based on chemiluminescent detection of hydrogen peroxide coupled with flow-injection analysis Y .L. Huang, Department
S.Y. Li, B.A.A. oj Enzyme
Dremel,
Technology,
Gesellschajt Braunschweig,
(Received
17 May
1990;
revision
U. Bilitewski fir
and R.D. Schmid
Biotechnologische
Forschung
GntbH,
F. R. G.
accepted
25 September
1990)
A flow-injection analysis (FIA) system for the on-line determination of glucose in animal cell cultures is described. The system is based on immobilized glucose oxidase (GOD). The hydrogen peroxide generated in the enzyme reaction is determined via a highly sensitive chemiluminescent reaction with luminol. Based on the measurement of the maximum emitted light intensity, the system was able to analyse hydrogen peroxide over the concentration range of lo-’ to 10-l M. For glucose determination, the system has a linear range of 10m5 to 5 x lo-’ M glucose, with an r.s.d. of 3% at the 1 mM level (5 measurements). The influence of luminol and buffer concentrations, pH and temperature on the chemiluminescent reaction were investigated. The enzyme reactor used was stable for more than 4 weeks in continuous operation, and it was possible to analyse up to 20 samples per h. The system has been successfully applied to on-line monitoring of glucose concentration during an animal cell culture, designed for the production of human antithrombin III factor. Results obtained with the FIA system were compared with off-line results, obtained with a Yellow Springs Instrument Company Model 27 (YSI). FIA; Chemiluminescence culture
Correspondence
Forschung
GmbH.
of luminol; On-line determination;
IO: Y.L. Huang, Dept. D-3300 Braunschweig,
of Enzyme F.R.G.
Technology,
Gesellschaft
Glucose; Animal cell
fir
Biotechnologische
162
Introduction Numerous animal cell cultures are carried out at industrial and laboratory scale for the production of biomass and products for medical application, such as interleukin II (IL-II; Wirth et al., 1987) human antithrombin III (AT-III; Wirth et al., 1989) and interferon (Reiser and Hauser, 1987). Glucose, as the main carbon source, is of essential importance for animal cell culture. Moreover, glucose consumption is also an indicator of cellular metabolic activity. Undesirable consequences of excess glucose concentration, such as catabolite repression (Demain, 1972) might be avoided by controlled feeding. The cultivation of baby hamster kidney cells (BHK) for the production of human antithrombin III protein (AT-III), for example, requires careful monitoring of the glucose level, as glucose is a limiting factor for both cell growth and protein production (Wirth et al., 1989). To date, animal cell reactors are equipped with some essential sensors for on-line monitoring of physical parameters e.g. pH, PO,, redox potential and temperature. The analysis of cultivation substrates such as glucose or lactate is usually achieved by off-line methods (Damen and Pettitt, 1980; Hesse et al., 1984; YSI, 1988). However these methods, which involve manual intervention in sampling or analysis have usually limited applicability since response time is typically too slow to be compatible with feedback control requirements and frequent sampling poses a risk of contamination. Hence, there is a requirement for efficient and reliable sensors for on-line monitoring of physiological parameters such as glucose or lactate, common substrates of animal cell culture (Sigma Catalog, 1988). Except for the determination of glucose based on its reducing properties, two methods have been developed to enhance the accuracy of analysis. First, glucose concentration can be quantified by liquid chromatography (Brinkmann, 1987) however these techniques require expensive instrumentation and often sample derivatization, and second, enzymatic assays for glucose determination have been developed, using either glucose oxidase (GOD), glucose dehydrogenase (GDH) (Persson et al., 1985; D’Costa et al., 1986) or glucose-6-phosphate dehydrogenase (G-6-PDH; Kunst et al., 1984). Various methods based on enzymes employ the technique of FIA coupled with the use of amperometric oxygen (Ruzicka and Hansen, 1979) H,O, (Persson et al., 1985; Mohler and Looser, 1967) and mediated electrodes (Bradley et al., 1989) or by utilizing the absorbance (Wolff and Mottola, 1978), fluorescence (Moody et al., 1986) or chemiluminescence (Peterson and Vurek, 1984; Mason, 1983). The latter mentioned chemiluminescent technique, in particular the chemiluminescence of luminol, is known to have a high sensitivity, a wide linear range and a good operational stability (Kricka and Carter, 1982; Grayeski, 1987). For glucose determination, the enzymatically generated hydrogen peroxide given in Eq. (1) is related to the intensity of light emitted by the following chemiluminescent reaction given in Eq. (2). P-D-glucose + 0, + H,O
Eiz f
luminol + H,O,
3-aminophthalate
~$l$,)
D-ghCOniC
acid + H,Oz + light (425 nm)
163
The presence of catalysts, e.g. horseradish peroxidase, and fairly basic conditions (pH 11-12) are required to obtain the emission of sufficient light by the chemiluminescent reaction (Gorus and Schram, 1979; Whitehead et al., 1979; Bostick and Hercules, 1975; Petersson et al., 1986; Auses et al., 1975; Malovolti et al., 1984; Grayeski, 1985) (Eq. 2). Unfortunately, the high pH conditions required for efficient light yields are incompatible with many enzyme reactions, for example the oxidation of glucose using glucose oxidase (GOD), which has a pH optimum of 6.5. Previously, attempts have been made to overcome this problem; for instance the use of alternative dyes such as peroxyoxalate, which were used in mixed aqueous-organic solvent mixtures under mild pH conditions (Seitz, 1978; Willams et al., 1976; Tsuji and Arakawa, 1985; Van Zoonen et al., 1986; Watanabe et al., 1986). Surfactants were used in luminol chemiluminescent systems, such as hexadecyltrimethylammonium chloride (CTAC) or cetyltrimethylammonium bromide (CTAB) (Igarashi and Hinze, 1988; Hitoshi and Hinze, 1987; Abdel-Latif and Guilbault, 1989; Khmelnitskii et al., 1984) to achieve chemiluminescence at mild pH conditions. But the application of these methods for on-line monitoring has some inherent disadvantages, such as low reproducibility, multiphase reaction and low sample frequency (Seitz, 1978; Hitoshi and Hinze, 1987; Pilosof and Nieman, 1982; Bostick and Hercules, 1975). As a continuous-flow assay technique, FIA has the advantages of being able to exactly control volumes, to mix patterns and residence times (Ruzicka and Hansen, 1988). Hence FIA offers the possibility of separating the reaction sequence of the chemiluminescent detection of glucose (Eqs. 1 and 2) into two steps in the flow system and allows the emission of sufficient luminol chemiluminescent light. Neither surfactant, organic solvent, nor membrane is required to form a pH gradient in the flow system. This paper describes a computer-controlled flow injection system based on luminol chemiluminescence for on-line glucose monitoring in animal cell (BHK) culture. The system performs glucose determinations in two separate chemical reaction steps. Firstly, the glucose was oxidized by immobilized glucose oxidase at pH 7.0. Secondly, the enzymatically generated hydrogen peroxide was propelled and mixed at a confluent point with a basic luminol solution to initiate the luminol reaction. The zone-sampling technique, introduced by Bergamin (Clark et al., 1989) was adopted in order to widen the linear analytical range.
Experimental
Reagents Glucose oxidase (GOD grade I, EC 1.1.3.4, from Aspergilllrs niger, 280 U mgg’) and peroxidase (POD grade II, EC 1.11.1.7, from horseradish, 200 U mg-‘) were obtained from Boehringer, Mannheim, F.R.G. Aminopropyl-derived controlled-pore glass (AMP-CPG-550 A, 120-200 mesh) was obtained from Fluka, Neu-Ulm,
164
F.R.G. Luminol (5amino-2,3-dihydro-1,4-phthalazinedione) was purchased from Boehringer, Mannheim, F.R.G. /&D-Glucose was purchased from Sigma, Munich, F.R.G. Hydrogen peroxide (30%), potassium hydroxide and potassium phosphate were purchased from Merck, Darmstadt, F.R.G. All chemicals were of analytical reagent grade and were used as received. All dilutions were made with distilled water. Luminol solution (1 mM) was made by dissolving 0.177 g of luminol in 5 ml KOH (2 N) which was then diluted to 1 litre with 0.05 M potassium phosphate buffer, pH 7.0. The pH of the final solution was adjusted to 12.5 with KOH (2 N). The concentration of POD in the luminol solution was 8 U ml-‘. The prepared luminol solution was stored for more than 30 h at room temperature prior to use. Aqueous solutions of hydrogen peroxide from 10-l M to 10m8 M were made by subsequent dilution of 30% hydrogen peroxide. Glucose standards covering the range lo-* M to lo-’ M were prepared by sequential dilution of 100 mM P-D-glucose, at least 24 h before use, to ensure attainment of mutarotational equilibrium, and were stored at 4°C. Enzyme
immobilization
The enzyme immobilization procedure adopted is as follows: 20 mg GOD were dissolved in 5 ml 0.01 M potassium phosphate buffer (pH 7.0) and were then mixed with 100 mg of AMP-CPG pre-activated with 2.5% glutaraldehyde. After 24 h storage at 4°C the enzyme immobilized on CPG was packed into an enzyme reactor (polycarbonate tube 0.5 mm i.d. 40 mm long). Nylon net 400 mesh (Reichelt Chemie, Heidelberg, F.R.G.) was used to retain the CPG in the reactor. Flow system
A schematic diagram of the flow system is shown in Fig. 1. The FIA system, consisting of pumps obtained from Jungkeit, Bovenden, F.R.G., and injection valves obtained from Tecator, Rodgau, F.R.G., was developed in-house and was adopted for performing zone sampling. For this purpose, the flow system consisted of two independent parts, each behaving as a single flow-injection system. The flow-rate of each carrier stream was set at 0.38 ml min-’ using pump tubes purchased by Tygon Verder, Dusseldorf, F.R.G. For on-line sampling from the bioreactor, a 0.22 pm microfiltration membrane (Sterivex-GS, Millipore, Bedford, U.S.A.) was used. 120 ~1 of the sterile filtered sample was introduced into the first carrier stream Cl and was dispersed with the aid of a magnetically stirred gradient chamber GC (200 pl), which causes a wide concentration gradient within this zone. For further dilution of the sample, an aliquot (120 ~1) of this dispersed zone was introduced into the second carrier stream C2, carefully controlling the time difference of the two injections by a computer (an industrial timer PS-3 from Klockner Moller, Bonn, F.R.G., was used). Different analytical ranges could be achieved by changing this time span. This second zone passed through the enzyme reactor ER, where the glucose was oxidized sufficiently by dissolved oxygen to form gluconic acid and
165
.-c% 0 “”
--------I r---------------
Fig. 1. FIA arrangement for on-line monitoring of glucose concentration during an animal cell culture. P. peristaltic pump; I, injection valve; Cl, C2. carrier streams for sample dilution and enzyme reaction (50 mM potassium buffer, pH 7.0); C3, luminol reagent; ER, enzyme reactor; W. waste; CC. gradient chamber; M, mixing ‘T’; TP, time processor; D. detector.
hydrogen peroxide. Mixing in a confluent ‘T’ with luminol solution C3, the enzymatically generated hydrogen peroxide was determined by the intensity of light emitted in the chemiluminescent reaction. The experiment was carried out at room temperature. The intensity of the light emitted was determined using a luminometer (AMKO Light Technology Instruments GmbH, F.R.G.), equipped with a 100 ~1 flow-through cell (100 ~1, LKB, Sweden). The flow-through cell was directly attached to the photomultiplier tube operated at -700 V (R928, Hamamatsu, Japan). Reference method The glucose concentration of samples taken was determined off-line with a YSI-glucose analyser Model 27 (Yellow Springs Instrument Company, USA) according to the manufacturer’s instructions (YSI, 1988).
Results The following optimization investigations using the zone sampling technique.
were performed,
without
predilution
Effect of luminol concentration and buffer concentration The influence of luminol concentration on the measurement of light intensity was investigated. At a POD concentration of 8 U ml-’ the maximum light intensity was
1 E-5
1 E-4
1 E-3 Luminol
1 E-2
(M)
Fig. 2. Effect of pH for the chemiluminescent reaction. Conditions: C2: 50 mM potassium phosphate buffer, pH 7.0. C3: 1 mM luminol solution in 50 mM potassium phosphate buffer, 8 U ml- ’ POD. injection of 200 PM H,Oz, room temperature.
obtained using a 1 mM luminol solution (data not shown). It was also observed that a buffer concentration of 50 mM was optimal for the enzyme reaction. So 1 mM luminol solution and 50 mM potassium phosphate buffer were utilized for the experiment. Effect
of pH
By the injection of hydrogen peroxide directly into the luminol carrier stream, C3, the optimum pH dependence of the luminol reaction was investigated. It was found that the signal increased remarkably at a pH of 12.5 (Fig. 2). The optimum pH of the phosphate buffer, C2, for glucose oxidase was also investigated, within the pH range 5.5 to 7.0 no influence on the glucose oxidase reaction was observed (data not shown). So the pH values of luminol and buffer were controlled at 12.5 and 7.0 separately throughout the experiment. Calibration
of the system
The hydrogen peroxide and hence glucose concentrations were obtained by comparison of sample peak height with calibration peak heights. A log-log calibration plot for the determination of hydrogen peroxide is shown in Fig. 3. The response is linear in the range 5.0 X 10-3-10-7 M with a RVAL = 0.982 (r.s.d. < 38, n = 5). The signal to noise ratio (S : N = 100) was high. Hence, the detection range was limited by the sensitivity of the PMT rather than by the background signal. The corresponding log-log calibration plot for glucose determination is shown in Fig. 4. The response was linear in the range 5 X 10-*-l X lo-’ M glucose with a RVAL 0.997 (r.s.d. < 38, n = 5).
167
1 .OE4 1000.0 7 5 x .z
0
6
--
100.0 -10.0 _-
s 2 d T
1 .OE-3, l.OE-7
l.OE-6
1 .OE-5
1 .OE-4
Hz02
l.OE-3
l.OE-2
0.1
(M)
Fig. 3. Calibration curve of hydrogen peroxide. Conditions: C2: 50 mM 7.0. C3: 1 mM luminol solution in 50 mM potassium phosphate buffer, rate 0.38 ml min-‘. room temperature.
potassium phosphate pH 12.5, 8 U ml-’
buffer, POD;
pH flow
On-line determination of glucose during a BHK culture with predilution using the zone sampling technique Different linear ranges could be obtained by selecting different zone sampling times. For on-line cultivation control a loo-fold dilution was required, this was obtained by using a zone sampling time of 110 s. The typical responses of the computer-controlled FIA to different concentrations of glucose in the range from 0.0 mM to 30 mM showed that the calibration curves were linear responses in the small concentration range rather than in the whole concentration range (data not shown). Therefore the calibration curves were calculated over the range 0.0 mM-5.0 mM with a slope of 0.22 (r.s.d. 3%, n = 5) and over the range 5.0 mM-30.0 mM with a slope of 0.40 (r.s.d. 3%, n = 5, RVAL = 0.995).
0.01-c 1 .OE-5
1 .OE-4
1 .OE-3 glucose
Fig. 4. Calibration
curve
of glucose.
1 .o :-2
(M) Conditions:
as Fig. 3.
168
-A-A 0.
18 16 E .5 ti 8 2 m
i
y:-**
+,.
. 4*&--,,
1
.>r
8-
.oQ
6-
l P
o/Biotechnology, 18 (1991) Elsevier Science Publishers 016816569100080M
BIOTEC
161-112 B.V. 0168-1656/91/$03.50
161
00597
On-line determination of glucose concentration throughout animal cell cultures based on chemiluminescent detection of hydrogen peroxide coupled with flow-injection analysis Y .L. Huang, Department
S.Y. Li, B.A.A. oj Enzyme
Dremel,
Technology,
Gesellschajt Braunschweig,
(Received
17 May
1990;
revision
U. Bilitewski fir
and R.D. Schmid
Biotechnologische
Forschung
GntbH,
F. R. G.
accepted
25 September
1990)
A flow-injection analysis (FIA) system for the on-line determination of glucose in animal cell cultures is described. The system is based on immobilized glucose oxidase (GOD). The hydrogen peroxide generated in the enzyme reaction is determined via a highly sensitive chemiluminescent reaction with luminol. Based on the measurement of the maximum emitted light intensity, the system was able to analyse hydrogen peroxide over the concentration range of lo-’ to 10-l M. For glucose determination, the system has a linear range of 10m5 to 5 x lo-’ M glucose, with an r.s.d. of 3% at the 1 mM level (5 measurements). The influence of luminol and buffer concentrations, pH and temperature on the chemiluminescent reaction were investigated. The enzyme reactor used was stable for more than 4 weeks in continuous operation, and it was possible to analyse up to 20 samples per h. The system has been successfully applied to on-line monitoring of glucose concentration during an animal cell culture, designed for the production of human antithrombin III factor. Results obtained with the FIA system were compared with off-line results, obtained with a Yellow Springs Instrument Company Model 27 (YSI). FIA; Chemiluminescence culture
Correspondence
Forschung
GmbH.
of luminol; On-line determination;
IO: Y.L. Huang, Dept. D-3300 Braunschweig,
of Enzyme F.R.G.
Technology,
Gesellschaft
Glucose; Animal cell
fir
Biotechnologische
162
Introduction Numerous animal cell cultures are carried out at industrial and laboratory scale for the production of biomass and products for medical application, such as interleukin II (IL-II; Wirth et al., 1987) human antithrombin III (AT-III; Wirth et al., 1989) and interferon (Reiser and Hauser, 1987). Glucose, as the main carbon source, is of essential importance for animal cell culture. Moreover, glucose consumption is also an indicator of cellular metabolic activity. Undesirable consequences of excess glucose concentration, such as catabolite repression (Demain, 1972) might be avoided by controlled feeding. The cultivation of baby hamster kidney cells (BHK) for the production of human antithrombin III protein (AT-III), for example, requires careful monitoring of the glucose level, as glucose is a limiting factor for both cell growth and protein production (Wirth et al., 1989). To date, animal cell reactors are equipped with some essential sensors for on-line monitoring of physical parameters e.g. pH, PO,, redox potential and temperature. The analysis of cultivation substrates such as glucose or lactate is usually achieved by off-line methods (Damen and Pettitt, 1980; Hesse et al., 1984; YSI, 1988). However these methods, which involve manual intervention in sampling or analysis have usually limited applicability since response time is typically too slow to be compatible with feedback control requirements and frequent sampling poses a risk of contamination. Hence, there is a requirement for efficient and reliable sensors for on-line monitoring of physiological parameters such as glucose or lactate, common substrates of animal cell culture (Sigma Catalog, 1988). Except for the determination of glucose based on its reducing properties, two methods have been developed to enhance the accuracy of analysis. First, glucose concentration can be quantified by liquid chromatography (Brinkmann, 1987) however these techniques require expensive instrumentation and often sample derivatization, and second, enzymatic assays for glucose determination have been developed, using either glucose oxidase (GOD), glucose dehydrogenase (GDH) (Persson et al., 1985; D’Costa et al., 1986) or glucose-6-phosphate dehydrogenase (G-6-PDH; Kunst et al., 1984). Various methods based on enzymes employ the technique of FIA coupled with the use of amperometric oxygen (Ruzicka and Hansen, 1979) H,O, (Persson et al., 1985; Mohler and Looser, 1967) and mediated electrodes (Bradley et al., 1989) or by utilizing the absorbance (Wolff and Mottola, 1978), fluorescence (Moody et al., 1986) or chemiluminescence (Peterson and Vurek, 1984; Mason, 1983). The latter mentioned chemiluminescent technique, in particular the chemiluminescence of luminol, is known to have a high sensitivity, a wide linear range and a good operational stability (Kricka and Carter, 1982; Grayeski, 1987). For glucose determination, the enzymatically generated hydrogen peroxide given in Eq. (1) is related to the intensity of light emitted by the following chemiluminescent reaction given in Eq. (2). P-D-glucose + 0, + H,O
Eiz f
luminol + H,O,
3-aminophthalate
~$l$,)
D-ghCOniC
acid + H,Oz + light (425 nm)
163
The presence of catalysts, e.g. horseradish peroxidase, and fairly basic conditions (pH 11-12) are required to obtain the emission of sufficient light by the chemiluminescent reaction (Gorus and Schram, 1979; Whitehead et al., 1979; Bostick and Hercules, 1975; Petersson et al., 1986; Auses et al., 1975; Malovolti et al., 1984; Grayeski, 1985) (Eq. 2). Unfortunately, the high pH conditions required for efficient light yields are incompatible with many enzyme reactions, for example the oxidation of glucose using glucose oxidase (GOD), which has a pH optimum of 6.5. Previously, attempts have been made to overcome this problem; for instance the use of alternative dyes such as peroxyoxalate, which were used in mixed aqueous-organic solvent mixtures under mild pH conditions (Seitz, 1978; Willams et al., 1976; Tsuji and Arakawa, 1985; Van Zoonen et al., 1986; Watanabe et al., 1986). Surfactants were used in luminol chemiluminescent systems, such as hexadecyltrimethylammonium chloride (CTAC) or cetyltrimethylammonium bromide (CTAB) (Igarashi and Hinze, 1988; Hitoshi and Hinze, 1987; Abdel-Latif and Guilbault, 1989; Khmelnitskii et al., 1984) to achieve chemiluminescence at mild pH conditions. But the application of these methods for on-line monitoring has some inherent disadvantages, such as low reproducibility, multiphase reaction and low sample frequency (Seitz, 1978; Hitoshi and Hinze, 1987; Pilosof and Nieman, 1982; Bostick and Hercules, 1975). As a continuous-flow assay technique, FIA has the advantages of being able to exactly control volumes, to mix patterns and residence times (Ruzicka and Hansen, 1988). Hence FIA offers the possibility of separating the reaction sequence of the chemiluminescent detection of glucose (Eqs. 1 and 2) into two steps in the flow system and allows the emission of sufficient luminol chemiluminescent light. Neither surfactant, organic solvent, nor membrane is required to form a pH gradient in the flow system. This paper describes a computer-controlled flow injection system based on luminol chemiluminescence for on-line glucose monitoring in animal cell (BHK) culture. The system performs glucose determinations in two separate chemical reaction steps. Firstly, the glucose was oxidized by immobilized glucose oxidase at pH 7.0. Secondly, the enzymatically generated hydrogen peroxide was propelled and mixed at a confluent point with a basic luminol solution to initiate the luminol reaction. The zone-sampling technique, introduced by Bergamin (Clark et al., 1989) was adopted in order to widen the linear analytical range.
Experimental
Reagents Glucose oxidase (GOD grade I, EC 1.1.3.4, from Aspergilllrs niger, 280 U mgg’) and peroxidase (POD grade II, EC 1.11.1.7, from horseradish, 200 U mg-‘) were obtained from Boehringer, Mannheim, F.R.G. Aminopropyl-derived controlled-pore glass (AMP-CPG-550 A, 120-200 mesh) was obtained from Fluka, Neu-Ulm,
164
F.R.G. Luminol (5amino-2,3-dihydro-1,4-phthalazinedione) was purchased from Boehringer, Mannheim, F.R.G. /&D-Glucose was purchased from Sigma, Munich, F.R.G. Hydrogen peroxide (30%), potassium hydroxide and potassium phosphate were purchased from Merck, Darmstadt, F.R.G. All chemicals were of analytical reagent grade and were used as received. All dilutions were made with distilled water. Luminol solution (1 mM) was made by dissolving 0.177 g of luminol in 5 ml KOH (2 N) which was then diluted to 1 litre with 0.05 M potassium phosphate buffer, pH 7.0. The pH of the final solution was adjusted to 12.5 with KOH (2 N). The concentration of POD in the luminol solution was 8 U ml-‘. The prepared luminol solution was stored for more than 30 h at room temperature prior to use. Aqueous solutions of hydrogen peroxide from 10-l M to 10m8 M were made by subsequent dilution of 30% hydrogen peroxide. Glucose standards covering the range lo-* M to lo-’ M were prepared by sequential dilution of 100 mM P-D-glucose, at least 24 h before use, to ensure attainment of mutarotational equilibrium, and were stored at 4°C. Enzyme
immobilization
The enzyme immobilization procedure adopted is as follows: 20 mg GOD were dissolved in 5 ml 0.01 M potassium phosphate buffer (pH 7.0) and were then mixed with 100 mg of AMP-CPG pre-activated with 2.5% glutaraldehyde. After 24 h storage at 4°C the enzyme immobilized on CPG was packed into an enzyme reactor (polycarbonate tube 0.5 mm i.d. 40 mm long). Nylon net 400 mesh (Reichelt Chemie, Heidelberg, F.R.G.) was used to retain the CPG in the reactor. Flow system
A schematic diagram of the flow system is shown in Fig. 1. The FIA system, consisting of pumps obtained from Jungkeit, Bovenden, F.R.G., and injection valves obtained from Tecator, Rodgau, F.R.G., was developed in-house and was adopted for performing zone sampling. For this purpose, the flow system consisted of two independent parts, each behaving as a single flow-injection system. The flow-rate of each carrier stream was set at 0.38 ml min-’ using pump tubes purchased by Tygon Verder, Dusseldorf, F.R.G. For on-line sampling from the bioreactor, a 0.22 pm microfiltration membrane (Sterivex-GS, Millipore, Bedford, U.S.A.) was used. 120 ~1 of the sterile filtered sample was introduced into the first carrier stream Cl and was dispersed with the aid of a magnetically stirred gradient chamber GC (200 pl), which causes a wide concentration gradient within this zone. For further dilution of the sample, an aliquot (120 ~1) of this dispersed zone was introduced into the second carrier stream C2, carefully controlling the time difference of the two injections by a computer (an industrial timer PS-3 from Klockner Moller, Bonn, F.R.G., was used). Different analytical ranges could be achieved by changing this time span. This second zone passed through the enzyme reactor ER, where the glucose was oxidized sufficiently by dissolved oxygen to form gluconic acid and
165
.-c% 0 “”
--------I r---------------
Fig. 1. FIA arrangement for on-line monitoring of glucose concentration during an animal cell culture. P. peristaltic pump; I, injection valve; Cl, C2. carrier streams for sample dilution and enzyme reaction (50 mM potassium buffer, pH 7.0); C3, luminol reagent; ER, enzyme reactor; W. waste; CC. gradient chamber; M, mixing ‘T’; TP, time processor; D. detector.
hydrogen peroxide. Mixing in a confluent ‘T’ with luminol solution C3, the enzymatically generated hydrogen peroxide was determined by the intensity of light emitted in the chemiluminescent reaction. The experiment was carried out at room temperature. The intensity of the light emitted was determined using a luminometer (AMKO Light Technology Instruments GmbH, F.R.G.), equipped with a 100 ~1 flow-through cell (100 ~1, LKB, Sweden). The flow-through cell was directly attached to the photomultiplier tube operated at -700 V (R928, Hamamatsu, Japan). Reference method The glucose concentration of samples taken was determined off-line with a YSI-glucose analyser Model 27 (Yellow Springs Instrument Company, USA) according to the manufacturer’s instructions (YSI, 1988).
Results The following optimization investigations using the zone sampling technique.
were performed,
without
predilution
Effect of luminol concentration and buffer concentration The influence of luminol concentration on the measurement of light intensity was investigated. At a POD concentration of 8 U ml-’ the maximum light intensity was
1 E-5
1 E-4
1 E-3 Luminol
1 E-2
(M)
Fig. 2. Effect of pH for the chemiluminescent reaction. Conditions: C2: 50 mM potassium phosphate buffer, pH 7.0. C3: 1 mM luminol solution in 50 mM potassium phosphate buffer, 8 U ml- ’ POD. injection of 200 PM H,Oz, room temperature.
obtained using a 1 mM luminol solution (data not shown). It was also observed that a buffer concentration of 50 mM was optimal for the enzyme reaction. So 1 mM luminol solution and 50 mM potassium phosphate buffer were utilized for the experiment. Effect
of pH
By the injection of hydrogen peroxide directly into the luminol carrier stream, C3, the optimum pH dependence of the luminol reaction was investigated. It was found that the signal increased remarkably at a pH of 12.5 (Fig. 2). The optimum pH of the phosphate buffer, C2, for glucose oxidase was also investigated, within the pH range 5.5 to 7.0 no influence on the glucose oxidase reaction was observed (data not shown). So the pH values of luminol and buffer were controlled at 12.5 and 7.0 separately throughout the experiment. Calibration
of the system
The hydrogen peroxide and hence glucose concentrations were obtained by comparison of sample peak height with calibration peak heights. A log-log calibration plot for the determination of hydrogen peroxide is shown in Fig. 3. The response is linear in the range 5.0 X 10-3-10-7 M with a RVAL = 0.982 (r.s.d. < 38, n = 5). The signal to noise ratio (S : N = 100) was high. Hence, the detection range was limited by the sensitivity of the PMT rather than by the background signal. The corresponding log-log calibration plot for glucose determination is shown in Fig. 4. The response was linear in the range 5 X 10-*-l X lo-’ M glucose with a RVAL 0.997 (r.s.d. < 38, n = 5).
167
1 .OE4 1000.0 7 5 x .z
0
6
--
100.0 -10.0 _-
s 2 d T
1 .OE-3, l.OE-7
l.OE-6
1 .OE-5
1 .OE-4
Hz02
l.OE-3
l.OE-2
0.1
(M)
Fig. 3. Calibration curve of hydrogen peroxide. Conditions: C2: 50 mM 7.0. C3: 1 mM luminol solution in 50 mM potassium phosphate buffer, rate 0.38 ml min-‘. room temperature.
potassium phosphate pH 12.5, 8 U ml-’
buffer, POD;
pH flow
On-line determination of glucose during a BHK culture with predilution using the zone sampling technique Different linear ranges could be obtained by selecting different zone sampling times. For on-line cultivation control a loo-fold dilution was required, this was obtained by using a zone sampling time of 110 s. The typical responses of the computer-controlled FIA to different concentrations of glucose in the range from 0.0 mM to 30 mM showed that the calibration curves were linear responses in the small concentration range rather than in the whole concentration range (data not shown). Therefore the calibration curves were calculated over the range 0.0 mM-5.0 mM with a slope of 0.22 (r.s.d. 3%, n = 5) and over the range 5.0 mM-30.0 mM with a slope of 0.40 (r.s.d. 3%, n = 5, RVAL = 0.995).
0.01-c 1 .OE-5
1 .OE-4
1 .OE-3 glucose
Fig. 4. Calibration
curve
of glucose.
1 .o :-2
(M) Conditions:
as Fig. 3.
168
-A-A 0.
18 16 E .5 ti 8 2 m
i
y:-**
+,.
. 4*&--,,
1
.>r
8-
.oQ
6-
l P
