Biosensors& Bioelectronics7 (1992) 569-514

A chemiluminescence fiber-optic biosensor system for the determination of glutamine in mammalian cell cultures M. V. Cattaneo, K. B. Male & J. H. T. Luong* Biotechnology

Research Institute, National

Research Council

(Received 3 February

Canada,

Montreal,

Quebec, Canada

H4P 2R2

1992; accepted 4 May 1992)

Abstract: Achemiluminescencc fiber-optic biosensor system has been developed for determining glutamine in hybxidoma cell cultures producing monoclonal antibodies against viral surface antigens. Glutaminase and glutamate oxidase (GLO) were immobilized onto aminopropyl glass beads via glutaraldehyde activation separately and packed in a column. Two separate columns containing immobilized GLO and catalase were placed upstream to eliminate endogenous glutamate. In the presence of ferricyanide, luminol reacted with hydrogen peroxide released from the enzymatic reactions to produce a chemiluminescence (CL) light signal which was detected and quantitated with a fiber-optic system. In combination with flow injection analysis it was possible to process samples virtually identically, thus avoiding difficulties in reproducing the CL signal. There was an excellent linear relationship between the CL response and standard glutamine concentration in the range 10e6 to 1O-3 M. A complete analysis could be performed in 2 min including sampling and washing. Each immobilized enzyme column was stable for at least 300 repeated analyses without any loss of activity. When the biosensor system was used for

the determination of glutamine in spent mammalian cell cultures, the values obtained compared well with those of high-performance liquid chromatography, thus validating the applicability of the CL fiber-optic system. Keywords: chemiluminescence, fiber-optic, glutamine, glutamate, glutaminase, glutamate oxidase, mammalian cells, hydrogen peroxide, luminol, ferricyanide.

INTRODUCTION

and

Glutamine is one of the medium components most utilized by mammalian cells (Butler et al., 1983; Thomas, 1986). Supplementation of glutamine has resulted in increased cell density

(Geaugey et al., 1989). However, an increase of glutamine beyond a certain limit produced ammonia at toxic or inhibitory levels (Doyle & Butler, 1990). In another extreme, cultivation under depleted glutamine caused severe growth limitation (Adamson ef al., 1987). Therefore, it is

*To whom correspondence

important

should be addressed.

096%5663/92/$05.00 Q 1992 Elsevier Science Publishers

Ltd.

higher

secretion

to

regulate

of monoclonal

glutamine

antibody

during

the 569

M. V: Cattaneo,& B. Male, .I H. T. Luong

Biosensors& Bioelectronics

course of cultivation. A rapid monitoring of glutamine could be incorporated as part of a control strategy to optimize the required glutamine uptake. It is somewhat difficult to envision highperformance liquid chromatography (HPLC), the most commonly used technique, as a truly online detection system because of its inherently long sampling and washing times. A recently developed glutamine biosensor employing immobilized glutaminase (GAH) and glutamate oxidase (GLO) and a hydrogen peroxide electrode could be adapted for on-line application (Cattaneo et al., 1992). Although applicable, the detection of hydrogen peroxide is subject to several electrochemically interfering substances such as ascorbic acid, uric acid, tyrosine, iron( etc. In addition, very few of the present amperometric sensors are useful below lo-100~~~ due to interferences and background signals, especially when detection occurs in real samples (Carr & Bowers, 1980). Detection based on the chemiluminescence (CL) reaction between hydrogen peroxide and luminol in the presence of ferricyanide can improve both the selectivity and limit of detection: fenicyanide

luminol + H202 -

pH 10.7

3-aminophthalate +N2+hv (1)

By using a photodiode incorporated in a miniaturized flow system Petersson et al. (1986) obtained a linear CL signal for glucose in the range lob2 M to 10T6 M. In this study, a flow-injection fiber-optic system based on immobilized GAH and GLO has been developed for the determination of glutamine in mammalian cell cultures based on the following reactions: glutamine

+ Hz0 2

glutamate + NH3 (2)

glutamate + 02 z a-ketoglutarate pH5’2 + H202 + NH3

(3)

Hydrogen peroxide released from the enzymatic reactions reacted with luminol in the presence of ferricyanide to produce light, which was detected and quantitated by a photomultiplier attached to a detector cell. The applicability of the fiber-optic biosensor system for the determination of glutamine in mammalian cell cultures in terms of selectivity, sensitivity, and stability is also presented and discussed in detail. 570

EXPERIMENTAL

Luminol, glutamine, glutamate, glutaraldehyde, glutaminase GAH (EC 3.5.1.2), catalase (EC 1.11.1.6), ascorbic acid and acetaminophen (4-acetamidophenol) were purchased from Sigma (St Louis, MO). Ferricyanide and L-tyrosine were purchased from Aldrich (Milwaukee, WI). Citric and uric acids were obtained from Anachemia (Champlain, NY) and American Chemicals (Montreal, QC), respectively. L-Glutamate oxidase (GLO, mol. wt 140 000) extracted from Streptumyces sp. Xl 19-6 (Kusakabe et al., 1983) was purchased- from Yamasa Shoyu Co. (Choshi, Chiba, Japan). LuminoVferricyanide solutions preparation A 5 mM luminol stock solution was prepared by dissolving luminol in a small quantity of base (2 M) and diluting to the appropriate volume in deionized water. Luminol (O-125 InM) and ferricyanide (30 mM) in 200 IIIM glycine buffer pH 11.2 were mixed in the flow line just prior to the inlet to the detection cell (Fig. 1). Immobilization of enzymes Glutaminase/glutamate oxidase immobilized on aminopropyl glass beads Four hundred milligrams of aminopropyl glass beads (80-120 mesh and 70 nm pore size) were washed extensively with phosphate buffer saline (PBS) (9 gl-’ sodium chloride, 20 mM phosphate, pH 7) split in two equal batches, each batch activated with 3 ml of 2.5% (w/v) glutaraldehyde in PBS for 2-3 h at room temperature (20-22 “C). The resulting glass beads were then washed extensively with 20 IIIM phosphate buffer pH 5.2 to remove unreacted glutaraldehyde and used for subsequent enzyme immobilization. A 3 ml solution of GAH (16.7 U ml-’ and 0.18 mg protein ml-‘) in 20 mM phosphate buffer pH 5.2 was contacted with one batch of the activated aminopropyl glass beads. Similarly, a 3 ml solution of GLO (8.3 U ml-’ and O-3 mg protein ml-‘) was covalently immobilized with the other batch of beads. The beads were slowly rotated end-over-end in capped test tubes overnight at 4°C. The assay for protein amount

Chemiluminescenceftic

Biosensors & Eioeltvtronics

biosensor

To waste

I

Amplifier recorder

I

Fig. 1. Schematic diagram of the CLJber-optic system.

indicated no activity as well as no protein in the supematant of each batch. After immobilization the beads were mixed together and packed in a 15 cm length of tygon tubing (2.54 mm i.d.) furnished with silanized glass wool at both ends to retain the beads; O-2 g of beads will pack into approximately a 12 cm column length. The GAH/GLO column was stored in 20 mM phosphate buffer pH 5-2 supplemented with 200 mM sucrose. Glutamate oxidase and catalase on aminoprapyl glass beads

Three milliliters of GLO (8.3 U ml-’ and 1.7 mg protein ml-‘) or 10~1 of bovine liver catalase (42 000 U mg-’ and 64 mg protein ml-‘) solutions in PBS (pH 7) were immobilized with separate 250 mg batches of activated aminopropyl glass beads. Apparatus As shown in Fig. 1, a peristaltic pump (flow injection analysis (FIA) pump 1000, FIAtron Laboratory Systems, WI, USA) delivered the reagents at a preset flow rate. A 113 yl sample was injected into this stream by a motorized injection valve (FL4 valve 2000, FIAtron Laboratory Systems). The sample or phosphate buffer flowed through enzyme columns containing in turn GLO, catalase and GAHlGLO and then it was mixed with fe~~anide/~ycine and luminol just prior to reaching the detection chamber. The detection chamber, consisting of a reflecting polished steel drum of 100~1 internal volume,

was enclosed in a black nylon casing and equipped with sample inlet and outlet, air inlet and fiber-optic ports. The chemiluminescence signal was detected and amplified with Oriel components including a 36-inch glass fiber-optic bundle (model 77525) coupled to a photomultiplier tube (model 77344) and a readout device (model 7070). The photomultiplier tube (PM~-applied voltage was set at SSOV, unless specified otherwise. A FIAdapte? module provided further filtering and tenfold amplification of the light signal. Continuous light output was measured, integrated and recorded on a Spectra-Physics SP4270 integrator. Measurement of glutamine levels The GLO and catalase columns were placed in the sample flow line before the GLO/GAH column to remove endogenous glutamate. Glutamine standards from a 200 IIIM HPLC and mammalian cell culture standard supema~nts were diluted accordingly in 20 mM phosphate buffer pH 5.2 containing 50 IIIM NaCl. The culture samples were taken from culture supematants of murine hybridoma cells producing monoclonal antibodies against red blood cell antigens. The fresh medium normally used in the culture (DMEM, Sigma, St Louis, MO) contained 10% fetal bovine serum and 2 mM glutamine further augmented to about 4 mM prior to use. The samples were analyzed for glutamine by the fiber-optic biosensor and by standard HPLC (Cattaneo et al., 1992). 571

M. K Cattaneo, K 3. Male, f. H. T. Luong

RESULTS AND DISCUSSION Optimizationof the luminol-ferricyaaide-peroxide system

100

80

In successive optimization experiments luminol, ferricyanide, pH, applied potential and flow rate were tested in turn while keeping the other variables constant. A series of experiments was first performed to study the effect of the potential applied to the photomultiplier on the detection sensitivity. The linear response increased tenfold as the applied potential to the photomultiplier was increased from 700 to 95OV. The maximum potential applied of the photomultiplier used in this study is 1000V and operation at this potential resulted in a highly unstable baseline signal. Consequently, the photomultiplier detector was set at 950 V. The response signal increased with increased luminol concentration without reaching a maximum, in agreement with the work of Bostick & Hercules (1975). The selected luminol COnwntratiOn

was

0.125

mM

SinCe

SigII&ti-

noise ratios greater than 100 to 1 were easily achievable. The response signal was reduced considerably by removing air mixing to the detection chamber. The instability of luminol with time has been well reported (Auses et al., 1975). The stock luminol solution was prepared in deoxygenated water to slow down the degradation of luminol in the presence of oxygen under basic conditions (White & Bursey, 1964, White et al., 1964):

60

JO

20

0 10

I

100

Ferricyanide

(mM)

Fig 2. Fkrrkyanide optimization (PMT at 950 K O-125 mu luminol, pH 10.6, IOJIMH,O, in 20 mMphosphate buffer pH 5-2, 60 ml h-l flow rate).

buffer streams is 5.2 and that of the luminol stream is 12. After the four streams were merged as shown in Fig. 1, the pH of the effluent from the detector cell was about 10.6-10.7. To maintain the pH of the sample 20 mM phosphate was used. Any increase in the sample buffer strength beyond 20 mM phosphate would result in a decrease of the effluent pH and a resulting decrease in the response signal (figure not shown). Using standard hydrogen peroxide, increasing the flow rate gave increasing CL readings within the range studied (Fig. 3) and a flow rate of 1 ml min-’ was chosen to provide at least 30 6000

luminol + 02 + OH- 3-aminophthalate +N,+hv

(4

The luminol solution did not show any sign of immediate instability, as evidenced by the same response obtained for standards at the beginning and end of a 5-6 h analysis. Since the ferricyanide ~ncen~tion versus response plot reached a maximum for ferricyanide levels around 30 mM, this concentration was used in subsequent experiments (Fig. 2). The luminol-hydrogen peroxide reaction is highly dependent on pH and proceeds maximally at pH 10.7 (Bostick & Hercules, 1975).In order to achieve this pH optimum, concentrated glycine buffer (200 mM, pH 11.2) was added to the ferricyanide reagent. The pH of the sample and 572

5000

1000 0

0

10

20

30 Flowrate

40

$0

60

70

80

(mllhr)

Fig. 3. Flow rate optim~atio~ (PMT at 7&I K IOmM luminol, 30 mM &FeCI+l& pH 10.6, f m&f Hz02 in deionized water).

analyses per hour. The higher respanse with increased flow rate could be due to a combination of reduced eontaet time between reagents and more light emitted per unit time reaching the fiber-optic. The signal response to hydrogen peroxide was linear in the range 10N6to 10V3M.

Responses of the biosensor

The biosensor system did nut respond to the standard sulution containing either hydrogen peroxide (up to I mM) or glutamate (up to 1 mM). Such an obsession validates the applicabili~ of the biosensor system for the measurement of glutamine regardless of the presence of any endogenous glutamate. The glutamine standards used fur the analysis &owed a linear response in the range fO-3 to 1O-6M with a currelation coefficient of 0995 (slope = 8 pApM_“). The ~p~ducibili~ was +3+9%for 22 ~pe~ted sample analyses of 176~M glutamine. Sim~arly, a good reproducibility (k3*9%) was observed when 4.2pM glutamate was assayed repeatedly (24 analyses) in the absence of the GLO and catalase columns. The de~~ina~o~ of glutamine and/or glutamate could be performed in 2 min, including sampling and washing. Stability considerations

During repeated analyses, 50 mM NaCl was added to fhe running bufFer to prevent any binding between interfering compounds and the enzyme column, This was essential since the activity of the immobilized en~mes decreased rapidly if 50 mM NaCl was not used. The anticipated binding between unknown compounds and the enzyme column was later dem~nst~ted since the original activity of the GAHIGLO spent column running in the absence of salt could be restored by reconditioning with 2 M NaCl. The GAH/GLG column was stabIe for at least two months when stored at 4°C in 20 MM phosphate, 200 mM sucrose, pH 52. The GLO and catalase columns were also stable for at least two months when stored at 4°C in 2~rn~ phosphate buffer pH 5.2. The enzyme columns were used for at least 300 consecutive analyses without any si~i~cant loss of their originaI activity.

Interference considerations

Selected substances known to interfere with the arnpe~rne~~ detection ~plati~~m vs sib&silver chloride at +O-7V) were added to a standard hyd~~n peroxide solu~on (200pM) to decade any changes in the osmose light signal. Electrochemically interfering compounds such as uric acid and acetaminophen when present at 1 mM co~~n~a~~~ were observed to depress the light signal by as much as 30%.The interference of urio acid on the chemilumineseence response was porously reported (Bostiek & Hereules, 1975) and various methods have been su~ested for ~rno~ng this substance if it is present in bigb amounts (3 mM) (Williams e?tal,, 1976).It is worth noting that uric acid and acetaminophen are not known to be present in any significant quantity in the mammalian cell cultures tested in this study. of the CL fi~r~~c sensor to ~u~~~e/ glutamate in mamma&w cell cultares R~~nse

The fiber-optic biosensar and HPLC methods were used in the estimation af &&amine concentration in mammalian cell. culture supernatants. This pulsed batch culture ex~~ment las~ng over 200 h had glutamine inje~ed ~~~~cally to prolong the lifespan of the culture. Since the culture medium contains several amino acids and other unknown compounds, it is of importance to investigate the non-specific reaction of such chemicals with 1uminoI in the presence of f&cyanide. Repeated injections of the cell culture samples to the CL system in the absence of the GLG/G~ enzyme ~l~rn~ produced the normal steady baseline signal, thus confirming no interference of the sample constituents on the CL reaction. Similar to the injection of glu~mine standard, each analysis could be performed in 2 min, and the glutamine condensation determined by the b&sensor mimicked well with those of HPLC (Fig. 4). The difference between the two methods is on average less than 10%. Alternatively, the biosensor values obtained when plotted against those of HPLC resulted in a straight line with a slope of 1a038and a curreiatioa lenient of O-95, Such good agreement thus validated the &ability and applicability of the ~he~u~nesc~n~e fiberoptic system for the meas~~ment of glu~mine levels in mammalian cell cultures. 573

M. V: Cattaneo, K. B. Male, JVH. T Luong

Biosensors & Bioelectronics Carr,

t

0

’

0

J

50

100

150

200

250

Time (hrs) Fig. 4. Glutamine determination by HPLC and CLfiberoptic biosensor of hybridoma cell culture.

REFERENCES Adamson, S. R, Behie, L. A, Gaucher, G. M. & Lesser, B. H (1987). Metabolism of hybridoma cells in suspension culture: evaluation of three commercially available media. In Commercial Fknktion of Monoclonal Antibodies, ed S. S. Seaver. Marcel Dekker, New York, pp. 17-34. Auses, J. P., Cook, S. L. & Maloy, J. T. (1975). Chemiluminescent enzyme method for glucose. Anal. Chem., 47,244-9. Bostick, D. T. & Hercules, D. M. (1975). Quantitative determination of blood glucose using enzyme induced chemiluminescence of luminol. Anal. Chem., 47,447-52. Butler, M., Imamura, T., Thomas, J. & Thilly, W. (1983). High yields from microcarrier cultures by medium perfusion. d Cell Sci., 61, 351-64.

574

P. W. Jr Bowers, L. D. (1980). Immobilized Enzymes in Analytical and Clinical Chemistry. John Wiley, New York, p. 251. Cattaneo, M. V., Luong, J. H. T. & Mercille, S. (1992). Monitoring glutamine in mammalian cell cultures using an amperometric biosensor. Biosensors and Bioelectronics, 7,329-34. Doyle, C. & Butler, M. (1990). The effect of pH on the toxicity of ammonia to a murine hybridoma. J. Biotechnol., 15, 91-100. Geaugey, V., Duval, D., Geahel, I., Marc, A. & Engasser, J. M. (1989). Influence of amino acids on hybridoma cell viability and antibody secretion. Cytotechnology, 2, 119-29. Kusakabe, H., Midorikawa, Y., Fujishima, T., Kuninaka, A. & Yoshino, H. (1983). Purification and properties of a new enzyme, L-glutamate ox&se, from Streptomyces sp. X-119-6 grown on wheat bran. Agric. Biol. Chem., 47, 1323-8. Petersson, B. A, Hansen, E. H. & Ruzicka, J. (1986). Enzymatic assay by flow injection analysis with detection by chemiluminescence: determination of glucose, creatinine, free cholesterol and lactic acid using an integrated FL4 microconduit. Anal. Len., 19, 649-65. Thomas, J. N. (1986). Nutrients, oxygen and pH. In Mammalian Cell Technology, ed. W. G. Thilly. Buttenvorths, Boston, pp. 109-30. White, E. H. & Bursey, M. M. (1964). Chemiluminescence of luminol and related hydrazides: the light emission step. .l Am. Chem. Sot., 86941-2. White, E. H., Zatirou, O., Kagi, H. H. & Hill, J. H. M. (1964). Chemiluminescence of luminol: the chemical reaction. J. Am. Chem. Sot., 86,940-l. Williams, D. C., Huff, F. & Seitz, W. (1976). Glucose oxidase chemiluminescence measurement of glucose in urine compared with the hexokinase method. Clin. Chem., 22, 372-4.