Two FIA-based biosensor systems studied for bioprocess monitoring

Journal o f Biotechnology, 31 (1993) 345-356 © 1993 Elsevier Science Publishers B.V. All rights reserved 0168-1656/93/$06.00

345

BIOTEC 00968

Two FIA-based biosensor systems studied for bioprocess monitoring Th. Scheper b, W. Brandes a H. Maschke c, F. P16tz a and C. Miiller b a Institutfiir Technische Chemie, Universitiit Hannover, Callinstr. 3, D-30167Hannover, Germany b Institut fiir Biochemie, Westfiilische Wilhelms- Universitiit Miinster, Germany c Genentech Inc., 460 Pt. San Bruno Blvd., South San Francisco, CA 94080, USA (Received 13 May 1992; revision accepted 23 January 1993)

Summary In this paper, two different FIA-based biosensor systems are described for application to different biotechnologically relevant purposes. In the first system, single fiber optodes were used to determine the pH, urea and penicillin V concentrations. A two-channel system was developed for the simultaneous monitoring of different variables to increase the analysis accuracy. This system was used for monitoring the penicillin V concentration during a cultivation of Penicillium chrysogenum. The second system described is a calorimetric immunoassay based on the use of an enzyme thermistor. A sandwich assay with protein A immobilized on a solid support for the determination of various IgGs was established. A fusion protein of protein A and fl-galactosidase obtained from a recombinant E. coli strain was used in the labelling and detection reaction. This system is designed for future application in bioprocess monitoring. Biosensor; Flow injection analysis; FIA; Optode; Calorimetry; Immunoanalysis

Introduction Biosensors used in flow injection analysis (FIA) systems offer interesting advantages over conventional analysis techniques for bioprocess monitoring. The sensiCorrespondence to: Thomas Scheper, Institut fiir Biochemie, Westf~ilische Wilhelms-Universit~it MOnster, Wilhelm-Klemm-Str. 2, D-48149 MOnster, Germany.

346 tivity and selectivity of biosensors can be used, while the disadvantages of these sensors (e.g., drift problems) can be excluded by the FIA technique (KittsteinerEberle et al., 1989; Schmid, 1990; Ogbomo et al., 1991). This combination of biosensors and flow injection analysis techniques is called biosensor system in this paper. However, the success of such biosensor analysis systems depends on the knowledge available on the bioprocess. When this knowledge is not available, it often is hard to run the analyses under optimal analysis conditions. Thus, this knowledge must be gained by application of such biosensor systems under real bioprocess conditions and by developing flexible biosensor systems which can be used for a wide variety of analyses or for the simultaneous monitoring of different variables. Two biosensor systems will be described in this paper: one for the simultaneous optical detection of different analytes and the other for the calorimetric immunoanalysis of IgGs with appropriate binding affinity to protein A.

Materials and Methods

Optodes During the last years a large number of publications appeared describing the development of optical sensors for different analytes (Scheller et al., 1989; Seitz, 1984, 1988; Wolfbeis, 1986, 1989). Only few authors used these sensors for substrate determination in real processes. It was our aim to develop optical sensors (optodes) for monitoring of penicillin and urea concentrations. The optodes were based on the pH-sensitive fluorescence properties of special organic dyes. These pH optodes served as transducers for the biosensors. The enzymes and the fluorescent dye were immobilized by cross-linking with glutaraldehyde. The substrates diffused into the membrane and the enzyme catalyzed reaction produced a pH shift, which was monitored by the change in the fluorescence intensity of the dye.

Chemicals. Polyvinylalkohol (Sigma), aminofluorescein (Aldrich), urease (Sigma), penicillinase (Calbiochem), penicillin G amidase (gift from Hoechst), BSA (Aldrich), Isothiocyanate (Sigma). Media. The penicillin V cultivations were performed in a synthetic medium with phenoxyacetic acid as a precursor, as described in M611er et al. (1992a,b). A typical hemodialysis buffer was used in the urea monitoring experiments (102 mM NaC1, 2 mM KC1, 1.25 mM CaC12, 0.5 mM MgCI2, 33 mM NaAc). Optodes. All experiments were performed with single 1 mm fibers (Torray, PF-U-CD 1501). Standard couplers (TP 110 Coupler, Laser Components) were used to connect the sensing fiber with the signal transmitting fibers.

347

I 0

t~

o~ o

O

3rn

m, O

lamp

filter

o

/~//r.-I

L

J dichroitic mirror

fiber optic Fig. 1. Schematic design of the modular filter fluorometer. A halogen tungsten lamp is used as light source. The excitation light is coupled into the single fiber via a dichroitic mirror and a microscope objective. A photomultiplier is used for detection of the fluorescent light.

Fluorescence photometer.

All measurements were performed in a fluorescence p h o t o m e t e r which was developed especially for the application of single fibers as optodes. Fig. 1 shows the design of the photometer. A 5 W halogen tungsten lamp served as light source. The light was filtered by an interference filter (480 nm) guided through a dichroitic mirror (Inst. fiir Quantenoptik, Universidit Hannover) and focused at the end of an optical fiber. Via the same filter, the backward fluorescence light passed the dichroitic mirror, was filtered at 530 nm with an appropriate interference filter, and measured by a photomultiplier tube. A special version of this system was developed for the two-channel measurements. Here, the excitation light is focussed on a bifurcated fiber and guided through a mechanical chopper (Standford Instruments) system. Two optodes can be irradiated with different frequencies, as shown in Fig. 2. The fluorescence signals from each optode have the same modulation, and a lock-in amplifier ( E G & G, model 8210) is used for the demodulation of both signals after detection at the photomultiplier.

Calorimetric immunoassay The experiments were performed in an enzyme thermistor device described in detail in earlier publications (Danielsson and Mosbach, 1987; Danielsson et al., 1989; Danielsson, 1990; Hundeck et al., 1990). In these systems, the heat generated

348

0

light source bifurcated fiber

Fig. 2. Schematic

design of the two-channel fluorometer system. A chopper wheel amplifier are used for the simultaneous analysis via two optodes.

and

a lock-in

during an enzymatic detection reaction is monitored. Thus, various analytes can be monitored in biotechnological processes, even simultaneously (Danielsson and Mosbach, 1987; Danielsson et al., 1989; Hundeck et al., 1990). In the experiments, a new simple sandwich assay between immobilized protein A, different IgGs, and a fusion protein from protein A and P-galactosidase (Hellebust et al., 1989) was established. Thus, IgG bound to the immobilized protein A could be labelled with the fusion protein. Afterwards, lactose was injected into the system. This substrate could be converted by the fusion protein as a function of the /3-galactosidase activity bound during the assays. In a different column the glucose generated during the previous step was analyzed via an enzyme thermistor device. The experimental setup is shown in Fig. 3. Chemicals. Protein A (SPA) (Sigma P 6650; CNBr-Sepharose 4B (Sigma C-9142); Lysozyme (Boehringer, Mannheim, 1243004); glucose oxidase (Sigma G-7016); catalase (Sigma C-401, rabbit IgG (Sigma, I-5006), mouse IgG2a (Sigma, M-7769), mouse IgG2b (Sigma, M-7644). Carrier buffer: 0.1 M KPP, pH 7.0, 1 mM MgCl,, flow rate: 0.6 ml min-‘. For the detection of the bound P-galactosidase activity lactose samples (5 mg ml-‘) were injected. For the elution of the sandwich, a 0.2 M glycine-HCl buffer (pH 2.2) was used to regenerate the SpA column. Fusion protein The fusion protein from protein A (from Staphylococcus aureus) and /3-galactosidase (from Escherichia coli) is produced in the recombinant E. coli strain JM103 [pRK248cIts, pRITlB1. The plasmid pRITlB (Hellebust et al., 1989) is used for excess production of the fusion protein under the control of the pn-promoter. The p.-promoter is controlled by c1 protein produced by plasmid pRK248cIts in its temperature-sensitive mutant cIts857. The resulting expression system JM103 [pRK248cIts, pRITlB] can be induced for the production of the

349 waste

~

>nj:lon SpA

rabbit IgG cellysate lactose elution buffer

witching valve

xx

OD

\

~, N \ \ \ \ \ \ \ \ \ \ \ " q enzyme-thermistor waste

Fig. 3. Schematic design of the Thermometric Enzyme Linked Immunosorbent Assay (TELISA) system used in this paper. A Lund-type enzyme thermistor was used for glucose analysis. Glucose is generated in the sandwich ELISA reaction performed in the SpA column.

fusion protein via a temperature shift from 30°C to 42°C. Both plasmids are transformed into competent JM103 cells (Chung and Miller, 1988) and stored in glycerol cultures at -80°C. The cells grow at 30°C in 200 ml cultures in M9 medium (Sambrook et al., 1989). At a n OD6o0n m o f 2, 15 m| of a 20-fold LB medium is added and the temperature is raised to 42°C. The fusion protein is enriched intracellularly. Cells were harvested, spun down (10 min, 10000g, 4°C) and washed with washing buffer (0.1 M NaC1, 10 mM Tris-HC1, 1 mM EDTA, pH 8). Afterwards, the cells were resuspended in 15 ml SolI (50 mM glucose, 25 mM Tris-HCl, 1 mM EDTA, pH 8), and 30 mg lysozyme were added. This mixture was incubated for 30 min on ice, and the fusion-proteincontaining supernatant was used after centrifugation (30 min, 42 000 g, 4°C).

Results and Discussion

pH Optodes. The pH-dependent fluorescence of fluorophores immobilized on the tip of the single fibers was used for the optical pH measurement. The configuration of a pH optode is shown in Fig. 4. For instance, aminofluorescein or fluorescein isothiocyanate were immobilized in a polyvinyl alcohol membrane cross-linked with glutaraldehyde (Zhujun et al., 1989) on the tip of the fibers. A mixture of 1 p.1 of a reaction mixture (50 ~1 of 5% polyvinyl alcohol, 10/zl 2.5% glutaraldehyde and 10 /zl 4 N HC1) were placed on tip of the fiber and were

350

pH-optode

fiber--optic

" "

fluorescein covalently bound to polyvinylalcohol Fig. 4. Principle of the optical pH sensor. On tip of the 1 mm fiber, the fluorescent dye was immobilized as described in the text. allowed to polymerize for 3 - 4 min. The optimal measuring p H range of the o p t o d e is between 5 and 8 (Fig. 5) with an accuracy of 0.01 p H unit. W h e n the enzymes were immobilized on the p H o p t o d e or together with the pH-sensitive dyes in the membrane, a biosensor is obtained. T h e analytes react to acidic or basic products in an enzyme-catalyzed reaction, generating a p H change in the membrane. Thus, the fluorescence intensity of the dye is influenced.

Urea optodes. A urea biosensor was obtained by coimmobilizing urease and FITC-labelled bovine serum albumine (BSA) in the m e m b r a n e . Here, 1 /xl of a reaction mixture ( 1 0 / z l B S A - F I T C (100 mg m l - 1 ) and 10 tzl enzyme solution (100 mg m l - 1 ) ) w e r e placed on tip of the fiber and allowed to polymerize for 5 - 1 0 min in a glutaraldehyde-containing gas phase. This urea o p t o d e was used for the determination of urea in hemodialysis buffers. T h e urea concentration during the

4ooo_ 16000::]j

.

.

,

,

° signal

(~

o differentiation 1 oooo

(0

a

/

6oooi 40002 20000

i

3

i

i

i

J

4

5

6 pH

7

8

9

J

10

Fig. 5. Calibration curve obtained with the pH optode (fluorescence dye aminofluorescein). It becomes obvious from the data that the optimal analysis region is in the range of pH 5.5 to 8 with a resolution of 0.01 pH unit.

351

6-portvalve 0o~0

Jffer

dialysisbuffer pH 5.6 sensor

I

stirrer

standard solutions

Fig. 6. Experimental set-up for urea monitoring in a simulated hemodialysis using a gradient mixer system. The gradient mixer is used to supply the exact urea concentration and buffer flow as in a real hemodialysis procedure. hemodialysis treatment ranges from 20 - 50 m M to below 10 m M (Thavarungkul et al., 1991). To stop the treatment at the optimal point, it is necessary to monitor the urea concentration on-line. The sensor was interfaced to a simulated dialysis experiment as shown in Fig. 6. The dialysis buffer containing between 3 and 10 m M urea in the gradient mixing system was diluted continuously with buffer to obtain optimal analysis conditions. This on-line monitoring is shown in Fig. 7. A good correlation was obtained between optically measured data and the urea concentration in the dialysis simulation experiment (deviations below 3.5%). At certain time intervals, calibration cycles were started automatically. The lifetime of the sensor was in the range of 6 days.

Penicillin optodes.

A penicillin optode was prepared for the analysis of cultivation samples. Here, the enzyme penicillin-G-amidase was coimmobilized with the B S A - F I T C in the m e m b r a n e on the fiber tip (see above). In the experiments, samples from a Penicillium chrysogenum cultivation containing penicillin V were analyzed at different concentrations (M611er et al., 1992a). Since the signal from an enzyme optode is based on a p H optode, the signal was extremely sensitive to changes in the buffer concentration and the p H value of the sample. To lower the influence of these variables, simultaneous monitoring of at least two different variables ( p H and analyte concentration) was necessary to obtain data with low standard deviations. A p H optode and a penicillin V optode were used simultaneously. The influence of the p H on the signals were calibrated in penicillin-free

352

14000 calculated data calibration 1

1200010000o 0

80006000-

measured / data

40002000calibration 2

0

50

1O0

150 time [mini

200

250

Fig. 7. Comparison of on-line monitored data (including calibration) and real concentrations in the gradient mixer system (calculated data). The fiber was directly placed in the buffer flow as described in Fig. 6.

cultivation samples at different pH values (Fig. 8). In penicillin-containing medium, the signals from both optodes were used for data processing. With the known pH dependence of the penicillin optode, it is possible to calculate the penicillin concentration independently from the pH value (Fig. 9). A good correlation between calibration and cultivation data could be obtained (standard deviation below 1.5%). The sensor was stable for several months, since the enzyme is extremely stable (Brand et al., 1991a, b). Immunoanalysis

In all TELISA (Thermometric Enzyme Linked Immunosorbent Assay) experiments immunoassays based on the sandwich principle were used, as shown in Fig. 10 (Mattiasson, 1977; Danielsson and Larson, 1990). Protein A from Staphylococcus aureus (SPA) was immobilized on CNBr-sepharose and filled into a flow-through column (see Fig. 3). Buffer was pumped continuously through the entire system. IgG-containing samples were injected via the injection valve. The IgGs were bound to the immobilized SpA. Afterwards, a fusion-protein-containing solution was injected, and a sandwich was formed. IgG in the SpA-column was labelled by this sandwich. After a washing step the valve prior to the enzyme thermistor was switched in such a way that the buffer flow had to pass the enzyme thermistor. The washing step is necessary to avoid unspecific binding of the fusion protein in the whole biosensor system (e.g., coils, enzyme column). When no washing step was included,

353

5( 4~ 4( 0 C

3!

0

3( 0

2!

_,2'

21 1!

pH 7.1

lq 5 0

ro

',

Penicillin V concentration [g I - i] Fig. 8. Influence of the pH on the optical penicillin V determination. Calibration graphs were obtained at different pH values. t h e b l a n k value of t h e g a l a c t o s i d a s e activity i n c r e a s e s with e a c h analysis cycle. H e r e , t h e flow p a s s e d a c o l u m n with i m m o b i l i z e d glucose oxidase a n d catalase. In o r d e r to analyze t h e /3-galactosidase activity b o u n d in t h e sandwich assay, lactose was i n j e c t e d into t h e e n t i r e system. L a c t o s e is c o n v e r t e d into glucose a n d

35

o~ 25 20 •o= 15 ~ ~= 10

0

2

4

6

8

10

12

14

Penicillin V concentration [g I"t] Fig. 9. Comparison between penicillin V concentrations in real cultivation samples determined by HPLC or the two-channel optode system, v, calibration; e, samples.

354

Fig. 10. Principle of sandwich assay. Protein A is immobilized on Sepharose support material. The IgG binds to the protein A. A fusion protein of protein A and /3-galactosidase binds to the IgG and finally forms the sandwich.

galactose by the fusion protein. The glucose can be detected sensitively in the enzyme thermistor via the heat generated during the enzymatic conversion of the glucose (Danielsson, 1990; Hundeck et al., 1992). In Fig. 11 the heat signals are shown as a function of the rabbit IgG concentration. The influence of MgC12 on the/3-galactosidase activity in the sample is small, but it increases the sensitivity in the range up to 100 ~g m1-1. The optimal concentration for rabbit IgG ranges from 10 to 400/zg ml- ~, showing that the assay can only be used for relatively high IgG concentrations. Thus, the TELISA system presented is useful for bioprocess monitoring where high antibody concentrations are obtained, e.g., in hybridoma cell cultivation. Before the next analysis cycle was started, the IgG and the fusion protein were eluted. The valve shown in Fig. 3 had to be switched back, and elution buffer was 70. "G

60-

o

so 40~

_~

I

30-

E

20100 0,0

0,2

0,4

016 rabbit IgG [mg

• no MgCI 2

018

1,0

1,2

ml "1]

[] lmM MgCI 2

Fig. 11. Temperature signals as a function of IgG concentration. Carrier buffers with and without MgCIz were used.

355 100

_

M

80 o o.

E

60-

40 20 ¸

// []

0,0

0,5

13---

1,0

1,5

2,0

2,5

mouse IgG [mg ml "1] • IgG 2a

o IgG 2b

Fig. 12. Calibration curves obtained for two subclasses of mouse IgGs. The affinity of the IgG 2b subclass to SpA is low in comparison to the affinity to IgG 2a.

pumped through the system. Over 50 analysis cycles could be performed with this TELISA system. The standard deviation was in the range of 4 - 5%. In principle, this assay system can be used for the detection of all kinds of IgGs with an appropriate binding affinity to protein A. The affinity might change between the various types and subclasses of these immunoglobulins. These differences become obvious in Fig. 12, where two subclasses of mouse IgG were tested. Valuable results were obtained only for the IgG subclass 2a, while the detection of subclass 2b caused problems.

Conclusions

These two examples of FIA-based biosensor systems are intended for the use in bioprocess monitoring. The simultaneous analysis of different analytes provides a detailed insight into the sample composition and increases the sensor reliability. More global systems, such as the immunodetection assays based on protein A for the analysis of IgG, offer the possibility of using such sensors in a wide field of application, such as IgG production processes in general. However, each biosensor system must be tested thoroughly for its special area of application for monitoring bioprocesses.

Acknowledgements

The authors gratefully acknowledge the financial support of the Deutsche Forschungsgemeinschaft.

356

References Brand, U., Brandes, L., Koch, V., Kullick, T., Reinhardt, B., Rfither, F., Scheper, T., Schiigerl, K., Wang, S., Wu, X., Ferretti, R., Prasad, S. and Wilhelm, D. (1991a) Monitoring and control of biotechnological production processes by Bio-FET-FIA-sensors. Appl. Microbiol. Biotechnol. 36, 167-172. Brand, U., Reinhardt, B., Riither, F., Scheper, T. and Schiigerl, K. (1991b) Bio-field-effect transistors for process control in biotechnology. Sensors and Actuators B 4, 315-318. Chung, C.T. and Miller, R.M. (1988) A rapid and convenient method for the preparation and storage of competent bacterial cells. Nucl. Acids Res. 16, 3580-3583. Danielsson, B. and Mosbach, K. (1987) In: A.P.F. Turner, I. Karube and G.S. Wilson (Eds.) Biosensors: Fundamentals and Applications. Oxford Science Publications, pp. 575-596. Danielsson, B., Flygare, L. and Velev, T. (1989) Biothermal analysis performed in organic solvents. Anal. Lett. 22, 1417-1428. Danielsson, B. (1990) Calorimetric biosensors. J. Biotechnol. 15, 187-200. Danielsson, B. and Larson, P.-O. (1990) Specific monitoring of chromatographic procedures. Trends Anal. Chem. 9, 223-227. Hellebust, H., Murby, M., Abrahamsen, L., Uhlen, M. and Enfors, S.-O. (1989) Different approaches to stabilize a recombinant fusion protein. Bio/Technology 7, 165-168. Hundeck, H.G., Hiibner, U., Lfibbert, A., Scheper, T., Schmidt, J., WeiB, M. and Schubert, F. (1990) Development and application of a four-channel enzyme thermistor system for bioprocess control. In: F. Scheller and R.D. Schmid (Eds.), Biosensors: Fundamentals, technologies and applications. GBF Monogr. 17, VCH, Weinheim, Germany, pp. 322-330. Kittsteiner-Eberle, R., Ogbomo, I. and Schmidt, H.-L. (1989) Biosensing devices for the semi-automated control of dehydrogenase substrates in fermentations. Biosensors 4, 75-85. Mattiasson, B. (1977) A general enzyme thermistor based on specific reversible immobilization using the antigen-antibody interaction. FEBS Lett. 77, 107-116. M611er, J., Niehoff, J., Hotop, S., Dors, M. and Schiigerl, K. (1992a) Influence of the preculture on the process performance of the penicillin V production in a 100 I airlift tower loop reactor. Appl. Microbiol. Biotechnol. 37, 157-163. M611er, J., Niehoff, J., Dors, M., Hiddessen, R. and Schfigerl, K. (1992b) Influence of the medium composition on the penicillin V production in a stirred tank reactor. J. Biotechnol. 25, 245-259. Ogbomo, I., Kittsteiner-Eberle, R., Englbrecht, U., Prinzing, U., Danzer, J. and Schmidt, H.-L. (1991) Flow-injection systems for the determination of oxidoreductase substrates: applications in food quality control and process monitoring. Anal. Chim. Acta 249, 137-143. Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning, 2nd edition. Cold Spring Harbor Laboratory, NY. Scheller, F., Schubert, F., Pfeiffer, D., Hintsche, R., Dransfeld, I., Renneberg, R., Wollenberger, U., Riedel, K., Pavlova, M., Kiihn, M. and Miiller, H.G. (1989) Research and development of biosensors: a review. Analyst 114, 653-662. Schmid, R.D., Ed. (1990) Flow injection analysis (FIA) based on enzymes or antibodies. GBF Monogr. 14, VCH, Weinheim, Germany. Seitz, W.R. (1984) Chemical sensors based on fiber optics. Anal. Chem. 56, 16A-34A. Seitz, W.R. (1988) Chemical sensors based on immobilized indicators and fiber optics. Crit. Rev. Chem. 19(2), 135-173. Thavarungkul, P., H~kanson, H., Hoist, O. and Mattiasson, B. (1991) Continuous monitoring of urea in blood during dialysis. Biosensors & Bioelectronics 6, 101-107. Wolfbeis, O.S. (1986) Analytical chemistry with optical sensors. Fresenius' Z. Anal. Chem. 325, 387-392. Wolfbeis, O.S. (1989) Optochemical sensors (optodes) and their applications. Kontakte (Darmstadt) 2, 30-34. Zhujun, Z., Zbang, Y., Wangbai, M., Russel, R. and Shakhsher, Z.M. (1989) Poly(vinyl alcohol) as a substrate for indicator immobilization for fiber-optic chemical sensors. Anal. Chem. 61,202-205.