Microbial biosensor array with transport mutants of Escherichia coliK12 for the simultaneous determination of mono-and disaccharides

Michael Held a, Wolfgang Schuhmann b, Knut Jahreis c, Hanns-Ludwig Schmidt a,*a Lehrstuhl fur Biologische Chemie, TU Munchen, Vottingerstrassc 40, D�/85350 Freising, Germany

b Lehrstuhl fur Analytische Chemie, Elektroanalytik & Sensorik, Ruhr-Universitat Bochum, D-44780 Bochum, Germanyc Lehrstuhl fur Genetik, Universitat Osnabruck, D-49069 Osnabruck, Germany

Received 2 January 2002; accepted 26 March 2002

Dedicated to Prof. Frieder Scheller on the occasion of his 60th birthday.

Abstract

An automated flow-injection system with an integrated biosensor array using bacterial cells for the selective and simultaneous

determination various mono- and disaccharides is described. The selectivity of the individually addressable sensors of the array was

achieved by the combination of the metabolic response, measured as the O2 consumption, of bacterial mutants of Escherichia coli

K12 lacking different transport systems for individual carbohydrates. k-Carrageenan was used as immobilization matrix for

entrapment of the bacterial cells in front of 6 individually addressable working electrodes of a screen-printed sensor array. The local

consumption of molecular oxygen caused by the metabolic activity of the immobilized cells was amperometrically determined at the

underlying screen-printed gold electrodes at a working potential of �/600 mV vs. Ag/AgCl. Addition of mono- or disaccharides for

which functional transport systems exist in the used transport mutant strains of E. coli K12 leads to an enhanced metabolic activity

of the immobilized bacterial cells and to a concomitant depletion of oxygen at the electrode. Parallel determination of fructose,

glucose, and sucrose was performed demonstrating the high selectivity of the proposed analytical system. # 2002 Published by

Elsevier Science B.V.

Keywords: Whole-cell biosensor; Bacterial biosensors; Flow-injection analysis; Sensor array; Transport mutants; Sugar analysis

1. Introduction

The great demand for the identification and determi-

nation of different sugars in various media, e.g. blood,

urine, beverages, nutrients has led to the development of

a variety of analytical methods. Among those, biosen-

sors have found considerable interest, mainly due to the

inherent selectivity of the used biological recognition

element. Specific enzymes are most frequently used as

selectivity elements in biosensors and corresponding

devices are excellent tools for the determination of a

given monosaccharide even in the presence of disacchar-

ides. Attempts for the parallel determination of different

sugars using coupled enzymes in specifically adapted

flow-injection systems were in principle successful (Og-

bomo et al., 1991; Schmidt et al., 1996) either using

enzymatically catalyzed consumption of the monosac-

charide prior to the determination of the respective

disaccharide or by subtraction of the monosaccharide

concentration from the sum of the total sugar signal

(Schuhmann, 1991; Becker and Schmidt, 2000). How-

ever, the direct selective determination of individual

sugars in complex mixtures using biosensor systems

remains a challenging task.

In recent years, the development of whole-cell bio-

sensors has found increasing interest on the one hand

due to the possibility of whole cells to convert complex

substrates using specific metabolic pathways (Riedel et

al., 1989; Bousse, 1996; D’Souza, 2001), and on the

other hand due to potential applications of whole-cell

biosensors for the monitoring of typical sum para-

meters, which cannot be monitored using enzyme-based

sensors, such as toxicity (Campanella et al., 1997; Evans

et al., 1998), biological oxygen demand (Riedel et al.,* Corresponding author. Address: Prielhofweg 2, 84036, Landshut,

Germany. Tel./fax: �/49-871-44-497

Biosensors and Bioelectronics 17 (2002) 1089�/1094

www.elsevier.com/locate/bios

0956-5663/02/$ - see front matter # 2002 Published by Elsevier Science B.V.

PII: S 0 9 5 6 - 5 6 6 3 ( 0 2 ) 0 0 1 0 3 - 3

1990; Li and Tan, 1994), nitrification inhibitors (Konig

et al., 1998), xenobiotic compounds (Beyersdorf-Radeck

et al., 1998), or heavy metals (Corbisier et al., 1999). The

opportunity to module metabolic activities of specificcells can additionally be used for drug screening (Wu et

al., 2001) and combinatorial approaches for drug

discovery (Durick and Negulescu, 2001). Additionally,

cell-based biosensors were successfully applied for the

specific determination of single compounds such as e.g.

glucose (Katrlik et al., 1996), fructose (Heim et al.,

1999), xylose (Reshetilov et al., 1997), or alcohols

(Reshetilov et al., 2001; Gonchar et al., 1998). A generaladvantage of these biosensors is, that living cells are

continuously repairing their integrated enzyme activities

and enzyme cascades, without any doubt an advantage

with respect to an improved long-term stability of labile

biological recognition elements.

In this communication, two aspects concerning the

improvement of the selectivity of cell-based biosensors

are introduced. On the one hand, mutants of Escherichia

coli K12 lacking specific properties in their carbohy-

drate transport systems are used as biological recogni-

tion element. On the other hand, an array configuration

of six independently addressable cell-based biosensors

with different selectivity towards selected sugar compo-

nents is used and provided the opportunity to conceive

an automatic flow-injection system for the simultaneous

determination of different mono- and disaccharides.

2. Experimental

2.1. Cell culture

Strains of E. coli K12 with defects in their carbohy-

drate transport systems have been isolated as described

elsewhere (Lengeler et al., 1981). Table 1 shows the used

mutant strains and their individual properties to use

different sugars as energy source.The bacterial cells were grown in a medium contain-

ing 10 g peptone, 5 g yeast extract and 5 g NaCl in 1 l of

distilled-deionized water (LB0 medium) with 1% (w /v ) of

the mono- or disaccharide for which the function of the

carbohydrate transport system should be induced. The

inoculated medium was incubated in a shaker bath at

29 8C for about 20 h. After this time the cells showed

maximum induction for the substrate in question andthey were in a stationary phase.

The cells were harvested by centrifugation (5000 rpm,

12 min) and washed in saline (0.9 % (w /v ) NaCl). The

obtained cell pellet was re-suspended in a volume of

saline that was equivalent to 1% of the initial volume of

the culture medium. The number of alive cells in the

culture medium was determined by successive dilution

and cultivation on LB0 agar. A concentration between2�/1010 and 8�/1010 cells ml�1, from where re-suspen-

sion with 1�/109 and 4�/109 cells ml�1 were obtained.

Cells not used immediately could be stored at 4 8C for

at least 1 week.

2.2. Immobilization of cells on the electrode surface

The cells were immobilized in front of the electrodes

in small ‘pockets’ (volume 3�/4 ml) that had been cut into

a 1-mm thick teflon sheet placed onto the electrode

array (Fig. 1). k-Carrageenan (2% (w /v )) was dissolved

in saline at a temperature of 70�/80 8C. An aliquot of 50

ml of this solution was allowed to cool down to

Table 1

Transport mutant strains of E. coli K12 used for the construction of cell-based biosensors (� and � represents the presence and the absence of the

related carbohydrate transport system)

E. coli K12 strain Glucose Fructose Galactose Lactose Sucrose Maltose

LLR220 � � � � � �JWL184 � � � � � �LR2-168 � � � � � �LLR101 � � � � � �KJ167 � � � � � �KJ175 � � � � � �S136 � � � � � �

Fig. 1. Scheme of the electrode array and the spacer forming the

immobilization pockets in front of the electrodes (A). The thick-film

transducer consists of 15 gold electrodes which can be individually

contacted. In order to prevent large iR-drops in the flow channel, each

working electrode is in close proximity of an overall large counter-

electrode. Schematic representation of the biosensor integrated in the

flow-through electrochemical cell (B).

M. Held et al. / Biosensors and Bioelectronics 17 (2002) 1089�/10941090

approximately 40�/50 8C in a pipette tip before mixing

it with 50 ml of the E. coli suspension prepared as

described above. The mixture was immediately pipetted

into the pocket in front of that platinized electrodewhich should exhibit the sugar selectivity imposed by the

chosen mutant strain. The number of alive cells was thus

between 1.5�/109 and 6�/109 cells per electrode pocket.

Gel formation immediately started and the gel

strength was increased by cooling at 4 8C and subse-

quent contact with the buffer containing 0.1 M KCl

(adapted after Chibata et al., 1987). The electrodes were

in general continuously tested in the flow system.Alternatively, they were stored in LB0 buffer in the

refrigerator at 4 8C.

2.3. Electrode configuration and flow-injection system

The electrode array consists of 14 small (1.5�/2 mm2)

and one larger (2�/7 mm2) gold electrodes arranged in a

U-form on an alumina plate (Scholze et al., 1991; Popp

et al., 1995). Six of the smaller electrodes were activated

for the reduction of O2 by platinization (Fig. 1A). Forthis, O2 was removed by Argon bubbling from a

solution containing 4 mg ml�1 H2PtCl6 in 0.1 M KCl,

which was pumped through the flow-through electro-

chemical cell. Platinization was achieved by three

potential cycles between �/500 and �/400 mV vs. Ag/

AgCl with a sweep rate of 10 mV s�1 following a

previously described procedure (Schuhmann, 1998).

A schematic diagram of the whole flow system isshown in Fig. 2 The buffer (9.9 g K2HPO4, 3.6 g

KH2PO4, 2 g (NH4)2SO4, 7.44 g KCl in 1 l of water, pH

7.0) or sample solution was pumped through an air-

bubble trap to the flow-through electrochemical cell

containing the electrode array and finally to waste.

Each platinized electrode was addressed as working

electrode by one of the channels of a home-built six-

channel potentiostat with the working potential set to �/

600 mV vs. Ag/AgCl. The other, non-platinized gold

electrodes served as counter electrodes. The change in

current after addition of mono- or disaccharide solution

was recorded by means of a PC using a 14-bit AD-card.

Simultaneously, the magnetic snap-valves were actuated

by the PC via a TTL-compatible digital IO-board and arelais board.

Carbohydrate samples (2�/4 mM in LB0 buffer) were

usually injected for a time interval of 2 min, and sample

injection was repeated after about 30 min, a delay time

which was in general sufficient to reach again the

stationary background current

3. Results and discussion

3.1. Response of cell-based biosensors to carbohydrate

injection

In the presence of a ‘recognized’ nutrient (i.e. a mono-

or disaccharide for which the cells posses a functioning

carbohydrate transport system) the cells which were

grown in a minimal medium immediately start to

metabolize the nutrient using molecular oxygen as

terminal electron acceptor. Thus, the oxygen concentra-tion in the diffusion zone of the electrode is depleted,

leading to a decrease of the diffusion-limited oxygen

reduction current, as a measure for the metabolism

activity of the immobilized cells.

Fig. 3 shows a typical response obtained for an

injection of a carbohydrate solution for 2 min. As an

example, the response obtained with the mutant E. coli

K12 LR2-168 for lactose is shown, but similar responseswere obtained using other mono- or disaccharides with

other strains of E. coli K12 (not shown). The O2-

reduction current drops within about 5 min after a

change from pure buffer to a carbohydrate-containing

solution and a current shift from a background value to

a new stationary response is observed, representing the

metabolic activity of the immobilized cells in the

presence of the nutrient. When the carbohydrate solu-tion is replaced by buffer (after 2 min), the signal

recovers and reaches the background level about 15�/20

min later. The time course of the signal is not only

dependent on the substrate concentration but also on

substrate and oxygen diffusion to the sensor surface as

Fig. 2. Schematic set up of the flow-injection system with integrated

whole-cell biosensor array.

Fig. 3. Response curve obtained by a 2 min injection of 2 mM lactose

at a sensor with immobilized E. coli K12 LR2-168.

M. Held et al. / Biosensors and Bioelectronics 17 (2002) 1089�/1094 1091

well as on transport and metabolization of the substrate

within the cells. The observed tailing of the peak-type

response is due to the fact that substrate uptake is a fast

process while complete metabolization of the substrateneeds more time (Lengeler, 1993). Since the nutrient is

not supplied continuously but only for a certain time (2

min), the slow onset and the tailing may be explained by

an overlay effect of the slow diffusional mass transport

into the immobilization layer and the time necessary for

the total oxidation of the nutrient in the cell by the

complex cell metabolism. Background and substrate-

proportional signals are reproducible and the signalheight (i.e. the degree of oxygen depletion) is a function

of the concentration of the carbohydrate (Section 3.2)

and of the duration of the injection of the mono- or

disaccharide solution (Fig. 4).

In general, a linear relationship between the carbohy-

drate injection time and signal height is obtained for a

nutrient contact of the immobilized cells of up to 4 min,

approaching a steady-state maximum value for aninjection time of 10 min and longer.

3.2. Selectivity of the sensor

With respect to a selective determination of carbohy-

drates in various complex mixtures of mono- and

disaccharides, a biosensor array was conceived using

different transport mutant strains of E. coli K12. To

demonstrate the principle feasibility of this approach wehave selected those mutants which posses carbohydrate

transport systems for only one carbohydrate out of the

mixture.

In order to measure the individual concentrations of

the individual sugars from a mixture of an aqueous

solution containing glucose, fructose and sucrose, the

strains E. coli K12 KJ167, E. coli K12 LR2-168 and E.

coli K12 KJ175 were selected. They were grown in LB0

medium in the presence of 1% (w /v ) of glucose, fructose,

or sucrose, respectively, in order to induce the relevant

transport system. Cells of each of the three strains were

then immobilized in front of different working electro-

des of the sensor array, and solutions containing one of

the three sugars were pumped through the flow-through

electrochemical cell.The results shown in Fig. 5 unequivocally demon-

strate the high selectivity imposed by the specific

features of the used mutant strains. Each of the three

sensors only generates the signals for one of the three

sugars in correspondence with the carbohydrate trans-

port system of the cell strains. In order to investigate

mixtures of different sugars strains with specific carbo-

hydrate transport systems have to be selected (Table 1).Simultaneous determinations of glucose (K12 JWL184)

and lactose (K12LR2-168), of glucose (K12 JWL184)

and galactose (K12S136), of lactose and galactose, and

of glucose, galactose and lactose were successfully

performed (data not shown). Mixtures of different

carbohydrates and of carbohydrates together with

potentially interfering compounds have not yet been

tested so far.

3.3. Quantitative determination of sugar concentrations

and functional stability of cell-based biosensors

The signal obtained with the cell-based biosensors islinearly increasing with the concentration the related

mono- or disaccharide. A typical calibration curve is

shown in Fig. 6 for the determination of maltose using

immobilized E. coli LLR101 that was grown in a

medium containing 1% maltose. The linear range is up

to about 4 mM for maltose and other disaccharides, and

up to about 2.5 mM for monosaccharides (data not

shown).The coefficient of variation in a series of six measure-

ments of a given sample is typically 2�/3%. For example,

the coefficient of variation (n�/6) for the determination

of a 3 mM glucose solution using E. coli K12 JWL184 is

2.1%.

As the analytical system is based on the metabolic

activity of living cells, sterilization is not possible. Thus,

after about 20�/30 h foreign micro-organisms growing in

Fig. 4. Dependence of the sensor response on the injection time of the

sample solution. Measurements of 1 mM lactose at a sensor with

immobilized E. coli K12 LR2-168.

Fig. 5. Measurements with E. coli K12 KJ167 (top), LR2-168 (center)

and KJ175 (bottom). Injections of glucose, fructose, and sucrose at

concentrations of 1.5 and 3.0 mM.

M. Held et al. / Biosensors and Bioelectronics 17 (2002) 1089�/10941092

the tubing of the flow-injection system increasingly

interfere with the measurements by consuming the

nutrients in the sample solution. In order to overcome

these problems sodium azide was added to the sample

solution in a concentration as low as 0.02 mg ml�1. In

preliminary experiments this concentration did not show

any influence on the response characteristics of the cell-

based biosensors (most probably due to the short

intermittent contact of the sensors with the sample

solution) but significantly reduced bacterial growth in

the sample solutions and within the tubings of the flow

system for about 3 days. For the evaluation of the

functional stability of the whole-cell biosensors, freshly

prepared sensor arrays were integrated in the flow

system, and a sample containing a known concentration

of a substrate was injected periodically.

Sample injection was repeated every 80 min for an

interval of 2 min. In Fig. 7 the operational stability is

displayed for the determination of 3 mM fructose using

a sensor based on E. coli K12 JWL184. The signal

shows an increase of the signal during the first day of

operation (�/20 measurements) and a continuous slow

decrease during the following 5 days of continuous

operation. Obviously, the living cells retain their meta-

bolic activity for at least 6 days when they are in

intermittent contact with an appropriate nutrient.

4. Conclusions

A significant improvement of selectivity for the

simultaneous determination of mono- and disaccharidesin mixtures can be achieved by the application of whole-

cell biosensors using mutants of bacterial cells which

lack components of specific carbohydrate transport

systems in combination with an array configuration of

six independently addressable sensors. The selective

determination of various mono- and disaccharides

with different transport mutant strains of E. coli K12

was demonstrated allowing finally the simultaneousdetermination of glucose, fructose, and sucrose using

three sensors with different cell strains. The presented

approach may be extended to more complicated mix-

tures by selecting additional transport mutants and a

mathematical treatment (e.g. fuzzy logic and/or neuro-

nal networks) of the signals obtained from sensors based

on bacterial strains with overlapping specificity.

Acknowledgements

The authors are grateful to Prof. Dr Norbert Hampp

and Dr Anton Silber, University of Marburg, FRG for

the generous gift of the thick-film multi-sensor arrays.

References

Becker, T.M., Schmidt, H.-L., 2000. New ways of enzymatic two-

substrate determinations in flow injection systems. Anal. Chim.

Acta 421, 7�/18.

Beyersdorf-Radeck, B., Riedel, K., Karlson, U., Bachmann, T.T.,

Schmid, R.D., 1998. Screening of xenobiotic compounds degrading

microorganisms using biosensor techniques. Microbiol. Res. 153,

239�/245.

Bousse, L., 1996. Whole cell biosensors. Sens. Actuat. B*/Chemical

34, 270�/275.

Campanella, L., Favero, G., Mastrofini, D., Tomassetti, M., 1997.

Further developments in toxicity cell biosensors. Sens. Actuat. B*/

Chemical 44, 279�/285.

Corbisier, P., van der Lelie, D., Borremans, B., Provoost, A., de

Lorenzo, V., Brown, N.L., Lloyd, J.R., Hobman, J.L., Csoregi, E.,

Johansson, G., Mattiasson, B., 1999. Whole cell- and protein-based

biosensors for the detection of bioavailable heavy metals in

environmental samples. Anal. Chim. Acta 387, 235�/244.

D’Souza, S.F., 2001. Microbial biosensors. Biosensors Bioelectron. 16,

337�/353.

Durick, K., Negulescu, P., 2001. Cellular biosensors for drug

discovery. Biosensors Bioelectron. 16, 587�/592.

Evans, M.R., Jordinson, G.M., Rawson, D.M., Rogerson, J.G., 1998.

Biosensors for the measurement of toxicity of wastewaters to

activated sludge. Pesticide Sci. 54, 447�/452.

Gonchar, M.V., Maidan, M.N., Moroz, O.M., Woodward, J.R.,

Sibirny, A.A., 1998. Microbial O2- and H2O2-electrode sensors

for alcohol assays based on the use of permeabilized mutant yeast

cells as the sensitive bioelements. Biosensors Bioelectron. 13, 945�/

952.

Fig. 6. Calibration curve for maltose lactose at a sensor with

immobilized E. coli K12 LLR101.

Fig. 7. Functional stability of the flow-injection system with integrated

whole-cell biosensors. Injection of 3 mM fructose at time intervals of

80 min at a sensor with immobilized E. coli K12 JWL184.

M. Held et al. / Biosensors and Bioelectronics 17 (2002) 1089�/1094 1093

Heim, S., Schnieder, I., Binz, D., Vogel, A., Bilitewski, U., 1999.

Development of an automated microbial sensor system. Biosensors

Bioelectron. 14, 187�/193.

Katrlik, J., Svorc, J., Rosenberg, M., Miertus, S., 1996. Whole cell

amperometric biosensor based on Aspergillus niger for determina-

tion of glucose with enhanced upper linearity limit. Anal. Chim.

Acta 331, 225�/232.

Konig, A., Riedel, K., Metzger, J.W., 1998. A microbial sensor for

detecting inhibitors of nitrification in wastewater. Biosensors

Bioelectron. 13, 869�/874.

Lengeler, J., Auburger, A.-M., Mayer, R., Pecher, A., 1981. Mol. Gen.

Genet. 183, 163.

Lengeler, J.W., 1993. Antonie van Leeuwenhoek 63, 275.

Li, F., Tan, T.C., 1994. Monitoring BOD in the presence of heavy

metal ions using a poly(4-vinylpyridine)-coated microbial sensor.

Biosensors Bioelectron. 9, 445�/455.

Ogbomo, I., Kittsteiner-Eberle, R., Englbrecht, U., Prinzing, U.,

Danzer, J., 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.

Popp, J., Silber, A., Brauchle, C., Hampp, N., 1995. Sandwich enzyme

membranes for amperometric multi-biosensor applications*/im-

provement of linearity and reduction of chemical cross-talk.

Biosensors Bioelectron. 10, 243�/249.

Reshetilov, A.N., Iliasov, P.V., Donova, M.V., Dovbnya, D.V.,

Boronin, A.M., Leathers, T.D., Greene, R.V., 1997. Evaluation

of a Gluconobacter oxydans whole cell biosensor for amperometric

detection of xylose. Biosensors Bioelectron. 12, 241�/247.

Reshetilov, A.N., Trotsenko, J.A., Morozova, N.O., Iliasov, P.V.,

Ashin, V.V., 2001. Characteristics of Gluconobacter oxydans B-

1280 and Pichia methanolica MN4 cell based biosensors for

detection of ethanol. Proc. Biochem. 36, 1015�/1020.

Riedel, K., Renneberg, R., Wollenberger, U., Kaiser, G., Scheller,

F.W., 1989. Microbial sensors: fundamentals and application for

process control. J. Chem. Technol. Biotechnol. 44, 85�/106.

Riedel, K., Lange, K.P., Stein, H.J., Kuhn, M., Ott, P., Scheller, F.,

1990. A microbial sensor for BOD. Water Res. 24, 883.

Schmidt, H.-L., Becker, T., Ogbomo, I., Schuhmann, W., 1996. Flow-

injection analysis systems with immobilized enzymes. Improvement

of applicability by integration of coupled reactions, separation

steps and background correction. Talanta 43, 937�/942.

Scholze, J., Hampp, N., Brauchle, C., 1991. Enzymatic hybrid

biosensors. Sens. Actuat. B*/Chemical 4, 211�/215.

Schuhmann, W., 1991. Amperometric substrate determination in flow-

injection systems with polypyrrole-enzyme electrodes. Sens. Actuat.

B*/Chemical 4, 41�/49.

Schuhmann, W., 1998. Methods in biotechnology, vol. 6: enzyme and

microbial biosensors: techniques and protocols. In: Mulchandani,

A., Rogers, K. (Eds.), Enzyme Biosensors Based on Conducting

Polymers. Humana Press, Totowa, pp. 143�/156.

Wu, Y.C., Wang, P., Ye, X.S., Zhang, G.Y., He, H.Q., Yan, W.M.,

Zheng, X.X., Han, J.H., Cui, D.F., 2001. Drug evaluations using a

novel microphysiometer based on cell-based biosensors. Sens.

Actuat. B*/Chemical 80, 215�/221.

M. Held et al. / Biosensors and Bioelectronics 17 (2002) 1089�/10941094