Biosensors & Bioelectronics 16 (2001) 467– 480

www.elsevier.com/locate/bios

The application of the bioelectric recognition assay for the

detection of human and plant viruses: definition of operational
parameters
S. Kintzios a,*, E. Pistola a, J. Konstas a, F. Bem b, Th. Matakiadis a,
N. Alexandropoulos c, I. Biselis d, R. Levin e
a
Laboratory of Plant Physiology, Agricultural Uni6ersity of Athens, 75 Iera Odos, 11855 Athens, Greece
b
Benaki Phytophathological Institute, 14561 Athens, Greece
c
Microbiology Department, Hippokration Hospital, 11527 Athens, Greece
d
Laboratory of Animal Husbandry, Agricultural Uni6ersity of Athens, 11855 Athens, Greece
e
Osmotek Ltd., PO Box 550, Kiryat Weizmann, Reho6ot 76120, Israel

Abstract

The bioelectric recognition assay (BERA) is a novel biosensory method based on a unique combination of a group of cells, their
immobilization in a matrix that preserves their physiological functions and the expression of the cell interaction with viruses as
a change in electrical properties. A BERA sensor consists of an electroconductive, tube-like probe containing components of
immobilized cells in a gel matrix. Cells are selected to specifically interact with the virus under detection. In this way, when a
positive sample is added to the probe, a characteristic, ‘signature-like’ change in electrical potential occurs upon contact between
the virus and the gel matrix. In the present study, we demonstrate that BERA can be used for the detection of viruses in humans
(hepatitis C virus) and plants (tobacco and cucumber viruses) in a remarkably specific, rapid (1 – 2 min), reproducible and
cost-efficient fashion. The sensitivity of the virus detection with BERA (0.1 ng) is equal or even better than with advanced
immunological, cytological and molecular techniques, such as the reverse transcription polymerase chain reaction. Moreover, a
good storability of the sensors can be achieved without affecting their performance. The potential use of portable BERA
biosensors in medicine, for mass screening purposes, as well as for the detection of biological warfare agents without prior
knowledge of a specific receptor-molecule interaction is discussed. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Biosensor; Cell immobilization; Electrophysiological response; Virus; Hepatitis C; Plant pathogenic

1. Introduction quinone (Rawson et al., 1989; Marty et al., 1998).

A biosensor can be defined as a device incorporating
a biological sensing element connected to a transducer
(Eggins, 1996). In recent years there has been a rapid
increase in the number of diagnostic applications based
on biosensors including live, intact cells. Cell-based
biosensors used for the detection of environmental pol-
lutants (commonly herbicides), are usually based on the
immobilization of microorganisms on a Clark oxygen
electrode or they include an amperometric detection of
an electron mediator such as ferrycyanide or p-benzo-
* Corresponding author. Tel.: + 30-1-529-4292/202-7041; fax: +
30-1-529-4286.
E-mail address: skin@aua.gr, spiroskintzios@usa.net (S. Kintzios).
Cell-based biosensors are likely to have improved sta-
bility, higher biocatalytic activity and low cost in their
favour; however, they have poor selectivity and can
suffer from long recovery times compared to enzymic
biosensors. In addition, the need for biosensors with a
shelf life of weeks or months precludes the use of fresh
cells requiring time-consuming harvesting and
preparation.
Cellular responses to various compounds under de-
tection can be evaluated by measuring the cellular
electric properties, in particular the electric potential,
which reflects changes of a network of inter-related
metabolic reactions. The interior of animal and plant
cells is electrically negative with respect to the exterior.

0956-5663/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 5 6 - 5 6 6 3 ( 0 1 ) 0 0 1 6 1 - 0
468 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
The magnitude of this potential difference is generally
between 5 and 90 mV (in some cases 100– 500 mV)
(Tsong and Astumian, 1988), with most of the potential
being developed across the membrane (Shapiro, 1982).
Thus, the cell membrane is polarized between its layers,
having a nominal capacitance of 1 mF/cm2 and a nomi-
nal resistance of 104 Siemens/cm2. Rapid changes in the
electrical potential and ion flux across cellular mem-
branes occur within minutes of specific ligand binding
to transmembrane receptors (Jaffe, 1976). Iwata et al.
(1999) measured membrane currents using the whole-
cell patch clamp technique to study effects of rabies
virus infection on ion channels in mouse neuroblastoma
cells. By using the same technique, Wang et al. (1994)
demonstrated that the influenza A virus M2 protein
had an ion channel activity in mammalian cells. In
addition, potentials can result from changes of the lipid
composition of the cell membrane in response to
changes in the proximate cellular environment (‘home-
ophasic adaptation responses’) (Vigh et al., 1998). In
the absence of any interaction, cells retain a rest poten-
tial, i.e. a time average potential that is sustained by
metabolic energy through ion pumps (e.g. H+ -AT-
Pases) or electron transport chains.
Membrane interactions are commonly studied in
their electrophysiological aspect by patch clamp tech-
niques, which rely on the fact that the ionic flow across
a cell membrane can be measured as an electrical
current if the membrane potential is held constant
(Orwar and Jardemark, 1998; Zhang and Hamill, 2000).
A cell or part of the cell is firmly attached by suction to
the tip of a glass microelectrode, and a highly sensitive
feedback current-to-voltage converter measures sub-pi-
coampere currents. Shapiro (1982) developed a method
for detecting changes in the membrane potential of
individual cells, after incubation with a membrane-per-
meant, fluorescent ionic dye that becomes distributed
on opposite sides of the cell membrane as a function of
membrane potential. Rosenthal and Shapiro (1983)
used this method in order to rapidly assay the binding
of Epstein-Barr virus on receptor-bearing B
lymphocytes. They observed that changes in the mem-
brane potential differed in time course and direction
with respect to the cell’s capacity to internalize the virus
and the cell type. Kovacs and Borkholder (1998)
demonstrated the effective use of electrodes small
enough (10 mM diameter) to study single cell/electrode
coupling and monitored cellular viability and position
by impedance measurement. Since the immobilization
of a single neuron to the gate of a FET by Fromherz et
al. (1991) (which allowed for the first direct bioelec-
tronic signal transfer) biosensors carrying olfactory re-
ceptor proteins have been used for odor detection as
electronic noses. Some recent applications rely on the
use of whole, intact organs (in particular insect anten-
nae) for the detection of volatile compounds by record-
ing electric signals produced thereof (Koch et al., 1989;
Schutz et al., 1999; Schroth et al., 1999).
However, the measurement of the membrane poten-
tial, membrane conductance and membrane electromo-
tive force across a cell surface is usually complicated,
due the zoning effects and the ‘cable property’ (Ogata
et al., 1983; Smith, 1983), i.e. when current is applied
via external electrodes on a cell membrane, an apprecia-
ble error in the measurement of the change in mem-
brane voltage occurs, due to the effect of additional
membrane-area specific conductance effects. The paral-
lel impedance of the uncovered electrode area and the
impedance of cell– cell contact areas also interfere with
the accurate measurement of an individual cell’s
impedance characteristics. By using secondary screens,
several hours or days may be required for electrophysi-
ologically assaying compounds interacting with integral
membrane proteins, such as receptors and ion channels.
It would, therefore, be desirable to develop a system
for assaying the cell potential not on an isolated mem-
brane region but rather as a result of processes associ-
ated with the whole cellular metabolism and preferably
on a tissue level. We have previously reported (Kintzios
et al., 2000) on the development of a novel method for
the qualitative and/or quantitative detection of biologi-
cal and chemical compounds on the basis of their
specific interaction with appropriately immobilized, vi-
able cells and the measurement of the change of the
electric potential that is caused by the aforementioned
interaction. For this purpose, we constructed a biosen-
sor comprising of an electroconductive gel matrix with
a cellular sensory material (the input transducer) and an
appropriate microelectrode/data acquisition system (the
output transducer). In the present report we demon-
strate that the bioelectric recognition assay (BERA) can
be used for the qualitative and quantitative detection of
viruses in humans (hepatitis C virus) and plants (to-
bacco and cucumber viruses) in a remarkably specific
and reproducible fashion. The sensitivity of the virus
detection with BERA is equal or even better than with
advanced immunological, cytological and molecular
techniques, such as the polymerase chain reaction
(PCR), which have also been used in our study.
2. Experimental
2.1. Biosensor system and working principle
Each biosensor comprised of cells immobilized in a
matrix of appropriate material and in such a way that
the viability and functional integrity of the sensory
material was preserved, along with its specific mode of
interaction with the agent(s) under detection. The cellu-
lar sensory material was selected to (i) respond specifi-
cally to the compound under detection, and (ii) express
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480 469

this specific response via changes of the electric
potential.
Cells and electrodes were arranged in the sensor in
the main configuration, which is schematically repre-
sented in Figs. 1 and 2. The sensor consisted of an
electroconductive, tube-like probe made of polypropy-
lene (4×0.8 cm2) and containing components of immo-
bilized cells in a gel matrix. A matrix volume of 1.28 ml
was obtained. A 50 ml syringe was used for applying the
sample. The measuring electrode laid in the area of
sample application (top of the probe, 3 mm below the
matrix fill line) while the reference electrode laid in the
opposite position (3 mm from the bottom of the probe).
The electrodes were made from pure silver, electro-
chemically coated with an AgCl layer, having a diame-
ter of 0.2 mm and were inserted into the tube for 7 mm.
Both electrodes were connected via coaxial cable to the
recording device, which comprised an Advantec PCL-
711TM PC I/O card. The Analog-to-Digital Converter
(ADC) of this computer card was recording the signal
(voltage). The ADC was a 16-bit, 5-scale unipolar/bipo-
lar converter. To the shortest scale (+ /− 325 mV
bipolar voltage) the ADC can accomplish an accuracy
of 0.01 mV (accuracy= range of measurement (325×
2+ 1)/216 (bit analysis of the ADC)=0.009918 or 0.01
mV). The software responsible for the recording of the
signal and processing of data was a modified version of
the Advantec GenieTM v3.0.
At the beginning of each assay, and prior to sample
application, a minor, non-zero potential exists between
the electrodes, possibly due to differences in the cellular
microenvironment (for example, immobilized cells at
the are of the measurement electrode are in closer
contact with air, thus an increased respiration can be

Fig. 1. (a) Main outline of the BERA-biosensor configuration. (b) Working principle of the sensor: (A) Before sample application (B) After
application of a positive sample (Vo, sensor rest potential, V(t), biosensor response as a function of time).
470 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480

Fig. 2. (a) Equivalent electric circuit of the BERA matrix-cell sensor system. 1, the measuring electrode; 2, the reference electrode, Gi,j,
conductance and Ci,j, capacitance of each immobilized cell (i= 1…n and j = 1…m), R, resistance of the matrix (gel and pores with solution). (b)
Outline of dual biosensor system for cancelling out the effects of various properties of the sample solution.

observed). In addition, the data acquisition circuit is the moment of sample application to the top of the
configured in such a way that the reference electrode is probe (Fig. 1(b), point A), the compounds under detec-
maintained at a virtual ground potential. In theory, at tion will interact with the part of the sensory cellular
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
material in the area of the measuring electrode, causing
a change of its electric properties. At the same time,
however, the part of the sensory cellular in the area of
the reference electrode will retain the initial value of
these properties (rest potential). In this way, an electric
potential will be created between the two electrodes. As
the sample constituents will progressively move towards
the bottom, they will react with increasingly more parts
of the cellular material so that a change in electrical
potential is occurring over time until a new ‘steady-
state’ value is obtained (Fig. 1(b), point B).
From another point of view, the BERA matrix-cell
sensor system can be considered as a series of n cell/ma-
trix layers acting both as bipolars and as capacitors
whereas the potential varies in relation to the total
impedance between the measuring electrode (Fig. 2(a),
point 1) and the reference electrode (Fig. 2(a), point 2).
Each immobilized cell has a conductance Gi,j and a
capacitance Ci,j (where i =1…n and j= 1…m) depend-
ing on its position within the probe. In addition, and in
particular at higher gel densities, there is an additional
resistance R of the matrix (gel and pores with solution).
Thus, the sensor resembles an RC circuit consisting of
a group of n capacitors serially connected to each other:
upon sample application, the initially reacting part of
the sensor (the first cell/matrix layers or ‘capacitors’) is
electrically charged until a certain maximum value is
achieved. The equivalent capacitative time constant of
the sensor is characteristic and specific for each bio-
chemical agent. Thus, a given virus demonstrates a
unique pattern of biosensor response over a definite
range of concentrations, like a ‘signature’. From a
practical point of 6iew, the response of the biosensor
should be plotted against a series of dilutions of the
sample under investigation. A second biosensor bearing
cellular material with a different sensitivity should be
used in order to cancel out the possible presence of
other compounds demonstrating similar response pat-
terns. In similar fashion, the effects of various proper-
ties of the sample solution, such as pH, conductivity
and ionic strength, on the sensor response were can-
celled out by using a suitable reference solution in a
dual biosensor system (Fig. 2(b)). It is obvious, how-
ever, that results can vary with minor changes in the
biosensor configuration, e.g. the distance between the
two electrodes.
2.2. Procedure
2.2.1. Detection of the hepatitis C 6irus
Biosensors were constructed under sterile conditions
by immobilizing liver cells (a porcine cell line) in vitro
selected to have a specific response against the hepatitis
C virus (HCV). After cell detachment from the culture
vessel by adding trypsine/EDTA for 10 min at 37 °C
and cell concentration by centrifugation (6 min, 1200
471
rpm, 25 °C), cells (at a density of 4× 106/ml) were
mixed with a 1% (w/v) low melting point agarose
solution in PBS-Dulbecco’s medium. This was done at
37 °C in order to avoid agarose solidification. The
cell-agarose mixture was transferred into an appropri-
ately configured box bearing Ag/AgCl electrodes, as
described in Section 2.1.
2.2.2. Detection of plant pathogenic 6iruses
2.2.2.1. Tobacco biosensors. Biosensors were con-
structed under sterile conditions by immobilizing proto-
plasts of the tobacco (Nicotiana tabacum L.) cultivar
‘Samsun’, which is susceptible to Tobacco rattle 6irus
(TRV) but resistant against Cucumber green mottle
mosaic 6irus (CGMMV).
Protoplasts were isolated from tobacco leaves by
preplasmolysing 0.5 g of them in 20 ml of CPW solu-
tion (Reinert and Yeoman, 1982) supplemented with
0.7 M mannitol for 1 h and then incubating them in 20
ml solution of the same composition and additionally
supplemented with 3 mg pectinase (8.5 units/mg, from
Aspergilus niger) and 2 mg cellulase (9.5 units/mg, from
Trichoderma 6iridae) for 20 h. One microlitre of proto-
plast solution (at a density of 6× 106 cells/ml) was
centrifuged at 14.000 rpm (20 min, 25 °C) and the
pellet was resolved and mixed with a 1% (w/v) low
melting point agarose solution in water. This was done
at 37 °C in order to avoid agarose solidification. The
cell-agarose mixture was transferred into an appropri-
ately configured box bearing Ag/AgCl electrodes, as
described in Section 2.1.
In addition, and in order to study the effect of
immobilized cell concentration on the sensor response,
tobacco biosensors bearing either 3× 106 or 6 ×106
cells/ml were constructed and used in the virus assay.
2.2.2.2. Cucumber biosensors. According to the proce-
dure described above, a set of biosensors was con-
structed by immobilizing protoplasts (at a density of
3× 106 cells/ml) of the cucumber (Cucumis sati6us L.)
cultivar ‘20202-G’, which is susceptible against both
viruses (CGMMV and TRV) used in this study.
2.3. Assay
2.3.1. Detection of the hepatitis C 6irus
Twenty microlitre of negative controls (30 individual
HCV-free blood samples) or HCV-positive sample (30
individual clinical blood samples) were applied. Sam-
ples were clinically obtained from patients at the Hip-
pokration Hospital (Athens, Greece) without revealing
their identity prior to assaying. In some cases, control
and sample blood was infected with other kind of
viruses, such as the hepatitis B virus (analytical data
not shown). The samples were assayed for HCV with
472 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
reverse transcription polymerase chain reaction (RT-
PCR), as essentially described by Garcia et al. (2000).
2.3.2. Effect of an interferon (Betaferon ®) on the
biosensor response
In an additional experiment, 20 ml of an interferon
solution (Betaferon, Schering AG) (29 million units/mg)
has been administered to liver cell-biosensors 30 min
after the application of blood samples from infected
patients and the sensor response was measured.
Betaferon is a genetically engineered, modified human
interferon beta with significant antiviral activity
(Anonymous, 1995). Prior to application, the protein
(0.75% w/v) was diluted in a 0.54% (w/v) sodium
chloride solution.
2.3.3. Detection of plant pathogenic 6iruses
2.3.3.1. Virus purification. Virus isolates were purified
from infected Nicotiana tabacum cv. Samsun leaves (a
Greek isolate of TRV) or infected cucumber (Cucumis
sati6us) leaves (the watermelon isolate PL104 of CG-
MMV), as described previously (Lister and Bracker,
1969; Tung and Knight, 1972; Bem, 1987; Bem and
Vassilakos, 1996; Brown et al., 1996).
2.3.3.2. Sample application. (a) Tobacco sensors: For the
quantitative determination of the CGMMV or TRV
virus in each assay, gradually increasing virus concen-
trations (40 ml of buffer containing 1, 3, 6.6, 9, 13.3, 20,
30, 40, 60, 70, 80, 90 and 100 ng purified virus) and the
control solution (phosphate buffer pH 7.4) were
applied.
In addition, we tested the response of the biosensors
against samples containing both CGMMV and TRV
viruses, at a quantity of either 9 or 20 ng.
(b) Cucumber sensors: For the quantitative determi-
nation of the CGMMV virus in each assay, gradually
increasing virus concentrations (40 ml of buffer contain-
ing 1, 3, 6.6, 9 and 13.3 ng purified virus) and the
control solution (phosphate buffer pH 7.4) were
applied.
(c) Effect of immobilized cell density: In order to
study the effect of immobilized cell concentration on
the tobacco sensor response, biosensors bearing either
3 × 106, 6×106 or 12 ×106 cells/ml were tested against
CGMMV or TRV virus solutions (with virus quantities
of 6.6, 13.3, 20 and 60 ng)
2.3.3.3. Assay of crude sap extracts. Crude sap extracts
were prepared from cucumber plants infected with CG-
MMV by homogenizing leaves in phosphate buffer (pH
7.2) (1 g leaf tissue: 10 ml buffer). Infected leaves were
removed from plants at 2, 3, 4, 6 and 11 days after
inoculation. The extracts were assayed for CGMMV
with tobacco sensors prepared as in Section 2.2.2.1, as
well as with ELISA and PCR, as described in follow-
ing: F(ab%)2-ELISA tests for CGMMV were done as
described by Ploeg et al. (1992) using specific antiserum
against the PL104 isolate of the virus. Total plant RNA
was isolated according to Verwoerd et al. (1989). RT-
PCR for CGMMV was performed essentially as de-
scribed by MacFarlane (1996).
2.3.3.4. Cytological assays. Membrane potential
changes were monitored by the cellular uptake of a
lipophilic cationic fluorochrome, the distribution of
which is affected by the potential difference across the
cell membrane, as described by Rosenthal and Shapiro
(1983). Briefly, plant (tobacco and cucumber) proto-
plasts were incubated in 3 ml of PBS buffer (pH 7.4,
106 cells/ml) and stained with 3 mg of 3,3 dipropylthiad-
icarbocyanide iodide. Samples were analysed before
and after the addition of 40 ml buffer containing various
quantities of either CGMMV or TRV (6.6, 13.3, 20 and
60 ng), after excitation at 580 nm. Fluorescence emis-
sion at 688–690 nm was measured with a Jasco FP-920
Fluorescence detector.
2.4. Storage of the biosensors
In order to test the effect of storage conditions on the
performance of BERA biosensors, the following experi-
ments were carried out:
(a) Biosensors which have been prepared as described
above (Sections 2.2.1 and 2.2.2) were stored for 2
months at −5 °C and then used for virus assays in
exactly the same manner as in Sections 2.3.1 and 2.3.3.
(b) Biosensors were also prepared as described above
(Sections 2.2.1 and 2.2.2) but cells were immobilized in
a modified procedure, which involved covalent bonding
of membrane fragments to the gel matrix (analytical
protocol not described). The biosensors were stored for
2 months at room temperature and then used for virus
assays in exactly the same manner as in Sections 2.3.1
and 2.3.3.
2.5. Chemicals
All solvents and chemicals used were of analytical
quality (from Sigma Company, St. Louis). Water was
double distilled.
2.6. Experimental design
Experiments were set up in a randomized complete
block design and each experiment was repeated three
times. In each application, a set of 15 biosensors was
tested against each individual sample. Different series
of measurements were compared by means of Student’s
t-test.
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480 473

3. Results and (b). There was a significant and fully reproducible

difference between the liver BERA sensor response to
3.1. Detection of the hepatitis C 6irus the HCV-positive samples and the control, even if
control and/or sample blood were infected with other
The results of the detection are presented in Fig. 3(a) kinds of viruses (Fig. 3(a)). The virus elicited a devia

Fig. 3. (a) Results of the detection of the HCV in a human blood sample using a BERA-sensor based on porcine liver cells. The original recording
of the biosensor system is presented. The arrow indicates the moment of the application of the virus. Dashed line indicates biosensor response to
HCV-negative sample (control). (b) Comparative response of a BERA-sensor based on porcine liver cells against the HCV in a human blood
sample before and after the treatment of the sensor with Betaferon®. (1, control (non-infected sample); 2, HCV-infected sample; 3, HCV-infected
sensor treatment with Betaferon®; 4, control treatment with Betaferon®). Variations in the response against the infected samples (bar 2) may be
due to quantitative differences between samples (mean values, 15 biosensors, n =45).
474 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
Fig. 4. Results of the response of a tobacco BERA biosensor against the TRV virus (3 ng). The original recording of the biosensor system is
presented. The arrow indicates the moment of the application of the virus.
tion of the biosensor potential from the average rest
value of +4 (control) to + 9 mV (sample) (Fig. 3(b),
bar 2).
3.2. Effect of an interferon (Betaferon ®) on the
biosensor response
On the contrary, the addition of Betaferon® to
biosensors previously (30 min) treated with HCV-posi-
tive samples caused a reduction of the rest potential to
− 6 mV (Fig. 3(b), bar 3), possibly associated with
inhibition of virus replication. The addition of
Betaferon® to non-infected biosensors caused an in-
crease of the potential to 12 mV, indicating an elicita-
tion of cellular response similar to viral infection (Fig.
3(b), bar 4). Application of a non-functional placebo
(0.54% sodium chloride solution) did not have any
effect.
3.3. Detection of plant pathogenic 6iruses
3.3.1. Tobacco sensors
A remarkable difference was observed between the
sensor response to different viruses and the control.
The biosensor response to each virus solution was
estimated as the difference of the sensor potential be-
fore and after sample application (Fig. 4). For both
viruses, this response generally declined with increasing
virus concentration. However, at low concentrations,
the tobacco pathogenic strain TRV elicited a higher
response than the non-pathogenic strain CGMMV,
while at concentrations higher than 0.5 ng/ml this effect
was reversed. In every assay, measurements taken by 45
different biosensors did not vary more than 0.00001%
of the average response i.e. independent measurements
of a virus concentration were virtually identical. Inter-
estingly, at 0.5 ng/ml both viruses caused a very low
biosensor response (Fig. 5(a)). Therefore, the response
did not increase with virus concentration in a linear
fashion. If, however, recorded data were logarithmically
(− 1/log − 1) transformed, then a negative linear corre-
lation could be observed between the sensor response
and virus quantities in the range of 1–70 ng (r 2 = −
0.97132, TRV) or 50–100 ng (r 2 = − 0.9492, CG-
MMV). However, when we tested the biosensors
against samples containing both CGMMV and TRV
viruses (at a quantity of either 9 or 20 ng) the observed
response was virtually identical to the response against
only TRV, which is 6irulent on ‘Samsun’ tobacco (ana-
lytical results are not shown).
Under the experimental conditions described in the
present study, a detection limit of  0.01 ng/ml was
achieved for both viruses.
3.3.2. Cucumber sensors
The cucumber protoplast biosensor response against
TRV (i.e. the difference of the sensor potential before
and after sample application) generally declined with
increasing virus quantities (Fig. 5(b)), however, a mini-
mum was observed at 6.66 ng. A similar minimum,
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480 475

within the quantity range of 1– 13.3 ng has been ob-
served in the response of the tobacco sensor against
TRV (Fig. 5(a)). As for the tobacco sensors, the vari-
ability of virus detection with BERA did not exceed
10 − 5% of the median value i.e. independent measure-
ments of a virus concentration were virtually identical.
3.3.3. Effect of immobilized tobacco cell density
Increasing the density of immobilized tobacco proto-
plasts from 3 to 12 million cells/ml caused varying
changes of the biosensor response against CGMMV or
TRV (Fig. 6(a) and (b)). Therefore, no linear relation-
ship between cell density and the biosensor response
against either virus could be observed.
3.3.4. Assay of crude sap extracts
The time course study of cucumber plants inoculated
with CGMMV was conducted on a daily basis for a
total period of 24 days after inoculation.
(a) By assaying the crude sap extract of cucumber
plants infected with CGMMV with tobacco BERA
biosensors we have been able to detect the presence of
the virus even on the second day after inoculation.
(b) Virus was detected by F(ab%)2-ELISA only after
the sixteenth day after inoculation.
(c) No virus was detected by RT-PCR in RNA
extracted from leaves of inoculated plants, sampled in
days 2, 3, 4, 6 and 11 after inoculation. It is possible

Fig. 5. Dose-response curve of (a) the tobacco biosensor to the CGMMV virus (non-virulent on tobacco) or the TRV virus (virulent) (b) the
cucumber biosensor to the CGMMV virus (virulent on cucumber) (mean values, 15 biosensors, n = 45). Error bars represent a variability in
response B0.00001%.
476 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480

Fig. 6. Dose-response curve of the tobacco biosensor to (a) the CGMMV virus (non-virulent on tobacco) (b) the TRV virus (virulent on tobacco)
in relationship to different immobilized cell densities (open boxes: 3 ×106 cells/ml, shaded boxes: 6 × 106 cells/ml, solid boxes: 12 × 106 cells/ml)
(mean values, 15 biosensors, n=45).

that virus concentration during this period was below
the detection threshold of the method. Alternatively,
this could be due to a possible uneven distribution of
the virus in the plant.
3.4. Assay of plant cell membrane potential with a
fluorescent dye
The addition of either CGMMV or TRV virus to a
plant protoplast solution caused an instant change of
the cell membrane potential. For cucumber, this change
never exceeded 15% of the control value (cells untreated
with virus). Increasing CGMMV quantity from 6.6 to
13.3 ng caused a slight increase of the potential, but
higher virus concentrations did not (Fig. 7(a)). Tobacco
cell membrane potential fluctuated with virus concen-
trations, however, an increasing trend was observed
towards higher concentrations (Fig. 7(b)). Any
change, however, did not exceed 1125% of the control
values.
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480 477

3.5. Effect of storage on biosensor performance 4. Discussion

All assays were also carried out with biosensors
which either (a) contained viable cells and were stored
for 2 months at −5 °C or (b) contained covalently
bonded cells and were stored for 2 months at room
temperature. In every case, the observed pattern of
response was identical to the pattern obtained with
freshly prepared sensors (analytical results not shown).
Therefore, although the actual half-life of the biosen-
sors has not been estimated, their storage in either way
did not affect the replicability of measurements.
In the present study, we investigated the potential
application of the BERA biosensory system for the
detection of viruses in humans (HCV) and plants (CG-
MMV, TRV). For this purpose, BERA sensors were
constructed consisting of an electroconductive probe
containing immobilized cells in an agarose matrix. Cells
were selected to specifically interact with the virus
under detection. When a positive sample was added to
the probe, a change in electrical potential was measured
upon contact between the virus and the gel matrix.

Fig. 7. Changes of the membrane potential (expressed as relative fluorescence) of (a) tobacco and (b) cucumber protoplasts incubated in a
fluorescent dye (3,3 dipropylthiadicarbocyanide iodide) before and after the addition of CGMMV or TRV virus. Increasing treatment numbers
(1 –4) indicate increasing virus quantities (6.6, 13.3, 20 and 60 ng).
478 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
Although based on a rather simple concept, BERA
could be used in a remarkably specific and reproducible
fashion. The sensitivity of the virus detection with
BERA is equal to advanced techniques, such as the
PCR, which have also been used in the present study.
The results of the assay with a fluorescent dye demon-
strated that rapid changes of the plant cell potential are
indeed associated with virus interactions. In contrast to
BERA, however, this method was impractical and not
suitable for quantitative determinations.
There already exist a considerable number of biosen-
sory methods based on electrophysiological effects (see
the Section 1 for more details). Some of them use cells,
or assay electric properties, or use immobilized compo-
nents. However, no other method has been previously
reported to combine these three elements. In addition,
most of the electrophysiologically operating biosensors
include microelectrodes in direct contact with a single
cell, which makes the setup impractical and expensive.
Contrary to those methods, a BERA sensor comprises
electrodes that are inserted into a matrix bearing a
group of cells. In this way, the sensor represents, more
or less, the natural tissue where the interaction takes
place. Although BERA is still in the stage of develop-
ment, we feel that the results are encouraging and this
approach is advantageous or can efficiently comple-
ment immunological or molecular methods, since:
(I) Irrespective of the field of desired application, it is
possible to find or isolate (by appropriate means of
selection) cells with a differential response to different
chemical or biological compounds, so that the desired
specificity is created.
(II) The construction and operation of BERA biosen-
sors is quite cost-efficient.
(III) The biosensor is portable and the assay rapid
(almost instant), which makes it suitable for mass
screening purposes, as well as for the detection of
biological warfare agents.
(IV) Immobilized cells can retain unalterable their
properties for a comparatively long time (\2– 3
months), at least under the conditions described in this
study.
(V) The measured response to a virus is closely linked
to its mechanism of virulence, particularly when whole,
intact cells are used as the biosensory material.
(VI) The method has an outstanding repeatability
(variability B 10 − 5). Most biosensory methods usually
have a variability of at least 5% between measurements.
As a biosensory method, BERA could be efficiently
used in drawing a first conclusion on the presence of a
given virus in an unknown sample. A particularly valu-
able application of the present method in medicine is
the possible detection of viruses and other microorgan-
isms before the host develops antibodies against them.
The detection of these antibodies is the basis of stan-
dard diagnostic methods (such as ELISA and immuno-
histochemical methods). However, the time required for
antibody production may be considerably long (e.g.
several months), during which infection will remain
undetected. For example, detection of the HCV in
blood by means of an antibody assay is not possible
before 15–24 weeks have elapsed since infection (Aach
et al., 1991). In addition, identification is particularly
troublesome because available antibodies mainly bind
to non-specific regions on the protein envelope of the
virus. The application of such assays is further limited
by the variability due to differences in reagent specific-
ity between different manufacturers or testing laborato-
ries, as well as the non-specific elevation of antibodies
due to other causes (e.g. chronic liver disease, liver
cirrhosis and hepatocellular carcinoma) (Anonymous,
1989; Houghton et al., 1991). The amount of HCV
genome in infected patients is too low to be detected by
conventional hybridization techniques and the applica-
tion of more advanced techniques such as RT-PCR
(Dash et al., 2000) is necessary. The sensitivity of
RT-PCR is : 0.1 ng per assay (Markoulatos et al.,
1999; Dash et al., 2000), which is equal to the virus titer
detectable by BERA in our experiments. However,
commercially available PCR tests are usually expensive
(which prohibits their use on mass sample screening)
and the method often requires separate laboratory
space and equipment in order to avoid sample contam-
ination with exogenous viral DNA (Weiner et al., 1990;
Chan et al., 1991; Romeo et al., 1993; Maier, 1995).
Contrary to PCR, which detects viral DNA or RNA
sequences, BERA assays the virus in its complete,
virulent form. In this way, an accurate correlation can
be established between the assay and the actual status
of infection and virus replication in the host, providing
important information on the viral load during anti-vi-
ral therapy. BERA could also complement the mass
detection of virulent compounds, enabling simplified
tests by the lay population and in developing countries.
Applications also include screening new vaccines and
pharmaceuticals, such as using liver cell biosensors in
order to study receptor antagonists used in pharmaceu-
ticals meant to affect the liver.
Further work is in progress in order to improve the
standardization of the method (in particular the biosen-
sory material) and to properly evaluate the reliability,
by performing a very large number of tests in a wide
variety of environmental conditions. There is a need to
improve our understanding of the fundamental pro-
cesses taking place during cell-virus interactions within
the probe. This interaction has a number of compo-
nents, which can be classified in three groups: (i) inter-
action with the cell membrane (ii) interaction with the
whole cellular physiology and (iii) interaction with
other parts of the probe, such as the gel and the
solution gathered in to the matrix pores. In addition, it
is necessary to specify the various resistance and capac-
 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
itance components present into the probe, before and
after the sample application, to define the components
of the signal and factor out noise (which was rather
high, :1 mV) from various sources, such as the thin,
low impedance Ag/AgCl microelectrodes (increased in-
terference noise) or unshielded components of the ac-
quisition circuit. The signal-to-noise ratio was not
improved even though we used a high resolution, high
impedance computer interfaced data acquisition card
and sophisticated software for the recording of the
signal and processing of data. Noise could have been
reduced either by pre-amplification of the signal, or by
signal frequency modulation; this was not done, how-
ever, because we wanted to monitor the actual magni-
tude of changes of extremely rapid variations in electric
characteristics of the cell in real time. By applying
high-intensity (0.5–1.5 kV cm − 1) electric field pulses to
the immobilized sensory material, cell conductivity
(Knorr and Angersbach, 1998) could be increased, thus
increasing the sensitivity of the method. Another prob-
lem is that no prior calibration of individual sensors is
yet possible, since this might affect their performance
upon application of the sample. In addition, the re-
sponse of a BERA sensor is dependent on the concen-
tration of the agent under detection. In co-operation
with an international panel of partners, we currently
develop BERA sensors for detecting a number of
viruses such as enteroviral poliovirus, human cy-
tomegalovirus and sheep poxvirus. Our final goal is the
dissemination of BERA to other research parties so
that the scope of applications of the method will be
expanded to the possible extent.
Acknowledgements
The authors wish to acknowledge the following per-
sons: Mr P. Tzomakas (from Tzomakas & Co, Athens)
for the kind provision of fine quality silver wire for the
construction of microelectrodes, Mr P. Alekos (Scher-
ing AG) for providing us with Betaferon® and placebo
solutions, Prof Y. Clonis (Laboratory of Enzyme Tech-
nology, Agricultural University of Athens) and Dr N.
Markoulatos (Hellenic Pasteur Institute) for their ad-
vise on our work and finally C. Varveri, N. Vassilakos,
F. Georgacopoulou, A. Kataki, M. Barberaki, A.
Tsipra and P. Tsoukalas for their technical assistance.
References
Aach, R.D., Stevens, C.E., Hollinger, F.B., Mosley, J.W., Peterson,
D.A., Taylor, P.E., Johnson, R.G., Barbosa, L.H., Nemo, G.J.,
1991. Hepatitis C virus infection in posttransfusion hepatitis. An
analysis with first- and second-generation assays. New Engl. J.
Med. 325, 1325.
479
Anonymous, 1989. Enzyme Immunoassay for the quantitative mea-
surement of alpha-fetoprotein in human serum and amniotic fluid.
Abbott AFP-EIA Monoclonal, Abbott Laboratories.
Bem, F., 1987. A serious disease in tobacco fields of Pieria county,
Greece, caused by the tobacco rattle virus. Fourth National
Phytopathological Conference, pp. 49 – 50.
Bem, F., Vassilakos, N., 1996. Eighth National Phytopathological
Conference, Phytopatologia Mediterranea, pp. 49 – 50.
Brown, D.J.F., Robertson, W.M., Neilson, R., Bem, F., Robinson,
D.J., 1996. Characterization and vector relation of a serologically
distinct isolate of tobacco rattle tobravirus (TRV) transmitted by
Trichodorus similis in northern Greece. Eur. J. Plant Pathol. 102,
61 – 68.
Chan, S.W., Simmonds, P., Mcomish, F., Yap, P.L., Mitchell, R.,
Dow, B., Follett, E., 1991. Serological responses to infection with
three different types of hepatitis C. Lancet 338, 1391.
Dash, S., Saxena, R., Myung, J., Rege, T., Tsuji, H., Gaglio, P.,
Garry, R.F., Thung, S.N., Gerber, M.A., 2000. HCV RNA levels
in hepatocellular carcinomas and adjacent non-tumorous livers. J.
Virol. Meth. 90, 15 – 23.
Eggins, B., 1996. Biosensors – An Introduction. Wiley & Teubner,
Chichester.
Fromherz, P., Offenhausser, A., Vetter, T., Weiss, J., 1991. A neuron-
silicon junction: a retzius cell of the leech on an insulated-gate
field-effect transistor. Science 252, 1290.
Garcia, F. Jr., Garcia, F., Roldan, C., Lopez, I., Martinez, N.M.,
Alvarez, M., Bernal, M.C., Hernandez, J., Maroto, M.C., 2000.
Detection of HCV and GBV-CHGV RNA in peripheral blood
mononuclear cells of patients with chronic type C hepatitis.
Microbios 103, 7 – 15.
Houghton, M., Wiener, A., Han, J., Kuo, G., Choo, Q.L., 1991.
Molecular biology of the hepatitis C viruses: implications for
diagnosis, development and control of viral disease. Hepatology
14, 381 – 388.
Iwata, M., Komori, S., Unno, T., Minamoto, N., Ohashi, H., 1999.
Modification of membrane currents in mouse neuroblastoma cells
following infection with rabies virus. Brit. J. Pharmacol. 126,
1691 – 1698.
Jaffe, L.A., 1976. Fast block to polyspermy in sea urchin eggs is
electrically mediated. Nature 261, 68 – 71.
Kintzios, S., Pistola, E., Panagiotopoulos, P., Bomsel, M., Alexan-
dropoulos, N., Bem, F., Varveri, C., Ekonomou, G., Biselis, J.,
Levin, R., 2000. Bioelectric recognition assay (BERA). Sixth
World Congress on Biosensors, San Diego, USA.
Knorr, D., Angersbach, A., 1998. Impact of high-intensity field pulses
on plant membrane permeabilization. Trends Food Sci. Tech. 9,
185 – 191.
Koch, U.T., de Kramer, J.J., Milli, R., Sauer, A., 1989. Anordnung
zur Registrierung von Elektroantennogrammen unter Freilandbe-
dingungen. DE-3906004.
Kovacs, T.A., Borkholder, D.A., 1998. Microelectronic biosensor and
method of monitoring cultured cells. PCT application Nr. WO
98/54294.
Lister, R.M., Bracker, C.E., 1969. Virology 37, 262.
MacFarlane, S.A., 1996. J. Virol. Meth. 56, 91 – 98.
Maier, K.P., 1995. Hepatitis, fourth ed. Georg Thieme Verlag,
Stuttgart.
Markoulatos, P., Samara, V., Siafakas, N., Plakokefalos, E., Spyrou,
N., Moncany, M.L.J., 1999. Development of a quadriplex poly-
merase chain reaction for human cytomegalovirus detection. J.
Clin. Lab. Anal. 13, 99 – 105.
Marty, J.L., Leca, B., Noguer, T., 1998. Biosensors for the detection
of pesticides. Anal. Magazine 26, 144 – 149.
Ogata, K., Chilcott, T.C., Coster, H.G.L., 1983. Spatial variation of
the electrical properties of Chara australis. I. Electrical potentials
and membrane conductance. Aust. J. Plant Physiol. 10, 339.
480 S. Kintzios et al. / Biosensors & Bioelectronics 16 (2001) 467–480
Orwar, O., Jardemark, K., 1998. Detection of biologically active
molecules by use of pre-activated cell-based biosensors in liquid-
based separation systems. PCT application WO 98/55870.
Ploeg, A.T., Brown, D.J.F., Robinson, D.J., 1992. The association
between species of Trichodorus and Paratrichodorus vector ne-
matodes and serotypes of tobacco rattle tobravirus. Ann. Appl.
Biol. 121, 619 – 630.
Rawson, D.M., Willmer, A.J., Turner, A.P.F., 1989. Whole-Cell
biosensors for environmental monitoring. Biosensors 4, 299 – 311.
Reinert, J., Yeoman, M., 1982. Plant Cell Tissue Culture. Springer,
Berlin.
Romeo, J.M., Ulrich, P., Busch, M.P., Vyas, G.N., 1993. Analysis of
hepatitis C virus RNA prevalence and surrogate markers of
infection among seropositive voluntary blood donors. Hepatology
17, 188.
Rosenthal, K.S., Shapiro, H.M., 1983. Cell membrane potential
changes follow Epstein-Barr virus binding. J. Cell. Physiol. 17,
39– 42.
Schroth, P., Schoening, M.J., Kordos, P., Lueth, H., Schutz, S.,
Weissbecker, B., Hummel, H.E., 1999. Insect-based BioFETs with
improved signal characteristics. Biosensors Bioelectron. 14, 303 –
308.
Schutz, S., Weissbecker, B., Koch, U.T., Hummel, H.E., 1999. Detec-
tion of volatiles by diseased potato tubers using a biosensor on the
basis of intact insect antennae. Biosensors Bioelectron. 14, 221 – 228.
Shapiro, H.M., 1982. Cytological assay procedure. US Patent
4.343.782.
Smith, J.R., 1983. The electrical properties of plant cell membranes:
cable properties involving the plasmalemma, tonoplast and cyto-
plasm in Chara. Aust. J. Plant Physiol. 10, 329.
Tsong, T.Y., Astumian, R.D., 1988. Electroconformational coupling:
how membrane-bound ATPase transduces energy from dynamic
electric fields. Ann. Rev. Physiol. 50, 273.
Tung, J.S., Knight, C.A., 1972. Virology 48, 574.
Verwoerd, T.C., Dekker, B.M.M., Hoekema, A., 1989. Nucleic Acids
Res. 17, 2362.
Vigh, L., Maresca, B., Harwood, J.L., 1998. Does the membrane’s
physical state control the expression of heat shock and other genes?
TIBS 23, 369.
Wang, C., Lamb, R.A., Pinto, L.H., 1994. Direct measurement of the
influenza A virus M2 protein ion channel activity in mammalian
cells. Virology 205, 133 – 140.
Weiner, A.J., Kuo, G., Bradley, D.W., Bonino, F., Saracco, G., Lee,
C., Rosenblatt, J., Choo, Q.L., Houghton, M., 1990. Detection of
hepatitis C viral sequences in non-A, non-B hepatitis. Lancet 335,
1 – 3.
Zhang, Y., Hamill, O.P., 2000. On the discrepancy between whole-cell
and membrane patch mechanosensitivity in Xenopus oocytes. J.
Physiol. 15, 101 – 115.