Automated Water Analyser Computer Supported System (AWACSS)

Automated Water Analyser Computer Supported System (AWACSS)
Biosensors and Bioelectronics 20 (2005) 1509–1519
Automated Water Analyser Computer Supported System (AWACSS)
Part II: Intelligent, remote-controlled, cost-effective, on-line,
water-monitoring measurement system
Jens Tschmelaka,∗ , Guenther Prolla , Johannes Riedta , Joachim Kaiserb , Peter Kraemmerb , Luis
Bárzagac , James S. Wilkinsond , Ping Huad , J. Patrick Holed , Richard Nudde , Michael Jacksone ,
Ram Abukneshaf , Damià Barcelóg , Sara Rodriguez-Mozazg , Maria J. López de Aldag , Frank
Sacherh , Jan Stienh , Jaroslav Slobodnı́ki , Peter Oswaldi , Helena Kozmenkoi , Eva Korenkovái ,
Lı́via Tóthováj , Zoltan Krascsenitsj , Guenter Gauglitza
a
Institute of Physical and Theoretical Chemistry (IPTC), Eberhard-Karls-University of Tuebingen, Auf der Morgenstelle 8,
72076 Tuebingen, Germany
b Siemens AG, CT PS 6, Paul-Gossen-Str. 100, 91050 Erlangen, Germany
c Siemens AG, CT SM ICA, Otto-Hahn-Ring 6, 81730 Munich, Germany
d Optoelectronics Research Centre, Southampton University, Highfield, Southampton SO17 1BJ, UK
e Central Research Laboratories Limited, Dawley Road, Hayes, Middlesex UB3 1HH, UK
f Division of Life Sciences, King’s College London, 150 Stamford Street, London SE1 9NN, UK
g IIQAB, Department of Environmental Chemistry, CID-CSIC, c/Jordi Girona 18, 08034 Barcelona, Spain
h DVGW-Technologiezentrum Wasser, Karlsruher Str. 84, 76139 Karlsruhe, Germany
i Environmental Institute, Okružná 784/42, 97241 Kos, Slovak Republic
j Water Research Institute, Nábr. arm. gen. L. Svobodu 5, 81249 Bratislava, Slovak Republic
Received 21 May 2004; received in revised form 16 July 2004; accepted 22 July 2004
Available online 26 August 2004
Abstract
A novel analytical system AWACSS (Automated Water Analyser Computer Supported System) based on immunochemical technology has
been evaluated that can measure several organic pollutants at low nanogram per litre level in a single few-minutes analysis without any prior
sample pre-concentration or pre-treatment steps.
Having in mind actual needs of water-sector managers related to the implementation of the Drinking Water Directive (DWD) [98/83/EC,
1998. Council Directive (98/83/EC) of 3 November 1998 relating to the quality of water intended for human consumption. Off. J. Eur.
Commun. L330, 32–54] and Water Framework Directive (WFD) [2000/60/EC, 2000. Directive 2000/60/EC of the European Parliament and
of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Off. J. Eur. Commun. L327,
1–72], drinking, ground, surface, and waste waters were major media used for the evaluation of the system performance.
The first part article gave the reader an overview of the aims and scope of the AWACSS project as well as details about basic technology,
immunoassays, software, and networking developed and utilised within the research project. The second part reports on the system performance,
first real sample measurements, and an international collaborative trial (inter-laboratory tests) to compare the biosensor with conventional
anayltical methods. The systems’ capability for analysing a wide range of environmental organic micro-pollutants, such as modern pesticides,
endocrine disrupting compounds and pharmaceuticals in surface, ground, drinking and waste water is shown. In addition, a protocol using
∗
Corresponding author. Tel.: +49 7071 29 74668; fax: +49 7071 29 5490.
E-mail address: [email protected] (J. Tschmelak).
URL: http://barolo.ipc.uni-tuebingen.de (J. Tschmelak).
0956-5663/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.bios.2004.07.033
1510
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
reconstitution of extracts of solid samples, developed and applied for analysis of river sediments and food samples, is presented. Finally,
the overall performance of the AWACSS system in comparison to the conventional analytical techniques, which included liquid and gas
chromatographic systems with diode-array UV and mass spectrometric detectors, was successfully tested in an inter-laboratory collaborative
trial among six project partners.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Automated Water Analyser Computer Supported System (AWACSS); Optical immunosensor; On-line biosensor monitoring; Emerging pollutants;
Environmental analysis; Immunosensor network
1. Introduction
Industrial activities, extensive farming, sewage water,
and many other human activities are polluting waterways
throughout Europe. Rivers, channels, lakes, oceans, ground
water, and even drinking water can be contaminated by a
variety of organic pollutants that can have adverse effects
on aquatic life and pose risks to human health (Barceló and
Hennion, 1997a, 1997b). Examples include herbicides, insecticides, fungicides, antibiotics, drugs, detergents, oils, industrial waste or by-products, endocrine disrupting compounds,
and carcinogens. Some man-made substances being released
into environment are toxic to humans via food chain. The new
water-related environmental legislation in Europe clearly defines a need for regular monitoring of wide, and ever growing,
range of organic substances down to low nanogram per litre
levels (2000/60/EC, 2000; 98/83/EC, 1998). Since many of
the so-called priority substances (2000/60/EC, 2000) to be
monitored in water are rather difficult to analyse European
Union expert groups working at the implementation of the
Water Framework Directive (WFD) (2000/60/EC, 2000) put
a considerable effort to propose a range of techniques and
methods for their monitoring. True enforcement demands
more frequent monitoring of water catchment areas and also
industrial plants need greater control of their waste water to
meet the demands of increased regulation.
Currently, to analyse the different environmental pollutants, quantitative water analysis is performed with standard
analytical methods such as high-performance liquid chromatography (HPLC), or gas chromatography coupled to mass
spectroscopy (GC–MS) (Clement et al., 1997). The contaminants are invariably present in complex matrices and at low
concentrations and, therefore, efficient sample concentration and clean-up methods followed by common analytical
techniques equipped with sensitive and selective detectors
are usually techniques of choice (Barceló, 2000). Usually,
these methods need trained manpower and automation is
limited, therefore, these methods are labour intensive, expensive, and time consuming. Despite the fast developments
of these techniques over the past decades, water-sector managers and researchers still strive for a water-monitoring device, which would be robust, cost-effective, automated and
able to measure several tens of organic pollutants at low
nanogram per litre level in a short time, preferably without any time-consuming sample concentration step and prior
sample pre-treatment. Another sought for feature is a possibility of instrument’s remote control, automated data process-
ing and generation of alarm signals when pollutant’s concentration exceeds the pre-set threshold value. The use of such
on-line sensors follows also new developments in the drinking water legislation and introduction of risk assessment/risk
management approach (World Health Organization, 2003).
Immunochemically based techniques are a popular alternative to standard methods. Key features of immunoassays,
such as their specificity, sensitivity and speed of analysis
led to the development of numerous environmental applications (Barceló, 2000) and qualify them as suitable candidates to be used for environmental monitoring. Still, most of
the commercially available techniques suffer from not fully
described cross-reactivity of target analytes, matrix effects,
limited availability of antibodies and not having capability
of multi-analyte analysis (Oubina et al., 2000). Recently,
the field of array-based technology for multianalyte detection has exploded. These techniques take advantage of the
two-dimensional layout of recognition elements to allow simultaneous detection and quantification of multiple analytes
(Rowe Taitt et al., 2002).
In the EU project AWACSS (Automated Water Analyser
Computer Supported System; EVK1-CT-2000-00045), a
multi-analyte immunoassay-based system has been constructed and successfully tested for analysis of aqueous samples. The first part article gave the reader an overview of the
aims and scope of the AWACSS project as well as details
about basic technology, immunoassays, software, and networking developed and utilised within the research project.
This second part now reports on the system performance,
first real sample measurements, and an international collaborative trial (inter-laboratory tests) to compare the biosensor
with conventional anayltical methods.
2. Experimental
2.1. Materials
Common chemicals of analytical grade were purchased
from Sigma–Aldrich, Deisenhofen, Germany, or Merck
KGaA, Darmstadt, Germany. The pesticides atrazine, isoproturon, and propanil were purchased as PESTANAL® ,
the antibiotic sulphamethizole and the hormone estrone
were purchased as VETRANAL® analytical standards from
Riedl-de Haën Laborchemikalien GmbH & Co. KG, Seelze,
Germany. The polymer plasticiser bisphenol A was also ordered from Sigma–Aldrich. The fluorescent dye CyDyeTM
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
Cy5.5 was purchased from Amersham Biosciences Europe
GmbH, Fribourg, Germany. The aminodextrans AmdexTM
with 40 and 170 kDa molecular weight were purchased from
Helix Research Company, Springfield, OR, USA. Labelling
and purification of antibody were carried out as described in
the product information sheet supplied with the labelling kit
from Amersham Biosciences Europe GmbH. UV–vis spectra were recorded using a Specord M500 spectrophotometer
from Carl Zeiss Jena GmbH, Jena, Germany. The spatially
resolved surface modification was performed using a parallel
micro-dispensing system (TopSpot) with high-performance
piezostack actuation system and integrated temperature adjustment from HSG-IMIT/IMTEK, Germany.
2.2. Instrumentation
The AWACSS instrument employs fluorescence-based detection of the binding of fluorophore-tagged biomolecules to
the surface of an optical waveguide chip. The fibre-pigtailed
chip, driven by a semiconductor laser, consists of a waveguide circuit which distributes excitation light to 32 separate
sensing patches on the chip surface. Bio/immunochemistry is
used to sensitise each of the 32 patches to a specific analyte
and a micro fluidic system is used to automatically handle
the sample injection over the sensor surface, enabling rapid,
simultaneous and high-sensitivity fluorescence detection of
up to 32 pollutants. A fibre-coupled detection array is used
to monitor the 32 separate fluorescence signals, and software
has been developed for control of the optics and fluidics and
data acquisition and processing for the fluorescence signals,
laser power, and ambient and chip temperature.
An HTC PAL autosampler with cycle composer software
(CTC Analytics AG, Zwingen, Switzerland) was used for dilutions, sample preparations (transferring 100 ␮L of the antibody stock solution to 900 ␮L of the sample followed by
one or two mixing strokes) and the sample transfer to the
AWACSS instrument. Liquid handling and data acquisition
are fully automated and computer controlled. One measurement cycle with washing steps, injection of the sample and
regeneration of the surface takes about 15–18 min.
Key features of the instrument, chip fabrication, chip characterisation, hardware requirements, and further details of the
instrumentation are described in the first part article published
in a previous issue of the same journal.
2.3. Immunochemistry
The immunochemistry utilised in the project takes advantage of a binding inhibition test that requires antibodies directed against specific analytes and analyte derivatives
that can be covalently bound to a transducer surface. The
previously immobilised aminodextran layer is used to reduce non-specific binding to the surface. Dried (immobilised)
aminodextran layers on a glass substrate showed a thickness
between one and three nm. The thickness of welled aminodextran layers (aminodextran layers on a glass substrate in
1511
contact with buffer solution after a few minutes) were also
verified by spectroscopic ellipsometry experiments and these
experiments yielded values between 100 and 150 nm.
A huge number of polyclonal antibodies and their corresponding analyte derivatives have been produced for a variety of organic micro-pollutants. After being purified and
labelled with a fluorescent marker, they were developed into
immunoassays and used in the project. The sample containing
the analyte is incubated in solution with the labelled specific
antibody. Therefore, 100 ␮L of the antibody stock solution
are mixed with 900 ␮L of the sample by an autosampler and
are incubated for approximately 5 min. The antibody binds
the analyte during the incubation step until a well-defined
condition of the reaction is reached. When the sample is
pumped over the sensor surface, only the antibodies with
free binding sites can bind to the surface. For the binding inhibition assay to be quantitative, the binding of the antibody
to the surface must be mass transport-limited. This allows
the signal to be a function of the diffusion rate to the surface
and not of the kinetics of the surface binding. The number
of high affinity binding sites on the surface has to be much
higher than the number of antibodies used for one measurement. To be sure, that the binding is mass transport-limited,
we use small amounts of antibodies and on the sensor surface
we immobilise a huge excess of antigen derivatives. This was
demonstrated by additional reflectometric interference spectroscopy (RIfS) measurements as already described in literature (Glaser, 1993). The surface evaluation was performed
with covalently immobilised peptide nucleic acid (PNA)
for the detection of different endocrine disruptors by the
above mentioned label-free detection method. Within these
experiments, a hybridisation capacity with DNA oligonucleotides of 180 fmol mm−2 on PNA surfaces has been reported (Kroger et al., 2002). Other experiments to evaluate
a covalent strategy for immobilisation of DNA microspots
suitable for microarrays with label-free and time-resolved
optical detection of hybridisation resulted in hybridisation
capacities of approximately 300 fmol mm−2 (Jung et al.,
2001). We achieved linear correlations between the increasing fluorescence signal and the antibody concentration used
within TIRF experiments. The linear behavior of the fluorescence signal shows that no saturation effects can be
observed even with highest antibody concentrations. Therefore, the immobilised huge excess of antigen derivatives
in comparison to the used amounts of antibodies could be
verified.
2.4. Detection
The AWACSS instruments are based on evanescent field
technology. Laser light is coupled into an optical transducer and guided down the integrated optical (IO) chip.
The transducer surface is chemically modified in spatially
distinct loci with analyte derivatives. Analyte-specific antibodies are labelled with a fluorescent marker (CyDyeTM
Cy5.5), which upon binding to the transducer surface are
1512
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
Fig. 1. Schematic set-up of the developed sampling system for on-line monitoring which allows direct intake of river water and its continuous transfer to the
autosampler of the AWACSS system.
excited in the evanescent field. The emitted light is then
collected for detection with 32 polymer fibres. The design
allows for the simultaneous measurement of multi-analyte
spots.
Analyte recognition is based on a binding inhibition assay. Analyte derivatives are immobilised onto the transducer
surface prior to the assay. Next, analyte-specific antibodies
labelled with fluorescent markers are incubated with the analyte samples. After the short incubation period, the analyte
solution flows over the transducer. Only analyte-specific antibodies with free binding sites will bind to the transducer
surface whereas, at the same time, antibodies that have two
analyte molecules bound to each epitope will not bind to the
surface. The surface bound labelled antibodies are excited
in the evanescent field and the fluorescence is detected. As
a result, an inverse analyte signal is measured, with samples
having low analyte concentrations giving rise to high fluorescence signals and samples with high analyte concentrations
resulting in low fluorescence.
over night at room temperature. A 10-fold volume excess of
methanol precipitated the aminodextran conjugate. The supernatant was removed and the conjugate was freeze-dried.
The IO chips were cleaned in a freshly prepared mixture
(ratio 2:3) of hydrogen peroxide (30% H2 O2 ) and concentrated sulphuric acid (65% H2 SO4 ) for approximately 10 min
and rinsed with Milli-Q water. After drying under a nitrogen flow, 25 ␮L of (3-glycidyloxypropyl)trimethoxysilane
(GOPTS) were applied to the surface and reacted for up
to 60 min. The silanised surface was rinsed with dry acetone and dried under a flow of nitrogen. Subsequently, the
aminodextran conjugates were dissolved in Milli-Q water
at a concentration of 1.0–2.0 mg mL−1 and were immobilised by a parallel spotting device TopSpot from HSGIMIT, Villingen-Schwenningen, Germany (http://www.hsgimit.de) and IMTEK (Institute for Microsystem Technology),
Fribourg, Germany (http://www.imtek.de).
2.5. Immobilisation
A sampling system for on-line monitoring has been developed which allows direct intake of river water and its
continuous transfer to the autosampler of the AWACSS
system. A schematic representation of the set-up is in
Fig. 1.
Active esters were prepared with the derivatives,
which are analyte molecules modified with a spacer
containing a carboxyl group. Approximately 5.0 mg of
the derivative were dissolved in 100 ␮L of dry N,Ndimethylformamide (DMF). N-Hydroxysuccinimide (NHS)
and N,N -dicyclohexylcarbodiimide (DCC), each in 1.1-fold
molar excess (referring to the amount of analyte derivative) were added to the solution. After stirring for several
minutes, the solution was kept over night at room temperature. Finally, the solution was centrifuged (12,000 rpm)
at approximately 4 ◦ C and the supernatant was stored under refrigeration. 50 mg aminodextran were dissolved in a
mixture of 500 ␮L Milli-Q water and 500 ␮L DMF. The active ester solution was added, mixed thoroughly and kept
2.6. Sampling
2.7. Measurement
For the measurements, we used a polyclonal IgG antibody from sheep and a suitable analyte derivative. The entire sample volume was 1000 ␮L. For a calibration routine,
900 ␮L of spiked Milli-Q water was automatically mixed by
the autosampler with 100 ␮L of an antibody stock solution
containing the antibodies and ovalbumin from chicken eggs
(OVA) in 10-fold phosphate buffered saline (PBS) (10-fold
PBS: pH 6.8, 1500 mM sodium chloride, 100 mM potassium
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
phosphate monobasic). After a defined incubation time, this
mixture was measured using the biosensor set-up. The experimental design for a calibration routine consists of nine
independent blank (Milli-Q water) measurements and nine
concentration steps (each measured as three replica) of the
analyte (spiked Milli-Q water). For all concentration steps
and the blank measurements (nine replica), the mean value
and the standard deviation (S.D.) for the replica was automatically calculated by the measurement control unit. The
measured signal for the mean value of the blanks was set to
100% and all spiked samples could be obtained as a relative
signal below this blank value. To fit the data set a logistic fit
function (Dudley et al., 1985) (parameters of a logistic function: A1 , A2 , x0 , and p) with three free parameters (A2 , x0 ,
and p) was used. A1 , as the upper asymptote was fixed to 100
% (relative signal for mean value of the blanks) and A2 is the
lower asymptote. The range between A1 and A2 is the dynamic
signal range. The inflection point is given by the variable x0
and represents the analyte concentration, which corresponds
to a decrease of 50% of the dynamic signal range (IC50 ). The
slope of the tangent in this point is given by the parameter
p. In compliance with the IUPAC rules (The Orange Book)
(Inczedy et al., 1998), the LOD is calculated as 3 times the
S.D. of the blank measurements (S.D.blank ) and the LOQ is
calculated as 10 times the S.D. of the blank measurements
(S.D.blank ). All statistical procedures and further calculations
are included within the AWACSS evaluation software package.
1513
3. Results and discussion
3.1. Selection of AWACSS compounds
A set of new immunochemical reagents was developed
for selected organic micro-pollutants to be analysed in environmental water samples. Major criteria for selection of
target analytes were: (i) their presence in the environment;
(ii) existing environmental legislation (98/83/EC, 1998;
2000/60/EC, 2000); (iii) technical possibilities to prepare
sufficiently selective polyclonal antibodies and their corresponding analyte derivatives; and (iv) marketing considerations. More than 20 different polyclonal antibodies were
isolated within the project and their corresponding analyte
derivatives were synthesised. A priority was given to classes
of pesticides, endocrine disrupting compounds, WFD Priority Substances, industrial pollutants and pharmaceuticals (for
a list of AWACSS compounds, see Table 1).
3.2. Regular monitoring of AWACSS compounds in
European rivers
Sampling sites: Four project partners (IIQAB, Department
of Environmental Chemistry, CID-CSIC, Barcelona, Spain;
DVGW-Technologiezentrum Wasser, Karlsruhe, Germany;
Environmental Institute, Kos, Slovak Republic; Water Research Institute, Bratislava, Slovak Republic) of the watermonitoring group have systematically investigated various
Table 1
A list of AWACSS compounds monitored in the surface water, ground water, municipal/industrial waste water and sediment samples within the years 2001–2003
with the number of positive findings and their concentration ranges
Compounda
No.b
Concentrationc
Matrixd
Commente
Alachlor
Pyrene
Benzo[a]pyrene
Fluorene
Fluoranthene
DEHP
Bisphenol A
Nonylphenol
Benzene
Toluene
Xylene
Trichloroetylene
Atrazine
Prometryn
Ametryn
Terbuthylazine
Simazine
Benzenesulfonamide
Caffeine
1
115
73
51
118
116
49
5
11
9
7
19
21
4
1
1
2
3
15
0.11
0.05–633
0.15–36.94
0.09–267
0.02–717
1–2115
0.06–35.19
0.35–87.57
0.6–177.5
1.1–447.3
1.0–31.9
0.1–22852
0.2–4.46
0.13–1252
0.22
0.14
0.1–0.5
0.42–4.86
1.3–112
SED
SED
SED
SED
SED
SED
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SW, GW, WW
SK, WFD
SK
SK, WFD
SK
SK, WFD
SK, WFD
SK
SK, WFD
SK, WFD
SK
SK
SK
SK, WFD
SK
SK
SK
SK, WFD
SK
SK
a
b
c
d
e
Antibody and analyte derivative(s) developed or in the process of preparation.
Number of positive findings.
Concentration range in mg kg−1 for sediment samples (SED) and in ␮g L−1 for water samples (SW, GW, and WW).
SED: sediment; SW: surface water; GW: ground water; WW: waste water.
SK: Slovak Republic; WFD: compound present on the list of Water Framework Directive Priority Substances.
1514
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
Fig. 2. Characteristic pollution patterns by atrazine (a), bisphenol A (b), and sulphamethoxazole (c) at the River Rhine (Germany) by GC–MS. Sampling sites:
Mainz, Karlsruhe, and Duesseldorf. Results were obtained in the period 2001–2003.
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
1515
Table 2
Calculated relative signal values with standard deviations for all analytes and concentration steps
Concentration (␮g L−1 )
Propanil
Atrazine
Isoproturon
Sulphamethizole
Bisphenol A
Estrone
0
0.009
0.027
0.09
0.27
0.9
2.7
9
27
90
100.00 ± 1.91
96.43 ± 2.02
92.87 ± 3.38
75.69 ± 1.83
49.85 ± 1.77
26.95 ± 1.35
17.44 ± 0.36
12.41 ± 0.61
9.39 ± 0.81
7.87 ± 0.68
100.00 ± 2.37
92.15 ± 2.19
89.66 ± 3.27
71.96 ± 3.49
54.96 ± 3.69
41.15 ± 1.53
32.78 ± 1.86
27.41 ± 0.98
22.58 ± 0.91
19.15 ± 0.63
100.00 ± 2.04
96.45 ± 0.83
92.05 ± 3.96
70.47 ± 3.48
40.62 ± 2.23
18.24 ± 0.57
12.52 ± 0.27
9.85 ± 0.26
8.24 ± 0.51
7.10 ± 0.24
100.00 ± 2.79
95.55 ± 2.23
90.49 ± 2.24
80.13 ± 1.80
70.13 ± 1.32
58.67 ± 0.10
52.73 ± 1.06
45.70 ± 1.14
39.59 ± 0.88
33.91 ± 0.92
100.00 ± 0.93
98.01 ± 2.33
95.84 ± 0.87
86.75 ± 2.27
81.76 ± 2.22
70.34 ± 2.55
58.67 ± 1.54
49.31 ± 2.25
40.25 ± 1.39
34.22 ± 1.64
100.00 ± 0.56
95.59 ± 2.04
91.62 ± 2.48
56.93 ± 2.48
32.34 ± 1.58
21.69 ± 0.41
18.65 ± 0.46
16.53 ± 1.00
15.64 ± 1.56
14.63 ± 0.23
A1 (%)
A2 (%)
x0 (␮g L−1 )
p
LOD (␮g L−1 )
100
9.34 ± 0.87
0.23 ± 0.01
1.05 ± 0.04
0.019
100
20.66 ± 1.65
0.22 ± 0.02
0.74 ± 0.05
0.010
100
8.39 ± 0.52
0.17 ± 0.01
1.23 ± 0.04
0.020
100
32.88 ± 2.41
0.47 ± 0.10
0.57 ± 0.05
0.018
100
28.10 ± 2.45
1.70 ± 0.29
0.58 ± 0. 04
0.008
100
16.72 ± 1.05
0.09 ± 0.01
1.46 ± 0.12
0.007
Resulting parameters (with standard deviations) for the determined logistic fit functions and the validation parameter LOD for all analytes.
types of water around Europe in order to select the analyte
panel applicable to real-world situations and future monitoring sites for the AWACSS system. Also, water matrix and
cross-reactivity effects related to the AWACSS immunoassay
chemistry were addressed.
In Spain, sampling sites at the Llobregat River, supplying
drinking water to the city of Barcelona, two of its tributaries,
the Ebro River and the Mediterranean Sea were regularly
monitored by RIANA and LC–MS techniques. The occurrence of target analytes was monitored also in the ground
water serving as a source for abstraction of drinking water. A
removal efficiency of the detected pollutants was studied at
the successive water treatment steps applied in the Sant Joan
Despi waterworks.
The AWACSS monitoring network in Germany consisted
of three sampling sites alongside the River Rhine. Among
the techniques applied for monitoring were GC–MS and
LC–DAD UV. Typical examples of pollution patterns by
selected analytes over the years 2001–2003 are given in
Fig. 2a–c. It is shown on the example of atrazine (Fig. 2a) that
pesticide monitoring strongly depends on the seasonal variations, while pollution by industrial chemicals (bisphenol A,
Fig. 2b) and pharmaceuticals (sulphamethoxazole, Fig. 2c)
is relatively constant.
Slovak partners monitored four sampling sites on the Nitra
River: Nedozery, a reference site upstream of major pollution
sources; Prievidza, at the outlet of the municipal waste water
treatment plant; Novaky, at the outlet of a discharge channel of large chemical industry and Chalmova, approximately
6 km downstream from the Novaky Chemical Plant. Both
GC–MS screening and target analysis of AWACSS analytes
using GC–MS, HPLC–DAD UV and ELISA were performed
on a regular basis. Occurrence of AWACSS analytes was evaluated also within the National screening programmes of organic pollutants and pesticides in the Slovak Republic within
the period 2001–2003. Concentrations ranges of AWACSS
compounds detected in real-world samples are summarised
in Table 1.
3.3. Development and testing of a sampling device
A sampling system was designed and realised that enables
the continuous transfer of water from a river (or any other
water source) to the autosampler of the AWACSS system
(Fig. 1). The system included a pumping system, a valve
for controlling the flow rates and the portion of water that
is provided to the AWACSS system and a filter unit which
allows the filtration of the water sample with a mesh size of
one micrometer. By exchanging the (commercially available)
filter cartridge, other filters with different pore sizes could be
used.
In a sampling experiment at River Rhine, it could be shown
that with this experimental set-up, a life-time of 3 months
could be achieved before an exchange of the filter cartridge
becomes necessary because a blocking of the filter occurs, assuming that the average turbidity of the river water is 15 FNU
(for River Rhine most often it is lower) and that a flow rate of
1 L h−1 through the sampling vessel is applied. With this flow
rate, the water in a 250 mL vessel is completely exchanged
every 15 min.
3.4. Studies on the performance of the biosensor
3.4.1. Multi-analyte measurements
Feasibility studies on multi-analyte measurements were
performed with the RIANA system for a mixture of atrazine,
isoproturon and estrone. The results showed that sensitivity
and selectivity of individual analytes remained unchanged
compared to measurements of individual compounds. A good
agreement between the results obtained by biosensor technique and the on-line SPE–LC–MS method were obtained for
each analyte (Rodriguez-Mozaz and Reder, 2004). Follow-up
1516
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
Fig. 3. The resulting set of calibration curves for atrazine, bisphenol A, estrone, isoproturon, sulphamethizole, and propanil which were measured in parallel
on a multi-analyte transducer. For all compounds, the calculated LOD is below 0.020 ␮g L−1 .
determinations by HPLC–DAD UV method showed a small
positive bias towards higher values obtained by the immunosensor for most of the tested analytes. In the final stage
of the project, the multi-analyte determination capability of
AWACSS was successfully tested with a mixture of six analytes. The AWACSS IO chip was modified with derivatives
of atrazine, bisphenol A, estrone, isoproturon, sulphamethizole, and propanil according to the previously described immobilisation protocol. Then, a simultaneous calibration from
0 to 90 ␮g L−1 analyte concentration in Milli-Q water with
mixed analytes and an antibody stock solution containing
the six corresponding polyclonal antibodies (anti-atrazine,
anti-bisphenol A, anti-estrone, anti-isoproturon, anti-mixed
sulphonamides, and anti-propanil) was performed. The resulting set of calibration curves is shown in Fig. 3. For
all compounds, the calculated LOD is below 0.020 ␮g L−1
and all validation parameters, calculated relative signals, and
standard deviations are summarised in Table 2.
This multi-analyte calibration demonstrated the possibility to quantify pesticides from three different classes
(triazines, phenylurea herbicides and anilides), endocrine disrupting compounds (bisphenol A), steroid hormones (estrone), and pharmaceuticals (sulphamethizole) within one
single measurement cycle, which only takes approximately
18 min. No cross-reactivity effect was observed for any of the
tested analytes.
3.4.2. Matrix effects
As expected from the mode of antibody action, matrix
effects were observed for different types of samples. The effects were studied in various types of water matrices, including Milli-Q, ground, surface and mixed municipal/industrial
waste water. Typical changes in profiles of calibration curves
are shown for estrone by Rodriguez-Mozaz and Reder (2004),
where the most pronounced effect was observed in the waste
water sample. It has been concluded that at analysis of such
complex samples a care should be taken to establish calibration curves in similar matrix. Several strategies were
proposed and tested to avoid the matrix effects: (i) adjustment of ionic strength and pH of the sample; (ii) dilution
of the sample; (iii) sample clean-up; and/or (iv) enrichment
of the sample on SPE column and reconstitution of the
extract.
One of the major advantages of biosensors compared to
conventional techniques is their ability to analyse selectively
target analytes in complex matrices. Therefore, a protocol
was developed allowing for analysis of solid matrices such
as sediments and food samples. In the procedure one gram
of solid sample is extracted in ultrasonic bath with 10 mL
acetonitrile or methanol, the upper layer of extract (5 mL)
is diluted in Milli-Q water in a way to reduce concentration of organic solvent to 2% and further treated/analysed as
a liquid sample. Feasibility of the approach was proven in
the inter-laboratory study comparing results obtained by online SPE–LC–DAD UV system and ELISA for atrazine and
alachlor spiked into the Nitra River sediments and baby food
(strawberry) samples at 50 and 500 ng g−1 levels. Recoveries
rates of both analytes ranged from 89 to 108%.
A separate study on the effects of residual amounts of organic solvents acetonitrile and methanol on the immunoassay
performance was conducted with ELISA for several antibodies by Abuknesha and Griffith (submitted for publication),
Abuknesha et al. (submitted for publication) and Abuknesha
and Luk (submitted for publication).
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
1517
Table 3
Results obtained by AWACSS system within the collaborative trial
Compound
Assigned value (␮g L−1 )a
Mean
Atrazine
Level 1
Level 2
Assigned value (ng g−1 )a ,b
Milli-Q water matrix
(␮g L−1 )
Mean (%)
CV
0.095
0.950
0.11
1.12
116
118
9.6
8.3
Bisphenol A
Level 1
0.097
Level 2
0.974
0.08
1.25
82
128
4.0
16.9
Estrone
Level 1
Level 2
0.08
1.04
105
136
7.0
4.4
0.076
0.763
Sediment matrix
Mean (ng g−1 )
Mean (%)
45
450
49.0
384
109
85
5.0
3.6
44.5
445
51.7
423
116
95
27.8
7.5
36
360
25.4
333
71
93
3.2
10.7
(%)c
CV (%)c
Milli-Q water and sediment samples were spiked at 0.1 and 1.0 ␮g L−1 (water samples) and 50 and 500 ng g−1 (sediment samples), respectively.
a Corrected for recoveries of analytes on the SPE cartridges and blank measurements.
b Corrected for extraction recoveries from sediments.
c Calculated from three measurements.
3.5. Comparison of AWACSS performance to
conventional analytical techniques
The overall performance of the AWACSS system in comparison to the conventional analytical and immunosensor
techniques was tested in the inter-laboratory collaborative
trial among the project partners. In total, eleven different analytical set-ups were used in six laboratories, among them four
automated on-line SPE–LC–DAD UV systems (Slobodnı́k
et al., 1993), on-line SPE–LC–FLD, on-line SPE–LC–MS
(Slobodnı́k and Brinkman, 2000), off-line SPE/LVI–GC–MS
(Korenková et al., 2001), two RIANA prototypes (Tschmelak
et al., 2004a, 2004b, 2004c, 2004d, 2004e, 2004f), ELISA and
AWACSS. The tested matrices were Milli-Q water and freezedried 63 ␮m fractions of river sediments from the Nitra River.
Each of them was spiked with three analytes: (1) atrazine as
a representative of pesticide class, being also on the list of
WFD Priority Substances; (2) bisphenol A—industrial pollutant known as an endocrine disrupting compound; and (3)
estrone—hormone with endocrine disrupting effects, often
present in outlets of municipal waste water treatment plants.
Spiking levels were 0.1 and 1.0 ␮g L−1 in water matrices and
50 and 500 ng g−1 in sediment. In order to prevent decomposition of analytes during transport, water samples were directly loaded onto small 10 mm × 2.0 mm i.d. SPE cartridges
packed with polymeric sorbent, which fitted into the automated sample preparation device PROSPEKT available in all
partner’s laboratories. Sediment samples were first extracted
by ultrasonic extraction into acetonitrile, extract diluted in
Milli-Q water and concentrated on the same cartridges. Each
sample has been prepared in triplicate and sets of cartridges,
including blanks, were distributed among the partners. Cartridges were measured either directly by techniques using
on-line SPE-LC set-ups (Brinkman et al., 1994) or eluted by
1 mL acetonitrile, which was then reconstituted into 20 mL
volume by Milli-Q water (resulting in an acetonitrile concentration of 2%) for analysis by RIANA, ELISA and off-line
SPE/GC–MS. For more details on the procedures, the reader
is addressed to a reference (Slobodnı́k et al., 2004).
The results showed that, in terms of accuracy, AWACSS
performance in Milli-Q water and sediment samples is fully
comparable to conventional chromatography-based techniques (see Tables 3 and 4). Recoveries of atrazine and
bisphenol A ranged from 82 to 126% compared to the assigned values, while those for estrone were between 71 and
136%. In general, the AWACSS results were less biased towards higher values in comparison to ELISA and RIANA
immunosensor techniques. An evaluation of the collaborative
trial using the Z-score methodology as an expression of deviation of the measured from assigned value showed that none of
the results obtained by AWACSS would be excluded from the
evaluation. Here, it should be mentioned that analysis of reconstituted sediment extracts by on-line SPE–LC–DAD UV
techniques was possible only by using time-consuming mathematical deconvolution and/or subtraction of the blank signals. As regards the reproducibility, all results by AWACSS
had a coefficient of variation lower than 17% (Table 3),
the only exception being bisphenol A at the lower spiking level in sediment (27.8%). The results were well within
the range obtained by both chromatography-based and other
Table 4
Comparison of atrazine determinations in Milli-Q water obtained by conventional liquid chromatography-based analytical techniques, ELISA, the
immunosensor RIANA, and the AWACSS instrument
Mean (␮g L−1 )
CV (%)
Conventional methods
SPE–HPLC–DAD UV (Lab 1)
SPE–HPLC–DAD UV (Lab 2)
SPE–HPLC–DAD UV (Lab 3)
0.11
0.14
0.07
14.3
3.5
20.8
Immunochemistry methods
RIANA
AWACSS
ELISA
0.16
0.11
0.14
24.1
9.6
3.6
Spiked concentration: 0.1 ␮g L−1 ; number of measurements = 3.
1518
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
immunosensor techniques (for an example of Milli-Q water,
see Table 4).
4. Conclusions
Two prototypes of the AWACSS system were designed
and constructed within the project and over 20 antibodies
and their analyte derivatives were developed. Analytical performance of AWACSS systems was tested in comparison to
a wide range of conventional chromatography-based analytical techniques, such as on-line SPE–LC–DAD UV, on-line
SPE–LC–MS and GC–MS systems. Feasibility studies on
the multi-analyte analysis and/or matrix effects were conducted with immunosensor techniques RIANA and ELISA.
Among the tested matrices were surface, ground, drinking
and waste water and, using special sample preparation protocol, sediment samples. The results showed that AWACSS
is fully comparable to conventional analytical techniques in
terms of accuracy, repeatability and reproducibility. Detection limits of all tested analytes were in the low nanogram
per litre range, while selectivity allowed for trace analysis
even in complex matrices such as sediment extracts. Time of
a single analytical run was less than 18 min and during the system validation more than 70 analyses were performed within
a day in a fully automated regime. Four water-monitoring
groups from the project team were extensively assessing various fields of potential application of the system and measurement sites that could accommodate the final instruments
for testing. The fact that monitoring of many of the AWACSS
compounds is required by the present environmental EU legislation (98/83/EC, 1998; 2000/60/EC, 2000) and that most
of the compounds are being frequently detected in real water
samples all over Europe gives good perspectives to the system to be placed among the current state-of-the-art analytical
instruments.
Acknowledgements
This work was funded by the “Automated Water Analyser
Computer Supported System” (AWACSS) (EVK1-CT-200000045) research project supported by the European Commission under the Fifth Framework Programme and contributing to the implementation of the Key Action “Sustainable
Management and Quality of Water” within the Energy, Environment and Sustainable Development. The group of Damià
Barceló (CSIC) acknowledges the Ministerio de Ciencia y
Tecnologı́a (Project PPQ 2000-3006-CE) for funding. Maria
J. López de Alda acknowledges her Ramon y Cajal contract
from the Spanish Ministry of Science and Technology. Jens
Tschmelak is a scholarship holder and Guenther Proll is participant of the research training group “Quantitative Analysis
and Characterisation of Pharmaceutically and Biochemically
relevant Substances” funded by the Deutsche Forschungsgemeinschaft (DFG) at the Eberhard-Karls-University of Tuebingen.
References
98/83/EC, 1998. Council Directive (98/83/EC) of 3 November 1998 relating to the quality of water intended for human consumption. Off.
J. Eur. Commun. L330, 32–54.
2000/60/EC, 2000. Directive 2000/60/EC of the European Parliament and
of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Off. J. Eur. Commun. L327,
1–72.
Abuknesha, R.A., Griffith, H.M.T. A broad specificity antiserum to steroid
estrogens generated with 17␤-estradiol-17-O-hemiglutarate-KLH immunogen: assessment of binding properties and use to develop an
immunoassay for total natural steroid estrogens and ethinyl estradiol
in water. Anal. Bioanal. Chem., submitted for publication.
Abuknesha, R.A., Griffith, H.M.T., Ossowska, K. A highly sensitive and
specific ELISA for estrone and use in the analysis of surface water
samples. Anal. Bioanal. Chem., submitted for publication.
Abuknesha, R.A., Luk, C. Generation and assessment of a polyclonal
antiserum to a mixture of six sulfonamides: potential use of a
mixed-specificity antiserum in screening for pluralities of sulfonamide
residues. Anal. Bioanal. Chem., submitted for publication.
Barceló, D., Hennion, M.C., 1997a. Trace Determination of Pesticides
and Their Degradation Products in Water. Elsevier.
Barceló, D., Hennion, M.C., 1997b. Sampling of polar pesticides from
water matrices. Anal. Chim. Acta 338 (1–3), 3–18.
Barceló, D. (Ed.), 2000. Environmental Analysis: Sample Handling and
Trace Analysis of Pollutants—Techniques, Applications and Quality
Assurance. Elsevier.
Brinkman, U.A.Th., Slobodnı́k, J., Vreuls, J.J., 1994. Trace-level detection
and identification of polar pesticides in surface water: the SAMOS
approach. Trends Anal. Chem. 13 (9), 373–381.
Clement, R.E., Yang, P.W., Koester, C.J., 1997. Environmental analysis.
Anal. Chem. 69 (12), 251–288.
Dudley, R.A., Edwards, P., Ekins, R.P., Finney, D.J., McKenzie, I.G.,
Raab, G.M., Rodbard, D., Rodgers, R.P., 1985. Guidelines for
imunnoassay data processing. Clin. Chem. 31 (8), 1264–1271.
Glaser, R.W., 1993. Antigen-antibody binding and mass transport by convection and diffusion to a surface: a two-dimensional computer model
of binding and dissociation kinetics. Anal. Biochem. 213 (1), 152–161.
Inczedy, J., Lengyel, T., Ure, A.M., 1998. Compendium of Analytical
Nomenclature—The Orange Book, third ed. Blackwell Science.
Jung, A., Stemmler, I., Brecht, A., Gauglitz, G., 2001. Covalent strategy for immobilization of DNA-microspots suitable for microarrays
with label-free and time-resolved optical detection of hybridization.
Fresenius J. Anal. Chem. 371 (2), 128–136.
Korenková, E., Matisová, E., Slobodnı́k, J., 2001. Large volume injection
in capillary gas chromatography. Chemicke Listy 95 (9), 528–539.
Kroger, K., Jung, A., Reder, S., Gauglitz, G., 2002. Versatile biosensor
surface based on peptide nucleic acid with label free and total internal reflection fluorescence detection for quantification of endocrine
disruptors. Anal. Chim. Acta 469 (1), 37–48.
Oubina, A., Ballesteros, B., Carrasco, P.B., Galve, R., Gascon, J., Iglesias, F., Sanvincens, N., Marco, M.-P., 2000. Immunoassays for environmental analysis. In: Barceló, D. (Ed.), Environmental Analysis:
Sample Handling and Trace Analysis of Pollutants—Techniques, Applications and Quality Assurance. Elsevier, pp. 287–339.
Rodriguez-Mozaz, S., Reder, S., López de Alda, M., Gauglitz, G., Barceló, D., 2004. Simultaneous multi-analyte determination of estrone,
isoproturon and atrazine in natural waters by the RIver ANAlyser
(RIANA), an optical immunosensor. Biosens. Bioelectron. 19 (7),
633–640.
Rowe Taitt, C., Anderson, G.P., Lingerfelt, B.M., Feldstein, M.J., Ligler,
F.S., 2002. Nine-analyte detection using an array-based biosensor.
Anal. Chem. 74 (23), 6114–6120.
Slobodnı́k, J., Brinkman, U.A.Th., 2000. LC/MS interfacing systems in
environmental analysis: applications to polar pesticides. In: Barceló,
D. (Ed.), Environmental Analysis: Sample Handling and Trace Anal-
J. Tschmelak et al. / Biosensors and Bioelectronics 20 (2005) 1509–1519
ysis of Pollutants—Techniques, Applications and Quality Assurance.
Elsevier, pp. 935–1001.
Slobodnı́k, J., Groenewegen, M.G.M., Brouwer, E.R., Lingeman, H.,
Brinkman, U.A.Th., 1993. Fully automated multi-residue method for
trace-level monitoring of polar pesticides by liquid chromatography.
J. Chromatogr. 642 (12), 359–370.
Slobodnı́k, J., Oswald, P., Kozmenko, H., Korenková, E., Proll, G.,
Tschmelak, J., Sacher, F., Stien, J., Rodriguez-Mozaz, S., Tóthová,
L., Krascsenits, T., 2004. Analysis of environmental pollutants in water and sediment samples by the multi-analyte biosensor AWACSS:
comparison of performance to conventional analytical techniques.
Manuscript in preparation.
Tschmelak, J., Kumpf, M., Proll, G., Gauglitz, G., 2004a. Biosensor for
seven sulphonamides in drinking, ground, and surface water with difficult matrices. Anal. Lett. 37 (8), 1701–1718.
Tschmelak, J., Proll, G., Gauglitz, G., 2004b. Immunosensor for estrone
with an equal limit of detection as common analytical methods. Anal.
Bioanal. Chem. 378 (3), 744–745.
1519
Tschmelak, J., Proll, G., Gauglitz, G., 2004c. Sub-nanogram per litre
detection of the emerging contaminant progesterone with a fully automated immunosensor based on evanescent field techniques. Anal.
Chim. Acta, doi:10.1016/j.aca.2004.06.031.
Tschmelak, J., Proll, G., Gauglitz, G., 2004d. Ultra-sensitive fully automated immunoassay for detection of propanil in aqueous samples:
steps of progress toward sub-nanogram per liter detection. Anal.
Bioanal. Chem., doi:10.1007/s00216-004-2695-x.
Tschmelak, J., Proll, G., Gauglitz, G., 2004e. Verification of performance
with the automated direct optical TIRF immunosensor (River Analyser) in single and multi-analyte assays with real water samples.
Biosens. Bioelectron. 20 (4), 743–752.
Tschmelak, J., Proll, G., Gauglitz, G., 2004f. Immunosensor for estrone
with an equal limit of detection as common analytical methods. Anal.
Bioanal. Chem. 378 (3), 744–745.
World Health Organization, 2003. Guidelines for Drinking Water Quality,
third ed. World Health Organization, Geneva, Switzerland.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement