High-throughput multiplexed competitive immunoassay for pollutants sensing in water

Article abstract of DOI:10.1021/ac302133u

The present study described the development and evaluation of a new fully automated multiplex competitive immunoassay enabling the simultaneous detection of five water pollutants (okadaic acid (OA), 2-chloro-4-ethylamino-6- isopropylamino-1,3,5-triazine (atrazine), 2.4-dichlorophenoxyacetic acid (2,4-D), 2,4,6-trinitrotoluene (TNT), and 1,3,5-trinitroperhydro-1,3,5-triazine (RDX)). The technology is taking advantage of an optical-clear pressure-sensitive adhesive on which biomolecules can be immobilized and that can be integrated within a classical 96-well format. The optimization of the microarray composition and cross-reaction was performed using an original approach where probe molecules (haptens) were conjugated to different carriers such as protein (bovine serum albumin or ovalbumin), amino-functionalized latex beads, or dextran polymer and arrayed at the surface of the adhesive. A total of 17 different probes were then arrayed together with controls on the adhesive surface and screened toward their specific reactivity and cross-reactivity. Once optimized, the complete setup was used for the detection of the five target molecules (less than 3 h for 96 samples). Limits of detection of 0.02, 0.01, 0.01, 100, and 0.02 μg L^-1 were found for OA, atrazine, 2,4-D, TNT, and RDX, respectively. The proof of concept of the multiplex competitive detection (semiquantitative or qualitative) of the five pollutants was also demonstrated on 16 spiked samples.

Full text of DOI:10.1021/ac302133u

Article
pubs.acs.org/ac
High-Throughput Multiplexed Competitive Immunoassay for
Pollutants Sensing in Water
Cloe Desmet,† Loic J. Blum,† and Christophe A. Marquette*,†,‡
́
†Equipe Gen
́
ie Enzymatique, Membranes Biomimet
ulaires et Supramoleculaires, Universite Lyon 1-CNRS 5246 ICBMS, Bat
Villeurbanne, Cedex, France
́
iques et Assemblages Supramolec
́
ulaires, Institut de Chimie et Biochimie
Molec
́
́
́
̂
iment CPE-43, bd du 11 Novembre 1918-69622
‡AXO Science SAS, 34 Rue du Mail, 69004 Lyon, France
S
* Supporting Information
ABSTRACT: The present study described the development
and evaluation of a new fully automated multiplex competitive
immunoassay enabling the simultaneous detection of five
water pollutants (okadaic acid (OA), 2-chloro-4-ethylamino-6-
isopropylamino-1,3,5-triazine (atrazine), 2.4-dichlorophenoxy-
acetic acid (2,4-D), 2,4,6-trinitrotoluene (TNT), and 1,3,5-
trinitroperhydro-1,3,5-triazine (RDX)). The technology is
taking advantage of an optical-clear pressure-sensitive adhesive
on which biomolecules can be immobilized and that can be
integrated within a classical 96-well format. The optimization
of the microarray composition and cross-reaction was performed using an original approach where probe molecules (haptens)
were conjugated to different carriers such as protein (bovine serum albumin or ovalbumin), amino-functionalized latex beads, or
dextran polymer and arrayed at the surface of the adhesive. A total of 17 different probes were then arrayed together with
controls on the adhesive surface and screened toward their specific reactivity and cross-reactivity. Once optimized, the complete
setup was used for the detection of the five target molecules (less than 3 h for 96 samples). Limits of detection of 0.02, 0.01, 0.01,
100, and 0.02 μg L−1 were found for OA, atrazine, 2,4-D, TNT, and RDX, respectively. The proof of concept of the multiplex
competitive detection (semiquantitative or qualitative) of the five pollutants was also demonstrated on 16 spiked samples.
and terrorist activities worldwide have resulted in extensive
he presence of toxic contamination traces in water having
Tadverse effect on human health and wildlife is an ongoing contamination of soil and groundwater by TNT or RDX.4
Atrazine and 2,4-dichlorophenoxyacetic acid (2,4-D) are
organochlorine herbicides used in global intensive agriculture.
Because of their widespread use and persistence, their
accumulation in soil and groundwater became a real public
health issue. In the European Union, concentration limits for
individual pesticides and the total pesticide content in drinking
water are 0.1 and 0.5 μg L−1, respectively.5 In the U.S., the EPA
sets the maximum allowed concentrations for common
pesticides such as atrazine to 3 μg L−1.6
The detection of these different types of molecules is an
important environmental, security, and health concern for the
global community. The simultaneous presence of more than one
of these pollutants in water is now often performed, and
multiparametric analytical tools are required. Currently,
detection of herbicides is mostly performed by chromatography
(gas chromatography (GC), high-pressure liquid chromatog-
raphy (HPLC)), mass spectrometry (mass spectrometry-time-
of-flight (MS-TOF), mass spectrometry-ion trap (MS-IT), mass
spectrometry-quadrupole (MS-Q)), and capillary electropho-
concern. These pollutants can be pesticides/herbicides coming
from intensive agricultural activities, industrial side products,
pharmaceutical chemistry inherited compounds (drugs, anti-
biotic), explosives molecules from terrorists and military
activities, or toxins produced by marine animals or phytoplank-
ton blooms. Okadaic acid (OA), for example, is a diarrheic toxin
produced by toxicogenic dinoflagellates. This phycotoxin
accumulates in the digestive glands of shellfish without a toxic
effect to its host. Nevertheless, as shellfish is a part of human
alimentation chain, this toxin accumulation can be harmful,
causing diarrheic shellfish poisoning (DSP), which is charac-
terized by gastrointestinal disorders.1
Explosive molecules, such as 2,4,6-trinitrotoluene (TNT) or
1,3,5-trinitroperhydro-1,3,5-triazine (RDX), are also considered
as toxic compounds since they may cause severe health problems
in both animals and humans populations.2,3 These troubles
include anemia, abnormal liver function, cataract development,
and skin irritation. Besides, the Environmental Protection
Agency (EPA) has determined that TNT and RDX are possible
human carcinogens. TNT and RDX pollute the environment
through the contamination of waste waters and solid wastes
coming from either the explosive industry and the related
bombing activity or the bomb recycling. Thus, different military
Received: July 26, 2012
Accepted: October 30, 2012
Published: October 30, 2012
© 2012 American Chemical Society
10267
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
resis (CE) methods. For instance, the EPA recommends GC/MS
and GC/electron capture detection (ECD) for analysis of
atrazine and 2,4-D in water samples.7−9 Nevertheless, these
methods are still time-consuming and rely on sophisticated
equipments and laborious sample preparation. In order to cope
with these issues, there is a growing demand from the
environmental monitoring community for inexpensive and
sensitive analytical devices that are reliable, rapid, and capable
of high-throughput testing. Among the different types of sensors
recently developed for this purpose (electrochemical sensors,
mass sensors, and optical sensors10−13), immunoassay like
systems appear to be the most popular thanks to their high
selectivity and to the standardization and automation possibil-
ities of their protocols. Particularly, monoparametric immuno-
sensors have been proven to be valuable tools to detect trace
amounts (ppm to ppb) of specific contaminants such as
pesticides, toxins, or explosives in environmental samples.14,15
However, most of the previous systems dedicated to the
detection of environmental pollutants were focused on few
toxicants only and rarely integrating different pollution sources
(i.e., intensive agricultural activities, industrial side products,
pharmaceutical chemistry, military activities, and phytoplank-
tons).16−19 For example, Morais et al. recently described an on-
disk multiplexed immunoassay for the detection of pesticides and
antibiotics.20 In a similar way Weller et al. described a “parallel
affinity sensor array” dedicated to the direct and indirect
immunodetection of contaminants in water thanks to a glass slide
biochip.21
In the present communication, we report a new multiplex
immunoassay for the simultaneous detection of environmental
pollutants coming from intensive agricultural activities, military
activities and phytoplankton blooms (atrazine, 2,4-D, RDX,
TNT, and okadaic acid (OA)). For the development of this
multiplex immunoassay, we have been screening 17 carrier-
hapten conjugates toward their reactivity and cross-reactivity
with specific antibodies. Using microarrays in a 96-well plate
format enabled us to generate large amount of experimental data
in highly controlled conditions, facilitating the optimization of
the first five-plex competitive immunoassay. The operational
multiparameter tool was demonstrated to be able to simulta-
neously detect atrazine, 2,4-D, RDX, TNT, and OA with limits of
detection relevant to the EU and U.S. regulations.
Jackson Immuno-Research, U.S. LowCross-Buffer was pur-
chased from Candor Bioscience, Germany. Anti-2,4-D E2G2
monoclonal antibodies were kindly provided by Dr. Milan
Franek. RDX-ovalbumin (OVA-RDX) and dextran-RDX were
kindly provided by Prof. Stephano Girotti.
Hapten-Carrier Conjugates Synthesis. In order to
integrate in a microarray the small molecule hapten probes
(about 300 g mol−1 for explosives and pesticides and 800 g mol−1
for toxins), conjugates of higher molecular weight were
synthesized. Seven different haptens (three pesticides, atrazine,
2,4-D, and 2,4,5-T; two explosives, TNT and RDX; one toxin,
okadaic acid; one control, 4-BBA) were conjugated to three
different carriers (latex-amino, dextran-amino, and BSA). For the
okadaic acid conjugate, BSA only was used on the basis of the
good results obtained previously with this probe.22
For all haptens, the conjugation was performed by coupling
the primary amine function of the carriers with the carboxylic
acid function of the hapten after their carbodiimide activation.
Briefly, the probes were activated via their carboxylic acid
function by a pretreatment in 1,4-dioxane at the concentration of
1.36 g L−1 in the presence of 3.9 g L−1 N-hydroxysuccinimide and
14.8 g L−1 N,N′-dicyclohexylcarbodiimide. After an incubation
time of 30 min, the dicyclourea precipitate was eliminated by
centrifugation, and 20 μL of the supernatant were added to either
(i) 500 μL of a 10 g L−1 BSA solution, (ii) 250 μL of a latex-amino
beads solution, or (iii) 250 μL of a dextran-amino solution (2
μM). All carrier solutions were prepared in 0.1 M carbonate
buffer, pH 11. The obtained solutions were then incubated under
stirring for 2 h at room temperature for the coupling process to
be completed. The formed BSA-conjugates were then separated
from the nonreacted species on a desalting chromatography
column (Sephadex G-25 M). The formed latex- and dextran-
conjugates were separated from the nonreacted species by
successive centrifugation/rinsing cycles using microcon 3000
columns. The conjugates were stored in 0.1 M acetate buffer, KCl
0.1 M, pH 5.5 at 4 °C.
These three carriers were used because of their difference in
terms of available amino groups and to evaluate the possible
cross-reactivity toward carrier. For BSA, there are 59 lysine
residues and 30−35 are accessible for the coupling reaction.23
The polybead amino microspheres contain an unknown number
of NH2, but they are believed to exhibit a high density of amino
groups. The dextran amino contains 146 mol of NH2/mol.
In order to characterize the conjugation, the protein
conjugates were studied using UV or mass spectrometry, and
the determined molar ratio for each pollutant is shown in
Supporting Information 1.
EXPERIMENTAL SECTION
■
Pollutants, Antibodies, and Reagents. Polybead Amino
Microspheres (1.00 μm) were purchased from Polyscience, U.S.
Dextran Amino (500 000 Da) was supplied by Life Technolo-
gies, United Kingdom. BCIP/NBT, 2-chloro-4-ethylamino-6-
isopropylamino-1,3,5-triazine (atrazine), 2,4-D, 2,4,5-trichloro-
phenoxyacetic acid (2,4,5-T), 4-benzoylbenzoic acid (4-BBA),
okadaic acid from Prorocentrum concavum, picrylsulfonic acid
(TNBS) solution 1 M in H2O, bovine serum albumin (BSA), and
alkaline phosphatase labeled antimouse IgG (Fc specific)
antibodies developed in goat were obtained from Sigma-Aldrich,
France. 2,4,6-Trinitrotoluene (TNT) and RDX were purchased
from LGC Standards, France. Atrazine-BSA and antiatrazine
6F36 monoclonal antibodies developed in mouse were supplied
by Euromedex, France. Anti-RDX antibodies and anti-TNT A1
monoclonal antibodies developed in mouse were supplied by
Strategic Diagnostics, U.S. Antiokadaic acid 7E1 monoclonal
antibodies developed in mouse were purchased from Santa Cruz
Biotechnology, U.S. Alkaline phosphatase-labeled antimouse
IgG antibodies (H+L) developed in goat were supplied by
Microarray Preparation. The obtained conjugate solutions
were spotted with the addition of 0.5 g L−1 bromophenol blue as
the spotting control. A volume of 0.8 nL of each solution were
arrayed with a 450 μm spot to spot pitch in a 49-spots matrix
format (7 × 7 array) on an optical clear pressure sensitive
adhesive (TKL, AXO Science, France) using a piezoelectric
spotter (sciFLEXARRAYER S3, SCIENION, Germany).
For the first part of the study (matrix I) concerning the
screening of the conjugates and the selectivity tests, all the
conjugates were deposited in duplicates, with triplicates of
positive control (alkaline phosphatase labeled streptavidin), and
triplicates of seven nonspecific interaction controls (BSA-4-BBA,
latex-4-BBA, dextran-4-BBA, BSA, latex, dextran, and bromo-
phenol blue).
For the second part of the study (matrix II) concerning the
calibration of the test and the pollutants multiplex detection, the
10268
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Figure 1. Competitive multiplex immunoassay setup: (a) composition of the different matrixes used and (b) principle of the adhesive microarray
assembly with the bottomless plate.
Figure 2. Workflow of the automated competitive multiplex detection of the five pollutants.
matrix was composed of 5 replicates of the chosen conjugates,
one for each pollutant, triplicates of the positive control, and
triplicates of the seven nonspecific interaction controls (Figure
1a). The spotted adhesive support was then directly assembled
with a bottomless 96-well plate (Greiner bio-one SAS, France)
thanks to its adhesive property (Figure 1b).
Automated Immunoassay Protocol. General Protocol.
Automated immunoassays were carried out on an EVO75 robot
(TECAN, France) specially equipped with a heater and a washer.
The general immunoassay protocol involved the following steps:
(i) the microarrays were saturated at 37 °C for 10 min with a 1:5
dilution of LowCross Buffer in phosphate buffer saline (0.1 M,
pH 7.4) as a blocking agent to minimize nonspecific interaction
signal. (ii) Then, the specific antipesticide, antiexplosive, and
antitoxin antibodies in saturation buffer were incubated at 37 °C
on the microarrays for 1 h, followed by an incubation with (iii)
alkaline phosphatase labeled antimouse IgG antibodies at a
concentration of 0.25 mg L−1 in saturation buffer. (iv) Finally,
200 μL of a BCIP/NBT ready-to-use solution were added in each
well for signal generation at 37 °C for 30 min. The original color
of the BCIP/NBT solution was light yellow whereas the product
of the enzymatic reaction generates a purple precipitate on the
spots. Between each incubation steps, and at the end of
colorimetric revelation, the adhesive in each well was washed 3
times with 400 μL of phosphate buffer saline (PBS).
For colorimetric imaging and signal acquisition, the 96-well
plate bottom was imaged using an ordinary flatbed scanner (HP
Scanjet 3770, Hewlett-Packard) in greyscale (from 0 to 65 535
arbitrary units (au)) with a 2400 dpi resolution. Image analysis
and signal quantification were performed using GenePix Pro 5.0
software (Axon Instrument). The signal intensity per spot was
calculated as the median intensity for all pixels included in a
circular feature defining the spot and corrected using a local
background evaluation. The resulting net specific intensity of a
given probe was calculated as the mean intensity of all replicates
and corrected using the corresponding control.
10269
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Table 1. Selected Hapten-Carrier Conjugates and Their Reactivity with the Different Antibodies
Automated Direct Immunoassay. The automated protocol
used the matrix I microarrays. The only deviation from the
general protocol is the first antibodies incubation step (ii), in
which concentration ranges of each specific antibody were
incubated separately on the microarrays.
Automated Competitive Immunoassay. For this type of
assay, the protocol was modified as presented in Figure 2 and
used on matrix II microarrays. For the first antibodies incubation
step (ii), a mixture of the anti-2,4-D, antiatrazine, anti-RDX, anti-
TNT, and anti-OA specific antibodies at an optimized
concentration of 0.1, 0.1, 1, 0.2, and 0.01 mg L−1, respectively,
were incubated at 37 °C for 1 h in each well, in the presence of
different concentration ranges of each pollutant in order to
calibrate the competitive immunoassay.
pollutants (Table 1) are OA, 2-chloro-4-ethylamino-6-isopropy-
lamino-1,3,5-triazine (atrazine), 2,4-D, TNT, and RDX, five
molecules with strong structural similarities. Since the five
pollutants will have to be detected simultaneously using a
competitive multiplex assay, it is then worth to foresee strong
cross-reactivity issues between the different immobilized haptens
and the different specific antibodies. The first step of the
development of the multiplex competitive immunoassay was
then the optimization of the haptens immobilization and cross-
reactivity.
Microarray Preparation. In the described device, the
adhesive support plays a double role: it enables one to
immobilize microarrays of the hapten conjugates and to assembly
these microarrays with the 96-well bottomless microtiter plates.
The choice for an adhesive support was based on previous study,
comparing polystyrene surfaces classically used for ELISA,
porous nitrocellulose/cellulose acetate membrane, and adhesive
surface.26 The adhesive surface showed very good results in terms
of colorimetric staining with the alkaline phosphatase precipitate
and a very low background of nonspecific signal. Moreover,
thanks to the support transparency, the colorimetric result was
easily recorded. Another additional interesting feature of this
adhesive material for probe immobilization is its possible
integration in a complex system thanks to fast and easy assembly
through its adhesive property.27
RESULTS AND DISCUSSION
■
Multiplex immunoassays involve multiple antibody/target
couples, each one having its own optimum working conditions.
Furthermore, the probability of cross-talking interferences with
all the other couples evolves more than exponentially while the
multiplexing number increases.24,25 Optimizing multiplex
immunoassays can then be rapidly a real experimental nightmare,
and the automation/throughput of our assay is then here a real
plus to rapidly test multiple assay conditions. In the field of water
quality monitoring, except in the case of biological pollutions by
pathogens, most of the target molecules are small organic
compounds with highly similar structures, usually involving
unsaturated aromatic. In the present study, the five target
Probes Screening toward Selectivity. For the develop-
ment of the present five-plex immunoassay, we have been
screening 17 carrier-hapten conjugates toward their reactivity
10270
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Figure 3. Cross-reactivity study between immobilized probes and antibodies. Graphical representation of the multiplex signals obtained on the different
probes as a function of the different antibodies concentrations.
and cross-reactivity with specific antibodies. Different carrier
molecules (BSA, OVA, dextran, and latex beads) were used in
order to produce the most suitable immobilization platform for
each pollutant. Indeed, a strong effect of the probe immobiliza-
tion chemistry is usually observed for the competitive immuno-
assay of small target molecules (<500 Da). In addition to the five
target molecules (atrazine, 2,4-D, RDX, TNT, and OA), two
haptens (4-BBA and 2,4,5-T) were chosen for their structural
10271
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Table 2. Cross-Reactivity of the Different Antibodies with the Different Analytes in Solution
similarities and potential cross-reactivity with anti-2,4-D anti-
bodies. 4-BBA was used as an additional control, and 2,4,5-T was
expected to be valuably used for the detection of its structural
analogue 2,4-D. The experiments described below were realized
with the optimized conditions, determined thanks to a
checkerboard titration of the antigen and antibodies simulta-
neously (data not shown).
As a first raw experiment, the specific reactivity and cross-
reactivity of the 17 carrier-hapten conjugates were evaluated in
order to select the conjugates giving the most sensitive result and
the lowest cross-reactivity. Figure 3 presents the reactivity
profiles of the different conjugates in the presence of increasing
concentrations of specific antibodies. This representation
enabled us to identify the different cross-reactivities and to
select the best compromise between reactivity and specificity. As
can be seen in Figure 3, from the 17 tested conjugates, 7 hapten-
carriers were giving significant specific signal (BSA-OA, BSA-
atrazine, BSA-2,4-D, OVA-RDX, BSA-TNT, latex-TNT, and
dextran-TNT). Nevertheless, reactivity was also observed
between numerous partners of the assay which force us to finely
tune the antibodies concentrations and the hapten-carrier
conjugates used.
The minimum signal accepted for the selection of a conjugate
and of an antibody concentration was set to 6 000 au, in order to
have a strong colorimetric signal for the maximum reactivity, and
to be able to clearly see the decrease in competitive format. When
looking closely to the profiles, the anti-2,4-D antibodies reacted
only with the BSA-2,4-D conjugate and cross-reacted with all the
TNT probes, with BSA-atrazine, with OVA-RDX, and with BSA-
2,4,5-T but only at a concentration higher than 0.1 mg L−1. The
fine-tuning of the anti-2,4-D concentration (0.01 mg L−1, green
bar in Figure 3a) permitted then to eliminate the nonspecific
signals on the RDX, TNT, and atrazine conjugates. The
antiatrazine antibodies cross-reacted with all BSA conjugates,
i.e., BSA-OA, BSA-2,4,5-T, BSA-2,4-D, BSA-TNT, and even the
BSA-4-BBA control, but also with OVA-RDX and latex-TNT
conjugates. The tuning of the antiatrazine antibodies concen-
tration was not a satisfying solution to completely avoid cross-
reaction, and the best compromise between reducing cross-
reaction and keeping a strong specific signal on BSA-atrazine was
then found to be an antiatrazine antibodies concentration of 1 mg
L−1 (green bar in Figure 3b). The anti-RDX antibodies reacted
only with the OVA-RDX conjugate and slightly cross-reacted
with BSA-TNT and with latex-TNT. By selecting an anti-RDX
antibodies concentration of 1 mg L−1 (green bar in Figure 3c), a
maximum specific signal can be reached while lowering the cross-
reaction, except on BSA-TNT and latex-TNT conjugates. The
anti-TNT antibodies reacted specifically with all of the TNT-
conjugates and cross-reacted weakly with all of the 2,4,5-T
conjugates and with OVA-RDX. The optimization of the anti-
TNT antibodies concentration (0.1 mg L−1, green bar in Figure
3d) permitted one to avoid almost all cross-reaction signals while
keeping a high enough specific signal. Finally, the anti-OA
antibodies were shown to generate strong specific signals and
weak cross-reactivity with BSA-atrazine, OVA-RDX, latex-2,4,5-
T, and all of the TNT-conjugates. One more time, the selection
of an anti-OA antibodies concentration of 0.1 mg L−1 (green bar
in Figure 3e) completely circumvents cross-reactivity while
keeping a high enough specific signal.
Once the different antibodies concentrations were optimized,
the hapten carriers were selected according to the specific signal
and cross-reactivity they generated from each specific antibody.
BSA-2,4-D, BSA-atrazine, OVA-RDX, dextran-TNT, and BSA-
OA were the conjugates giving the highest specific signal, and
latex-TNT and BSA-TNT were discarded because of their strong
reactivity with anti-RDX and antiatrazine antibodies. The 2,4,5-T
conjugates, initially included in the study as an analogue useful
for the detection of 2,4-D, was also abandoned because of its
large cross-reactivity with other antibodies.
Table 1 presents the selected conjugates and the remaining
cross-reactivity levels. As can be seen, significant cross-reactivity
still occurred when using antiatrazine or anti-RDX antibodies,
anticipating difficulties in the analytical treatment of the
competition curves for these two pollutants.
A study of the cross-reactions of the analytes in solution was
also performed. For that purpose, a cocktail of all antibodies was
incubated with each analyte separately. The results of this study
are summarized in Table 2. For each probe, the cross-reactivity or
competitive interference of a given analyte with the different
antibodies was defined as the percentage of signal decrease at the
inflection point of the calibration curve (specific signal
corresponding to 50%).15
According to the previously calculated cross-reactivity of the
antibodies with the immobilized probes and the present cross-
reactivity with the analytes in solution, the atrazine detection is
foreseen to be hardly reliable. This issue shall in future
development be specifically targeted. Methods were previously
reported in order to evaluate and to avoid most of the cross-
reaction issues. For example, Schuetz et al. described the
selection of hapten structures for indirect immunosensors array,
testing different haptens to obtain the best reactivity and
selectivity with anti-TNT and anti-2,4-D antibodies.28 Jones et al.
also described the complexity of multianalyte analysis with
antibodies and proposed a mathematical method to improve the
experimental performance of these immunoassays and applied it
to the detection of herbicides.29Chemometric approaches and
10272
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Figure 4. Calibration curves for the competitive detection of (a) 2,4-D, (b) TNT, (c) OA, (d) RDX, and (e) atrazine. Curves were fitted using a four-
parameter logistic function. (f) Limits of detection of the different pollutants, calculated using 3 × mean standard deviation.
the use of a recognition pattern to classify the molecular analogue
could also be an interesting tool for the determination of cross-
reactivities and could improve the reliability of analyte
identifications.30
Monoplex Calibration Curves. In order to be able later on
to determine in a multiplex format the concentration of the five
different pollutants, each of the five assays were calibrated
independently using a cocktail of the five specific antibodies and
different concentration ranges of the pollutants.
Figure 4 depicts the different calibration curves obtained for
the detection of 2,4-D, TNT, OA, atrazine, and RDX. As a matter
of fact, the competition occurs for all target molecules and the
specific signal obtained on each probe decreases with the
pollutant concentration, as expected in a competitive immuno-
assay. Moreover, the reproducibility of the measurement was
high with a signal mean standard deviation of 9.6% (40
measurements, 5 replicates, 2 runs).
Figure 4f presents the limit of detection (LOD) for the five
pollutants, determined using the four-parameter logistic fitting
(Supporting Information 2). The obtained LOD (100 μg L−1 for
TNT, 0.02 μg L−1 for RDX, 0.02 μg L−1 for OA, 0.01 μg L−1 for
atrazine, and 0.01 μg L−1 for 2,4-D) were in good agreement with
the EU and U.S. regulations.3,5,6 Moreover, the sensitivity of our
immunoassays is at least as good as the one obtained with
previously described immunosensors using the same antibodies,
except for TNT detection.15,21,31−33
respectively. These percentages have to be correlated to the
highest cross-reactivity observed (Tables 1 and 2) on the
different conjugates, which might generate stable signal, even at
high competitor concentrations. As can be seen in Figure 3, the
OVA-RDX conjugate is the only RDX conjugate that gave a
signal with anti-RDX antibodies, but we also found an important
cross-reactivity signal with antiatrazine and anti-2,4-D antibodies.
Considering these results, the plateau obtained on the RDX
competition curve can be attributed to nonspecific recognition
from antiatrazine and anti-2,4-D antibodies, which are not
displaced by RDX in solution. Nevertheless, we chose to keep
this probe in our multiplex immunoassay because of its
importance for the detection of explosive molecules, and even
if the upper limit of quantification is low, our assay still can give
qualitative information about the presence or absence of this
molecule in the tested samples. The complexity of the involved
immunochemical reactions occurring in the present system, i.e.,
five different antibodies having different cross-reactivity toward
five different immobilized haptens, was thus here plainly
evidenced.
Pollutants Detection Using the Multiplex Competitive
Immunoassay. In order to evaluate the assay performances for
the pollutants multiparametric detection with a semiquantitative
method, spiked water samples were prepared with various
concentrations of the different targets. The composition of the
samples is given in Table 3 together with the recovery values (R)
calculated using the concentrations determined thanks to the
equations of the four-parameter fitting curves determined
previously (Supporting Information 2).
Analyzing the atrazine and RDX calibration curves reveals an
abnormal behavior at high competitor concentrations. Indeed,
even if the signal levels off at high concentrations, it never reaches
a zero value (or even close) and still remains at approximately
45% and 70% of the maximum signal for atrazine and RDX,
As can be seen, 82.5% of the spiked concentrations were
successfully determined using the present multiplex competitive
10273
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Table 3. Recovery Values Obtained during the Multiplexed Detection of Various Spiked Water Samples
10274
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
Notes
assay (green boxes in Table 3). Matching concentrations were
defined as significantly close values (i.e., 85% ≤ R ≤ 115%).
Moreover, every spike concentration known to be lower than the
LOD of a particular target were considered to be successfully
determined when a zero value was calculated.
Besides, 13.75% of the calculated concentrations were found
to be significantly different but not incoherent (70% ≤ R < 85%
and 115% < R ≤ 130%) from the spiking values (orange boxes in
Table 3). Then, 3.75% of the concentrations were overestimated
(R < 70% and R > 130%), leading to false positive results (red
boxes in Table 3). Interestingly, atrazine was the only target to
generate a false positive but only in the presence of RDX, 2,4-D,
or TNT, evidencing one more time the impact of the cross-
reactivity of the antiatrazine with the other haptens and analytes
(Tables 1 and 2) on the analytical results.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors are grateful to Dr. Milan Franek for the anti-2,4-D
supply and to Prof. Stefano Girotti and co-workers for the
ovalbumin-RDX and dextran-RDX supplies. This work has been
supported in part by the European Commission Program STREP
- FP7-SEC-2010-1-Bomb Factory Detection by Networks of
Advanced Sensors.
REFERENCES
■
(1) European Food Safety Authority. Eur. Food Saf. Authority J. 2008,
589, 1−62.
(2) Levine, B. S.; Furedi, E. M.; Gordon, D. E.; Barkley, J. J.; Lish, P. M.
Fund. Appl. Toxicol. 1990, 15, 373−380.
CONCLUSIONS
■
(3) Standards Division Office of Drinking Water Report; U.S.
Environmental Protection Agency: Washington, DC, 1988.
(4) Shankaran, D. R.; Kawaguchi, T.; Kim, S. J.; Matsumoto, K.; Toko,
K.; Miura, N. Anal. Bioanal. Chem. 2006, 386, 1313−1320.
(5) Off. J. Eur. Communities, 1998; pp 32−54.
(6) U.S. National Primary Drinking Water Regulations 40CFR141; U.S.
Environmental Protection Agency: Washington, DC, 1999.
(7) Office of Ground Water and Drinking Water Report; U.S.
Environmental Protection Agency: Cincinnati, OH, 2000.
(8) Office of Ground Water and Drinking Water Report; U.S.
Environmental Protection Agency: Cincinnati, OH, 2005.
(9) Office of Research and Development National Homeland Security
Research Center Research Report; U.S. Environmental Protection Agency
2005.
(10) Meaney, M. S.; McGuffin, V. L. Anal. Bioanal. Chem. 2008, 391,
2557−2576.
(11) Singh, S. J. Hazard. Mater. 2007, 144, 15−28.
(12) Li, H.; Wang, J. X.; Pan, Z. L.; Cui, L. Y.; Xu, L. A.; Wang, R. M.;
Song, Y. L.; Jiang, L. J. Mater. Chem. 2010, 21, 1730−1735.
(13) Chen, W.; Wang, Y.; Bruckner, C.; Li, C. M.; Lei, Y. Sens.
Actuators, B: Chem. 2010, 147, 191−197.
(14) Mallat, E.; Barcelo, D.; Barzen, C.; Gauglitz, G.; Abuknesha, R.
TrAC, Trends Anal. Chem. 2001, 20, 124−132.
(15) Anderson, G. P.; Moreira, S. C.; Charles, P. T.; Medintz, I. L.;
Goldman, E. R.; Zeinali, M.; Taitt, C. R. Anal. Chem. 2006, 78, 2279−
2285.
(16) Shlyapnikov, Y. M.; Shlyapnikova, E. A.; Simonova, M. A.;
Shepelyakovskaya, A. O.; Brovko, F. A.; Komaleva, R. L.; Grishin, E. V.;
Morozov, V. N. Anal. Chem. 2012, 84, 5596−5603.
(17) Bhand, S.; Surugiu, I.; Dzgoev, A.; Ramanathan, K.; Sundaram, P.
V.; Danielsson, B. Talanta 2005, 65, 331−336.
In this proof of concept study, we demonstrated the applicability
of an adhesive-based microarray in a 96-well plate format for the
multiparametric detection of three different types of small
molecule water pollutants.
The optimization of the immobilized probes, of the protocol
conditions, and of the cross-reactions was highly facilitated
thanks to the fully automated protocol possible only because of
the standard architecture of the tool (96-well plate). Moreover,
with less than 3 h assay duration for 96 samples and whatever the
number of parameter determined, our system was shown to be
compatible with high-throughput requirements.
Nevertheless, as expected when dealing with multiplex
immunoassays, cross-reactivity of the assay components was a
real bottleneck. The optimization of the assay conditions and
hapten-carrier compositions helped us partially overcome this
problem, but interferences still remain which are generating false
positive results.
In previous reports, the Pla-Roca group proposed an
interesting approach based on antibody colocalization in order
to surmount the major issue of cross-reaction in multiparametric
immunoassays.25 With their system, the specific antibodies did
not interact with each other or with the other probes and the
cross-reactions are avoided. Another alternative, possible only
when using the present adhesive microarray, shall be to combine
the multiplexing properties of the microarray with the ultrahigh
throughput of a 384 or 1536-well plate.34 Separating the cross-
reacting species in different microwells shall then lead to more
accurate analytical results while keeping the multiplex analysis
possible. Indeed, up to 100 spots per well can be obtained in 384
format and 25 in a 1536 format. This is clearly an interesting
avenue for future developments of the present water monitoring
system.
(18) Wang, Y.; Liu, N.; Ning, B. A.; Liu, M.; Lv, Z.; Sun, Z. Y.; Peng, Y.;
Chen, C. C.; Li, J. W.; Gao, Z. X. Biosens. Bioelectron. 2012, 34, 44−50.
(19) Girotti, S.; Ferri, E.; Maiolini, E.; Bolelli, L.; D’Elia, M.; Coppe, D.;
Romolo, F. S. Anal. Bioanal. Chem. 2011, 400, 313−320.
(20) Morais, S.; Tortajada-Genaro, L. A.; Arnandis-Chover, T.;
Puchades, R.; Maquieira, A. Anal. Chem. 2009, 81, 5646−5654.
(21) Weller, M. G.; Schuetz, A. J.; Winklmair, M.; Niessner, R. Anal.
Chim. Acta 1999, 393, 29−41.
ASSOCIATED CONTENT
* Supporting Information
■
S
Additional information as noted in text. This material is available
free of charge via the Internet at http://pubs.acs.org.
(22) Marquette, C. A.; Coulet, P. R.; Blum, L. J. Anal. Chim. Acta 1999,
398, 173−182.
(23) Muller, S.; Regenmortel, M. H. V. V.; Muller, S. In Laboratory
Techniques in Biochemistry and Molecular Biology; Elsevier, 1999; Vol. 28,
pp 79−131.
AUTHOR INFORMATION
Corresponding Author
*E-mail: christophe.marquette@univ-lyon1.fr.
■
(24) Luo, W.; Pla-Roca, M.; Juncker, D. Anal. Chem. 2011, 83, 5767−
5774.
Author Contributions
(25) Pla-Roca, M.; Leulmi, R. F.; Tourekhanova, S.; Bergeron, S.;
Laforte, V.; Moreau, E.; Gosline, S. J. C.; Bertos, N.; Hallett, M.; Park,
M.; Juncker, D. Mol. Cell. Proteomics 2012, DOI: 10.1074/
mcp.M111.011460.
The manuscript was written through contributions of all authors.
All authors have given approval to the final version of the
manuscript.
10275
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276

Analytical Chemistry
Article
(26) Le Goff, G. C.; Corgier, B. P.; Mandon, C. A.; De Crozals, G.;
Chaix, C.; Blum, L. J.; Marquette, C. A. Biosens. Bioelectron. 2012, 35 (1),
94−100.
(27) Corgier, B. P.; Mandon, C. A.; Le Goff, G. C.; Blum, L. J.;
Marquette, C. A. Lab Chip 2011, 11, 3006−3010.
(28) Schuetz, A. J.; Winklmair, M.; Weller, M. G.; Niessner, R. Fresenius
J. Anal. Chem. 1999, 363, 625−631.
(29) Jones, G.; Wortberg, M.; Hammock, B. D.; Rocke, D. M. Anal.
Chim. Acta 1996, 336, 175−183.
(30) Cheung, P. Y. K.; Kauvar, L. M.; Engqvistgoldstein, A. E.; Ambler,
S. M.; Karu, A. E.; Ramos, L. S. Anal. Chim. Acta 1993, 282, 181−192.
(31) Franek, M.; Kolar, V.; Granatova, M.; Nevorankova, Z. J. Agric.
Food Chem. 1994, 42, 1369−1374.
(32) Girotti, S.; Eremin, S.; Montoya, A.; Moreno, M. J.; Caputo, P.;
D’Elia, M.; Ripani, L.; Romolo, F. S.; Maiolini, E. Anal. Bioanal. Chem.
2010, 396, 687−695.
(33) Rabbany, S. Y.; Lane, W. J.; Marganski, W. A.; Kusterbeck, A. W.;
Ligler, F. S. J. Immunol. Methods 2000, 246, 69−77.
(34) Mandon, C. A.; Berthuy, O. I.; Corgier, B. P.; Le Goff, G. C.;
Faure, P.; Marche, P. N.; Blum, L. J.; Marquette, C. A. Biosens.
Bioelectron. 2013, 39, 37−43.
10276
dx.doi.org/10.1021/ac302133u | Anal. Chem. 2012, 84, 10267−10276