Influence of the hapten design on the development of a competitive ELISA for the determination of the antifouling agent Irgarol 1051 at trace levels

Article abstract of DOI:10.1021/ac980241d

Enzyme-linked immunosorbent assays (ELISAs) with a high detectability have been developed for determination of the antifouling agent Irgarol 1051. The features of the resulting assays have been rationalized by using molecular mechanic calculations (MM2+) to correlate the chemical structure of different immunizing haptens and the corresponding avidities of the obtained antisera. The ability of Irgarol 1051 to compete for the antibody binding sites with 11 horseradish peroxidase enzyme tracers, differing in the chemical structures of the hapten, has been investigated. The present paper demonstrates that high-quality antibodies and, therefore, immunoassays reaching very low detection limits could be predicted by molecular modeling studies of the analyte conformations and of the immunizing haptens' geometries, hydrogen-bonding capabilities, and electronic distributions. Two of the ELISAs obtained have been optimized to obtain reproducible immunoassays. The dynamic ranges of both assays are between 30 and 200 ng/L, and the limits of detection are ~16 ng/L. The reported immunoassays have been evaluated and validated by analyzing spiked and real seawater samples. Irgarol 1051 has been found to be present in two of the geographical locations analyzed at concentration levels dependent on the time of year. The analytical results obtained with these immunoassays have been validated by chromatographic methods.

Full text of DOI:10.1021/ac980241d

Anal. Chem. 1998, 70, 4004-4014
Influence of the Hapten Design on the
Development of a Competitive ELISA for the
Determination of the Antifouling Agent Irgarol
1051 at Trace Levels
Berta Ballesteros,†,‡ Damia` Barcelo´,‡ Francisco Sanchez-Baeza,† Francisco Camps,† and
Marı´a-Pilar Marco*,†
Departments of Biological Organic Chemistry and Environmental Chemistry, C.I.D.sC.S.I.C.,
Jorge Girona, 18-26, Barcelona, Spain
other triazines, Irgarol 1051 is mainly present in the dissolved
phase. However, its low solubility in water (Irgarol 1051, 7 mg/
L; atrazine 33 mg/ L; terbutryn 25 mg/ L)) and high partition
coefficient (Irgarol 1051 Koc ) 3.0; atrazine Koc ) 2.1; terbutryn
Koc ) 2.8)3 explains its reported association with sediments.3
Regarding its undesirable effects over the aquatic environment,
it has been reported by long-term studies that a significant
decrease of the photosynthetic activity of the periphyton occurs
at concentrations ranging from 0.25 to 1 nM (0.063-0.25 µg/ L).
On the other hand, it is worth noting the reported toxicity of
Irgarol 1051 in the rainbow trout assay with an EC50 of 0.86 mg/
L, 10-fold higher than that of the atrazine.4
Before 1992, contamination of coastal waters by Irgarol 1051
was unknown. The first report pointing to the presence of this
agent on the coastline of the Coˆte d’Azur at concentrations ranging
from 0.11 to 1.7 µg/ L appeared in 1993.5 One year later, coastal
water contamination by Irgarol 1051 was also found in southern
England at slightly lower, but still significant, concentrations
(0.002-0.5 µg/ L).6 Recently, the presence of Irgarol 1051 was
observed in other regions of Europe such as the western coast of
Sweden7 and the freshwater lake of Geneve.4 Similarly, we have
also found contamination by Irgarol 1051 on the Mediterranean
Spanish coast.8
Enzyme-linked immunosorbent assays (ELISAs) with a
high detectability have been developed for determination
of the antifouling agent Irgarol 1 0 5 1 . The features of the
resulting assays have been rationalized by using molecular
mechanic calculations (MM2 +) to correlate the chemical
structure of different immunizing haptens and the corre-
sponding avidities of the obtained antisera. The ability
of Irgarol 1 0 5 1 to compete for the antibody binding sites
with 1 1 horseradish peroxidase enzyme tracers, differing
in the chemical structures of the hapten, has been
investigated. The present paper demonstrates that high-
quality antibodies and, therefore, immunoassays reaching
very low detection limits could be predicted by molecular
modeling studies of the analyte conformations and of the
immunizing haptens’ geometries, hydrogen-bonding ca-
pabilities, and electronic distributions. Two of the ELISAs
obtained have been optimized to obtain reproducible
immunoassays. The dynamic ranges of both assays are
between 3 0 and 2 0 0 ng/ L, and the limits of detection are
∼16 ng/ L. The reported immunoassays have been evalu-
ated and validated by analyzing spiked and real seawater
samples. Irgarol 1 0 5 1 has been found to be present in
two of the geographical locations analyzed at concentration
levels dependent on the time of year. The analytical
results obtained with these immunoassays have been
validated by chromatographic methods.
The methods of analysis applied to Irgarol 1051 determination
are common to other triazines: GC/ NPD (gas chromatography
with nitrogen and phosphorus detector), GC/ MS (mass spec-
trometry detection), HPLC/ DAD (high-performance liquid chro-
matography with diode array detection), or HPLC/ MS that
reaches very low detection limits (5 ng/ L) by preconcentrating
high sample volumes (0.5-1 L). It is known that, among other
benefits, immunoassays are able to simultaneously process many
Pilot survey studies of coastal water contamination have been
so far mainly focused on tributyltin (TBT), which has recently
been restricted after the regulations introduced by the European
Community and the Mediterranean region.1,2 The herbicide
Irgarol 1051 [2-(methylthio)-4-(tert-butylamino)-6-(cyclopropylamino)-
s-triazine] has replaced this agent as the biocide in antifouling
paints in combination with copper- and zinc-based agents. As with
(3) Tolosa, I.; Readman, J. W.; Blaevoet, A.; Ghilini, S.; Bartocci, J.; Horvat, M.
Mar. Pollut. Bull. 1 9 9 6 , 32, 335-341.
(4) Toth, S.; Becker-van Slooten, K.; Spack, L.; de Alencastro, L. F.; Tarradellas,
J. Bull. Environ. Contam. Toxicol. 1 9 9 6 , 57, 426-433.
(5) Readman, J. W.; Liong Wee Kwong, L.; Gronding, D.; Bartocci, J.; Villeneuve,
J.-P.; Mee, L. D. Environ. Sci. Technol. 1 9 9 3 , 27, 1940-1942.
(6) Gough, M. A.; Fothergill, J.; Hendrie, J. D. Mar. Pollut. Bull. 1 9 9 4 , 28,
613-620.
* Corresponding author: (phone) 34 93 400 61 00; (fax) 34 93 204 59 04;
(e-mail) mpmqob@cid.csic.es.
† Department of Biological Organic Chemistry.
‡ Department of Environmental Chemistry.
(1) UNEP; UNEP(OCA)/ MED IG. 1/ 5, 1989.
(2) Europeenne, C.; European Community: Journal officiel des Communautes
Europeennes, 1989; pp 19-23.
(7) Dahl, B.; Blanck, H. Mar. Pollut. Bull. 1 9 9 6 , 32, 342-350.
(8) Ferrer, I.; Ballesteros, B.; Marco, M.-P.; Barcelo, D. Environ. Sci. Technol.
1 9 9 7 , 31, 3530-3535.
4004 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
S0003-2700(98)00241-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/29/1998

samples reaching very low detection limits without need of cleanup
or preconcentration steps, thus increasing throughput analysis.
Water samples are ideal matrixes for immunoassay performance.
Immunochemical methods have been demonstrated to be excel-
lent screening tools for analyzing water samples with a high salt
content (>35‰) such as seawater.9 In studies on the coastal water
contamination by Irgarol 1051, we have recently reported the
development of an immunoassay10 for the detection of this
antifouling agent. Antibodies were raised against an immunogen
prepared with a hapten lacking the cyclopropyl group of the
molecule. The research presented in this paper deals with an
improved immunoassay to analyze Irgarol 1051 in seawater
environmental samples at very low concentration levels. It is
known that immunoassay features are strongly determined by the
chemical structure of the immunizing hapten.11,12 Some authors
have already discussed the usefulness of using computer molec-
ular modeling studies of the haptens to predict their minimum
energy conformations and rationalize immunoassay results.13,14 In
this work, molecular modeling studies (MM2+) have been used
to interpret the obtained results and to establish a correlation
between the sensitivity of the immunochemical analytical method
and the participation of the different groups of the irgarol molecule
in antibody binding.
operates at a wavelength of 337 nm and a maximum output of 6
mW. Polystyrene microtiter plates were purchased from Nunc
(Maxisorb, Roskilde, Denmark). Washing steps were carried out
using a SLY96 PW microplate washer (SLT Labinstruments
GmbH, Salzburg, Austria). Absorbances were read using a
Multikskan Plus MK II microplate reader (Labsystems) at a single-
wavelength mode of 450 nm. The competitive curves were
analyzed with a four-parameter logistic equation using the software
Genesys (Labsystems Helsinki, Finland) and GraphPad Prism
(GraphPad Sofware Inc., San Diego, CA). Unless otherwise
indicated, data presented correspond to the average of at least
two well replicates. Immunochemicals were obtained from Sigma
Chemical Co. (St. Louis, MO). Chemical reagents were purchased
from Aldrich Chemical Co. (Milwaukee, WI). Pesticide standards
such as cyanazine, desmetryne, ametryne prometryne, propazine,
terbutylazine, terbutrin, and terbumeton used for cross-reactivity
studies were purchased from Riedel De H¨aen (Seelze, Germany)
and Dr. Ehrenstorfer (Augsburg, Germany). Atrazine and si-
mazine, used in synthesis, and Irgarol 1051, used as standard,
were obtained as a gift from Ciba-Geigy (Barcelona, Spain).
Buffers. PBS is 0.2 M phosphate buffer, 0.8% saline solution
and unless otherwise indicated the pH is 7.5. PBS-BSA is PBS
containing a 0.1% of BSA. Borate buffer is 0.2 M boric acid-
sodium borate, pH 8.7. Coating buffer is 0.5 M carbonate-
bicarbonate buffer, pH 9.6. PBST is 0.2 M phosphate buffer, 0.8%
saline solution, 0.05% Tween 20. PBST at double concentration
is 2× PBST. Citrate buffer is a 0.1 M solution of sodium citrate,
pH 5.5. The substrate solution contains 0.01% 3,3′,5,5′-tetrameth-
ylbenzidine (TMB; handle with care, carcinogenic compound) and
0.004% H2O2 in citrate buffer.
Molecular Modeling. The calculations were performed using
the Hyperchem 4.0 software package (Hyperube Inc., Gainesville,
FL). Theoretical geometries and electronic distributions were
evaluated for Irgarol 1051 and related compounds such as
terbumeton, propazine, and prometrine and for haptens 2 a, 4 b,
4 c, 4 d, and 4 e (all of them as amide derivatives). The eight main
conformers possible for each of them (defined by the relative
orientation of the triazine ring substituents) were fully optimized
using molecular mechanics model MM+ for the minimum energy
levels. These final geometries were used as starting points in
the semiempirical quantum mechanical model PM3 calculations.
Molecular electronic distributions were derived from the last
model.
EXPERIMENTAL SECTION
Thin-layer chromatography (TLC) was performed on 0.25-mm,
precoated silica gel 60 F254 aluminum sheets (Merck, Gibbstown,
NJ). 1H and 13C NMR spectra were run with a Varian Unity-300
spectrometer (300 MHz for 1H and 75 MHz for 13C) and the
chemical shifts are expressed as δ, ppm. Infrared spectra were
measured on a Bomen MB120 FT-IR spectrophotometer (Hart-
mann & Braun, Que´bec, Canada). Mass spectra were recorded
on a MD-800 capillary gas chromatograph with MS quadrupole
detector (Fisons Instruments, VG, Manchester, U.K.), and data
are reported as m/ z (relative intensity). The ion source temper-
ature was set at 200 °C; a 15 m × 0.25 mm i.d. × 0.15 µm (film
thickness) DB-225 fused-capillary column (J&W, Folsom, CA) was
used; He was the carrier gas employed at 1 mL/ min. GC
conditions were as follows: temperature program, 60 to 90 °C
(10 °C/ min), 90 to 220 °C (6 °C/ min), 220 °C (10 min); injector
temperature, 250 °C. Hapten conjugates were analyzed by
measuring their absorbance on a Uvikon 820 UV/ visible spectro-
photometer (Kontron, Zurich, Switzerland) on scan mode (200-
400 nm). The matrix-assisted laser desorption/ ionization mass
spectrometer (MALDI-MS) used for analyzing the protein conju-
gates was a time-of-flight (TOF) mass spectrometer Bruker Biflex
III (Bruker, Kalsruhe, Germany) equipped with a laser unit that
Synthesis of the Haptens. The chemical structures of the
haptens used in this paper are shown in Table 1. The preparation
of haptens 2 e and 4 d has been previously described.10 The
synthesis of the carboxylic acid derivatives of atrazine (2 b, 2 d,
4 a) and simazine (2 a, 2 c, 4 b) has been reported.15 Haptens 2 f,
4 c, and 4 e were prepared as follows.
(4 ,6 -Dichloro[1 ,3 ,5 ]triazin-2 -yl)cyclopropylamine (1 d).
Following similar procedures as already described,10,15 cyclopro-
pylamine (3.9 mg, 56 mmol) was reacted with cyanuric chloride
(10 g, 54.2 mmol) in the presence of N,N-diisopropylethylamine
(9.75 mL, 56 mmol) in diethyl ether (450 mL) to provide 10.07 g
(91% yield) of crystalline 1 d: IR (KBr, cm-1) 3428 (st N-H), 3143,
3095, and 3016 (st C-H), 1592 (δ HN<), 1565 and 1544 (st Cd
(9) Gascon, J.; Oubin˜ a, A.; Onnefjord, P.; Ferre, I.; Hammock, B. D.; Marko-
Varga, G.; Marco, M.-P.; Barcelo´, D. Anal. Chim. Acta 1 9 9 6 , 330, 41-51.
(10) Ballesteros, B.; Barcelo, D.; Camps, F.; Marco, M.-P. Anal. Chim. Acta 1997,
347, 139147.
(11) Marco, M.-P.; Gee, S.; Hammock, B. D. Trends Anal. Chem. 1 9 9 5 , 14, 415-
425.
(12) Szurdoki, F.; Bekheit, H. K. M.; Marco, M.-P.; Goodrow, M. H.; Hammock,
B. D. In New Frontiers in Agrochemical Immunoassay; Kurtz, D. A., Skerritt,
J. H., Stanker, L., Eds.; AOAC International: Arlington, VA, 1995; pp 39-
63.
(13) Holtzapple, C. K.; Carlin, R. J.; Rose, B. G.; Kubena, L. F.; Stanker, L. H.
Mol. Immunol. 1 9 9 6 , 33, 939-946.
(15) Gasco´n, J.; Ballesteros, B.; Oubin˜ a, A.; Gonzalez- Martinez, M.-P.; Puchades,
R.; Maquieira, A.; Marco, M.-P.; Barcelo´, D. Anal. Chim. Acta 1 9 9 7 , 347,
149-162.
(14) Lee, N. J.; Skerritt, J. H.; McAdam, D. P. J. Agric. Food Chem. 1 9 9 5 , 43,
1730-1739.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4005

CdN), 1519 (st CAr-N), 1455, 1395, and 1380 (δ sy CH2 and CH3);
1H NMR (200 MHz, CDCl3, δ, ppm) 0.54 (m, 2H, CH2, cyclopro-
pyl), 0.76 (m, 2H, CH2, cyclopropyl), 1.42 (s, 9H, 3 CH3), 2.70 (m,
1H, CHN), 5.39 (br s, 1H, NH (cyclopropyl)), 6.25 (br s, 1H, NH
(tert-butyl)); 13C NMR (75 MHz, CDCl3, δ, ppm) 7.11 (2 CH2),
23.55 (CHN), 28.57 (3 CH3), 51.59 ((CH3)3CN), 164.81 (CN3),
166.72 (CN3), 167.94 (CN2Cl); EIMS m/ z (relative intensity) 241
Table 1. Homology of the Chemical Structures of the
Haptens Used as Enzyme Tracers with the Target
Analyte Irgarol 1051a
(M•+, 35), 226 (M•+ - CH3, 40), 184 (M•+ - But, 25), 170 (M•+
But - CH2, 100).
-
hapten
R1
R2
R3
S-[4-(ter t-Butylamino)-6-(cyclopropylamino)[1,3,5]triazin-
2-yl]thiopropionic Acid (4c). A solution of 3-mercaptopropionic
acid (241 mg, 2.3 mmol) in 85% KOH (273 mg, 4.1 mmol), absolute
degassed ethanol (20 mL) was added to a stirred solution of 3 c
(0.5 g, 2.1 mmol) in the same solvent (20 mL) under argon
atmosphere. The reaction mixture was treated according to the
optimized procedure described before15 to yield 416 mg of a yellow
oil, which contained 80% of 4 c (51% yield) and 20% of the
group I
2 a
2 b
2 c
2 d
2 e
2 f
Cl
Cl
Cl
Cl
Cl
Cl
ethyl
isopropyl
ethyl
isopropyl
tert-butyl
(CH2)3COOH cyclopropyl
(CH2)3COOH
(CH2)3COOH
(CH2)3COOH
(CH2)3COOH
(CH2)3COOH
group II
4 a
4 b
S(CH2)2COOH ethyl
S(CH2)2COOH isopropyl
SCH3
SCH3
ethyl
ethyl
(CH2)3COOH
group III 4 d
4 e
group IV 4 c
target
analyte
tert-butyl
(CH2)3COOH cyclopropyl
S(CH2)2COOH tert-butyl
cyclopropyl
cyclopropyl
1
dithiopropionic acid (ratio determined by H NMR). About 50
Irgarol SCH3 tert-butyl
1051
mg of the crude mixture was purified by reversed-phase chroma-
tography (10 g of Isolute C18) to obtain 30 mg of pure 4 c: IR
(KBr, cm-1) 3432 (st N-H), 3282 (st O-H), 3093, 2967, 2929,
and 2871 (st C-H), 1710 (st CdO), 1556 (st CdN), 1519 (st CAr-
N), 1453, 1409, and 1351 (δ sy CH2 and CH3); 1H NMR (200 MHz,
DMSO-d6, δ, ppm) 0.48 (m, 2H, CH2 cyclopropyl), 0.63 (m, 2H,
CH2 cyclopropyl), 1.36 (s, 9H, 3 CH3), 2.63 (t, CH2COO), 2.67 (m,
J ) 3.6, CHN), 3.14 (br s, 2H, CH2S), 6.87 (s, 1H, NH), 7.46 (s,
1H, NH); 13C NMR (75 MHz, DMSO-d6, δ, ppm) 6.03 (2 CH2),
20.97 (CHN), 23.18 (CH2S), 28.68 (3CH3), 34.53 (CH2COOH),
50.23 ((CH3)3CN), 163.55 and 164.80 (2CN3 and CNS), 171.82
(CO); methyl ester EIMS m/ z (relative intensity) 325 (M•+, 90),
310 (M•+ - CH3, 45), 294 (M•+ - OCH3, 20), 268 (20), 254 (M•+
- COOCH3, 40), 238 (30), 224 (35), 210 (35), 182 (100).
4 -[N-4 -(Cyclopropylamino)-6 -(methylthio)[1 ,3 ,5 ]triazin-
2 -yl]aminobutyric Acid (4 e). A solution of sodium methanethi-
olate (258 mg, 3.7 mmol) in deoxygenated absolute ethanol (50
mL) was added under argon atmosphere to a stirred solution of
2 f (1 g, 3.7 mmol) in the same solvent (50 mL). The mixture
was refluxed for 6 h. After 3 and 5 h, additional sodium
methanethiolate (258 mg, 3.7 mmol) was added in ethanol (20
mL) until 2 f was no more detected by TLC analysis (hexane/
diethyl ether/ acetic acid, 5:2:0.01). The ethanol was removed
under reduced pressure, and the residue was dissolved in 5%
NaHCO3 and washed with CH2Cl2. The aqueous layer was
acidified with 1 N HCl to pH 3.9 to precipitate the acid 4 e and
the suspension extracted with ethyl acetate. The organic layer
was dried with Na2SO4 and filtered and the solvent evaporated to
a
Italic type entry indicates homology with the target analyte.
1
N) 1509 (st CAr-N), 1453 and 1394 (δ sy CH2); H NMR (300
MHz, CDCl3, δ ppm) 0.57 (m, 2H, CH2 cyclopropyl), 0.87 (m, 2H,
CH2 cyclopropyl), 2.80 (m, 1H, CHN), 6.23 (br s, 1H, NH); 13C
NMR (75 MHz, CDCl3, δ, ppm) 7.33 (CH2), 24.05 (CH), 167.22
(CN3), 169.61 (CN2Cl), 171.11 (CN2Cl); EIMS m/ z (relative
intensity) 204 (M•+, 10), 189 (M•+ - CH3, 100). Anal. Calcd for
C6H6N4Cl2 (205.05): C, 35.15; H, 2.95; N, 27.32; Cl, 34.58. Found:
C, 35.22; H, 3.10; N, 27.27; Cl, 34.54.
4-[N-4-Cyclopropylamino)-6-chloro[1,3,5]triazin-2-yl]ami-
nobutyric Acid (2 f). According to the procedures already
described for similar compounds,10,15 the triazine 1 d (3 g, 14.63
mmol) was reacted with 4-aminobutyric acid (1.66 g, 16.12 mmol)
in the presence of N,N-diisopropylethylamine (5.1 mL, 14.63
mmol) in absolute ethanol (370 mL) for 14 h at room temperature
and 5 h at reflux until complete disappearance of the starting
material. Extraction of the reaction mixture yielded 1.27 g (32%)
of pure 2 f: IR (KBr, cm-1) 3267 (st N-H), 3282 (st O-H), 3128,
3093, 3012, and 2939 (st C-H), 1697 (st CdO), 1620 (δ HN<),
1
1589 and 1556 (st CdN), 1446 and 1407 (δ sy CH2); H NMR
(300 MHz, DMSO-d6, δ, ppm) 0.47 (m, 2H, CH2 cyclopropyl), 0.63
(m, 2H, CH2, cyclopropyl), 1.71 (m, J ) 6.9, 2H, CH2), 2.19 (m,
2H, CH2COO), 2.72 (m, J ) 3.9, 1H, CHN), 3.24 (m, 2H, 2 CH2N),
7.83 (m, 2H, 2NH); 13C NMR (75 MHz, DMSO-d6, δ ppm) 5.82,
6.01 (2 CH2, cyclopropyl), 23.32 (CH2), 24.38 (CHN), 24.54 (CH2-
COO), 31.88 (CH2N), 164.78 (CN3), 165.43 (CN3), 167.30 (CN2-
Cl), 174.59 (CO); methyl ester EIMS m/ z (relative intensity) 285
(M•+, 85), 270 (M•+ - CH3, 100), 254 (M•+ - OCH3, 50), 238 (65),
226 (M•+ - COOCH3, 25), 212 (M•+ - CH2COOCH3, 75), 198 (M•+
- (CH2)2COOCH3, 50).
give 720 mg (69% yield) of the desired product 4e: IR (KBr, cm-1
)
3429 (st N-H), 3280 (st O-H), 3131, 3011, 2956, and 2923 (st
C-H), 1703 (st CdO), 1666 (δ HN<), 1552 (st CdN), 1510 (st
1
C-N), 1479, 1450, and 1415 (δ sy CH2 and CH3); H NMR (300
MHz, DMSO-d6, δ, ppm) 0.46 (m, 2H, CH2, cyclopropyl), 0.60 (m,
2H, CH2, cyclopropyl), 1.71 (m, CH2), 2.22 (t, J ) 7.0, CH2COO),
2.36 (s, CH3S), 2.73 (m, CHN), 3.24 (m, CH2N), 7.28 (br s, 2H,
2NH); 13C NMR (75 MHz, DMSO-d6, δ, ppm) 6.31 (2 CH2), 14.51
(CH3S), 23.39 (CH), 23.47 (CH2), 24.81 (CH2COO), 31.36 (CH2N),
164.31 (CN3), 165.35 (CN3), 170.38 (CN2S), 174.47 (CO); methyl
2-(Cyclopropylamino)-6-chloro-4-(ter t-butylamino)[1,3,5]-
triazine (3 c). Following the described synthetic methods,15
cyclopropylamine (1.96 mL, 28.2 mmol) was reacted with 1 c10 (6
g, 7.1 mmol) in the presence of N,N-diisopropylethylamine (4.92
mL, 28.2 mmol) in absolute ethanol (100 mL) to obtain 6.19 g
(94% yield) of the desired product 3 c: IR (KBr, cm-1) 3432 (st
N-H), 3093, 2965, 2931, and 2868 (st C-H), 1573 and 1550 (st
ester EIMS m/ z (relative intensity) 297 (M•+, 90), 282 (M•+
-
4006 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

CH3, 75), 266 (M•+ - OCH3, 30), 224 (M•+ - CH2COOCH3, 25),
149 (100).
(0.0037-100 nM in PBST) or samples were added to the coated
plates (50 µL/ well) and incubated for 45 min at room temperature.
Subsequently, the ET (4 e-HRP 1/ 8000 or 2 a-HRP 1/ 2000,
respectively, in PBST, 50 µL/ well) was added and incubated for
15 min more at room temperature. The plates were washed as
described before and the substrate solution was added (100 µL/
well). The enzymatic reaction was stopped after 30 min at room
temperature with 50 µL of 4 M H2SO4, and the absorbances were
read at 450 nm. The standard curve was fitted to a four-parameter
logistic equation according to the following formula: y ) A -
B/ [1 + (x/ C)D] + B,16 where A is the maximal absorbance, B is
the minimum absorbance, C is the concentration producing 50%
of the maximal absorbance, and D is the slope at the inflection
point of the sigmoid curve.
Cross-Reactivity Determinations. Stock solutions of differ-
ent structurally related triazine pesticide were prepared (100 mM
in DMSO) and stored at 4 °C. Standard curves were prepared in
PBST (0.06-200 nM) and each IC50 determined in the competitive
experiment described above. The cross-reactivity values were
calculated according to the following equation: (I50(irgarol)/ I50-
(triazine derivative)) × 100.
Conjugation to Carrier P roteins. (A) Antigens. Haptens
4 c, 4 d, and 4 e were covalently attached through the carboxylic
acids to the lysine groups of keyhole limpet hemocyanin (KLH ),
bovine serum albumin (BSA), conalbumin (CONA), and ovalbu-
min (OVA) using the mixed anhydride method as previously
described.10 The conjugates were lyophilized and stored at -80
°C. Stock solutions of 1 mg/ mL were prepared with PBS buf-
fer and stored in aliquots at -20 °C. A working aliquot was kept
a 4 °C.
(B) Enzyme Tracers (ETs). Haptens 2 a-f and 4 a-e were
coupled to horseradish peroxidase (HRP) through their corre-
sponding N-hydroxysuccinimide (NHS) esters using dicyclohexy-
lcarbodiimide (DCC) as described.15 Stock solutions (0.25 mg/
mL) of these ETs were prepared in PBS-BSA/ saturated (NH4)2SO4
(1:1) and stored in aliquots under an argon atmosphere at 4 °C.
Hapten Density Analysis. Hapten densities of 4 c-e-BSA
conjugates were determined by MALDI-TOF-MS by comparing
the molecular weight of the standard BSA with that of the
conjugates. MALDI spectra were obtained by mixing 2 µL of the
matrix [(E)-3,5-dimethoxy-4-hydroxycinnamic acid, 10 mg/ mL in
CH3CN/ H2O (70:30), 0.1% TFA] with 2 µL of a solution of the
conjugates or proteins [10 mg/ mL in CH3CN/ H2O (70:30), 0.1%
TFA]. Similar analysis could not be performed with the corre-
sponding KLH conjugates used as immunogens because of the
wide range of high molecular weights reported for this protein.
Since both BSA and KLH conjugates were prepared simulta-
neously, a parallelism was assumed regarding the extent of the
conjugation reactions.
Matrix Effect Studies. Seawater (blank sample from Medi-
terranean open sea) and filtered seawater (0.45-µm filter) samples
were used to prepare Irgarol 1051 standard curves and compare
them to the curve prepared in MilliQ water. In these studies,
the ET was diluted in 2× PBST. The competitive immunoassay
was carried out as described above.
Analysis of Spiked and Real Sample Waters. Seasalt water
was prepared by dissolving seasalts (Sigma Chemical Co.) to a
concentration of 35‰ in MilliQ water. Environmental samples
were collected in the Masnou Marina (Barcelona, Spain), Alfacs
Bay, Fangar Bay (Ebre Delta, Tarragona, Spain), Jucar river,
Albufera, Malvarosa, and Cullera (Valencia, Spain). Samples were
buffered with 10% PBST 10×, and measured 1-, 2-, or 4-fold diluted
by ELISA.
Immunization of the Rabbits. Female New Zealand white
rabbits weighing 2-4 kg were immunized with 4 c-KLH (rabbits
13, 14, and 15), 4 d-KLH (rabbits 16, 17, and 18), and 4 e-KLH
(rabbits 19, 20, and 21) according to the immunization protocol
already described.15 Evolution of the antibody titer was assessed
by measuring the binding of serial dilutions of the antisera to
microtiter plates coated with 4 c-BSA, 4 d-BSA, or 4 e-BSA,
respectively. Antiserum was obtained by centrifugation and stored
at -80 °C in the presence of 0.02% NaN3.
Screening of the Antisera (As) and ETs. The avidity of each
antiserum for the different ETs was determined by measuring the
binding of serial dilutions (1/ 1000 to 1/ 64000 in PBST, 100 µL/
well) of the ETs 2 a-f-HRP and 4 a-e-HRP (1 h at room
temperature) to microtiter plates coated (overnight, 4 °C) with
12 different dilutions of each sera (1/ 1000-1/ 256000 in coating
buffer, 100 µL/ well). Optimal concentrations for ETs and antisera
were chosen to produce absorbances of 0.7-1 unit of absorbance
in 30 min. With these concentrations, the ability of the analyte
Irgarol 1051 to compete with the ETs for the antibody binding
sites was investigated. These experiments were carried out by
adding serial dilutions of the analyte (1000-0.002 nM in PBST,
50 µL/ well) and the ETs (appropriately diluted in PBST, 50 µL/
well) to the antisera-coated plates.
RESULTS AND DISCUSSION
Hapten Design and Synthesis. In our search for the best
hapten molecule to obtain high-quality antisera, we investigated
the MM2+ optimized geometry of Irgarol 1051. When calcula-
tions were carried out as usual (vacuum conditions), the eight
conformers showed similar energies, which suggested similar
molecular populations. The situation changed when the water
dielectric constant and a network of water molecules was
introduced in the MM+ model. In this case, one of the conform-
ers appeared slightly more favored (∼10 kJ/ mol). This conformer
had the two voluminous alkyl substituents oriented opposite to
the methylthio group (see Figure 1A). The cyclopropyl group
was placed perpendicularly outside of the plane defined by the
triazine ring, which indicates a facial asymmetry. In contrast, the
tert-butyl group covered an important space region more or less
equally distributed on the two faces of the plane defined by the
triazine ring. An interesting feature was the manifested acces-
sibility of the two NH groups and of the two nitrogen atoms of
the aromatic ring. This fact potentially allows the establishment
of hydrogen bonds with other molecules such as the antibody or
Optimized Competitive ELISAs. Microtiter plates were
coated with the antiserum in coating buffer (As15 1/ 16000 or As16
1/ 8000, 100 µL/ well) overnight at 4 °C, covered with adhesive
plate sealers. The following day, the plates were washed 5 times
with PBST buffer (300 µL/ well). Serial dilutions of Irgarol
(16) Tjissen, D. H. Practice and Theory of Enzyme Immunoassays; Elsevier:
Amsterdam, 1985.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4007

Figure 1. Ball and stick models of the minimum energy conformations of Irgarol 1051, haptens 4c, 4d, and 4e, using molecular mechanics
model MM+ and quantum mechanical model PM3 calculations. Calculations have been performed using the corresponding amide derivatives
to mimic the conjugated hapten. Haptens 4c and 4d showed the greatest similarity to the analyte. The elements are presented in the following
manner: light blue, carbon; dark blue, nitrogen; gray, hydrogen; yellow, sulfur; red, oxygen; and green, chlorine.
the solvent itself (see Figure 1A). Additionally, the aliphatic region
limited by the tert-butyl and the cyclopropyl groups defines an
asymmetric space area able to participate in hydrophobic interac-
tions. Finally, the third position of the triazine ring, occupied by
the methylthio group, confers special properties to the molecule
due to the electronegativity and size of the sulfur atom.
In principle, high-affinity antibodies should afford immu-
nochemical techniques with superior detectability. The molecular
recognition relays on the molecular shape, defined by the
geometry, and on low-energy interactions such as hydrogen
bonding, hydrophobic interactions, and electrostatic and dipole-
dipole forces together with π-π complementary ring bonding.
The preliminar analysis performed with the irgarol molecule
suggested that all three groups could play an important role in
the antibody binding and therefore a decision on which substituent
had to be removed to place the spacer arm was difficult. For this
reason, we addressed the synthetic preparation of three haptens
(see Figure 2) to evaluate the participation of each of the
substituents of the triazine ring in the antibody recognition.
Analogously, by raising antisera against these three immunizing
haptens, we would obtain a broader range of immunoassay
Figure 2. Chemical structures of the three immunizing haptens
coupled to KLH. In each hapten, one of the ring substituents of the
selectivities.
The preparation of the haptens was accomplished by nucleo-
Irgarol 1051 molecule has been replaced by a spacer arm.
philic substitution of the chlorine atoms of the cyanuric chlorine
as previously reported for the preparation of hapten 4 d,10
introducing the spacer arm on the last step of the synthetic
pathway. Thus, hapten 4 c was obtained from 1 c,10 by sequential
nucleophilic substitution with cyclopropylamine and mercapto-
propionic acid (see Figure 3, scheme 1). As previously reported,
triazine carboxylic haptens containing a sulfur atom showed
amphoteric properties that made difficult their isolation and
purification from the reaction mixture. We attributed this behavior
to the greater ability of the sulfur atom to confer charge to the
triazine ring by resonant effect, when compared to that of the
chlorine atom. This effect probably causes an increase of the
4008 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

were just identified by their chemical shifts in the 1H NMR spectra
(δ 2.42, SCH3 monosubstituted; δ 2.51, 2SCH3 disubstituted) and
by GC/ MS according to their retention times (21.03 and 24.68
min) and their molecular ions and fragments in the mass spectra
(monosubstituted, m/ z 216 one chlorine isotopic pattern; disub-
stituted, m/ z 228 without chlorine isotopic pattern). The same
kind of compounds was obtained when 1 c was reacted with
sodium methanothiolate to obtain hapten 4 d.10 Attempts to
separate these tert-butyl derivatives by normal-phase chromatog-
raphy led to the corresponding dealkylated amino derivatives
because of the acidic properties of the SiO2 used as stationary
phase that promoted the elimination of the tert-butyl group.
Haptens 4 c, 4 d, and 4 e were covalently coupled to KLH and
used as immunogens to obtain antisera. The yield of the coupling
procedure was evaluated on the corresponding BSA derivatives
by MALDI-MS (4 c-BSA, 6.2; 4 d-BSA, 2.7; and 4 e-BSA, 3.1).
Despite the low molecular ratio observed, the antisera obtained
showed a high antibody titer when tested against the different
ETs.
Screening of Competitive Immunoassays. In a direct
competitive immunoassay we determine the ability of the analyte
to compete with the formation of the HRP tracer-antibody (Ab)
complex. Two equilibrium reactions take place simultaneously
leading to the analyte-Ab (Ka) and HRP tracer-Ab (K′a) immu-
nocomplexes. In such assays, the ratio between both affinity
constants is the key to obtaining a good immunoassay with very
low detection limits. To our knowledge, nobody has made an
exhaustive investigation of what should be the optimum value of
Ka/ K′a ratio. Affinity constants of polyclonal antibodies cannot
usually be adequately assessed due to the heterogeneous multiple
specificities. However, this would provide us with significant
information for the design of appropriate competitors (haptens
that, conveniently attached to an enzyme, could compite with the
free analyte for the binding sites of the immobilized antibodies).
To solve this lack of information, we screened a battery of 11 HRP
tracers with the aim of finding the hapten giving the immunoassay
with the best detectability. Simultaneously, we tried to establish
some kind of correlation between the chemical structure of the
haptens used as competitors and the IC50’s of the immunoassays
obtained. The chemical structures of the haptens used for the
preparation of ETs are shown in Table 1. We can distinguish four
groups of haptens with an increasing degree of structural homol-
ogy with the target analyte: (i) haptens 2 a-2 d do have only the
triazine ring in common with the irgarol molecule; (ii) haptens
2 e-2 f, 4 a, and 4 b share with the analyte one of the substituents
of the triazine ring; (iii) haptens 4 d and 4 e retain two of the
substituents of the irgarol molecule, and finally, (iv) hapten 4 c
shows the highest degree of homology with the target analyte.
The preparation of these competitors was carried out as previously
reported10,15 and as described in the Experimental Section. The
screening for competitive immunoassays allowed evaluation of
both the influence of the chemical structure of the immunizing
haptens and the quality of the different competitors to provide
useful immunoassays.
Figure 3. Synthetic pathways used to prepare the immunizing
haptens 4c (scheme 1) 4d, and 4e (scheme 2). During the preparation
of haptens 4d and 4e, attempts to react the intermediates 1c and
1d, first with sodium methylthiolate and subsequently with the
aminobutyric acid, gave mixtures of the mono- and dimethylthio-
substituted derivatives 5a/5b and 6a/6b, respectively (scheme 3).
Conditions: (i) RNH2/N(i-Pr)2Et, Et2O, -20 °C; (ii) RNH2/N(i-Pr)2Et,
EtOH; room temperature; (iii) HS(CH2)2COOH, KOH/EtOH, 75 °C;
(iv) H2N(CH2)3COOH/ N(i-Pr)2Et, EtOH; room temperature; (v) NaSCH3/
EtOH, ∆; R ) t-Bu or cycloPr.
basicity of the ring that could explain the higher solubility of the
S-derivatives in the acidic aqueous media used to precipitate and
isolate the other triazine carboxylic haptens. The 1H NMR spectra
also evidenced this effect in the smaller chemical shift values of
the protons in R position to the amino group of the alkyl
substituents. The higher charge density of the ring determines
a more electronegatively polarized Ctriazine-N bond to the nitrogen
causing a shielding effect on these protons in R. Thus, the protons
of the cyclopropyl group of the hapten 4 c appeared at 0.48 and
0.63 ppm, while for its precursor 3 c, with a chlorine atom, the
corresponding absorptions showed up at δ 0.54 and 0.76, respec-
tively. Similar effects were observed in other S-substituted
triazines when compared with their respective chlorinated deriva-
tives (i.e., 4 a and atrazine; data not shown). Overall, these
observations corroborated the convenience of keeping the thio-
ether function on the immunizing hapten.
Hapten 4e was prepared by replacing one of the chlorine atoms
by cyclopropylamine in a first step to obtain 1d. The aminobutyric
acid spacer arm was introduced in a second step, and finally, the
methylthio group on the third position of the triazine ring was
incorporated (see Figure 3, scheme 2). Attempts to react 1 d with
sodium methanothiolate first, introducing the spacer arm in the
last step, were unsuccessful. In this case, because of the high
nucleophilicity of the thio group a mixture of mono- and disub-
stituted methylthio derivatives 5 and 6 was obtained (see Figure
3, scheme 3). In contrast, a double nucleophilic substitution of
the chlorine atoms was never observed when the amino group
was the nucleophile. These compounds were not isolated but
Influence of the Immunizing Hapten Chemical Structure
on the Immunoassay. The competitive immunoassays obtained
with each of the combinations and their IC50’s are reported in
Table 2. A first look at these results shows that the immunizing
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4009

Table 2. IC50’s Obtained When Screening As13-15
(4c-KLH), As16-18 (4d-KLH), and As19-20 (4e-KLH)
against 2a-2f, 4a-4e-HRP Tracers for Competitive
Immunoassaysa
Figure 4. Standard Irgarol 1051 calibration curves of the best
competitive immunoassays obtained from each immunizing hapten:
4c-KLH (As15/4e-HRP), 4d-KLH (As16/2a-HRP), and 4e-KLH (As19/
4d-HRP). Curves were obtained by following the standard ELISA
protocol described in the Experimental Section and using two-well
replicates.
immunizing hapten, 4 d seems to keep most of the important
structural features of the irgarol molecule. Regardless of the
absence of the cyclopropyl group, the spacer arm partially imitates
this group when the nitrogen-alkyl angle is considered (see
Figure 1A and C), being the NH group free to establish hydrogen
bonds with the antibodies while showing similar asymmetric ring
facial distribution.
a
The gray scale indicates increasing degree of homology with the
target analyte Irgarol 1051.
The decrease of the antisera avidity for the target analyte when
the tert-butyl group was absent from the chemical structure of
the immunizing hapten 4 e has been explained by several criteria
(see Figure 1A and D). The spatial atom geometry of hapten 4 e
is very different from that of the analyte, with a lower steric volume
and restrictions. On the other hand, the absence of the bulky
tert-butyl group causes a lack of hydrophobic interactions. Thus,
the antibodies obtained are not able to stabilize the immunocom-
plex by establishing hydrophobic interactions with this important
hydrophobic area of the Irgarol 1051 molecule. Finally, the
hydrogen-bonding capabilities are not equal for hapten 4 e and
Irgarol 1051. The antibodies produced are prone to establish
equal hydrogen bonds with both NH groups, which are equally
accessible in this hapten, but not in the Irgarol 1051, where one
of them is less accessible because of the steric hindrance caused
by the presence of the tert-butyl group.
After all the above-mentioned considerations, we conclude that
the As raised against immunizing haptens possessing the tert-
butyl group are able to establish more efficient interactions with
the analyte. This fact explains the greater number of immunoas-
says showing sufficient detectability for the anaysis of trace levels
of Irgarol 1051 in environmental seawater samples.
Influence of the Chemical Structures of the ETs on the
Immunoassay. As we mentioned above, the sensitivity of the
immunoassay is highly dependent on the ratio of the affinity
constants participating in the equilibria. Looking at the IC50’s
obtained for a single immunogen and the 11 HRP tracers
screened, it can be observed that the chance of obtaining an assay
with high detectability are greater if the chemical structure of the
ET differs to a high extent from that of the analyte, especially for
low-avidity antisera. Thus, the number of antisera giving usable
assays is higher when the analogy of the ET is lower (groups I
and II). For example, the three antisera (As13, As14, and As15;
haptens 4c and 4d gave the best antisera. Both of them provided,
already in the screening step, several useful competitive immu-
noassays with acceptable IC50’s. Hapten 4 c gave the highest
number of assays with great detectability. Any of the combina-
tions afforded immunoassays with IC50’s higher than 10 µg/ L. In
contrast, the immunizing hapten 4 e did not yield any useful
immunoassay; only As19 was able to afford some assays showing
IC50 values close to 1 µg/ L, which is insufficient for direct trace
analysis although the detectability could perhaps have been
improved by varying the immunoassay conditions.
In light of these results, we intended to find an explanation
for the low quality of the immunizing hapten 4 e when compared
to 4 c and 4 d. Considering the difference of behavior of hapten
4 e as an immunizing hapten, it seems clear that the tert-butyl
group, which is absent in this molecule, must be the most
important antigenic determinant of the analyte. In contrast,
despite the also important steric magnitude of the cyclopropyl
group (see Figure 1A), this group seems to have minimum
participation in stabilizing the immunocomplexes. We compared
the MM+ optimized geometries of these haptens and the irgarol
molecule (see Figure 1) and attempted to evaluate the noncovalent
interactions that could participate in the stabilization of the
analyte-antibody immunocomplexes in each case. Figure 1B-D
exhibits the optimized geometries of the haptens. The more
stable conformer of hapten 4 c shows the greatest structural
similarity with the analyte; the potential binding interactions
commented above are nearly equal (see Figure 1A and B).
However, with small differences, both immunizing haptens 4 c
(with cyclopropyl group) and 4 d (without this substituent) gave
a very similar number of competitive immunoassays (see Table
2). Therefore, although hapten 4 c could be selected as the best
4010 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

Table 3. Effect of the Ionic Strength in the
Immunoassay Parametersa
PBS,
M
n
Amax
Amin
IC50, pM
slope
R2
0.02
2
4
6
0.707 ( 0.051 0.001 ( 0.001 295 ( 31 1.9 ( 0.1 0.995
0.630 ( 0.041 0.007 ( 0.005 330 ( 32 1.2 ( 0.1 0.989
a
Amax, Amin, IC50, and slope are the parameters A, B, C, and D,
respectively, of the four-parameter logistic equation used to fit the
standard curves (see Experimental Section). The values shown were
obtained with the immunoassay AS15/ 4 e-HRP and are the average of
n calibration curves run on two separate ELISA plates.
Figure 6. Effect of the pH on the immunoassay As16/2a-HRP. The
ELISA is operative between pH 2.5 and 10.5. Only a small decrease
in detectability and maximal absorbance was observed at acidic pHs
whereas no significant changes occurred at basic pHs (see text for
details). Assays were run simultaneously on three different ELISA
plates.
petitive immunoassays using antisera raised against 4 c-KLH, had
values higher than 1 (average slope values for As13 1.06 ( 0.16,
As14 0.97 ( 0.15, and As15 1.91 ( 0.43). High slope values have
been often associated with high-affinity antibodies, which is in
agreement with the higher structural homology of the immunizing
hapten 4c and Irgarol 1051. Figure 4 shows the calibration graphs
of the As/ ET combinations selected for each of the three
immunogens evaluated. It can be observed how the best immu-
noassay obtained with the antisera raised against 4 e-KLH (As19/
4 d-HRP) does not reach the necessary detection limit although
it shows a good slope and Amax/ Amin ratio.
Immunoassay Optimization and Evaluation. Immunoas-
says As15/ 4 e-HRP and As16/ 2 a-HRP were selected for further
optimization and evaluation of the immunoassay protocol. Several
parameters were studied such as incubation time, ionic strength,
and pH. From all of them, the incubation time was the most
determinant to improve immunoassay detectability. In these
studies, Irgarol 1051 was incubated on the antibody-coated plates
during increasing periods of 0, 10, 20, 30, 40, and 50 min before
adding the ET, which was subsequently incubated for 60, 50, 40,
30, 20, and 10 min, respectively (competitive step). This strategy
has proved to be useful for other assays developed in this15 and
other laboratories.17 Figure 5 shows the shift of the calibration
curves for the immunoassays investigated. Both immunoassay
combinations showed the same behavior: a preincubation of the
analyte with the antibody and reduction of the competitive step
Figure 5. Effect of preincubating the analyte on the antibody-coated
plates during different periods of time using the immunoassays As15/
4e-HRP and As16/2a-HRP. An increase of the incubation time
enhanced the detectability of the immunoassay. For each As/ET
combination, ELISAs were carried out in three separate plates each
of them containing two-well replicate curves of the analyte incubated
on the antibody-coated plates during periods of 0, 10, 20, 30, and 40
min.
see Table 2) obtained when immunizing with 4 c-KLH are able to
give useful assays (IC50’s 0.1-0.3 µg/ L) with ETs from group I;
As13 and As15 also do it with ETs from group II, and finally, only
As15 is able to afford assays if the competitors belong to groups
III and IV. As14, showing the lowest titer of the group, was only
able to give assays with IC50 lower than 1 µg/ L, if the ETs were
from group I. Similarly, when the immunizing hapten was 4 d-
KLH, the three As raised were able to provide useful competitive
immunoassays when the tracers belong to group I, but only As17
was able to give assays showing enough detectability with the
competitors from groups II, III, and IV (see Table 2). In contrast,
for As with high avidity for the analyte, we hypothesize, although
it should be proved with other antisera, that it is possible to obtain
good assays even under homologous or quasi-homologous condi-
tions (see the behavior of As15 and As17 in Table 2).
It is worth noting the lower value of the slopes of the im-
munoassays developed with the antisera raised against 4 d-KLH
(average slope values for As16 1.1 ( 0.41, As17 0.69 (0.11, and
As18 0.86 ( 0.22). In contrast, the slopes of most of the com-
(17) Weller, M. G.; Weil, L.; Niessner, R. Mikrochim. Acta 1 9 9 2 , 108, 29-40.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4011

Table 4. Parameters of the Optimized ELISAs for Irgarol 1051a
assay
n
Amax
Amin
IC50, pM
slope
working range, pM
LOD, pM
As15/ 4 e-HRP
As16/ 2 a-HRP
9
4
0.742 ( 0.045
0.867 ( 0.161
0.007 ( 0.079
0.019 ( 0.009
293 ( 79
304 ( 83
1.7 ( 0.3
1.3 ( 0.1
114-668
110-848
63 ( 24
62 ( 14
a
Amax, Amin, IC50, and slope are the parameters A, B, C, and D, respectively, of the four-parameter logistic equation used to fit the standard
curves (see Experimental Section). The values shown for As15/ 4 e-HRP correspond to the average of nine curves run on three separate plates
simultaneously. The values shown for As16/ 2 a-HRP correspond to four curves run on four separate plates on two different days. The working
range was determined by the concentration values giving 80 and 20% of the control zero absorbance. The LOD is the concentration producing 90%
of the control zero absorbance.
Table 5. Selectivities of the Immunoassays Obtained against the Immunogens 4c-KLH, 4d-KLH, and 4e-KLH
As15/ 4 e-HRP
As16/ 2 a-HRP
As19/ 4 d-HRP
As17/ 4b-HRP
analyte
R1
R2
R3
IC50, nM
CR, %
IC50, nM
CR, %
CR, %
IC50, nM
CR, %
Irgarol 1051
terbutryne
terbumeton
SCH3 tert-butyl
cyclopropyl
ethyl
methyl
ethyl
isopropyl
ethyl
methyl
isopropyl
ethyl
0.325 ( 0.075
0.34
0.49
1.31
2.75
3.37
6.31
7.68
7.37
100
88
52
29
15
12
6
5
3
13
3
0.244 ( 0.098 100
100
148
18
10
8
1.79 ( 0.35 100
SCH tert-butyl
OCH3 tert-butyl
tert-butyl
SCH3 isopropyl
SHC3 isopropyl
SCH3 isopropyl
2.25
nda
nd
0.49
2.42
nd
14
ncb
nc
45
9
20.8
35.8
8
5
tertbutylazine Cl
nc
305
92
34
21
16
prometryne
ametryne
desmetryne
propazine
atrazine
0.71
8
2.34
6.25
6.78
8.89
nc
nc
15
5
8
3
Cl
Cl
Cl
Cl
isopropyl
isopropyl
C(CN)(CH3)2 ethyl
ethyl ethyl
4.39
nd
nd
3.5
nc
nc
nc
cyanazine
simazine
2.46
15.17
594
27.6
0.2
5.2
nd
a
nd, not detected at concentrations below 100 nM. b nc, not cross-reacted.
period improved immunoassay detectability. Incubation periods
of 15-20 min for the competitive step were chosen as a
compromise between detectability and maximal absorbance of the
assay.
the immunoassay As15/ 4e-HRP that produces a narrower working
range. Immunoassay As16/ 2 a-HRP with a slope of 1.34 confers
enough precision while it broadens the range of concentrations
that can be measured.
Ionic Strength. This was considered an important physical
parameter to study since this immunoassay was addressed to
analyze seawater samples with a high salinity content. To evaluate
the influence of the ionic strength on the immunoassay features,
assays were carried out at increasing PBS concentrations varying
from 0.2 to 2 M. Table 3 shows how the immunoassay parameters
were not significantly changed by increase of the salt content of
the buffers, except for a slight diminution of the maximal
absorbance and a decrease of the slope value. In light of this
result, 0.2 M PBS was selected for further experiments.
pH Effect. Similarly, the immunoassay reported in this paper
is stable under a very broad range of pH. Figure 6 shows the
calibration graphs obtained by varying the pH of the assay buffer
from 2.5 to 10.5. Only a small decrease of the sensitivity and of
the maximal absorbance was observed at acidic pH (n ) 3, IC50
90 ( 34 and 80 ( 15 ng/ L; Amax 0.794 and 0.878, for pH 2.5 and
3.5, respectively) when compared with the assay parameters at
pH 7.8 (n ) 4, IC50 76 ( 21 ng/ L; Amax 1.102). From pH 5.5 to
basic pH, the immunoassay parameters did not differ significantly
(n ) 3, IC50’s 73 ( 23 and 75 ( 29 ng/ L; Amax 1.046 and 1.096 for
pH 5.5 and 10.5, respectively).
Immunoassay Specificity. The immunoassays obtained with the
three immunogens were evaluated according to their selectivity
versus other structurally related compounds. The results are
summarized in Table 5. With the As15/ 4 e-HRP, the importance
of the tert-butyl group for antibody recognition was again noted.
These triazine herbicides possessing a methylthio group in their
structure are only slightly recognized in this system (i.e., prom-
etryne 15%); however, the tert-butyl group constitutes a stronger
antigenic determinant (i.e., terbutylazine 29%). As a consequence,
the best recognized triazines are those possessing both of the
above groups (i.e., terbutryn 88%). The cross-reactivities obtained
could map the recognition elements for each Ab/ ET pair. As an
example, structural and electronic features of terbumeton, prom-
etryne, and propazine were analyzed by molecular modeling,
compared to those of irgarol, and related to the observed cross-
reactivity values. In electronic terms, only the change of the
methylthio group for a chlorine atom or a methyloxy group
modifies the net charge of the triazine ring and the electronic
distribution but not the net dipolar moment of the molecule. The
isopropylamino group (propazine and prometryne) partially mim-
icked the steric and hydrophobic properties of the tert-butyl group,
although its conformational restrictions are lower and all the main
conformers could be equally populated.
Table 4 summarizes the parameters defining the calibration
graph of the optimized immunoassays As15/ 4 e-HRP and As16/
2a-HRP. Both immunoassays show a good signal and an excellent
sensitivity. The only difference is the high value of the slope of
Surprisingly, the immunoassay As16/ 2 a-HRP was the most
specific for Irgarol 1051. Regardless of the presence of the tert-
4012 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

Figure 7. Effect of the seawater matrix in the ELISA As15/4e-HRP.
Seawater collected from the open sea to ensure the absence of the
analyte was split in two fractions and one of them was filtered before
its use to prepare the standard curve. Assays were carried out on
two separate plates using three-well replicates.
Table 6. Acuracy of the Assaya
sample
spike level, nM
n
measured, nM recovery, %
seasalt water
0.2
0.4
0.6
0.8
1.5
3.98
2
2
3
3
6
2
0.27 ( 0.11
0.37 ( 0.06
0.61 ( 0.06
0.98 ( 0.12
1.41 ( 0.35
4.35 ( 0.07
135
93
101
122
94
Malvarosa
109
a
Spiked water samples were buffered measured in the ELISA 2-
and 4-fold diluted using two-well replicates.
butyl and the S-methyl groups on the chemical structure of the
immunizing hapten, a pesticide such as terbutryne containing both
groups was only slightly recognized, while terbumeton and
terbutylazine did not cross-react at the concentrations studied (up
to 100 nM). Until now we have not found any explanation for
this behavior other than animal variability. In contrast, the
immunoassay As17/ 4 b-HRP, raised against the same immunogen
showed the expected selectivity (see Table 5, italics).
Figure 8. Accuracy studies to assess the correlation between
HPLC/DAD and ELISA measurements using spiked MilliQ water
samples (A) and seasalt water samples (B). Whereas the slopes were
very close to 1 for water samples, a slight bias was observed in both
analytical techniques in the presence seasalts. ELISA analyses were
carried out using two-well replicates of 1-, 2-, and 4-fold PBST-diluted
water samples.
Finally, immunoassay As19/ 4 d-HRP, recognizes best those
analytes possessing S-methyl and -isopropyl groups, such as
prometryne (305%) or ametryne (92%). Probably the isopropyl
group can mimic the cyclopropyl group of the immunizing hapten,
thus explaining the high extent of recognition of the pesticide
prometryne with two isopropyl groups in its chemical structure.
The presence of the bulky tert-butyl group in pesticides such as
terbutryne or terbutylazine reduces drastically the recognition,
despite the presence of the S-methyl and -ethyl groups that could
also mimic the spacer arm of the hapten 4 e. The reasons are
probably the same as those mentioned above for the lack of
recognition of Irgarol 1051 by these antibodies.
was observed if seawater was filtered just before the ELISA
analysis (see Figure 7).
The accuracy of the assay was studied by spiking water samples
at different levels (0.2-1.5 nM). In this experiment, the recoveries
obtained were always close to 100% (see Table 6). Similarly, we
compared the optimized ELISA with HPLC/ DAD by measuring
spiked MilliQ and seawater samples. The correlation obtained
between spiked and measured concentrations was always good,
and the slope of the linear regression equations was close to 1 in
both techniques (ELISA 1.00; HPLC 1.11) when measuring spiked
MilliQ water (see Figure 8, A). The slope was lower (slope 0.91)
when ELISA measurements were performed on spiked artificial
seawater samples which is in agreement with the matrix effect
previously mentioned (see Figure 8B). However, it must also be
noted that a slight bias in the oposite direction was also observed
for the chromatographic results (slope 1.15). When analyzing the
goodness of the correlation between the HPLC and the ELISA
results, the regression coefficients (r2) were 0.999 and 0.996, and
the standard deviations of the residuals (Sy‚x) 0.02 and 0.03 for
Effect of a Seawater Matrix. Seawater was collected from the
open sea to ensure the absence of the analyte. One fraction was
filtered before its use in the ELISA. The studies demonstrated
that seawater does not significantly affect immunoassay perfor-
mance; however, because of the slightly larger absorbance
observed at low concentration values, measurements carried out
without filtering the sample may underestimate irgarol concentra-
tion in environmental samples. In contrast, no effect of the matrix
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4013

ation of the antisera obtained against the three immunizing
haptens prepared, 4 c, 4 d, and 4 e, has revealed that the cyclo-
propyl group does not really play an important role in stabilizing
the immunoclomplex. In fact, the antiserum obtained against
hapten 4 d shows a high avidity comparable to 4 c where all the
original substituents are present. On the other hand, hapten 4 e
has yielded antisera with poor recognition of the irgarol molecule.
These results have been interpreted by considering the more
favored conformations adopted in aqueous solutions by the analyte
and the three haptens studied. The present paper demonstrates
that high-quality antibodies can be anticipated by molecular
modeling studies of the analyte conformations and of the im-
munizing hapten geometries, hydrogen-bonding capabilities, and
electronic distributions. Similarly, we show how sensitivity and
detectability requirements of competitive immunoassays can be
modulated by selecting the appropriate chemical structure of the
hapten to be coupled to the enzyme. We have selected the two
best antisera (As15 and As16) from the two best immunogens
(4 c-KLH and 4 d-KLH, respectively) and developed direct com-
petitive ELISAs reaching very low limits of detection. These
assays have been optimized and evaluated in terms of effect of
selected physicochemical conditions, selectivity, accuracy and
performance on seawater matrixes. Because of their simplicity,
sensitivity, selectivity and high-throughput sample processing, the
immunoassays presented here have proven to be useful tools to
carry out monitoring studies of the contamination by the antifoul-
ing agent Irgarol 1051 in coastal areas.
Table 7. Results from the Analysis of Irgarol 1051 in
Real Seawater Samplesa
sample
ELISA, pM
HPLC/ DAD, pM
Masnou (4/ 96)
Masnou (8/ 96)
Masnou (1/ 97)
Cullera
>668
<LOD
<LOD
<LOD
<LOD
<LOD
254
1284b
86b
12b
<LOD
<LOD
<LOD
278
Malvarosa
Jucar river (7/ 97)
Jucar river (9/ 97)
Jucar river (10/ 97)
209
ntc
a
The samples were buffered and analyzed 2- and 4-fold diluted by
ELISA (As15/ 4 e-HRP). b Results reported by Ferrer et al.8 The LOD
of the ELISA method is 63 ( 24 pM; the LOD of the HPLC method is
0.5 pM. c nt, not tested.
the MilliQ (y ) 0.94x + 0.05) and the seasalt (y ) 0.82x + 0.02)
water samples, respectively.
Finally, real samples were collected and measured with the
optimized immunoassay As15/ 4 e-HRP and the results compared
with those obtained by HPLC/ DAD (Table 7). The only treatment
applied to the water samples prior to the analysis was filtering of
the water and adding a small amount of PBS buffer. The results
obtained matched very well those obtained by the chromato-
graphic method. These results also show that Irgarol 1051 may
be a ubiquitous contaminant of coastal waters of the Mediter-
ranean as was pointed out in previous studies.3,5,8
ACKNOWLEDGMENT
CONCLUSIONS
This work has been supported by the EEC Environment
To rationalize hapten design and immunoassay development,
Program (Contracts ENV4-CT95-0066 and ENV4-CT97-0476),
“ACE” MAST-III (PL-971620) and CICYT (AMB96-7808-CE and
AMB98-1048-C04-01). We thank Dr. Josefina Casas (Animal
Breeding Facility, C.I.D.sC.S.I.C.), Dr. Oriol Bulvena, M-Angeles
Garc´ıa, and Manel Gonzalez (Laboratorios Vin˜ as, Barcelona,
Spain) for their kind assistance in animal treatments and Ciba-
Geigy (Barcelona, Spain) for supplying samples of simazine,
atrazine, and Irgarol 1051.
in this paper we have studied the participation of the different
antigenic determinants of the irgarol molecule on eliciting antibod-
ies. High antiserum avidities are desirable to develop sensitive
immunochemical techniques able to detect environmental pollut-
ants at the trace level. Since antibody recognition is based on
the establishment of noncovalent interactions, it is evident that
the chemical structure and electronic distribution of the target
analyte plays an important role. The immunizing hapten analyte
should preserve these features although the location of the spacer
arm introduces certain variations. Chemical conjugation of the
irgarol molecule to carrier proteins needs to replace one of the
ring substituents to introduce an appropriate linker. The evalu-
Received for review March 4, 1998. Accepted June 18,
1998.
AC980241D
4014 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998