Development of an Immunochemical Technique for the Analysis of Trichlorophenols Using Theoretical Models

Article abstract of DOI:10.1021/ac991336y

An immunoassay has been developed for trichlorophenol analysis on the basis of theoretical chemistry modeling studies. These data have allowed us to choose the optimum chemical structure of the immunizing hapten according to realistic similarities with the target analyte. The synthesis of this hapten and the subsequent application of an appropriate immunization protocol have lead to the production of poryclonal antibodies against the target analyte. A homologous direct competitive ELISA has been developed that can be carried out in about 1 h. It has a limit of detection of 0.2 ± 0.06 μg/L (1.01 ± 0.3 nM) and it has been proven to tolerate a wide range of ionic strengths and pH values. Thus, the assay has acceptable features in samples with ionic strength between 4 and 56 mS/cm and pH values between 5.5 and 9.5. Studies on the selectivity of this immunoassay have demonstrated a high recognition of the corresponding brominated analogues. Other phenolic compounds do not interfere significantly in the analysis of 2,4,6-trichorophenol using this immunochemical technique. The accuracy of the assay has been evaluated using certified and spiked samples.

Full text of DOI:10.1021/ac991336y

Anal. Chem. 2000, 72, 2237-2246
Development of an Immunochemical Technique for
the Analysis of Trichlorophenols Using Theoretical
Models
Roger Galve, Francisco Camps, Francisco Sanchez-Baeza, and M.-Pilar Marco*
Department of Biological Organic Chemistry, Institute of Chemistry and Environmental Research of Barcelona Josep Pascual
Vila (IIQAB-CSIC), Jorge Girona, 18-26, 08034-Barcelona, Spain
intermediates in the production of phenoxy herbicides and other
important chemicals. The generalized use of bleaching processes
in the past decades has also contributed to their presence. Pulp
and paper factories have been important sources of the occurrence
of trichlorophenols in the environment. For example, between 0.6
and 1.6 mg/ L of AOX (adsorptive organic halogen) were estimated
to be emitted daily by the Finish paper industries in the late 1980s.6
Chlorophenols are also formed in incinerators during the combus-
tion of organic matter in the presence of chlorine or chlorine-
containing compounds.15,16
Although the use of chlorophenol derivatives as preservatives
has recently dropped down in many developed countries and
whitening and disinfection procedures use alternatives to hy-
pochlorite processes, exposure of the population still occurs via
a contaminated environment,6,17 domestic preservatives, edible
products,11,18 etc. Thus, chlorophenols were found in Swedish
riverine sediments at low microgram per gram levels, 40-50 km
downstream from cellulose-plant discharges.19 In fact, it is worth
noting that the amount of AOX found in several American and
European pulp and paper samples was significantly high (6 -1265
mg/ L), even if in some occasions these were marketed as TCF
(total chlorine free) or ECF (elemental chlorine free) products.20,21
Another important fact relative to the widespread exposure to
chlorophenols is that polychlorinated dibenzodioxins (PCDD),
including the most toxic dioxin congener 2,4,7,8-tetrachlorod-
ibenzo-p-dioxin (TCDD), and polychlorinated dibenzofurans
An immunoassay has been developed for trichlorophenol
analysis on the basis of theoretical chemistry modeling
studies. These data have allowed us to choose the opti-
mum chemical structure of the immunizing hapten ac-
cording to realistic similarities with the target analyte. The
synthesis of this hapten and the subsequent application
of an appropriate immunization protocol have lead to the
production of polyclonal antibodies against the target
analyte. A homologous direct competitive ELISA has been
developed that can be carried out in about 1 h. It has a
limit of detection of 0 .2 ( 0 .0 6 µg/ L (1 .0 1 ( 0 .3 nM)
and it has been proven to tolerate a wide range of ionic
strengths and pH values. Thus, the assay has acceptable
features in samples with ionic strength between 4 and 5 6
mS/ cm and pH values between 5 .5 and 9 .5 . Studies on
the selectivity of this immunoassay have demonstrated a
high recognition of the corresponding brominated ana-
logues. Other phenolic compounds do not interfere sig-
nificantly in the analysis of 2,4,6-trichorophenol using this
immunochemical technique. The accuracy of the assay
has been evaluated using certified and spiked samples.
Chlorophenols are widely distributed in many compartments
of the ecosystem such as soil,1-4 water,2,5,6 tissues,7 and body
fluids.7-14 Their presence is especially due to their use as antistain
agents for wood products since the early 1930s and also as
* To whom correspondence should be sent. Phone: 34 93 400 61 00. FAX:
34 93 204 59 04. Email: mpmqob@cid.csic.es.
(1) Alonso, M. C.; Puig, D.; Silgoner, I.; Graserbauer, M.; Barcelo, D. J.
Chromatogr., A 1 9 9 8 , 823, 231-239.
(11) Angerer, J.; Heinzow, B.; Schaller, K. H.; Weltle, D.; Lehnert, G. Fresenius’
J. Anal. Chem. 1 9 9 2 , 342, 433-438.
(12) Pekari, K.; Luotamo, M.; J¨arvisalo, J.; Lindroos, L.; Aitio, A. Int. Arch. Occup.
(2) Lee, M. R.; Yeh, Y. C.; Hsiang, W. S.; Hwang, B. H. J. Chromatogr., A 1 9 9 8 ,
Environ. Health 1 9 9 1 , 63, 57-62.
806, 317-324.
(13) Hill, R. H., Jr.; Head, S. L.; Baker, S.; Gregg, M.; Shealy, D. B.; Bailey, S. L.;
Williams, C. C.; Sampson, E. J.; Needham, L. L. Environ. Res. 1 9 9 5 , 71,
99-108.
(3) Santos, F. J.; Jauregui, O.; Pinto, F. J.; Galceran, M. T. J. Chromatogr., A
1 9 9 8 , 823, 249-258.
(4) Jauregui, O.; Moyano, E.; Galceran, M. T. J. Chromatogr., A 1 9 9 8 , 823,
(14) Rosenberg, C.; Kontsas, H.; Tornaeus, J.; Mutanen, P.; Pfa¨ffli, P.; Vubui, H.;
Ja¨ppinen, P. Organohalogen Comp. 1 9 9 5 , 26, 81-84.
(15) Angerer, J.; Goen, T.; Lehnert, G. Organohalogen Compd. 1 9 9 3 , 14, 163-
167.
241-248.
(5) Jin, X. H.; Zhu, M.; Conte, E. D. Anal. Chem. 1 9 9 9 , 71, 514-517.
(6) Trapp, S.; Rantio, T.; Paasivirta, J. Environ. Sci. Pollut. Res. Int. 1 9 9 4 , 1,
247-252.
(16) Wrbitzky, R.; Goen, T.; Letzel, S.; Frank, F.; Angerer, J. Int. Arch. Occup.
(7) Mukerjee, D. J. Air Waste Manage. Assoc. 1 9 9 8 , 48, 157-165.
(8) Kauppinen, T.; Kogevinas, M.; Johnson, E.; Becher, H.; Bertazzi, P. A.; Bas
Bueno de Mesquita, H.; Coggon, D.; Green, L.; Littorin, M.; Lynge, E.;
Matthews, J.; Neuberger, M.; Osman, J.; Plannet, B.; Pearce, N.; Winkel-
mann, R.; Saracci, R. Am. J. Ind. Med. 1 9 9 3 , 23, 903-920.
(9) Kontsas, H.; Rosenberg, C.; Tornaeus, J.; Mutanen, P.; Jappinen, P. Arch.
Environ. Health 1 9 9 8 , 53, 99-108.
Environ. Health 1 9 9 5 , 68, 13-21.
(17) Kitunen, V. H.; Salkinoja-Salonen, M. S. Chemosphere 1 9 9 0 , 20, 1671.
(18) Hill, R. H., Jr.; Ashley, D. L.; Head, S. L.; Needham, L. L.; Pirkle, J. L. Arch.
Environ. Health 1 9 9 5 , 50, 277-280.
(19) Heeb, N. V.; Dolezal, I. S.; Buhrer, T.; Mattrel, P.; Wolfensberger, M.
Chemosphere 1 9 9 5 , 31, 1407.
(20) Clapp, R. T.; Truemper, C. A.; Aziz, S.; Reschke, T. Tappi J. 1 9 9 6 , 79, 111-
113.
(10) Kontsas, H.; Rosenberg, C.; Pfa¨ffli, P.; Ja¨ppinen, P. Analyst (Cambridge, UK)
1 9 9 5 , 120, 1745-1749.
(21) Blaney, C. A.; Hossain, S. U. CHEMTECH 1 9 9 7 , 27, 48-51.
10.1021/ac991336y CCC: $19.00 © 2000 American Chemical Society
Published on Web 04/06/2000
Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2237

(PCDFs) are common impurities. The presence of chlorophenols
has been considered as a prerequisite for the formation of dioxins.
Thus, on a German plant the estimated median concentration of
TCDD in 1955-1970 was 5 ppm on 2,4,5-T products, but could
reach values between 10 and 50 ppm in some processes.
Measurements made in the 1970s revealed that the median
concentration of TCDD was about 1 ppm and lower than 0.1 ppm
after the introduction of a new dioxin extraction procedure in
1981.8 As a consequence, TCDD has been detected in the lipid
fraction of the serum of the production workers. In an Austrian
plant, in 1990 an average level of 340 pg/ g of blood lipid was
determined and all of workers showed symptoms of chloracne
and neurological diseases.22-24 Chlorophenols have thus been
named as predioxins25 and can be contemplated as indicators of
the formation of PCDDs. Therefore, routine monitoring of chlo-
rophenols in urine or other environmental and food matrixes may
be important not only because of their own risk effects but also
owing to their association with dioxins.
Because of the toxicity and persistence of these compounds
in the environment, chlorophenols are included in the priority list
of the EU and US environmental protection agencies. Determi-
nation of phenolic compounds in water is especially difficult as
reported in a few interlaboratory studies26 due to their wide range
of polarities and high vapor pressure that enhances volatilization
during the extraction and cleanup procedures. There is a need
for developing more efficient methods for extraction from envi-
ronmental matrixes. Usual methods of analysis require acidifica-
tion of the sample, liquid-liquid or solid-phase extraction,
derivatization, and cleaning up the sample, and the analysis finally
takes place by GC/ ECD or GC/ MS, reaching limits of detection
between 0.2 and 1 µg/ L (1.01 and 5.06 nM, respectively), starting
from significant amounts of sample.1-5 In this context, the
advantages of immunochemical techniques are well-known be-
cause of their reliability, low cost, speed of analysis, easy of use,
portability, selectivity, and detectability.27 Additionally, immuno-
chemical methods can detect extremely small quantities of the
target analyte with little sample preparation.
application of some of these theoretical criteria to the development
of an immunoassay for determination of trichlorophenol has led
to a highly sensitive and reliable analytical immunochemical
technique.
EXPERIMENTAL SECTION
Thin Layer Chromatography (TLC) was performed on 0.25-
mm, precoated silica gel 60 F254 aluminum sheets (Merck,
Gibbstown, NJ) and unless otherwise indicated the mobile phase
employed was hexane/ Et2O, 1:1.1H and 13C NMR spectra were
obtained with a Varian Unity-300 (Varian Inc., Palo Alto, CA)
spectrometer (300 MHz for 1H and 75 MHz for 13C). Infrared
spectra were measured on a Bomen MB120 FT-IR spectropho-
tometer (Hartmann & Braun, Que´bec, Canada). Gas chromatog-
raphy-mass spectrometry was performed on a MD-800 capillary
gas chromatograph with MS quadrupole detector (Fisons Instru-
ments, VG, Manchester, UK) and data are reported as m/ z
(relative intensity). The ion-source temperature 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, 80-220 °C (10 °C/ min), 220 °C (10 min);
injector temperature, 250 °C. The MALDI-MS (matrix-assisted
laser desorption ionization mass spectrometer) used for analyzing
the protein conjugates was a time-of-flight (TOF) mass spectrom-
eter Bruker Biflex III (Bruker, Kalsruhe, Germany) equipped with
a laser unit that operates at a wavelength of 337 nm and a
maximum output of 6 mW. The pH and the conductivity of all
buffers and solutions were measured with a pH meter pH 540
GLP and a conductimeter LF 340, respectively (WTW, Weilheim,
Germany). Polystyrene microtiter plates were purchased from
Nunc (Maxisorp, 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, Helsinki,
Findland) at a single wavelength mode of 450 nm or on a
SpectamaxPlus (Molecular Devices, Sunnyvale, CA). The competi-
tive curves were analyzed with a four-parameter logistic equation
using the software Genesys (Labsystems), SoftmaxPro v2.6
(Molecular Devices), and GraphPad Prism (GraphPad Sofware
Inc., San Diego, CA). Unless otherwise indicated, data presented
correspond to the average of at least two well replicates. Immuno-
chemicals were obtained from Sigma Chemical Co. (St. Louis,
MO). Chemical reagents were purchased from Aldrich Chemical
Co. (Milwaukee, WI).
The key step in developing an immunoassay for small mol-
ecules is the design and preparation of optimum haptens as
immunogens and competitors. Recently it has been demonstrated
that theoretical molecular models and calculations can be ex-
tremely useful tools for the immunochemists to develop assays
under rational bases and to adequately direct the immunoassay
features.28-31 In this paper we present the way in which the
(22) Neuberger, M.; landvoigt, W.; Denrtl, F. Int. Arch. Occup. Environ. Health
1 9 9 1 , 6, 325-327.
Buffers. PBS is 10 mM phosphate buffer/ 0.8% saline solution,
and unless otherwise indicated the pH is 7.5. Borate buffer is 0.2
M boric acid/ sodium borate pH 8.7. Coating buffer is 50 mM
carbonate-bicarbonate buffer pH 9.6. PBST is PBS with 0.05%
Tween 20. 2X PBST is PBST double-concentrated. Citrate buffer
is a 40 mM solution of sodium citrate pH 5.5. The substrate
solution contains 0.01% TMB (tetramethylbenzidine) and 0.004%
H2O2 in citrate buffer.
(23) Vena, J.; Boffetta, P.; Becher, H.; Benn, T.; Bueno-De-Mesquita, H. B.;
Coggon, D.; Colin, D.; Flesch-Janys, D.; Green, L.; Kauppinen, T.; Littorin,
M.; Lynge, E.; Mathews, J. D.; Neuberger, M.; Pearce, N.; Pesatori, A. C.;
Saracci, R.; Steenland, K.; Kogevinas, M. Environ. Health Perspect. Suppl.
1 9 9 8 , 106, 645-653.
(24) Neuberger, M.; Kundi, M.; Jager, R. Toxicol. Lett. 1 9 9 8 , 96-97, 347-350.
(25) Wrbitzky, R.; Angerer, J.; Lehnert, G. Gesundheitswesen 1 9 9 4 , 56, 629-
635.
(26) Aquacheck; WRC: Medmenham, UK, 1994.
Molecular Modeling and Theoretical Calculations. Molec-
ular modeling was performed using the Hyperchem 4.0 software
package (Hyperube Inc, Gainesville, FL). Theoretical geometries
(27) Vanemon, J. M.; Gerlach, C. L. J. Microbiol. Methods 1 9 9 8 , 32, 121-131.
(28) Ballesteros, B.; Barcelo, D.; Sanchezbaeza, F.; Camps, F.; Marco, M. P. Anal.
Chem. 1 9 9 8 , 70, 4004-4014.
(29) Holtzapple, C. K.; Carlin, R. J.; Rose, B. G.; Kubena, L. F.; Stanker, L. H.
Mol. Immunol. 1 9 9 6 , 33, 939-946.
(30) Lee, N. J.; Skerritt, J. H.; McAdam, D. P. J. Agric. Food Chem. 1 9 9 5 , 43,
1730-1739.
(31) Wang, S.; Allan, R. D.; Skerritt, J. H.; Kennedy, I. R. J. Agric. Food Chem.
1 9 9 8 , 46, 3330-3338.
2238 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

and electronic distributions were evaluated for 2,4,6-trichlorophe-
nol and potential haptens (all of them as amide derivatives) using
semiempirical quantum mechanics MNDO32 and PM333 models.
All the calculations were performed using standard computational
chemistry criteria. Theoretical calculations regarding pKa values
were carried out using the ACL/ PKa 1.2 software package
(Advanced Chemistry Development Inc., Toronto, ON, Canada)
at the Department of Analytical Chemistry (University of Lund,
Sweden).
3-(3-Hydroxy-2,4,6-trichlorophenyl)-propanoic acid 5. A solution
of 0.5 N NaOH (9 mL, 4.5 mmol, 3 equiv) was added to a solution
of the ester 4 (200 mg, 0.7 mmol) in THF (11.3 mL). The reaction
mixture was stirred for 12 h at RT until the complete disappear-
ance of the starting material was judged to have occurred by TLC.
The solvent was evaporated and the residue was dissolved in 0.5
N aq NaOH, washed with Et2O, and acidified with 1 N HCl. The
aqueous layer was extracted with ethyl acetate, dried with MgSO4,
filtered, and evaporated to dryness under reduced pressure to
1
obtain 170 mg of the hapten 5 with a 90% yield. H NMR (200
Synthesis of the Hapten. 3-(3-Hydroxyphenyl)-propanoic acid
2. An alloy of Na-Pb (5g, 10% Na, 21.75 mmol, 3.6 equiv in terms
of Na) was added slowly to a stirred solution of (3-3-hydroxy-
phenyl) propenoic acid 1 (5 g, 3.05 mmol) in 5% aqueous NaOH
(4 mL) at RT. A slight N2 stream was used to remove the hydrogen
formed during the reaction. After 4 h at RT the color of the
reaction mixture had changed from an intense to a pale yellow.
The mixture was acidified to pH 1 with 0.1 N HCl and extracted
with ethyl acetate. The organic layer was washed with NaCl sat
solution, dried with MgSO4 anhyd, filtered, and evaporated to
dryness under reduced pressure to obtain the acid 2 (0.45 g) with
an 89% yield. 1H NMR (200 MHz, DMSO-d6);δ 2.49 (t, J ) 7.5 Hz,
2H, -CH2COO-), 2.73 (t, J ) 7.5 Hz, 2H, PhCH2-), 3.37 (bs,
1H, -OH), 6.56-6.65 (ca, 2HAr ortho, 1HAr para), 7.06 (dd, J )
7.9 Hz, J ) 8 Hz, 1HAr meta), 9.25 (bs, 1H, COOH).
MHz, DMSO-d6) δ: 2.40 (t, J ) 8.4 Hz, 2H, -CH2COO-), 3.07
(t, J ) 8.4 Hz, 2H, PhCH2-), 7.59 (s, 1HAr meta), 10.43 (bs, 1H,
COOH). 13C NMR (75 MHz, DMSO-d6) δ: 26.8 (C-3), 31.7 (C-2),
120.5 (C-4′), 123.6 (C-2′), 123.7 (C-6′), 127.8 (C-5′), 135.6 (C-1′),
148.5 (C-3′), 172.6 (C-1).
P reparation of the P rotein Conjugates. Immunogen. Fol-
lowing described procedures,34 the hapten (21.6 mg, 0.08 mmol)
was reacted with tributylamine (21 µL, 0.08 mmol) and isobutyl-
chloroformate (2.4 µL, 0.08 mmol) in DMF (dimethylformamide
200 µL) and added to KLH (keyhole limpet hemocyanin 20 mg).
Antigen. Simultaneously, a BSA (bovine serum albumin)
conjugate was prepared by the same procedure but using 10.8
mg of the hapten for 20 mg of the protein. This conjugate was
used to assess conjugation by MALDI-TOF-MS.
Enzyme Tracer. According to described procedures,35 the
hapten (2.7 mg, 10 µmol) was reacted with NHS (N-hydroxysuc-
cinimide, 5.7 mg, 50 µmol) and DCC (dicyclohexylcarbodiimide,
20.6, 100 µmol) in DMF (200 µL) and added to HRP (horseradish
peroxidase, 2 mg).
P olyclonal Antisera. Three female New Zealand white rabbits
(rabbits 43, 44, and 45) weighing 1-2 kg were immunized with
5 -KLH according to the immunization protocol previously
described.35 Evolution of the antibody titer was assessed by
measuring the binding of serial dilutions of the antisera to
microtiter plates coated with 5-BSA. After an acceptable antibody
titer was observed, the animals were exsanguinated and the blood
collected on vacutainer tubes provided with a serum separation
gel. Antiserum was obtained by centrifugation and stored at -80
°C in the presence of 0.02% NaN3.
Methyl 3-(3-hydroxyphenyl)-propanoate 3 . Three drops of H2-
SO4 concentrated were added to a solution of the acid 2 (400 mg,
2.41 mmol) in MeOH (5 mL), placed on a round-bottom flask
provided with a CaCl2 tube. After 6 h at RT the reaction was
completed according to the TLC analysis. The solvent was
evaporated under reduced pressure and the oily residue sus-
pended with 5% aqueous NaHCO3 and extracted with ethyl ether.
The organic layer was dried over MgSO4 anhyd, filtered, and
evaporated to dryness to afford the desired ester 3 (430 mg) with
1
an 88% yield. H NMR (200 MHz, CDCl3); δ 2.63 (t, J ) 7.6 Hz,
2H, -CH2COO-), 2.90 (t, J ) 7.6 Hz, 2H, PhCH2-), 3.68 (s, 3H,
-OCH3), 6.03 (s, 1H, -OH), 6.68-6.77 (ca, 2HAr ortho, 1HAr para),
7.15 (dd, J ) 7.7 Hz, J ) 7.6 Hz, 1HAr meta).
Methyl 3-(3-hydroxy-2,4,6-trichlorophenyl)-propanoate 4 . A mix-
ture of SO2Cl2 (1.5 mL, 18.7 mmol) and Et2O anhyd (2.5 mL, 23.92
mmol) was added dropwise to a stirred solution of the ester 3
(0.35 g, 1.94 mmol) in CH2Cl2 anhyd (2.5 mL) under an Ar
atmosphere. The reaction mixture was stirred at RT for 5 h and
additional amounts of SO2Cl2 (150 µL, 1.86 mmol) and Et2O anhyd
(200 µL, 1.91 mmol) were added. After 1 h more the reaction was
completed according to the TLC analysis, affording the ester 4
(0.51 g) with a 93% yield. Melting point: 86 °C. IR, ν (KBr, cm-1):
3250 (-OH st), 1710 (CdO), 1458 (COO- st), 1166 (C-O st),
(34) Ballesteros, B.; Barcelo´, D.; Camps, F.; Marco, M.-P. Anal. Chim. Acta 1997,
347, 139-147.
(35) Gasco´n, J.; Oubin˜ a, A.; Ballesteros, B.; Barcelo´, D.; Camps, F.; Marco, M.-
P.; Gonza´lez-Martinez, M.-A.; Morais, S.; Puchades, R.; Maquiera, A. Anal.
Chim. Acta 1 9 9 7 , 347, 149-162.
(36) Marco, M.-P.; Gee, S.; Hammock, B. D. Trends Anal. Chem. 1 9 9 5 , 14, 415-
425.
(37) Oubin˜ a, A.; Ballesteros, B.; Bou Carrasco, P.; Galve, R.; Gasco´n, J.; Iglesias,
F.; Marco, M.-P. In Sample Handling and Trace Analysis of Pollutants:
Techniques, Applications and Quality Assurance; Barcelo´, D., Ed.; Elsevier
Science B. V.: Amsterdam, in press.
(38) Skerritt, J. H.; Lee, N. Approaches to the Synthesis of Haptens for
Immunoassay of Organophosphate and Synthetic Pyrethroid Insecticides.
In Immunoassays for Residue Analysis: Food Safety; Beier, R. C., Stanker, L.
H., Eds.; ACS Symposium Series 621; American Chemical Society: Wash-
ington, D.C., 1996; pp 124-149.
(39) Semmelhack, M. F.; Hall, H. T. J. Org. Chem. 1 9 7 4 , 96, 7091-7092.
(40) Semmelhack, M. F.; Hall, H. T. J. Org. Chem. 1 9 7 4 , 96, 7092-7094.
(41) Semmelhack, M. F.; Schmalzm, H.-G. Tetrahedron Lett. 1 9 9 6 , 37, 3089-
3092.
(42) Fischer, A.; Leary, G. J.; Topsom, R. D. J. Chem. Soc. B 1 9 6 7 , 686.
(43) Dewar, M. J. S.; Dieter, K. M. J. Am. Chem. Soc. 1 9 8 6 , 108, 8075-8086.
(44) Stull, D. R.; Prophet, J. JANAF Thermochemical Tables 1971, NSRDS-NBS37.
(45) Tabei, K.; Hiranuma, H.; Amemiya, N. Bull. Chem. Soc. Jpn. 1966, 39, 1085-
1086.
1
869 (ArC-H δ oop). H NMR (200 MHz, CDCl3) δ: 2.56 (t, J )
8.4 Hz, 2H, -CH2COO-), 3.23 (t, J ) 8.4 Hz, 2H, PhCH2-), 3.72
(s, 3H, -OCH3), 5.90 (s, 1H, -OH), 7.34 (s, 1HAr meta). 13C NMR
(75 MHz, CDCl3) δ: 27.1 (-OCH3), 31.9 (C-3), 51.9 (C-2), 119.4
(C-4′), 122.1 (C-2′), 125.7 (C-6′), 128.2 (C-5′), 135.8 (C-1′), 147.1
(C-3′), 172.6 (C-1). EM, m/ z (%): 282 (M+, 13), 247 (100), 223
(17), 222 (10), 209 (40), 205 (47). Anal. Calcd for C10H9Cl3O3: C,
42.36; H, 3.20; Cl, 37.51. Found: C, 42.45; H, 3.17; Cl, 37.57.
(32) Dewar, M. J. S.; Thiel, W. J. Am. Chem. Soc. 1 9 7 7 , 99, 4899-4917.
(33) Stewart, J. J. P. J. Comput. Chem. 1 9 8 9 , 10, 209-221.
Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2239

Immunochemistry. The appropriate dilutions of the antisera
and the enzyme tracer were established after a 2D checkerboard
titration assay performed as described with the three antisera
obtained.28 Finally antiserum 43 was selected for further optimiza-
tion and evaluation of the resulting competitive assay.
Optimized ELISA Protocol. Microtiter plates were coated with
the antiserum 43 (1/ 8000 in coating buffer 100 µL/ well) overnight
at 4 °C. The following day the plates were washed four times with
PBST and the different concentrations of the analyte standard,
crossreactants, or samples in 5 mM PBS buffer (50 µL/ well) were
added followed by the solution of the enzyme tracer 5HRP (1/
8000 diluted in 5mM PBS, 50 µL/ well). After 30 min at room
temperature, plates were washed again four times with PBST. The
substrate solution was added (100 µL/ well) and the enzymatic
reaction stopped after 30 min at room temperature with 4N H2-
SO4 (50 µL/ well). The absorbances were measured 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, 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.
Figure 1. Chemical structures of the target analyte and four
potential immunizing haptens used to raise antibodies against 2,4,6-
TCP. The arrows indicate possible derivatization positions.
consuming steps of the antibody-development process. Therefore,
one must thoroughly evaluate the synthetic effort before starting
to produce antibodies against an organic molecule. Nowadays, it
is possible to make computer-modeling studies in order to predict
the suitability of a particular chemical structure of a hapten for
raising antibodies against the target analyte.28-31
Cross-Reactivity Determinations. Stock solutions of different
phenolic compounds were prepared (1 mM in DMSO) and stored
at 4 °C. Standard curves were prepared in PBS (10 000-0.64 nM)
and each IC50 determined in the competitive experiment described
above. The cross-reactivity values were calculated according to
the following equation: (I50 trichlorophenol/ I50 phenolic derivative)
× 100.
In our search for the best hapten molecule to obtain high-
quality antisera against 2,4,6-TCP, we considered four different
potential haptens (see Figure 1, A-D). From the synthetic point
of view, hapten B can be easily prepared in just one step using
the analyte as starting material. However, the hydroxyl group is
blocked in this hapten and, additionally, certain chlorophenoxyacid
herbicides could be then also recognized by the antibodies raised
against that hapten. Observing the planar chemical structures,
one can already be aware that hapten A is the only one preserving
all the functional groups of the target analyte (chlorine atoms and
hydroxyl group), since in haptens C and D one of the chlorine
atoms has been replaced by the spacer arm at either the para or
the ortho position, respectively. From both of them, hapten C
appeared as the most suitable one since the area covered by the
phenolic group and the two chlorine atoms in ortho was more
exposed to the immune system.
In a retrosynthetic analysis for the preparation of hapten A we
considered the possibility of introducing an alkyl chain at the meta
position. However, it is well-known in organic chemistry that it is
difficult to derivatize this position in phenolic compounds using
the traditional Friedel-Crafts alkylation or acylation reactions. The
opposite reactivity, which means the nucleophilic substitution in
the aromatic ring, has been described if electron-withdrawing
groups are present.39-41 Thus, nucleophilic substitution at the meta
position with a high regioselectivity has been reported by
preparing first arene-tricarbonyl-chromium (0) complexes41 of the
corresponding aromatic compound.
Accuracy. (a) Blind Spiked Samples. This parameter was
assessed by preparing 10 different blind spiked samples in milliQ
water and measuring them in the ELISA. Measurements were
performed in triplicate and in three separate plates run on different
days.
(b) Certified water Samples. Two certified water samples, Group
18B and A29, provided by Aquacheck WRC (Medmenham, UK)
and Yorkshire Environmental (LEAP Scheme, Bradford, UK),
respectively, were used to measure 2,4,6-TCP and compare the
results with the target value and the average value obtained by
several laboratories which analyzed this sample by chromato-
graphic methods. The Aquacheck sample was a mixture of chloro-
form, bromodichloromethane, dibromochloromethane, bromo-
form, trichloroethylene, tetrachloroethylene, carbon tetrachloride,
pentachlorophenol, trichlorophenol, phenol, 2- and 4-chlorophenol,
2,4-dichlorophenol, 2-bromophenol, benzene, toluene, ethylben-
zene, styrene, o-xylene, m-xylene, and p-xylene. The Yorkshire
Environmental sample contained 4-chloro-3-methylphenol, 2-chlo-
rophenol, 2,4-dichlorophenol, 2,4-dimethylphenol, 2,4-dinitrophe-
nol, 2-methyl-4,6-dinitrophenol, 2- and 4-nitrophenol, pentachlo-
rophenol, phenol, 2,4,5-trichlorophenol, and 2,4,6-trichlorophenol.
After these previous considerations and before starting any
synthetic effort, we decided to perform theoretical studies on the
suitability of haptens A-C for antibody production against 2,4,6-
TCP. Using MNDO models we could really see that the geom-
etries of the haptens and the target analyte, at their minimum
energetic levels, did not vary significantly (see Figure 2). The
charge density showed, in contrast, slight differences as we will
discuss later. Working with phenolic compounds both the neutral
compound and the corresponding conjugated base coexist to-
RESULTS AND DISCUSSION
Hapten Design and Synthesis. The production of antibodies
against a small molecule such as 2,4,6-TCP requires preparation
of a hapten which mimics as much as possible the analyte.36-38
The introduction of a linker in the molecule may cause deforma-
tions of the molecular geometry as well as changes in its
electronical distribution. On the other hand, often the chemical
synthesis of the hapten is laborious and one of the most time-
2240 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

equilibrium. Looking at the graphed differences between this
hapten and the target analyte, it is thus clearly demonstrated that
hapten B was not an appropriate hapten to raise antibodies against
2,4,6-TCP, since it cannot mimic the analyte regarding electronic
distribution. Hapten C has, again, a more negative total charge
and especially at the para position because of the effect of the
linker. In contrast, hapten A mimicked almost perfectly the
behavior of the 2,4,6-TCP in both situations.
However, all these studies are only valid if the tendency of
these organic acids to give protons to the media is the same. This
fact is indicated by their pKa values, but it is not described in the
literature for all of them. The pKa values can be estimated
theoretically using the information available in the databases for
other similar compounds. However, another strategy that has the
same meaning is to determine the deprotonation enthalpies (DPE)
of the proton-transfer reactions of these compounds. According
to the Gibbs equation
Figure 2. Stick and wedges display of the optimized geometries of
2,4,6-TCP and haptens A-C according to MNDO models. Calcula-
tions have been made using the corresponding amide derivatives to
mimic the conjugated haptens. The elements are presented in the
following manner: Light blue, carbon; dark blue, nitrogen; gray,
hydrogen; red, oxygen; green, chlorine.
∆G0 ) ∆H0 - T∆S0
∆G0 ) -2.303RT log Keq
gether in an aqueous media, their ratio depending on the pH. For
example, with a pKa of 6.23,42 one can assume that both the acid
and conjugated base of 2,4,6-TCP will coexist in a physiological
media or in the assay buffer (see Figure 3).
∆H0 ) -2.303RT log Keq + T∆S0
on proton-transfer reactions
Charge distribution in the aromatic ring for both species was
calculated at their minimum energetic conformations for haptens
A-C and the target analyte. Figure 3 shows the results obtained
on each atom of the ring positions and the sum of those charges.
Thus, as organic acids, the differences between the analyte and
the haptens were only small and the balance of the total charge
was positive for all compounds except for hapten C, due to the
lower inductive effect produced by the alkyl chain replacing the
chlorine atom. However, the differences were greater when
considering the charge distribution in a situation where the
compounds would exist as conjugated bases. The balance of
charge in the aromatic ring was, in this case, negative for all
molecules except for hapten B, which cannot participate in this
Keq ) Ka and -logKa ) pKa
thus
DPE ) +2.303 RTpKa + T∆S0
Therefore, DPE is directly related to the pKa value in the gas
phase. Dewar et al. proved that semiempirical theoretical models
provide reliable data regarding PA (proton affinities) and DPE
values if compared with experimental data.43 Since
Figure 3. Total charge and punctual charges in the phenolic forms and as the corresponding conjugated bases. The relative amount of these
two species depends on their pKas. Hapten A is the one which mimicks the behavior of the target analyte better in both situations.
Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2241

Table 1. Calculated DPE and PKa Valuesa
Table 2. Features of the Competitive Immunoassays
Obtained by Immunizing with 5KLH and Using 5HRP as
Tracera
compound
∆Hf(ArOH)
∆Hf(ArO-)
DPE
pKa
2,4,6-TCP
hapten A
hapten C
-1373
-2638
-2937
-1362
-2631
-2922
378
374
383
6.59 ( 0.23
6.63 ( 0.28
7.37 ( 0.25
antiserum
Amax
Amin
slope
IC50 (µg/ L)
r2
43
44
45
0.747
0.526
0.830
0.062
0.105
0.108
-0.97
-0.86
-0.82
5.5
29.9
10.1
0.997
0.991
0.994
a
The enthalpies and pKa values were calculated using appropriate
software as described in the Experimental Section. Enthalpies are
expressed in kcal/ mol. The formation enthalpy considered for the
proton was 367.2 kcal/ mol.
a
The parameters are extracted from the four-parameter equation
used to fit the standard curve. Curves were run in duplicate using
twelve different concentrations.
∆H0) ∆Hfp - ∆Hfr
monochlorination of phenols using sulfuryl chloride with the
advantage that the reaction takes place at room temperature in
CH2Cl2 using ethyl ether as a base. Additionally, this approach
was successfully employed by Corey et al.50 for the chlorination
of 3,5,-dimethylphenol with a high yield and the benzylic positions
were not affected. These precedents prompted us to apply this
procedure to the chlorination of the methyl ester 3 using three
equivalents of SO2Cl2. The reaction was completed after 1.30 h
with a yield close to 90%. The hydrolysis of the ester gave the
desired final product, 5 (hapten A), with a global yield of 70%
from the starting acid 1 (see Figure 4).
by calculating the formation enthalpy of the reagents (organic
acids, ∆Hfr) and the products (conjugated bases and proton, ∆Hfp)
for 2,4,6-TCP and haptens A and C, we determined the DPE values
of the different equilibria according to the following equation
∆H0 )
[∆Hfp (H+) + ∆Hfp (ArO-)] - ∆Hfr (ArOH) ) DPE
The ∆Hfp of the H+ was considered to be 367.2 kcal/ mol,
according to the literature.43,44 Table 1 shows the formation
enthalpies calculated for the studied phenols and their conjugated
bases using a MNDO semiempirical model as well as the DPE
values of the corresponding deprotonation equilibrium reactions.
The calculated DPE values indicate clearly the acidic character
of these compounds but the differences encountered between the
calculated values did not show large variations. This fact was
additionally supported by theoretical calculations made for the
pKa values of the analyte and haptens A and C using suitable
software (see Table 1). All these data demonstrate the validity of
the theoretical model used to choose the hapten which best
mimics our target analyte. In summary, hapten A mimics well the
geometry, charge distribution, and acidic character of the target
analytesparameters that can be considered as appropriate de-
scriptors of its physicochemical behavior.
With all these precedents in mind we addressed to the
preparation of hapten A. A more accurate retrosynthetic analysis
suggested the possibility to start from a commercially available
compound lacking the chlorine atoms in the aromatic ring, but
possessing the appropriate function at the meta position. Thus,
3(3-hydroxyphenyl)-propenoic acid 1 was selected as starting
material. Reduction of the double bound was accomplished by
treating the acid with Na-Pb alloy (10% Na) in basic aqueous
media (5% NaOH).45 Next step consisted on introducing the
chlorine atoms avoiding chlorination of the benzylic position. The
direct chlorination of phenols using Cl2 has been described (g).46
Similarly, the use of N-halosuccinimides (NCS, N-chlorosuccin-
imide) appears in the literature as a source of halonium ions that
react as electrophiles in front of the aromatic ring in solvents with
a high dielectric constant (i.e., DMF) or in the presence of
perchloric acid.47,48 Masilamani and Rogic49 reported the controlled
IMMUNOCHEMISTRY
The 3-(3-hydroxy-2,4,6-trichlorophenyl)-propanoic acid 5 was
conjugated to KLH and to BSA following the mixed anhydride
method. The coupling reaction was verified by analyzing the BSA
derivative by MALDI-TOF-MS. By comparing the observed mo-
lecular weight with that of the intact protein we could estimate
an average of 7.4 haptens covalently attached to each molecule
of BSA, which means a yield ranging from 21 to 25% (considering
that the amount of accessible lysine residues is between 30 and
35). This yield was considered sufficient to start the immunization
protocol. Thus, three rabbits were inoculated with 100 µg of the
5-KLH conjugate and boosted each month for six months until
no significant increase in the antibody titer was observed.
The same hapten was conjugated to HRP and used as an
enzyme tracer (5-HRP). We found it more convenient to use, in
this case, a different coupling procedure to avoid potential
interferences due to side reactions.51 We used the active ester
method and verified conjugation by analyzing the recognition of
the enzyme tracer by the antisera.
Once the necessary immunoreagents had been prepared, we
established the appropriate concentrations by a 2D-checkerboard
titration experiment where the avidities of different dilutions of
the antiserum immobilized on a microtiter plate versus different
concentrations of the enzyme tracer were measured. Using these
immunoreagent concentrations we could obtain three different
competitive immunoassays for each antiserum. Table 2 shows the
features of the direct ELISAs obtained; all of them showed a good
signal and an appropriate assay slope. Similarly, the regression
coefficient of the four-parameter equation used to fit the standards
(46) Carey, F. A.; Sundberg, R. J. Advanced Organic Chemistry; Plenum Press:
New York, 1977.
(47) Challis, B. C.; Challis, J. A. In Comprehensive Organic Chemistry; Barton,
D., Ollis, W. D., Eds.; Pergamon Press: Oxford, 1985; Vol. II, p 1067.
(48) Goldberg, Y.; Alper, H. J. Org. Chem. 1 9 9 3 , 58, 3072-3075.
(49) Masilamani, D.; Rogic, M. M. J. Org. Chem. 1 9 8 1 , 46, 4486-4489.
(50) Corey, E. J.; Letavic, M. A.; Sarshar, S. Tetrahedron Lett. 1 9 9 4 , 35, 7553-
7556.
(51) Gendloff, E. H.; Casale, W. L.; Ram, B. P.; Tai, J. H.; Pestka, J. J.; Hart, L.
P. J. Immunol. Methods 1 9 8 6 , 92, 15-20.
2242 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

Figure 4. Synthetic pathway leading to hapten A (product 5). The hapten is obtained with a global yield around 70% by using this procedure.
Figure 5. The length of the competitive step did not produce any
improvement in the immunoassay performance (Amax/IC50 ratio). A
Figure 6. Variation of the immunoassay features with the ionic
decrease of the incubation time produced an enhancement of the
strength (expressed in mS/cm) of the media. The best immunoassay
detectability but it was accompanied by a significant reduction of the
performance occurs with a conductivity of 8.16 mS/cm. Detectability
immunoassay signal. The data presented correspond to the average
slowly diminishes when the ionic strength is increased. An inflection
of two assays run on different plates. Each standard curve was run
point is produced when the media has a conductivity of 29.1 mS/cm.
in duplicate.
The data presented are extracted from the four-parameter equation
used to fit the standard curve. Data correspond to the average effect
observed for each pH value, on two assays run in different microtiter
plates. Each standard curve was run in duplicate for each assay.
was very good for all assays. From all of them, antiserum 43
showed the best performance regarding detectability (IC50 5.5 µg/
L, 27.83 nM) when compared with the antisera 44 and 45 (IC50
ment of the assay detectability was observed when diminishing
the ionic strength of the media (see Figure 6), indicating perhaps
that ionic species may compete in establishing electrostatic
interactions with the antibody for the target analyte, due to its
polar nature.55 A similar effect has been described for other polar
analytes such as 3,5,6-trichloro-2-pyridinol,56 while opposite results
have been reported for nonpolar analytes, which behavior can be
explained by an enhancement of the hydrophobic interactions
caused by the high ionic strength of the media.57,58 Anyway, it is
worth noting that the assay performed well under a wide range
of ionic-strength values, which is a positive issue regarding the
potential application of this assay to the analysis of complex
matrixes (wastewater, urine, etc.) with high ionic-strength values.
Thus, from 8.16 to 56.1 mS/ cm (5-50 mM in terms of PBS), the
29. 9 and 10 µg/ L, respectively). Thus we continued investigating
the combination antiserum 43/ 5-HRP for further optimization and
evaluation.
IMMUNOASSAY OPTIMIZATION AND EVALUATION
Preincubation Time. The analyte was incubated in the anti-
serum coated plates at different periods of time (0, 10, 20, 40,
and 60 min) before adding the enzyme tracer to proceed with
the competitive step. This strategy has proven to be useful in this
and other laboratories to improve immunoassay detectability;28,35,52
however, in this case no significant variation was observed on any
of the parameters of the assay (i.e., IC50 4.8, 4.2, 4.2, 4.3, 4.4 µg/ L
at different incubation times from 60 to 0 min).
Length of the Competitive Step. As observed with other assays,28
a reduction of the duration of the competitive step led to a
decrease of the assay signal simultaneous with an increase in the
detectability (see Figure 5). Incubation periods around 30 min
were chosen as a compromise between detectability and maximal
absorbance of the assay, although in further experiments we tried
to find out conditions in which the signal could be improved using
short incubation periods, without success.
(52) Weller, M. G.; Weil, L.; Niessner, R. Mikrochim. Acta 1 9 9 2 , 108, 29-40.
(53) Oubin˜ a, A.; Barcelo´, D.; Marco, M. P. Anal. Chim. Acta 1 9 9 9 , 387, 266-
279.
(54) Oubin˜ a, A.; Ballesteros, B.; Galve, R.; Barcelo´, D.; Marco, M.-P. Anal. Chim.
Acta 1 9 9 9 , 387, 255-266.
(55) Jefferis, R.; Deverill, I. In Principles and practice of immunoassays; Price, C.
P., Newman, D. J., Eds.; Stockton Press: New York, 1991.
(56) Manclus, J. J.; Montoya, A. J. Agric. Food Chem. 1 9 9 6 , 44, 3710-3716.
(57) Abad, A.; Primo, J.; Montoya, A. J. Agric. Food Chem. 1 9 9 7 , 45, 1486-
1494.
(58) Abad, A.; Moreno, M. J.; Pelegri, R.; Martinez, M. I.; Saez, A.; Gamon, M.;
Montoya, A. J. Chromatogr., A 1 9 9 9 , 833, 3-12.
Ionic Strength. As reported for other immunoassays,28,53,54 an
increase in the ionic strength produced a concomitant decrease
of the maximal absorbance of the assay. In contrast, an improve-
Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2243

Figure 8. Calibration curve of the optimized 2,4,6-TCP immuno-
assay. The data presented correspond to the average and the
standard deviation of nine assays run on three different plates. The
curves were run in triplicate. See Table 3 for the features of the
optimized immunoassay.
Figure 7. Effect of the pH in the 2,4,6-TCP immunoassay. Several
standard curves were prepared using 5 mM PBS at different pH
values and added to the Ab coated plates. The results reported are
extracted from the four-parameter equation used to fit the standard
curve. Data correspond to the average effect observed for each pH
value, on two assays run in different microtiter plates.
Table 3. Features of the Optimized Immunoassay
Antiserum 43/5HRPa
IC50 varied only from 3.95 ( 0.07 µg/ L (n ) 2) to 6.03 ( 0.5
µg/ L (n ) 2) and the Amax from 0.582 ( 0.034 (n ) 2) to 0.445 (
0.034 (n ) 2) units of absorbance, respectively. With an ionic
strength of 116 (100 mM) mS/ cm, the immunoassay still worked
with acceptable features (Amax) 0.368 ( 0.028; slope 0.90 ( 0.07,
IC50 ) 7.76 ( 0.6 µg/ L, r2 ) 0.997). Because of the higher
detectability encountered and good assay signal, a PBS 5 mM
with an ionic strength of 8.16 mS/ cm was employed for subse-
quent experiments.
Amax
Amin
IC50, µg/ L
slope
dynamic range
LOD, µg/ L
r2
0.501 ( 0.042
0.006 ( 0.003
2.76 ( 0.26
-0.82 ( 0.07
0.53 ( 0.11 to 14.88 ( 1.68
0.22 ( 0.07
0.993 ( 0.003
a
The parameters are extracted from the four-parameter equation
used to fit the standard curve. The data presented correspond to the
average of nine calibration curves run in three different plates. Each
curve was build using three-well replicates.
Concentration of Tween 20. It has been reported that low
concentration values or the absence of detergent may enhance
the affinity of the antibody for analytes such as chlorpyrifos,59
permethrin60 or endosulfan.30 This is why we decided to examine
the influence of Tween 20 in the analytical characteristics of the
2,4,6-TCP immunoassay. An improvement of 38% on the assay
detectability was observed in the absence of the detergent without
affecting the Amax. Since no significant differences were observed
in terms of the coefficient, subsequent experiments were thus
carried out in the absence of the detergent.
Effect of pH. The assay performed better in neutral or basic
media, and it was inhibited below pH 4. The IC50 did not
significantly vary between pH 5.5 and 9.5, but the maximal
absorbance of the assay reached a maximum at pH 6.5 and then
slowly decreased until pH 9.5. Higher pHs inhibited, again, the
assay signal (see Figure 7). Although optimum conditions regard-
ing detectability and maximal signal are placed around pH 7, the
results show that the assay may be useful in a range between 5.5
and 9.5. Similar behavior has been observed with other immu-
noassays for phenolic compounds.53,54 Because of their acidic
character, pH changes cause a shift of the phenol/ phenolate ratio
present in the solution and consequently in the antibody-analyte
equilibrium reaction. The greater tolerance to the basic conditions
(pH > pKa) indicates that participation of the phenolate in the
stabilization of the immunocomplex may be larger than that of
the phenol. These results support, again, the idea that electrostatic
interactions may play an important role in the recognition of these
analytes by the antibody.
Figure 8 shows the standard curve, and Table 3 summarizes
the parameters defining the calibration graph of the immunoassay
antiserum 43/ 5HRP carried out after the evaluation of the above-
mentioned immunoassay characteristics. The assay can be run
in about 1 h in 5 mM PBS and in the absence of detergent. The
limit of detection (90% of the assay response at zero dose) is
0.21 ( 0.06 µg/ L (1.06 ( 0.03 nM), and the working range (20-
80% of the assay response at zero dose) is between 14.88 ( 1.68
and 0.53 ( 0.11 µg/ L (75.29 ( 8.50 to 2.68 ( 0.55 nM).
Immunoassay Specificity. We evaluated the potential interfer-
ence of a group of 23 structurally related compounds by preparing
calibration curves of all of them and measuring them in the assay.
The degree of recognition was directly related to the presence of
the three chlorine atoms at the ortho and para positions as in the
target analyte, and as expected, any change introduced at the
opposite site where the spacer arm was placed affected much more
strongly the binding to the antibody (see Table 4). Thus, 2,3,4,6-
tetrachlorophenol, possessing the three necessary chlorine atoms,
was recognized 86% (considered 100% the recognition of the
analyte), while 2,3,5,6-tetrachlorophenol, lacking the chlorine atom
at the para position and possessing two chlorine atoms at the meta
positions cross-reacted only 3%. Similarly, 2,4,5-trichlorophenol,
mimicking half of the chemical structure of the immunizing
hapten, was recognized 41%, while 2,3,4-trichlorophenol and 2,3,6-
TCP were not recognized at all. Analytes with only two, one, or
no chlorine atoms had a negligible (e3%) or void interference in
the assay. It is also worth noting that blocking the hydroxy group
as an ether reduced drastically the recognition; thus, 2,4,6-
(59) Manclus, J. J.; Montoya, A. J. Agric. Food Chem. 1 9 9 6 , 44, 4063-4070.
(60) Stanker, L. H.; Bigbee, C.; Emon, J. V.; Watkins, B.; Jense, R. H.; Morris,
C.; Vanderlaan, M. J. Agric. Food Chem. 1 9 8 9 , 37, 834-839.
2244 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

Table 4. Interference Caused by Structurally Related
Chemicals, Expressed by Their IC50 and the
Percentage of Cross-Reactivitya
phenolic
cmpds
IC50
brominated IC50
no.b
(nM)
% CR
9.8
86
3
analogues (nM) %CR
5
4
PCP
92.3
PBP
95.6
24
2,3,4,6-TtCP 15.2
2,3,5,6-TtCP 463.1
3
2,4,6-TCP
2,4,6-TCA
2,3,4-TCP
2,4,5-TCP
2,3,5-TCP
2,3,6-TCP
2,4-DCP
15.7 ( 2.59 100
2,4,6-TBP
0.8 1550
>10 000
<0.001
1730
0.5
39.7
41
Figure 9. Graph showing the correlation between the spiked and
the measured values. Blind samples were prepared by spiking milliQ
water with different concentrations of 2,4,6-TCP. The data shown
correspond to the average and the standard deviation of the results
obtained from the immunochemical analysis of the blind samples run
in triplicate during three different days. The dotted line corresponds
to a perfect correlation (slope ) 1).
130.7
12
>10 000
259.5
<0.001
3.5
2
1
2-B-4-CP
2,4-DBP
67.2
33.5
11.1 136
2,5-DCP
2,6-DCP
3,4-DCP
2-CP
4-CP
4-C-3-MeP
4-NP
2666.6
0.7
827.4
2.3
2,6-DBP
291
5.2
>10 000
>10 000
>10 000
>10 000
>10 000
>10 000
<0.001
<0.001
<0.001 4-BP
<0.001
<0.001
<0.001
117.1
14
For the spiked blind samples, a slight overestimation of the
concentration was observed (see Figure 9), as evidenced by the
slope of the linear regression equation that is slightly higher than
1 (slope ) 1.19). The consequence of this is a small increase in
the risk of false positives. As a screening technique it is desirable
to avoid false negatives, and this never occurred. The correlation
between spiked and measured values had a very good coefficient
of correlation (r2 ) 0.992). Certified samples were a mixture of
different analytes as described in the Experimental Section. Table
5 shows the sample composition and the 2,4,6-TCP concentration
value measured by ELISA and by liquid chromatography. The
overestimation observed for sample A was attributed to the
presence of significant concentrations of other chlorinated phe-
nolic compounds such as 2,4-dichlorophenol, pentachlorophenol,
and 2,4,5-trichlorophenol (see above for cross-reactivity). Similarly,
sample B contained pentachlorophenol, 2,4-dichlorophenol, and
2-bromophenol that could also contribute to the ELISA measured
value. Nevertheless, the chromatographic measured values did
not match the target value either, probably because of the
difficulties for those highly polar compounds to be adequately
extracted and quantitated as mentioned in the introductory
section.1,5,61,62
0
phenol
a
Cross-reactivity is expressed as a percent of the IC50 of the 2,4,6-
TCP/ IC50 phenolic compound. b Number of halogens. B, bromo; C,
chloro; Me, methyl; CP, chlorophenol; DCP, dichlorophenol; TCP,
trichlorophenol; TtCP, tetrachlorophenol; PCP, pentachlorophenol; BP,
bromophenol; DBP, dibromophenol; TBP, tribromophenol; PBP, penta-
bromophenol; TCA, trichloroanisol.
trichloroanisol was not recognized in the assay at concentrations
up to 10 000 nM (2.11 mg/ L), even if it had the three chlorine
atoms in the right position. The reason for this lack of recognition
should be found again in the incapability to produce the phenolate
species, which seems to be the one participating the most in the
stabilization of the immunocomplex. As it was demonstrated by
theoretical models, blocking of the hydroxy group produced great
electronic differences with the target analyte. Surprisingly, bro-
minated phenols were highly recognized in this assay, especially
those keeping the geometry of the immunizing hapten, and in
the same degree as the chlorinated analytes. Hence, 2,4,6-tri-
bromophenol cross-reacted 1550% and 2,4-dibromophenol 136%.
When the bromine atom in para was a chlorine, the crossreactivity
dropped to 33%. If only the ortho positions had bromines the
interference was only 5.2%. In every case the recognition of the
brominated analytes was always greater than that of the chlori-
nated analyte analogues (see Table 4). We have not yet found an
explanation for this behavior; that will be the object of further
studies using computer models and other series of halogenated
analytes.
CONCLUSIONS
Molecular geometry and charge distributions are important
parameters determining the ability of these analytes to establish
noncovalent binding forces. For ionized molecules, it is also
interesting to consider the changes produced in the above
physicochemical parameters as a consequence of this property.
An initial knowledge of some of these features helps the immu-
nochemist to rationalize hapten design. Thus, using molecular
Immunoassay Accuracy. To perform this study, 10 blind samples
and two certified samples were analyzed with the optimized ELISA.
Table 5. Results Obtained from the Analysis of Certified Samples
composition (µg/ L)a
2,4,6-TCP results
chromatography
203.3
2.15
2,4,6-TCP
sample
2-BP
2.59
2-CP
2,4-DCP
PCP
149
2,4,5-TCP
139
4-CP
4.2
(target value)
ELISA
A
B
113
1890
99.1
1.27
123
2.98
244
3.88
2.14
a
Only the concentration values of those analytes potentially interfering in the assay are reported in the table. Sample A was provided by Yorkshire
Environmental. Sample B is from Aquacheck.
Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2245

modeling tools we have been able to develop an immunoassay
for the detection of 2,4,6-TCP, reaching reasonable detectability
limits, and have found explanations for the behavior of the assay.
Thus, it has been demonstrated by theoretical calculations that
blocking the phenolic group in order to introduce a linker is a
bad strategy for raising antibodies against a phenolic compound.
Similarly, analytes where this function is protected such as the
corresponding anisole, are not recognized by the antibodies raised.
The ELISA reported can simultaneously process many samples
in about 1 h. The assay is reliable and shows accuracy in the
measurement of 2,4,6-trichlorophenol in water that is similar to
that of the chromatographic method. Only brominated phenolic
compounds have shown a significant interference in the assay,
although their presence in environmental water samples has rarely
been reported.63
ACKNOWLEDGMENT
This work has been supported by the EC Programs (Contracts
ENV4-CT97-0476 and IC15-CT98-0910) and by CICYT Especial
Actions (AMB98-1640-CE and ALI99-1379-CE). We thank Dr.
Josefina Casas (Animal Breeding Facility of the CSIC), Dr. Oriol
Bulbena, M.-Angeles Garc´ıa and Manel Gonza´lez (Laboratorios
Vin˜ as, Barcelona, Spain) for their kind assistance in animal
treatments. We also would like to thank Professor Damia` Barcelo´
(IIQAB-CSIC, Barcelona, Spain) for his advice and for supplying
certified water samples. Likewise, we are also pleased with the
interesting discussions and advice provided by Dr. Lluis Julia`
(IIQAB-CSIC, Barcelona, Spain) regarding chlorination proce-
dures.
Received for review November 19, 1999. Accepted
February 8, 2000.
(61) Castillo, M.; Puig, D.; Barcelo´, D. J. Chromatogr. 1 9 9 7 , 778, 301-311.
(62) Jahr, D. Chromatographia 1 9 9 8 , 47, 49-56.
(63) Nomani, A. A.; Ajmal, M.; Ahmad, S. Environ. Monit. Assess. 1 9 9 6 , 40, 1-9.
AC991336Y
2246 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000