Detection of Chemicals by a Reporter Immunoassay: Application to

Jun 10, 2004 - of the antibodies toward the reporter. Fluoride was used as a model, and silylated estradiol was used as a reporter. Upon silylation wi...
0 downloads 0 Views 59KB Size
Download PDF
Anal. Chem. 2004, 76, 4286-4291

Detection of Chemicals by a Reporter Immunoassay: Application to Fluoride Marie-Astrid Sagot,† Florence Heutte,† Pierre-Yves Renard,‡ Fre´de´ric Dolle´,§ Philippe Pradelles,† and Eric Ezan*

CEA, Service de Pharmacologie et d’Immunologie, Service de Marquage Mole´ culaire et de Chimie Bioorganique and IRCOF, UMR CNRS6014, 76131 Mont St-Aignan, and CEA, Service Hospitalier Fre´ de´ ric Joliot CE-Saclay, 91191 Gif-sur-Yvette, France

This report describes a concept in which an immunoassay is used indirectly to quantify a nonantigenic very low molecular weight compound participating in a chemical reaction with a haptenic reporter. The detection limit of each reagent is, therefore, governed only by the affinity of the antibodies toward the reporter. Fluoride was used as a model, and silylated estradiol was used as a reporter. Upon silylation with N-O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) or N-O-bis(dimethylterbutylsilyl) trifluoroacetamide (MTBSTFA), estradiol is no longer recognized by antibodies specific to estradiol. After reaction with hydrofluoric acid (HF) or fluoride salts (KF, CsF, NaF), its immunoreactivity is restored, and native estradiol is formed and is detected by immunoassay. The level of synthesized estradiol is dependent on the concentration of fluoride. A fluoride detection limit of 0.3 µg/L (15 nM) is obtained. Potential interference with other acids has been eliminated by choosing the silyl group (trimethylsilyl vs tert-butyldimethylsilyl) and by selecting optimal reaction conditions for the desilylation. The method has been applied to the detection of fluoride salts in natural waters (range 0.28-9.0 mg/L) and in an atmosphere artificially contaminated with HF between 8 and 160 µg/m3 in the parts-per-billion range. This indirect immunoassay combines simplicity and high sensitivity and, therefore, can be used in field monitoring. Finally, the extension of the concept to other chemicals is discussed. Immunoassays were initially designed to detect large molecules such as proteins and natural antigens in complex biological media. Yet, since antibodies can be raised specifically against small molecules, they have subsequently been used to measure a wide range of biological and chemical compounds. However, antibodies raised against compounds with a molecular weight under 200250 Da have limited physical interactions, which reduces their affinity, thus limiting their suitability for the development of sensitive immunoassays.1 For instance, whereas affinities in the nanomolar range are commonly encountered for steroids (mo* To whom corespondence should be addressed. Mailing address: Service de Pharmacologie et d’Immunologie, Baˆt 136, CE Saclay, 91191 Gif-sur-Yvette, France. Phone: 33-1-69-08-73-50. Fax: 33-1-69-08-59-07. E-mail: [email protected]. † CEA, Service de Pharmacologie et d’Immunologie. ‡ Service de Marquage Mole ´ culaire et de Chimie Bioorganique and IRCOF. § CEA, Service Hospitalier Fre ´ de´ric Joliot CE-Saclay.

4286 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

lecular weights, MW, in the range of 250-300 Da), they decrease to the micromolar range for smaller structures.1-4 The challenge of extending immunoassays to small inorganic or organic analytes has led to original strategies, which involve specific immunogen preparation and sample pretreatment. The size of the analyte is artificially increased in order to transform it into a larger compound with more antibody-binding sites. For instance, this has been described for histamine and methyl phosphonic acid for which chemical derivatization mimics the structure of the immunogen.5-7 In the same way, immunoassays for metals have been described with antibodies obtained after immunization of the metals bound to EDTA or glutathione.8,9 In all these examples, the apparent affinity of the antibodies is increased by a magnitude of more than 3 logs, and the immunoassay’s limit of detection reaches the nanogram-per-milliliter range. Yet these approaches apply only to compounds that can be easily derivatized chemically in biological or environmental samples and are, therefore, restricted to a few specific analytes. Here, we propose an alternative in which the analyte to be assayed modifies the affinity of an antigen and, thus, provides the basis for an indirect immunoassay of the analyte. This is supported by the fact that antibodies are restricted to structural features borne by the antigen, minor chemical modifications of which, such as the introduction of hydroxyl, methyl or silyl groups, may significantly decrease or even abolish antibody recognition. Therefore, when an antigen is chemically modified by reaction with an organic compound or a catalyst, it loses its capacity to bind to a specific antibody. Inversely, a nonantigenic structure can be rendered antigenic after chemical modification. For a particular chemical reaction, it is expected that the detection (1) Chappey, O.; Debray, M.; Niel, E.; Scherrmann, J. M. J. Immunol. Methods 1994, 172, 219-25. (2) Brimfield, A. A.; Hunter, K. W., Jr.; Lenz, D. E.; Benschop, H. P.; Van Dijk, C.; de Jong, L. P. Mol. Pharmacol. 1985, 28, 32-39. (3) Cailla, H. L.; Racine-Weisbuch, M.S.; Delaage, M. A. Anal. Biochem. 1973, 56, 394-407. (4) Harrison, R. O.; Brimfield, A. A.; Nelson, J. O. J. Chem. 1989, 958-964. (5) Ashley, J. A.; Lin, C. H.; Wirshing, P.; Janda, K. D. Angew. Chem., Int. Ed. 1999, 38, 1793-1795. (6) Waldum, H. L.; Sandvik, A. K.; Brenna, E.; Schulze-Sognen, B.; Scand. J. Gastroenterol. Suppl. 1991, 180, 32-39. (7) Morel, A.; Darmon, M.; Delaage, M. Mol. Immunol. 1990, 27, 995-1000. (8) Blake, D. A.; Jones, M.; Blake, R. C., II; Pavlov; A. R.; Darwish, I. A.; Yu, H. Biosens. Bioelectron. 2001, 16, 799-809. (9) Darwish, I. A.; Blake, D. A. Anal. Chem. 2002, 74, 52-58. 10.1021/ac030395f CCC: $27.50

© 2004 American Chemical Society Published on Web 06/10/2004

of the antigen may be proportional to the concentration of the entities involved in the chemical transformation, whatever their size and structure. If low molecular weight compounds participate in the reaction, their quantification by indirect immunoassay is, therefore, possible. In this report, and as a model, we present an indirect immunoassay for fluoride (MW 19) based on its ability specifically to hydrolyze silylated alcohols. Fluoride was chosen for at least two reasons: first, it appeared to be a good model, since its size is far below that of the smaller compounds for which immunoassays have been described, and second, fluoride in the atmosphere is a pollutant requiring fast and sensitive monitoring which is a potential characteristic of immunoassays. Hydrofluoric acid (HF) is used for several industrial applications, such as the industrial manufacture of aluminum and plastics and in the nuclear industry. It is highly corrosive and causes health hazards, such as irritation, skin and lung burning, and alteration of bone metabolism. Although environmental methods for HF determination exist,10-19 an immunoassay that is able to combine sensitivity (atmospheric concentrations in the range of 0.08-1.2 mg/m3) and simplicity represents an attractive alternative. In this report, we describe the design and optimization of an extremely sensitive assay of both HF and fluoride salts. EXPERIMENTAL SECTION Reagents. Salts, acids, and solvents (HPLC grade, purity>99.5%) were from Sigma (St Louis, Mo) or Merck (Darmstadt, Germany). Natural waters (Vittel, Saint-Yorre, Saint-Amand, San Pellegrino, Arcens, Talians, Badoit, Celestin, Vauban) were obtained at the local department store. Zymafluor (Laboratoires Richard, Sauzet, France) was purchased at the local pharmacy (Saclay, France). Estradriol was from Sigma. Monoclonal antibody against estradiol was produced by conventional hybridization techniques after immunization of mice with an estradiol-bovine serum albumin conjugate in position six of estradiol (conventional steroid numbering). The antibody and its use in enzyme immunoassay for estradiol detection have been described elsewhere.20 The tracer was synthesized by coupling a 6-carboxymethyl estradiol derivative to acetylcholinesterase (AChE).21 The purified enzyme was obtained from Spi-Bio (Massy, France). Enzyme activities were measured using Ellman’s reagent (Spi-Bio, Massy, France), an AChE substrate comprising 2.2 g of acetylthiocholine and 1.0 g of dithionitrobisbenzoic acid (Sigma) in 200 mL of 0.05 M phosphate buffer, pH 7.4. Hydrofluoric acid (HF) is one of the (10) Gomez-Gomez, M.; Palacios Corvillo, M. A.; Rica, C. C. Analyst 1988, 113, 1109-1112. (11) Jones, P. Anal. Chim. Acta 1992, 258, 123-129. (12) Oszwaldowski, S.; Lipka, R.; Majewski, T.; Jarosz, M. Analyst 1998, 123, 1529-1533. (13) Moritz, W.; Bartholoma¨us, L.; Roth, U.; Filipov, V.; Vasiliev, A.; Terentjev, A. Anal. Chim. Acta 1999, 393, 49-57. (14) Yuchi, A.; Sakurai, J.; Tatebe, A.; Hattori, H.; Wada, H. Anal. Chim. Acta 1999, 387, 189-195. (15) Bayon, M. M.; Rodriguez Garcia, A.; Garcia Alonso, J. I.; Sanz-Medel, A. Analyst 1999, 124, 27-31. (16) Li, H. B.; Chen, F. Fresenius’ J. Anal. Chem. 2000, 368, 501-504. (17) Itai, K.; Tsunoda, H. Clin. Chim. Acta 2001, 308, 163-71. (18) DiCesare, N.; Lakowicz, J. R. Anal. Biochem. 2002, 301, 111-116. (19) Kaiser, E.; Rohrer, J.; Campbell, F. J. Chromatogr., A 2003, 997, 259-267. (20) Buscarlet, L.; Grassi, J.; Cre´minon, C.; Pradelles, Ph.; Dupret-Carruel, J.; Jolivet, M.; Mons, S. Anal. Chem. 1999, 71, 1002-1008. (21) Porcheron, P.; Morinie`re, M.; Grassi, J.; Pradelles, P. Insect Biochem. 1989, 19, 117-119.

strongest and most corrosive acids known. HF burns penetrate deeply into skin and muscle tissue and cannot be treated by simply flushing the area with water. Therefore, special safety precautions are necessary when using this chemical. Solutions should be handled in an approved fume hood or ventilated place. Protective gloves (natural rubber or neoprene) and eye protection should be used. The materials should not be taken internally. Washing hands and gloves frequently with water is wise when working with even dilute HF. Synthesis of Silylated Estradiol. To demonstrate the feasibility of the approach, estradiol (1 g/L in DMF) was silylated in the presence of 40 µL of N-O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) or N-O-bis(dimethylterbutylsilyl)trifluoroacetamide (MTBSTFA) for 20 min at room temperature. Derivatives were not further purified or analytically characterized. Silylation was demonstrated indirectly by loss of immunoreactivity with antibodies specific to estradiol (cross-reaction less than 1%). In final assay developments, estradiol-17-dimethyl-tert-butylsilyl ether (E17Si) was synthesized according to a literature method.22 Dimethyl-tertbutylsilyl chloride (110 mg) was added to a solution of estradiol (60 mg) in DMF (5 mL), which was then shaken for 1 h at room temperature. The solvent was evaporated, and the residue was purified by preparative thin-layer chromatography using cyclohexane-AcOEt (4:1) as developing solvent to yield 60 mg of estradiol-bis-3,17-dimethyl-tert-butylsilyl ether. Tetra-n-butylammonium fluoride (80 mg) was added to 60 mg of the disilylated compound dissolved in 10 mL of tetrahydrofuran, and the reaction mixture was heated for 4 h at 50 °C with shaking. The reaction mixture was then diluted with AcOEt, washed with water, and dried with anhydrous sodium sulfate. Purification was carried out by chromatography on silica gel using diethyl ether-hexane (7: 3) for the elution yielding 30 mg (45% global yield) of E17Si, mp 128-130°, 1H NMR (CDCl3, conventional steroid numbering): δ 0.79 (3H, 18-CH3), 0.93 (9H, Si-C4H9), 1.19-1.51 (12H), CH2), 2.3 (1H, CH), 2.80 (2H, CH), 3.73 (1H-17RCH), 6.56 (1H, 1-H), 6.63 (1H, 2-H), 7.14 (1H, 4-H). Using a previously described immunoassay,20 anti-estradiol monoclonal antibody showed a cross-reaction of 0.3% with E17Si. Enzyme Immunoassay. The assays were performed in 96well microtiter plates (ImmunoModule F8 Maxisorp, Nunc, Denmark) coated at 5 mg/L in 50 mM phosphate buffer pH 7.4 (one night at 22 °C) with goat polyclonal antibodies specific for mouse immunoglobulins (Jackson ImmunoResearch, ref 115-005044, West Grove, PA). Before use, coated plates were washed three times with 0.01 M phosphate buffer pH 7.4 containing 0.05% Tween 20 (washing buffer, 300 µL/well and three wash cycles) (Autowasher 96, Labsystems, Eragny, France). Tracer and antiserum were diluted in 0.1 M phosphate buffer pH 7.4 with 0.15 M NaCl, 5 mM EDTA, 0.1% bovine serum albumin, and 0.01% sodium azide (assay buffer). The assay was performed in a total volume of 150 µL. Reagents were dispensed as follows: 50 µL of standards or sample, 50 µL of enzymatic tracer, and 50 µL of antiserum. After incubation at room temperature for 1 h, the plates were washed three times as described above, and Ellman’s reagent (200 µL) was dispensed into each well and incubated in the dark without shaking. After 1 to 2 h of enzymatic reaction, the plate was read at 414 nm (Multiskan RC, Labsystems, Eragny, France). (22) Wilson, N. S.; Keay B. A. Tetrahedron Lett. 1997, 38, 187-190.

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4287

All measurements for standards and samples were made in duplicate. Assay Protocol and Optimization. Optimization of assay conditions and desilylation were performed by modifying various parameters: the nature of silyl groups (trimethylsilyl vs tertbutyldimethylsilyl), the concentration of silylated estradiol (between 2 and 4000 µg/L), the nature and percentage of solvent during the desilylation (ethanol, methanol, DMF, or DMSO at 50 or 95% in H2O), the reaction time (5 min to 6 h) and temperature (20-95 °C) of desilylation. Optimized conditions led to the following protocol. Silylated estradiol (E17Si) was diluted at a concentration of 400 µg/L in 95% DMSO in water (50 µL) containing EDTA (10 mM) and was reacted with the desilylating agent (50 µL in 95% DMSO) for 30 min at 65 °C. After desilylation, the mixture was diluted 10-fold in assay buffer, and estradiol was measured by enzyme immunoassay. Since, and as described previously, the sample is also diluted three times with the antibody and the tracer, the final dilution of the solvent and estradiol is 30. The extent of desilylation was indirectly estimated by the ability of the estradiol formed to inhibit binding between the enzymatic tracer and antibody. This inhibition was plotted versus the concentration of the desilylating agent. Application to Samples. The assay was applied to commercial products containing known concentrations of fluoride (natural waters and a tooth remineralization product containing NaF). Furthermore, to mimic measurement of atmospheric fluoride, HF at various concentrations was placed (50 µL) at the bottom of a 750-mL Falcon bottle. A 10-cm polypropylene tube (i.d. 1 mm) was then connected to the bottle from which the air was aspirated (flow rate, 4 L/hour) for 4 h and passed through 0.5 mL of 0.05 M phosphate buffer pH 5.5 in a 1-mL polypropylene tube. HF was completely evaporated in order to obtain artificial atmospheric concentrations between 8 and 160. HF recovery was measured from a standard curve containing known concentrations of HF. RESULTS Principle of the Assay. The aim of the study was to demonstrate that antibodies may be used to monitor chemical reactions in which the antigenicity of one reagent is restored after reaction. This can lead to an indirect sensitive measurement of low molecular weight nonimmunogenic compounds participating in the reaction process. As an application, we chose fluoride, which is known to remove silyl groups, such as trimethylsilyl or dimethyl terbutyl silyl ethers used to protect hydroxyl groups. Silylation of hydroxyl, amino, or carboxylic acid functions is a common derivatization procedure used to increase the volatility and hydrophobicity of organic compounds before their quantification by gas chromatography or as a protective group in chemical synthesis.23,24 For instance, during the silylation process for alcohols, an active hydroxyl group undergoes etherification by a silyl group such as is present in BSTFA or MTBSTFA. The stability to hydrolysis of the newly introduced silyl ether function is closely related to structural and steric features of the molecule. The stability scale of the formed silyl ether in acidic conditions and with respect to temperature has been estimated as follows: tertiary silyl ethers > secondary silyl ethers . primary silyl ethers (23) Li, D.; Park, J.; Oh, J. R. Anal.Chem. 2001, 73, 3089-3095. (24) Halket, J. M.; Zaikin, V. G. Eur. J. Mass. Spectrom. 2003, 9, 1-21.

4288 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

Figure 1. Principle of the assay. Excess silylated estradiol, here estradiol-17-dimethyl-tert-butylsilyl ether (E17Si), is reacted with fluoride or hydrofluoric acid. Estradiol is formed and detected by a competitive enzyme immunoassay. Antibodies specific to estradiol with only slight recognition (0.5%) of E17Si are used. The part of estradiol which is supposed to be recognized20 by the antibodies and the position of the immunogen are indicated (a and b, respectively). Controls are performed in the absence of fluoride. The immunoassay signal is inversely related to the fluoride concentration.

> phenolic silyl ethers.25,26 Among the desilylation reagents, fluoride-bearing agents have the highest reactivity, related to the high affinity of fluoride toward silicon atoms. For instance, HF molecule can attack the Si-O bond on any surface configuration. As a consequence, both the Si-O bonds and the H-F bond will break, restoring an alcohol and an Si-F compound. With the aim of detecting hydrofluoric acid and as a model, we chose silylated estradiol, which might undergo transformation into estradiol after interaction with HF. The scheme of the reaction is depicted Figure 1. With an antibody which largely differentiates silylated estradiol and estradiol, it is expected that the detection of estradiol is directly related to the quantity of HF reacting with silylated estradiol. The sensitivity of the detection of HF was investigated by studying the effects of the nature of silylated groups, the interference of solvents, and the conditions of the chemical reaction. Demonstration of the Concept. Since the specificity of antibodies to steroids is driven by the coupling position of the hapten to the carrier protein, we used a monoclonal antibody produced after immunization of mice with bovine serum albumin conjugated to estradiol at position 6.20 Modifications at position 17 in structures such as ethynylestradiol, estriol, or estrone virtually abolished the immunoreactivity (cross-reaction of 1% or less) and indicated an epitope located roughly opposite the phenolic ring. Silylation of both estradiol hydroxyl groups (phenol at position 3 and secondary alcohol at position 17) was initially performed with the silylating agents BSTFA or MTBSTFA following protocols which have been extensively used for the analysis of steroids by gas chromatography/mass spectrometry.23,24 The two compounds showed cross-reactivities of 0.5%, indicating a loss of immunoreactivity following silylation, which had occurred at least at position 17. The possibility of cleaving the O-Si bond of silylated estradiol was demonstrated by reaction with 1 N HCl or HF for 15 min, which completely restored the immunogenicity of modified estradiol, since standard curves constructed with either estradiol or with HCl-reacted silylated estradiol were identical (data not shown). (25) Greene, T. W.; Wuts, P. G. M. Protective groups in organic synthesis, 3rd ed.; Wiley & Sons: New York, 1999. (26) Kocienski, P. J. Protective Groups; Enders, R., Noyori, R., Trost, B. M., Eds.; Thieme: Stuttgart, 1994.

We found that the yield of desilylation and the solubility of silylated estradiol were favored in the presence of a solvent. To be able to perform a straightforward assay and with the goal of routine applications, such as field testing, it was necessary to choose a solvent that was compatible with the buffer used for the immunological detection, thus avoiding any extraction step. The first experiments were therefore performed with dimethylformamide, for which concentrations below 10% induced only weak interference with tracer-antibody binding. Since the HF detection was in a context of atmospheric analysis, it was necessary to determine in parallel whether other volatile components could interfere with the assay. We tested acids known to be present at high concentrations and also likely to remove silyl groups. Compared to BSTFA silylated estradiol, the better selectivity of MTBSTFA silylated estradiol in acidic conditions was demonstrated with hydrochloric acid, being 100-fold less potent than HF (data not shown). The same result was obtained with sulfuric and nitric acids. To improve the specificity, we investigated desilylation conditions in buffered or basic media and according to various conditions of time and temperature. The desilylation process was optimized by performing the reaction in potassium phosphate buffer 50 mM, pH 5.5 for 1 h at 56 °C. Under these conditions, none of the strong acids (HCl, H2SO4, and HNO3 up to concentrations of 1 mM) effected desylilation of silylated estradiol, and no interference with the assay of HF was detected. At concentrations of 1 mM, nucleophiles such as 4-nitrophenol; 4-nitroimidazole; tetrabutylamine salts; and inorganic halogens, such as potassium iodide and sodium iodide, did not interfere with the assay. As expected, fluoride salts NaF, CsF, or KF were comparable to HF in their potency, since HF in solution is neutralized into fluoride. Optimization of the Assay. Following these initial developments, several parameters were extensively studied in order to sensitize the assay. These experiments were performed with estradiol monosilylated at position 17 with a tert-butyldimethylsilyl group (E17Si), which presented a cross-reactivity of 0.3% compared to that of estradiol. As indicated above, we found that the desilylation efficacy for this compound was also dependent on the nature and concentration of the solvent. To optimize this parameter, we first investigated the potential effect of water-miscible solvents on the estradiol immunoassay characteristics. As shown in Table 1, tracerantibody binding was only moderately affected when the organic solvents tested were used at final concentrations of 1.7 or 3.2%. These concentrations were chosen because they corresponded to the final solvent concentrations obtained after a 30-fold dilution of crude desylilation reaction mixture in either 50 or 95% of solvent. Ethanol or methanol least interfered with antibody binding, whereas DMSO and DMF had a more significant effect on the binding and the affinity of the antibody. We compared the yield of the desylilation reaction using these solvents under various conditions of time (5 min to 6 h) and temperature (25, 65, or 95 °C). E17Si (5 µg/L) was incubated with potassium fluoride at 10 µM in various solvents. After reaction, the mixtures were diluted 10-fold in assay buffer, and the efficiency of the desilylation was determined by measuring the estradiol released. No measurable desilylation was observed with methanol or ethanol. In contrast, estradiol was effectively formed in the presence of the highest

Table 1. Effect of the Solvent on Tracer-Antibody Binding and Estradiol Assay Sensitivity solvent assay buffer methanol ethanol DMF DMSO

%a

antibody-tracer bindingb

IC50c (ng/L)

1.7 3.2 1.7 3.2 1.7 3.2 1.7 3.2d

0.95 0.90 0.95 0.95 0.82 0.81 0.73 0.65 0.48

46 39 43 82 75 105 160 50 82

a Final concentration of the solvent in the estradiol immunoassay. Enzymatic activity in absorbance units, expressing tracer-antibody binding. c Concentration of estradiol inhibiting 50% of initial tracerantibody binding. d Condition chosen in the final assay. b

Figure 2. Calibration curve and cross-reaction profiles of the indirect fluoride immunoassay. Each point of the standard curve is the mean ( SD of five independent experiments performed on 5 different days. Five microliters of fluoride (KF) were reacted with 95 µL of 200 µg/L of E17Si in 95% DMSO, and estradiol was measured by enzyme immunoassay after 10-fold dilution of the desilylation crude reaction mixture in assay buffer. The presence of estradiol resulted in inhibition of initial tracer-antibody binding (reference, 100% in the absence of fluoride). The concentrations indicated are the initial concentrations of KF in reaction with E17Si. The curve for HCl is representative of the other tested acids (sulfuric, perchloric, nitric, TFA). Formic and acetic acids were not reactive at a concentration of 10-3 M.

concentrations of DMF and DMSO, with a maximal desilylation yield in the range of 20-25% for DMSO. Therefore, DMSO was chosen because the desilylation reaction was faster and interfered less with the assay. Desilylation reaction temperatures of 65 or 95 °C were optimal whatever the reaction time. We tested assay sensitivity by varying the concentration of silylated estradiol and found that the optimal initial concentration was 400 µg/L. If HF is reacted with a lower concentration of E17Si, it may be expected that the production of estradiol will be too low to induce a significant inhibition of tracer-antibody binding. In contrast, higher concentrations may interfere with the final assay as a result of cross-reactivity with E17Si. Using these optimized conditions, we found an increase in assay sensitivity of more than 100-fold compared to the initial assays performed in 10% DMF. On the basis of the concentration able to induce a 10% shift of the signal found in the absence of fluoride (HF or KF), a fluoride detection limit of 0.3 µg/L (15 nM) (mean of five independent experiments) was obtained (Figure 2). Under these new conditions, stability toward acids was checked again. The strong acids tested (HCl, HNO3, HClO4, H2SO4, trifluoroacetic acid) were able to remove E17Si silyl ether only at Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4289

Table 2. Analysis of Fluoride Content in Commercial Natural Mineral Waters brand Saint-Yorre Celestin Vauban Saint-Amand Arcens Badoit San Pellegrino Talians Vittel mean ( SD

concna mg/L

foundb mg/L

recovery %

9 6 2 2 1.3 1 0.61 0.35 0.28

11 5.5 2.6 1.4 0.91 0.95 0.65 0.32 0.30

122 92 130 70 70 95 107 91 107 98 ( 21

a Indicated on the label. b Intra- and interprecision of 15-20%; n ) 2-4.

a concentration of 1 mM, that is, a potency at least 5000-fold below that of fluoride (Figure 2). None of the other chemicals (monoand bivalent salts; weak acids, such as formic or acetic acids) tested at the concentration of 1 mM were effective. Applications. The assay was used to evaluate fluoride concentrations in nine natural waters (range 0.28-9 mg/L) which were diluted 50- to 1000-fold before assay. The mean recovery was 98 ( 21% (Table 2). After dissolution in water, we assessed Zymafluor, a drink containing 1.14 g/L of NaF which is used as fluoridated complement to prevent dental caries. The mean of five different dilutions performed in duplicate indicated a concentration of 1.0 ( 0.11 g/L. As described in the Experimental Section, a special device was also manufactured in order to create artificial concentrations of HF between 10 and 200 ppb. Air was collected in 0.5 mL of 50 mM phosphate buffer pH 5.5, and the medium was incubated with silylated estradiol in order to be assayed. The mean recovery was 63% when artificial concentrations were in the range of 8 to 160 µg/m3. Since the assay is able to detect 0.3 µg/L of HF, the potential limit of detection is close to 8 ng/m3 if 100 L of HF-contaminated air is pumped through the recuperation medium. DISCUSSION We have described an indirect immunoassay that specifically detects low concentrations of fluoride. Our approach is based on monitoring the time-course of a chemical reaction between HF and a reporter immunogenic organic compound of higher molecular weight. This principle may be applied to numerous chemical reactions in which a reagent may be detected by immunoassay. Owing to the infinite nature of compounds that may participate in any chemical transformation, our approach may be extended to inorganic or low-molecular-weight organic compounds inaccessible to classical detection by specific antibodies. Noteworthily, this could also be used when the targeted analyte is the catalyst of the reaction, since for most catalyzed reactions, the reaction rates are proportional to catalyst concentration. The potential interest is that the level of detection is, thus, governed by the affinity of the antibody for the reporter, which traditionally allows sensitivity in the range of 0.1 to 10 nM when used in immunoassays. Various assay formats are possible, since antibodies to either the appearing or disappearing entity may be used. In our example, a competitive immunoassay was used, but an 4290 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

immunometric format in which the modified antigen is absorbed on a solid phase may also be used with a potentially higher sensitivity. The conditions for which this approach is feasible are (1) the pattern of antibody recognition is favorable to one of the reagents; (2) the antibody affinity allows sensitive detection; (3) the chemical conditions (time, temperature, pressure) for which the reaction is possible do not require special equipment in order to be adaptable to routine or field analysis; (4) the reaction medium is compatible with the immunological reaction; and (5) the yield of the chemical reaction is satisfactory, and the appearance or disappearance of the antigen is proportional to the concentration of the analyte for which the assay is intended. On the basis of these considerations, it seems obvious that the main parameters governing the sensitivity of the assay are reaction yield and antibody affinity. Theoretically, the analyte detection limit is that of the immunoassay. In our demonstration, we found that under the reaction conditions, the desilylation yield was not optimized and was limited to 20-30% but nonetheless gave a detection limit of ∼15 nM, taking into account a sensitivity of the estradiol immunoassay of 0.3 nM and a dilution factor necessary to reduce the influence of the solvents used for the desilylation. A second parameter is the respective affinities of the chemical components. If the reagent that is likely to be modified has a cross-reactivity up to 1%, a sensitive detection is always possible, providing the reagent concentrations are optimized. In this case, it is conceivable that assay sensitivity will be improved if modification of the reagent occurs with the highest possible yield. Finally, the solvent used for the reaction should be reasonably compatible with the immunological reaction. Although it may be removed by extraction, this step decreases the rapidity and simplicity of the method for specific applications, such as field testing. In the latter case, alternatives to immunoassay, such as field-portable gas chromatography, may be used. Nevertheless, the use of water-miscible solvents as demonstrated here and elsewhere does not hamper immunoassays, providing their concentration is adjusted to minimize their effect on antibody binding.27,28 It should be pointed out that for some ligands, binding reactions may also occur with non-water-miscible solvents.29,30 Our work was intended not only to describe a particular application of the immunoassay but also to propose a potential application in environmental monitoring. According to the Toxic Release Inventory of the U.S. Environmental Protection Agency,31 HF is the sixth most common toxic pollutant, with 67 million pounds released into the atmosphere. On the basis of this report, HF represents about one-tenth of the acids (mainly hydrochloric and sulfuric acids) released by industry. Different methods have been described or are commercially available to analyze fluoride in biological materials or environmental samples. The Occupational Safety and Health Administra(27) Russell, A. J.; Trudel, L. J.; Skipper, P. L.; Groopman, J. D.; Tannenbaum, S. R.; Klibanov, A. M. Biochem. Biophys. Res. Commun. 1989, 158, 80-85. (28) Matsuura, S.; Hamano, Y.; Kita, H.; Takagaki, Y. J. Biochem. 1993, 114, 273-278. (29) Aston, J. P.; Hitchings, E. J.; Ball, R. L.; Weeks, I.; Woodhead, J. S. J. Immunoassay 1997, 18, 235-246. (30) Weetall, H. H. J. Immunol. Methods 1991, 136, 139-142. (31) U.S. Environmental Protection Agency; Toxic Release Inventory database; accessed June 30, 2001.

tion adopted an ion-specific electrode procedure with a detection limit of 1 mg/L.32 Published methods use fluorometry, spectrophotometry, ion chromatography, or molecular absorption spectrometry.10-19 Their detection limits range from 0.2 to 5000 µg/L. Fluorometric methods appear to be the most sensitive but suffer from marked interferences by several anions or cations, such as copper, iron, and aluminum and, thus, require a separation step. Compared to these methods, the immunoassay presented here has several advantages. First, it is highly sensitive and free of interference by other ions. Second, it is rapid (desilylation and enzyme immunoassay may be performed in less than 2 h) and does not require special technology, since the colorimetric (32) Occupational Safety and Health Administration. Analytical Laboratory, Method No. ID-110. OSHA Manual of Analytical Methods, December 1988 (revised Feb. 1991).

detection can be performed visually or by a portable spectrophotometer for field applications. The minimal requirement for sample volume is