Broad-Specific Antibodies for a Generic Immunoassay of Quinolone

Apr 1, 2009 - Freund's adjuvant (IFA), and complete Freund's adjuvant (CFA) were purchased from Sigma Chemical Co. 3,3′,5,5′-Tetramethyl- benzidin...
0 downloads 0 Views 1MB Size
Download PDF
Anal. Chem. 2009, 81, 3246–3251

Broad-Specific Antibodies for a Generic Immunoassay of Quinolone: Development of a Molecular Model for Selection of Haptens Based on Molecular Field-Overlapping Limin Cao,† Dexin Kong,‡ Jianxin Sui,† Tao Jiang,§ Zongyan Li,† Lei Ma,† and Hong Lin*,† Food Safety Laboratory and School of Medicine and Pharmacy, Ocean University of China, Qingdao, 266003, P. R. China, and Shandong Provincial Research Center for Bioinformatics Engineering and Technique, Center for Advanced Study, Shandong University of Technology, Zibo 255049, P. R. China A new molecular model for quinolone haptens was developed based on molecular field-overlapping. The quanlitive modeling of 3-D conformations showed that the conformation difference among quinolones is caused mainly by the different substitutes at the 1 and 7 positions. The 8-substitute also showed some effect by its inter-reaction with the 1-substitute. The conformational similarity of 27 quinolones to each other was for the first time calculated and exploited for a selection of haptens according to desired broad specificity of corresponding antibodies. The developed model was preliminarily validated with antibodies against different quinolones. A significant positive correlation (R ) 0.7793) was observed between calculated overlapping coefficients of haptens and the cross-reactivity of corresponding polyclonal antibodies (Pabs), which confirmed the overall accuracy of the developed model and its application in quantitative structure-activity relationship analysis. On the basis of molecular modeling results, the strategy for the production of broad specific antibodies against quinolones was suggested and the potentiality of several candidates was predicted. Quinolones is a group of synthetic antibiotics in human medicine and in veterinary treatment. Because of their broadspectrum and high antibacterial efficiency, they have became one of most widely used antibacterial agents since 1990s.1,2 To control the hazards of residues in edible animals, many countries and areas have set strict maximal residue limit (MRL) for quinolones in food stuff.3,4 Therefore there is a growing interest in analytical techniques for such antibiotics. * To whom correspondence should be addressed. E-mail: [email protected]. Address: College of Food Science and Engineering, Ocean University of China, 5 Yushan Road, Qingdao, P. R. China. Phone: 0086-532-82032203. Fax: 0086-53282032389. † Food Safety Laboratory, Ocean University of China. ‡ Shandong University of Technology. § School of Medicine and Pharmacy, Ocean University of China. (1) Bhanot, S. K.; Singh, M.; Chatterjee, N. R. Curr. Pharm. Des. 2001, 7, 311–335. (2) Zhou, W. C.; Zhou, H. Y. Chin. J. Pharm. 2005, 36, 309–312. (3) Herna´ndez-Arteseros, J. A.; Barbosa, J.; Compano ˜, R.; Prat, M. D. J. Chromatogr., A 2002, 945, 1–24.

3246

Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

Given the fact that over 20 quinolones have been applied in clinical practice, simultaneous determination of such drugs becomes necessary for construction of an effective monitoring system. This task can be partly solved by chromatograph-based techniques. Several papers have described the simultaneous determination of 6-8 quinolones in food using liquid chromatography (HPLC) or liquid chromatography-tandem mass spectrometry (HPLC-MS).5-7 These techniques are accurate and sensitive but time-consuming, expensive, and greatly rely on laboratory facilities as well as experimental skills. All these characteristics make them more suitable for confirmatory analysis than fast screening of drug residues. A microbiological test is often used for routine screening of quinolones and other antimicrobial agents, but it takes usually about 20 h, and the sensitivity is much lower compared with HPLC based methods.8 Therefore it is important to develop alternative techniques for fast and sensitive screening of multiple quinolones, especially those that can be easily performed without expensive equipment. Now there is an emerging trend to develop broad-specific antibodies for high-throughput immunoassays. Such antibodies usually showed similar cross reactivity to many members of a group of chemical contaminants and so can be hopefully exploited as broad selective receptors for development of simple and lowcost screening analysis.9 Precise prediction of antibodies’ specificity is an important precondition for generation of broad-specific antibodies, but until now it is still a tough task due to the absence of effective technical tools. In previous studies, molecular modeling hasbeenprovedapromisingtoolforthestudyofthestructure-specificity relationship and the selection of haptens. Several papers have described its application for preparation of broad-selectivity

(4) Lin, W. X., Ed. The Compilation of Residue Limits Stands for Pesticides and Veterinary Drugs in Foodstuffs in The World; Dalian Maritime University Press: Dalian, China, 2002. (5) Marazuela, M. D.; Moreno-Bondi, M. C. J. Chromatogr., A 2004, 1034, 25–32. (6) Idowu, O. R.; Peggins, J. O. J. Pharm. Biomed. Anal. 2004, 35, 143–153. (7) Schneider, M. J.; Donoghue, D. J. J. Chromatogr., B 2002, 780, 83–92. (8) Hee-Jung Cho, A. M.; Abd El-Aty, A. G.; Sung, G. M.; Yi, H.; Seo, D. C.; Kim, J. S.; Shim, J. H.; Jeong, J. Y.; Lee, S. H.; Shin, H. C. Biomed. Chromatogr. 2008, 22, 92–99. (9) Spinks, C. A. Trends Food. Sci. Technol. 2000, 11, 210–217. 10.1021/ac802403a CCC: $40.75  2009 American Chemical Society Published on Web 04/01/2009

antibodies against food contaminants such as triazines.10-12 On the basis of conformation analysis and comparative molecular field analysis (CoMFA), Hotzaple et al. and Shen et al. provided much valuable information on the structural features of quinolone haptens that affect the specificity of corresponding antibodies.13,14 However these retrospective studies did not characterize different quinolones quantitatively to provide more definite guilds for design and selection of haptens. In this paper, on the basis of the calculation of the filedoverlapping coefficients of the quinolones, a new quantitative model was developed to describe the conformational features of haptens and to predict the specificity of corresponding antibodies. The model was preliminarily validated with antibodies against four different quinolone haptens, and the strategy on selection of quinolone haptens was also disccussed for the generation of broadspecific antibodies. EXPERIMENTAL SECTION Reagents. Quinolones were purchased from Veterinary Medicine Supervisory Institute of China (Beijing, China). Stock solutions (200 µg/mL) of quinolones were prepared in NaOH (0.01 mol/L) and diluted by phosphate buffers (pH 7.4, 0.01 mol/L, PBS). Bovine serum albumin (BSA), ovalbumin (OVA), Nhydroxysuccinimide (NHS), N-(3-dimethylaminopropyl)-N′-carbodiimide hydrochloride (EDC), ethylenediamine (EDA), incomplete Freund’s adjuvant (IFA), and complete Freund’s adjuvant (CFA) were purchased from Sigma Chemical Co. 3,3′,5,5′-Tetramethylbenzidine (TMB) was purchased from Amresco. Goat antimouse IgG-HRP was purchased from Zhongshan Biology Co. (Beijing, China). N,N′-Dimethylformamide (DMF) was purchased from Shanghai Chemical Co. All the other chemicals were analytical grade. Female BALB/C mice were from Center for New Drugs Evaluation of Shandong University (Jinan, China). Monoclonal antibodies (Mabs) against enrofloxacin from different cell strains were provided by Beijing Biosynthesis Biotechnology Co. (Beijing, China). Methods. Molecular Modeling of Quinolone Haptens. The technique of molecular modeling, which means simulating molecular behavior in silico based on quantum mechanics or molecular mechanics theory, was wildly used in life science, such as computer aided drug design (CADD), structure-activity relationship (SAR), etc. Our simulating was performed with Cerius 2, a molecular modeling software package supplied by Accelrys Inc.15 First, the structural models of quinolones were constructed, minimized with molecular mechanism method and put into a database. Then, one by one, taking the lowest-energy conformation of each molecule as a template, all of the other molecules were flexibly aligned to this molecule. Alignment was field-based, and the potential probe was set to a sp3 hybridization C atom. Then, (10) Kim, Y. J.; Cho, Y. A.; Lee, H. S.; Lee, Y. T. Anal. Chim. Acta 2003, 494, 29–40. (11) Paul, J.; Elmar, M.; Reinhard, R. Anal. Chim. Acta 1995, 315, 279–287. (12) Nathalie, D. B.; Pichon, V.; Hennion, M. C. J. Chromatogr., A 2003, 999, 3–15. (13) Holtzapple, C. K.; Buckley, A.; Stanker, H. J. Agric. Food Chem. 1997, 45, 1984–1900. (14) Wang, Z. H.; Zhu, Y.; Ding, S. Y.; He, F. Y.; Beier, R. C.; Li, J. C.; Jiang, H. Y.; Feng, C. W.; Wang, Y. P.; Zhang, S. X.; Kai, Z. P.; Yang, X. L.; Shen, J. Z. Anal. Chem. 2007, 79, 4471–4483. (15) Accelrys Software, Inc. Cerius 2, version 4.10L; Accelrys Software, Inc.: San Diego, CA, 2005.

the overlap coefficient (molecular field similarity) was calculated based on both steric and electrostatic fields, each holding a 50% proportion. According to the similarity premise that structurally similar molecules have similar properties or similar biological activities,16,17 the molecule pairs with higher overlap coefficient were supposed to possess a higher cross-recognization ratio. A total of 27 quinolone haptens including pipemidic acid (PPA), enrofloxacin (ENR), norfloxacin (NOR), ciprofloxacin (CIP), lomefloxacin (LOM), ofloxacin (OFL), sparfloxacin (SPA), danofloxacin (DAN), sarafloxacin (SAR), pefloxacin (PEF), enofloxacin (ENO), levofloxacin (LEV), balofloxacin (BAL), difloxacin (DIF), benofloxacin (BEN), marbofloxacin (MAR), frefloxacin (FRE), caderofloxacin (CAD), gatifloxacin (GAT), moxifloxacin (MOX), grepafloxacin (GRE), gemifloxacin (GEM), trovafloxacin (TRO), prulifloxacin (PRU), pazufloxacin (PAZ), flumequine (FLU), and oxolinic acid (OA) were investigated with the developed model. Hapten Conjugation. BSA and OVA were used as carriers for the immunogens and coating antigens, respectively, and BSA was precationized before its conjugation with quinolones according to the method described by Holtzapple et al.:13 200 µL of EDA, 240 mg of BSA, and 250 mg of EDC were sequentially added to 10 mL of PBS, and the mixture was reacted at 20 °C overnight under frequent shaking. Then the EDA-treated BSA was dialyzed against PBS (0.01 mol/L, pH 7.0) to remove free EDA. The synthesis of the enrofloxacin-BSA complex was according to methods described by Liu et al.:18 15 mg of enrofloxacin, 5.2 mg of NHS, and 9.3 mg of EDC were dissolved in 500 µL of DMF. The mixture was shook for 1.5 h at room temperature and then centrifuged at 4 °C for 30 s (BR-4i, Jouan, 3000 rpm). The supernatant was added to 2.0 mL of PBS (containing 33% DMF and 25 mg of cationized BSA) and stirred overnight at 4 °C in the dark. The conjugates were first dialyzed against PBS containing 33% (v/v) DMF and then reduced the concentration of DMF gradually to pure PBS. After dialysis and lyophilization, the conjugates were kept at -80 °C until further use. Other quinolone-carrier protein conjugates were prepared in the same way and with the same hapten-carrier molecular ratio as that of enrofloxacin-BSA. Antibody Production. A total of 10 mg of immunogen (protein content) was dissolved in 1.0 mL of physiological saline and mixed with CFA (v/v ) 1:1). The mixture was used to immunize three 7-week-old female BALB/c mice intraperitoneally with a dose of 0.3 mL/mouse. Booster injections were carried out in the same procedure every two weeks, except that IFA was used instead of CFA. Three mice were injected with physiological saline alone and used as controls. Blood samples were collected 7 days after the third booster injection and then centrifuged at 2 000 rpm for 10 min, and the supernatant was removed and stored at -80 °C until further use. Enzyme-Linked Immunosorbent Assay (ELISA). A competitive indirect ELISA (ci-ELISA) was used to determine the specificity of antibodies. The operation was at first optimized by the checkerboard test;18 and when the absorbance value at 450 nm was about 1.0, the corresponding concentrations of coating (16) Johnson, M. A., Maggiora, G. M., Eds. Concepts and Applications of Molecular Similarity; John Wiley: New York, 1990. (17) Wermuth, C. G. Drug Discovery Today 2006, 11, 348–354. (18) Liu, C. E.; Lin, H.; Cao, L. M.; Jiang, J. J. Fish. China 2005, 29, 534–539.

Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

3247

Table 1. Optimized Parameters for ci-ELISA

test groups

coating antigen concentration (µg/mL)

IgG-HRP dilution

antibody dilution

Pab anti-NOR Pab anti-ENR Pab anti-OFL Pab anti-SPA MEF03 MEF07

20 50 10 20 50 50

10 000 10 000 20 000 30 000 10 000 10 000

40 000 50 000 80 000 20 000 200 2 000

antigens (hapten-OVA conjugates), antibodies, and goat antimouse IgG-HRP were chosen as the working concentrations for ELISA experiments (Table 1). Microplates were coated with coating antigens of working concentration (100 µL/well) in carbonate buffer (0.1 mol/L, pH 9.6) and incubated at 37 °C for 2.0 h and then were washed with PBST (pH 7.4, 0.01 M PBS containing 1% v/v Tween 20). A volume of more than 200 µL of OVA (0.1%, in PBS) was added to each well to block nonspecific binding sites on the plastic surface. After incubation at 37 °C for 2.0 h, microplates were washed and then antibodies of working concentrations (suitable dilutions of antiserum or cultures of monoclonal cell clones, 100 µL/well) were added, followed by addition of serial dilutions of competitors (100 µL/well). After incubation at 37 °C for 2 h, the plates were washed with PBST and goat antimouse peroxidase (IgG-HRP) of a working concentration in PBS was added (100 µL/well) and incubated at 37 °C for 2 h. The plates were washed with PBST, and then the TMB substrate solution (500 µL of 2 mg/mL TMB-DMF and 40 µL of 0.6% H2O2 diluted with 10 mL of citrate-phosphate buffer, pH 5.0) was added (100 µL/well). The plates were incubated for 20 min at 37 °C in the dark, followed by the addition of stopping solution (2 mol/L H2SO4, 50 µL/well). The absorbance value at 450 nm was determined with a plate reader (Multiskan MK3, Laboratory Systems). Triplicate experiments were performed for each concentration of competitors. Blank wells were treated in the same way as the test wells except that PBS replaced the competitor solutions. The absorbance values at 450 nm obtained in the presence of various competitor concentrations and without competitor (maximal signal) were referred to as B and B0, respectively. The inhibition ratio was calculated as follows: inhibition (%))

(

)

1-B × 100 B0

A calibration curve was prepared by plotting log [competitor concentration] versus percent inhibition, and the 50% binding inhibition value (IC50) was the concentration of the competitor when the inhibition reached 50%. The cross-reactivity of antibodies was calculated as follows: cross-reactivity (%) )

(

)

IC50 value of hapten × 100 IC50 value of competitor

RESULTS AND DISCUSSION Molecular Modeling of Quinolone Haptens. The quantitative molecular superimposing results of the 27 quinolones were summarized in Figure 1. The 27 quinolones can be approximately divided into 3 groups according to calculated overlap coefficients. 3248

Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

PEF, MAR, BEN, NOR, CIP, BEN, and ENO consist of group 1 which showed high conformational similarity (average overlap coefficient more than 0.91) to most quinolones. SAR, DIF, TRO, PRU, OA, and FLU represent group 2 that possessed the least conformational features common to the whole quinolone family (columns 9, 14, 22, 23, 26, 27, with an average overlap coefficient less than 0.85, mostly comprised of blue and dark purple squares). The rest of the quinolones belonged to group 3 which demonstrated average overlap coefficients ranging from 0.85 to 0.91 for the quinolones investigated. Further investigation of the overlapped 27 quinolones (Figure 2) demonstrated that there existed significant conformational difference at position 7 and the area around positions 1 and 8. The conformational features at other positions were almost the same and so indicated very limited influence on the conformational difference among quinolones. On some quinolone molecules possessing fluorine or oxygen substitutions at position 8 (such as LOM and SPA), a significant ringlike conformation was observed around positions 1 and 8. Combined with quantitative molecular superimposing results, these modeling results allowed us to assume the following reasons for the conformational difference among quinones: (i) different substitutions at positions 1 and 7, especially their difference in the size and electronic properties, will result in a significant conformational difference among quinolones, which could be illustrated by the low overlap coefficients of quinolones possessing a fluoroohenyl ring (such as SAR, DIF, and TRO) at position 1 to other quinolones with alkane groups at the same position. (ii) Although never indicated by any previous study, there seems to exist a significant interaction between the fluorine or oxygen atoms of position 8 and the alkane groups of position 1 on some quinolones; such an interaction may be caused by electronic attraction or other forces and also contributes a lot to the conformational difference among quinolones. Considering that much difference in position 7 was introduced by a few quinolones (such as PRU and MOX) not very popular in present applications, the conformation around positions 1 and 8 seemed more important to select haptens for production of broad-specific antibodies against quinolones. It is very interesting to note the unexpected high conformational similarities of MAR, BEN, and OFL to most other quinolones. The reason may lie in following structural features of these quinolones: (i) several quinolones (such as LEV, FLU, and PAZ) have very similar chemical structures with MAR, BEN, and OFL, especially that they all possess a six-membered-ring around positions 1 and 8. (ii) As indicated in the last section, many quinolones (LOM, SPA, CAD, MOX, BAL, and FRE) could form a ringlike conformation around positions 8 and 1 by the interaction between fluorine (or oxygen) atoms and alkane substitutions, which significantly increases the conformational similarity of these molecules to MAR and BEN. (iii) The ring structure around positions 1 and 8 makes the conformation of this area in the middle of simple ones (such as ENR, CIP, NOR with alkanes) and complex ones (such as SAR and DIF with fluoroohenyl rings), which helps MAR, BEN, and OFL to demonstrate homogeneous overlap coefficients to different quinolones; this assumption was supported by the high overlap coefficients of MAR and BEN to SAR and DIF. The calculated overlap coefficients of OFL to SAR and DIF were a little lower than expected; this may be caused by

Figure 1. The overlap coefficients contour plot: Each column, comprised of 27 colored squares, represents coefficients for 27 objective compounds. The abbreviated template names were given on the top of the grid. From left to right, also bottom to top: PPA, ENR, NOR, CIP, ENO, LOM, OFL, LEV, SAR, SPA, BAL, CAD, DAN, DIF, BEN, MAR, FRE, GAT, MOX, GRE, GEM, TRO, PRU, PAZ, PEF, FLU, and OA.

Figure 2. The basic structure of quinolone molecules (A) and 3D conformations of 27 overlapped quinolones.

the unsettled flexibility of OFL or kinetic factors during the binding process. Validation of the Developed Model with Polyclonal Antibodies. To validate the accuracy of the developed molecular model, four polyclonal antibodies against different quinolones were prepared. The IC50 values of these antibodies for the corresponding haptens ranged from 36.67 to 125.89 µg/L, and their

Figure 3. The general analysis of the correlation between the crossreactivity (%) of the antiquinolone polyclonal antibodies and the overlap coefficients of haptens with different quinilones. The crossreactivity values over 100% and below 1% were regarded as 100% and 0.5%, respectively. The data of the inserted table were crossreactivity (%) of antibodies to different quinolones, and each value represents the average of three measurements.

cross reactivity to different quinolones were determined to fit calculated overlap coefficients with haptens. It is usually believed that antibodies will exhibit higher binding affinities to a target with increased conformational analogy of the target to haptens.9 As shown Figure 3, there is a significant linear relationship between cross-reactivity of the antibodies and the calculated overlap coefficients of the competitors with haptens. Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

3249

The correlation coefficient was calculated as 0.7793, which confirmed the overall accuracy of the developed molecular model. Compared with other antibodies, anti-OFL antibodies demonstrated some different properties. Its cross-reactivity to sarafloxacin was much higher than that of other antibodies. This confirmed the comparable high conformational similarity of OFL to quinolones with fluoroohenyl rings at position 1, which has been assumed in Molecular Modeling of Quinolone Haptens. However, the low cross-reactivity of this antibody to sparfloxacin was unexpected; it conflicts with the calculated conformational similarity (more than 0.91) and the very high binding affinity demonstrated by antisparfloxacin antibody to ofloxacin. A possible reason is that after conjugation to carrier proteins, there are some special conformational changes on the ofloxacin molecules which decreased its structural similarity to sparfloxacin. Some electronic and hydrophobic groups both from haptens and proteins may contribute to the postconjugation conformational changes of haptens by their effects on hydrogen bonds, van der Waals forces, hydrophobic interactions, and electronic bonds. However, with the existing techniques it is still very difficult to accurately describe the conformational changes during hapten-carrier conjugation. Validation of the Developed Model with Monoclonal Antibodies. For further validation of the developed molecular modeling technique, the specificity of anti-ENR Mabs was investigated. The Mabs were produced by two different cell strains and were designated as MEF03 and MEF07, respectively. The immunogens and the immunization method for the preparation of two Mabs were the same as those for anti-ENR polyclonal antibodies. As shown in Figure 4, two Mabs demonstrated significantly different cross-reactivity to most competitors. The results cannot fit well with the overlap coefficients calculated, and the correlation coefficients were only 0.6952 and 0.5277 for MEF03 and MEF07, respectively. The different specificity demonstrated by MEF03 and MEF07 could be due to a genetic difference of the cell strains. Such a genetic difference may be very slight but enough to significantly change the specificity of Mabs to the analogues of the haptens. Similar results were also investigated by Shen at el.,14 where antiquinolone Mabs from different strains demonstrated quite different specificity to the same competitors. For this reason, it seems very difficult to explain and predict the intergroup specificity of Mabs from a single cell strain by molecular modeling of haptens or hapten-carrier conjugates. However, this problem should not exist for polyclonal antibodies, which in fact represent the average property of numerous cells. The assumption was confirmed by a high linear correlation coefficient (0.8424) between cross-reactivity of polyclonal anti-ENR antibodies and overlap coefficients of ENR to other quinolones (Figure 5). Suggestions for Preparation of Broad-Specific Antibodies Against Quinolones. For generation of a broad-specific antibody, it is necessary to make sure that even after their conjugation with carriers, the hapten molecules still demonstrate conformations very similar to those of the most targeted analogues.9 For this proposal, haptens consisting of the most common structural features of the drug family were usually selected as preferred candidates and the binding strategy should also be carefully designed. 3250

Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

Figure 4. The relationship between the cross-reactivity (CR, %) of monoclonal antienrofloxacin antibodies and overlap coefficients of enrofloxacin with different quinolones. Monoclonal antibodies were MEF03 (A) and MEF07 (B), respectively. Each value of crossreactivity (CR) represented the average of three measurements, and the cross-reactivity values over 100% and below 1% were regarded as 100% and 0.5%, respectively.

Now the commonly exploited quinolone-carrier conjugation was directly performed through an amide linkage between active carboxylic groups at position 3 of the quinolones and the amino groups on the carrier proteins.13,14,18,19 Recently, modification of position 7 was also reported where a carboxyl acid group was introduced on the piperazinyl ring of NOR and SAR molecules to conjugate with the carriers.20 Both above binding operations were easy to carry out, but the structural difference in positions 1 and 7 among different quinolones remained after conjugation and so the resulting antibodies usually demonstrated uneven binding affinity with many quinolones analogues. For production of more broad-specific antibodies against quinolones, position 1 may be a more desirable binding site based on following assumptions: (i) the overlapped 3D conformation of quinolones (Figure 2) demonstrated that the structural difference around position 1 is much more significant than that in other positions. (ii) This structural difference in position 1 will be effectively diminished by substitution of original diverse inactive moieties with some active groups (19) Duan, J. H.; Yuan, Z. H. J. Agric. Food Chem. 2001, 49, 1696–1700. (20) Huet, A. C.; Charlier, C.; Tittlemier, S. A.; Singh, G.; Benrejeb, S.; Delahaut, P. J. Agric. Food Chem. 2006, 54, 2822–2827.

Figure 5. The relationship between the cross-reactivity (%) of polyclonal antienrofloxacin antibody and overlap coefficients of enrofloxacin with different quinolones. The cross-reactivity values over 100% and below 1% were regarded as 100% and 0.5%, respectively.

such as a carboxyl acid group or amino groups. (iii) Common structural features in most quinolones are helpful to obtain a generic antibody, a carboxylic acid group, a ketone group, and a piperazinyl group, and will be retained after conjugation with carriers. However, because of the difficulty and complexity of chemical modifications required, this strategy seems uneasy to carry out and has not been tried until now. Another possible problem is that the substitution of some big groups (such as fluoroohenyl rings) with small active moieties at position 1 may introduce significant conformational differences to the quinolone skeloten, therefore resulting in decreased binding affinity of the corresponding antibodies to the quinolone targets. On the basis of the calculation of the hapten’s overlap coefficients to the targets, quantitative comparison and selection of haptens could be performed to produce antibodies with the desired specificity to the different quinolones. Among the present haptens, MAR, BEN, PEF, and NOR were suggested by our molecular modeling as suitable haptens to develop broad-selective antibodies for most quinolones. This hypothesis has been partly verified by a previous study where anti-NOR antibodies were exploited for simultaneous detection of six quinolones in a food matrix.14 PEF, MAR, and BEN demonstrated higher average overlap coefficients than NOR and may be more suitable haptens for the production of group-specific antibodies for quinolones. This potential has not been indicated by any previous study; in our

next papers, the production and characterization of anti-PEF antibodies and anti-MAR antibodies and their application for screening of multiple quinolones will be described. Antibodies against different quinolones have been described in several papers13,14,20 but may be due to the significant conformational difference between the fluoroohenyl rings and the alkane groups at position 1; these antibodies did not demonstrate high cross-reactivity to quinolones with different types of position-1 substitutions and therefore could not be used to develop a more generic immunoassay. The high overlap coefficients of MAR and BEN with SAR and DIF, as well as the higher cross-reactivity to SAR demonstrated by anti-OFL antibodies than that of other antiquinolone antibodies, allowed us to suggest that this task might be solved by designation of new quinolone haptens: they also possess a ringlike conformation around positions 1 and 8 similar to those of MAR, BEN, and OFL but can mimic quinolones of both groups 1 and 3 much better by precise structural modification. CONCLUSIONS A novel molecular model based on the molecular field overlapping coefficient was developed for quinolones haptens. Besides qualitative characterization of the 3D conformations of 27 quinolones, their conformational similarity to each other was for the first time calculated and exploited for the selection of haptens and predication of the antibodies’ specificities. The developed model was preliminarily validated with antibodies against different quinolones. A significant positive correlation was observed between the calculated overlapping coefficients of the haptens and the cross-reactivity of the antibodies. These results demonstrated that the developed molecular model can fit well with the real structure-activity relationship between the quinolone haptens and the corresponding antibodies, which confirmed its overall accuracy and allowed us to suggest it as a useful tool for selection of quinolone haptens according to the desired specificity of antibodies. ACKNOWLEDGMENT This work was supported by the National Natural Science Funding of China (Grant No. 30400336) and the Scientific Research Funding for Agriculture (Grant No. NYHYZX 07-046). The first two authors contributed equally to this work.

Received for review November 12, 2008. Accepted March 16, 2009. AC802403A

Analytical Chemistry, Vol. 81, No. 9, May 1, 2009

3251