Molecular Modeling Application on Hapten Epitope Prediction: An

Article pubs.acs.org/JAFC

Molecular Modeling Application on Hapten Epitope Prediction: An Enantioselective Immunoassay for Ofloxacin Optical Isomers Hongtao Mu,† Hongtao Lei,*,† Baoling Wang,† Zhenlin Xu,† Chijian Zhang,† Li Ling,† Yuanxin Tian,‡ Jinsheng Hu,§ and Yuanming Sun*,† †

Guangdong Provincial Key Laboratory of Food Quality and Safety, College of Food Science, South China Agricultural University, Guangzhou, Guangdong 510642, People’s Republic of China ‡ School of Pharmaceutical Sciences, Southern Medical University, Guangzhou, Guangdong 510515, People’s Republic of China § Department of Plant and Environmetal Protection Sciences, University of Hawaii, Honolulu, Hawaii 96822, United States S Supporting Information *

ABSTRACT: To deepen our understanding of the physiochemical principles that govern hapten−antibody recognition, ofloxacin enantiomers were chosen as a model for epitope prediction of small molecules. In this study, two monoclonal antibodies (mAbs) mAb-WR1 and mAb-MS1 were raised against R-ofloxacin and S-ofloxacin, respectively. The enantioselective mAbs have a high sensitivity and specificity, and the enantioselectivity is not affected by heterologous coating format reactions. The epitopes of the ofloxacin isomers were predicted using the hologram quantitative structure−activity relationship (HQSAR) and comparative molecular field analysis (CoMFA) approaches. The results consistently show that the epitope of the chiral hapten should be primarily composed of the oxazine ring and the piperazinyl ring and mAbs recognize the hapten from the side of this moiety. The enantioselectivity of mAbs is most likely due to the steric hindrance caused by the stereogenic center of the epitope. Modeling of chiral hapten−protein mimics reveals that ofloxacin isomers remain upright on the surface of the carrier protein. Suggestions to improve the enantioselectivity of antibodies against ofloxacin isomers were also proposed. This study provided a simple, efficient, and general method for predicting the epitopes of small molecules via molecular modeling. The epitope predictions for small molecules may create a theoretical guide for hapten design. KEYWORDS: ofloxacin enantiomers, enantioselective antibody, antigenic determinant, molecular modeling

■

INTRODUCTION Immunoassays based on antibody−antigen reactions are widely used during clinical environmental and food safety monitoring.1−3 Immunoassays, especially enzyme-linked immunosorbent assays (ELISAs), display high specificity, sensitivity, simplicity, rapidity, low cost, high throughput, and suitability for on-site analyses; these techniques have been successfully developed to detect toxic compounds with low molecular weights, such as pesticide residues, veterinary drug residues, environmental hormones, toxins, and prohibited food additives.4 The key step during the development of an immunoassay against small molecules is the production of an antibody with the proper specificity and a high affinity.5−9 However, most immunoassays for the determination of small molecules are designed on the basis of “trial and error”,10 especially when designing chiral immunizing and coating haptens. The reason for this approach may lie in the lack of understanding about the epitope of small molecules in hapten− antibody recognition. Although hapten−antibody recognition has been explored in some studies, prediction of the epitopes received insufficient attention.11−15 Even in the few studies that have tried to do the prediction, analyses were mainly based on empirical conjecture and quantitative assessments were far from enough.16,17 The shortcomings are due to the limitations in predicting the epitope of a hapten. The term epitope prediction or epitope mapping is derived from the realm of protein antigen−antibody © 2014 American Chemical Society

interactions. However, the method for predicting protein antigen epitopes is not suitable for haptens.18−22 Although high-resolution X-ray crystallography can give haptens the most detailed representation of conformational epitopes, crystallization of those complexes is idiosyncratic as well as time- and resource-intensive.23 In this work, cross-reactivity (CR) data are coupled with molecular modeling, particularly two- and threedimensional quantitative structure−activity relationships (2Dand 3D-QSAR), to predict the epitopes of haptens. Ofloxacin is a potent quinolone (QN) antibacterial agent used on both humans and livestock; it has a tricyclic ring structure with a methyl group attached to the asymmetric carbon at the C-3 position of the oxazine ring (Figure 1). Ofloxacin enantiomers were chosen as the model to predict small-molecule epitopes for three reasons. First, the stereogenic center is a unique indicator for epitopes. Second, ofloxacin is one of the quinolones and has a series of analogues. This would be appropriate for QSAR modeling and investigation of hapten−antibody recognition. Last but not least, ofloxacin enantiomers have very different environmental effects;24 therefore, it is necessary to monitor ofloxacin enantiomers in Received: Revised: Accepted: Published: 7804

October 3, 2013 June 6, 2014 June 6, 2014 July 29, 2014 dx.doi.org/10.1021/jf404449n | J. Agric. Food Chem. 2014, 62, 7804−7812

Journal of Agricultural and Food Chemistry

Article

Apparatus and Software. Ultraviolet−visible (UV−vis) spectra were recorded on a UV-160A Shimadzu spectrophotometer (Kyoto, Japan). The washing steps were carried out using a DEM-III microplate washer (Tuopu Analytical Instrument, Beijing, China). The absorbance readings were carried out at 450 nm with a Multiskan MK3 microplate reader (Thermo Fisher Scientific, Pittsburgh, PA). The competitive curves were analyzed with a four-parameter equation using the OriginPro 8.0 software (OriginLab Corporation, Northampton, MA). Preparation of Hapten−Protein Conjugates. The haptens of the ofloxacin enantiomers containing a carboxylic acid group on dihydroquinoline were synthesized and bound to carrier proteins using the carbodiimide method to prepare the immunogenic agents.25 BSA and OVA were used as carrier proteins for the immunogens and coating antigens, respectively. R-OFL and S-OFL were coupled to BSA separately for immunogens (R-OFL−BSA and S-OFL−BSA) and to OVA for coating antigens (R-OFL−OVA and S-OFL−OVA). The hapten−protein conjugates were prepared using the carbodiimide method with modifications. The hapten (10 mg) was dissolved with BSA or OVA (15 mg) in 0.9% normal saline (1.5 mL). EDC (30 mg) was added slowly to the hapten solution, and the mixture was stirred at room temperature for 30 min. The hapten−carrier protein conjugates were dialyzed against phosphate-buffered saline (PBS) at 4 °C for 72 h and then characterized with UV−vis spectrometry. Preparation of mAbs. Immunization. The immunization protocol was similar to those previously described.9 Briefly, eight BALB/c female mice were randomly separated into two equal groups, which were immunized with S-OFL−BSA and R-OFL−OVA, respectively. All of the mice aged 6−8 weeks were immunized intraperitoneally with the immunogens (50 μg for each mouse) at 2 week intervals. Blood samples were taken from the mice to detect the presence of antibodies against R-OFL or S-OFL using icELISA in a homologous conjugate-coated format on the eighth day after each immunization, starting 49 days after the first injection. The mouse that exhibited the best titer and inhibition rate for each hapten was selected as the donor of spleen cells for hybridoma production. Each selected mouse received a final intraperitoneal injection with 100 μg of immunogen, and the mice were sacrificed for cell fusion in the third day after the final injection. All experiments involving animals were performed in compliance with the relevant protective and administrant laws for laboratory animals of China and were conducted with the approval of the Institutional Authority for Laboratory Animal Care. Cell Fusion, Hybridoma Selection, and Cloning. The fusion of SP2/0 cells (saved in our laboratory) and spleen cells was performed with PEG 2000 to produce hybridoma cells. The positive hybridomas were selected and cloned by limited dilution. After the fourth subcloning, monoclonal hybridomas possessing S-OFL or R-OFL enantioselectivity were selected to intraperitoneally inject to each pretreated mice. After 10 days, ascites fluids were obtained. They were first purified by sequential precipitation with caprylic acid and ammonium sulfate26 and then purified using a protein-G column. The concentrations of purified mAbs were determined with UV−vis spectrometry. The isotype of mAbs were identified using IsoQuick Strips (Sigma-Aldrich). icELISA Protocol. The ELISA was established on the basis of the common protocol of icELISA.9 A total of 96 microtiter plates were coated with 100 μL of the coating antigen at the optimal dilution in CBS overnight at 4 °C, and then they were washed 2 times with PBST solution. Each well was blocked with 150 μL of 5% skim milk in PBS buffer for 3 h at 37 °C. After five washes, the plates were added with different concentrations of ofloxacin standard solutions (50 μL/well) and an appropriate concentration of antibody solutions (50 μL/well), both of which were diluted in PBST solution. Afterward, the plates were incubated for 40 min at 37 °C. After washing 5 times with PBST solution, the plates were incubated with 100 μL/well of HRP−IgG (diluted 1:5000 in PBST) for 30 min at 37 °C. After five washes, 100 μL of TMB solution was added to each well and incubated at 37 °C for 10 min; the color development was stopped by adding 50 μL/well of 2 mol/L H2SO4, and the absorbance was recorded at 450 nm (A450). The concentrations of antibodies and coating antigens had been

Figure 1. Chemical structures of ofloxacin stereoisomers. The common structure of QNs is labeled in blue.

animal byproducts and environmental samples with a simple, cost-effective, and high-throughput analysis method. In this study, ofloxacin isomers with a stereogenic center were chosen as a model while predicting the epitopes of small molecules. The ofloxacin isomers were conjugated to bovine serum albumin (BSA) to form immunogens using the carbodiimide method. Two monoclonal antibodies (mAbs) were raised by BALB/c mice immunization, cell fusion, and hybridoma selection. Then, the enantioselectivity and specificity of mAbs were characterized using indirect competitive ELISA (icELISA). On the basis of the CR data, the epitopes of the ofloxacin enantiomers were predicted by both a hologram quantitative structure−activity relationship (HQSAR) method and a comparative molecular field analysis (CoMFA) method. The conformational structures of epitope on the carrier protein were modeled for explaining the experimental phenomenon and providing a theoretical guide for hapten design.

■

MATERIALS AND METHODS

Chemicals and Apparatus. R-(+)-Ofloxacin (R-OFL), S(−)-ofloxacin (S-OFL), S-(−)-ofloxacin methyl ester (S-OFLM), and R-(+)-ofloxacin methyl ester (R-OFLM) were obtained from Daicel Chiral Technologies Company (Shanghai, China). Racemic ofloxacin (rac-OFL), pefloxacin (PEF) mesylate, gatifloxacin (GAT), norfloxacin (NOR) hydrochloride, sarafloxacin hydrochloride (SAR), garenoxacin (GAR), clinafloxacin (CLI), rufloxacin (RUF), ciprofloxacin hydrochloride (CIP), enrofloxacin (ENR), prulifloxacin (PRU), lomefloxacin (LOM), tosufloxacin (TOS), sparfloxacin (SPA), difloxacin (DIF), pazufloxacin (PAZ), marbofloxacin (MAR), moxifloxacin (MOX), pipemidic acid (PIP), nalidixic acid (NAL), and oxolinic acid (OXO) were purchased from Veterinary Medicine Supervisory Institute of China (Beijing, China). The structure of QNs is shown in Figure S1 of the Supporting Information. BSA, ovalbumin (OVA), 1-(3-(dimethylamino)propyl)-3-ethyl carbodiimide (EDC), 2,4,6-trinitrobenzene-1-sulfonic acid (TNBS), complete and incomplete Freund’s adjuvant, horseradish peroxidase−immunoglobulin G (HRP−IgG), 3,3′,5,5′-tetramethylbenzidine (TMB), the culture media RPMI-1640, hypoxanthine−aminopterin−thymidine (HAT), hypoxanthine−thymidine (HT), and poly(ethylene glycol) (PEG) 2000 were purchased from Sigma-Aldrich (St. Louis, MO). All of the chemicals and organic solvents, which were analytical grade or better, were obtained from a local chemical supplier (Yunhui Trade Co., Ltd., Guangzhou, China). Deionized water was prepared using a Milli-Q water purification system (Millipore, Bedford, MA). A total of 96 microtiter plates for ELISA were obtained from Yunpeng Technology Development Co., Ltd. (Xiamen, China). Female BALB/c mice, 6−8 weeks old, were provided by the Guangdong Medical Laboratory Animal Center (Guangzhou, China). Buffers. Carbonate−bicarbonate buffer (CBS, 50 mmol/L, pH 9.6) is used as the coating buffer. Phosphate-buffered saline with Tween 20 (PBST) contains 10 mmol/L phosphate buffer, 0.8% saline solution, and 0.05% Tween 20 (pH 7.4). The phosphate−citrate buffer is a 40 mmol/L solution of sodium citrate (pH 5.5). The substrate solution contains 0.01% tetramethylbenzidine (TMB) and 0.004% H2O2 in citrate buffer. 7805

dx.doi.org/10.1021/jf404449n | J. Agric. Food Chem. 2014, 62, 7804−7812

Journal of Agricultural and Food Chemistry

Article

optimized using checkerboard titration and standard curves. The conditions that resulting in the smallest IC50 were chosen as the optimized conditions (1.0 < Amax 450 nm < 1.2). The 50% inhibition value (IC50) was obtained from the four-parameter logistic equation of the sigmoidal curve27 using OriginPro 8.0 (OriginLab Corporation, Northampton, MA). CR Study. The specificity of the mAbs was determined using 24 ofloxacin analogues under optimized icELISA conditions. The CR values were calculated according to the following equation:

production. After cell fusion and four rounds of limited dilution screening, hybridomas named WR1 (obtained from the ROFL−BSA immunizing group) and MS1 (obtained from the SOFL−BSA immunizing group) that can secrete the highest specificity antibodies against the corresponding ofloxacin isomer (R-OFL and S-OFL, respectively) were obtained. Finally, mAbs were obtained from WR1 and MS1 by ascites production and purified using a protein-G sepharose column. The isotypes of mAb-WR1 and mAb-MS1 were all IgG1/κ. Enantioselectivity of Antibodies. Competitive ELISA can be divided in general into homologous and heterologous formats. In homologous icELISA (Ho-icELISA) formats, the same hapten is used for immunization and assay purposes, whereas in heterologous icELISA (He-icELISA) formats, the immunizing hapten and the competitor hapten (coating antigen hapten) differ in their molecular structures (e.g., ofloxcin isomer). icELISA using OVA conjugate as the coating antigen was applied to investigate the enantioselectivity and sensitivity of the obtained mAbs to ofloxacin stereoisomers. The optimal concentration of mAbs and coating antigens by checkerboard titration is shown in Table S1 of the Supporting Information. In Ho-icELISA, the optimized concentrations of R-OFL−OVA, mAb-WR1, S-OFL−OVA, and mAb-WS1 were 250.0, 40.0, 250.0, and 5.5 μg/L, respectively. Ho-icELISA results showed that mAb-WR1 selectively recognizes R-OFL and mAb-MS1 selectively recognizes S-OFL (Figure 2). The sensitivity (IC50) of mAb-WR1 to R-OFL (0.0344 μmol/L) is higher than that of S-OFL (0.3988 μmol/L) (Figure 2a and Table 1). Concurrently, the sensitivity of mAb-MS1 to S-OFL (0.0018 μmol/ L) is higher than that of R-OFL (0.0096 μmol/L) (Figure 2b and Table 1). Both mAbs predominately have enantioselectivity to the corresponding ofloxacin isomer. The effect of He-icELISA on enantioselectivity was also investigated. The results showed that icELISA based on heterologous coating antigens have no obvious effect on enantioselectivity of mAb-WR1 and mAb-MS1 (Table 2). For mAb-WR1, both the CR values of S-OFL in Ho-icELISA and He-icELISA were about 8%. Concurrently, for mAb-MS1, both the CR values of S-OFL in Ho-icELISA and He-icELISA were about 18%. Above all, we can conclude that the enantioselectivity of mAbs was not affected by the heterologous coating reaction model and the enantioselectivity of mAbs (mAb-WR1 and mAb-MS1) would be the intrinsic property of its own. Sensitivity and Specificity of Antibodies. The sensitivity and specificity data showed that mAb-WR1 and mAb-MS1 have a high sensitivity and specificity (Table 1). In Ho-icELISA of mAb-WR1 and mAb-MS1, 24 analytes were employed for the CR study. The IC50 of mAb-WR1 to R-OFL is 10 times larger than that of mAb-MS1 to S-OFL. That is to say, the sensitivity of mAb-MS1 is higher than that of mAb-WR1. A similar rule was also found in other antibodies, including anti-serum against ofloxacin stereoisomers in our studies. It seems that different enantiomers may be different in immunogenicity, especially in sensitivity. The CR studies showed that both mAb-WR1 and mAb-MS1 have a relatively high specificity (Table 1). Except for S-OFLM, R-OFLM, RUF, GAR, and MAR, the CR values of other QNs are lower than 2% or even not detected (ND). The CR values of PAZ (ND for mAb-WR1 and 0.2% for mAb-MS1) and DIF (0.7% for mAb-WR1 and 1.0% for mAb-MS1) were much lower, although they have a little bit of difference in R1, R2 (oxazine ring), or R3 (piperazinyl ring) groups compared to

CR (%) = IC50(ofloxacin isomer, μmol/L)/IC50(analyte, μmol/L) × 100 HQSAR. All molecular modeling calculations were conducted with Sybyl 7.3 software package (Tripos, Inc., St. Louis, MO) running on a personal computer. The sketched QN structures were optimized to global low-energy conformations using the standard Tripos force field with an 8 Å cutoff for non-bonded interactions in conjunction with Gasteiger−Hückel charges,28 a 0.005 kcal mol−1 Å−1 termination gradient, and a dielectric constant of 4. A database containing ofloxacin stereoisomers and its analogues was established using the Sybyl 7.3 software package. Hologram lengths, atom count in fragments, and information sources were optimized, and the best model with the least standard error was selected. In both HQSAR and CoMFA models, pIC50 values (pIC50 = −log IC50) in homologous format were chosen as dependent variables, while the HQSAR or CoMFA descriptors were chosen as independent variables. The HQSAR descriptors (independent variables) were related to pIC50 (dependent variable) using partial least-squares (PLS) analysis. Fragment contribution to pIC50 was illustrated by encoded colors. CoMFA. The optimal structures obtained above were put into a database. A database including all of the QN molecules previously sketched and optimized was built. The common structure of the QNs (4-oxo-4,7-dihydroquinoline-3-carboxylic acid nucleus) was chosen as the basis for the alignment. Experimental pIC50 (dependent variable) and CoMFA descriptors (independent variables) were correlated by PLS analysis with a leave-one-out (LOO) cross-validation and no cross-validation, respectively. The favored and disfavored contour maps could be visually investigated. Hapten−Carrier Conjugate Modeling. The chiral hapten− carrier conjugate was modeled by Sybyl 7.3. BSA was chosen as the model for the carrier protein, and the crystal structure of BSA [Protein Data Bank (PDB) ID: 4F5S] was obtained from the PDB. BSA was composed of two chains: chain A and chain B. A substructure containing 80 amino acids (AA) at the N terminus of chain A was chosen as the BSA mimic. The carboxyl groups from the ofloxacin isomers were randomly coupled with amino groups, forming a new amido linkage. The conformational structures of ofloxacin isomer− BSA conjugate mimics were optimized using staged minimization: first the ligands, then the side chains, then the biopolymer without C-α, and finally, minimize all. The weighted root-mean-square (RMS) distances of carbon atoms between the free and conjugated ofloxacin isomers were measured using the Fit Atoms module of Sybyl 7.3. The distances from the chiral carbon of the ofloxacin isomers to the nearest atom from the carrier protein residues were also measured.

■

RESULTS AND DISCUSSION Production of mAbs. Ofloxacin isomers are small molecules (MW = 361.37 < 1000) that must be conjugated to a carrier protein to elicit an immune response. Through the free carboxylic acid on the ofloxacin isomer, the molecule was conjugated with BSA or OVA using EDC. The BSA conjugates (S-OFL−BSA and R-OFL−BSA) were used as immunogens, and the OVA conjugates (S-OFL−OVA, R-OFL−OVA, and rac-OFL−OVA)were used as coating antigens. Two groups (four mice in each group) of mice were immunized with conjugates S-OFL−BSA and R-OFL−BSA, respectively. The mouse that showed the highest titer of antiserum and the best inhibition with corresponding hapten in icELISA in each group was selected for monoclonal antibody 7806

dx.doi.org/10.1021/jf404449n | J. Agric. Food Chem. 2014, 62, 7804−7812

Journal of Agricultural and Food Chemistry

Article

Table 1. Enantioselectivity, Sensitivity, and Specificity of mAb-WR1 and mAb-MS1 in Homologous Format (n = 3) mAb-WR1

mAb-MS1

analytes

IC50a

CR (%)

IC50

CR (%)

S-OFL rac-OFL R-OFL S-OFLM R-OFLM RUF GAR MAR NOR DIF ENR PEF CLI LOM GAT MOX TOS PAZ CIP SAR SPA PIP NAL PRU OXO

0.3988 0.0610 0.0344 0.0398 0.0025 0.0327 0.0244 0.5041 6.5761 4.9749 5.8431 5.6943 8.9403 5.6923 5.2418 NDb ND ND ND ND ND ND ND ND ND

8.6 56.4 100.0 86.4 1376.0 105.2 141.0 6.8 0.5 0.7 0.6 0.6 0.4 0.6 0.7 ND ND ND ND ND ND ND ND ND ND

0.0018 0.0020 0.0096 0.0001 0.0079 0.0026 0.0065 0.0097 0.1697 0.1742 0.3328 0.4923 0.2713 6.1420 ND 0.5235 0.6627 1.0490 29.7993 12.6117 ND ND ND ND ND

100.0 90.0 18.8 1800.0 22.8 69.2 27.7 18.6 1.1 1.0 0.5 0.4 0.7