Synthesis of haptens and production of antibodies to bisphenol A

Three immunizing haptens of bisphenol A (BPA), including two new haptens, were used to produce highly sensitive and specific polyclonal antibodies. The spacer arms of haptens for coupling to the protein carrier were located at different positions in BPA, and different length spacer arms were tested. Highly sensitive polyclonal antibodies were obtained and characterized using indirect competitive enzyme-linked immunosorbent assay (icELISA). Under optimized conditions, the half maximal inhibitory concentration (IC50) value of the best polyclonal antibody was 2.1 mg$L, based on coating heterogeneous antigens, and this optimal polyclonal antibody was highly sensitive toward BPA and displayed negligible crossreactivity with bisphenol B and bisphenol E. A sensitive icELISA method utilizing the polyclonal antibody was developed for the determination of BPA in milk. In spiked samples (5, 10 and 20 mg$L), the recovery ranged from 80% to 102% with a coefficient of variation (CV) value below 15.8%. The limit of detection of icELISA was 1.95 mg$L. These results indicate that the icELISA method is suitable for the detection of BPA in milk.


Introduction
Endocrine disrupting chemicals that bind receptors for estrogen, progesterone, androgen and other steroid hormones have been identified as potential estrogenic substances [1,2] , of which bisphenol A (BPA), generally known as bisphenol, is one example [1] . BPA shows a weak estrogenic effect similar to 17β-estradiol [3] . Furthermore, BPA is reported to induce secretion of prolactin [4] and is postulated to cause reproductive disorders including a decline in sperm count [5] . In addition, BPA is a key constituent of epoxy resins that are used as protective coating on metal cans to maintain the quality of canned food and beverages, and traces can be present in the enclosed food or drink [1,6] . BPA has also been found in air and dust samples in residential and commercial environments in Massachusetts [7] and in leachates from a waste water treatment plant and river water in Japan [6,8,9] . To reduce the hazard from ingesting food contaminated by BPA, a simple, rapid and sensitive analytical method for the detection of BPA is necessary.
Several methods for detecting BPA have been reported, including gas chromatography coupled with mass spectrometry [10] and liquid chromatography coupled with mass spectrometry [11] . These methods are highly sensitive and specific, but time-consuming and costly. Some highly sensitive immunoassays have been reported that are more convenient and faster than standard chromatographic methods [12][13][14][15][16][17][18] , but antibodies with a high affinity and sensitivity are essential for immunoassay development, and hapten design is a critical step in the production of suitable antibodies [19] . Researchers have derivatized the hydroxyl group of BPA with 2-5 carbon straight chains [12,15,20] and used these derivatives, as well as 4,4bis(4-hydroxyphenyl)pentanoic valeric acid (BHPVA) [18] , as haptens for coupling to carrier proteins (Fig. 1). However, adding haptens at other positions using different length spacer arms for producing highly sensitive antibodies have not been fully explored. Thus, the objectives of this study were to synthesize two new haptens for preparing highly sensitive polyclonal antibodies, and to develop an indirect competitive enzymelinked immunosorbent assay (icELISA) method for detecting BPA in milk.  manipulated in compliance with Chinese laws and guidelines (GKFCZ2001545) and according to the China Agriculture University (Beijing, China) regulations concerning protection of animals used for scientific purposes (2010-SYXK-0037).

Preparation of immunogens and coating conjugates
Haptens of OYPPP, HHPBA and BHPVA were covalently attached to BSA and OVA for preparing the immunizing and coating conjugates, respectively. To attach OYPPP to BSA or OVA, 66 mg of BSA (or 90 mg of OVA) and sodium bicarbonate (168 mg) were dissolved in 10 mL of deionized water, and 1 mL of DMF containing 30 mg of OYPPP was added dropwise and allowed to react for 2 days at room temperature.
All conjugates were purified by dialysis against PBS for 3 days and stored at -20°C. MALDI-TOF/MS was used to measure the hapten-protein molar ratios.

Polyclonal antibody production
Nine female New Zealand white rabbits (each weighing 2-3 kg) were used for immunization. The rabbits were injected intradermally at multiple sites on a shaved area of the back (about 30 sites) with 1 mg of immunogen in 0.2 mL of PBS emulsified with 0.2 mL of Freund's complete adjuvant. For booster injections, the same amount of immunogen was emulsified in incomplete Freund's adjuvant and injected subcutaneously at 10 sites at 3-week intervals for 18 weeks [23] . Blood specimens were collected 7 days after immunization. Antisera were tested by indirect ELISA and icELISA as described below.

ELISA optimization
The polyclonal antibody was characterized in terms of titer and sensitivity by indirect ELISA (iELISA) and indirect competitive ELISA (icELISA), and after confirmation of sensitivity, the antibody was used in subsequent experiments. The optimized method was consistent with that published previously [22] . The checkerboard method was used to optimize the concentration of coating antigen, and the coating buffer, temperature, pH and ionic strength were evaluated to establish a sensitive icELISA procedure. Lower IC 50 values and a suitable absorbance value (from 1.5 to 2.0) were selected as the criteria. The icELISA method was essentially carried out as described previously [22,24] . Briefly, 96-well microplates were coated with 100 mL per well of coating antigen in different buffer and incubated at 37°C for 2 h or 4°C overnight. Plates were washed three times with PBST and blocked with 200 mL of blocking buffer and incubated at 37°C for 1 h. After removal of blocking buffer, 50 mL per well of standard solution and 50 mL per well of antibody were added and incubated for 30 min at 37°C. Following a washing step, goat anti-mouse IgG-HRP (diluted 1:5000 in PBS, 100 mL per well) was added and incubated for another 30 min. After another washing step, 100 mL per well of substrate solution was added and incubated for 15 min. The reaction was stopped using 50 mL per well of stop solution. The optical density was measured at 450 nm and 630 nm with a Multiskan MK3 microplate absorbance reader.
The IC 50 value of BPA was calculated using a titration curve. The binding inhibition was determined for BPB, BPE, BHPVA, 17β-estradiol, diethylstilbestrol and nonylphenol. Cross-reactivity (CR) values were calculated as follows [24] : CR ¼ ðIC 50 of BPA=IC 50 of competitorÞ Â 100 2.7 Sample preparation BPA-free skimmed milk was supplied by the National Reference Laboratory for Veterinary Drug Residues (Beijing, China). Milk samples were centrifuged at 3500 g for 10 min at 4°C, and the upper fat layer was completely removed. Fat-free milk samples were then diluted twice with 0.1% Tween 20 prior to immunoassay. Twenty blank samples were subjected to icELISA as control, and the limit of detection (LOD) was calculated using the following equation:

Synthesis of haptens and conjugates
The design of haptens is important in the development of a specific antibody. A suitable hapten for immunization should exhibit a near perfect mimic of the target molecule structure, and share similar electronic and hydrophobic properties. Tethering of the hapten to an antigenically inert handle at a location distal to the determinant group(s) should then be performed to avoid masking or altering any functional groups [25,26] . Two basic approaches have been used for preparing BPA conjugates; one involving BHPVA [1,13,18] , and the other involving the attachment of various spacer arms between the hydroxyl group of BPA and the terminal carboxyl group of the carrier protein used for coupling, as demonstrated for BPA-GA, BPA-CPE, BPA-CME, BPA-CBE and BPA-CPE (Fig. 1a) [12,13,15,16,20] . In this study, to preserve both potential immunodominant epitope phenolic groups, BHPVA was used as the hapten for conjugation, and hapten OYPPP, which has one remaining phenolic group and the methyl group of BPA, was initially synthesized and coupled to the carrier protein. Furthermore, in order to avoid any other functions of BPA, 4-aminobutyric acid was introduced into the benzene ring in hapten HHPBA. The reaction scheme is shown in Fig. 1b

Screening of polyclonal antibodies
All nine antisera were detected following the sixth immunization using non-competitive ELISA to determine the optimum dilution of coating antigens and antibodies, and icELISA was then used to determine the sensitivity using homologous coating antigens. As shown in Table 1, IC 50 values and suitable absorbance values (from 1.5 to 2.0) were used as selection criteria. The three immunogens could induce a significant immune response, but the immunogen used OYPPP as hapten was the most active, compared with haptens of HHPBA and BHPVA. Although the hapten of HHPBA was generated purposely to avoid any other functions of BPA, the sensitivity of these Note: a , the antisera were evaluated by icELISA for primary screening. The antiserum dilution was 20000 and coating conjugate dilution was 5000. "-" indicates that the rabbit died.
antibodies was lower than that of other haptens, partly due to the lower hapten-protein molar ratio. The sensitivity of antibodies derived from the BHPVA hapten was lower than those derived from OYPPP, presumably because the two phenolic groups act as potential immunodominant epitopes, and the structure of BPA was altered slightly following coupling with the carrier protein. Serum from the rabbit induced by OYPPP-BSA was eventually selected for subsequent experiments.

Comparison of homogeneous and heterogeneous icELISA
Heterogeneous formats can greatly enhance the affinity of antibodies toward their targets compared with coating antigen or tracer hapten [19,24,27,28] by substantially improving the sensitivity. In this study, HHPBA-OVA and BHPVA-OVA were used as heterogeneous coating antigens, and OYPPP-OVA was used as a homogeneous coating antigen, in order to compare their sensitivities. The IC 50 value of icELISA with HHPBA-OVA as a heterogeneous coating antigen was 3.83 mg$L -1 , compared with 22.6 mg$L -1 for OYPPP-OVA (Fig. 2a). However, an IC 50 value of 81 mg$L -1 was obtained when BHPVA-OVA was used as the coating antigen, further indicating that the phenolic and methyl groups of BPA act as the major epitopes. HHPBA-OVA was therefore selected for subsequent experiments.

ELISA optimization
The effects of various factors on icELISA (Table 2) were investigated using B 0 and IC 50 as selection criteria. CB (0.05 mol$L -1 , pH 9.6), 37°C and pH 7.4 were chosen for the coating buffer, the incubation temperature and the optimal pH of the assay buffer, respectively. The ionic strength had a strong effect on the immunoassay. Specifically, increasing the NaCl concentration from 0.1 to 0.4 mol$L -1 resulted in lower B 0 values. However, a higher NaCl concentration (0.4 mol$L -1 ) may affect antigen-antibody binding, hence, 0.1 mol$L -1 NaCl in PBS buffer was considered the optimal ionic strength.

Sensitivity and specificity of the immunoassay
Based on the optimized conditions, standard solutions were diluted in PBS at concentrations of 0, 0.33, 1, 3, 9, 27, 81 mg$L -1 , and a standard curve for icELISA was determined (Fig. 2b), resulting in an IC 50 value of 2.1 mg$L -1 . IC 50 values of other structurally related compounds and CR values when the cross reactivity with BPA was set as 100% are shown Table 3. The antibody displayed lower CR values than BPB and BPE, but differences in CR values for other related compounds were negligible ( < 0.2%), confirming the high specificity of the polyclonal antibody for BPA. In addition, we summarized in Table 4 the haptens and  IC 50 values reported recently in the literature. It can be concluded that the icELISA we developed was more sensitive (expressed by IC 50 values) than some in the literature [1,12,15,16,18,31] , but similar to others in published literature [13,29,30] . We suggest that the new immunogens OYPPP-BSA and heterogeneous coating antigens HHPBA-OVA both contributed to enhance the sensitivity of the icELISA for the detection of BPA.

Assay performance in milk
To demonstrate the practicality of the icELISA method, milk samples were spiked with various concentrations of BPA, and the resulting LOD was 1.95 mg$L -1 . Recovery and coefficient of variation (CV) values are shown in

Conclusions
Three BPA immunizing haptens were used to produce polyclonal antibodies and all three immunogens were able to induce a significant immune response against BPA.