Food Chemistry 171 (2015) 98–107

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Analytical Methods

Hapten synthesis, monoclonal antibody production and developmentof a competitive indirect enzyme-linked immunosorbent assayfor erythromycin in milk

http://dx.doi.org/10.1016/j.foodchem.2014.08.1040308-8146/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +86 10 6273 2803; fax: +86 10 6273 1032.E-mail address: sjz@cau.edu.cn (J. Shen).

Zhanhui Wang a, Tiejun Mi a, Ross C. Beier b, Huiyan Zhang a, Yajie Sheng a, Weimin Shi a, Suxia Zhang a,Jianzhong Shen a,⇑a College of Veterinary Medicine, China Agricultural University, Beijing Key Laboratory of Detection Technology for Animal-Derived Food Safety, Beijing Laboratory For Food Qualityand Safety, Beijing 100193, People’s Republic of Chinab Food and Feed Safety Research Unit, Southern Plains Agricultural Research Center, Agricultural Research Service, U.S. Department of Agriculture, 2881 F&B Road, College Station,TX 77845-4988, USA

a r t i c l e i n f o

Article history:Received 25 April 2013Received in revised form 13 August 2014Accepted 23 August 2014Available online 6 September 2014

Keywords:Cross-reactivityEnzyme-linked immunosorbent assayErythromycinHapten incorporationMacrolidesMonoclonal antibody

a b s t r a c t

Erythromycin is an antibiotic used extensively in veterinary practice worldwide for treatment, preven-tion and growth promotion. In this work, monoclonal antibodies (Mabs) against erythromycin wereproduced and used to develop a competitive indirect enzyme-linked immunosorbent assay (ciELISA)for the determination of erythromycin in milk. A novel carboxyphenyl derivative of erythromycin(ERO-CMO) was synthesized and conjugated with bovine serum (BSA) for use as the immunogen or oval-bumin (OVA) as the coating antigen. Four hybridoma cell lines were isolated, which produced Mabs thatcompeted with erythromycin. The 6C1 and 5B2 Mabs had IC50 values for erythromycin of 14.40 and0.94 lg L�1, respectively. These Mabs demonstrated high cross-reactivity to the macrolides containing14-membered rings, but not to oleandomycin. No cross-reactivity was observed for 12 macrolides thatcontained 15 or 16-membered lactone rings or for 2 pleuromutilins. The ciELISA developed using the5B2 Mab afforded recovery values that ranged from 76.9% to 85.7% with only a 10-fold sample dilutionprior to analysis.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Erythromycin is one of the most commonly used macrolideantibiotics in veterinary medicine to treat respiratory diseasesand enteric infections in swine, cattle, sheep and poultry. It hasalso been used on a large scale as a feed additive and also deliveredvia drinking water for animal growth promotion (Díaz-Cruz &Barceló, 2007; McGlinchey, Rafter, Regan, & McMahon, 2008). Inhumans, erythromycin is often administered to those who areallergic to penicillin, and it has proven to be a safe and effectivetherapy for a number of commonly encountered infections (Rayet al., 2004). In aquaculture, erythromycin is used to treat infec-tions from Gram-positive bacteria, such as Lactococcus garvieae introut (Lucchetti et al., 2005). Erythromycin has low toxicity, butresidues in food animals might provoke allergic reactions in somehypersensitive individuals or lead to drug-resistant pathogenicbacteria (Thong, 2010; Wieczorek, Kania, & Osek, 2013). In

addition, the use of these antibiotics has resulted in their releaseinto the environment through different pathways; thereby, posinga potential risk to the ecosystem, as well as human and animalhealth by shifting the physiological profile of microbial communi-ties (Koike et al., 2007; Maul, Schuler, Belden, Whiles, & Lydy,2006). For these reasons, the European Union (EU), Ministry ofAgriculture (MOA) in China and other international bodies haveestablished maximum residue limits (MRLs) for erythromycin infood-producing animal species (Font et al., 2008). For example,the EU and MOA in China have set MRLs for erythromycin in milkat 40 lg L�1.

Currently, methods for the determination of erythromycin indifferent matrices are often based on chromatographic techniquesusing various detection systems (Avramov Ivic et al., 2008; Huet al., 2010; Minh, Lam, & Giao, 2011; Tao et al., 2012; Ye,Weinberg, & Meyer, 2007). Although these methods are sensitiveand selective, there is still a need for rapid and cost-effective alter-natives to screen large numbers of samples, and in particular foron-site detection. An effective alternative, based on specific anti-gen–antibody interactions are immunoassays, which are low cost

Z. Wang et al. / Food Chemistry 171 (2015) 98–107 99

and sensitive methods capable of screening large numbers of sam-ples. A few immunoassays have been developed to analyze forerythromycin in foods, such as radioimmunoassay, enzyme-linkedimmunosorbent assay (ELISA), and electrochemical ELISA (Ammidaet al., 2004; Campagnolo et al., 2002; Draisci et al., 2001; Situ &Elliott, 2005; Situ, Grutters, van Wichen, & Elliott, 2006; Tanakaet al., 1988; Yao & Mahoney, 1989). The antibodies used in thesestudies were either commercial or from unspecified sources. Theonly commercial immunoassay available for the detection of eryth-romycin is the Charm 6600/7600 system. The Charm system’s limitof detection (LOD) for erythromycin is 40 lg L�1 in milk; therefore,the Charm system is working at its LOD when measuring the MRL.

The key reagent in an immunoassay is the antibody. To theauthor’s best knowledge, no papers describing the production ofantibodies against erythromycin have been published. In the pres-ent study, we describe the synthesis of a hapten, production of amonoclonal antibody against erythromycin, and development ofa competitive indirect ELISA (ciELISA) for the determination oferythromycin in milk with improved sensitivity.

2. Experimental

2.1. Reagents and apparatus

Erythromycin A (91.6%), spiramycin (94%), josamycin (98%) andclarithromycin were obtained from TCI chemicals (Shanghai, P. R.China). The reference standard of azithromycin (100%), tulathro-mycin (100%) and desosaminylazithromycin (100%) were obtainedfrom Pfizer Pharmaceuticals Limited (New York, USA). Valnemulinand tiamulin were obtained from the Council of Europe’s EuropeanPharmacopoeia (Strasbourg, France). Tylosin and tilmicosin (93.7%)were obtained from the China Institute of Veterinary Drug Control(Beijing, P. R. China). Kitasamycin and acetylspiramycin (90.2%)were obtained from the China Pharmaceutical Biological ProductsAnalysis Institute (Beijing, P. R. China). Erythromycylamine wasobtained from Toronto Research Chemicals Inc. (North York,Ontario, Canada). Avermectin, roxithromycin (97%) and erythro-mycin ethyl succinate (98%) were obtained from Dr. EhrenstorferGmbH (Augsburg, Germany). Dirithromycin (>95%), oleandomycintriacetate, carboxymethoxylamine hemihydrochloride (CMO),bovine serum albumin (BSA), ovalbumin (OVA), N-hydroxy-succinimide (NHS), 1-(3-dimethylaminopropyl)-3-ethyl carbodi-imide (EDC), isobutyl chloroformate, and tributylamine wereobtained from Sigma–Aldrich Chemical Co. (St. Louis, MO, USA).The peroxidase-conjugated goat anti-mouse IgG and Dulbecco’sModified Eagle’s Media (DMEM) used for cell culture wereacquired from Huamei Biotech Co. (Beijing, P. R. China). Reagentgrade solvents and salts were supplied by Beijing ChemicalReagent Co. (Beijing, P. R. China). Incomplete Freund’s adjuvant(IFA), complete Freund’s adjuvant (CFA), and fetal calf serum wereobtained from Gibco BRL (Carlsbad, CA, USA). Deionized water wasprepared using a Milli-Q water purification system (Millipore, Bed-ford, MA, USA). Polystyrene microtiter plates were obtained fromCostar Inc. (Cambridge, MA, USA). The ELISA plate reader wasobtained from TECAN Inc. (Durham, NC, USA). Erythromycin-freeskimmed milk was supplied by the National Reference Laboratoryfor Veterinary Drug Residues (Beijing, P. R. China).

The hapten chemical structure was confirmed by high-perfor-mance liquid chromatography–tandem mass spectrometry (HPLC–MS/MS). Chromatography was performed on a Waters Alliance2690 LC system (Waters Corp., Milford, MA, USA) and thecolumn effluent was detected by a Quattro LC triple-quadrupole mass spectrometer (Micro-mass, Manchester, UK),which was connected to the LC system via an electrosprayionization (ESI) interface. Protein conjugates were evaluated by

matrix-assisted laser desorption/ionization time of flight mass spec-trometry (MALDI–TOF-MS) (Bruker, Daltonics, Billerica, MA, USA).

2.2. Buffers and solutions

Stock standard solutions (5 g L�1) of all antibiotics were pre-pared by dissolving an appropriate amount of each standard indimethylformamide (DMF). The individual stock solutions werestored at �20 �C in amber glass bottles. Working standards of eachantibiotic were prepared by diluting the stock standard solution inassay buffer and were stored at 4 �C.

The following buffers were used in the enzyme-linked immuno-sorbent assay (ELISA): (1) coating buffer was 0.05 mol L�1 carbon-ate buffer, pH 9.6; (2) blocking buffer consisted of 0.01 mol L�1 PBS,pH 7.4, 0.5% BSA, and 0.01% Tween 20; (3) washing buffer was0.01 mol L�1 PBS with 0.05% Tween 20; (4) antibody dilution bufferwas 0.01 mol L�1 PBS, containing 0.2% albumin; (5) the enzymelabeled secondary antibody dilution buffer was the antibody dilu-tion buffer containing 5% albumin; (6) substrate was 0.1% 3,3,5,50-tetramethylbenzidine (TMB) and 50% H2O2 in 0.05 M citrate buffer,pH 4.5; and (7) the stopping reagent was 2 mol L�1 H2SO4.

2.3. Hapten synthesis

Erythromycin A (100 mg) was dissolved in absolute ethanol(6 mL), and 44 mg of CMO was dissolved in water (2 mL) and thenadded dropwise to the erythromycin A solution. The pH wasadjusted to 5.5 by using a 1 mol L�1 NaHCO3 solution. The mixturewas held at 50 �C for 5 h before cooling to room temperature.Dichloromethane (20 mL) was added and the organic phase wasremoved under reduced pressure, leaving a brown oily substance.The hapten ERY-CMO was confirmed by HPLC–MS/MS.

2.4. Preparation of protein conjugates

The immunogens were prepared by conjugating the hapten toBSA by a mixed anhydride method (Rajkowski, Cittanova,Desfosses, & Jayle, 1977). The hapten (5 mg or 10 mg) was dis-solved in 2 mL of DMF in an ampule, and the solution was cooledby placing the ampule into cold ethanol (4 �C). Then 10 lL of tri-ethylamine and 10 lL of isobutyl chloroformate were added. Theresulting mixed anhydride solution was stirred at room tempera-ture for 20 min before 5 mL of 10 g L�1 BSA in carbonate buffer(pH 9.6) was added dropwise, followed by continuous stirring atroom temperature for 6 h. Two immunogens, ERY-CMO-BSA7.5

and ERY-CMO-BSA15 (where the subscripts reflect the ratio of themoles of hapten to the moles of BSA in the reaction mixture), dif-fering in the extent of hapten incorporation, were dialyzed againstPBS (pH 7.0) at 4 �C for 72 h.

The coating antigens were prepared by conjugating the samehapten to OVA using the NHS ester method: the hapten (5 mg)was dissolved in 1 mL of DMF, and the solution was cooled withcold ethanol (4 �C). NHS (5 mg) and EDC (5 mg) were added tothe hapten solution and stirred at room temperature overnight.OVA (10 mg) was dissolved in 3 mL of carbonate buffer (pH 8.0)and added drop-wise to the NHS solution with continuous stirring,and then further stirred at room temperature for 4 h. The ERY-CMO-OVA conjugates were dialyzed against PBS (pH 7.0) at 4 �Cfor 72 h, and characterization of the protein conjugates was per-formed by MALDI–TOF-MS by comparing the observed molecularweights of the prepared conjugates with the unreacted protein.

2.5. Production of monoclonal antibodies

Mice were euthanized by cervical dislocation and manipulatedin compliance with Chinese laws and guidelines (GKFCZ2001545)

100 Z. Wang et al. / Food Chemistry 171 (2015) 98–107

and according to the China Agriculture University regulations con-cerning protection of animals used for scientific purposes (2010-SYXK-0037).

Antibody production was similar to that of Zhang, Wang,Nesterenko, Eremin, and Shen (2007). Briefly, eight female BALB/c mice, 6–8 weeks old, were immunized subcutaneously with eachreceiving 50 lg of immunogen in 0.25 mL of 0.9% NaCl and 0.25 mLof CFA. Two, four and 6 weeks after the initial injection, animalswere boosted with 25 lg of immunogen in IFA. One week afterthe third and fourth injection, sera were collected from the mouseeye socket and assayed by ELISA. The mouse showing the highestantibody titer and sensitivity was sacrificed and splenocytes werefused with SP2/0 myeloma cells using polyethylene glycol (PEG2000). The fused cells were propagated in hypoxanthine–aminopterin–thymidine (HAT) medium, and plated in six 96-wellmicro-culture plates. Eight to ten days after cell fusion, culturesupernatants were screened for the presence of antibodies thatrecognized erythromycin. The hybridomas excreting antibodiesspecific to erythromycin were subcloned by the limiting dilutionmethod (Mercader, Suárez-Pantaleón, Agulló, Abad-Somovilla, &Abad-Fuentes, 2008). The clones of interest were transferred fromthe 96-well plate cultures to 24-well plates containing 0.5 mL ofculture medium. After the hybridomas achieved dense growth inthe 24-well plates, they were transferred to 20 mL culture flasks.Hybridoma cells were collected, centrifuged, and the supernatantswere stored at �20 �C until used.

2.6. ELISA methods

A non-competitive ELISA was carried out to determine the seratiter using the following procedure: polystyrene microtiter plates(96-well) were coated with coating antigen (100 lL well�1) andincubated at 4 �C overnight. The plates were washed three timeswith washing buffer, and then blocked with blocking buffer(300 lL well�1) at 37 �C for 2 h. Antibody (50 lL well�1) and assaybuffer (50 lL well�1) were added to each of the wells. The plateswere incubated for 1 h at 37 �C and the unbound compounds wereremoved by washing three times with the washing buffer. Goatanti-mouse IgG-HRP (1/3000 in PBS, 100 lL well�1) was addedand incubated at 37 �C for 1 h and then washed three times withthe washing buffer. The substrate solution (100 lL well�1) wasadded and incubated at 37 �C for 30 min before adding 2 mol L�1

H2SO4 (100 lL well�1). Absorbance values were measured at450 nm.

The ciELISA procedure was similar to that of the noncompetitiveELISA. The only difference was in the addition of the antibody. Inthe ciELISA procedure, 50 lL of antibody prepared in assay bufferand 50 lL of the selected antibiotic standard solution (or sample)were added to the wells instead of 50 lL of PBS. The absorbance(OD) at 450 nm was then read.

2.7. Curve fitting and statistical analysis

The OD values were plotted against the analyte concentrationon a logarithmic scale, and the generated sigmoidal curve wasmathematically fitted to the following four-parameter logisticequation using the OriginPro 7.5 software (OriginLab Corporation,Northampton, MA, USA):

Y ¼ ðA� DÞ½1þ ðX=CÞB þ D�

where A = response at high asymptote, B = the slope factor,C = concentration corresponding to 50% specific binding (IC50),D = response at low asymptote, and X = the calibrationconcentration.

The LOD of the assay was defined as the concentration of ana-lyte that provided a 10% reduction of the ODmax (IC10) (Lee, Ahn,Park, Ko, & Kim, 2002). The dynamic range of the assay was estab-lished between the values of IC20 and IC80 (Brady, 1995).

Cross-reactivity values were calculated according to the follow-ing equation:CR (%) = (IC50molar (ERY, lmol L�1)/IC50molar (analogs,lmol L�1)) � 100.

2.8. Matrix effects of milk

Erythromycin standard curves were prepared in PBS containingdifferent proportions of milk. Non-specific interferences producedby the milk were evaluated by preparing the erythromycin stan-dard in several different dilutions of milk and comparing thesestandard curves with the erythromycin standard curve preparedonly in PBS.

2.9. Sample preparation

The accuracy and precision of the ciELISA was evaluated by per-forming recovery studies with spiked milk samples at 10, 20 and30 lg L�1, which were then diluted 10-fold in assay buffer to avoidmatrix interferences. Each sample was determined in triplicate,and the average absorbance values were interpolated with a stan-dard curve prepared in assay buffer with each data point being anaverage of three determinations.

3. Results and discussion

3.1. Preparation of ERY-CMO and its conjugates

Erythromycin contains a 14-membered lactone ring and twosugars (L-cladinose and D-desosamine) and has a molecular weightof 733.93; however, it still belongs in the small molecule category,in which members do not elicit immunogenicity. To generate anantibody to erythromycin, it must be conjugated to an immuno-genic macromolecule to elicit an immune response (Shelver,Shappell, Franek, & Rubio, 2008). There are five hydroxyls and oneketone group in the chemical structure of erythromycin, whichcan theoretically be used to link it to a carrier protein (Fig. S1, seeAppendix A. Supplementary data). To ensure the selectivity of thereaction, CMO was used as a cross-linking reagent, which undergoescondensation with the exclusive ketone group of erythromycin atposition 9 to form the oxime (Fig. S1, see Appendix A. Supplemen-tary data). The hapten ERY-CMO, was identified by HPLC–MS/MS,as shown in Fig. S2 (see Appendix A. Supplementary data), wherethe molecular ions (m/z) of ERY-CMO are 805.5 and 807.5 in nega-tive ion mode and positive ion mode, respectively, indicating thatCMO was successfully conjugated with erythromycin A.

The ERY-CMO hapten was coupled to BSA in molar ratios of hap-ten:protein (7.5:1 and 15:1) using the mixed anhydride method(Fig. S1, see Appendix A. Supplementary data). A shift in the molec-ular weight of the molecular ion was observed in comparison withthe control protein (Fig. S3, see Appendix A. Supplementary data),demonstrating that the ERY-CMO hapten had been conjugated tothe carrier protein. The ERY-CMO-BSA7.5 mean coupling densitywas 0.61, and for ERY-CMO-BSA15 the mean coupling density wasdetermined to be 1.56. Fig. S3 (see see Appendix A. Supplementarydata.) shows the MALDI–TOF mass spectra obtained for the BSA ref-erence sample, ERY-CMO-BSA7.5, and ERY-CMO-BSA15.

3.2. Antibody production

The parameters for the antisera obtained from each mousefrom the final bleeding are shown in Table S1 (see Appendix A.

Table 1ciELISA assay parameters of four Mabs obtained from mice immunized with ERY-CMO-BSA7.5 and ERY-CMO-BSA15 using erythromycin as the competitora.

Antibodies ERY-CMO-BSA15 ERY-CMO-BSA7.5

Antisera 6C1 5D1 Antisera 5B2 6D9

Antibody dilution 1/1600 1/200,000 1/50,000 1/800 1/100,000 1/100,000Antigen dilution 1/4000 1/50,000 1/50,000 1/900 1/30,000 1/50,000ODmax 1.729 1.864 1.633 1.617 1.866 1.554IC50 (lg L�1) 16.6 14.40 26.4 2.91 0.94 8.90

a Mean values of three independent determinations.

1E-4 1E-3 0.01 0.1 1 10 100 1000 10000-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

OD

Val

ues

Erythromycin (µg L-1)

6C1 5B2

Fig. 1. Standard curves for erythromycin with Mab 6C1 and 5B2. Each pointrepresents the mean ± standard deviation from three determinations.

Z. Wang et al. / Food Chemistry 171 (2015) 98–107 101

Supplementary data). Generally, the mice in the group receivingimmunogen ERY-CMO-BSA15 had higher titers and IC50 values thanthose receiving ERY-CMO-BSA7.5. All eight mice immunized withERY-CMO-BSA15 showed an antisera dilution over 1/800 with thelowest ODmax of 1.283. The mice immunized with ERY-CMO-BSA7.5 had antisera dilutions below 1/1000 with the highest ODmax

of 1.617 (Table S1, see Appendix A. Supplementary data). Followingthe evaluation of antisera titer, the antisera sensitivities for eachmouse were obtained by calculating the IC50 of each antisera.The results are summarize in Table S1 (see Appendix A.Supplementary data) and show that all of the mice receivingERY-CMO-BSA15 had affinity to erythromycin and also IC50 valuesranging from 16.6 to 39.4 lg L�1. However, in the group of micereceiving ERY-CMO-BSA7.5, only two mice had affinity to erythro-mycin and surprisingly low IC50 values of 2.91 and 3.72 lg L�1

were recorded. The antisera screening results indicate (1) therelationship between antibody sensitivity and titer was not pro-portional, (2) an immunogen having a hapten:protein ratio of less

Table 2ELISA parameters using monoclonal antibodies 6C1 and 5B2 in buffer and milka.

Parameter 6C1b

Ab dilution (lg L�1) 122Ag dilution (lg L�1) 15.6ODmax 1.864 ± 0.132ODmin 0.088 ± 0.00354Slope 0.74 ± 0.09Detection range (IC20–IC80) (lg L�1) 2.22 ± 0.21–86.05 ± 2.24IC50 (lg L�1) 14.40 ± 0.15Sample dilution –LOD (lg L�1) 0.65 ± 0.13R2 0.999 ± 0.005

a The parameters are extracted from the four parameter equation used into fit the stab The data presented correspond to the average of three standard curves run three di

than 8:1 can often generate antibodies, although a hapten:proteinratio in the range of 8:1–25:1 is generally recommended for thebest antibody production, and (3) reducing the extent of haptenincorporation in the immunogen produced a more sensitiveantibody.

Mouse 8 in the ERY-CMO-BSA15 group and mouse 2 in the ERY-CMO-BSA7.5 group were both used to produce monoclonal antibod-ies because of their relatively higher sensitivities. A total of fourhybridomas were obtained. Two clones were from mouse 8 inthe ERY-CMO-BSA15 group (6C1 and 5D1) and two clones werefrom mouse 2 in the ERY-CMO-BSA7.5 group (5B2 and 6D9). Thesemonoclonal antibodies were prepared as ascites without furtherpurification and then characterized (Table 1). As seen in Table 1,the optimum antibody dilution for 6C1 and 5B2 was 1/200,000and 1/100,000, which improved the antibody titers by 125-foldcompared to the corresponding antisera and the ODmax remainedbetween 1.5 and 2.0, while the IC50 of the ascites increased by1.2–3.1 times, respectively. The opposite trend was observed for5D1 and 6D9 when compared to their corresponding antisera;i.e., they had increased antibody titer but decreased sensitivity(Table 1). The typical standard curves for erythromycin, preparedwith 6C1 and 5B2, are shown in Fig. 1, and Table 2 summarizesthe parameters defining the standard curves for the ciELISA. TheLODs achieved were 0.65 and 0.057 lg L�1 for 6C1 and 5B2, respec-tively. In particular, the low LOD value achieved with the 5B2-based ciELISA did not compromise the detection of the low levelsnecessary for erythromycin monitoring.

3.3. Cross-reactivity

Investigations on specificity of the antibody produced arecrucial for assessment of the results. The specificities of 6C1 and5B2 were evaluated with 17 macrolides and two pleuromutilins,as shown in Table 3. Both Mabs had similar recognition patterns,but different affinities. The Mabs, 6C1 and 5B2, exhibited highcross-reactivity to the macrolide antibiotics with 14-memberlactone rings; i.e., erythromycin (100% and 100%), erythromycinethyl succinate (70.5% and 43.7%), erythromycylamine (32.5% and16.7%), dirithromycin (242% and 157%), roxithromycin (21.3% and

5B2b ciELISAb

333 33325.9 25.91.878 ± 0.017 1.692 ± 0.00140.045 ± 0.00707 0.038 ± 0.00640.84 ± 0.11 0.83 ± 0.080.16 ± 0.04–6.01 ± 0.57 0.85 ± 0.09–32.3 ± 0.070.94 ± 0.02 3.5 ± 0.11– 100.057 ± 0.032 0.3 ± 0.0240.999 ± 0.009 0.993 ± 0.004

ndard curve. Each curve was built using three well replicatesfferent assays.

Table 3Cross-reactivity of several structurally related analogs to the Mabs in the ciELISAa.

Antibiotics Structure Ring carbons/MWb

6C1 5B2

IC50 CRc

(100%)IC50 CRa

(100%)lg L�1 lmol L�1 lg L�1 lmol L�1

Erythromycin 14/733.93 14.40 0.0196 100 0.94 0.00128 100

Erythromycin ethylsuccinate

14/862.05 23.91 0.0278 70.5 2.53 0.00293 43.7

Erythromycylamine 14/734.96 44.35 0.0603 32.5 5.64 0.00767 16.7

Dirithromycin 14/835.07 6.80 0.0081 242 0.81 0.00814 157

Roxithromycin 14/837.05 77.22 0.0922 21.3 7.23 0.00864 14.8

102 Z. Wang et al. / Food Chemistry 171 (2015) 98–107

Table 3 (continued)

Antibiotics Structure Ring carbons/MWb

6C1 5B2

IC50 CRc

(100%)IC50 CRa

(100%)lg L�1 lmol L�1 lg L�1 lmol L�1

Clarithromycin 14/747.95 50.19 0.0671 29.2 3.59 0.00480 26.7

Oleandomycin 14/785.85 >10000d –e – – – –

Azithromycin 15/785.01 1472 1.874 1.05 107.53 0.1370 0.93

Desosaminylazithromycin 15/590.79 >10000 – – – – –

Tulathromycin 15/806.23 >10000 – – – – –

(continued on next page)

Z. Wang et al. / Food Chemistry 171 (2015) 98–107 103

Table 3 (continued)

Antibiotics Structure Ring carbons/MWb

6C1 5B2

IC50 CRc

(100%)IC50 CRa

(100%)lg L�1 lmol L�1 lg L�1 lmol L�1

Spiramycin 16/843.07 >10000 – – – – –

Acetylspiramycin 16/885.09 >10000 – – – – –

Kitasamycin 16/785.96 >10000 – – – – –

Josamycin 16/884.07 >10000 – – – – –

Tylosin 16/916.11 >10000 – – – – –

Tilmicosin 16/ 869.13 >10000 – – – – –

Avermectins 16/873.08 >10000 – – – – –

104 Z. Wang et al. / Food Chemistry 171 (2015) 98–107

1E-4 1E-3 0.01 0.1 1 10 100 1000-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

OD

Vau

les

Erythromycin (µg L-1)

PBS (0.94 µg L-1)

1/10 diluted milk (0.35 µg L-1)

1/5 diluted milk (0.37 µg L-1)

milk (0.46 µg L-1)

Fig. 2. Comparison of ciELISA curves obtained from standards prepared in PBS, milk, 1/5 diluted milk, and 1/10 diluted milk.

Table 3 (continued)

Antibiotics Structure Ring carbons/MWb

6C1 5B2

IC50 CRc

(100%)IC50 CRa

(100%)lg L�1 lmol L�1 lg L�1 lmol L�1

Valnemulin –/564.82 >10000 – – – – –

Tiamulin –/493.74 >10000 – – – – –

a Each data point is the average of three experimental evaluations.b The number of atoms in the lactone nucleus and the molecular weight of macrolide antibiotics.c Cross-reactivity is expressed as a percent of the IC50 (lmol L�1) of the erythromycin divided by the IC50 (lmol L�1) of the other antibiotics tested.d >10000 in the table means that antibiotics at the concentration of 10000 lg L�1 cannot decreased in absorbance values.e Not calculation.

Z. Wang et al. / Food Chemistry 171 (2015) 98–107 105

14.8%) and clarithromycin (29.2% and 26.7%), respectively, exceptfor oleandomycin (where both Mabs had cross-reactivity>10,000 lg L�1). Other macrolide antibiotics containing 15-mem-bered lactone rings (desosaminylazithromycin and tulathromycin)or 16-membered lactone rings (spiramycin, acetylspiramycin,kitasamycin, josamycin, tylosin, tilmicosin and avermectins) orthe pleuromutilins (valnemulin and tiamulin) were not recognizedby either Mab. However, azithromycin, a 15-member lactone ringmacrolide, was recognized by both Mabs with IC50 values of1.874 and 0.137 lmol L�1 for Mabs 6C1 and 5B2, respectively.The recognition patterns of the Mabs showed that the occurrenceof a 14-membered lactone ring in the macrolide is necessaryfor high antibody affinity. The number of lactone ring atoms and

substituent groups on the lactone ring inevitably change the con-formation and electron distribution of the macrolide antibiotic,meaning they have a significant impact on antibody recognition.

We also observed that the distance between the substituentgroup on the lactone ring and the conjugation site of the haptenand carrier protein is important to antibody recognition. Thesubstitution at position 6 (clarithromycin), position 9 (erythromy-cylamine, dirithromycin and roxithromycin), and D-desosamine(erythromycin ethyl succinate) were not detrimental to antibodybinding because these substitutions are close to the conjugationsite (Table 3). However, the lack of one methyl group on L-cladi-nose of oleandomycin resulted in no antibody recognition eventhough oleandomycin does possess a 14-membered lactone ring.

Table 4ELISA recovery using skimmed milk spiked with erythromycin (n = 3)a.

Spiked amount(lg L�1)

Recovery(%)

Intra-assay RSD(%)

Inter-assay RSD(%)

10 85.7 6.1 11.320 78.4 5.1 9.830 76.9 5.7 8.1

a Each value represents the average of three independent experiments.

106 Z. Wang et al. / Food Chemistry 171 (2015) 98–107

In this case, the methyl group is away from the conjugation siteand would be expected to play a role in antibody recognition.The same conclusion also can be arrived at by comparing the IC50

values of erythromycin to azithromycin (a 15-membered lactonering) and desosaminylazithromycin (a degradation product ofazithromycin). The structural difference between azithromycinand desosaminylazithromycin lies only in the position of thesubstituted carbon-3 on the macrolide ring, which is a good dis-tance from the conjugation site. Both Mabs showed affinity toazithromycin (IC50 values of 1.874 and 0.137 lmol L�1). However,there was no affinity by either Mab to desosaminylazithromycin(IC50 values >10,000 lg L�1).

3.4. Matrix effect

The matrix effect of milk could hinder the quantification of thetarget of interest. When using an ELISA method for screening pur-poses, the sample preparation should be as simple as possible.Often times the influence of the sample matrix can be avoided bya simple dilution with assay buffer before analysis. Therefore, themilk matrix effect was evaluated by preparing erythromycinstandard curves in 0, 5, and 10-fold diluted milk and comparingthe data with the erythromycin standard curves prepared in onlybuffer. With an increase in the dilution factor, the maximum ODvalues increased along with relatively stable and improved IC50

values (Fig. 2). It was observed that inhibition curves withoutinterference were obtained after a 10-fold dilution of the milk,indicating that skimmed milk could be measured directly follow-ing dilution. The parameters of the ciELISA used with 10-folddiluted milk are shown in Table 2. The limit of detection of theciELISA was 0.3 lg L�1, which is more than 100-fold higher thanthe MRL of erythromycin in milk (40 lg L�1). The working rangeof the ciELISA was 0.85–32.3 lg L�1. The higher sensitivity of theciELISA was helpful for detection of erythromycin in milk at therequired residue level.

3.5. Recovery

The accuracy of the ELISA developed for erythromycin wasassessed by measuring spiked raw milk samples and estimatingthe overall recovery. Milk was fortified at different erythromycinconcentrations (10, 20 and 30 lg L�1). Three independent determi-nations of erythromycin from 10-fold diluted milk were conductedusing the developed ciELISA. Recoveries ranging from 76.9% to85.7% were obtained, with intra-assay RSD ranging from 5.1% to6.4% and inter-assay RSD ranging from 8.1% to 11.3% (Table 4).

4. Conclusions

An erythromycin hapten and high specificity Mabs to erythro-mycin were prepared. The effects of hapten incorporation on thetiter and affinity of the antibodies to erythromycin were investi-gated. The results of the study showed that the lower degree ofhapten incorporation resulted in a lower titer but a higher degree

of Mab sensitivity. A Mab-based ciELISA for the determination oferythromycin in milk was developed with a LOD of 0.3 lg L�1 inskimmed milk. Milk samples can be directly analyzed withoutcomplicated sample pre-treatment other than a 10-fold dilution.The ciELISA developed showed high cross-reactivity to the macro-lides containing a 14-membered ring, but not for oleandomycin.No cross-reactivity was observed for other macrolides tested,which contained a 15 or 16-membered lactone ring, or for pleuro-mutilins. The recoveries for erythromycin in skimmed milk were76.9–85.7% with the intra-assay RSD ranging from 5.1% to 6.1%and inter-assay RSD ranging from 8.1% to 11.3%.

Acknowledgements

This work is supported by grants from the Trans-CenturyTraining Programme Foundation for Talents by the Ministry ofEducation (NCET-12–0529) and the State Key Program of theNational Natural Science of China (No. 30830082).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2014.08.104.

References

Ammida, N. H. S., Volpe, G., Draisci, R., delli Quadri, F., Palleschi, L., & Palleschi, G.(2004). Analysis of erythromycin and tylosin in bovine muscle using disposablescreen printed electrodes. Analyst, 129, 15–19.

Avramov Ivic, M. L., Petrovic, S. D., Mijin, D. Z., Vanmoos, F., Orlovic, D. Z.,Marjanovic, D. Z., et al. (2008). The electrochemical behavior of erythromycin Aon a gold electrode. Electrochimica Acta, 54, 649–654.

Brady, J. F. (1995). Interpretation of immunoassay data. In J. O. Nelson, A. E. Karu, &R. B. Wong (Eds.), Immunoanalysis of agrochemicals: Emerging technologies, ACSsymposium series 586 (pp. 266–287). Washington, DC: American ChemicalSociety.

Campagnolo, E. R., Johnson, K. R., Karpati, A., Rubin, C. S., Kolpin, D. W., Meyer, M. T.,et al. (2002). Antimicrobial residues in animal waste and water resourcesproximal to large-scale swine and poultry feeding operations. Science of TheTotal Environment, 299, 89–95.

Díaz-Cruz, M. S., & Barceló, D. (2007). Recent advances in LC–MS residue analysis ofveterinary medicines in the terrestrial environment. TrAC Trends in AnalyticalChemistry, 26, 637–646.

Draisci, R., delli Quadri, F., Achene, L., Volpe, G., Palleschi, L., & Palleschi, G. (2001). Anew electrochemical enzyme-linked immunosorbent assay for the screening ofmacrolide antibiotic residues in bovine meat. Analyst, 126, 1942–1946.

Font, H., Adrian, J., Galve, R., Estévez, M.-C., Castellari, M., Gratacós-Cubarsi, M., et al.(2008). Immunochemical assays for direct sulfonamide antibiotic detection inmilk and hair samples using antibody derivatized magnetic nanoparticles.Journal of Agricultural and Food Chemistry, 56, 736–743.

Hu, X., Wang, P., Yang, J., Zhang, B., Li, J., Luo, J., et al. (2010). Enhancedelectrochemical detection of erythromycin based on acetylene blacknanoparticles. Colloids and Surfaces B: Biointerfaces, 81, 27–31.

Koike, S., Krapac, I. G., Oliver, H. D., Yannarell, A. C., Chee-Sanford, J. C., Aminov, R. I.,et al. (2007). Monitoring and source tracking of tetracycline resistance genes inlagoons and groundwater adjacent to swine production facilities over a 3-yearperiod. Applied Environmental Microbiology, 73, 4813–4823.

Lee, J. K., Ahn, K. C., Park, O. S., Ko, Y. K., & Kim, D. H. (2002). Development of animmunoassay for the residues of the herbicide bensulfuron-methyl. Journal ofAgricultural and Food Chemistry, 50, 1791–1803.

Lucchetti, D., Fabrizi, L., Esposito, A., Guandalini, E., Di Pasquale, M., & Coni, E.(2005). Simple confirmatory method for the determination of erythromycinresidues in trout: A fast liquid–liquid extraction followed by liquidchromatography–tandem mass spectrometry. Journal of Agricultural and FoodChemistry, 53, 9689–9694.

Maul, J. D., Schuler, L. J., Belden, J. B., Whiles, M. R., & Lydy, M. J. (2006). Effects of theantibiotic ciprofloxacin on stream microbial communities and detritivorousmacroinvertebrates. Environmental Toxicology Chemistry, 25, 1598–1606.

McGlinchey, T. A., Rafter, P. A., Regan, F., & McMahon, G. P. (2008). A review ofanalytical methods for the determination of aminoglycoside and macrolideresidues in food matrices. Analytica Chimica Acta, 624, 1–15.

Mercader, J. V., Suárez-Pantaleón, C., Agulló, C., Abad-Somovilla, A., & Abad-Fuentes,A. (2008). Production and characterization of monoclonal antibodies specific tothe strobilurin pesticide pyraclostrobin. Journal of Agricultural and FoodChemistry, 56, 7682–7690.

Z. Wang et al. / Food Chemistry 171 (2015) 98–107 107

Minh, N. P., Lam, T. B., & Giao, N. T. (2011). Simultaneous determination oferythromycin A in giant prawn and tilapia in mekong region by stripping squarewave voltammetry. European Journal of Food Research & Review, 1, 1–14.

Rajkowski, K. M., Cittanova, N., Desfosses, B., & Jayle, M. F. (1977). The conjugationof testosterone with horseradish peroxidase and a sensitive enzyme assay forthe conjugate. Steroids, 29, 701–713.

Ray, W. A., Murray, K. T., Meredith, S., Narasimhulu, S. S., Hall, K., & Stein, C. M.(2004). Oral erythromycin and the risk of sudden death from cardiac causes. TheNew England Journal of Medicine, 351, 1089–1096.

Shelver, W. L., Shappell, N. W., Franek, M., & Rubio, F. R. (2008). ELISA forsulfonamides and its application for screening in water contamination. Journalof Agricultural and Food Chemistry, 56, 6609–6615.

Situ, C., & Elliott, C. T. (2005). Simultaneous and rapid detection of five bannedantibiotic growth promoters by immunoassay. Analytica Chimica Acta, 529,89–96.

Situ, C., Grutters, E., van Wichen, P., & Elliott, C. T. (2006). A collaborative trial toevaluate the performance of a multi-antibiotic enzyme-linked immunosorbentassay for screening five banned antimicrobial growth promoters in animalfeedingstuffs. Analytica Chimica Acta, 561, 62–68.

Tanaka, Y., Kimura, K., Komagata, Y., Tsuzuki, K., Tomoda, H., & Omura, S. (1988).Radioimmunoassay for erythromycin derivatives. Journal of Antibiotics (Tokyo),41, 258–260.

Tao, Y., Yu, G., Chen, D., Pan, Y., Liu, Z., Wei, H., et al. (2012). Determination of 17macrolide antibiotics and avermectins residues in meat with acceleratedsolvent extraction by liquid chromatography–tandem mass spectrometry.Journal of Chromatography B, 897, 64–71.

Thong, B. (2010). Update on the management of antibiotic allergy. Allergy AsthmaImmunology Research, 2, 77–86.

Wieczorek, K., Kania, I., & Osek, J. (2013). Erythromycin-resistant Campylobacter colifrom slaughtered animals as a potential public health risk. Veterinarni Medicina,58, 352–358.

Yao, R. C., & Mahoney, D. F. (1989). Enzyme immunoassay for macrolide antibiotics:Characterization of an antibody to 23-amino-O-mycaminosyltylonolide. Appliedand Environmental Microbiology, 55, 1507–1511.

Ye, Z., Weinberg, H. S., & Meyer, M. T. (2007). Trace analysis of trimethoprim andsulfonamide, macrolide, quinolone, and tetracycline antibiotics in chlorinateddrinking water using liquid chromatography electrospray tandem massspectrometry. Analytical Chemistry, 79, 1135–1144.

Zhang, S., Wang, Z., Nesterenko, I. S., Eremin, S. A., & Shen, J. (2007). Fluorescencepolarisation immunoassay based on a monoclonal antibody for the detection ofsulphamethazine in chicken muscle. International Journal of Food Science &Technology, 42, 36–44.