A Monoclonal Antibody-Based ELISA for Multiresidue Determination of Avermectins in Milk

Due to the widespread use and potential toxicity of avermectins (AVMs), multi-residue monitoring of AVMs in edible tissues, especially in milk, has become increasingly important. With the aim of developing a broad-selective immunoassay for AVMs, a broad-specific monoclonal antibody (Mab) was raised. Based on this Mab, a homologous indirect enzyme-linked immunosorbent assay (ELISA) for the rapid detection of AVMs in milk was developed. Under the optimized conditions, the IC50 values in assay buffer were estimated to be 3.05 ng/mL for abamectin, 13.10 ng/mL for ivermectin, 38.96 ng/mL for eprinomectin, 61.00 ng/mL for doramectin, 14.38 ng/mL for emamectin benzoate. Detection capability (CCβ) of the ELISA was less than 5 ng/mL and 2 ng/mL in milk samples prepared by simple dilution and solvent extraction, respectively. The optimized ELISA was used to quantify AVMs in milk samples spiked at different amounts. The mean recovery and coefficient of variation (CV) were 95.90% and 15.42%, respectively. The Mab-based ELISA achieved a great improvement in AVMs detection. Results proved this broad-selective ELISA would be useful for the multi-residue determination of AVMs in milk without purification process.


Introduction
Avermectins (AVMs) are insecticidal/miticidal compounds derived from the soil bacterium, Streptomyces avermitilis. Five AVMs (Table 1), namely abamectin (ABM), ivermectin (IVM), eprinomectin (EPR), doramectin (DOR) and emamectin (EMA) are widely used in agriculture and food-producing animals for the treatment of a broad spectrum of parasitic diseases. AVMs are very effective against parasites at extremely low doses. However, toxicology research has showed that an overdose of AVMs could cause a combination of clinical side effects ranging from mild to extremely severe, including death [1]. Given the potential hazard that AVMs pose to human and animal health, maximum residue limits (MRLs) are established in many countries. Joint FAO/WHO Expert Committee on Food Additives recommend the MRLs for IVM, DOR, and EPR in milk are 10, 15, 20 μg/L, respectively [2]. In the China, the MRL for IVM in milk is 10 μg/L, whereas the use of ABM and DOR is prohibited in cattle producing milk for human consumption [3]. The conventional methods for detection of AVMs involving high performance liquid chromatography (HPLC) and liquid chromatography/mass spectrometry (LC/MS) are sensitive and reliable [4][5][6][7]. However, these applications are relatively time-consuming and usable only on a laboratory scale with expensive instruments. Since immunoassays have proven to be a powerful tool for high throughput and on-site screening analysis for surveillance and monitoring purpose in a wide variety of food matrices [8], broad-selective immunodetection prior to chromatographic determination may be an attractive approach for multi-residue monitoring. If the total quantity of two or more AVMs detected in a sample is less than the MRL, the sample needs no further inspection. x Immunochemical methods involve the use of antibodies, which are the key components of all immunoassays because their quality greatly contributes to the sensitivity and selectivity [9]. Some antibodies have been produced against single AVM [10][11][12][13][14][15][16], whereas few broad-specific antibodies have been reported. Schmidt et al. [17] described a monoclonal antibody (Mab) with high specificity, but the Mab could recognize only two AVMs. In our previous work [18], a broad-selective ELISA for three AVMs using a polyclonal antibody (Pab) was developed. Since Mab offered a more definite specificity than Pab and an unlimited production, in this work, a broad specific Mab was produced. Moreover, a broad-selective ELISA based on this Mab was developed, and applied to detect all the five AVMs in milk samples.

Antigen Synthesis
AVMs are small molecules and must be conjugated to a carrier protein to elicit an immune response. The design of the required immunogen is critical for the success of rapid immunoassay. AVMs have three available hydroxy groups (Table 1), and especially the 4′-OH can be modified to conjugate with a carrier protein for the production of antibodies against AVMs [17]. Moreover, 4′-O-succinoyl-ABM had been successfully used to produce broad-specific Pab against AVMs in our previous research [18]. In this study, the same immunogen was used for the Mab production, and the structure of ABM was used as the determinant group.
For assay purposes, 4′-O-succinoyl-ABM-OVA, 5-ABM-OVA and 5-ABM oxime-OVA were synthesized as coating antigens. Conjugate formation was confirmed spectrophotometrically. UV-Vis spectra showed qualitative differences between the carrier protein and conjugates in the region of maximum absorbance of ABM. Molar ratios of the conjugates ranged from 2 to 15.

Mab Production and Characterization
Since a single B-lymphocyte produces a single type of antibody molecule, screening for the hybridomas which could secrete broad specific Mabs, would be a pivotal process. In order to choose broad specific Mabs, indirect ELISA for IVM detection was performed to screen hybridomas. Five cell fusion experiments were conducted. After screening, two selected Mabs (named 2C11, 3A9) showed good recognition towards IVM. The hybridomas were injected intraperitoneally into mineral oil-primed mice to produce ascites and then purified. Their sensitivity and specificity were investigated by indirect ELISAs using three coating antigens ( Table 2). Neligible CR was observed from moxidectin, which belongs to milbemycins family and the structure is similar to AVMs but lacks the disaccharide group. Sensitivity results of ABM with the two Mabs and three coating antigens were very similar (IC 50 values ranging from 38.97 to 65.62 ng/mL). When the CR of ABM was set as 100%, broad selectivity was evaluated by the other four AVMs. The combination of Mab 2C11 and 4′-O-succinoyl-ABM-OVA was selected as the optimal assay for broad selective detection, with which the CR values were 50.46%, 14.17%, 19.82% and 47.81% for IVM, EPR, DOR and EMA benzoate, respectively.

Assay Optimization
Immunoassays are usually carried out under physiological conditions, and frequently influenced by several nonspecific parameters such as temperature, time, pH, ionic strength, and presence of organic solvent. With the aim of improving immunoassay performance, the influence of several nonspecific parameters on assay characteristics was examined. Maximal absorbance (A max )/IC 50 ratio was a convenient estimate of the influence of some factors on the ELISA sensitivity, the higher ratio indicating higher sensitivity [21].

Assay Buffer
Four combinations of Mab dilution buffer and standard preparation buffer were performed to investigate the effect of assay buffer. As shown in Table 3, the IC 50 value was significant increased when ABM standards were dissolved in PBS, which implied that the concentration of K + had a great impact on the ELISA property. Apart from A max /IC 50 ratio, dynamic range, and limit of detection were also taken into account. Therefore, PB was selected to dissolve both Mab and standards.

pH Effect
The plots of assay parameters (A max , IC 50 ) as a function of pH value were depicted in Figure 1. The ability of Mab to recognize the coating antigen (A max ) decreased gradually (from 1.89 to 0.38) as pH value increased from 5 to 9. In addition, the recognition of ABM (IC 50 ) did not change markedly (from 6.19 to 4.06 ng/mL) until pH value increased to pH 9. On the basis of the result that maximal A max /IC 50 ratio was 0.32 for pH 6, followed pH 6.5 and of the fact that antigen-antibody reaction was more stable in neutral environments, pH 6.5 seemed to be a reasonable choice for the competition step. Figure 1. Effect of pH on ELISA performance. For the competition step, standards and Mab were dissolved in PB of different pH, and the GαM HRP was diluted in PB of pH 7.1. Incubation temperature was 37 °C.

Methanol Concentration
Since AVMs are highly lipophilic, and organic solvents are often used to extract analytes from samples in the immunoassays, it is necessary to use a water-miscible organic solvent in the assay buffer for ELISAs. Methanol was the most popular solvent for immunoassays [22,23], and 10% methanol was selected previously for the Pab-based ELISA against AVMs [18]. In this work, the effect of solvent on the ELISA performance was evaluated by preparing standard curves in buffers containing various concentration of methanol ( Figure 2). On the one hand, maximum absorbance went up as methanol concentration increased. On the other hand, the immunoassay showed a tendency to decrease the sensitivity of the immunoassay (lower IC 50 value) with increase of methanol concentration. Judged by the A max /IC 50 ratio, the optimal concentration of methanol in buffer was 5%.

Incubation Temperature
Both temperature and time are important physical parameter for the immunoassay. In general terms, the higher temperature was, the shorter the incubation time needed, and an increase of the incubation time resulted in higher maximum absorbance [24,25]. To investigate the effect of incubation temperature on ELISA performance, the incubation time was set at 30 min, and competitive curves were performed at both 37 °C and 25 °C. As shown in Figure 3, higher sensitivity was observed when the incubation temperature was at 25 °C. Moreover, in the case of incubation at 37 °C, the intra-assay coefficient of variation (CV) increased evidently. Therefore, incubation temperature of 25 °C was selected for further study.  Taking all these factors into account, the optimized conditions for the AVMs immunoassay summarized as follows: The coating antigen was 4′-O-succinoyl-ABM-OVA (diluted to 1:10,000), Mab was 2C11 (diluted to 1:70,000), room temperature (25 °C) for all incubations, overnight incubation (4 °C) for the coating step, an incubation time of 30 min for the competitive step, PB buffer (10 mM, pH 6.5) for Mab dilution and 5% methanol-PB for standard preparation. ELISAs were performed under the above-selected conditions to establish standard curves for five AVMs. As seen from Figure 4, the IC 50 value in assay buffer was 3.05 ng/mL for ABM, 13.10 ng/mL for IVM, 38.96 ng/mL for EPR, 61.00 ng/mL for DOR and 14.38 ng/mL for EMA benzoate. Concerning the reproducibility, average intra-assay CVs of the IC 50 value were 4.87%, 12.01%, 7.93%, 6.48% and 9.12% for ABM, IVM, EPR, DOR and EMA benzoate, respectively. The assay for ABM was the most sensitive compared with the literature [16][17][18], and the sensitivity for IVM was also higher than those reported in Refs. [12][13][14]18,20]. In our previous research [18], it was observed that EPR could be well recognized by the Pab, in which CR was 145.40% (CR for ABM was set as 100%), whereas in this work, the weak recognition of EPR (CR = 7.82%) could be attributed to a more definite specificity of the Mab than the Pab. On the other hand, EMA benzoate could be also recognized by the Mab (CR = 21.21%), which meant that the Mab had broader selectivity than the Pab.

Sample Preparation by Simple Dilution
Analysis of real samples by immunoassay usually requires some sort of sample pretreatment to circumvent the matrix effect. Dilution, which was considered as an effective means [26], significantly decreased matrix interference of samples, though it simultaneously caused reduction of assay sensitivity due to the shift of the dynamic range. In order to investigate the effect of milk matrix on the ELISA performance, different dilution factors were used to prepare ABM standard curves, and total factors were set comparing with the standard curve in buffer. As shown in Figure 5, the assay parameters showed no noticeable difference between ABM standard curves obtained at a dilution of 1:10 (v/v) and in buffer. Therefore, to balance the sensitivity and the dynamic range of the immunoassay, milk samples were diluted to 1:10 (v/v) for detection of AVMs. To evaluate the detection capability of the ELISA, twenty blank samples were selected and replicates of the samples were spiked with ABM at 5 ng/mL, and the blank samples and the spiked samples were determined simultaneously by the developed ELISA. Results showed that none of the responses of the spiked samples overlap with the range of responses of the blank samples. Thus the detection capability (CCβ) of the ELISA was less than 5 ng/mL in milk with simple dilution preparation.
The accuracy and precision of the ELISA were represented by recovery and coefficient of variation (CV), respectively (Table 4). To evaluate the LOD of the ELISA, twenty blank samples were systematically included in the analysis. For IVM and EPR, the values of blank samples were obviously lower than the lowest concentration of the corresponding standard curve. Thus the LOD for IVM and EPR was 7.81 and 15.60 ng/mL, respectively. On the other hand, based on the mean value of twenty blank samples plus three times of the mean standard deviation, the LOD for ABM, DOR and EMA benzoate was 3.97, 30.05 and 11.93 ng/mL, respectively. The mean recoveries at two spiked levels were 83.29%, 101.31%, 88.56%, 75.65% and 101.73% for ABM, IVM, EPR, DOR and EMA benzoate, respectively. The mean intra-assay CVs for ABM, IVM, EPR, DOR and EMA benzoate were 19.28%, 13.39%, 10.17%, 15.71% and 12.97%, respectively. Moreover, the mean inter-assay CV was also satisfactory, which was 14.10% in average and ranged from 6.40% to 20.64%. These results confirmed that the established Mab-based ELISA could detect IVM and EPR at their MRL levels (CAC, 2011) in milk sample with simple dilution preparation.

. Sample Preparation by Solvent Extraction
To improve the detection capability of the ELISA in milk, the dilution factor was decreased by solvent extraction in sample preparation. The assay parameters showed no noticeable difference between ABM standard curves obtained in assay buffer and in extracted milk matrix (dilution factor was 1:4, v/v). Based on the mean value of twenty blank samples plus three times of the mean standard deviation, the LODs for ABM, IVM and EMA benzoate were 1.28, 2.94 and 3.15 ng/mL, respectively. For EPR and DOR, the values of blank samples were obviously lower than the lowest concentration of the corresponding standard curve. Thus the LOD for EPR and DOR was 6.24 and 12.50 ng/mL, respectively. When twenty blank samples were selected and replicates of the samples were spiked with ABM at 2 ng/mL, none of the responses of the spiked samples overlapped with the range of responses of the blank samples. Therefore, when milk was prepared by solvent extraction, the CCβ of the ELISA was less than 2 ng/mL. Additionally, milk samples were also analyzed by the ELISA after being fortified with AVMs. As shown in Table 5, the mean recoveries at two spiked levels were between 69.05% and 130.89%. The mean assay CV, which was 16.63% on average and ranged from 7.08% to 28.40%, was also satisfactory. These results confirmed that the established Mab-based ELISA was a potential screening tool for AVM monitoring in milk samples. hybridomas were subcloned by limiting dilution. Stable antibody-producing clones were expanded and stored in liquid nitrogen. Hybridomas that could produce Mabs with broad-selectivity and high sensitivity were injected intraperitoneally into pristane-primed mice to produce ascites. Then, Mabs were separated and purified by salt precipitation (with caprylic acid-ammonium sulfate) as described by Svendsen et al. [20]. Selectivity of these Mabs was investigated by indirect competitive ELISA using all the three coating antigens. And the cross-reactivity (CR) was calculated by the following equation:

Competitive Indirect ELISA
For competition assays, the concentrations of antibody and coating antigen were optimized by checkerboard titration. After each step, plates were washed four times with PBST. The ELISA was run as described in a previous paper [18]. Briefly, microtiter plates were coated with the optimized concentrations of antigens in CB (100 µL/well) by incubation at 4 °C overnight. Nonspecific binding sites were blocked with the blocking buffer (200 µL/well) at 37 °C for 2 h. Afterward, serial dilutions (50 µL/well) of the analyte were added, followed by adding 50 µL/well of Mab at a previously determined concentration. The mixture solution was allowed to incubate at 37 °C for 30 min and then 100 µL per well of diluted (1/10,000) GαM HRP was added. After another 30 min incubation, 100 µL/well of substrate solution was added. After incubation for 15 min, the reaction was stopped by adding 50 µL of 2 M H 2 SO 4 and absorbance at 450 nm was measured.
Competitive curves were obtained by plotting inhibition (inhibition = B/B 0 × 100%) against the logarithm of analyte concentration. Sigmoid curves were simulated by means of Origin 7.0 software. From the equations, IC 50 values (i.e., analyte concentrations at which the binding of the antibody to the coating conjugate were inhibited by 50%) were determined to assess the assay sensitivity.

ELISA Optimization
Assay buffer. The effect of buffering capacity of assay solution on ELISA performance was studied using PB and PBS to dissolve Mab and ABM standards. Four competitive curves were performed with above Mab and ABM standards to assess the effect of the assay buffer (PB and PBS).
pH effect. PBs were adjusted to different pH values (4.0-9.0) that were prepared by changing the amounts of Na 2 HPO 4 and NaH 2 PO 4 , whereas the concentrations of NaCl and KCl remained at 274 and 54 mM, respectively. These buffers were used to prepare ABM standard solutions and the antibody solutions that were employed for the competitive ELISA to obtain the standard curves.
Methanol concentration. The assays' tolerance to organic solvent was evaluated between 0 and 30% methanol concentration (v/v). In this case, competitive curves were performed from the ABM standards dissolved in PB (pH 6.5) containing different amounts of methanol and Mab in PB (pH 6.5).
Incubation temperature. Competitive curves were performed from the ABM standards dissolved in 5% methanol-PB (pH 6.5) and Mab in PB (pH 6.5). The incubation temperature was set at 37 °C and 25 °C to assess the influence of incubation temperature.

Milk Analysis
Different milk samples were purchased from local markets. All the control samples were previously checked as AVMs free using HPLC method at the National Reference Laboratory for Veterinary Drug Residue (Beijing, China).
Two sample pretreatment procedures were performed. Simple dilution: To investigate the effect of milk matrix on assay performance, milk sample (0.5 mL) was diluted to different volumes (0.5, 1, 2.5, 5 and 10 mL) with 5% methanol-PB (pH 6.5). Then 50 μL portions were used for the selected ELISA test to investigate the matrix interference on the ABM standard curve. For spiked samples, 1 mL of standard solution was spiked to 10 mL milk sample, and diluted to 1:10 (v/v) by 5% methanol-PB (pH 6.5). Solvent extraction: Milk sample (3 mL) was added acetonitrile (9 mL) and hexane (3 mL), and the mixture was shaken for 1 min with a vortex mixer. After centrifugation (4,000 rpm, 10 min), a portion of the acetonitrile layer (1 mL) was transferred and evaporated at 50 °C under a nitrogen stream. The dry residue was dissolved in 5% methanol-PB (pH 6.5, 1 mL). Determination of spiked samples was performed by interpolating their mean absorbance of triplicates in the standard curve run on the same plate.
The assay validation of the ELISA was carried out according to the related content of Commission Decision 2002/657/EC and guidelines for the validation of screening methods for residues of veterinary medicines. The detection capability (CCβ) is the smallest content of the analyte that may be detected, identified and/or quantified in a sample with an error probability of β (<5%).
The limit of detection (LOD) was defined as the lowest amount of analyte in a sample that could be detected but not necessarily quantified exactly, which was based on the mean value of 20 blank samples plus three times of the mean standard deviation. The accuracy and precision of the method were represented by recovery and coefficient of variation (CV), respectively. The precision of the ELISA method was analyzed by repeated analysis of the spiked samples and comparison of the intraand inter-assay CVs. Intra-assay variation was measured by four replicates of each spiked concentrations. And the inter-assay variation was based on the results of four different days.

Conclusions
A broad-specific Mab (2C11) for AVMs was produced. The optimized Mab-based indirect competitive ELISA was sensitive and accurate. Recovery results proved this broad-selective ELISA would be useful for the multi-residue determination of AVMs in milk.