Determining β-lactam antibiotics in aquaculture products by modified QuECHERS combined with ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS)

2.1 Materials and reagents

Certified reference materials of penicillin G (purity of 99.5%), cephalexin (purity of 98.3%), ampicillin (purity of 99.0%), amoxicillin (purity of 99.5%), cephapirin (purity of 99.4%), cefuroxime (purity of 99.1%), cefazolin (purity of 98.5%), ceftiofur (purity of 99.8%), cefquinome (purity of 98.5%) and penicillin G-d7 (purity of 98.0%) were obtained from Dr. Ehrenstorfer (Augsburg, Germany) (Table.A.1). HPLC-grade acetonitrile (ACN) and methanol (MeOH) were purchased from Merck (Darmstadt, Germany). Acetic acid (AA) and formic acid (FA) of LC-MS grade were obtained from Fisher Scientific Inc. (Pittsburgh, PA, USA). The primary secondary amine (PSA, 40–63 µm), Clearnet alumina N (100–300 mesh), graphitized carbon black (GCB, 120–200 mesh), florisil (F), and C₁₈ were purchased from ANPEL Laboratory Technologies Inc. (Shanghai, China). Anhydrous sodium sulfate (Na₂SO₄) was procured from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). The ultrapure experimental water (≥18.2 MΩ) was produced by a Milli-Q purification system (Millipore, Bedford, USA).

The sample muscles of aquaculture products were purchased from the local supermarkets (Xilang, Guangzhou, China). The muscle homogenate was extracted from Penaeus Orientalis, Cyprinus Carpio, Channa Argus, Aristichthys Nobilis, Ctenopharyngodon Idella, and stored at −20 °C until analysis. All samples were freshwater aquaculture products. The blank muscle samples of Ctenopharyngodon Idella (nine β-lactam antibiotics not detected) used in the validation process were collected at the aquaculture base of Pearl River Fisheries Research Institute, Chinese Academy of Fishery Science, where none of these antibiotics were used.

2.2 Preparation of standard solution

Stock standard solutions of nine β-lactam antibiotics and penicillin G-d7 (Internal standard) were prepared at 1.0 mg/mL in a mixture ACN:water (3:7,v/v), and stored at −20℃ in the dark for one month. The mixed intermediate solution was prepared at 10 μg/mL and stored at −20 ℃ in the dark.

2.3 Sample preparation

Approximately 2.0 ± 0.1 g of muscle samples were added to 15 mL centrifuge tubes and spiked 50 µL of the internal standards (1.0 µg/mL), after a vortex-mixing, 5 mL of 1% FA in ACN and 2.0 g Na₂SO₄ to each tube. Later, the tube was homogenized using a vortex oscillator (T25 Ultra-Turrax, IKA, Germany) for 1 min and then centrifuged at 5000 rpm for 5 min. The supernatant was transferred into a 10 mL plastic centrifuge tube containing 200 mg of C₁₈ and 500 mg of Na₂SO₄. The mixture was vortexed for 1 min and centrifuged for 5 min at 5000 rpm, and the supernatant was transferred to another 10 mL glass tube. The extraction solution was dried with nitrogen at 45 ℃. After vaporizing, the residues were dissolved in 1 mL of the aqueous solution of 10% acetonitrile containing 0.1% formic acid and filtered through 0.22 µm nylon syringe filters before the UHPLC-MS/MS analysis.

2.4 The instrument conditions

2.5 Method validation

The QuEChERS and UHPLC-MS/MS techniques were employed to determine nine β-lactam antibiotics in aquaculture products. Validation parameters (Linearity, sensitivity, trueness, precision, and matrix effect) were evaluated according to the guidelines of SANTE/12682/2019, using blank samples (SANTE 2019). The information related to the blank sample is described in 2.1.

The sensitivity of the method was performed by determining the limit of detection (LOD) and limit of quantification (LOQ) of each antibiotic. The LOQ was the lowest concentration at which the method was validated, and the quantifier multiple reaction monitoring (MRM) transition of each compound was evaluated at S/N of ≥ 3; and ≥ 10, respectively.

The samples spiked with β-lactam antibiotics at LOQ, 2 × LOQ and 4 × LOQ (n = 6 at each level) in the muscle of P.orientalis, C.carpio, C.argus, A.nobilis, and C.idella were analyzed by the recovery experiments to validate the trueness and precision of the method. Intermediate precision (inter-day precision) was evaluated by the analysis of six spiked samples in the muscle of C.idella at 5.0 µg/kg in three different days. The analyses were carried out in triplicate. The results were expressed as the percentages of relative standard deviations (%RSD). The RSD within 20%, and the recoveries ranging from 70% to 120% of the nominal concentrations are the primary means to detect drug residues (Chun et al. 2005).

2.6 Matrix effects

Matrix effects (ME) refer to the direct or indirect effect of co-eluting substances in the matrix with response to the target analysis, leading to changes in the ionization efficiency of the analysis and enhancing or weakening the signal response of the target analysis (Chatterjee et al. 2016; Shin et al. 2018). Several studies have reported that high protein and fat contents in aquaculture products can easily cause matrix enhancement or suppression (Kruve et al. 2008). In this study, the matrix effect of nine β-lactam antibiotics in different aquaculture products was analyzed. The matrix effect was calculated by the following equation (Chatterjee et al. 2016; Guedes et al. 2016): $$\mathit{ME}=\frac{S_m-S_s}{S_s}\times 100$$

Where S_m is slope of the matrix-matched calibration curve, and S_s is the slope of the curve obtained by the standard solvent. The matrix effect can be ignored when its absolute value is less than 20%. When the absolute value of ME is between 20% and 50%, ME is considered medium. If the absolute values of ME are greater than 50%, the matrix effect exerts a significant influence (Chatterjee et al. 2016; Guedes et al. 2016).

2.7 Analyses of real samples

In this study, 20 samples were collected from freshwater aquaculture of C.argus in Guangdong Province. A pooled sample was prepared from at least 3 fishes. The homogenate of muscle was extracted and stored at −20℃ until analysis. The pretreatment method and chromatographic conditions established in this study were used to determine the residues of nine β-lactam antibiotics.

2.8 Data statistics

Data were collected by Waters MassLynx V4.1 software, and the qualitative and quantitative analyses of the test results were performed using the MassLynx V4.1 SCN843 data processing software. Statistical analyses were performed using Origin 2020 (Origin Lab, Northampton, MA, U.S.A.) and Excel 2016. Data were presented as mean ± standard deviation (SD).

3.1 Optimization of instrument conditions

The UHPLC-MS/MS technique was developed for determining nine β-lactam residues in MRM mode. Approximately, 100 ng/mL of the mixed standard solution containing nine β-lactam antibiotics were injected into the mass spectrometry system, and the mass spectrum parameters, such as the decluttering voltage and collision energy, were optimized. The detailed optimized MRM parameters are summarized in Table 1. One parent ion, two daughter ions (Table 1), and the highest intense one was utilized for quantification, while the secondary strength ion was monitored for identification (Moreno-González and García-Campaña 2017).

Adding a certain amount of acid in the mobile phase can improve the ionization efficiency, peak shape, and separation effect in the positive ESI mode (Lopes et al. 2012; Moreno-González and García-Campaña 2017). Therefore, the ammonium acetate (5 mmol/L), formic acid, and acetic acid (0.1% in aqueous phase and organic phase) were optimized, while MeOH and ACN were evaluated as organic phases. The results indicated that the signal intensities of ACN as the organic mobile phase were higher than MeOH. The best ionization was observed with the addition of formic acid. Therefore, the percentages of FA were also optimized, and finally, 0.1% FA was employed as an additive in the aqueous and organic phases. Additionally, the gradient elution was optimized to achieve the best separation and peak efficiencies. At the beginning of the gradient, a high proportion of the aqueous phase (90%) was used to accelerate the elution of hydrophilic interfering substances. Additionally, a high percentage of organic phase was set before the end of the gradient to flush the column and resist interference caused by target residues (Table.A1). The ion chromatograms extracted from a standard solution at 50 ng/mL for nine β-lactam antibiotics (quantification transition was shown) are depicted in Fig.A1.

3.2 Optimization of mobile phase in chromatography

The composition of the dissolving agent is associated with the elution effect and response strength of the target compounds. Moreover, these agents play a significant role in obtaining a larger response strength during detection by dissolving the agents as much as possible. Therefore, the composition of the dissolving solution was optimized in this study. The response strengths of the target compounds in 10% acetonitrile–water, 30% acetonitrile–water, 50% acetonitrile–water, 10% ACN-water (containing 0.1% FA), and 10% ACN-water (containing 0.3 %FA) were compared. The peak area of 50 ng/mL β-lactam antibiotics was relatively high in 10% ACN and 10% ACN containing 0.1% FA (Fig. 1), indicating excellent elution ability of β -lactam antibiotics. The ampicillin, ceftiofur, and cefuroxime exhibited the highest response strength in 10% ACN aqueous (containing 0.1% FA). Therefore, 10% ACN solution with 0.1% FA was used as the dissolving solution in this study.

Close

Fig. 1

Peak area of β-lactam antibiotics in different solving liquid（50 ng/mL).

Export to PPT

3.3 Optimization of extraction solvent

The selection of extraction solvent has significant importance in detecting the extraction effect of the target analytes in the matrix. The commonly used extraction agents for antibiotic extraction are methanol, acetonitrile, and acidified acetonitrile (Dasenaki and Thomaidis 2015). Herein, 100 ng of the β-lactam antibiotic standard mixed solution was mixed with a blank sample of grass carp and the recoveries of methanol, acetonitrile, 1% formic acid acetonitrile, and 1% acetate acid acetonitrile were compared as extraction agents. The spiked recovery results of each compound in these extraction agents are shown in Fig. 2. The recoveries of penicillin, cefquinome, and amoxicillin were lower than 80% upon using methanol as an extraction agent. The recoveries of cephalexin and cefquinome in pure acetonitrile were lower than 70%, while the recovery of ampicillin in acetonitrile with acetic acid was lower than 60%).When 1.0% FA-ACN was used as an extraction agent, the spiked recoveries of nine β-LA ranged from 86.9% to 111.7%, meeting the requirements of drug residue analysis.

Close

Fig. 2

The effect of extractants on recoveries of β-lactam antibiotics (n = 3).

Export to PPT

3.4 Selection of a purifying agent

Aquaculture products contain many impurities, such as protein and fat, which could easily induce target based matrix interference. Therefore, purifying the sample and reducing the interference through the impurity peak method are highly necessitated to reduce the target-based matrix co-extract interference. PSA, GCB, C₁₈, neutral alumina and Florisil are the commonly used purifying agents in the QuEChERS method. The PSA is primarily used to remove organic acids, sugars, and fatty acids, while C₁₈ is used for fat and cholesterol removal. GCB can effectively adsorb pigment and neutral alumina can remove fat and pigments. Florian silica is a highly polar and highly active compound, and can adsorb nonpolar and moderately polar compounds (Chatterjee et al. 2016; Ruiz-Medina et al. 2012). The type and dosage of the purifying agents were optimized by the matrix labeling method. 100 ng of β-LA mixed with the standard solution was added into the blank grass carp matrix, extracted with 1.0% formic acid acetonitrile, and purified with 100 mg of five sorbents, respectively. The recovery rates of each adsorbent were compared, and each adsorption step was repeated three times. The results showed that the four adsorbents except for C₁₈ had strong adsorbents for β-LA and the spiked recoveries were less than 70%. When C₁₈ was used as the adsorbent, the spiked recoveries of nine β-lactam antibiotics ranged from 90% to 130% (Fig. 3a). Therefore, C₁₈ was selected as the purifying agent in this method. The dosages of C₁₈ were optimized at the concentrations of 50, 100, 200, 300 and 500 mg, and the recovery rate and purification effect were compared to obtain a better purification effect. When the dosage of the purifying agent was 200 mg, the spiked recoveries of 9 β-lactam antibiotics ranged from 87.44% to 106.22%(Fig. 3b) without any impurity peak interference, meeting the requirements of drug residue detection. Therefore, 200 mg C₁₈ was selected as the adsorbent for QuEChERS in this study.

Close

Fig. 3

Effect of adsorbent type (a) and dosage (b) on spiked recovery rate (n = 3).

Export to PPT

3.5 Method validation

3.5.3

3.5.3 Matrix effect

The matrix effect produced by the aquatic product matrix co-extract could affect the trueness and sensitivity of the method. Therefore, in this study, the matrix standard curve and solvent standard curve were measured under the established pretreatment method and chromatographic determination conditions. The matrix effect was calculated according to the formula in 2.5. The matrix effects of nine β-lactam antibiotics in five aquaculture products (P.orientalis, C.carpio, C.argus, A.nobilis, C.idella) were compared and analyzed and the results are depicted in Fig. 4.The results demonstrated that the substrate effect of β-lactam antibiotics was significantly affected by the substrate type of aquaculture products. The matrix effects of cefazolin, ceftiofur, and cephalexin were within ± 20%, indicating that the matrix effect of these three drugs in the five aquaculture products was within the acceptable range. Generally, the matrix effect of most β-lactam compounds falls between ± 20% and ± 50%. In addition, there is a moderate matrix effect. The matrix effect of cefuroxime in C.argus was less than −50%, showing strong matrix inhibitiory effect (de Sousa et al. 2012; Guedes et al. 2016). Therefore, it is recommended to adopt a matrix calibration curve while determining the actual samples to reduce the matrix effect on the trueness of test results.

Close

Fig. 4

Matrix effects of nine β -lactam compounds in different aquatic product substrates (n = 3).

Export to PPT

3.6 Analyses of actual samples

The established QuEChERS method was applied for the residue analysis of β-lactam antibiotics in 20C.argus samples obtained from the local market, and cefuroxime and cefazolin residues were detected (Table.A3). Cefuroxime was detected in 6 samples with the concentration ranging from 11.51 ± 1.03 µg/kg to 75.15 ± 3.61 µg/kg, and the residual concentration in the 4 samples was greater than the MRL value in the cattle muscle (20 µg/kg); while cefazolin was detected in 8 samples with the concentrations ranging from 15.21 ± 2.54 µg/kg to 38.12 ± 2.58 µg/kg.