Screening of 49 antibiotic residues in aquatic products using modified QuEChERS sample preparation and UPLC-QToFMS analysis

Instrumental

ACQUITY H-CLASS UPLC and Xevo G2-S Q-ToF mass spectrometer (Waters, Milford, MA, USA) with electrospray ionization source were used. A 3–30K high speed refrigerated centrifuge (SiGMA, Ronkonkoma, NY, USA), MS3 digital vortex mixer (IKA, Königswinter, Germany), laborata 4000 efficient rotary evaporator (Heidolph, Schwabach, Germany), multi Reax oscillator (Heidolph, Schwabach, Germany), N-EVAP™ 112 (Organomation Associates, Berlin, MA, USA) and Milli-Q water purification system (Millipore, Burlington, MA, USA) were used for sample preparation.

LC conditions

The separation of mixed antibiotic standard solutions were achieved on a Waters Acquity UPLC BEH C18 silica column (100 mm × 3.0 mm, 1.7 μm). A gradient LC elution method was employed by 0.1% formic acid aqueous solution as mobile phase A and methanol as mobile phase B.

The gradient elution was as follows: 10% B at 0–3 min, 10–100% B at 3–15 min, 100% B at 15–18 min, 100–10% B at 18–18.1 min and 10% B at 18.1–21 min. The injection volume, flow rate, sample manager and column temperature were set at 10 μL, 0.3 mL/min, 10 °C and 40 °C, respectively. All target antibiotics were eluted, and the column was cleaned and equilibrated.

MS conditions

MS experiments were operated using electrospray ionization (ESI) in the positive mode. The optimum MS parameters were as follows: mass collection range 50–1,000 Da; capillary voltage 3.0 kV; ion source temperature 120 °C; desolvation temperature 450 °C; cone gas flow 50 L/h; desolvation gas flow rate 800 L/h and core voltage 40 V.

QToFMS screening for 49 antibiotic residues was performed using MS^E mode. The simultaneous acquisition of accurate-mass full-spectrum at low and high collision energy are allowed in MS^E mode, where the low collision energy (LE) spectrum provides useful information on the parent molecules and the main fragment ions were obtained commonly in the high collision energy (HE) function. In this study, LE was set as 6 V and HE was set from 10 eV to 40 eV. Leucine enkephalin, a commonly used peptide, was employed here as a reference material to tune MS instruments in every 10 s.

Sample extraction

For the purpose of optimizing extraction of the antibiotic residues for different substrates of aquatic products including grass Carp, Penaeus vannamei and Scylla serrata, ethyl acetate and acetonitrile mixed with different amounts of acetic acid were compared. As shown in Fig.2, 3% acetic acid acetonitrile was used as the extractant, and the average recoveries of 49 antibiotics in three matrices were 75.3%, 76.7%, 81.8%, respectively, which were higher than using 1% acetic acid-acetonitrile (v:v), 5% acetic acid-acetonitrile (v:v), and ethyl acetate for the extraction. Intriguingly the acidity of the extractant has a great effect on the quinolones. The sequence of recoveries of quinolones from low to high was ethyl acetate, acetonitrile, 1% acetic acid acetonitrile, 3% acetic acid acetonitrile, 5% acetic acid acetonitrile when each of them was performed as the extractant. The possible reason is that quinolones, which are amphoteric, are easily soluble in acidic or alkaline such as acetic acid solutions. From these results, 3% acetic acid acetonitrile was chosen as the optimum composition of solvents for the extraction buffer.

Purification procedure

Five most commonly used sorbents were investigated in this experiment, including PSA, C18, ALU-N, PSA-C18 mixture, PSA-ALU-N mixture. The purification effects on grass Carp, Penaeus vannamei and Scylla serrata were shown in Fig. 3. It is obvious that ALU-N gets an inferior purification effect probably because ALU-N has a certain adsorption effect on antibiotics especially quinolones. The highest average recoveries of all 49 antibiotics in three matrices were achieved using PSA-C18, overall.

Afterwards, the amounts of salting-out agents (anhydrous Na₂SO₄ and NaCl) and sorbents (PSA and C18) were optimized using L₉(3⁴) orthogonal experimental design at three levels (Table 2). The results indicated that satisfactory recoveries of 49 antibiotics were observed when 3g Na₂SO₄/1 g NaCl and 50 mg PSA/200 mg C18 were conducted.

After optimization, the average recoveries of 49 antibiotics in grass Carp, Penaeus vannamei and Scylla serrata reached 83.4%, 88.4%, and 88.8% respectively, while this procedure provided the best results for the majority of target antibiotics. In summary, this improved QuEChERS process for antibiotic extraction in aquatic products is fast, effective, economical and eco-friendly.

Identification

As listed in Table 1, each of the 49 target antibiotics was measured in MS^E mode by one precursor ion and at least two product ions. Meanwhile, retention time was also required to provide vital information to identify specific antibiotics.

Linear range, regression equation, limits of detection and limits of quantitation

The series of solvent-based standard solutions were prepared according to “Chemicals and Solutions” and were then determined by UPLC-QToFMS. The calibration curves were obtained from the relationship between the analyte concentration (X, μg/L) and the analyte peak areas/internal standard peak area, providing the linear equation and the correlation coefficient for each analyte. The linear ranges were 1–100 μg/L for each examined analyte with correlation coefficients of greater than 0.9888. The limits of detection (LODs) were evaluated with signal-to-noise ratio (S/N) of 3 and the limits of quantification (LOQs) were evaluated with S/N of 10. LODs and LOQs of solvent-based calibration curves were in the range of 0.01–1.33 μg/L and 0.04–4.42 μg/L, respectively.

Matrix effects

Aquatic products are rich in proteins and unsaturated fatty acids, as well as they contain a variety of vitamins, minerals, trace elements and so on. Complex components cause ubiquitous matrix effects (signal suppression and enhancement) during the LC–MS/MS analysis which may strongly affect the quantitative accuracy and reproducibility in this study (Guo et al., 2016). Here, the matrix effects of three subtracts were evaluated by comparing the calibration curves of the target antibiotics prepared in solvent and in the matrix (Hernando et al., 2007), which is calculated as: $$\text{Matrix effect}(\%)=({\text{Slope}}_{\text{matrix-matched standard curve}}/{\text{Slope}}_{\text{solvent-based standard curve}}-1)\times 100$$

Three sets of blank matrix samples were introduced to the mixed standard solution of different concentrations (1, 5, 10, 25, 50, 100 μg/L). As listed in Table 3, among the three matrices of grass Carp, Penaeus vannamei and Scylla serrata, matrix effects could still encountered in determining several antibiotics such as lincomycin hydrochloride, clindamycin hydrochloride and tylosin. Therefore, matrix-matched standard curves were applied to mitigate matrix effects for quantification of 49 antibiotics. The results of the regression analysis showed that the correlation coefficients (R²) of the matrix-matched standard curves of 49 antibiotics in grass Carp, Penaeus vannamei and Scylla serrata ranged from 0.9900 to 0.9999, 0.9851 to 0.9998, 0.9908 to 0.9997, respectively which indicated excellent linearity.

Based on data obtained from matrix-matched standard curves of 49 antibiotics in grass Carp, Penaeus vannamei and Scylla serrata, the range of the LODs were 0.05–2.40 μg/kg, 0.08–2.00 μg/kg and 0.10–2.27 μg/kg, respectively. And LOQs were in the range of 0.16–8.00 μg/kg, 0.25–6.66 μg/kg and 0.32–7.56 μg/kg, respectively. Hereby, the results of all the LODs and LOQs exhibited in Table 3 in this research were satisfactory as compared with the MRLs.

Recovery and precision

In order to investigate the accuracy and precision of this method, recovery experiments were conducted at different spiking levels of 10, 50, 100 μg/kg (Table 4). Among the 49 antibiotics, except for lincomycin hydrochloride whose recoveries were less than 60%, the recoveries of other antibiotics in three matrices were generally greater than 70%. These results indicated that this method had a satisfactory stability and could meet the actual detecting requirements of 49 antibiotics in aquatic products.