Highly-Selective Analytical Strategy for 90 Pesticides and Metabolites Residues in Fish and Shrimp Samples

The analysis of pesticide residues in aquatic products is challenging due to low residue levels and the complex matrix interference. In this study, we developed a simple, fast method for the trace analysis of 90 pesticides and metabolites in aquatic products. The analytes covered a wide polarity range with log Kow (log octanol-water partition coefficient) ranging from −1.2 to 6.37. Grass carp (Ctenopharyngodon idellus) and prawn (Penaeus chinensis) samples were chosen to validate the quantification method. The samples were extracted by 0.2% formic-acetonitrile, cleaned by solid-phase extraction (PRiME HLB), and analyzed by high performance liquid chromatography−tandem mass spectrometry. The results showed good linearities for the analytes and were observed in the range of 0.05–50 μg/L. The recoveries of the method were within 50.4–118.6%, with the relative standard deviations being lower than 20%. The limits of quantifications (LOQs) of the method were in the range of 0.05–5.0 μg/kg, which were superior to values compared with other research. The developed method was applied to detect pesticide residues in prawn samples from eastern coastal areas of China. Three herbicide residues of diuron, prometryn, and atrazine were detected in prawn samples. The method was sensitive and efficient, which is of significance in expanding the screening scope and improving the quantitative analysis efficiency in aquatic products.


Introduction
Pesticides play a key role in enhancing crop production and controlling pests. However, the extensive use of pesticides in modern agriculture has resulted in agricultural nonpoint source pollution causing risk hazards for water [1,2]. Pesticides can enter the aquatic environment through various ways, such as drift, irrigation, wash-off, drainage, and container and equipment cleaning practices [3]. These pesticides in the aquatic ecosystems may cause adverse effects on water pollution and the diversity of aquatic biota. More importantly, these pesticides in water could bioaccumulate in non-target species and even amplify at higher concentrations along the food chain; ultimately, they could cause risks to humans [1,4].
In aquatic products, fish and shrimp, the main providers of high-quality protein, polyunsaturated fatty acids, and vitamins and minerals, are an important part of a healthy diet [5,6]. Nonetheless, they can also be exposed to pesticides. For example, fish species are at a higher nutritional level in the aquatic food chain and easily bioaccumulate pesticides [3], and shrimp are considered benthic organisms that easily accumulate pesticides from sediments [7][8][9]. Therefore, it is very vital to develop analytical methods to detect pesticide residues in aquatic products for health reasons or ecological reasons [10].
In the past, and presently, the top concerns were relevant to the analytical methods of persistent organic pollutants because of their specific semi-volatile, persistent, and bioaccumulative characteristics [11][12][13][14][15][16][17][18]. As for the currently used pesticides, a comparison of

Optimization of the LC-MS/MS
For the analysis of 90 pesticides and metabolites, a standard solution of each analyte (10 µg/L) was prepared and directly injected into the autosampler to obtain the optimal mass spectrum conditions. The MS/MS conditions of each analyte were optimized through the following pattern: MS2 SCAN mode, SIM mode, Product ion mode, and MRM mode. Among the analytes, 83 analytes were detected using the ESI positive ion mode and 7 analytes were detected using the ESI negative ion mode. The multiple reaction monitoring (MRM) transitions, retention time, fragmentor, and collision energy of each analyte are summarized in Table S1. For the mobile phase, water was chosen as the aqueous phase because of the presence of the ESI negative ion mode. The MRM chromatograms of 90 pesticides and metabolites were shown in Figure 1.  Table S1. For the mobile phase, water was chosen as the aqueous phase because of the presence of the ESI negative ion mode. The MRM chromatograms of 90 pesticides and metabolites were shown in Figure 1. The dynamic multiple reaction monitoring (dMRM) data acquisition mode can yield a higher response, symmetrical peak shape because the mass spectrometer scans only certain channels within a specified time window so that more scans can be obtained [33,34]. In this study, the MRM mode and dMRM mode were compared. As shown in Figure S1, the dMRM mode indicated a 3-to 10-fold higher response than the MRM mode. Therefore, the dMRM mode was used to achieve a high sensitivity and superior peak shape.

Optimization of Sample Preprocessing
The choice of extraction solvents was significant in obtaining a satisfactory recovery. In the residue analysis, acetonitrile (ACN), ethyl acetate, and acetone were the common extraction solvents. However, ACN was chosen as the preferred solvent because it extracted fewer coextracted matrix components [22]. Meanwhile, it was also the major com- The dynamic multiple reaction monitoring (dMRM) data acquisition mode can yield a higher response, symmetrical peak shape because the mass spectrometer scans only certain channels within a specified time window so that more scans can be obtained [33,34]. In this study, the MRM mode and dMRM mode were compared. As shown in Figure S1, the dMRM mode indicated a 3-to 10-fold higher response than the MRM mode. Therefore, the dMRM mode was used to achieve a high sensitivity and superior peak shape.

Extracting Procedure
The choice of extraction solvents was significant in obtaining a satisfactory recovery. In the residue analysis, acetonitrile (ACN), ethyl acetate, and acetone were the common extraction solvents. However, ACN was chosen as the preferred solvent because it extracted fewer coextracted matrix components [22]. Meanwhile, it was also the major component of the mobile phase and could be directly injected into the column without evaporation and reconstitution [35]. In this study, the extraction efficiencies were examined using pure ACN, 0.2% formic acid/ACN (FA/ACN), ACN/water (8:2), and ACN/water (8:2) with 0.2% FA. ACN/water (8:2) was selected because it had been used to analyze pesticide residues in fishery products [10]. The results of different extraction solvents are shown in Figure 2a. The recoveries of 90 pesticides and metabolites using ACN/water (8:2) and ACN/water (8:2) with 0.2% FA were not satisfactory, and the recoveries of almost 26% and 27% of analytes were less than 60%. The recoveries of two analytes (methamidophos and hydroxyatrazine) using ACN were less than 60%. The addition of formic acid in ACN improved the recovery of methamidophos (74%) and hydroxyatrazine (64%). The RSDs of ACN and 0.2% FA/ACN as extraction solvents are compared in Figure 2b. Furthermore, smaller RSDs were observed when 0.2% FA/ACN was used as the extracting solvent. Thus, 0.2% FA/ACN was selected as the optimal extraction solvent for the target analytes.

Clean-Up Conditions
Aquatic samples were rich in fat components, and they had an adverse impact on quantitative results. In this study, different types of SPE were tested for recovery. In the initial experiment, SPE cartridges (the Oasis HLB cartridge and Cleanert PEP cartridge), which belonged to hydrophilic-lipophilic balanced copolymer, were selected for the purification of the 90 analytes [4]. It was previously reported that the efficiency of SPE was determined by the variety of sorbents, type of the loading solvents and elution solvents, etc. [29,36].
Therefore, different varieties of combinations (including the variety of sorbents, type of the loading solvents and the elution solvents) were designed to optimize several relevant conditions affecting the clean-up efficiency. As shown in Figure 3a, with the PEP SPE clean-up (S1, S2, S3, and S4 treatment), 30%, 51%, 52%, and 70% of the analytes were in the recovery of 70-120%, respectively. With the HLB SPE cleanup (S5, S6, S7, and S8 treatment), 57%, 49%, 64%, and 47% of the analytes were in the recovery of 70-120%, respectively. The recoveries of the target compounds that could not meet the recovery requirements under different types of clean-up conditions were displayed in Figure 3b. The results may be due to the fact that the target analytes have broad physical and chemical properties (log Kow: −1.2-6.37), therefore, very hydrophobic analytes were too strongly retained to be eluted with ACN or methanol (MEA) (Figure 3b). In the case of PRiME HLB, the recoveries of 85% of the analytes were in the range of 70-120%, and the recoveries of 14% of the analytes were 60-70%. Thus, we preliminarily adopted PRiME HLB SPE for purification based on the consideration of recoveries.

Validation of the Proposed Method
The validation results of the analytical method in terms of linearity, sensitivity, precision, and accuracy are presented in Tables 2 and S2. Good linearities (R 2 > 0.9902) were observed in solvent-and matrix-matched calibration curves. The LOQs of all the analytes were 0.05-5.0 µg/kg. Our results showed that the selectivity and linearity were favorable for the residue analysis at all target concentrations of the pesticides. In this next experiment, different volumes of eluent were optimized including 1 mL, 2 mL, 3 mL, and 4 mL. The results showed that 85% of the analytes were in the recovery of 70-120% with only 3 mL eluent volumes (Figure 3c). Therefore, 3.0 mL was chosen as the elution volume for PRiME HLB in order to save organic solvents.
Conventional SPE procedures consist of four steps: pre-equilibrating, loading, rinsing, and eluting steps for purification of the sample. The passthrough PRiME HLB column consists of two steps: loading the extraction solution onto the cartridges, and then collecting the eluates. Therefore, this method avoids the tedious steps and reduces solvent consumption and saving in comparison with traditional SPE [22].

Validation of the Proposed Method
The validation results of the analytical method in terms of linearity, sensitivity, precision, and accuracy are presented in Tables 2 and S2. Good linearities (R 2 > 0.9902) were observed in solvent-and matrix-matched calibration curves. The LOQs of all the analytes were 0.05-5.0 µg/kg. Our results showed that the selectivity and linearity were favorable for the residue analysis at all target concentrations of the pesticides.
All analytes were pre-spiked into samples at the concentration levels of 0.05, 0.5, 5.0, and 50.0 µg/kg (n = 5). The average recovery values ranged from 50.4% to 117.4%, with the intra-RSDs being less than 19.8% in grass carp samples. The average recovery values ranged from 60.5% to 118.6%, with intra-RSDs being less than 19.5% in prawn samples. The inter-RSD of the method ranged from 0.5% to 19.1% in grass carp, and 0.3% to 18.5% in prawn, respectively. Our results showed that the recoveries and RSDs can meet the requirements of residue analysis. The matrix effects were assessed by comparing the slope of the matrix standard curve to the solvent standard calibration curve. The ME ranged from 9% to 115% in grass carp, and ranged from 10% to 166% in prawn. The reason for the difference in matrix effect may be related to the physicochemical properties of analytes and the texted matrix, which was observed in other studies [37][38][39]. To eliminate the effect of matrix effects, the matrix-matched calibration curve was selected for the quantitative analysis. Therefore, the developed method can meet the requirements of pesticide residues in grass carp and prawn samples.

Real Samples Analysis and Health Risk Assessment
Of the pesticides studied, three pesticides were detected at values greater than the LOQ. The concentration of dinuron from Tianjin was detected among all samples, with 0.92 µg/kg maximum concentration. Prometryn were detected in all samples from Dalian, and the concentration ranged from 0.05 to 0.06 µg/kg. The concentration of atrazine ranged from LOQ to 0.05 µg/kg in Tianjin, from LOQ to 0.58 µg/kg in Dalian, and from LOQ to 0.28 µg/kg in Qingdao. In the study, dinuron (log Know = 2.87, BCF = 9.7), prometryn (log Know = 3.34, BCF = 85), and atrazine (log Know = 2.7, BCF = 4.3) were detected in samples, probably because of its high log Kow and BCF values. Similarly, dinuron were detected in rice-crayfish systems in Jiangsu Province, with the highest value being 11.5 µg/kg [40]. Prometryn also were detected in the shellfish samples and the concentration ranged from 1.0 µg/kg to 33.6 µg/kg in coastal areas in China [26]. The concentrations of atrazine in aquatic organisms in the Xiangshan Harbor ranged from 2.37 to 39.2 µg/kg (dw), with an average concentration of 13.2 ± 11.8 µg/kg [41].
The risk assessment of detected pesticides in prawn samples was shown in Table S3. The acute and chronic risk quotients were calculated, ranging from 0.0029% to 0.0288%, and 0.0002% to 0.0289%, respectively. This indicated that there was not a significant risk to human health.

Sample Preparation
Grass carp and prawn samples were selected to validate the method. In addition, prawn samples collected from eastern coastal areas of China were used for the real sample analysis. The areas were mainly distributed in Dalian (Liaoning Province), Tianjin, Yantai (Shandong Province), and Qingdao (Shandong Province). A total of 32 prawn samples comprising 8 of each area were used for this work. The prawn meat was homogenized, and stored at −20 • C for the next analysis.

Extraction and Cleanup
An amount of 2.0 g of each sample was weighed into a 10 mL centrifuge tube and extracted with 4 mL of 0.2% FA/ACN solution. The samples were vortexed on the multifunctional vortex mixer at 2500 rpm for 10 min. Subsequently, 0.4 g of MgSO 4 and 0.1 g of NaCl were added, after which the samples were vortexed for 5 min and centrifuged at 8000 rpm for 5 min. An amount of 2 mL of supernatant was prepared for purification.
Method 1: The supernatant (2.0 mL) was evaporated to dryness at 30 • C and was redissolved in 2 mL loading solution for subsequent clean-up. Before sample loading, Oasis HLB and Cleanert PEP cartridges were preequilibrated with 3 mL of methanol and 3 mL of deionized water. Then, the redissolved extracts were loaded into a SPE cartridge. The cartridge was eluted with different solvents and then the eluent was evaporated to dryness by nitrogen at 30 • C. The extract was adjusted to 1 mL with ACN. After filtration with 0.22 µm nylon syringe filters, the final extracts were transferred into an autosampler vial at −20 • C until analysis.
Method 2: The supernatant (2.0 mL) was passed through a PRiME HLB column and there was no activation requirement for PRiME HLB cartridges. Then, the column was eluted by 3.0 mL of ACN. All of the solutions were collected and evaporated to dryness by nitrogen at 30 • C. The extract was adjusted to 1 mL with ACN. After filtration with 0.22 µm nylon syringe filters, the final extracts were transferred into an autosampler vial at −20 • C until analysis.

LC-MS/MS Analysis
The analysis of 90 pesticides and metabolites was performed on an Agilent 1290 UPLC system coupled to a G6470A triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Separation was achieved with a Poroshell PFP column (2.1 × 100 mm, 2.7 µm particle size, Agilent) under 30 • C column temperature. The mobile phase was made up of water and ACN. The gradient of mobile phase is shown in Table S4. The injection volume was 5.0 µL, and the flow in the analysis procedure was 0.3 mL/min. MS/MS acquisition was carried out by dMRM mode for 90 pesticides and metabolites and the cycle time was 500 ms. The instrument allows for the simultaneous detection of positive and negative ionization modes without affecting method sensitivity, and 9 pesticides and metabolites were analyzed in a single needle injection and the whole analysis time was 17 min (83 analytes in the positive mode and 7 analytes in the negative mode). Other MS detection conditions were set as follows: sheath gas temperature, 350 • C; sheath gas flow, 11 L/min; drying gas temperature, 350 • C; drying gas flow, 10 L/min; nebulizer, 45 psi; and nozzle voltage, 500 V.

Method Validation
The optimized method was verified to evaluate the applicability of the method according to SANTE guidelines [23]. These parameters included linearity, recovery, RSD, LOQ, and ME. The solvent-and matrix-matched calibration curves were set at concentration levels of 0.05, 0.1, 0.5, 1, 2, 5, 10, 20, and 50 µg/L. Good linearity was observed when the correction coefficient (R 2 ) of the analyte was greater than 0.99. Four different spiking concentrations on a single day (intra-day) and over three consecutive days (inter-day) were performed to verify the accuracy and precision of the method. The LOQ was the lowest spiked level of the validation that meets the method criteria for recovery and precision. ME was confirmed by comparing the slope of the matrix standard curve to the solvent standard calibration curve.

Risk Assessment
The acute and chronic risk assessment of consuming these prawn samples was investigated. The results were calculated according to the formula [42]: IESTI = (HR × LP)/bw %HQa = IESTI/ARfD × 100 where, IESTI is the estimate of short term intake (mg/kg bw), HR indicates the highest concentration (mg/kg), LP means large portion (g/d), and the large portion of prawn is 300 g/d for the general population [43]. bw is the average weight for the general population (60 kg) [42], %HQa is the acute risk quotient, and ARfD is the acute reference dose (mg/kg bw). EDI = (STMP × F)/bw %HQ C = EDI/ADI × 100 Here, EDI is the estimated daily intake (mg/kg bw), STMR indicates the median concentration (mg/kg), F means the average consumption of prawn (g/d), and the average consumption of prawn is 100 g/d for the general population [43]. bw is the average weight for the general population (60 kg) [42], %HQc is the chronic risk quotient, and ADI is the acceptable daily intake (mg/kg bw).

Conclusions
In this study, a strategy based on the purification of the SPE method combined with HPLC-MS/MS for the analysis of 90 pesticides and metabolites in fish and shrimp was proposed. The developed method, with good linearity (>0.9902), specificity, trueness (50.4-118.6%), precision (<19.8%), and sensitivity (0.05-5.0 µg/kg), could meet the requirements for the simultaneous analysis of 90 pesticides and metabolites. The method had the advantages of simple operation and high sensitivity. The results of this study may be helpful for understanding the residue levels in aquatic biota, and the residues in aquatic biota would help us perform risk assessments.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/molecules28104235/s1, Figure S1: Total ion response through MRM mode and dynamic MRM mode (red: MRM; blue, dMRM mode); Table S1: Experimental parameters and HPLC-MS/MS conditions of 90 pesticides and metabolites in ESI; Table S2: Inter RSDs and matrix effect of 90 pesticides and metabolites in grass carp and prawn samples; Table S3: Acute and chronic risk assessment of detected pesticides in prawn samples; Table S4