Multi-residue analysis of 203 pesticides in strawberries by liquid chromatography tandem mass spectrometry in combination with the QuEChERS method

ABSTRACT Herein, a QuEChERS (quick, easy, cheap, effective, rugged, and safe) procedure together with liquid chromatography and tandem mass spectrometry was applied to the multi-residue analysis of 203 pesticides in strawberries. To reduce the amount of co-eluents, we compared several dispersive solid-phase extraction cleanup procedures and found that cleanup using octadecylsilane provided the best recoveries. Spiked-sample recoveries of 70–120% with relative standard deviations of <20% were obtained for all tested pesticides except for quinmerac. Matrix-matched calibration curves exhibited good linearities (R2 > 0.99), and the limits of quantification (LOQs) were lower than the maximum residue limits recommended by the European Commission and Korea Food and Drug Administration. The developed method was applied to real samples, revealing that their pesticide contents exceeded the corresponding LOQs. The obtained results demonstrate that octadecylsilane-based dispersive solid-phase extraction is a rapid and high-throughput cleanup procedure for the multi-residue analysis of pesticides in strawberries.


Introduction
The high nutritional value of fruits makes them important food components (Ruiz-Torralba, Guerra-Hernández, & García-Villanova, 2018); however, fruits can also be dietary sources of toxic substances such as pesticides. For example, pesticide residues are frequently detected in strawberries, which are very popular agricultural raw materials (Sharma, Nagpal, Pakade, & Katnoria, 2010). The increasing public concern over the potential health risks posed by pesticide residues in food has strengthened the management of agricultural products for improved food quality and safety. To ensure food-supply safety, maximum residue limits (MRLs, expressed in mg kg −1 ) have been established to regulate the use of pesticides (European Commission, 2019;KFDA, 2019). In Korea, MRLs have been set for most pesticides in a variety of animal-origin and agricultural products, and a uniform value of <0.01 mg kg −1 has been set for those for which no specific MRLs exist. To monitor MRL compliance in food commodities, multi-residue analysis methods have been developed and validated (Yang, Luo, Duan, Li, & Liu, 2018). However, the optimization of sample preparation and chromatographic conditions for multi-residue analysis remains challenging due to the large variation of analyte chemical and physical properties.
Various instrumental techniques, including gas chromatography (GC) with flame ionization or electron capture detection, liquid chromatography (LC) with UV, evaporative light scattering, or fluorescence detection, and GC-MS or MS/MS, have been used to sensitively detect multi-residues (Farha et al., 2018;Guo et al., 2018;Lee et al., 2016). Above all, LC coupled with tandem mass spectrometry (LC-MS/MS) and electrospray ionization (ESI) operating in multiple reaction monitoring (MRM) mode has proven to be the most powerful analytical technique for the quantification of pesticides and has been widely used for their multi-residue analysis owing to its high sensitivity and selectivity (Lee et al., 2016). High-resolution mass spectrometry techniques such as quadrupole time-offlight (QTOF) MS are advantageous, allowing quantification and identification to be accomplished in a single run without the need for compound-specific optimization (Muehlwald, Buchner, & Kroh, 2018). Although the main drawback of the MRM technique is blinded non-targeted contaminants in food commodities, LC-MS/MS with MRM is quantitatively more sensitive and selective than full-scan TOF/MS.
Even though LC-MS/MS with MRM provides high sensitivity and selectivity, sample preparation is crucial for the accurate detection of multiple pesticide residues (Hamdy Abdelwahed, Khorshid, El-Marsafy, & Souaya, 2019). The multi-residue analysis of pesticides in real samples is considered to be a difficult task because of the high complexity and diversity of sample matrices. Strawberries contain some troublesome pesticideanalysis co-eluents such as pigments, polyphenols, and sugars (Forbes-Hernandez et al., 2016). Sample preparation using the QuEChERS (quick, easy, cheap, effective, rugged, and safe) method involves simple extraction and cleanup procedures. The original QuEChERS method introduced by Anastassiades, Lehotay, Štajnbaher, and Schenck (2003) has been modified to be more suitable for the analysis of some pesticides, including those with planar structures, different polarities, and pHdependent characteristics. The dispersive solid-phase extraction (d-SPE) technique was developed simultaneously with the QuEChERS method to minimize interference (Song et al., 2019). In d-SPE, a primary secondary amine (PSA) sorbent removes polar interferents (organic acids and sugars), while octadecylsilane (C 18 ) is used to remove non-polar co-eluents such as lipids, and graphitized carbon black (GCB) is used to remove pigmented matrices such as chlorophyll and carotenoids. Various sorbent types have been evaluated to improve cleanup and recovery in the QuEChERS method. Oshita and Jardim (2014) reported that PSA minimizes interference and presents the best cleanup during the determination of some pesticide residues in strawberries. Koesukwiwat, Lehotay, Miao, and Leepipatpiboon (2010) determined 126 pesticides in strawberries employing QuEChERS (acetate buffered version) extraction combined with PSA, C 18 , and GCB cleanup, and achieving satisfactory recoveries. Among the~500 pesticides for which MRLs are set in Korea for animal-origin and agricultural products, more than 150 are LC-amenable. Hence, sensitive detection methods and effective cleanup procedures are required to monitor a wide range of pesticide residues.
Herein, we introduce a method based on LC-MS/MS and QuEChERS with acetate-buffered acetonitrile for the simultaneous determination of 203 LC-amenable pesticides in strawberries, comparing different d-SPE cleanup techniques to improve cleanup efficiency for the tested pesticides. LC-MS/MS combined with the selected QuEChERS method was validated and used to determine the levels of pesticides in real strawberry samples.

Sample preparation
The AOAC Official Method 2007.01 (AOAC, 2007) was followed for extraction. Extraction was performed using an AOAC extraction kit containing 6 g MgSO 4 and 1.5 g sodium acetate. Each sample (15 g) was placed in a 50-mL centrifuge tube to which 1% acetic acid in acetonitrile (1%, 15 mL) was added as the extraction solvent. Triphenylphosphate was spiked directly into the centrifuge tube as the internal standard to a concentration of 1 µg mL −1 . The centrifuge tube was shaken for 1 min, charged with the QuEChERS AOAC extraction kit, shaken QuEChERS extraction was carried out using the AOAC (6 g MgSO 4 , and 1.5 g sodium acetate) extraction kit and cleanup was performed using cleanup 1 (50 mg PSA), cleanup 2 (50 mg PSA +50 mg C 18 ), cleanup 3 (50 mg PSA + 50 mg GCB), cleanup 4 (50 mg PSA +50 mg C 18 + 50 mg GCB), or cleanup 5 (50 mg C 18 ).
vigorously for 10 min, and centrifuged at 4,000 g for 10 min at 4°C. Various d-SPE cleanup methods were then applied to the extracted analytes. Five d-SPE cleanup methods were used: Cleanup 1 (50 mg PSA and 150 mg MgSO 4 ), Cleanup 2 (50 mg C 18 , 50 mg PSA, and 150 mg MgSO 4 ), Cleanup 3 (50 mg GCB, 50 mg PSA, and 150 mg MgSO 4 ), Cleanup 4 (50 mg C 18 , 50 mg GCB, 50 mg PSA, and 150 mg MgSO 4 ), and Cleanup 5 (50 mg C 18 and 150 mg MgSO 4 ). Each tube was strongly vortexed for 1 min and then centrifuged at 12,000 rpm for 5 min at 4°C. The extract was filtered through a syringe with a 0.22-μm nylon membrane filter and transferred into a vial for analysis by LC-MS/MS.

LC-MS/MS
An Agilent LC 1200 HPLC system (Agilent Technologies, Santa Clara, CA, USA) coupled to a 4000 QTRAP mass spectrometer equipped with a turbo ion-spray ionization source (AB SCIEX, Foster City, CA, USA) was used to analyze pesticides. An Agilent Zorbax Eclipse Plus C 18 column (2.1 × 100 mm, 1.8 μm; Santa Clara, CA, USA) was used for chromatographic separation. The above separation was carried out with a binary mobile phase containing (A) 5 mM ammonium acetate and 0.1 vol% formic acid in water, and (B) 5 mM ammonium acetate and 0.1 vol% formic acid in methanol. A linear binary mobile phase gradient was used: 95% A at 0 min, 60% A at 1.5 min, 40% A at 2 min, 30% A at 3.5 min, 20% A at 15 min, 0% A from 16 to 22.5 min, and 95% A from 23 to 27 min. The flow rate, column temperature, and injection volume equaled 0.2 mL min −1 , 40°C, and 2 μL, respectively. The pesticides were ionized in positive ESI mode and determined using scheduled MRM with the following general settings: curtain gas pressure = 30 psi, ion-source gas (1) pressure = 50 psi, ion-source gas (2) pressure = 55 psi, source temperature = 400°C, ion spray voltage = 5500 V. The MRM transitions, retention times, collision energies, and declustering potentials of analytes are summarized in the Appendix.

Validation study and matrix effects
The acceptability of the developed method for the analysis of target pesticides was validated following the European Commission SANTE/11813/2017 (European Commission, 2017) and ICH/2005/Q2/R1 (ICH, 2005) protocols. Linearities were determined using matrix-matched calibration curves with spiked blank samples at five concentrations (0.005, 0.01, 0.02, 0.05, and 0.1 mg kg −1 ). All coefficients of determination (R 2 > 0.99) were acceptable.
Recoveries (%) and precisions, in terms of repeatability and reproducibility, were determined by analysis of blank samples spiked with standard solutions at three concentrations (0.01, 0.05, and 0.1 mg kg −1 ), with recoveries calculated using Equation (1). Repeatabilities and reproducibilities were determined based on the results of at least six replicate analyses performed on the same day and on different days, respectively. Precision was expressed as the relative standard deviation (RSD%) of replicate analyses.
Limits of detection (LODs) and limits of quantification (LOQs) were determined using five independently spiked pesticide concentrations (0.005, 0.01, 0.02, 0.05, and 0.1 mg kg −1 ) and were calculated based on the standard deviations of response and slope as LOD = 3.3 σ/s, LOQ = 10 σ/s, where σ is the standard deviation of the response, and s is the slope of the matrix-matched calibration curve.
To compensate for matrix effects (%MEs), the slopes of pesticide standards in solvent and in extracts were compared, with Equation (2) used to calculate %ME: %ME ¼ slope of calibration curve in the extract slope of calibration curve in solvent À 1 Â 100 (2)

Optimization of LC-MS/MS conditions
Optimization of mobile phase composition is important for the analysis of trace amounts of analytes by LC-MS, as it leads to satisfactory sensitivity and good ionization. Herein, ammonium acetate and formic acid were used as mobile phase additives to improve ionization efficiency, peak intensities, and target analyte separation (Berlioz-Barbier et al., 2014). Ammonium ions suppress the formation of sodium adducts under acidic conditions, increasing sensitivities and response consistency for some pesticides by facilitating the formation of [M + H] + and [M+ NH 4 ] + ions (Hiemstra & de Kok, 2007). When water (containing 0.1 vol% formic acid and 5 mM ammonium acetate) and methanol (containing 0.1 vol% formic acid) were used as the mobile phases, analyte peaks eluted separately and featured good shapes ( Figure 1). A slightly longer run time (a total of 27 min) was needed when the Eclipse Plus C 18 RRHD (2.1 × 100 mm, 1.8 µm) column was used; however, this column provided satisfactory separation and resolution (Figure 1), and all analytes were eluted between 5 and 21 min (Table A1). The MS operating conditions were optimized for maximum sensitivity to identify and quantify the target analytes. As shown in Table A1, the MS/MS parameters, including declustering potential and collision energy, were optimized through the direct infusion of each analyte with a syringe pump at a flow rate of 10 µL min −1 . Two transitions were chosen for each pesticide to avoid false negatives in the MRM mode, and the most intense product ion was chosen as the quantifier ion.

Cleanup procedure
When Cleanup 3 (PSA and GCB) and Cleanup 4 (PSA, GCB, and C 18 ) were used, 84 and 103 out of 203 pesticides, respectively, showed recoveries lower than those observed for Cleanup 1 (PSA alone) ( Figure 2). Although GCB is commonly applied to remove compounds with planar structures, such as chlorophyll and carotenoids, it absorbs and retains numerous planar pesticides Nguyen et al., 2008). The low recoveries (>70%) observed for planar pesticides such as pyrimethanil, terbufos, and thiabendazole (Table 1)  Eleven pesticides showed recoveries of less than 70% when fortified at 0.1 mg kg −1 using d-SPE with Cleanups 1-4 (Table 1). Nine sulfonylurea herbicides and one quinolinecarboxylic acid herbicide showed low recoveries in the 3.0-69.2% range with Cleanups 1-4 using GCB and/or PSA. The original QuEChERS method, employing PSA as the sorbent, showed weak recoveries (<70%) for sulfonylurea herbicides (Kaczyński & Łozowicka, 2017), which was ascribed to the known ability of PSA and GCB to absorb some weakly acidic herbicides such as sulfonylureas (Kaczyński & Łozowicka, 2017;Liu et al., 2015). A modified QuEChERS method employing C 18 cleanup was introduced by Lee et al. (2016) to overcome this issue; consistent with their results, the best recoveries for sulfonylurea herbicides in strawberries were observed for Cleanup 5 with C 18 (Table 1). This outcome was rationalized by the fact that the C 18 sorbent has a lower affinity for acidic herbicides than PSA or GCB and features excellent performance during extract purification (Liu et al., 2015). The combination of C 18 and PSA (Cleanup 2) resulted in slightly higher recoveries of sulfonylurea herbicides, which, however, remained below 70%. Carbendazim exhibited poor recovery when PSA (Cleanup 1) or a combination of PSA and GCB (Cleanups 3 and 4) was used. Guan, Tang, Chen, Xu, and Li (2013) ascribed the low recovery of carbendazim to the high amount of absorbent (i.e. GCB) strongly absorbing non-polar compounds. As a result, with the exception of quinmerac, 10 pesticides with low recoveries in Table 1 showed better recoveries (74.6-95.5%) when Cleanup 5 (with C 18 alone) was used. Consequently, cleanup with C 18 was selected for the multi-residue analysis of pesticides in strawberries because of its high extraction efficiency. Quinmerac, which is a pH-dependent acidic compound (pKa = 4.31 at 25°C), was out of the recovery criterion range in the cases of all five cleanup methods.

Method validation and matrix effects
To evaluate the extraction and d-SPE cleanup procedure based on the use of C 18 , we validated our method in terms of linearity, accuracy, precision, LOD, and LOQ following the European Commission SANTE/11813/2017 guidelines (European Commission, 2017) and the ICH/2005/Q2/R1 (ICH, 2005) protocol.
Calibration curves for 203 pesticides were constructed by the matrix-matching method at concentrations of 0.005, 0.01, 0.02, 0.05, and 0.1 mg kg −1 in blank sample extracts. As shown in Table 2, all matrix-matched calibration curves were linear, with coefficients of determination (R 2 ) exceeding 0.995 for all tested pesticides. According to the SANTE/   Figura 2. Recuperaciones de LC-MS/MS para diversos procedimientos de limpieza a un nivel de pico de 0.1 mg kg −1 . Limpieza 1 (50 mg de PSA y 150 mg de MgSO 4 ), limpieza 2 (50 mg de C 18 , 50 mg de PSA y 150 mg de MgSO 4 ), limpieza 3 (50 mg de GCB, 50 mg de PSA y 150 mg de MgSO 4 ), limpieza 4 (50 mg de C 18 , 50 mg de GCB, 50 mg de PSA y 150 mg de MgSO 4 ), y limpieza 5 (50 mg de C 18 y 150 mg de MgSO 4 ).  We found that 202 pesticides exhibited recoveries of 71-100% and 75-102% at fortification levels of 0.01 and 0.1 mg kg −1 , respectively. Halosulfuron-methyl showed a recovery of 69.2% at the lowest fortification level; however, acceptable recoveries were observed at 0.05 and 0.1 mg kg −1 . Although the recoveries of quinmerac were less than 70% at all fortification levels, repeatability and reproducibility were in the acceptable precision criterion ranges; hence, screening for this pesticide was not problematic. Although the lowest fortification level of carbofuran exceeded the EU-MRL level, the validated method provided satisfactory results at a lower level of the Korea-MRL. Precisions were acceptable in all cases, with %RSDs lower than 20%, and the repeatability and reproducibility RSDs were less than 14 and 16% for all pesticides, respectively. LODs ranged from 0.001 to 0.005 mg kg −1 . LOQs of QuEChERS extraction was carried out using the AOAC (6 g MgSO 4 , and 1.5 g sodium acetate) extraction kit and cleanup was performed using cleanup 5 (50 mg C 18 ). * LMR uniforme para plaguicidas sin LMR en Corea. NA: LMR no disponible actualmente para la fresa analizada. La extracción de QuEChERS se realizó con el kit de extracción AOAC (6 g de MgSO 4 y 1.5 g de acetato de sodio) y la limpieza se realizó empleando limpieza 5 (50 mg C 18 ).
0.002-0.01 mg kg −1 obtained for the target analytes were lower than the lowest points of each linear range. All pesticides exhibited LOQs lower than the lowest MRLs based on EU and Korean requirements. Matrix effects (MEs) due to biological samples may result in quantification errors caused by the enhancement or suppression of ionization during LC-MS/MS (Guo et al., 2018). Figure 3 presents the retention times and MEs of each pesticide, revealing that the LC-MS/MS responses of 69 tested pesticides exhibited no ME (between −20 and 20%) in the 6-20 min retention time range. More than 70% of pesticides showed medium-level suppression effects (between −20 and −50%); however, several pesticides showed strongly suppressed signals, with MEs of less than −50%. Taylor (2005) reported that LC-MS/MS signals are mainly suppressed by matrix co-eluents. To overcome this issue, alternative approaches such as the use of matrix-matched or standard addition calibration curves have been applied for accurate quantification (Hanafi, Dasenaki, Bletsou, & Thomaidis, 2018). We also used the matrix-matching method for pesticide quantitation. With the exception of quinmerac, the MEs of most pesticides did not affect their quantitation; and adequate recoveries and acceptable precisions were observed at all spiking levels (Table 2).

Application of the developed method to real-life sample analysis
The validated method was used to analyze pesticide residues in real strawberries. Ten commercial samples were collected from local markets in Jeonju and Wanju, Korea. Two fungicides (metalaxyl and tricyclazole) and two insecticides (carbofuran and dinotefuran) were detected in the real strawberry samples (Table 3). Metalaxyl is used in broadcast soil applications and is frequently detected in fruits and crops. Pous, Ruíz, Picó, and Font (2001) reported that LC-MS is a powerful method for determining metalaxyl in strawberries at concentrations below the MRL. Tricyclazole, which is classified by the WHO as a moderately hazardous pesticide (WHO, 2005), was also concurrently found in these samples, albeit at concentrations below the EU and Korean MRLs of 0.01 mg kg −1 . Dinotefuran accounts for more than 25% of globally used pesticides (van der Sluijs et al., 2013), and numerous studies on this pesticide in crops and vegetables have been reported (Morrissey et al., 2015). Some countries have banned the use of carbofuran on crops because of its high toxicity, suggesting less toxic carbosulfan as a substitute (Song et al., 2018). As shown in Table 3, all detected pesticides were present at levels below the MRLs recommended by Korea (KFDA, 2019) and the EU (European Commission, 2019); hence, no sample was found to be non-compliant. In conclusion, LC-MS/MS combined with the QuEChERS method and C 18 purification was concluded to be suitable for pesticide detection and multi-residue analysis in strawberries.

Conclusion
Herein, we demonstrated that LC-MS/MS in combination with QuEChERS d-SPE/C 18 extraction can be used to analyze multiple pesticide residues. The validated method provided satisfactory results (linearities, recoveries, and precisions), enabling the qualitative and quantitative analysis of 203 pesticides in strawberries. The introduced approach also provided adequate LOQs compliant with EU and Korean MRLs and was applied to monitor pesticide residues in commercial strawberries. Metalaxyl, tricyclazole, carbofuran, and dinotefuran were found in the tested samples at levels less than the corresponding LOQs. No MRLs were exceeded in all samples. Thus, AOAC extraction using acetate buffer and d-SPE using C 18 in combination with LC-MS /MS was concluded to be a rapid, sensitive, and reliable method for the multi-residue analysis of pesticides in strawberries. However, further studies on pesticide residue monitoring in strawberries are required to assess the involved risks and determine cumulative exposure to pesticides through dietary intake.

Disclosure statement
No potential conflict of interest is reported by the authors.

Funding
This research was supported by the Main Research Program (E0187200-02) of the Korea Food Research Institute funded by the Ministry of Science and ICT.