Trace level determination of pyrethroid and neonicotinoid insecticides in beebread using acetonitrile-based extraction followed by analysis with ultra-high-performance liquid chromatography–tandem mass spectrometry

Journal of Chromatography A, 1316 (2013) 53–61

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Trace level determination of pyrethroid and neonicotinoid insecticides in beebread using acetonitrile-based extraction followed by analysis with ultra-high-performance liquid chromatography–tandem mass spectrometry Barbara Giroud, Antoine Vauchez, Emmanuelle Vulliet ∗ , Laure Wiest, Audrey Buleté Université de Lyon, Institut des Sciences Analytiques, UMR5280 CNRS Université Lyon 1, ENS-Lyon, 5 rue de la Doua, 69100 Villeurbanne, France

a r t i c l e

i n f o

Article history: Received 26 August 2013 Received in revised form 25 September 2013 Accepted 25 September 2013 Available online 4 October 2013 Keywords: Modified QuEChERS Traces Pyrethroids Neonicotinoids Beebread Sample preparation

a b s t r a c t Beebread is among the matrices suspected of contaminating honeybee. To better understand this contamination, the study aimed to develop an efficient, sensitive and reliable analytical method for the trace analysis of pesticides in beebread. This study focuses specifically on the insecticides pyrethroids and neonicotinoids and some of their metabolites. It describes the development and validation of an original analytical approach that consists of one simple extraction method based on modified QuEChERS followed by a selective and sensitive analysis by UHPLC–MS/MS to determine the target compounds in beebread. The method was validated using a quadratic fit. RSD values below 20% were obtained, except for 5-hydroxy-imidacloprid and imidacloprid at 0.5 ng/g, which exhibited RSDs of 25% and 21%, respectively. The intra-day precision was less than 10% for many of the investigated compounds. The inter-day precision varied between 2% and 36%, depending on the compound and the concentration. The recoveries varied from 53% to 119%, with averages of 83, 81 and 77% for the extraction of beebread samples spiked at 0.5, 5 and 10 ng/g, respectively. The LOD values for all the substances were below ng/g, with the exception of 6-chloronicotinic acid (LOD = 1.7 ng/g). The method was then applied to the analysis of 32 beebread samples and revealed the presence of 7 of the target substances. The most frequently detected pesticides belonged to the neonicotinoid family and were generally present at low concentrations, but in some cases exceeded 170 ng/g (acetamiprid and thiacloprid). Some pyrethroids were also detected (lambda-cyhalothrine and bifenthrine), but at very low levels. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The honeybee is a land-air insect that participates in the pollination of more than 80% of flowering plants due to its foraging activity. It contributes to both plant biodiversity and improving the quantity and quality of fruit and vegetable production. Globally, this value has been estimated to be D 153 billion [1], making the honeybee an essential target for conservation. The European Parliament therefore adopted, on 25 November 2010 [2], a resolution addressing the challenges in the beekeeping sector requiring that research be conducted on honeybee mortality. Importantly, honeybees have suffered from high mortality worldwide over the past fifteen years. Their annual mortality is estimated to be 30–40%, and the complete destruction of hives is regularly reported. Various factors are

∗ Corresponding author. Tel.: +33 4 37 42 36 08. E-mail addresses: [email protected] (B. Giroud), [email protected] (A. Vauchez), [email protected] (E. Vulliet), [email protected] (L. Wiest), [email protected] (A. Buleté). 0021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2013.09.088

believed to be contributing to this decline, including new and reemerging pathogens, habitat loss, pests and nutritional stress [3]. Pesticide use has also been implicated; however, the nature and extent of the relationship between honeybee decline and pesticide uses is very difficult to establish [4–6]. To better understand the involvement of pesticides in the decline of honeybees, several laboratories have developed reliable and sensitive analytical methods for the determination of pesticides in honeybees as well as in bee products, including incoming products such as pollen and transformation products such as wax and honey. Kubic et al. [7] identified residues of thiophanatemethyl, vinclozolin and iprodione in pollen, bee bread and, to a lower extent, in the honey from beehives in cherry trees treated with the three fungicides during the blooming period. One year later, the same team found evidence for the contamination of the same three matrices by two fungicides, captan and difenoconazole, following their application to apple trees [8]. In a broader study, Mullin et al. demonstrated that high levels of miticides and agrochemicals are present in honeybees as well as their wax and pollen in North American apiaries [9]. Similarly, in a recent

54

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

study conducted in our laboratory, we developed a multi-residue method for the analysis of 80 pesticides in honeys, honeybees and pollens [10]. The application of this method to over 100 samples of each matrix revealed a relatively high percentage of contaminated matrices, primarily by pesticides used to combat varroa but also by fungicides like carbendazime. Beebread is among the matrices that can contaminate honey bee. It is derived from the transformation of plant pollen by biochemical processes caused by the enzymes in the saliva and gastric fluid of the honeybee and is essential for the spawning and survival of honeybees during the winter months. Researchers at the French Agency for Food, Environmental and Occupational Health & Safety (ANSES) were among the first to propose the possibility of the contamination of honey by beebread [11]. They hypothesised that the honeybees could store the contaminated pollen during the honey-season and that the deferred consumption of beebread in winter could cause colony losses. However, at present, few studies have focused on the analysis of pesticide residues in beebread. Recently, a Spanish study revealed the presence of 16 pesticides in beebread [12], and other results in the USA [9,13] quantified active substance levels up to several mg/kg. These data indicate the need to better investigate this matrix to understand its possible involvement in the decline of honeybees. Moreover, beebread contains pollens gathered by honeybees throughout the year. It therefore could be used as a long-term surveillance matrix. Lastly, the investigation of beebread can provide information on the contaminants possibly involved in the decline of honeybees, especially in winter. In light of these concerns, the aim of this study was to develop a simple, fast, sensitive and reliable analytical method for the trace analysis (on the order of ng/g) of pesticides in beebread. This study focuses specifically on the pyrethroid and neonicotinoid families of insecticides and some of their metabolites, as they have recently been introduced on the market, act at low doses per hectare and are undergoing increasing use. One of the advantages of their use, as advanced by chemical companies, is the low dose required: approximately 100 times lower than those of more traditional organochlorines, carbamates or organophosphorous insecticides. However, they are also more toxic (up to 5000 times) and affect non-target insects. Some of them, such as imidacloprid or thiamethoxam (neonicotinoids) or fenvalerate (pyrethroid), have exhibited adverse effects on honeybees [14–16]. Beebread is a very complex matrix that represents a particular analytical challenge for pesticide residue analysis. Its composition varies according to the origin of pollen but is mainly composed of water, proteins, carbohydrates, lipids, inorganic elements and various other components such as decanoic acid, gamma globulin, nucleic acids, vitamins B and C, pantothenic acid, biopterin, neopterin, acetylcholine, reproductive hormones and many other components [17]. Currently, very few methods have been developed for the analysis of organic contaminants in this matrix. The few methods proposed are based on extraction with solvent [8,12] or with concentrated hydrochloric [7] acid followed by one or several clean-up steps [8,12]. A sample preparation method known as QuEChERS (quick, easy, cheap, rugged effective and safe) has been introduced by Anastassiades et al. [18] for the extraction of pesticides from fruits and vegetables. The method consists of a salting-out liquid–liquid extraction using acetonitrile as an organic solvent followed by a clean-up by dispersive solid phase extraction (dSPE). The methods based on QuEChERS approach are being increasingly used and now represent the most commonly applied extraction methods for the determination of pesticide residues in food samples [19–21]. In this study, we developed and validated the first acetonitrile-based method for determining pesticides in beebread.

Currently, neonicotinoids are typically analysed by liquid chromatography (LC) [22,23], whereas pyrethroids are generally analysed by gas chromatography (GC) in environmental matrices, liquids or solids [24]. Although mass spectrometry (MS) offers better selectivity for pyrethroids, electron capture detectors (ECD) are often preferred because of their robustness and high sensitivity for compounds containing one or more halogen groups. More recently, a method involving two-dimensional chromatography (GC × GC-ToF-MS) was optimised for the multi-residue analysis of pesticides, including some pyrethroids. The method improved separation and prevented co-elution problems [25]. One drawback of GC for the analysis of pyrethroids, however, is the existence of cis–trans isomerisation and the degradation of molecules during analysis [26]. Some research groups have therefore used LC coupled with photochemically induced fluorescence detection [27] or mass spectrometry (LC–MS) [26,28] for the analysis of pyrethroids in aqueous solutions. One of our objectives was to develop an analytical method using UPLC–MS/MS for the determination of pyrethroids at sub ng/g levels, which has, to our knowledge, never before been achieved. Our method also addressed neonicotinoids and the metabolites of the two families of insecticides, allowing the simultaneous analysis of all the compounds. Thus, the objective of the study was the development and validation of an original analytical approach that consists of one simple extraction method followed by a selective and sensitive analysis by UPLC–MS/MS to detect both pyrethroids and neonicotinoids and some of their metabolites in beebread at ng/g levels. This method was applied to samples collected from bee hives to verify its robustness as well as obtain preliminary data on the contamination of beebread. 2. Materials and methods 2.1. Standards and reagents Analytical standards of imidacloprid, thiamethoxam, clothianidine, acetamiprid, esfenvalerate thiacloprid, bifenthrine, deltamethrine, lambda-cyhalothrine, imidacloprid-d4 and cypermethrine-d6 were obtained from Sigma–Aldrich (Saint Quentin Fallavier, France). The standard for 6-chloronicotinic acid was purchased from Dr Ehrenstorfer (Augsburg, Germany). All the standards were ≥97.5% purity, with the exception of cypermethine (92.0%). The olefin metabolite of imidacloprid (named olefin in the paper) was synthesised by Orga-Link (Magny les Hameaux, France); 5-hydroxy-imidacloprid was furnished by UMR 406 INRA UAPV (Avignon, France). Individual stock solutions were prepared at concentrations of 1000 mg/L in methanol (MeOH) and were stored at −23 ◦ C. Working solutions were prepared by the appropriate mixture and dilution of the stock solutions. Acetonitrile (ACN), MeOH (UPLC-MS grade), ammonium formate and ammonium acetate were obtained from Biosolve Chimie (Dieuze, France). Heptane (LC grade), acetic acid, formic acid, triethylamine (TEA) and ammonium hydroxide solution (NH3 , aq; 25% in water) were furnished by Sigma–Aldrich. Pure water was obtained from a MilliQ device from Millipore (Saint-Quentin-enYvelines, France). QuEChERS extract tubes were obtained from Agilent Technologies (Massy, France). The acetate buffer contained 1.5 g of NaOAc and 6 g of MgSO4 , while the citrate buffer contained 1 g of NaOCitrate, 4 g of MgSO4 , 1 g of NaCl and 0.5 g of disodium citrate sesquihydrate. Various dispersive SPEs were tested. PSA (mix III), PSA/C18 (mix VI) and PSA/C (mix V) were obtained from Macherey-Nagel (Hoerdt, France). The Florisil phase was purchased from SDS (Peypin, France).

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

2.2. Sampling procedure The blank beebread was collected during summer 2011 from hives on Ouessant Island, located in the Atlantic Ocean near the French coast, where pesticide use is strictly prohibited. During the study, the samples were evaluated to confirm their lack of contaminants. A total of 32 other samples were collected during the beekeeping seasons in 2012 from various French regions. The blank beebread and the various samples were stored at −20 ◦ C until analysis. 2.3. QuEChERS-based extraction A 2 g aliquot of beebread was weighed in a 50 mL polypropylene centrifuge tube. Volumes of 5 mL of pure water, 5 mL of heptane and 10 mL of ACN with TEA 2% were introduced into the tube. A 200 ␮L volume of a solution that contained the internal standards at 100 ␮g/L was added together with a ceramic bar (Agilent Technologies). The mixture was then vigorously shaken by vortex for 15 s. A packet of acetate buffer was added and the tube was immediately manually shaken for 10 s to prevent the coagulation of MgSO4 and swirled on a vortex mixer for 20 s to homogenise the sample. The mixture was then centrifuged at 5000 g for 2 min. An 8 mL aliquot of the supernatant (ACN phase) was transferred to a 15 mL tube and incubated for 15 h at −18 ◦ C. Afterwards, a 6 mL volume of the extract was transferred to a 15 mL centrifuge tube containing 150 mg of PSA and 900 mg of MgSO4 then swirled on a vortex mixer for 10 s. Subsequently, the extract was centrifuged again (5000 rpm for 2 min) and 4 mL of the supernatant was transferred to a glass tube. Lastly, the solvent was evaporated to dryness under a gentle stream of N2 at 40 ◦ C. The dry residue was dissolved in 400 ␮L MeOH. Lastly, 40 ␮L were added to 160 ␮L of pure water for the UPLC–MS/MS analysis. 2.4. UPLC–MS/MS analysis Liquid chromatography was performed using an H-Class UPLC system from Waters (Saint Quentin en Yvelines, France). Various

55

chromatographic columns were evaluated. Lastly, the separation was performed with a Kinetex Phenyl-Hexyl (100 × 2.1 mm, 2.6 ␮m) column from Phenomenex (Le Pecq, France). The mobile phases were (A) 0.01% acetic acid with 0.04 mmol/L ammonium acetate in MilliQ water and (B) MeOH and were applied using the following gradient: 5–90% (B) for 7 min, followed by 100% (B) for 2 min. The column was then equilibrated at the initial conditions for 3 min. The flow rate was 0.4 mL/min, the oven temperature was 60 ◦ C and the injection volume was 2 ␮L. The chromatographic system was coupled to a Xevo TQ-S triplequadrupole mass spectrometer from Waters equipped with a new StepWave ion guide. Electrospray ionisation was performed in the positive mode with the following optimised parameters: capillary voltage 3200 V, desolvation temperature 450 ◦ C, source temperature 150 ◦ C, and nitrogen desolvation and nebuliser gas flows 900 L/h and 150 L/h, respectively. For each compound, the IntelliStartTM Software was used to automatically select the m/z value for the precursor ion as well as product ions, capillary and cone voltage, desolvation gas flow and collision energy. Thus, two multiple reaction monitoring (MRM) transitions were optimised. The target ion transition with the highest intensity (MRM1) was used for quantitation, whereas the second target ion transition (MRM2) was used for confirmation. For the pesticide esfenvalerate, a third transition, MRM3, was followed. For each neonicotinoid and their metabolites, the protonated molecular ions [M + H]+ were chosen as the precursor ions, whereas for the five pyrethroid compounds, the ammonium adducts [M + NH4 ]+ were selected. The ion transitions, cone voltages, collision energies and dwell times for the analytes are displayed in Table 1. 2.5. Validation of the method The performance of the method was evaluated based on the recommendations provided by the International Conference Harmonisation ICH/2005 directives [29]. The method was performed over 3 days, according to the plan presented in Fig. 1. The mathematical model providing the best fit to the calibration curve was determined by analysing spiked beebread matrices

Table 1 Retention time (tR ), transitions used for the quantification (MRM1) and confirmation (MRM2), dwell time and source parameters. Compound

tR (min)

Transitions

Dwell time (s)

Cone voltage (V)

Collision energy (eV)

6-Chloronicotinic acid

1.75

0.003

Thiamethoxam

3.18

Olefin

3.18

5-Hydroxy-imidacloprid

3.28

Clothianidine

3.37

Imidacloprid

3.74

Acetamiprid

4.04

Thiacloprid

4.48

Lambda-cyhalothrine

8.16

Cypermethrine

8.22

Deltametrine

8.28

Esfenvalerate

8.29

Bifenthrine

8.37

MRM1 158.2 → 122.1 MRM2 158.2 → 78.0 MRM1 292.2 → 211.1 MRM2 292.2 → 181.1 MRM1 254.2 → 171.1 MRM2 254.2 → 205.1 MRM1 272.2 → 191.1 MRM2 272.2 → 225.1 MRM1 250.2 → 169.0 MRM2 250.2 → 132.0 MRM1 256.2 → 175.1 MRM2 256.2 → 209.2 MRM1 223.2 → 126.1 MRM2 223.2 → 187.1 MRM1 253.2 → 126.1 MRM2 253.2 → 186.0 MRM1 467.2 → 225.1 MRM2 467.2 → 141.1 MRM1 433.2 → 191.1 MRM2 433.2 → 416.1 MRM1 523.1 → 281.0 MRM2 523.1 → 506.0 MRM1 437.3 → 167.1 MRM2 437.3 → 125.1 MRM3 437.3 → 420.2 MRM1 440.2 → 181.1 MRM2 440.2 → 166.0

28 20 22 22 18 18 16 16 20 20 16 16 28 28 20 20 6 6 20 20 10 10 30 30 30 32 32

18 21 12 24 18 18 16 14 12 14 18 14 20 12 20 12 12 46 12 8 14 8 12 44 6 14 44

0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.025 0.003 0.025 0.003

0.003

56

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

3 days - 6 concentration levels C1

C2 1st day 2 x 6 concentrations 3 x C2 recovery at C2

C3

C4

C5 2nd day 2 x 6 concentrations 3 x C5 recovery at C5

C6 3rd day 2 x 6 concentrations 3 x C6 recovery at C6

Fig. 1. Validation plan conducted on three separate days.

over a range of 6 concentrations (C1–C6) from 0.1 to 10 ng/g, with the exception of 6-chloronicotinic acid (between 5 and 500 ng/g) and bifenthrin (between 0.25 and 25.0 ng/g). This model was evaluated in triplicate on 3 separate days for each compound. The data obtained also allowed the determination of the intermediate precision for 3 concentrations (C2, C5 and C6). The repeatability was determined by performing two additional extractions for C2 on the first day, C5 on the second day and C6 on the third day. The intermediate precision (or inter-day precision) and repeatability (or intra-day precision) were expressed as the relative standard deviation (RSD, %) of the series of 3 measurements for the different concentrations. The recoveries were determined at the C2, C5 and C6 concentrations by analysing beebread samples spiked at the different concentrations and comparing the areas obtained to those of beebread extracts spiked following the sample preparation. The limit of detection (LOD) was quantified as the analyte concentration that produced a peak signal of 3 times the background noise from the chromatogram for the MRM2 transition. The limit of quantification (LOQ) was quantified as the analyte concentration that produced a peak signal of 10 times the background noise from the chromatogram for the MRM1 transition. 2.6. Data analysis and quantification The analytes were identified by both their chromatographic characteristics and their MRM-specific fragmentation. Thus, analyte identification was based on three criteria: (1) the comparison of the retention times of the analyte and a standard compound (±2.5%), (2) the presence of MRM1 and MRM2, and (3) the comparison of the specific ratios of MRM1/MRM2 with respect to the ratios of the analytical standards (deviation <20%). This allows meeting unequivocal identification criteria recommended by the ICH standards [29]. The data processing was performed using the software MassLynx 4.1 from Waters. A matrix-matched calibration was used for quantification. A range of 6 points was used, from 0.1 to 10 ng/g, with the exception of 6-chloronicotinic acid (from 5 to 500 ng/g) and bifenthrin (from 0.25 to 25.0 ng/g). Moreover, imidacloprid-d4 and cypermethrined6 were used as internal standards for the neonicotinoids and pyrethroids, respectively. 3. Results and discussion 3.1. Chromatographic conditions The optimisation of the chromatographic conditions for the multi-residue analysis of pyrethroids and neonicotinoids is challenging due to the different physico-chemical properties of the two families. While pyrethroids are apolar (log P between 6.15 and 7.23), neonicotinoids exhibit a polar character (log P between 0.8 and 1.26). Appropriate chromatographic conditions must allow not only the simultaneous analysis of the two families of substances but also the separation of individual insecticides within the same family.

The chromatographic conditions were optimised with respect to the column, the mobile phase composition, and finally the nature of the solvent and the volume used for injection. In a first approach, we conducted tests using a column previously employed in our laboratory for the multi-residue analysis of contaminants in bee products [10]. Different compositions and gradients of water/methanol or water/acetonitrile as the mobile phase revealed that the ionisation of pyrethroids is favoured with methanol, which was therefore chosen as the solvent. Subsequently, several aqueous phases were tested, comprising MilliQ water with ammonium formate or acetate, then the effect of incorporating acid into the aqueous phase was evaluated by adding either formic or acetic acid. Pyrethroids were observed to ionise more efficiently in the presence of ammonium acetate at a concentration of 0.4 mmol/L. These tests revealed also that the best compromise was to incorporate acetic acid into the mobile phase at a low concentration of 0.01%. Lastly, the effect of the desolvation temperature was evaluated within a range of 250–650 ◦ C. The results indicated that the optimal ionisation of pyrethroids and neonicotinoids occurred at 450 ◦ C and 650 ◦ C, respectively. A temperature of 450 ◦ C was chosen to favour the ionisation of pyrethroids. Using these conditions, a variety of columns with different geometries and bonded phases was then evaluated using various gradients composed of MeOH/water with acetic acid and ammonium acetate. The chromatographic conditions were optimised to obtain an even distribution of the analytes and to prevent coelution with highly polar matrix compounds near the dead volume. The conditions were also selected to obtain the best signal-to-noise ratio. The C18-type columns proved too apolar for our compounds and did not allow sufficient selectivity for aromatics. The columns including a polar amide group embedded in a C14-alkyl chain adversely affected the peaks by inducing asymmetry and reducing their resolution. The pentafluorophenyl and Phenyl-Hexyl phases were then compared. PFP columns provide less resolution and inferior peak intensities compared with the Phenyl-Hexyl column, reducing the signal to noise ratio. The Phexyl-Hexyl phase was therefore selected for subsequent analyses. This column allows separation of the target compounds in approximately 8.5 min (Fig. 2). 3.2. Sample preparation protocol based on modified QuEChERS The analysis of complex matrices such as beebread requires rigorous sample preparation to achieve an analysis that is repeatable and satisfies the required detection limits (here, the ng/g range). Because the 13 studied molecules exhibit different characteristics and physical/chemical properties, the optimisation of the sample preparation protocol was challenging. The QuEChERS-based extraction method consists in a liquid–liquid extraction principle. Analytes are extracted from the matrix by an organic solvent that is subsequently salted out from an aqueous matrix. Because the beebread matrix is complex, an additional purification step (dispersive SPE) was necessary to limit the presence of interfering substances. Indeed, one of the greatest drawbacks of LC–ES–MS is the perturbation of the signal by co-extracted substances from the sample matrix. These beebread components can induce either ion suppression or enhancement of the analyte in the ES interface, leading to sources of error during the quantification. The presence of the matrix effect in the mass spectrometric analysis was evaluated by comparing the peak areas of the standards in the mobile phase with those of the same quantities of standards added to the spiked samples following extraction. The response of each compound in the mobile phase was designated as the 100% response value. To determine the optimal extraction procedure, several parameters were studied: the choice of buffer, the properties and volumes

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

57

Fig. 2. UPLC–MS/MS chromatogram corresponding to a mixture of the of the 13 target compounds in H2 O/MeOH (80/20) at 100 ␮g/L and with a 2 ␮L injection volume (1: 6-chloronicotinic acid; 2: thiamethoxam; 3: olefin; 4: 5-hydroxy-imidacloprid; 5: chlothianidine; 6: imidacloprid; 7: acetamiprid; 8: thiacloprid; 9: lambda-cyhalothrine; 10: cypermethrine; 11: deltamethrin; 12: esfenvalerate; 13: bifenthrine).

of the solvent, the effect of freezing, and the final conditions under which the extract is dissolved. 3.2.1. Choice of buffer Two standards are used with regard to the buffer type: the American standard (AOAC), which involves the use of acetate buffer [30], and the European standard, EN 15662, which involves the use of citrate buffer [31]. The original QuEChERS method, introduced by Anastassiades et al. in 2003 [18], employed ACN as the extraction solvent. Since this time, the majority of extraction methods used for pesticides employ ACN. The responses obtained with the two buffers were compared by performing extractions with 5 mL of water and 10 mL of ACN and either one or the other buffer. The extracts obtained in both cases were evaporated, dissolved in 0.4 mL MeOH, and diluted by a factor of 5 in MeOH/water (20/80) prior to being injected into the UPLC–MS/MS system. The signal intensities thus obtained (Fig. 3) revealed that the acetate buffer was generally more efficient in extracting the target substances. Indeed, with the exception of thiacloprid, which provided a higher signal with the citrate buffer, the signals obtained using acetate were either higher or similar in intensity to those obtained using citrate. Therefore, the citrate-based buffer was excluded from additional experiments. 3.2.2. Properties and volumes of the solvents Analyses performed by the Karl–Fisher method revealed that the beebread samples were composed of 17% water on average. Consequently, 5 mL of water was added in the preliminary experiments to each 2 g sample of beebread to produce a total water content of 80%, as recommended by the standard [30]. Tests were then conducted to compare the recoveries obtained with additions of 3, 5 or 8 mL of water. No significant differences were observed between the samples, and thus the middle volume of 5 mL was used. In the final method, 2% TEA was added to 10 mL of ACN in the extraction step. Without the addition of TEA, 6-chloronicotinic acid was completely unrecovered during the purification step involving the PSA phase due to binding of the carboxylic acid. The use of a base to modify the pH was therefore necessary. The use of ammonia

(less toxic than TEA) was considered, but this base did not allow recovery of the acidic metabolite. Lastly, the strongly polar base TEA was chosen. The addition of heptane allows the extraction of lipids and thus limits their presence in the ACN phase. Volumes of 3 and 5 mL of heptane were compared; recoveries were higher with 5 mL of heptane. 3.2.3. Freezing step According to the NF standard [30], the use of a freezing step following the salting-out extraction facilitates the precipitation of sugar, wax and lipid residues, allowing their removal. This method has the advantage of a simplified clean-up and removes the need for additional equipment. Different freezing times were evaluated for our study. An 8 mL volume of the ACN phase recovered following the centrifugation that succeeded the extraction step was frozen for 1 h, 3 h, 5 h and 15 h. During this process, the lipids were precipitated while the target pesticides remained soluble in the acetonitrile phase. Immediately following freezing, the precipitate was removed and the acetonitrile was evaporated. The dry residue was dissolved in MeOH/water (20/80) prior to injection. Fig. 4 depicts the decrease in the matrix effect due to the freezing step. This decrease was most pronounced for neonicotinoid pesticides and their metabolites. A positive matrix effect, i.e., an increase in signal, was observed for esfenvalerate that was frozen for a short period. A negative matrix effect was observed, however, when the duration of freezing was increased to 15 h, achieving a decrease the same order of magnitude as that observed for the other neonicotinoids. 3.2.4. Optimisation of the dSPE clean-up We focused on dispersive SPE (d-SPE) for the development of an easy and rapid protocol for sample preparation. The extract recovered during the extraction was placed into a tube that contained a sorbent for purification and 900 mg MgSO4 for eliminating excess water. Several sorbents were tested: PSA, C, C18, C/C18 (50/50), PSA/C18 and Florisil. To determine the optimal clean-up conditions, both the recoveries during this step and the matrix effects were

58

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

acetate buffer citrate buffer

Signal intensities

10 8 6 4

imidacloprid

thiamethoxam

thiacloprid

olefin

lamdacyhalothrine

deltamethrine

cypermethrine

clothianidine

6chloronicotinic acid

acetamiprid

5-hydroxyimidacloprid

0

bifenthrine

2

Fig. 3. Influence of the choice of buffer on the signal intensities of the target compounds.

Fig. 4. Effect of freezing on the matrix effects of the target compounds.

considered. However, the results were summarised by comparing signal intensities (Fig. 5), which are indicative of the recovery and disposal of interfering substances. In this figure, 150 mg of each sorbent was used. The C18, C and Florisil sorbents caused a loss of signal for several compounds, particularly the pyrethroids lambda-cyalothrine, cypermethrine and, to a lesser extent, deltamethrine. The loss observed with C and C18 stemmed primarily from the poor recoveries of the most apolar substances (between 0 and 45% in the case of pyrethroids) following the purification step. Florisil is a polar sorbent with a basic pH and is generally used to extract non-polar to moderately polar compounds. This phase contains magnesium ions, which allow the retention of chlorinated pesticides. However,

it was unsuitable for our purification. The C18, C and Florisil sorbents were consequently abandoned. The phase that produced the most intense signals contained primary and secondary amines (PSA). Further experiments were conducted by varying the amount of PSA (from 75 to 500 mg); 150 mg provided the most intense signals and therefore the best signal to noise ratio. 3.2.5. Dissolution of the dry extract The properties and volumes of the solvents used for the resolubilisation and injection of the extract were also optimised. After evaporating to dryness, the extract obtained from the spiking of the blank matrix at 5 ng/g was dissolved in 400 ␮L of a solvent solution, with the solution compositions ranging from 100% MeOH to

Fig. 5. Signal intensities measured after the use of 150 mg of various clean-up phases.

26 19 27 26 3 28 27 34 14 11 149 22 17 21 23 2 22 21 15 3 14 128 8 36 7 7 2 8 12 21 8 16 67 83 (1) 71 (2) 89 (3) 83 (1) 74 (19) 82 (1) 76 (4) 75 (19) 53 (12) 54 (3) 109 (17) – – 93 (3) 83 (7) 85 (4) 97 (5) 119 (12) 109 (7) 103 (3) 69 (8) 76 (13) 85 (11) 77 (12) 53 (4) 87 (18) 89 (3) 66 (16) 89 (25) 83 (8) 75 (21) 91 (1) 85 (4) 84 (2) 74 (10) 89 (9) – – – −46 −36 −35 −46 −70 −37 −31 −61 −67 −78 79 −18 −76 0.05 1.5 1.250 0.160 0.200 0.068 0.096 2.600 0.100 0.180 – 5.7 2.1

c

a

b

0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 0.1–10 5–500 0.25–25 Thiamethoxam Olefin 5-Hydroxy-imidacloprid Clothianidine Imidacloprid Acetamiprid Thiacloprid Lambda-cyhalothrine Cypermethrine Deltametrine Esfenvalerate 6-Chloronicotinic acid Bifenthrine

C2 = 0.5 ng/g. C5 = 5 ng/g (except 6-chloronicotinic acid: 250 ng/g; and bifenthrine: 12.5 ng/g). C6 = 10 ng/g.

0.017 0.750 0.250 0.083 0.042 0.013 0.021 0.825 0.042 0.100 0.05 1.70 0.62 0.9978 0.9924 0.9949 0.9966 0.9927 0.9983 0.9975 0.9782 0.9910 0.9943 – 0.9618 0.9778

C5 C2 C5

Recovery ± %RSD

a b

matrix effect (%) LOQ (ng/g) LOD (ng/g) r2 model domain (ng/g) Compound

Table 2 Performances of the developed method.

The entire method, including the sample preparation protocol and the LC–MS/MS analysis, was validated by characterising the limits of detection and quantification, the mathematical best fit to the calibration curve, the extraction recovery, the matrix effects, and the repeatability, reproducibility and accuracy. The use of matrix-matched calibration standards was done. To meet the ICH standard, more than nine evaluations of intra- and inter-day precision (3 concentrations/3 replicates/3 days) were performed (with the exception of bifenthrine and 6-chloronicotinic acid). The elements of the validation are summarised in Table 2. It was not possible to validate the analytical protocol for esfenvalerate in the desired concentration range because the results for the calibration did not produce a defined model. Although the recoveries were favourable at concentrations of 5 and 10 ng/g with RSD values of less than 20%, the inter-day precision was poor (greater than 120% for the two concentrations tested). Consequently, the analysis of this compound was limited to its qualitative detection in environmental samples, and quantification was not achieved. For the remaining compounds, a quadratic model provided the best fit to the calibration curves. For chlothianidine, both linear and quadratic models were successfully applied, but the quadratic model was ultimately selected for all of the substances. Typically the linearity is assessed for the method validation. However, it has been previously observed that some analytical methods better fit the quadratic model, both in GC–MS [32] and LC–MS [33]. The bending of the calibration curve with electrospray ionisation could be explained by the properties of the analyte itself and the presence of other ionisable compounds [34]. Positive results were obtained for the intra-day precision, with RSD values below 20%, except for 5-hydroxy-imidacloprid and imidacloprid at 0.5 ng/g, which exhibited RSDs of 25% and 21%, respectively. The intra-day precision was less than 10% for many of the investigated compounds. The inter-day precision varied greatly depending on the compound and the concentration and was typically between 2% and 36%. Only two substances exceeded 30%. The recoveries varied from 53 to 119%, with averages of 83, 81 and 77% for the extraction of beebread samples spiked at 0.5, 5 and 10 ng/g, respectively. One of the major drawbacks of the use of electrospray is the suppression or enhancement of the analyte signal by co-extracted matrix substances. This matrix effect was quantified for a concentration of 5 ng/g (except 6-chloronicotinic acid: 250 ng/g and bifenthrine: 12.5 ng/g). A matrix effect was observed for all the substances. Signal suppression was most often observed, but an enhanced response was observed for esfenvalerate. The LOD values for all the substances were below ng/g, with the exception of 6-chloronicotinic acid (LOD = 1.7 ng/g). This sensitivity surpasses those reported in previously published studies

b

3.3. Validation of the method

59

Quadratic Quadratic Quadratic Quadratic Quadratic Quadratic Quadratic Quadratic Quadratic Quadratic – Quadratic Quadratic

C5b C2a C6

c

Inter-day precision (%RSD)

80/20 H2 O/MeOH. The results indicated that the presence of MeOH was necessary for the solubilisation of all of the target compounds. However, the presence of an organic solvent in the final extract affected the chromatographic separation and decreased the signal to noise ratio. The best compromise was found to consist of an 80/20 H2 O/MeOH solution. Lastly, the injection volume was chosen. This volume should maximise the introduction of analytes into the system without introducing too many residual interfering substances. To ensure the longest possible lifetime of our chromatographic column, we tested only low injection volumes from 2 to 5 ␮L. The injection volume 2 ␮L was chosen because it limited the introduction of interfering substances that would increase matrix effects, as illustrated in Fig. 6.

C6c

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

60

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

Fig. 6. The influence of the injection volume on the matrix effect of the target compounds. Table 3 Pyrethroids, neonicotinoids and metabolites residues in beebread samples. Compound

Positive sample

Sample < LOQ

Min. value

Max.value

6-Chloronicotinic acid Olefin 5-Hydroxy-imidacloprid Imidacloprid Cypermethrine Esfenvalerate Bifenthrine Lambda-cyhalothrine Deltametrine Acetamiprid Clothianidine Thiacloprid Thiamethoxam

1 0 0 8 0 0 2 2 0 2 0 24 12

1 – – 6 – – 2 1 – 0 – 16 7

– – – 0.3 – – –

100

7: MRM of 2 Channels ES+ TIC (imidacloprid) 1.63e5

3.80

(A)

%

3.90

0 100

1.50

2.00

2.50

3.00

3.50

4.00

(B)

4.50 4.52

5.00

4.50

5.00

5.50

6.00

6.50

7.00

7.50

8.00

8.50

5: MRM of 2 Channels ES+ TIC (thiacloprid) 5.35e7

%

0 100

1.50 1.07

2.00

2.50

3.00

3.50

4.00

5.50

6.00

6.50

7.00

8.00

8.50

1: MRM of 2 Channels ES+ TIC (6-chloronicotinic acid) 6.92e4

1.85

(C)

7.50

%

0

1.50

2.00

2.50

3.00

3.50

4.00

4.50

5.00

5.50

6.00

6.50

7.00

7.50

8.00

8.50

Time

Fig. 7. An example of the extracted MRM1 chromatograms of a beebread sample, indicating the presence of (A) imidacloprid, (B) thiacloprid and (C) 6-chloronicotinic acid.

B. Giroud et al. / J. Chromatogr. A 1316 (2013) 53–61

of bee products using multi-residue analysis [9,10,35–38] or the targeted analysis of limited substances or groups of substances [36,39]. These very low values were achieved through efficient extraction and purification and by employing the latest generation mass spectrometer. 3.4. Application to collected beebread samples Table 3 presents the compounds detected in 32 samples of beebread. Of the 13 targeted pesticides and metabolites, 7 were detected in at least one sample. A typical chromatogram of a beebread sample containing imidacloprid (0.31 ng/g), thiacloprid (177.1 ng/g) and 6-chloronicotinic acid (
61

References [1] N. Gallai, J.M. Salles, J. Settele, B. Vaissière, Ecol. Econ. 68 (2009) 810. [2] http://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//TEXT+TA+ P7-TA-2010-0440+0+DOC+XML+V0//FR&language=FR [3] D. vanEngelsdorp, M.D. Meixner, J. Invert. Pathol. 103 (2010) S80. [4] E. Stokstad, Science 316 (2007) 970. [5] L. Maxim, J.P. van der Sluijs, Environ. Res. Lett. 5 (2010) 1. [6] F.L.W. Ratnieks, N.L. Carreck, Science 327 (2010) 152. [7] M. Kubik, J. Nowacki, A. Pidek, Z. Warakomska, L. Michalczuk, W. Goszczynski, Apidologie 30 (1999) 521. [8] M. Kubik, J. Nowacki, A. Pidek, Z. Warakomska, L. Michalczuk, W. Goszczynski, Apidologie 31 (2000) 531. [9] C.A. Mullin, M. Frazier, J. Frazier, S. Ashcraft, R. Simonds, D. vanEngelsdorp, J.S. Pettis, PLoS ONE 5 (2010) e9754. [10] L. Wiest, A. Buleté, B. Giroud, C. Fratta, S. Amic, O. Lambert, H. Pouliquen, C. Arnaudguilhem, J. Chromatogr. A 1218 (2011) 5743. [11] AFFSA report 2008, Affaiblissements, mortalités de colonies d’abeilles. (1987–2007). [12] F.J. Orantes-Bermejo, A. Gómez Pajuelo3, M. Megías Megías, C. Torres, ˜ J. Apicult. Res. 48 (2010) 243. Fernández-Pínar, [13] D. vanEngelsdorp, J.D. Evans, C. Saegerman, C. Mullin, E. Haubruge, B.K. Nguyen, M. Frazier, J. Frazier, D. Cox Foster, Y. Chen, R. Underwood, D.R. Tarpy, J.S. Pettis, PLoS ONE 4 (2009) 1. [14] S. Suchail, D. Guez, L.P. Belzunces, Environ. Toxicol. Chem. 20 (2001) 2482. [15] M.E. Colin, J.M. Bonmatin, I. Moineau, C. Gaimon, S. Brun, J.P. Vermandere, Arch. Environ. Contam. Toxicol. 47 (2004) 87. [16] T. Iwasa, N. Motoyam, J.T. Ambrose, R.M. Roe, Crop Prot. 23 (2004) 371. [17] V.A. Isidorov, A.G. Isidorova, L. Sczczepaniak, U. Czyzewska, Food Chem. 115 (2009) 1056. [18] M. Anastassiades, S. Lehotay, D. Stajnbaher, F. Schenk, J. AOAC Int. 86 (2003) 412. [19] O.D. Prestes, C.A. Friggi, M.B. Adaime, R. Zanella, Quim. Nova 32 (2009) 1620. [20] A. Wiilkowska, M. Biziuk, Food Chem. 125 (2011) 803. [21] J. Fenik, M. Tankiewicz, M. Biziuk, Trends Anal. Chem. 6 (2011) 814. [22] S.Y. Liu, Z.T. Zheng, F.L. Wei, Y.P. Ren, W.J. Gui, H.M. Wu, G.N. Zhu, J. Agric. Food Chem. 58 (2010) 3271. [23] C. Mohan, Y. Kumar, Y.J. Madan, N. Saxena, Environ. Monit. Assess. 165 (2010) 573. [24] M.L. Feo, E. Eljarrat, D. Barcelo, Trends Anal. Chem. 7 (2010) 692. [25] K. Banerjee, S.H. Patil, S. Dasgupta, D.P. Oulkar, S.B. Patil, R. Savant, P.G. Adsule, J. Chromatogr. A 1190 (2008) 350. [26] A. Valverde, A. Aguilera, M. Rodriguez, M. Boulaid, J. Chromatogr. A 943 (2001) 101. [27] P. Parrilla Vazquez, A.R. Mughari, M. Martinez Galera, J. Chromatogr. A 118 (2008) 61. [28] M.D. Gil-García, D. Barranco-Martínez, M. Martínez-Galera, P. Parrilla-Vázquez, Rapid Commun. Mass Spectrom. 20 (2006) 2395. [29] International Conference on Harmonization, Q2 (R1): Validation of Analytical procedures: Text and Methodology, 2005, November. [30] AOAC Official Method, 2007.01, Pesticides Residues in Food by Acetonitrile Extraction and Partitioning with Magnesium Sulphate, 2007. [31] NF EN 15662, 2009.01, Méthode polyvalente de détermination des résidus des pesticides par GC-MS et SL/SM/SM avec extraction/partition avec de l’acétonitrile et nettoyage par SPE disperses (Méthode QuEChERS), 2009. [32] C. Przybylski, F. Hommet, J. Chromatogr. A 1201 (2008) 78. [33] M. Hiemstra, A. de Kok, J. Chromatogr. A 1154 (2007) 3. [34] R.B. Cole, Electrospray Ionisation Mass Spectrometry: Fundamentals, Instrumentation and Applications, Wiley-Interscience, New York, 1997. [35] S. Walorczyk, B. Gnusowski, J. Chromatogr. A 1216 (2009) 6522. [36] J. Serra-Bonvehi, J. Orantes-Bermejo, Pest. Manag. Sci. 66 (2010) 1230. [37] D. Tomasini, M.R.F. Sampaio, S.S. Caldas, J.G. Buffon, F.A. Duarte, E.G. Primel, Talanta 99 (2012) 380. [38] M. Garcia-Chao, M.J. Agruna, G.F. Calvete, V. Sakkas, M. Llompart, T. Dagnac, Anal. Chim. Acta 672 (2010) 107. [39] A. Kamel, J. Agric. Food Chem. 58 (2010) 5926. [40] M.I. Smodis Skerl, B. Velikonla, H. Basa Cesnik, A. Gregorc, Bull. Environ. Contam. Toxicol. 83 (2009) 374.