Analytical and Bioanalytical Chemistry

, Volume 405, Issue 28, pp 9251–9264

Simultaneous determination of residues in pollen and high-fructose corn syrup from eight neonicotinoid insecticides by liquid chromatography–tandem mass spectrometry

Authors

  • Mei Chen
    • Department of Environmental HealthHarvard School of Public Health
  • Erin M. Collins
    • Department of Environmental HealthHarvard School of Public Health
  • Lin Tao
    • Department of Environmental HealthHarvard School of Public Health
    • Department of Environmental HealthHarvard School of Public Health
Research Paper

DOI: 10.1007/s00216-013-7338-7

Cite this article as:
Chen, M., Collins, E.M., Tao, L. et al. Anal Bioanal Chem (2013) 405: 9251. doi:10.1007/s00216-013-7338-7

Abstract

The neonicotinoids have recently been identified as a potential contributing factor to the sudden decline in adult honeybee population, commonly known as colony collapse disorder (CCD). To protect the health of honeybees and other pollinators, a new, simple, and sensitive liquid chromatography-electrospray ionization mass spectrometry method was developed and validated for simultaneous determination of eight neonicotinoids, including acetamiprid, clothianidin, dinotefuran, flonicamid, imidacloprid, nitenpyram, thiacloprid, and thiamethoxam, in pollen and high-fructose corn syrup (HFCS). In this method, eight neonicotinoids, along with their isotope-labeled internal standards, were extracted from 2 g of pollen or 5 g of HFCS using an optimized quick, easy, cheap, effective, rugged, and safe extraction procedure. The method limits of detection in pollen and HFCS matrices were 0.03 ng/g for acetamiprid, clothianidin, dinotefuran, imidacloprid, thiacloprid, and thiamethoxam and ranged between 0.03 and 0.1 ng/g for nitenpyram and flonicamid. The precision and accuracy were well within the acceptable 20 % range. Selectivity, linearity, lower limit of quantitation, matrix effect, recovery, and stability in autosampler were also evaluated during validation. This validated method has been used successfully in analyzing a set of pollen and HFCS samples collected for evaluating potential honeybee exposure to neonicotinoids.

Keywords

Neonicotinoid insecticidesPollenHFCSQuEChERSLC-MS/MS

Introduction

The neonicotinoids are a group of systemic insecticides with a common mode of action that affects the central nervous system by acting as nicotinic acetylcholine receptor agonists, causing paralysis and then death. Several neonicotinoids are being commonly used as seed and soil treatments, and some are also directly applied to plant foliage [1]. The extensive use of neonicotinoids has led to increasing concern about toxicity to natural pollinators, and neonicotinoids have recently been identified as a potential contributing factor to the sudden decline in adult honeybee population, commonly known as colony collapse disorder (CCD) [24].

The main consideration when examining honey bee exposure to neonicotinoids is the contamination of pollen, as it is the main food source for honey bees. Given their systemic properties, neonicotinoids applied to the seed, and pre-treated soil are taken up by the roots and travel through the entire plant, including the flowers [57]. High-fructose corn syrup (HFCS) made from corn grown from neonicotinoid treated seeds could contain residue that could put honey bee colonies at risk of sub-lethal exposure if it is used to feed honey bees as an alternative to sucrose-based food by commercial beekeepers. Sub-lethal levels of neonicotinoid exposure could cause disorientation and associative learning problems in honeybees [8]. Considering the essential role of honeybees in honey production and pollination, it is therefore important to assess routinely neonicotinoid residues in pollen and HFCS to protect the health of honeybees and other pollinators.

The thermolability and high polarity of neonicotinoids make them difficult to be analyzed by gas chromatography (GC). Various analytical methods, such as ELISA [912], high-performance liquid chromatography (HPLC) with UV detection, or diode-array detection [1315] have been developed for measuring neonicotinoids in environmental and agricultural samples. Liquid chromatography coupled with mass spectrometry (LC-MS) [16, 17] or with tandem MS (LC-MS/MS) [1824] is currently the preferred analytical method because it has provided improved sensitivity, selectivity, and accuracy, as well as faster analysis. Most of the current LC-MS/MS methods reported for the determination of neonicotinoid residues are in fruits, vegetables, and other agriculture products [1825]. Only a few LC-MS/MS methods targeted neonicotinoids in pollen, and none of them monitored more than four neonicotinoids (not including the metabolites) [6, 2629]. Moreover, the detection limits (LOD) of these method ranged from 4.3 ng/g using 2–3 g of pollen samples to as low as 0.2 ng/g, but still using 15 g of pollen. Among a few published LC-MS/MS methods which analyzed eight neonicotinoids in fruits, vegetables, and bee matrices, the method limits of quantitation (LOQ) ranged between 2 and 5 ng/g [21, 25]. Although these sensitivities are sufficient to monitor neonicotinoids residues based on the maximum residue limits (MRL) in honey (no MRL available for pollen), which ranges from 10 to 200 [30], a more sensitive analytical method with as small a sample size as possible is needed to monitor the sub-lethal level of neonicotinoids exposure to bees. To the best of our knowledge, no analytical method has been published for neonicotinoid detection in HFCS.

The sample preparation steps in current analytical methods reported in determination of neonicotinoids in pollen often involve tedious liquid–liquid partitioning [6, 31] or time-consuming and expensive solid phase extraction [26] for sample extraction and cleanup. In recent years, the sample preparation procedure known as quick, easy, cheap, effective, rugged, and safe (QuEChERS) has been widely used for the extraction of a wide variety of pesticides from different sample matrices [24, 25, 32]. QuEChERS methodology was also used in a few studies in pollen [2729], but the shaking extraction potency and efficiency and sample cleanup in these methods are not sufficient enough. In addition, these methods can be further improved by incorporating the isotope-labeled neonicotinoids as internal standards (IS) for quantification [29] and adding IS from the beginning of the sample extraction [27]. Therefore, for routine and/or large-scale assessment of honeybee exposure, an improved analytical method is needed that can quantify all eight neonicotinoids simultaneously in complicated matrices, such as pollen and HFCS, with improved sensitivity using smaller sample sizes while being low cost and easily prepared.

In the present study, we reported a simple, rapid, and sensitive analytical method for simultaneous determination of acetamiprid, clothianidin, dinotefuran, flonicamid, imidacloprid, nitenpyram, thiacloprid, and thiamethoxam in pollen and HFCS by LC-MS/MS using an optimized and improved QuEChERS extraction procedure. In this QuEChERS procedure, we added an additional QuEChERS ceramic homogenizer in the first part of the extraction to further reduce the extraction time and ensure high extraction repeatability and use dispersive SPE containing both PSA and C18 to obtain cleaner samples extracts. The current method is the first method that simultaneously analyzes eight neoniconiods, which are all neonicotinoid insecticides currently available on the market in pollen and HFCS. This method uses matrix matched standard curves for quantitation and has shown better sensitivity than analytical methods currently reported in analyzing neonicotinoids in pollen and bee matrices. The method LOD is 0.03 ng/g for acetamiprid, clothianidin, dinotefuran, imidacloprid, thiacloprid, and thiamethoxam and ranges between 0.03 and 0.1 ng/g for nitenpyram and flonicamid in 2 g of pollen or 5 g of HFCS sample. The method has been validated in terms of selectivity, accuracy, precision, linearity, lower limit of quantification, matrix effect, recovery, and stability. We then successfully applied this method to analyze a set of field pollen and HFCS samples collected for evaluating the potential honeybee exposure to neonicotinoids.

Experimental

Chemicals and standard solutions

Acetamiprid, flonicamid, thiacloprid, thiamethoxam, and nitenpyram standard solutions were purchased from Accustandard (New Haven, CT, USA) with purity higher than or equal to 99.7 %. Imidacloprid and clothianidin were purchased from Sigma-Aldrich (St. Louis, MO, USA) with purity of 99.9 %. Dinotefuran was purchased from Chem service (West Chester, PA, USA) with purity of 99.2 %. The isotope labeled IS for imidacloprid-d4 (99.2 %), clothianidin-d3 (98.9 %), and thiamethoxam-d3 (99.8 %) were purchased from C/D/N Isotopes, Inc. (Quebec, Canada). LC-MS grade formic acid and ammonium formate were purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade reagents, including acetonitrile and water were purchased from JT Baker (Center Valley, PA, USA); 1.5 mL vials with 0.2-μm nylon filter was purchased from VWR international (Radnor, PA, USA). QuEChERS extraction salt package which includes 4 g MgSO4, 1 g NaCl, 1 g sodium citrate, and 500 mg of disodium citrate sesquihydrate in each salt pack, and 2 mL of QuEChERS dispersive SPE containing 50 mg PSA + 150 mg MgSO4, 50 mg PSA + 50 mg C18 + 150 mg MgSO4, or 50 mg PSA + 50 mg C18 + 150 mg MgSO4+GCB were purchased from Agilent Technologies (Santa Clara, CA, USA).

The individual stock standard solutions for each neonicotinoid were prepared in acetonitrile at a concentration of 100 μg/mL and then stored in amber glass vials at −20 °C until use. The intermediate standard solution containing eight neonicotinoids was prepared by aliquoting the appropriate volume of individual stock standards and diluting with acetonitrile. All working solutions for calibration, QCs were prepared by serial dilutions of this multi-component intermediate standard solution with acetonitrile. The initial IS stock solution of 1,000 μg/mL of imidacloprid-d4, clothianidin-d3, and thiamethoxam-d3 was prepared by dissolving 10 mg of the commercial standard compounds in 10 mL methanol and stored at −20 °C. The intermediate IS solution was prepared by aliquoting individual stock solutions and diluting with acetonitrile to give a final concentration of 5 μg/mL.

Instrumentation

The HPLC system consisted of a Shimadzu LC SCL-10AVP solvent delivery unit, an on-line solvent degasser, a gradient mixer and a system controller (Shimadzu Scientific, Columbia, MD, USA), coupled with a CTC-PAL autosampler (LEAP Technologies, Carrboro, NC, USA) for injecting samples. The mass spectrometer was API 4000 LC-MS/MS system (AB SCIEX, Framingham, MA, USA) equipped with a Turbo V IonSpray ionization source. Analyst software version 1.4.1 from AB SCIEX was used for data acquisition and processing; 1500 ShaQer from SPEX SamplePrep (Metuchen, NJ, USA) was used for mixing samples in the QuEChERS extraction procedure.

LC conditions

The chromatographic separation was performed on a YMC ODS-AQ column (100 mm × 2.1 mm, 3 μm particle size; YMC, Allentown, PA, USA) maintaining at the ambient temperature with YMC ODS-AQ guard cartridge (20 mm × 2.0 mm, 3 μm particle size) along with a pre-column inline filter (0.5 μm; Sigma-Aldrich, St. Louis, MO, USA). The CTC-PAL Leap cooling unit was set at 4 °C, and the sample injection volume was 10 μL. The mobile phases, consisted of water with 5-mm ammonium formate and 0.1 % formic acid as mobile phase A and acetonitrile/water (95:5, v/v) with 5-mm ammonium formate and 0.1 % formic acid as mobile phase B, was run at a flow rate of 170 μL/min. The mobile phase gradient was as follows: 0 % B for 1.3 min; linear increased to 100 % B from 1.3.0 to 2.3 min, and then maintained at 100 % B from 2.3 to 7.5 min, went back to 0 % B from 7.5 to 8.0 min and maintained at this proportion from 8.0 to 11.00 min.

Mass spectrometer conditions

The electrospray probe was operated in the positive ion mode. Ultra-pure nitrogen (N2, 99.995 %; Airgas, Radnor, PA, USA) was used as the nebulizer, curtain, and collision gas. The optimum operating conditions of the electrospray ionization (ESI) were as follows: nebulizing gas (GS1), turbo gas (GS2), curtain gas (CUR), collision activated dissociation gas (CAD), turbo-spray voltage, and turbo temperature were set to 38, 36, 20, 12 psi, +4.5 kV, and 480 °C, respectively. All analytes were quantified using the multiple reaction-monitoring (MRM) mode. The mass spectrometer was operated in unit resolution for both Q1 and Q3 in the MRM mode, with a dwell time of 110 ms per MRM channel with a 5 ms pause between scans. The MS/MS settings used in this method are listed in Table 1.
Table 1

Optimized LC-MS/MS acquisition parameters and retention times for each neonicotinoid insecticide

Compounds

Retention time (min)

Precursor ion (m/z)

Product ion (m/z)

DP (V)

CE (V)

Dinotefuran

5.14

203.2

113.2

41

17

129.2a

35

17

Nitenpyram

5.25

270.9

56.1

42

49

225.2a

38

17

Thiamethoxam

5.46

292.0

132.0

30

27

211.1a

28

17

Flonicamid

5.51

230.2

174.2

67

27

203.0a

67

25

Clothianidin

5.63

250.0

132.0

38

21

169.2a

35

19

Imidacloprid

5.70

256.2

175.1

35

25

209.1a

37

21

Acetamiprid

5.76

223.1

56.1

52

30

126.1a

50

27

Thiacloprid

5.94

253.1

90.1

50

55

126.1a

52

28

Imidacloprid-d4b

5.70

260.1

212.9

46

21

Clothianidine-d3c

5.63

253.1

172.1

51

19

Thiamethoxam-d3d

5.46

294.9

214.0

56

17

RT retention time, DP declustering potential, CE collision energy, DP declustering potential, CE collision energy

aProduct ion used for quantification.

bInternal standard for imidacloprid, acetamiprid, thiacloprid, and flonicamid

cInternal standard for clothianidin

dInternal standard for thiamethoxam, dinotefuran, and nitenpyram

The peak areas of analytes and their IS's were determined using analyst version 1.4.1. Identification of each analyte was based on its retention time and mass spectrum. Additionally, the stable isotope ISs served as a tool for the confirmation of the analytes in unknown samples by providing a chromatographic reference for each peak selection. For each analytical batch, a calibration curve containing eight neonicotinoids with slope, intercept, and correlation coefficient (r) was derived from weighted (1/x) linear least squares regression of the peak area ratio (analyte: IS) versus the concentration of the standards. The regression equation from the calibration curve was used to back-calculate the measured concentration of each standard and quality control (QC). The standard calibration was injected before and after all samples, including blanks, QCs, and unknowns, to monitor sensitivity changes.

Preparation of calibration and QC samples

A prior analysis of neonicotinoid-free pollen and HFCS samples, which were used as blank matrix samples, was performed and confirmed to have no contamination of neonicotinoids. These blank pollen and HFCS samples were used for calibration, while QCs and blanks for the analysis and validation. The calibration curves for 8 neonicotinoids at seven levels ranged from 0.1 to 100 ng/g for pollen and 0.1 to 50 ng/g for HFCS were prepared by adding aliquots of intermediate standard solutions to blank pollen or HFCS matrix. The QC samples at two concentration levels, 5 (QCL) and 50 ng/g (QCH) for pollen, and 2 (QCL) and 40 ng/g (QCH) for HFCS were prepared the same way in blank pollen or HFCS matrix. The standards and QCs were stored at −20 °C.

Sample preparation

Pollen samples

Blank pollens from honeybee hives at Worcester Massachusetts and blank HFCS of the year 2012 which were tested and found on contamination were used for method developments and validations. Pollen was manually grinded into fine powder and mixed well using a mortar and pestle. As shown in Fig. 1, 2 g of ground pollen was weighed into a 50-mL centrifuge tube into which 8 mL of water was added. The mixture was then shaken to dissolve until a homogeneous solution was obtained. Subsequently, 10 mL of acetonitrile, 3 mL of n-hexane, and 10 μL of intermediate isotope-labeled IS solution (imidacloprid-d4, clothianidin-d3, and thiamethoxam-d3) at 5 μg/mL (final 25 ng/g) were added to each tube. For calibration standards, QCL and QCH, blank pollen matrix was fortified with 10 μL of appropriate level of working standard solution. Double blanks and blanks were also prepared in parallel with and without IS added, respectively. The tube was then shaken vigorously for 30 seconds in ShaQer at 1500 stroks per minute. Next, one QuERChERS salt pack and one ceramic homogenizer (Agilent Technologies, Santa Clara, CA) were added to the suspension. Then, the tube was immediately shaken vigorously for 40 s in the shaker, and centrifuged for 4 min at 4,000×g. Afterwards, the top hexane layer was removed and 1 mL from the acetonitrile layer was transferred to a 2-mL QuEChERS dispersive SPE vial, containing 50 mg PSA, 50 mg C18, and 150 mg MgSO4. The vial was vortexed for 30 s and then centrifuged for 5 min at 13,500 rpm. Six hundred microliters of supernatant was transferred into a glass test tube and dried under gentle N2 stream using a TruboVap LV Evaporator (Zymark, Framingham, MA) at 40 °C. The dried residue was reconstituted in 200 μL of acetonitrile/water (15:85, v/v), and then filtered through a 1.5-mL vials (0.2-μm nylon filter) followed by centrifuging at 5,000 rpm for 3 min. One hundred fifty microliters of supernatant was transferred to an auto-sampler vial, and 10 μL was injected into the LC-MS/MS for analysis. Another final step was evaluated: 100 μL of supernatant was diluted with 100 μL (or 400 or 800 μL) of water, vortexed, and then centrifuged at 5,000 rpm for 3 min. The rest of the steps were the same as above.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-013-7338-7/MediaObjects/216_2013_7338_Fig1_HTML.gif
Fig. 1

Sample preparation procedures in pollen and HFCS

HFCS samples

Five grams of HFCS was weighted into a 50-mL centrifuge tube, in which 10 mL of water were added, and then shaken in a shaking water bath at 50 °C for 20 min until a homogeneous solution was obtained. Subsequently, 10 mL of acetonitrile, 20 μL of intermediate isotope-labeled IS solution (imidacloprid-d4, clothianidin-d3, and thiamethoxam-d3) at 5 μg/mL were added to the tube. The rest of the sample preparation procedures were the same as those for pollen except for no adding hexane (Fig. 1).

Assay validation

The analytical method was validated in order to demonstrate the recovery, sensitivity, accuracy, and precision of the measurements. All calibration samples were prepared as described above using blank pollen (or HFCS) matrix fortified with respective amounts of reference standards.

Prior to the sample analysis, a standard mixture of the targeted analytes was injected to LC-MS/MS to check the operational conditions. The linearity of the matrix-matched calibration curves was evaluated from three consecutively prepared batches at the concentration range of 0.1–100 ng/mL for pollen and 0.1–50 ng/mL for HFCS. The peak area ratios of analytes and the corresponding IS were used to calculate the correlation coefficients, intercepts, and slopes.

Fifteen replicates of QC samples generated at concentrations of 5 and 50 ng/g for pollen and 2 and 40 ng/g for HFCS from runs on three consecutive days were quantified using matrix-matched calibration curves to evaluate precision and accuracy at each concentration level to determine the intra- and interday validation. Precision was calculated as the relative standard deviation (RSD in percentage) for both intra- and interday variability. Accuracy describes the degree of closeness of the determination value to the true value and was evaluated by the recovery, which is the mean back-calculated concentrations from the matrix-matched calibration curve for each level of concentration divided by the theoretical fortified concentration. Acceptable intra- and interday precision was defined by a RSD <20 % per level. Accuracy was acceptable if the mean recovery is within 70–120 % [30]. The LOD were calculated based three times the signal-to-noise (peak to peak) ratio of the quantitative ion transitions, and the LOQ was calculated as ten times the signal-to-noise ratio of the quantitative ion transition in the matrix.

Matrix effect (ME) occurs when the co-elutes from the same sample matrix affects (either attenuates or enhances) the response of the analyte during quantitation by LC-MS/MS. To determine the matrix effect of pollen (or HFCS) on the ESI process, blank pollen (or HFCS) samples were extracted, dried, reconstituted and fortified with neonicotinoids standard solutions at QCL and QCH (post-extraction fortified sample), then compared with the same concentration level of neonicotinoid standard solutions in neat standard solutions injected directly into LC-MS/MS using the following equation:
$$ ME\%=\left(\mathrm{Mean}\;\mathrm{post}-\mathrm{extraction}\;\mathrm{peak}\;\mathrm{area}/\mathrm{Mean}\;\mathrm{neat}\;\mathrm{solution}\;\mathrm{peak}\;\mathrm{area}\right)\times 100 $$

The values of ME indicate either ion enhancement (>100 %), ion suppression (<100 %), or no matrix effect if a value of 100 % was obtained.

The total process recovery was calculated by dividing the peak area counts of individual extracted pollen (or HFCS) samples fortified with standard solutions by the mean area counts of neat standard solutions injected directly into LC-MS/MS.

Stability of standard stock solutions of neonicotinoids and ISs in acetonitrile used in the preparation of standards and QCs was investigated at both room temperature and −20 °C. The stabilities were calculated by comparing mean response ratios (area of response per unit of concentration) of stability solutions to mean response ratios of freshly prepared control solutions. The stability study was further evaluated under the conditions the samples were likely to encounter during the analytical process. The stability of extracted samples at QCL and QCH in six replicates in reconstitution solution at 4 °C (the temperature of the autosampler) for 2 days was evaluated. The mean analyte concentration at each level was compared with each mean concentration determined in the initial testing. The analytes were considered to be stable if degradation was <10 % of the concentration at day 0.

Determination the concentrations of 8 neonicotoids in pollen and HFCS samples

To demonstrate the suitability of this validated method in the routine analysis of neonicotinoids in pollen and HFCS samples, we analyzed 13 pollen samples from different honeybee hives in the central Massachusetts area and eight archived HFCS samples obtained from different sources between 2005 and 2009 using this LC-MS/MS method.

Results and discussion

Optimization of extraction conditions

Sample preparation is an important process that can selectively isolate the analytes of interest from the matrix and minimize or eliminate matrix components in the processed sample. Although both solid phase extraction and liquid-liquid extraction have been used in sample preparation for the determination of neonicotinoids in pollen, the QuEChERS method is clearly a better choice for its ease, low cost and increased sample throughput. The QuEChERS method is based on the extraction of pesticides from the sample by acetonitrile, partitioning with a mixture of sodium buffer and magnesium sulfate, followed by a dispersive-SPE step to remove residual water and clean up the samples. This methodology has been adapted in a few analytical methods in quantification of neonicotinoids in pollen [2729]. The sample extraction in this method development was carried out following the buffered QuEChERS procedures [33].

The first step of this procedure involved shaking extraction with acetonitrile and hexane. Acidified acetonitrile has been reported in this step [28], we chose to use acetonitrile without acid to increase the extraction efficiency since neonicotinoids would be protonated under acidic pH. Although lipids are important components in pollen [34], only one reported method using QuEChERS methodology for analyzing neonicotinoids in pollen included the step to remove lipids from acetonitrile extract [29]. Hexane is a very non-polar solvent with lower density than acetonitrile, therefore it can dissolve lipids in the acetonitrile extract and be easily removed after centrifugation. At the same time, neonicotinoids extracted into acetonitrile were not likely to be dissolved in hexane as neonicotinoids are polar compounds with a solubility of ≥100 mg/L in water [35]. Our preliminary experiments showed that adding hexane in this step of extraction resulted in cleaner extracts, lower background of chromatograms of blank pollen samples, and improved method sensitivity as compared with extraction without hexane. Two volumes of hexane (3 and 5 mL) were compared, and the differences in recoveries between them were within 7 % with 3 mL hexane showing slightly higher recoveries over 5 mL for most of the neonicotinoids. Therefore, we chose to use 3 mL of hexane. To ensure the extraction efficiency, we shaken the sample with acetonitrile and hexane first, then adding in the QuEChERS salt package, instead of doing extraction/partition as one step as previously reported. In QuEChERS extraction salt package, anhydrous magnesium sulfate served to partition water from the sample, and sodium chloride served to help to reduce polar co-extractives, such as sugars in pollen [36]. We chose citrate buffer (pH ∼8) instead of acetate buffer (pH ∼5) to enhance the performance of extraction and also provide a more optimized pH for d-SPE sorbent since neonicotinoids contain weak base or strong base groups. To further improve the shaking extraction potency and efficiency, we added a ceramic homogenizer during this shaking partitioning/extraction step. The homogenizer helped to break up salt agglomerates, further reduced extraction time, and ensured high extraction repeatability.

Unlike HFCS, the amount of pollen collected is often limited; therefore, we tried to develop a method using the lowest amount of pollen possible without sacrificing the sensitivity. Among different weights of pollen that we evaluated (2 g with 8 mL of water, 3 g with 10 mL of water, and 5 g with 12 mL of water), 2 g of pollen showed cleaner extracts, lower chromatogram backgrounds, and better sensitivity than the others. Therefore, we chose 2 g pollen in the extraction.

The sample cleanup steps in the previous QuEChERS methods reported are either not sufficient enough for the complex composition of pollen or time consuming [27, 29]. Although lipids should be much more soluble in hexane than in acetonitrile, a small amount of lipid, along with some polar components (e.g., fatty acid and sugar) could still be co-extracted from the samples into acetonitrile extracts. Therefore, not only PSA is necessary to remove acidic components (e.g., fatty acids) and sugar in pollen, but also C18 or C18 and GCB is need to absorb fatty components. Two types of d-SPE sorbents, PSA + C18 and PSA + C18 + GCB, were investigated. We found that either type of d-SPE sorbent is sufficient to clean up the extract with similar good recoveries and difference less than 6 % at both QCL and QCH levels. We chose d-SPE with PSA + C18 sorbent for the cleanup step because it was less expensive than the sorbent with GCB.

HFCS is widely used as a liquid sweetener consisting of different percentages of fructose and glucose and some oligosaccharides as minor components [37]. We used the similar QuEChERS extraction procedure without hexane for HFCS sample preparation. Two types of d-SPE sorbents of PSA and PSA + C18 were compared at the QCL and QCH levels. As shown in Fig. 2, PSA + C18 sorbents resulted in more consistent recoveries at both levels of QCs with less variation between triplicated samples (RSD ≤ 9 %) than PSA sorbents (RSD ≤ 16 %). C18 also did not add much extra cost to the sample preparation, therefore we chose d-SPE with PSA + C18 sorbent to clean HFCS extracts.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-013-7338-7/MediaObjects/216_2013_7338_Fig2_HTML.gif
Fig. 2

Recoveries of neonicotinoids fortified at 2 ng/g and 40 ng/g in blank HFCS matrices when the extracts were cleaned up by d-SPE with PSA or PSA + C18 as sorbents during the QueChERS procedure. Error bars depict the standard deviation (n = 3)

The final sample preparation step included reconstituting the extracts dried under N2 with the solution containing acetonitrile/water (15:85, v/v). The benefit of this step is that it results in cleaner extracts since the co-extractives that are insoluble in the reconstitute solution would be left behind after centrifuge and filtration.

Optimization of LC-MS conditions

The positive ion ESI MS/MS product ion spectra for each compound and IS were obtained by direct infusion of each analyte solution via a tee connection between the LC and the mass spectrometer. The protonated molecular ion [M+H]+ was selected as precursor ion for each compound, and two product ions generated from [M+H]+ of each analyte were selected based on ion abundance. After the mass spectrometer operating parameters were optimized for each analyte, the MRM method was built up using the most abundant and stable product ion of each analyte for quantification and the second abundant ion for confirmation. The optimized LC-MS/MS parameters for each compound are summarized in Table 1.

After the detection conditions were optimized, experiments were conducted to optimize the chromatographic conditions. We investigated several reversed phase columns, specifically the retention times, analyte responses, peak shapes, resolutions, and background interferences of these columns were compared using the same mobile phase with optimized gradients. Given that dinotefuran and nitenpyran have higher water solubility than the other six neonicotinoids, dinotefuran was eluted early (<1.5 min) as a broad peak on all columns tested except YMC ODS-AQ and the nitenpyram peak also showed a front tailing on some columns. Conventional C18 columns (such as YMC pro-C18, Water's XTerra MS C18, and Thermo Betasil C18 were tested) with higher density coverage by C18 chains than YMC ODS-AQ lose their lipophilic properties and ability to retain polar compounds in high aqueous content mobile phase because of the “chain folding” effect. YMC ODS-AQ is a unique C18 reversed-phase silica-based column with a relatively hydrophilic surface, therefore it can maintain its brush-like C18 structure and retain polar compounds in 100 % aqueous eluents. Because of its balanced hydrophilic and lipophilic nature, YMC ODS-AQ has strong retentions of all eight neonicotinoids compounds, including dinotefuran and nitenpyram. Overall, YMC ODS-AQ (100 mm × 2.1 mm, 3 μm) showed superior peak shape with adequate retention time for each analyte compared with other columns tested and was therefore chosen as the analytical column for this method.

The composition of the HPLC mobile phase was also optimized. Formic acid added in the mobile phase promoted the protonation of the analytes in positive ESI conditions, leading to better sensitivity with adequate retention time for each analyte. Mobile phase containing 0.1 % formic acid and 0.1 % formic acid with 5 mM ammonium formate were compared. Figure 3 shows 5 mM ammonium formate as an additive to the mobile phase significantly enhanced the response values for each analyte. All analytes also showed sharper peaks when 5 mM ammonium formate was added in aqueous phase, therefore we chose to add 0.1 % formic acid with 5 mM ammonium formate in the aqueous and organic mobile phase. The HPLC elution gradients and the flow rate were also optimized. Figure 4 shows this elution gradient can separate analytes from potential interferences with low background noise and give the robust chromatographic peaks. The total run time for this LC-MS/MS method was 11 min.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-013-7338-7/MediaObjects/216_2013_7338_Fig3_HTML.gif
Fig. 3

Responses of eight neonicotinoids using different mobile phase: (1) 0.1 % (v/v) formic acid in aqueous mobile phase A and organic mobile phase B; (2) 0.1 % (v/v) formic acid with 5 mM ammonium formate in aqueous mobile phase A and organic mobile phase B

https://static-content.springer.com/image/art%3A10.1007%2Fs00216-013-7338-7/MediaObjects/216_2013_7338_Fig4a_HTML.gifhttps://static-content.springer.com/image/art%3A10.1007%2Fs00216-013-7338-7/MediaObjects/216_2013_7338_Fig4b_HTML.gif
Fig. 4

Representative LC-MS/MS chromatograms of a blank pollen matrix fortified with eight neonicotinoids standards at 5 ng/g (A), a blank pollen matrix (B), and at LOQs (C)

We chose imidacloprid-d4, clothianidin-d3, and thiamethoxam-d3, three stable isotopic-labeled standards, as the IS and added them to all samples at the beginning of the sample preparation. Since IS has the same physical and chemical properties as the unlabeled analyte, the addition of IS can reduce variations in sample preparation and small volume fluctuations in injection, minimize matrix effects on quantitation, and also improve peak identification by providing a chromatographic reference (retention time reference) for peak selection of the analytes, especially at trace level analysis.

Assay validation

This LC-MS/MS method was validated in terms of selectivity, recovery, linearity, intra- and inter-day precision, and sensitivity.

The absence of matrix interference at the retention time of each analyte in ten blank pollens and HFCS matrix samples sufficiently demonstrated the selectivity and specificity of this method. This also means that these blank matrix samples were free of neonicotinoid residues. Method validation and matrix-matched calibrations were subsequently generated in these blank pollen and HFCS matrices.

Quantificaiton of low levels of neonicotinoids by LC-MS/MS in complex matrices, such as pollen and HFCS, is often challenging due to the potential matrix effect caused by the competition between the analytes and co-eluting compounds for ionization in ESI source. In order to obtain satisfactory extraction efficiency, and reliable and sensitive analytical results, matrix effect and recovery were investigated first in pollen matrix. The absolute recoveries of two approaches of the last step of sample preparation in QuEChERS method, dilution and drying followed by reconstitution before injection into the LC-MS/MS system, were compared. After the last step of QuEChERS extraction, the compounds were extracted into acetonitrile. Ideally, the injection sample should be prepared in the mobile phase, because injecting strong solvent (such as pure acetonitrile or methanol) into LC-MS/MS can cause sample precipitation on or before the column and also induce chromatographic issues, such as peak tailing. Therefore, the injection samples were prepared by diluting acetonitrile extracts with water to reduce organic content in injection solvent. The performance of such dilution was investigated by adding different volume of water to the acetonitrile extract (1:1, 1:4, and 1:8, v/v). Table 2 shows that the mean total process recoveries of all neonicotinoids were between 85 and 114 % when the acetonitrile extracts were diluted with water in 1/4 and 1/8 ratios. The recoveries were still in the same range for flonicamid, imidacloprid, acetamiprid, and thiacloprid when the acetonitrile extracts were diluted with same volume of water (1:1, v/v), but reduced to 72 to 80 % for dinotefuran, nitenpyram, thiamethoxam, and clothianidin.
Table 2

The mean total process recoveries (% RE) and matrix effects (% ME) of eight neonicotinoids in fortified pollen and high-fructose corn syrup (HFCS) (n = 5)

 

Dilutiona

Dinotefuran

Nitenpyram

Thiamethoxam

Flonicamid

Clothianidin

Imidacloprid

Acetamiprid

Thiacloprid

Pollen

 Mean REa% ± SD

  5 ng/g

0

60 ± 10

38 ± 4

52 ± 8

86 ± 10

66 ± 11

102 ± 7

99 ± 5

107 ± 10

1/1

78 ± 7

76 ± 11

78 ± 10

84 ± 16

78 ± 12

105 ± 7

104 ± 13

113 ± 8

1/4

87 ± 14

108 ± 17

110 ± 18

96 ± 16

109 ± 11

106 ± 5

105 ± 9

114 ± 7

1/8

94 ± 7

103 ± 12

111 ± 6

90 ± 14

103 ± 17

95 ± 10

108 ± 11

105 ± 12

  50 ng/g

0

58 ± 12

37 ± 4

56 ± 14

83 ± 10

70 ± 4

102 ± 15

102 ± 5

113 ± 10

1/1

80 ± 8

77 ± 5

72 ± 9

90 ± 14

73 ± 11

96 ± 4

105 ± 7

106 ± 7

1/4

85 ± 7

101 ± 12

108 ± 11

96 ± 18

93 ± 11

108 ± 19

103 ± 18

99 ± 19

1/8

89 ± 16

92 ± 18

96 ± 10

102 ± 20

92 ± 12

109 ± 8

98 ± 16

104 ± 10

 Mean ME% ± SD

  5 ng/g

0

66 ± 1

43 ± 2

59 ± 2

89 ± 4

73 ± 1

106 ± 3

105 ± 2

113 ± 1

  50 ng/g

0

68 ± 1

45 ± 2

60 ± 1

90 ± 4

74 ± 1

101 ± 3

109 ± 3

111 ± 3

 Mean accuracyb% ± SD

  5 ng/g

0

90 ± 3

102 ± 2

104 ± 3

93 ± 6

97 ± 5

94 ± 3

104 ± 2

105 ± 4

  50 ng/g

0

92 ± 5

105 ± 3

102 ± 3

96 ± 5

101 ± 4

99 ± 2

101 ± 4

103 ± 4

HFCS

 Mean ME% ± SD

  2 ng/g

0

88 ± 5

102 ± 9

101 ± 10

103 ± 9

101 ± 7

105 ± 7

104 ± 8

103 ± 9

  40 ng/g

0

93 ± 6

92 ± 7

93 ± 5

102 ± 8

103 ± 7

102 ± 8

97 ± 8

102 ± 4

 Mean accuracyb% ± SD

  2 ng/g

0

103 ± 3

91 ± 8

104 ± 6

95 ± 4

92 ± 3

97 ± 5

101 ± 2

104 ± 4

  40 ng/g

0

97 ± 2

92 ± 3

102 ± 4

98 ± 5

94 ± 2

97 ± 1

100 ± 1

99 ± 2

SD standard deviation

aRecovery for the QuEChERS extract diluted with different ratio of water (1/1, 1/4, and 1/8 (v/v)) or without dilution which means it was dried under N2 and then reconstitute with 15 % acentonitrile in water (B) (n = 5)

bAccuracy was evaluated by the recovery for the QuEChERS extract without dilution calculated from matrix matched calibration curve (n = 5)

We further investigated the mean total process recoveries of the second approach by evaporating acetonitrile extract to dryness, followed by reconstituting the residue in 15 % acetonitrile in water. Consistently good recoveries were obtained for flonicamid, imidacloprid, acetamiprid, and thiacloprid (83 to 113 %) in this approach, but not for dinotefuran, nitenpyram, thiamethoxam, and clothianidin (37 to 70 %). Matrix effects of the dried and reconstitute pollen samples in this second approach were evaluated. The results (Table 2) show that the matrix effect caused by signal suppression was observed for thiamethoxam and nitenpyram (43–60 %), and with less extent for dinotefuran and clothianidin (66–74 %) in pollen matrix. No matrix effect was observed for flonicamid, imidacloprid, and acetamiprid. Minor ion enhancement, which may come from improved ionization, was observed for thiacloprid with ≤13 % of the average intensity of the analyte ions enhanced at each level. As the mean total process recoveries for dinotefuran, nitenpyram, thiamethoxam, and clothianidin were close to their corresponding values of matrix effect, it suggests that the loss in their recoveries were mainly due to the matrix effect during ESI, instead of during the QuEChERS procedure.

The quantification based on the analyte/IS response ratios should not be influenced by the matrix effect and also give a measure of control on variability. This is because the deuterated IS should have the same or similar extraction recovery, the same degree of ionization response (including ion suppression, ion enhancement) in ESI mass spectrometry, and the same variations during sample preparation and injections process as the analytes. However, not each neonicotinoid has deuterated IS that is commercially available. The mean total process recoveries for three deuterated ISs thiamethoxam-d3, clothianidin-d3, and imidacloprid-d4 that are commercially available were 59, 68, and 99 % respectively. Based on these results of matrix effects and the best recoveries calculated from analyte/IS response ratios can be obtained, imidacloprid-d4 were chosen as ISs for the quantitation of imidacloprid, acetamiprid, flonicamid and thiacloprid, clothianidin-d3 for clothianidin, and thiamethoxam-d3 for thiamethoxam, dinotefuran and nitenpyran, respectively. Therefore, the recoveries using internal standardization by calculating the ratio of peak area ratio of analyte/IS eliminated the matrix effects, and brought the recovery values close to 100 % with RSD < 7 % for each analyte except nitenpyram, which were around 50 %. However, when the recoveries in pollen matrices were obtained by back calculating from matrix-matched calibration curve, the corresponding accuracies were 90–105 % (Table 2). Table 2 also shows that the accuracies for HFCS obtained by back calculating from matrix-matched calibration curve were between 91 and 104 %. These results fulfilled the requirements for mean recovery in the range of 70 to 120 % with relative standard deviation (RSD) ≤ 20 % [31].

Both matrix effects and dilution of injection solutions could negatively affect the sensitivity of the method. Although matrix-matched calibration and IS can eliminate or minimize the negative impacts of matrix effects on quantitation, they would not improve the method sensitivity. Therefore, we further investigated the influence of matrix effect (from the drying approach) and dilution on the sensitivities in pollen fortified with neonicotinoids. Table 3 shows that drying approach gave us better sensitivities for dinotefuran, nitenpyram, clothianidin, flonicamid, and thiacloprid with LOQ ranging from 0.1 to 0.5 ng/g than dilution approach (LOQs were 0.1–2 ng/g). The LOQs for imidacloprid, acetamiprid, and thiamethoxam in pollen from both preparation procedures were the same (≤0.5 ng/g). The LOQs for all neonicotinoids in HFCS were 0.1 ng/g, except 0.3 ng/g for nitenpyram. Comparing the two sample preparation approaches: dilution and drying the extracts followed by reconstitution, the dilution approach resulted in better recoveries and less matrix effects for all eight neonicotinoids, but with less sensitivity due to the broad or irregular peaks at low concentrations in the dilution samples. Although the drying approach could reduce recoveries (mainly due to matrix effects), the concentration step compensated the signal suppression from matrix effect, and would actually increase the sensitivity for dinotefuran, nitenpyram, clothianidin, flonicamid, and thiacloprid. Considering that improving sensitivity was our main goal, we chose to use the drying approach for pollen and HFCS sample preparation.
Table 3

The comparison of limits of detection (LOD) and limits of quantification (LOQ) from two procedures of the final step of the QuEChERS extraction in spiked blank pollen and high-fructose corn syrup (HFCS) samples (n = 5)

Compounds

Pollen

HFCS

1/1 dilutiona

Dryb

Dryb

LOD (ng/g)

LOQ (ng/g)

LOD (ng/g)

LOQ (ng/g)

LOD (ng/g)

LOQ (ng/g)

Dinotefuran

0.3

1.0

0.03

0.1

0.03

0.1

Nitenpyram

0.5

2.0

0.1

0.5

0.1

0.3

Thiamethoxam

0.05

0.1

0.03

0.1

0.03

0.1

Flonicamid

1.0

2.0

0.1

0.5

0.03

0.1

Clothianidin

0.1

0.5

0.03

0.1

0.03

0.1

Imidacloprid

0.03

0.1

0.03

0.1

0.03

0.1

Acetamiprid

0.03

0.1

0.03

0.1

0.03

0.1

Thiacloprid

0.1

0.5

0.03

0.1

0.03

0.1

aThe acetonitrile extract from QuEChERS was diluted with water (1:1, v/v)

bThe acetonitrile extract from QuEChERS was dried under N2 at 40 °C and reconstituted with 15 % acetonitrile in water

The obvious matrix effects in pollen for selected neonicotinoids in the drying approach demonstrated the importance of using matrix-matched calibration for quantification to counteract the matrix effect. In addition, the use of ISs can improve accuracy through different steps of the analytical method. Therefore, calibrations were prepared using isotope labeled ISs accompanied by matrix-matched solutions ranging from 0.1 to 100 ng/g for pollen and 0.1 to 50 ng/g for HFCS. The matrix-matched calibration generated in solutions with the exact same composition as the samples adequately counteracted the matrix effect in samples. The calibration curves in both pollen and HFCS showed good linearity within the given concentration ranges, with the correlation coefficient (r) exceeding 0.993.

We used twelve replicates of LOQ samples (0.1 ng/g for each of the neonicotinoids in pollen and HFCS, except 0.5 ng/g for nitenpyram and flonicamid in pollen and 0.3 ng/g for nitenpyram in HFCS) in three separate runs to evaluate the precision and accuracy at the lower end of the assay. As shown in Table 4, the precisions were 4–16 % for pollen and 2–11 % for HFCS, and the accuracies were 78–120 % for pollen and 68–115 % for HFCS. Those results not only met EU regulation on recovery 60–120 % with RSD of ≤30 % at the fortified level ≤10 ng/g [30], but also demonstrated acceptable precision and accuracy at LOQ levels for all analytes. The current method with LODs 0.03 ng/g for all neonicotinoids in 2 g of pollen (except 0.1 ng/g for nitenpyram and flonicamid) were more sensitive than LC-MS/MS methods currently reported in analyzing neonicotinoids in pollen, which ranging from 4.3 ng/g using 2–3 g of pollen to 0.2 ng/g using 15 g of pollen [6, 2629]. The high sensitivity made this method advantageous as compared with current analytical methods reported on measuring the trace amount of neonicotinoids in pollen or HFCS.
Table 4

Intra- and interday precision and accuracy for eight neonicotinoids spiked in pollen and high-fructose corn syrup (HFCS) matrix

Compounds

Analysis

Pollen

HFCS

Intraday (ng/g)

Interday (ng/g)

Intraday (ng/g)

Interday (ng/g)

LOQa (n = 4)

5 (n = 5)

50 (n = 5)

LOQa (n = 12)

5 (n = 15)

50 (n = 15)

LOQb (n = 4)

2 (n = 5)

40 (n = 5)

LOQb (n = 12)

2 (n = 15)

40 (n = 15)

Dinotefuran

RE (%)

78

100

97

85

108

96

76

103

100

84

104

102

RSD (%)

13

4

9

11

10

9

10

9

4

11

10

13

Nitenpyram

RE (%)

120

106

104

111

98

104

115

98

101

104

97

107

RSD (%)

8

3

4

9

6

8

9

5

10

7

9

6

Thiamethoxam

RE (%)

113

101

98

110

101

98

103

100

103

99

104

104

RSD (%)

9

5

3

11

4

6

2

5

8

8

3

6

Flonicamid

RE (%)

89

93

107

93

96

99

68

102

103

78

98

103

RSD (%)

12

3

5

16

5

7

11

10

8

10

8

6

Clothianidin

RE (%)

103

104

100

105

99

95

108

92

102

106

100

105

RSD (%)

13

5

5

10

3

7

9

6

1

11

8

3

RE (%)

91

100

102

94

99

105

87

93

101

92

100

106

Imidacloprid

RSD (%)

10

3

1

8

3

6

4

12

2

6

8

5

RE (%)

102

96

99

108

103

99

95

95

103

89

104

105

Acetamiprid

RSD (%)

10

5

5

7

3

3

5

8

2

9

8

9

RE (%)

103

99

99

100

101

98

93

94

95

104

98

98

Thiacloprid

RSD (%)

7

5

4

4

8

6

10

7

8

2

11

6

RE recovery, SD standard deviation

aLOQ: 0.1 ng/g for each of the neonicotinoids in pollen, except 0.5 ng/g for nitenpyram and flonicamid

bLOQ: 0.1 ng/g for each of the neonicotinoids in HFCS, except 0.3 ng/g for nitenpyram

The intra- and interday accuracy and precision were evaluated by assessing recovery and reproducibility in blank pollen and HFCS fortified with neonicotiniods at concentrations of 5 and 50 and 2 and 40 ng/g, respectively. The recoveries were calculated using matrix-matched calibration standards at concentration levels contained the fortifying level of the analytes. Results shown in Table 4 indicated good intra- and interday precisions (1 to 13 %) and accuracies (92–108 %) of this LC-MS/MS method.

The stabilities of the standard stock solutions at room temperature for at least 6 h and −20 °C for at least 90 days were established with differences between the stored and fresh solutions were ≤5 %. The stability of the extracted samples in reconstituted solution at the QCL and QCH levels in the auto-sampler were found stable at 4 °C at least for 24 h. These results showed that acceptable stability for stock solutions during the storage period, and extracted samples within the time required for analyzing the samples. Therefore, neonicotinoids had acceptable stability under the test conditions.

Application of the LC-MS/MS method

We analyzed 13 pollen samples collected from honeybee hives located in the central Massachusetts area in 2012. Among the eight neonicoinoids tested, only imidacloprid and thiamethoxam were found in several pollen samples, as shown in Table 5. The frequency of detection for imidacloprid and thiamethoxam in pollen samples were 85 and 15 %, respectively, with the concentrations ranging from 0.2 to 2.2 ng/g for imidacloprid and 0.1–0.3 ng/g for thiamethoxam. The levels of imidacloprid residue in pollen from our study were similar to those reported in the literature [7, 28, 31, 38]. However, the levels of thiamethoxam from our study were much lower than the reported levels (53.3 ng/g) but with higher detection frequency (15 vs 0.29 %) [27]. The LC/MS-MS method was also applied to the analysis of eight HFSC archived samples collected from 2005 to 2009. No detectable levels of neonicotinoids were found in HFCS samples tested.
Table 5

Concentrations of eight neonicotinoids detected in pollen and high-fructose corn syrup (HFCS) samples

Compounds

Pollen (n = 13)

HFCS (n = 8)

Concentration range (ng/g)

No. positive samples

Concentration range (ng/g)

No. positive samples

Dinotefuran

<LOQ

0

<LOQ

0

Nitenpyram

<LOQ

0

<LOQ

0

Thiamethoxam

<LOQ

0

<LOQ

0

Flonicamid

<LOQ

0

<LOQ

0

Clothianidin

<LOQ

0

<LOQ

0

Imidacloprid

0.2–2.2

10

<LOQ

0

Acetamiprid

<LOQ

0

<LOQ

0

Thiacloprid

0.1–0.3

2

<LOQ

0

<LOQ below limit of quantitation

Conclusions

We have developed an LC-MS/MS method coupled with the QuEChERS procedures for simultaneously analyzing eight neonicotinoids residues in pollen and HFCS samples. These eight compounds comprise the entire group of neonicotinoid insecticides with the regulation of European Union (EU) MRL for pesticide residues on fruits, vegetables and other agriculture products. The method has been validated and proven to be sensitive, accurate, and precise using 2 g of pollen or 5 g of HFCS. The method has been successfully applied to analyze pollen and HFCS samples collected for research purposes. These results have demonstrated the value of this method in routine monitoring of neonitocinoid residues in pollen and HFCS.

Acknowledgments

This project was supported by the Harvard-NIEHS Center for Environmental Health (ES000002) Pilot Project Program. The authors would like to thank Dr. Christine Austin (from the University of Sydney, Sydney, Australia) for her assistance in the method development and Ms. Michaela Kapp (at the Harvard School of Public Health) for her assistance in the sample collections as well as in the preparation of this manuscript. The authors declare no conflict of interest.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013