Efficient Identification of Deoxynivalenol in the Global Wheat Food Chain Testing Unprocessed, Processed, and Consumer Wheat Products using Automated Solid Phase Extraction Cleanup and Quick High Performance Liquid Chromatography Analysis Toni Hofhine, Horizon Technology Inc., Salem, NH USA Elizabeth Krantz, Dr. Cheri A. Barta, and Dr. Pamela Doolittle, University of Wisconsin, Madison, WI, USA Robert Buco, Richard Koeritz, and Zachary Lilla, Shimadzu Scientific Instruments, Marlborough, MA, USA Jennifer Claus, Kenneth Espenschied, and Michael Ye, Sigma-Aldrich, Bellefonte, PA, USA Key Words Deoxynivalenol, Mycotoxins, Vomitoxin, Solid Phase Extraction (SPE), Liquid-Liquid Extraction (LLE), Food and Drug Administration (FDA), European Union (EU), Ultra High Performance Liquid Chromatograph (uhplc) Introduction Mycotoxin testing in consumer food products has become increasingly important as global food trade increases, making it necessary to identify mycotoxins efficiently and accurately. Deoxynivalenol, also known as vomitoxin, is a mycotoxin commonly found in wheat feed crops. Because wheat is a highly-used raw agricultural export commodity in many counties for both animal feed and consumer food products, it is an important component in the diets of both humans and animals 1. Deoxynivalenol has the unique ability to withstand high processing temperatures, creating the need to perform food safety testing prior to trade and consumption of unprocessed, processed, and consumer wheat products. Currently, the United States has an advisory limit for deoxynivanol of 1 ppm (or 1000 μg/kg) in finished foodstuffs 2. Current European Union legislation sets the maximum level of deoxynivalenol in foodstuffs at 0.75 ppm (750 μg/kg) for unprocessed cereals marketed for direct consumer consumption and 0.2 ppm (200 μg/kg) for processed cereal-based foods and foods intended for babies and small children 3. The EU also has performance criteria in place for methods testing for mycotoxins in foodstuffs 4. Preparation of wheat samples generally requires liquid-liquid extraction (LLE) and solid phase extraction (SPE) steps to remove the matrix effects and allow deoxynivalenol to be easily detected and quantified by HPLC analysis, with the extraction via SPE using immunoaffinity columns (IAC). A gel sorbent in the IAC uses antibodies to bind the deoxynivalenol and wash away the interfering compounds prior to elution. Newer technologies are available that increase SPE efficiency. An alternative to IAC is to bind the interfering compounds and immediately elute the deoxynivalenol from a non-gel standard solid phase extraction (SPE) cartridges.
The SmartPrep Extractor, using new Supel Tox DON SPE cartridge technology, efficiently prepared a variety of wheat samples representing the agriculture and processed food chain prior to HPLC analysis. Sample matrices representing the food chain (unprocessed cracked wheat, unprocessed wheat berries, processed natural wheat germ, consumer shredded wheat cereal, and consumer infant whole wheat cereal) were selected for this application. Data presented concludes that deoxynivalenol is effectively recovered using automation from a variety of samples with and without natural deoxynivalenol present. Recoveries for all matrices meet both United States and European Union limits. Experimental Wheat samples were chosen to represent processed, semi-processed, lightly processed and commercially processed wheat products. The multi-step procedure used for these five wheat samples, from their packaged form through final analysis is outlined below. Details for each step (pre-preparation, liquid-liquid extraction, solid phase extraction, evaporation, and analysis) are outlined below. Instrumentation P a ge 2 Solvents Purified water Milli-Q Acetonitrile laboratory grade Methanol laboratory grade Standard Sigma-Aldrich Deoxynivalenol solution, 100 μg/ml in acetonitrile Samples purchased from local USA suppliers Wheat Germ Cracked Wheat Wheat Berries Shredded Wheat Cereal Infant Wheat Cereal SPE SmartPrep Extractor automated SPE cartridge system Supelco Supel Tox DON SPE Cartridge, 6 ml Evaporation XcelVap Automated Evaporation/Concentration System Purified compressed nitrogen HPLC Shimadzu Nexera XR uhplc system with UV detection Titan C18 UHPLC Column, 10 cm x 2.1 mm I.D., 1.9 μm particle size Titan C18 HPLC Guard Cartridge, 5 mm x 2.1 mm I.D., 1.9 μm particle size Titan Guard Cartridge Holder SmartPrep Extractor XcelVap Automated Evaporation/Concentration System
Sample Pre-Preparation and Storage A representative sampling of each wheat matrix combined at least two packaged sources (e.g., two boxes of cereal or two bags of wheat germ). Each sample was thoroughly mixed by a combination of grinding and mixing. If the sample was of a bulky consistency, such as the wheat berries, it was ground using ten 3-second pulses in a coffee grinder to a medium grind before being mixed and combined. Samples were stored in plastic reusable storage containers. Wherever possible, plastic or silanized glass materials were used throughout the method. Sample Preparation Liquid-Liquid Extraction Procedure Each prepared wheat sample included a control sample (non-spiked sample) to test for the presence of natural deoxynivalenol that was subtracted from spiked sample percent recoveries. Testing of extraction times was performed, with no difference in recovery found between 30 minutes of shaking and 15 minutes of shaking (Procedure 1). Procedure 1: Liquid-Liquid Extraction Procedure P a ge 3
Spiked samples at the US limits were prepared by pipetting 250 μl of a 100 μg/ml neat solution of deoxynivalenol into the 25 grams of sample, achieving a spike level of 1000 μg/kg, or 0.5 μg/ml (according to the preparation procedure). Spiked samples at the EU limits were prepared by pipetting 50 μl of a 100 μg/ml neat solution of deoxynivalenol into the 25 grams of sample. Images of each wheat sample following the liquid-liquid extraction procedure are shown below. Wheat Berries Cracked Wheat Wheat Germ Shredded Wheat Cereal Infant Wheat Cereal Sample Preparation Automated SmartPrep Extractor SPE Procedure & XcelVap Evaporation Procedure The SmartPrep Extractor automated the solid phase extraction process. From the extracted filtrate, 2 ml was automatically loaded onto the Supel Tox DON SPE cartridge using the SmartPrep Extractor (Procedure 2). Procedure 2: Automated SmartPrep Extractor SPE Procedure P a ge 4
SPE breakthrough tests were performed using a 0.5 µg/ml standard and a spiked shredded wheat sample, with the goal of determining the need to collect the load step. It was found that the breakthrough tests for both the standard and sample contained no deoxynivalenol. Discarding the load step may remove unretained compounds that would appear as extraneous peaks in the final analysis. It is recommended to discard the load step in order to reduce the overall time for evaporation and improve the final analysis step. Following SPE, the collected sample fraction was evaporated to dryness using the XcelVap evaporator. Reconstitution was performed with 1 ml of mobile phase (92:4:4 water:acetonitrile:methanol) and mixed with a transfer pipette. The reconstituted samples were transferred to HPLC plastic or silanized glass vials prior to HPLC injection. HPLC Analysis Conditions Deoxynivalenol samples were injected onto a Shimadzu Nexera XR uhplc system with single wavelength UV detector using a 2 µl injection, which is approximately 25X lower than traditional methods using older column technology. The new uhplc column technology of 1.9 µm particle size increased sensitivity and peak height, allowing for early elution of the deoxynivalenol peak between 2.8 and 4.0 minutes, with retention time consistency achieved within a run. Variation of retention time was due to the preparation of new mobile phase from run to run. Run time length of samples eluted sample matrix peaks prior to subsequent injections. Table 1: HPLC Conditions for Deoxynivalenol Analysis Flow Rate Column Guard Column Column Temperature 55 C Mobile Phase Injection Volume 2 µl Run Time Wavelength HPLC Conditions 0.6 ml/min Titan C18 uhplc Column, 10 cm x 2.1 mm I.D., 1.9 μm particle size Titan C18 HPLC Guard Cartridge, 5 mm x 2.1 mm I.D., 1.9 μm particle size 92:4:4 water:acetonitrile:methanol 7 minutes for standards; 20 minutes for spiked samples 220 nm P a ge 5
Results and Discussion Deoxynivalenol was spiked at the US advisory limits of 1000 µg/kg and at the EU level for infant cereals and foods at 200 µg/kg. Prepared control wheat samples were injected to obtain levels of natural deoxynivalenol. Control values were subtracted from spiked recoveries to obtain the spiked recovery values. Of the five matrices tested, wheat germ (Figure 1) and shredded wheat cereal (Figure 7) controls had detectable levels of deoxynivalenol at 255 µg/kg and 187 µg/kg, respectively. All control and 1000 µg/kg spiked matrices are shown in Figures 1-10 below. Figure 1: Wheat Germ Control Sample Figure 2: Wheat Germ Spiked at 1000 µg/kg Figure 3: Wheat Berries Control Sample Figure 4: Wheat Berries Spiked at 1000 µg/kg Figure 5: Cracked Wheat Control Sample Figure 6: Cracked Wheat Spiked at 1000 µg/kg P a ge 6
Figure 7: Shredded Wheat Control Sample Figure 8: Shredded Wheat Spiked at 1000 µg/kg Figure 9: Infant Wheat Cereal Control Sample Figure 10: Infant Wheat Cereal Wheat Spiked at 1000 µg/kg Sample injections were bracketed by standards. Duplicate injections of standards and samples were performed. Standards retention times within a run had a calculated %RSD of 1.5%. Day to day retention time variation was due to differences in newly mixed batches of mobile phase or when a new guard column was installed. Deoxynivalenol calibration standards were prepared and injected for quantitation of sample spikes. For the sample spikes at 1000 µg/kg, a 6-point standard curve was used at levels of 0, 0.1, 0.25, 0.5, 1, 2.5, and 5 µg/ml. Standards were injected to bracket the sample spikes. Linearity of standards was calculated for each run, with typical R2 values at 0.9993 or higher (Figure 11). For sample spikes at 200 µg/kg, a 7-point standard curve was used at levels of 0, 0.02, 0.04, 0.05, 0.1, 0.2, and 0.5 µg/ml. Standards were injected to bracket the sample spikes. Linearity of standards was calculated for each run, with typical R2 values at 0.9998 or higher (Figure 12). Figure 11: Deoxynivalenol 6-point linearity for runs containing sample spikes at 1000 µg/kg Figure 12: Deoxynivalenol 7-point linearity for runs containing sample spikes at 200 µg/kg P a ge 7
Figure 13: 0.5 µg/ml deoxynivalenol standard with a run time of 7 minutes and retention time of 2.9 minutes Figure 14: 0.5 µg/ml deoxynivalenol standard with a run time of 20 minutes and retention time of 3.8 minutes To avoid late eluting sample matrix peaks interfering with subsequent injections, sample injections had a run time of 20 minutes. Standards did not have late eluting peaks, allowing reduction of the run time to 7 minutes. Figures 13 and 14 show the 0.5 µg/ml deoxynivalenol standard retention time variation and run time change from 20 minutes to 7 minutes. The 0.5 µg/ml standard is the equivalent level to the 1000 µg/kg spiked samples. Recoveries for samples spiked at 1000 µg/kg ranged from 78.3% to 90.9% (Table 2). The average recovery was 85.1%, with a %RSD of 6.10%. There is no current US recovery performance criteria; however, this level of recovery is well within the EU performance criteria of 70-120% for deoxynivalenol recoveries for samples spiked at >500 µg/kg. Table 2: Recoveries and % CVs of Spiked Wheat Matrices at 1000 µg/kg Matrix Number of Replicates US Advisory Limits Recovery (%) @ 1000 µg/kg %CV Wheat Germ 3 90.9 1.0 Wheat Berries 3 78.3 0.8 Cracked Wheat 3 83.9 3.5 Shredded Wheat Cereal 3 82.2 2.5 Infant Wheat Cereal 4* 90.2 2.2 *includes an additional spiked sample extracted at 30 minutes P a ge 8
Table 3: Automated SmartPrep Extractor SPE recoveries of Spiked Wheat Matrices at 200 µg/kg Matrix Number of Replicates EU Advisory Limits Recovery (%) @ 200 µg/kg Wheat Germ 1 77.7 Wheat Berries 1 81.6 Cracked Wheat 1 83.6 Shredded Wheat Cereal 1 68.8 Infant Wheat Cereal 1 83.6 Recoveries for samples spiked at 200 µg/kg ranged from 68.8 83.6% (Table 3). The average recovery was 79.1%, with a %RSD of 6.2%. These values meet the EU recovery performance criteria of 60 110% for deoxynivalenol recoveries spiked at >100 and < 500 µg/kg. Conclusion The effectiveness of the Supel Tox DON SPE Cartridges in combination with the SmartPrep Extractor System efficiently prepared a variety of unprocessed to processed wheat samples representing the global wheat food chain prior to uhplc analysis. Using new uhplc column technology, deoxynivalenol sensitivity was increased 25X over typical analysis methods. Elution of deoxynivalenol enabled quick identification within 5 minutes. The method, as presented, was able to successfully quantitate deoxynivalenol US and EU limits from a wide variety of wheat sample matrices wheat germ, wheat berries, cracked wheat, shredded wheat cereal, and infant wheat cereal. Data presented concludes that deoxynivalenol is effectively recovered and passes performance criteria for % recovery. The level of automation used improves overall laboratory workflow by reducing scientist bias by treating all samples in the same manner. Removing bias improves reliability in reporting results without requiring additional sample runs. The limited need for manual preparation significantly decreases the typical laboratory workload for analysis of deoxynivalenol in foodstuffs and animal feed. P a ge 9
References 1. United States National Library of Medicine, National Institutes of Health website, http://www.ncbi.nlm.nih.gov/pmc/articles/ PMC2984136/ (last checked on June 14th 2014). 2. International Development Research Centre (IDRC) website on Food and Biodiversity, http://www.idrc.ca/en/resources/ Publications/Pages/ArticleDetails.aspx?PublicationID=565 (last checked on August 27th 2014). 3. European Union Site. http://eur-lex.europa.eu/legal-content/en/all/ uri=celex:02006r1881-20100701 (last checked on June 14th 2014). 4. Supelco Supel Tox DON data and LLE and SPE method parameters sheet, T712117, 2012. www.horizontechinc.com AN0961410_01 16 Northwestern Drive, Salem, NH 03079 USA Tel: (603) 893-3663 Email: Support-Service@horizontechinc.com