Anal. Chem. 2006, 78, 1789-1800
Detection, Confirmation, and Quantification of Staphylococcal Enterotoxin B in Food Matrixes Using Liquid Chromatography-Mass Spectrometry John H. Callahan,* Kevin J. Shefcheck, Tracie L. Williams, and Steven M. Musser
Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, 5100 Paint Branch Parkway, College Park, Maryland 20740
Although immunoassay-based methods are sensitive and widely used for measuring protein toxins in food matrixes, there is a need for methods that can directly confirm the molecular identity of the toxin in situations where immunoassay tests yield a positive result. A method has been developed that uses mass spectrometry to identify a protein toxin, staphylococcal enterotoxin B (SEB), in a model food matrix, apple juice. The approach employs ultrafiltration to remove low molecular weight components from the sample, after which the remaining high molecular weight fraction, containing the protein, is digested with trypsin. The tryptic fragments are separated from residual biopolymers and analyzed by liquid chromatography-electrospray mass spectrometry. The background is still sufficiently complex that tandem mass spectrometry (MS/MS) is used to confirm the identity of target peptides. Limits of detection are 80 ng of SEB for MS and 100 ng for full scan MS/MS, using a tryptic fragment as the analytical target. Lower detection limits can be obtained using selected ion monitoring and multiple reaction monitoring. The presence of SEB can be confirmed at concentrations as low as 5 parts-per-billion by increasing the size of the sample to 10 mL. The method is applicable to the detection of SEB in other water-soluble food matrixes. The identification, confirmation, and quantification of protein toxins in food matrixes is an important analytical problem. In the area of food safety, bacterial protein toxins, such as those from Staphylococcus aureus, Campylobacter jejuni, Vibrio cholerae, and Escherichia coli O157:H7, contribute to millions of food-related illnesses in the United States each year.1 Additionally, bacterial protein toxins can be produced in purified form and used as deliberate adulterants, posing a food security risk. Protein toxins ingested in food are well known to produce a variety of adverse health effects, from mild illness to death, and the doses at which heath effects are observed are low. For example, the family of staphylococcal enterotoxins (SEs) from S. aureus are estimated to contribute to ∼185 000 food-related illnesses in the United States each year1 and produce nausea, vomiting, and diarrhea * Corresponding author. E-mail:
[email protected]. (1) Mead, P. S.; Slutsker, L.; Dietz, V.; McCaig, L. F.; Bresee, J. S.; Shapiro, C.; Griffin, P. M.; Tauxe, R. V. Emerging Infect. Dis. 1999, 5, 607-625. 10.1021/ac051292v Not subject to U.S. Copyright. Publ. 2006 Am. Chem. Soc.
Published on Web 02/07/2006
within 12 h of ingestion of a variety of S. aureus-contaminated foods. Total ingested quantities as small as 100 ng (0.5 ng/mL in 200 mL of milk) are thought to have produced health effects in children;2 doses in the low-microgram range can produce effects in adults.3 Although this toxin is generally not fatal when ingested, it clearly presents a health risk and has the potential to incapacitate if used as a nontraditional adulterant in food. Thus, the development of methods for the analysis of SEs is important in itself; moreover, approaches developed for SEs can be used as models for other more toxic proteins, such as those from Clostridium botulinum, which can produce severe symptoms and even death for doses in the low-nanogram range.4 In cases of suspected protein toxin contamination, confirmation by direct detection of the toxin is necessary where the bacterium itself may not be present, such as when toxins are used as deliberate contaminants. Fortunately, there are well-established and sensitive immunoaffinity-based methods available for the detection of protein toxins such as SEs in food matrixes, many of which have detection limits in the 1-10 ng/g range. These include the antibody-based diffusion assay,5 the reverse passive latex agglutination,6,7 reverse passive hemaaglutination8 radioimmunoassay,9 and both displacement and sandwich enzyme-linked immunoassay (ELISA).10-15 The latter are available in commercially available kits with detection limits in foods in the low (2) Evenson, M. L.; Hinds, M. W.; Bernstein, R. S.; Bergdoll, M. S. Int. J. Food Microbiol. 1988, 7, 311-316. (3) Wang, A. C. L.; Bergdoll, M. S. Staphyloccocal Food Poisoning. In Foodborne Diseases; Cliver, D. O., Rieman, H. P., Eds.; Academic Press: San Diego, 2002; pp 232-233. (4) Gill, D. M. FEMS Microbiol. Rev. 1982, 46, 86-94. (5) Official Methods AOAC Int. 1991, (16th ed., Chapter 17), 37-41. (6) Igarishi, H.; Fujikawa, H.; Shingaki, M.; Bergdoll, M. S. J. Clin. Microbiol. 1986, 23, 509-512. (7) Silverman, S. J.; Knott, A. R.; Howard, M. Appl. Microbiol. 1968, 16, 10191023. (8) Miller, B. A.; Reiser, R. F.; Bergdoll, M. S. Appl. Environ. Microbiol. 1978, 36, 421-426. (9) Bergdoll, M. S. Staphylococcal food poisoning. In Foodborne Diseases; Cliver, D. O., Ed.; Academic Press: San Diego, 1990; pp 86-106. (10) Freed, R. C.; Evenson, M. L.; Reiser, R. F.; Bergdoll, M. S. Appl. Environ. Microbiol. 1982, 44, 1349-1355. (11) Kauffman, P. E. J. Assoc. Off. Anal. Chem. 1980, 63, 1138-1143. (12) Kuo, J. K. S.; Silverman, G. J. J. Food Prot. 1980, 43, 404-407. (13) Notermans, S.; Verjans, H. L.; Bol, J.; Van Schothorst, M. Health Lab. Sci. 1978, 15, 28-31 (14) Saunders, G. C.; Bartlett, M. L. Appl. Environ. Microbiol. 1977, 34, 518522. (15) Stiffler-Rosenberg, G.; Fey, H. J. Clin. Microbiol. 1978, 8, 473-479.
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1789
nanogram per gram range. Assays for SEs with detection limits in the subnanogram per milliliter range have been reported for matrixes such as urine and buffer using recently developed ELISA assays,16,17 as well as immunoaffinity-based time-resolved fluorescence,18 immunomagnetic-electrochemiluminescent detection (the ORIGEN system),19 and magnetoelastic detection.20 Immunoaffinity-based multichannel sensors for simultaneous detection of SEs and other toxins have also been reported, with detection limits in the nanogram per milliliter range for SEs in food matrixes (typically milk). These include systems based on fluorescence/ fiber optics,21 chemiluminescence,22 optical/evanescent wave attenuation,23 and surface plasmon resonance.24 The latter has been used to detect SEs in buffer and milk at the 5 ng/mL level in the direct mode (SE binding to antibody on surface). A similar approach has also been employed with a far simpler piezoeletricbased sensor to detect SEs in milk.25 Although they are sensitive, immunoaffinity-based methods rely on the antibody-antigen interaction to induce a signal for the analyte. If another molecule specifically or nonspecifically reacts with the antibody binding site, the assay may generate a false positive or false negative signal, respectively. Where positives are observed, it is desirable to have a confirmatory method that can make a direct measurement of an intrinsic property of the analyte. One such method, mass spectrometry (MS), has already been widely used in the analysis of adulterants in many foods.26,27 Electrospray ionization (ESI) MS28 and matrix-assisted laser desorption/ionization (MALDI) MS29 readily extend this capability to protein toxins and other high molecular weight biomolecule adulterants. Furthermore, proteomics-based methods can be used to obtain even more specific information;30-32 proteins are enzymatically digested to produce peptides, which can be readily separated by liquid chromatography (LC), ionized, and analyzed using collision-induced dissociation tandem mass spectrometry (MS/MS). The latter produces information about the primary sequence of amino acids in the peptide.33 Since a relatively short (16) Poli, M. A.; Rivera, V. R.; Neal, D. Toxicon 2002, 40, 1723-1726. (17) Lee, W. E.; Thompson, H. G.; Hall, J. G.; Bader, D. E. Biosens. Bioelectron. 2000, 14, 795-804. (18) Peruski, A. H.; Johnson, L. H., III; Peruski, L. F. J. Immunol. Methods 2002, 263, 35-41. (19) Kijek, T. M.; Rossi, C. A.; Moss, D.; Parker, R. W.; Henchal, E. A. J. Immunol. Methods 2000, 236, 9-17. (20) Ruan, C.; Zeng, K.; Varghese, O. K.; Grimes, C. A. Biosens. Bioelectron. 2004, 19, 1695-1701. (21) King, K. D.; Anderson, G. P.; Bullock, K. E.; Regina, M. J.; Saaski, E. W.; Ligler, F. S. Biosens. Bioelectron. 1999, 14, 163-170. (22) Yacoub-George, E.; Meixner, L.; Scheithauer, W.; Koppi, A.; Drost, S.; Wolf, H.; Danapel, C.; Feller, K. A. Anal. Chim. Acta 2002, 457, 3-12. (23) Rasooly, L.; Rasooly, A. Int. J. Food Microbiol. 1999, 49, 119-127. (24) Homola, J.; Dostalek, J.; Chen, S.; Rasooly, A.; Jiang, S.; Yee, S. S. Int. J. Food Microbiol. 2002, 75, 61-69. (25) Lin, H.-C.; Tsai, W.-C. Biosens. Bioelectron. 2003, 18, 1479-1483. (26) Careri, M.; Bianchi, F.; Corradini, C. J. Chromatogr., A 2002, 970, 2-64. (27) Careri, M.; Mangia, A. J. Chromatogr., A 2003, 1000, 609-635. (28) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989, 246, 64. (29) (a) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid Commun. Mass Spectrom. 1988, 2, 151-153. (b) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (30) Aebersold, R.; Goodlett, D. R. Chem. Rev. 2001, 101, 269-296. (31) Gygi, S. P.; Rist, B.; Griffin, T. J.; Eng, J.; Aebersold, R. J. Proteome Res. 2002, 1, 47-54. (32) Florens, L.; Washburn, M. P.; Raine, J. D.; Anthony, R. M.; Grainger, M.; Haynes, J. D.; Moch, J. K.; Muster, N.; Sacci, J. B.; Tabb, D. L.; Witney, A. A.; Wolters, D.; Wu, Y.; Gardner, M. J.; Holder, A. A.; Sinden, R. E.; Yates, J. R., 3rd; Carucci, D. J. Nature 2002, 419, 520-526.
1790
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
peptide sequence can often identify the protein in question, it is typically not necessary to identify all peptide fragments from a particular protein. Thus, MS can provide a direct or intrinsic measure of the analyte in one of two ways, by molecular weight or primary sequence. MS is also an intrinsically multianalyte technique, enabling the simultaneous analysis of multiple toxins without need for specific reagents. The LC/ESI MS analysis of intact staphylococcal enterotoxin B (SEB) has been previously reported by Kientz and co-workers,34 and they were able to detect quantities as small as 85 pg of SEB per chromatographic injection. They also showed that LC/MS of SEB tryptic digest fragments produced fragmentation spectra that definitively identified the peptide by both mass and amino acid sequence. This group has also demonstrated that MS is a useful tool for the analysis of other protein toxins, such as that from C. botulinum.35,36 Kawano and co-workers showed that a variety of proteins produced by S. aureus, including enterotoxins, can be characterized by LC/MS.37 In an application involving food, Nedelkov and co-workers detected SEB in milk and mushrooms using a combination of antibody extraction and MALDI MS.38 This latter study nicely eliminated the effects of matrix by using antibody extraction; however, the reliance on antibodies means that the approach cannot easily be extended to toxins for which antibodies have not been developed. The previously cited studies have shown that MS is a sensitive and specific approach for the identification, confirmation, and detection of repurified protein toxins. However, the identification of protein toxins in food matrixes constitutes a more difficult problem. Foods consist of large quantities of proteins, carbohydrates, and other biomolecules, many of which can be expected to produce significant signals by both electrospray and MALDI mass spectrometry. Both of these methods are affected by signal suppression due to competition for charge during ionization, so detection limits are expected to be different in the presence of complex background. In this paper, we discuss an approach for the confirmation and quantification of SEB in apple juice, a simple matrix that nevertheless presents difficult analytical challenges and serves as a useful model for approaches to this problem. Apple juice contains a wide molecular weight range of soluble saccharides, from monosaccharides on the low end of the range to polysaccharides that exceed 100 kDa in mass at the high end.39 These soluble components are not readily removed and can interfere with electrospray-based analyses, since even neutral polysaccharides can pick up charge and compete with protein signals. In this report, we discuss the use of whole protein LC/MS to measure SEB in a representative food matrix and present the (33) Hunt, D. F.; Yates, J. R., III; Shabanowitz, J.; Winston, S.; Hauer, C. R. Proc. Natl. Acad. Sci. U.S.A.1986, 83, 6233-6237. (34) Kientz, C. E.; Hulst, A. G.; Wils, E. R. J. J. Chromatogr., A 1997, 757, 5164. (35) van Baar, B. L. M.; Hulst, A. G.; de Jong, A. L.; Wils, E. R. J. J. Chromatogr., A 2002, 970, 95-115. (36) van Baar, B. L. M.; Hulst, A. G.; de Jong, A. L.; Wils, E. R. J. J. Chromatogr., A 2004, 1035, 97-114. (37) Kawano, Y.; Ito, Y.; Yamakawa, Y.; Yamashino, Y.; Horii, T.; Hasegawa, T.; Ohta, M. FEMS Microbiol. Lett. 2002, 189, 103-108. (38) Nedlekov, D.; Rasooly, A.; Nelson, R. W. Int. J. Food Microbiol. 2000, 60, 1-13. (39) Mehrlander, K.; Dietrich, H.; Sembries, S.; Dongowski, G.; Will, F. J. Agric. Food Chem. 2002, 50, 1230-1236.
limitations of this methodology. We then outline an approach that relies on initial sample separation combined with enzymatic digestion and tandem mass spectrometry to produce specific tryptic peptide fragments from the target protein in a food matrix. In turn, these selected peptide fragments and their MS/MS spectra can be used to identify and confirm SEB. The approach was also translated to a triple quadrupole MS, to utilize methods such as selected ion monitoring (SIM) and multiple reaction monitoring (MRM) that are more amenable to quantification. Ultimately, using SIM and MRM with increased sample sizes, SEB was quantified at the 5-10 ng/mL level in a model food matrix. Although SEB only comprises one of the SEs, and there are many other additional protein toxins, SEB is used as a model system in this study. A brief examination of other food matrixes indicates that the approach is applicable to foods other than apple juice. Although proteomics-based methods are widely used for semiquantitative analysis in biological matrixes such as albumindepleted serum, there have been few reports of the use of these methods for the quantitative analysis of target proteins in the presence of high concentrations of the complex biomolecule-based matrixes found in foods. EXPERIMENTAL SECTION Reagents. Staphylococcal enterotoxin B was obtained in purified form from Toxin Technology, Inc. (Sarasota, FL). The toxin was dissolved in water to prepare a 1 mg/mL stock solution, which was maintained in frozen form until diluted (serially) to make additional stock solutions of 100, 10, and 1 µg/mL (ng/ µL). Care should be taken in the handling of SEB, using gloves and protection at all times. Surfaces and materials exposed to SEB should be treated with bleach to destroy residual toxin. The trypsin used in this study was Promega stabilized trypsin (20 µg/50 µL) and was kept frozen at -40 °C until used. Enzymatic digestions were carried out at 37 and 60 °C for periods of 1-24 h, as noted below. All digestions were performed in 0.1 M NH4HCO3 at pH 8. Synthetic peptides VTAQELDYLTR, VTAQELDYVTR, and VTAQELDYVTRHYVVK were prepared at >95% purity by Sigma Genosys, Inc. (The Woodlands, TX). All were dissolved at ∼1 mg/ mL in water and serially diluted to give solutions of 0.1, 1.0, 10, and 100 µg/mL (ng/µL). All solutions were frozen if not in use and periodically remade from the stock as required. Leucine enkephalin was obtained from Sigma-Aldrich (St. Louis, MO). HPLC grade water and HPLC grade acetonitrile were obtained from Burdick and Jackson. Formic acid (Fluka) and acetic acid (Baker) were reagent grade. Apple juice was obtained from retail sources and was used as received. Materials. Filtration. Two types of molecular weight cutoff (MWCO) centrifuge (spin) filters were used in these studies. For volumes of 1 mL and less, 0.5-mL Ultrafree Centrifuge Filters (Millipore, Billerica, MA) with 5 or 10 kDa MWCO were employed. Samples were processed at 12000g, unless otherwise stated. For quantities of sample larger than 1 mL, 4-mL Amicon Ultra Centrifuge filters (Millipore) with 5 and 10 kDa MWCO ranges were used. Samples were processed at 4000g unless otherwise specified. Where sample volumes exceeded the capacity of the filter, the samples were processed in multiple poritions. In all cases, cutoff filters were washed with water before use to remove glycerol on the filter.
Solid-Phase Extraction (SPE). Solid-phase extraction was performed with Phenomenex (Torrance, CA) Strata X SPE anion exchange cartridges using a modification of the protocol suggested by the manufacturer. Cartridges were washed with acetonitrile and 0.1 M NH4HCO3. Samples were loaded in 0.25 mL of 0.1 M NH4HCO3 and washed with 1 mL of the same solvent. Samples were eluted with 0.25 mL of 70% acetonitrile and 30% H2O followed by 0.25 mL of 70% acetonitrile and 30% H2O with 1% formic acid. The eluates were combined. Sample Preparation Methods. Method 1. Direct measurements of SEB in apple juice: SEB standards were spiked into aliquots of apple juice directly and thoroughly mixed. Samples were analyzed by direct injection of 5-10-µL aliquots. Method 2. In situ sample digestion and spin filtration (UF): Known quantities of SEB (or surrogate peptide standards in recovery studies) were spiked into 1 mL of apple juice. The sample was then adjusted to pH 8 with 40 µL of 1 M NaOH and 100 µL of 1 M NH4HCO3. After adjustment, 2 µL (0.4 µg) of trypsin was added to the sample and digested at 60 °C for 18-24 h. The sample was then processed, in two aliquots, through a 0.5-mL, 10 kDa MWCO spin filter at 12000g for 20 min. The volume was reduced to dryness by vacuum evaporation and the sample was reconstituted in 0.1% formic acid. Method 3. Spin filtration and digestion (UF1): 1-mL apple juice samples were spiked with known quantities of SEB and spin filtered, in two portions, through a 0.5-mL, 5 kDa MWCO centrifuge filter. The filtrate was discarded and the retentate was washed with aliquots of 0.1 M NH4HCO3. The final retentate was then resuspended in 200 µL of 0.1 M NH4HCO3, 2 µL (0.4 µg) of trypsin was added, and the sample was digested for 18-24 h at 60 °C (standards were added at this point in surrogate studies). After digestion, the sample was acidified with 1.0% formic acid and evaporated under vacuum, followed by reconstitution in 25 µL of 0.1% formic acid. Method 4. Spin filtration/digestion/spin filtration (UF2 or “notch” filtration): Sample volumes of 1 mL were filtered in two aliquots through 5 kDa MWCO centrifuge filters, and the filtrate was discarded. The retentate was then washed twice with 0.25 mL of 0.1 M ammonium bicarbonate solution, taken up in 0.2 mL of this buffer, and transferred to a vial. The filter was then rinsed with 0.1 mL of the buffer, which was added to the recovered retentate. After digestion with 2 µL (0.4 µg) of trypsin for 18-24 h at 60 °C, the samples were cooled. The digest solution was added to a 0.5-mL, 10 kDa MWCO spin filter and centrifuged at 12000g for 20 min. A 200-µL sample of 0.1 M ammonium bicarbonate was added to the filter and collected into the filtrate to further wash the sample. The filtrate was then retained for further cleanup with anion exchange SPE cartridges according to the procedure outlined above. The eluate was then evaporated under vacuum for later reconstitution in a known volume. Other Food Matrixes. Four additional food matrixes, orange juice, fruit punch, green beans, and crackers, were tested. The notch filtration method was used for direct analysis of orange juice and fruit punch in a fashion similar to apple juice, with the exception that particulate matter was centrifuged from the sample prior to spin filtration. SEB was added directly to the liquid used to pack canned green beans, and the sample was then processed as with the juice samples. A 7-g sample of crackers was crushed Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
1791
and sonicated in 50 mL of water and then centrifuged to remove particulate. The decantate with water-soluble components was used as a test matrix, to which SEB was directly added. Internal Standards. Leu-enkephalin was used as an added internal standard in these studies. A stock solution with a concentration of 1 mg/mL was prepared in water and kept frozen. Serial dilutions were used to prepare 10 and 1 ng/µL standards, which were also kept frozen until used. In most cases, 15 µL of the 1 ng/µL standard (15 ng) was added to each sample after workup but before final evaporation. A digest internal standard (VTAQELDYVTRHYVVK) was also used in these studies. 17 ng (10 µL of 1.7 ng/µL solution) of VTAQELDYVTRHYVVK was added to solutions prior to digestion with trypsin. The peptide is cleaved to form VTAQELDYVTR, which is an analogue of a target peptide VTAQELDYLTR (replacement of valine for leucine at the ninth amino acid). Chromatography. Whole protein chromatography was performed on a Phenomenex Jupiter 300 C4 column (0.5 mm × 250 mm). The flow rate, unless otherwise specified, was 40 µL/min. A binary solvent system was used, consisting of solvents A (90% water, 10% acetonitrile, 0.5% acetic acid) and B (10% water, 90% acetonitrile, 0.5% acetic acid). The flow program consisted of 100% A for 5 min, a linear gradient of 0-50% B in 40 min, a linear gradient of 50-90% B in 20 min, followed by a 5-min hold at 90% B and a 20-min reequilibration at 0% B. The separations were carried out at 50 °C. Peptide separations were performed on a 0.32-mm-diameter, 150-mm Symmetry 300 C18 column (Waters, Inc., Milford, MA) at flow rates of 20 µL/min. A binary solvent system consisting of A (95% water, 5% acetonitrile, 0.5% formic acid) and B (10% water, 90% acetonitrile, 0.5% formic acid) was used with the following program: 100% A for 5 min, 0-50% B in 40 min, 50-90% B in 20 min, followed by a 5-min hold at 90% B and a 20-min reequilibration at 0% B. All peptide separations were carried out at 50 °C. When working with samples in apple juice, the high loads on the columns led to rapid degradation of the LC column, even when extensive cleanup was employed. To minimize column degradation, guard columns consisting of Phenomenex C18 guard cartridges were employed. While this led to some degradation of column resolution and peak broadening, the use of guard columns made the application of the approach more practical and cost-effective. Mass Spectrometry. Mass spectrometry studies were performed on a Micromass QTOF Micro quadrupole time-of-flight (QTOF), a Waters QTOF Premier (Waters, Inc.), or a Micromass Quattro Premier QqQ mass spectrometer. Electrospray ionization was employed at flow rates of 10-40 µL/min. The capillary voltage was 3 kV, the capillary temperature was 150 °C, the nebulizer flow was 500-600 L/h, and the cone gas flow rate was 50 L/h (
