Anal. Chem. 2009, 81, 4285–4295
Fully Automated Analysis of β-Lactams in Bovine Milk by Online Solid Phase Extraction-Liquid Chromatography-Electrospray-Tandem Mass Spectrometry Lina Kantiani,† Marinella Farre´,† Martin Sibum,‡ Cristina Postigo,† Miren Lo´pez de Alda,† and Damia´ Barcelo´*,†,§ Department of Environmental Chemistry, IDAEA-CSIC, c/Jordi Girona 18-26, 08034, Barcelona, Spain, Spark Holland B.V., P.O. Box 388, 7800 AJ Emmen, The Netherlands, and Catalan Institute for Water Research (ICRA), Parc Cientı´fic i Tecnolo`gic de la Universitat de Girona, Edifici Jaume Casademont, Porta A, Planta 1 - Despatx 13C/ Pic de Peguera, 15, E-17003 Girona, Spain A fully automated method for the detection of β-lactam antibiotics, including six penicillins (amoxicillin, ampicillin, cloxacillin, dicloxacillin, oxacillin, and penicillin G) and four cephalosporins (cefazolin, ceftiofur, cefoperazone, and cefalexin) in bovine milk samples has been developed. The outlined method is based on online solidphase extraction-liquid chromatography/electrospraytandem mass spectrometry (SPE-LC/ESI-MS-MS). Target compounds were concentrated from 500 µL of centrifuged milk samples using an online SPE procedure with C18 HD cartridges. Target analytes were eluted with a gradient mobile phase (water + 0.1% formic acid/methanol + 0.1% formic acid) at a flow rate of 0.7 mL/min. Chromatographic separation was achieved within 10 min using a C-12 reversed phase analytical column. For unequivocal identification and confirmation, two multiple reaction monitoring (MRM) transitions were acquired for each analyte in the positive electrospray ionization mode (ESI(+)). Method limits of detection (LODs) in milk were well below the maximum residue limits (MRLs) set by the European Union for all compounds. Limits of quantification in milk were between 0.09 ng/mL and 1.44 ng/mL. The developed method was validated according to EU’s requirements, and accuracy results ranged from 80 to 116%. Finally, the method was applied to the analysis of twenty real samples previously screened by the inhibition of microbial growth test Eclipse 100. This new developed method offers high sensitivity and accuracy of results, minimum sample pre-treatment, and uses for the first time an automated online SPE offering a high throughput analysis. Because of all these characteristics, the proposed method is applicable and could be deemed necessary within the field of food control and safety. Beta-lactam antibiotics, which consist of penicillins and cephalosporins, form one of the oldest and most important groups of * To whom correspondence should be addressed. E-mail:
[email protected]. † Department of Environmental Chemistry, IDAEA-CSIC. ‡ Spark Holland B.V. § Catalan Institute for Water Research (ICRA). 10.1021/ac9001386 CCC: $40.75 2009 American Chemical Society Published on Web 04/29/2009
antibiotics. They are widely used in veterinary medicine to treat and/or prevent bacterial infections and diseases, such as bovine mastitis, pneumonia, bacterial diarrhea, and bacterial arthritis,1 or as supplements that are provided illegally to promote growth in food-producing animals. Antibiotic residues in foods are of concern as they present potential health risks because of possible allergic reactions, carcinogenicity, and promotion of the spread of bacterial resistance to antibiotics used in human medicine.2 A high number of studies reported cases of allergic reaction in humans upon consumption of contaminated food.3-6 In addition, the whole fermentation process in the production of cultured dairy products, for example, yogurt, is disturbed and can be adversely affected by traces of antibiotics in raw milk.7 This may result in considerable economic loss as product batches may be downgraded or discarded. Producers are required to offer an antibiotic free product and are responsible to ensure illegal antibiotic residues are prevented. To reduce human exposure to β-lactam antibiotics, tolerance levels have been established by regulatory agencies for antibiotics that are approved to be used in food-producing animals. The European Union [2002/657/CE and amendments] has set specific maximum residue limits (MRLs) for each of these substances to protect human health.8 The presence of antibiotics in animal products can lead to heavy penalties being imposed on producers. In 2002, it is estimated that one tank in a thousand has an antibiotic failure in England.9 (1) Goto, T.; Ito, Y.; Yamada, S.; Matsumoto, H.; Oka, H. J. Chromatogr. A 2005, 1100, 193–199. (2) Bogialli, S.; Capitolino, V.; Curini, R.; Di Corcia, A.; Nazzari, M.; Sergi, M. J. Agric. Food Chem. 2004, 52, 3286–3291. (3) Dayan, A. D. Vet. Microbiol. 1993, 35, 213–226. (4) Tsujikawa, K.; Kuwayama, K.; Miyaguchi, H.; Kanamori, T.; Iwata, Y.; Inoue, H.; Kishi, T. J. Forensic Sci. 2008, 53, 226–231. (5) Schwartz, H. J.; Sher, T. H. Ann. Allergy 1984, 52. (6) Sundlof, S. F.; Cooper, J. Veterinary drug residues: Human health risks associated with drug residues in animal derived foods; American Chemical Society: Washington, DC, 1996. (7) Grunwald, L.; Petz, M. Anal. Chim. Acta 2003, 483, 73–79. (8) Implementing Council Directive 96/23/EC concerning the performance of analytical methods and the interpretation of results (2002/657/EC and amendments) Off. J. Eur. Commun. 2002, L2218. (9) Edmondson, P. Avoiding antibiotic residues in milk, 2002, http://www. milkproduction.com (accessed during August 2008).
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009
4285
The most commonly used antibiotics in food animals can be grouped into five major classes. These include the beta-lactams (β-lactams) (e.g., penicillins and cephalosporins), tetracyclines (e.g., oxytetracycline, tetracycline, and chlortetracycline), aminoglycosides (e.g., streptomycin and gentamicin), macrolides (e.g., erythromycin), and sulfonamides (e.g., sulfamethazine). In this study, we have dealt with the analysis of six penicillins and four cephalosporins in milk, compounds that are currently used most in food producing animals in Europe, and for which certain limitations on veterinary usage have been placed. The basic structure of penicillins consists of a β-lactam ring attached to a five-membered thiazolidine ring and a side chain, whereas that of cephalosporins has a six-membered dihydrothiazine ring attached to the β-lactam ring. For the compounds included in this study the EU set MRLs in milk-based matrixes are the following: Amoxicillin (AMOX) - 4 ng/ mL, Ampicillin (AMP) - 4 ng/mL, Penicillin G (PEN G) - 4 ng/ mL, Oxacillin (OXA) - 30 ng/mL, Cloxacillin (CLOX) - 30 ng/mL, Dicloxacillin (DICLOX) - 30 ng/mL, Cefoperazone (CEFOP) - 50 ng/mL, Cefazolin (CEFAZ) - 50 ng/mL, Cefalexin (CEFAL) - 100 ng/mL, and Ceftiofur (CEFT) - 100 ng/mL. Molecular structure, CAS number, molecular weight, and physical properties for the target analytes are provided in the Supporting Information, Table 1. Various screening methods using rapid test kits are being regularly employed to control chemical residues in milk. Recent developments include optical biosensors using surface plasmon resonance (SPR) and microplate assays based on the penicillinbinding protein PBP 2x*.10-12 However, although these tests have a wide range of selectivity, which can give evidence of one or more groups of antibiotics in a milk sample, they do not provide high specificity, which results in semiquantitative measurements and can also lead to false positive results. For this reason, sensitive and reliable analytical methods have been developed mainly employing liquid chromatography coupled to mass spectrometry (LC-MS) or tandem MS (LC-MS-MS) following a pre-concentration step, usually achieved by solid-phase extraction (SPE). To the present many single-analyte and multiresidue methods have been developed for the extraction and detection of one or more β-lactam compounds in milk and animal tissues and kidney.1,2,13-21 Although limits of detection presented in these studies are below the set MRLs, the chromatographic runs are usually long leading to low sample throughput. To our knowledge, all existing methods employ manual sample preparation, extraction, and pre-concentration, usually a lengthy and laborious (10) Cacciatore, G.; Petz, M.; Rachid, S.; Hakenbeck, R.; Bergwerff, A. A. Anal. Chim. Acta 2004, 520, 105–115. (11) Gustavsson, E.; Sternesjo, A. J. AOAC Int. 2004, 87, 614–620. (12) Lamar, J.; Petz, M. Anal. Chim. Acta 2007, 586, 296–303. (13) Holstege, D. M.; Puschner, B.; Whitehead, G.; Galey, F. D. J. Agric. Food Chem. 2002, 50, 406–411. (14) Becker, M.; Zittlau, E.; Petz, M. Anal. Chim. Acta 2004, 520, 19–32. (15) Bruno, F.; Curini, R.; Di Corcia, A.; Nazzari, M.; Samperi, R. J. Agric. Food Chem. 2001, 49, 3463–3470. (16) Riediker, S.; Diserens, J. M.; Stadler, R. H. J. Agric. Food Chem. 2001, 49, 4171–4176. (17) Daeseleire, E.; De Ruyck, H.; Van Renterghem, R. Rapid Commun. Mass Spectrom. 2000, 14, 1404–1409. (18) Turnipseed, S. B.; Andersen, W. C.; Karbiwnyk, C. M.; Madson, M. R.; Miller, K. E. Rapid Commun. Mass Spectrom. 2008, 22, 1467–1480. (19) Fagerquist, C. K.; Lightfield, A. R.; Lehotay, S. J. Anal. Chem. 2005, 77, 1473–1482. (20) Mastovska, K.; Lightfield, A. R. J. Chromatogr. A 2008, 1202, 118–123. (21) Msagati, T. A. M.; Nindi, M. M. Food Chem. 2007, 100, 836–844.
4286
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009
procedure, or dispersive SPE cleanup, a somewhat simplified and speeded up sample preparation,19,20 or supported liquid membrane for sample purification and enrichment, which reduces the cost of analysis.21 Recent reviews have included this particular group of antibiotics and their detection in animal-derived products.22-26 In the last years, fully automated extraction methods have been gaining attention for the detection and quantification of emerging pollutants and β-lactams in environmental samples and plasma.27-29 The main advantages of online SPE methods are the significant reduction in preparation time resulting in higher sample throughput and the small sample volume requirements needed for high sensitivity and accuracy. In the case of β-lactam antibiotics, an analytical method with an online SPE procedure and a short chromatographic elution time is crucial to avoid potential degradation of the compounds under study in the milk matrix. Following this need for fast and reliable analytical methods, the main objectives for the present work were as follows: To develop a fully automated online SPE-LC-MS/MS method for the detection and quantification of 6 penicillins and 4 cephalosporins in milk. To validate this new analytical method according to the EC requirements at levels below the set MRLs so that it can be applicable toward the control of milk samples by the food industry. To analyze selected analytes in real raw milk samples. To the authors’ knowledge this is the first analytical method reported for online analysis of the selected compounds in food samples. EXPERIMENTAL SECTION Chemicals and Reagents. Amoxicillin (T trihydrate), ampicillin, cefazolin (sodium salt), cefoperazone (sodium salt), ceftiofur, cefalexin, cloxacillin (sodium salt hydrate), dicloxacillin (sodium salt hydrate), oxacillin (sodium salt monohydrate), penicillin G (benzathine salt hydrate), and penicillin V (potassium salt), used as internal standard, were purchased from Sigma Aldrich (Madrid, Spain). All SPE cartridges used for sorbent screening tests and validation were supplied by Spark Holland (Emmen, The Netherlands). HPLC-grade water, methanol and acetonitrile were purchased from Merck (Darmstadt, Germany). Potassium dihydrogen phosphate and potassium hydroxide were obtained from Scharlau (Barcelona, Spain), and Formic acid (98-100%) (FA) was from Merck. Standard Solutions. Individual stock solutions were prepared by dissolving 1 mg of each compound in 1 mL of HPLC water in stained glass stopper bottles at room temperature. Oxacillin and dicloxacillin stock solutions were prepared in HPLC water/MeOH (50:50, v/v), while ceftiofur, penicillin G, and ampicillin, as well (22) (23) (24) (25) (26) (27) (28) (29)
Di Corcia, A.; Nazzari, M. J. Chromatogr. A 2002, 974, 53–89. Careri, M.; Bianchi, F.; Corradini, C. J. Chromatogr. A 2002, 970, 3–64. Stolker, A. A. M.; Brinkman, U. A. T. J. Chromatogr. A 2005, 1067, 15–53. Gentili, A.; Perret, D.; Marchese, S. TrAC, Trends Anal. Chem. 2005, 24, 622–634. Ricci, F.; Volpe, G.; Micheli, L.; Palleschi, G. Anal. Chim. Acta 2007, 605, 111–129. Postigo, C.; Lopez De Alda, M. J.; Barcelo´, D. Anal. Chem. 2008, 80, 3123– 3134. Feitosa-Felizzola, J.; Temime, B.; Chiron, S. J. Chromatogr. A 2007, 1164, 95–104. de Castro, A.; Fernandez, M.d. M. R.; Laloup, M.; Samyn, N.; De Boeck, G.; Wood, M.; Maes, V.; Lo´pez-Rivadulla, M. J. Chromatogr. A 2007, 1160, 3–12.
as penicillin V internal standard solution, were made up in HPLC water/ACN (50:50, v/v). All stock solutions were divided into small aliquots and stored in a refrigerator at -25 °C. To ensure stability of the compounds in solution, different aliquots were used for each freeze-thaw cycle, and fresh stock solutions were prepared monthly. Standard working mixtures, required for calibration curves and spiking tests, were prepared daily, at appropriate concentrations for each compound, by successive accurate dilutions of the stock solution in HPLC water. Samples and Sample Preparation. Initially, optimization of SPE parameters, as well as LC-MS-MS conditions, were performed with pasteurized milk (3.6 g fat, 3.0 g proteins) purchased from local markets. As a second step, raw milk samples, that is, not pasteurized milk, obtained from the Official Milk Laboratory of Catalonia (ALLIC) were used for the validation of the method and the applicability study. Raw milk samples were aliquoted into plastic bottles, and 150 µL of azidiol (sodium azide/chloramphenicol), a preservative used to inactivate the metabolism of bacteria, were added in every 45 mL of milk. Among various preservatives evaluated in a recent study by Sesˇkena R. et al.,30 azidiol was shown to have no significant effect on the fat or protein content in milk and to provide stable milk quality. The addition of this preservative, first evaluated in a study by Barcina et al. in 1987,31 increases the conservation time of milk to 5-6 days at 2-6 °C or 24 h at 20 °C and forms part of the standardized procedure used by ALLIC. The samples were vortex mixed for 1 min, analyzed straightaway, or stored at -25 °C until further use. The preparation of milk samples prior to SPE-LC-MS/MS analysis was as follows: 5 mL of milk sample were transferred into 50 mL glass centrifuge tubes and fortified with 400 µL of a 1 ng/mL PEN-V solution (i.e., 40 ng/mL in the final extract) by vortex mixing (1 min). The sample was then centrifuged at 8000 rpm for 15 min at room temperature. After partial separation of the protein and fat content, the middle aqueous phase was transferred into sample vials ready to be used for online SPE and LC-MS-MS analysis. Online SPE Procedure. Sample pre-concentration was performed using the online SPE Symbiosis Pharma System (Spark Holland, Emmen, The Netherlands). The system setup is shown in Figure 1 and consists of two integrated units: (i) the refrigerated Reliance autosampler, able to accommodate up to 24 of 96-well blocks, with two binary LC pumps, and (ii) the automated cartridge exchange (ACE) unit with two separate high-pressure dispenser (HPD) units for solvent handling. The HPD1 unit is the right side module and serves to carry solvents during the sample application and clean up. The HPD2 unit is the left side module and is responsible for the elution of the loaded cartridge. The integrated system is operated using the Symbiosis Pharma software (Spark, Holland) for Analyst versions 1.4.2 and 1.5 (Applied Biosystems). The defatted sample extracts were loaded onto the SPE system, and the following XLC sequence was used: The SPE C18 HD cartridges were conditioned by passing MeOH (1 mL), water (2 mL), and finally a buffer solution at pH 8.5 (2 mL), all at a flow (30) Sesˇkena, R.; Jankevica, L. Acta Universitatis Latviensis 2007, 723, 171– 180. (31) Barcina, Y.; Zorraquino, M. A.; Pedauye, J.; Ros, G.; Rinco´n, F. An.Vet. (Murcia) 1987, 3, 65–69.
Figure 1. Schematic representation of the online SPE system layout.
rate of 5 mL/min. 1 M buffer solution was prepared by diluting 109.5 g of KH2PO4 in 1 L of HPLC water. The pH was adjusted to 8.5 with the addition of 10 M KOH solution. A 500 µL portion of sample extract was then applied onto the cartridge using 2 mL of the same buffer solution as transfer solvent at 5 mL/ min. The SPE cartridge was then washed to remove interferences by passing 1.5 mL of the buffer solution and 3 mL of water, again at a flow rate of 5 mL/min. Once the SPE procedure, taking place at the left clamp of the Symbiosis Pharma, was completed, the cartridge was automatically moved onto the right side clamp, where the elution of the target analytes to the LC column begun. At the same time, a new cartridge was placed onto the left side clamp, and the next extraction sequence was initiated. Target analytes were eluted using the mobile phase solvents with a flow rate of 0.7 mL/ min at an elution time of 5 min. Following the elution step, and to avoid sample carry over, multiple valve and clamp washes were carried out with an aqueous solution of 5% ACN 0.2% FA. LC-ESI-MS/MS LC was performed using the Symbiosis Pharma system equipped with a 10 mL sample loop. The chromatographic separation of the compounds was achieved on a C12 Phenomenex Hydro-RP (50 mm × 2 mm i.d., 4 µm, 80 Å) preceded by a C12 MAX-RP security guard column (4 mm × 3 mm i.d.), or just stainless steel frits (Screen 1/8′′ OD, 90% for all cephalosporins.34 For these reasons, as well as the variability in sample’s components (proteins, fat, etc) between the different types of milk, matrix-matched calibration curves were always used to avoid significant quantification errors and therefore misinterpretation of results. Three analyzed raw milk samples were found to be negative with respect to β-lactam concentrations and another two samples contained one β-lactam but at a concentration lower than the corresponding quantification limit. AMOX, AMP, CEFAL, PENG, and CLOX were detected in real milk samples, and the relevant results are shown in Table 5. From a total of 20 samples, 15 samples were found to contain at least one target analyte at a quantifiable level. For one sample the concentration quantified for CLOX (48.68 ng/ mL) exceeded the permitted limit set by the EU (30 ng/mL) by more than 50%. In particular, CEFAL was found only in one sample and AMOX was found in four samples at a mean concentration of 1.57 4294
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009
ng/mL. Concentrations for AMP, PENG, and CLOX ranged between 0.02-1.28, 1.18-3.36, and 4.97-48.68, respectively. Mean concentrations were 0.23 ng/mL for AMP, 2.17 ng/mL for PEN G, and 18.77 ng/mL for CLOX. The presence of AMOX, CEFAL, AMP, and PENG was confirmed but not quantified (concentrations being lower than the LOQ but higher than the LOD) in one, four, one, and six samples, respectively. Figure 4 shows online SPE-LC-MS/MS chromatograms acquired for the samples containing the most pronounced concentrations, as well as for a standard solution of 10 ng/mL. This analytical method presents an easy and cost-effective procedure for the identification and quantification of selected antibiotic compounds in raw milk samples and is a useful tool for the confirmatory analysis of these compounds in previous positively screened milk samples in routine food analysis. CONCLUSIONS Numerous previously developed methods for the analysis of β-lactams in milk based matrixes involve lengthy and laborious steps for the pre-treatment of the samples, involving de-proteinization by the addition of solvents or strong acids, and evaporation usually under a stream of nitrogen. The SPE procedure is unquestionably the most important step when developing a multi-analyte method. Because of variations between physicochemical properties of the compounds under study, when optimizing different parameters one needs to be very cautious on the compromises that need to be made in order for all analytes to be detected under the specified requirements. In this study, the applicability, high sensitivity, and accuracy of a novel, fully automated method for confirmatory analysis, unequivocally identification and quantification of β-lactam antibiotics (six penicillins: amoxicillin, ampicillin, cloxacillin, dicloxacillin, oxacillin, and penicillin G; and four cephalosporins: cefazolin, ceftiofur, cefoperazone, and cefalexin) in bovine milk have been demonstrated. In addition, the method requires minimum sample preparation, allowing high sample throughput (10 min as analysis time). LODs in milk were well below the MRLs set by the European Union for all compounds, and LOQs in milk ranged between 0.09 ng/mL (AMP) and 1.44 ng/mL (PEN G). The method has been demonstrated to serve as a useful tool for identification and quantification of the selected compounds in previously positive screened samples by routine analysis. The analysis of spiked real samples showed no false negative or positive results, whereas the analysis of real samples revealed that 15 out of 20 samples contained quantifiable concentrations for five target β-lactams (AMOX, AMP, CEFAL, PENG, and CLOX). In fact, one of the samples presented a concentration for CLOX exceeding the maximum legal limits set for this compound in milk. ACKNOWLEDGMENT The work described in this article was supported by the European Union through project CONffIDENCE (Contract No. 211326). Lina Kantiani acknowledges the AGAUR (Generalitat de Catalunya, Spain) and Alexander S. Onassis public benefit foundation (F-ZD 029/2007-2008) for their economical support through the FI pre-doctoral grant. Cristina Postigo acknowl-
edges the European Social Fund and AGAUR (Generalitat de Catalunya, Spain) for their economical support through the FI pre-doctoral grant. The authors acknowledge Spark Holland for continuous support throughout the project and supply of the SPE cartridges used, Frederique van Holthoon (RikiltInstitute of Food Safety) and J.M. Fernandez (SIA Enginyers). This article reflects only the author’s views, and the EU is not liable for any use that may be made of the information contained therein.
SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review January 20, 2009. Accepted April 15, 2009. AC9001386
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009
4295
