Metabolic study of enrofloxacin and metabolic profile modifications in broiler chicken tissues after drug administration

Food Chemistry 172 (2015) 30–39

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

Metabolic study of enrofloxacin and metabolic profile modifications in broiler chicken tissues after drug administration F.J. Morales-Gutiérrez, J. Barbosa, D. Barrón ⇑ Department of Analytical Chemistry, Food and Nutrition Torribera Campus, University of Barcelona, Avda. Prat de la Riba, 171, Sta. Coloma de Gramenet, E-08921 Barcelona, Spain

a r t i c l e

i n f o

Article history: Received 12 March 2014 Received in revised form 2 September 2014 Accepted 6 September 2014 Available online 16 September 2014 Keywords: Chicken tissues Enrofloxacin Metabolites Multivariate data analyses Metabolic profile

a b s t r a c t In this work, the identification and distribution of the metabolites from enrofloxacin (ENR) in liver, kidney and muscle tissues from broiler chickens subjected to a pharmacological treatment was studied. In addition, qualitative analyses of changes in the metabolic profile in those tissues after drug administration were also investigated. As a result, a total of 31 different metabolites from ENR were identified, which ciprofloxacin (CIP) and desethylene-ENR were the major metabolites. After four days of withdrawal period, most of the metabolites were excreted, but residues of ENR and CIP still persisted in tissues at a concentration under the permitted maximum residue limit (MRL). Non-medicated, medicated and post-treatment samples of chicken tissues were clearly clustered according to their metabolite profile by principal component analysis and partial least squares discriminant analysis, which indicates that endogenous metabolites have not returned to their original levels after the withdrawal period. A total of 22 relevant mass features contributing to this separation as potential markers of chicken samples were tentatively identified. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction In modern agricultural practice, veterinary drugs are extensively used and administered as feed additives or via the drinking water as therapeutic, prophylactic and growth promoting agents, but the inappropriate and abusive use of these substances can leave residues in food products from animals. Specifically in the case of antibiotics, the concern about their residues in foodstuff and the misuse in humans has increased as a result of the transfer of antibiotic-resistant bacteria to man, toxicity and allergy problems and their illegal use as growth promoters (Blasco, Torres, & Picó, 2007; Fàbrega, Sánchez-Céspedes, Soto, & Vila, 2008; Stolker & Brunkman, 2005). In the last few years, the concern about the use of veterinary drugs in food-producing animals and their possible negative effects in the health of consumers has made the control of these residues in edible animal tissues mandatory at the EU. Maximum residue limits (MRLs) of antibiotics in foodstuffs of animal origin such as multiple animal tissues were established by the Commission Regulation (EU) No. 37/2010 (Commission Regulation (EU) No. 37/ 2010, 2010) for safe consumption.

⇑ Corresponding author. Tel.: +34 93 4033797/21277; fax: +34 93 4021233. E-mail address: [email protected] (D. Barrón). http://dx.doi.org/10.1016/j.foodchem.2014.09.025 0308-8146/Ó 2014 Elsevier Ltd. All rights reserved.

In the last recent years, the interest in monitoring those regulated compounds has increased. Up to this time, a large number of published articles have been focused on the development and validation of analytical methods to determine target substances and their main metabolites in several matrices (Blasco et al., 2007; Bogialli & Di Corcia, 2009; Clemente, Hermo, Barron, & Barbosa, 2006; Hermo, Nemutlu, Barbosa, & Barron, 2011; Macarov et al., 2012; Marazuela & Bogialli, 2009; MartínezHuélamo, Jiménez-Gámez, Hermo, Barrón, & Barbosa, 2009; Moreno-Bondi, Marazuela, Herranz, & Rodríguez, 2009; RomeroGonzález, Aguilera-Luiz, Plaza-Bolaños, Garrido Frenich, & Martínez Vidal, 2011), but there are few studies focused on the identification and determination of unknown metabolites and degradation products (Hermo, Gómez-Rodríguez, Barbosa, & Barrón, 2013), which could lead to unknown harmful effects for human health. Consequently, the analysis of metabolites and degradation products of food contaminants, especially those that are considered genotoxic or carcinogenic, is of great interest at present. Moreover, researchers have recently indicated the need of studying the influence of antibiotics on the endogenous metabolism to evaluate changes in metabolite levels (Drexler, Reily, & Shipkova, 2011; García-Reyes, Hernando, Molina-Díaz, & Fernández-Alba, 2007). Accordingly, metabolic alterations caused by the use of antibiotics in veterinary and human medicine might be of great interest in the research of new potentially toxic or

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

healthy compounds and to determine possible markers of the pharmaceutical treatments. To the best of our knowledge, only two studies have been previously reported related to the evaluation on metabolome modifications after pharmacological treatment in animals. Sun et al. (2013) evaluated the effects of Penicillin V on the metabolome of rats and more recently, Hermo, Saurina, Barbosa, and Barrón (2014) estimated the metabolome modifications in chicken after amoxicillin administration. Accordingly, due to the scarce information obtained, more efforts need to be taken in evaluating the effect on metabolic changes in animal tissues intended for human consumption after a pharmacological treatment with antibiotics, which could be of great interest in food control and safety applications. Enrofloxacin (ENR) is a widely used quinolone in poultry farms due its high antimicrobial activity against a wide variety of infections in animals. The present study was focused on the identification of residues of metabolites from ENR and the study of their distribution in liver, kidney and muscle tissues from broiler chickens slaughtered in different days after applying a pharmacological treatment with the antibiotic. Thus, besides of the identification of metabolites from ENR, differences in the metabolome caused by the administration of the drug followed by the tentative identification of markers from non-medicated, medicated and post-treatment samples of chicken liver, kidney and muscle tissues after applying the pharmacological treatment with ENR was also carried out. 2. Experimental 2.1. Reagents and materials Unless specified otherwise, all reagents were of analytical grade. Quinolones were purchased from different pharmaceutical firms: enrofloxacin (ENR) from Cenavisa (Reus, Spain), ciprofloxacin (CIP) from Ipsen Pharma (Paris, France) and norfloxacin (NOR), used as internal standard (IS), was supplied by Liade-Boral Quimica (Barcelona, Spain). Formic acid (FA), trifluoroacetic acid (TFA), acetic acid (HAcO), ammonia (NH3), potassium dihydrogenphosphate (KH2PO4), sodium hydroxide (NaOH), methanol (MeOH, HPLC grade) and acetonitrile (MeCN, HPLC and MS grade) were provided from Merck (Darmstadt, Germany). Ammonium acetate (NH4AcO, MS grade) was supplied by Sigma–Aldrich (St. Louis, MO, USA). Ultrapure water was obtained from a MilliQ system from Millipore (Billerica, MA, USA). Solid-phase extraction (SPE) cartridges Isolute ENV+ (3 mL/200 mg) were supplied by Biotage AB (Uppsala, Sweden). 2.2. Preparation of standard and working solutions Individual ENR, CIP and NOR stock solutions were prepared at a concentration of 100 mg L 1 in HAcO 0.050 mol L 1. The working solutions used to spike the chicken tissue samples were prepared from the individual stock solutions by appropriate dilution to obtain concentrations of 1 and 0.5 mg L 1. For the extraction procedures, 0.050 mol L 1 dihydrogenphosphate solution (adjusted to pH 5.0 with NaOH 0.1 mol L 1) and the hydroorganic solution TFA:H2O:MeCN (2:23:75, v/v/v) were also prepared. 2.3. Pharmacological treatment with ENR Chickens were medicated according to the pharmacological administration protocol fit for human consumption. The therapeutic treatment involved a daily dose of 10 mg/kg of ENR dissolved in the chicken drinking water during 4 days. Fresh pre-solutions of

31

the antibiotic and the medicated water were prepared every day just before it is offered to the animals. Four types of samples were analysed. Two male broiler chickens (A1 and A2, non-medicated chickens) randomly selected from the poultry farm were sacrificed and used as blanks, three male broiler chickens (A3, A4 and A5) slaughtered on the second day of the pharmacological treatment (2-day treated), three male broiler chickens (A6, A7 and A8) slaughtered on the fourth day of the pharmacological treatment (4-day treated) and two male broiler chickens (A9 and A10) slaughtered four days after pharmacological treatment ends (post-treatment). Recovery time was chosen according to the administered commercial product specifications, which indicate a withdrawal period of 4 days. All animals were handled and sacrificed according to the ethical protocols of the chicken producer farm. Chicken liver, kidney and muscle tissue samples from the 10 animals were analysed. Meat was minced, homogenised and stored at 20 °C until sample treatment (Section 2.4.1). For each type of tissue (muscle, liver and kidney), three independent replicates of the 10 specimens were analysed. Each extract was injected twice. 2.4. Sample preparation 2.4.1. Medicated animal samples An amount of 4 g (±0.1 mg) of minced chicken muscle or 2 g (±0.1 mg) of minced chicken kidney and liver was introduced into a 50 mL capped polypropylene centrifuge tube, adding then the IS (NOR) at a concentration of 300 lg kg 1 (Macarov et al., 2012; Morales-Gutiérrez, Hermo, Barbosa, & Barrón, 2014). Analytes were extracted with a mixture of 2 mL (1 mL for kidney and liver tissues) of MilliQ water and 20 mL of MeCN (10 mL for kidney and liver tissues). After shaking for 2 min, the mixture was centrifuged at 3500 rpm (5 min). The supernatant was then isolated and the organic solvent (MeCN) was eliminated by evaporation under N2 stream in a TurboVap system at 35 °C until 2 mL (1 mL for kidney and liver tissues) as final volume. After adding 25 mL of 0.050 mol L 1 dihydrogenphosphate solution (12.5 mL for kidney and liver tissues) to the remaining aqueous extract, the resulting mixture was processed by solid phase extraction (SPE). The Isolute ENV+ cartridges were activated with 2 mL of MeOH, 2 mL of MilliQ water and 2 mL of 0.050 mol L 1 dihydrogenphosphate solution at pH 5.0. The mixture was then loaded to the cartridge and washed with 3 mL of dihydrogenphosphate solution at pH 5.0 and 1 mL of MilliQ water, followed by the elution of the analytes with 5 mL of the hydroorganic solution TFA:H2O:MeCN (2:23:75, v/v/v) and 1 mL of MeCN. The produced SPE eluates were evaporated to dryness at 35 °C under N2 stream and reconstituted with 200 lL of MilliQ water (100 lL for kidney and liver tissues). Prior to injection, samples were filtered and stored at 20 °C. 2.4.2. Quantification of ENR and CIP The concentration of ENR and its major metabolite (CIP) were determined in the three studied tissues from the medicated chickens. Chicken liver, kidney and muscle blank samples obtained from a local supermarket in Barcelona (Spain) were firstly screened to ensure that they were free of antibiotics of interest. Blank samples were directly spiked at seven ENR and CIP concentrations levels in the range of 5–250 lg kg 1 and NOR (IS) at a concentration of 300 lg kg 1. Extraction of analytes was accomplished following the same procedure described above (Section 2.4.1). Calibration curves were constructed using analyte/IS peak area ratios versus analyte/IS concentration ratios. 2.5. LC–MS and LC–MS/MS conditions Identification of metabolites and metabolome modifications analyses were performed using an Accela HPLC system from

32

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

Thermo Fisher Scientific (Hemel Hempstead, UK) coupled to a linear ion trap quadrupole-Orbitrap (LTQ-Orbitrap) Velos-Hybrid FT Mass Spectrometer from Thermo Fisher Scientific (Hemel Hempstead, UK), equipped with a heated electrospray ionisation (HESI) interface. Mass spectrometry analyses were carried out on fullscan MS and product ion scan MS/MS modes with a mass range of 100–1000 Da. The resolving power was 30,000 and 15,000 for the full-scan and MS/MS events, respectively. Employing positive ionisation mode, a multiple component detection method was used as a default values of the parameters settings to carry out the different experiments. A source voltage of 3500 V and a capillary temperature of 300 °C were used as main parameters settings. High collision dissociation (HCD) energy of 40–70% was used for the MS/MS experiments. Data acquisition was performed by Xcalibur 2.2 QualBrowser from Thermo Fisher Scientific (Hemel Hempstead, UK). Quantification of ENR and CIP was performed using an HP Agilent Technologies 1100 LC system coupled to an API 3000 triplequadrupole mass spectrometer (PE Sciex) equipped with a turbo ion spray source. Mass spectrometry analyses were carried out on multiple reaction monitoring (MRM) and positive ionisation mode with a dwell time of 200 ms. LC–MS/MS conditions were optimised by direct injection of each quinolone individually at a concentration of 1 mg L 1. Optimised parameters were the following: capillary voltage 4500 V, nebulizer gas (N2) 10 (arbitrary units), curtain gas (N2) 12 (arbitrary units), collision gas (N2) 15 (arbitrary units), declustering potential (DP) 45 V (CIP, ENR) and 42 V (NOR), focusing potential: 200 V, entrance potential: 10 V. Drying gas (N2) was heated to 400 °C and introduced at a flow-rate of 4500 mL min 1. MS/MS product ions were produced by collision activated dissociation (CAD) of selected precursor ions. Two transitions were followed for each quinolone (Hermo, Barron, & Barbosa, 2008): transitions 360 ? 316 ([M + H-CO2]+, CAD energy 29 V) for ENR, 332 ? 314 ([M + H-H2O]+, CAD energy 32 V) for CIP and 320 ? 276 ([M + H-CO2]+, CAD energy 25 V) for NOR were chosen for quantification analyses and 360 ? 342 ([M + H-H2O]+, CAD energy 29 V) for ENR, 332 ? 288 ([M + H-CO2]+, CAD energy 27 V) for CIP and 320 ? 302 ([M + H-H2O]+, CAD energy 30 V) for NOR were chosen for confirmation analyses. Data acquisition and processing was performed by Analyst v. 1.4.2 from Applied Biosystems (Foster City, CA, USA). LC-LTQ-Orbitrap and LC-QqQ separations were carried out using a Waters Symmetry C8 50 mm  2.1 mm i.d. 5 lm (Milford, MA, USA), working at room temperature. In both cases, analyses were performed at a constant flow rate of 0.3 mL min 1 and the same binary solvent system: solvent A, 0.005 mol L 1 NH4AcO adjusted at pH 2.5 with FA and solvent B, MeCN. Gradient system was programmed as follows: initial 3% B, from 0 to 5 min B increased to 25%, from 5 to 6 min B increased to 35%, from 6 to 7 min B increased to 55% and then maintained at this percentage for 1 min. Finally B decreased to 3% in 1.5 min and maintained at this percentage for 3 min.

2.6. Data processing 2.6.1. Identification of metabolites The potential ENR metabolite ions from the medicated chicken tissues samples were performed by two data processing methods. The faster method consisted on calculating the metabolite ions from a default metabolite list in MetWorks 3.0.1 from Thermo Fisher Scientific (Hemel Hempstead, UK). The number of identified metabolites by this method was improved by applying an iterative process, i.e. each found m/z as a potential metabolite was used as a parent compound. Consequently, metabolites from multi-step biotransformation reactions could be identified.

On the other hand, MetWorks 3.0.1 was also used to process the accurate-mass full-scan raw data by multiple mass defect filter (MMDF) (Zhang, Zhang, Ray, & Zhu, 2009; Zhu, Zhang, Ray, Zhao, & Humphreys, 2006). ENR (m/z 360.1718), CIP (m/z 332.1405), the core substructure with m/z 263.0826, formed by the piperazine ring loss, and the glucuronide conjugation of ENR (m/z 536.2039), were used as MDF templates. The MDF window was set to ±40 mDa around the mass defects of the templates over a mass range of ±50 Da around the filter template masses. The use of MMDF technique as a pre-filter enabled the reduction of most of the false-positive peaks (endogenous components) and background interferences. Once the data was filtered, comparison of mass spectral data from medicated and non-medicated samples using mMass-Open Source Mass Spectrometry Tool (Strohalm, Hassman, Košata, & Kodícˇek, 2008; Strohalm, Kavan, Novák, Volny´, & Havlícˇek, 2010), enabled to differentiate the metabolite ions of interest from interference ions in the biological matrix, especially those ions that show a very low intensity by LC–MS. The m/z list obtained from the two methods was then filtered by assigning a molecular formula, ring double bonding equivalent number (RDBE) and isotopic pattern score matching for each compound (Tchoumtchoua, Njamen, Mbanya, Skaltsounis, & Halabalaki, 2013). Confirmation and structure elucidation of the identified metabolites were carried out by the MS/MS spectrum generated by the product ion scan of each ion. 2.6.2. Metabolomic data analysis In the first stage of the data processing, the accurate-mass fullscan raw data from LC-LTQ-Orbitrap analyses were converted to mzXML data files and imported to XCMS bioinformatics tool (Smith, Want, O’Maille, Abagyan, & Siuzdak, 2006; Tautenhahn, Boettcher, & Neumann, 2008) on R platform for peak detection, matching, grouping and retention time alignment of peaks across the samples. For feature (m/z and retention time) detection, centWave algorithm with mass deviation of 4 ppm, peak width (range wmin–wmax) of 10–60, signal to noise (S/N) ratio of 5 and intensity signal threshold of 10,000 counts was employed. XCMS processing was carried out in pairwise treatment combinations (non-medicated vs. medicated, medicated vs. post-treatment and non-medicated vs. post-treatment). Following this, the data from all three treatment combinations were combined using MetaXCMS (Tautenhahn, Patti, Kalisiak, Miyamoto, et al., 2011). Peak lists generated by the software contained all features characterised by its m/z value, retention time and peak abundance (intensity). Peak lists were then exported to csv data files and uploaded to the freely available web platform Metaboanalyst 2.0 (www.metaboanalyst.ca) (Xia, Mandal, Sinelnikov, Broadhurst, & Wishart, 2012). Prior to multivariate data analysis, data were firstly normalised by constant row sum and secondly, Pareto scaled (mean-centred and divided by the square root of standard deviation of each variable). Chemometric analysis included multivariate data analysis using (unsupervised) principal component analysis (PCA) and (supervised) partial least square discriminant analyses (PLS-DA). PCA was first used to investigate general interrelation between groups, including clustering and outliers among the samples. Supervised pattern recognition approach, using PLS-DA, was then applied to the classification of samples based on metabolic profile changes in chicken tissues according to the pharmacological treatment stage. Detection of the most significant features as being responsible for the separation between groups were chosen in accordance with the PLS-DA loading plots and the variable importance in the projection (VIP) and further analysed by a One wayANOVA & Post-hoc test (p < 0.01). Potential markers contributing to the complete separation between groups were then tentatively identified based on the accurate exact mass compared with those in on-line database resources.

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

3. Results and discussion 3.1. Metabolites of ENR in chicken tissues Chicken liver, kidney and muscle samples from medicated animals with ENR (Section 2.3) were analysed with the aim of identifying those metabolites that have been accumulated in the animal tissues after drug administration and the subsequent animal metabolism and therefore, can lead to health risks when pass to humans through the food chain. Furthermore, the distribution of those metabolites in the animal tissues was also investigated. 3.1.1. Identified metabolites Table 1 shows a summary of mass spectral data and main MS/ MS spectrum fragments of the identified metabolites from the chicken liver, kidney and muscle samples from the medicated animals with ENR. Proposed structures for each compound are shown in Fig. 1. Structures were tentatively identified by the MS/MS spectrum generated from each compound, due to the absence of commercial standards since most of them are novel compounds. According to the proposed structure for each compound (Fig. 1), main biotransformation reactions of ENR due to the animal metabolism were piperazine ring and aromatic core transformations. Piperazine ring transformations were mainly based on the addition of different functional groups (oxidation, hydroxylation, formylation, acetylation, N-demethylation and N-desethylation) in positions 1–4 of the ring, the partial or total piperazine ring cleavage and once the ring has been cleaved, the resulted compound undergoes further degradation by the addition of functional groups (acetylation and formylation) in positions 1 and 4 of the piperazine ring. On the other hand, main degradation steps involved in the aromatic core were the hydroxylation in one of the two available positions of the aromatic ring, as well as the replacement of the fluoride for a hydroxyl group. ENR was mainly metabolized in CIP (M4) and desethylene-ENR (M7), originated by the N-desethylation of ENR and the piperazine ring cleavage, respectively. M6, originated by the decarboxylation of the molecule and the addition of a hydroxyl group in the aromatic ring, M15 (N-formylciprofloxacin) and M22 (N-acetylciprofloxacin), formed by the N-desethylation and the subsequent addition of an N-formyl and N-acetyl group, respectively, showed also a great abundance. In previous works, the compounds M1, M3, M4, M7–M10, M12– 15, M20–M26 and M28 were described as microbiological biotransformation products (Karl, Schneider, & Wetzstein, 2006; Lykkeberg, Halling-Sorensen, & Jensen, 2007; Parshikov et al., 2000; Prieto, Moeder, Rodil, Adrian, & Marco-Urrea, 2011; Wetzstein, Schneider, & Karl, 2006), metabolite residues in animal tissues (Anadón et al., 2001; Morales-Gutiérrez et al., 2014) or residues in cow milk after drug administration for human consumption (Junza, Barbosa, Codony, Jubert, Barbosa, & Barron, 2014; Turnipseed, Storey, Clark, & Miller, 2011). In this study, the number of samples, the number of pharmacological treatment days, the number and the quality of the data processing techniques and above all, by including liver and kidney tissues at the study, enabled the identification of 12 new metabolites from ENR. The compound M2, formed by the oxidation of the piperazine ring in the position 3, followed by the N-desethylation and decarboxylation, the compound M6, the compound originated by the methylation of M6 in the position 3 of the piperazine ring (M11), the compounds formed by further degradation of desethylene-ENR (M7) leading to M5 (formed by the replacement of the fluoride for a hydroxyl group), M16 and M17 (formed by the N-formylation in the positions 1 and 4 of the cleaved piperazine ring) and the compounds M27, M29, M19 and M18, formed by consecutive hydroxylations in the positions 2 and 3 of the piperazine ring

33

and the subsequent N-desethylation, to our knowledge, have not been described previously in the literature as metabolites of ENR. In addition, the formation of the glucuronide ester (M31) and taurine conjugation (M30), which have not been described previously in the literature, were also identified as conjugated metabolites of ENR accumulated in the animal tissues. 3.1.2. Distribution of ENR and their main metabolites From the intensity showed by LC–MS analyses and considering that the different identified compounds have a similar level of ionisation since their chemical structure are related, estimation of the distribution (accumulation and elimination) of the main metabolites in the liver, kidney and chicken tissues were carried out (see Supplementary material (1)). Analyses corresponding to the 2-day treated (A3, A4, and A5) and 4-day treated (A6, A7 and A8) animal samples showed that metabolites M1, M2, M4–M7, M14, M24, M25 and above all the administered drug (ENR), were accumulated and had a great abundance in the liver, kidney and muscle tissues. Metabolites M11, M15, M17, M22 and M30 showed also a great abundance in liver and kidney tissues, but were scarcely or non-accumulated in the muscle tissues. Moreover, the effect of the continued medication with the drug became apparent in the metabolite M30, which increased slightly its abundance in the muscle chicken tissue when increasing the days of pharmacological treatment. On the other hand, metabolites M3, M9 and M13 had a great abundance in liver tissue, but were poorly accumulated in kidney and muscle tissues. In contrast, M26 showed a high intensity in muscle and kidney tissues, but was scarcely accumulated in the liver. Analyses corresponding to samples from animals A9 and A10, which correspond to the post-treatment samples with a withdrawal period of four days, showed that the animals metabolized and excreted most of the metabolites. However, the supplied antibiotic (ENR) and the major identified metabolite (CIP) still remain in the animal tissues with a great abundance. Not only CIP, as the major metabolite, but other minor metabolites were also observed in the recovery period samples. Metabolites M1, M2 and M7 were observed with a low abundance in the three studied tissues, except for M7, which showed also a great abundance in liver tissue samples, whereas residues of M15 and M17 were only detected in kidney tissue samples. Besides, M25 and M26 were also detected with a low abundance in muscle chicken tissue samples. 3.2. Quantification of ENR and CIP In order to ensure that the applied withdrawal period of the drug was adequate for the animal to eliminate the antibiotic or at least to decrease the antibiotic level concentration under the permitted MRL established by the European Commission Regulation (EU) No. 37/2010 (Commission Regulation (EU) No. 37/2010, 2010), the determination of ENR and CIP concentrations in the liver, kidney and muscle tissues from the 10 broiler chickens under the pharmacological treatment with ENR (Section 2.3) was carried out by LC-QqQ using the transitions given in Section 2.5 and NOR as the IS (Section 2.4.2). In this study, a withdrawal period of 4 days was chosen (Section 2.3) according to the administered commercial product specifications. The withdrawal period depends on the type of animal and antibiotic and above all, the administered dosage. In addition, the European Medicine Agency (EMA) recommends a withdrawal period within 3–4 days in chickens subjected to a pharmacological treatment with related products of ENR and a dosage of 10 mg/kg per day (European Medicine Agency (EMA)). Quinolones/fluoroquinolones – Article 35 referral – Annex I, II, III, 2010; (European Medicine Agency (EMA)). Baytril 10% oral solution and associated names: Annex I, II, III, IV, 2012) which is the

34

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

Table 1 Mass spectral data and main fragmentation pattern of the identified metabolites from ENR in liver, kidney and muscle tissues from medicated broiler chickens. Compound

RT (min)

m/z

ENR

4.53

360.1718

m/z error (ppm)

Molecular formula

RDBE

Isotopic pattern score (%)

Main MS/MS spectrum fragments 360.1724, 245.1089 263.0826, 177.0821 302.1303, 217.0772 306.1253, 227.0690 332.1410, 231.0566 332.1609, 243.1128 332.1773, 247.0880 334.1559, 245.1083 342.1813, 227.1177 346.1203, 235.0517 346.1565, 245.1089 346.1935, 275.1198 348.1363, 261.1039 348.1362, 217.0411 358.1769, 228.1598 360.1362, 243.0573 344.1404, 259.1236 362.1519, 240.1387 364.1307, 285.0749 364.1312, 218.0489 372.1920, 227.0813 374.1519, 275.0831 374.1521, 231.0569 374.1522, 243.0571 374.1722, 277.0827 376.1676, 244.0920 376.1670, 300.1415 376.1667, 259.1234 388.1303, 320.1048 392.1626, 289.0987 467.1766, 245.1089 536.2035, 257.1091

0.8

[C19H23N3O3F]+

9.5

98.7

+

M1

6.50

263.0826

0.2

[C13H12FN2O3]

8.5

95.9

M2

4.41

302.1299

1.4

[C16H17FN3O2]+

9.5

93.2

M3

3.73

306.1248

0.2

[C15H17FN3O3]+

8.5

92.9

1.2

+

[C17H19FN3O3]

9.5

99.3

+

M4

4.23

332.1405

M5

4.23

332.1605

1.2

[C17H22N3O4]

8.5

95.7

M6

3.14

332.1769

0.2

[C18H23FN3O2]+

8.5

92.6

M7

4.02

334.1561

1.1

[C17H21FN3O3]+

8.5

97.0

9.5

89.9

+

M8

3.87

342.1812

1.1

[C19H24N3O3]

M9

6.52

346.1198

0.5

[C17H17FN3O4]+

10.5

95.5

M10

4.25

346.1561

1.6

[C18H21FN3O3]+

9.5

90.5

3.0

+

8.5

87.7

+

M11

3.44

346.1925

[C19H25FN3O2]

M12

4.30

348.1354

1.1

[C17H19FN3O4]

9.5

92.7

M13

6.34

348.1354

0.5

[C17H19FN3O4]+

9.5

94.8

M14

3.52

358.1761

0.5

[C19H24N3O4]+

9.5

93.5

+

M15

7.13

360.1354

0.0

[C18H19FN3O4]

10.5

96.9

M16

6.87

362.1511

0.1

[C18H21FN3O4]+

9.5

93.1

M17

7.25

362.1511

0.1

[C18H21FN3O4]+

9.5

97.1

0.1

+

9.5

94.1

+

M18

3.93

364.1303

[C17H19FN3O5]

M19

4.41

364.1303

2.0

[C17H19FN3O5]

9.5

87.7

M20

4.38

372.1918

0.0

[C20H26N3O4]+

9.5

91.9

M21

3.17

374.1511

0.4

[C19H21FN3O4]+

10.5

96.2

+

M22

7.26

374.1511

0.1

[C19H21FN3O4]

10.5

96.6

M23

7.43

374.1511

1.2

[C19H21FN3O4]+

10.5

95.7

M24

4.16

374.1710

0.9

[C19H24N3O5]+

9.5

91.5

M25

4.49

376.1667

1.9

[C19H23FN3O4]+

9.5

91.1

+

M26

4.84

376.1667

0.5

[C19H23FN3O4]

9.5

95.2

M27

6.04

376.1667

1.4

[C19H23FN3O4]+

9.5

89.0

M28

6.40

388.1303

1.5

[C19H19FN3O5]+

11.5

93.5

0.6

+

9.5

87.2

M29

4.70

392.1616

[C19H23FN3O5]

+

M30

3.74

467.1759

0.4

[C21H28FN4O5S]

9.5

95.3

M31

4.51

536.2039

1.8

[C25H31FN3O9]+

11.5

85.2

dosage applied to broiler chickens according to the pharmacological administration protocol fit for human consumption. ENR and CIP were quantified from calibration curves constructed for each chicken tissue spiked with different ENR and CIP concentration levels in the range 5–250 lg kg 1 (Section 2.4.2). Each concentration level was prepared and assayed twice. For the analysis of chicken samples, three independent replicate extractions were prepared and injected into the LC–MS system.

342.1618, 316.1823, 286.0990, 257.1088, 245.0719, 223.0515, 217.0770, 204.0328, 274.1344, 260.1201, 245.1093, 231.0933, 286.1188, 268.1087, 263.0830, 245.0725, 314.1302, 288.1510, 268.1446, 245.1089, 314.1495, 296.1398, 287.1028, 261.0875, 312.1710, 301.1224, 276.1146, 261.1037, 314.1497, 296.1391, 289.0980, 263.0825, 324.1709, 298.1916, 268.1079, 239.1178, 328.1096, 305.0811, 287.0706, 258.0678, 328.1463, 302.1667, 285.1279, 257.1086, 326.1872, 316.1465, 302.1673, 290.1308, 330.1252, 305.0940, 279.0785, 265.0626, 330.1252, 263.0832, 247.0518, 230.0490, 314.1867, 287.1032, 265.0865, 243.1132, 342.1258, 318.0894, 301.0867, 272.0838, 316.1107, 302.1758, 290.1384, 268.1339, 344.1413, 316.1099, 268.1338, 263.0833, 344.1246, 326.1140, 321.0886, 298.1190, 344.1245, 306.1255, 263.0836, 236.0596, 328.2020, 297.1472, 257.1282, 241.0970, 346.1568, 314.0942, 289.0988, 286.0992, 356.1412, 300.0786, 272.0836, 243.0570, 356.1409, 315.1022, 286.0993, 259.1121, 356.1615, 328.1667, 303.0983, 285.0878, 358.1234, 332.1777, 305.0942, 261.1042, 358.1239, 344.1415, 330.1617, 315.1748, 348.1355, 332.1400, 314.1288, 287.1181, 360.1365, 348.0995, 342.1262, 330.0885, 374.1515, 348.1720, 334.1560, 314.1495, 449.1663, 344.1766, 316.1818, 288.1504, 360.1723, 342.1615, 316.1819, 286.0988,

Calibration curve parameters are summarised in Table 2 and the mean tissue concentration data of ENR and CIP, for each animal and tissue, are presented in Fig. 2. Samples from animals A3–A8 were diluted 25 times to decrease the concentration within levels of calibration curves range. Considering that the drug concentration in the animal tissues depends on the amount of medicated drinking water consumed by each animal per day, Fig. 2 shows that the three studied tissues

35

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

Fig. 1. Proposed structures for the identified metabolites from ENR in liver, kidney and muscle tissues from medicated broiler chickens.

Table 2 Calibration curve parameters for liver, kidney and muscle chicken tissues spiked with ENR and its main metabolite (CIP). NOR was used as IS at a concentration of 300 lg kg

a

a

a

Compound

Tissue

Slope (±s )

Y-intercept (±s )

Correlation coefficient (r)

ENR

Liver Kidney Muscle

7.4 (±0.5) 3.52 (±0.09) 3.7 (±0.1)

0.5 (±0.1) 0.23 (±0.04) 0.22 (±0.05)

0.98 0.996 0.994

CIP

Liver Kidney Muscle

2.1 (±0.1) 0.696 (±0.008) 1.33 (±0.02)

0.05 (±0.03) 0.015 (±0.003) 0.001 (±0.008)

0.98 0.9994 0.9990

s: Error.

Fig. 2. Concentration of ENR and CIP in liver, kidney and muscle tissues from medicated broiler chickens with ENR.

1

.

36

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

Fig. 3. PCA and PLS-DA score plots for chicken muscle (A and B), liver (C and D) and kidney (E and F). Classification was accomplished by grouping the non-medicated, 2-day treated, 4-day treated and post-treatment classes according to the differences in their metabolic profile.

37

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

presented a high absorption (amount) of ENR during the pharmacological treatment (A3–A8). Results showed also that the absorption of ENR were similar during the pharmacological treatment in kidney and muscle tissues, but absorption of ENR after 2 days was higher than after 4 days of pharmacological treatment in the case of liver tissue. In contrast, CIP presented a similar tissue penetration in liver and kidney tissues, but as Fig. 2 shows, penetration in muscle tissues was considerably lower. Considering the ENR and CIP MRLs established by the European Commission Regulation (EU) No. 37/2010 (Commission Regulation (EU) No. 37/2010, 2010), which are 300 lg kg 1 for liver, 200 lg kg 1 for kidney and 100 lg kg 1 for muscle for the sum of ENR and CIP, as it was expected, concentrations of ENR + CIP in the three studied tissues exceeded by far the permitted levels because the animals were in treatment. In the analyses of samples corresponding to the animals slaughtered four days after the end of the pharmacological treatment and as it was commented in Section 3.1.2, residues of ENR and its major metabolite (CIP) still remain in the animal tissues, but at a concentration of ENR + CIP under 30 lg kg 1 for all cases. Hence, a withdrawal period of four days was enough to decrease the concentrations levels under the MRLs, but it was not sufficient for the animal to metabolize and eliminate the drug and all the metabolites.

3.3. Effect of the pharmacological treatment on the metabolic profile In this work, the existence of differences in the metabolic profile of the chicken tissues in accordance with the pharmacological treatment stage, i.e. non-medicated, medicated and post-treatment animal samples, was evaluated by multivariate data analyses, using pattern recognition techniques (PCA and PLS-DA) as explained in Section 2.6.2.

The initial exploratory analysis, performed by the unsupervised PCA, reveals differences in samples according to the pharmaceutical treatment stage. PCA score plots for muscle (Fig. 3A), liver (Fig. 3C) and kidney (Fig. 3E) chicken tissues shows differentiated clustering between non-medicated, medicated (2-day treated and 4-day treated) and post-treatment samples, although 2-day treated and 4-day treated groups overlapped partially in all cases as expected. Data was then further analysed by PLS-DA to maximise the difference of metabolic profiles and consequently, to obtain better discrimination between groups. The quality of the PLS-DA models was evaluated by 10-fold cross validation using R2 and Q2 parameters. R2 provides a measure of how much variation is represented by the model and Q2 represent the predictive ability parameter. Results of R2 and Q2 indicated that PLS-DA models for each tissue had a high goodness of fit and good predictability ability since they were R2 = 0.983 and Q2 = 0.978 for muscle, R2 = 0.988 and Q2 = 0.971 for liver and R2 = 0.977 and Q2 = 0.944 for kidney. PLSDA score plots for muscle (Fig. 3B), liver (Fig. 3D) and kidney (Fig. 3F) chicken tissues showed also a clear segregation of the samples according to the therapeutic treatment stage, which indicates the existence of differences in the metabolic profile due to the drug administration. As for PCA score plots, 2-day treated and 4-day treated groups overlapped partially for liver and kidney tissues and were close to each other in the case of muscle samples. Thus, classification observed for medicated samples (2-day treated and 4-day treated) indicated a similarity in their metabolic profile, which means that the effect of the pharmacological treatment with the antibiotic on the metabolome of the animal tissues was maintained even increasing the number of days of therapeutic treatment. In accordance with the MS data, sets of features were completely different for the three classes, which means that the overall chemical behaviour of each sample class was very different. This

Table 3 Analytical data of selected features (markers) determined by LC-LTQ-Orbitrap and multivariate data analyses (PLS-DA). Features were selected according VIP values (VIP > 3) and significance level (p < 0.01). [M + H+]+

Error (ppm)

b c

Molecular formula

Tentative identification

VIP

Tissue

Trend NMa vs. Mb

Trend M vs. PTc

3.2 5.0 3.1 3.6 3.1 3.5 3.7 3.3 3.0 11.1 8.5 4.1 8.1 4.5 3.1 7.9 3.0 3.2 9.2 5.1 4.7 6.5 22.3 21.8 17.0 4.0 13.7 3.0

Liver Kidney Muscle Kidney Kidney Kidney Liver Muscle Liver Liver Liver Liver Muscle Liver Liver Muscle Liver Kidney Liver Kidney Muscle Kidney Liver Muscle Kidney Muscle Kidney Muscle

Up Down Up Down Down Down Up Up Up Up Up Up Down Up Up Down Up Down Up Up Up Down Up Up Up Up Up Up

Up Up Down Up Up Up Up Down Up Up Up Up Up Up Up Up Up Up Down Down Down Up Down Down Down Down Down Down

128.0707 134.1178 204.1229 212.0742 216.1230 217.1544

0.7 1.8 0.7 1.1 0.2 1.2

0.85 0.54 0.91 1.08 4.18 1.53

C6H9NO2 C6H15NO2 C9H17NO4 C10H13NO2S C10H17NO4 C10H20N2O3

Unknown Unknown Acetylcarnitine Benzylcysteine Unknown Dipeptide

218.1386 229.1548 231.1703 245.1862 265.1549 269.0881 279.1705 304.1658 307.0441 310.1761 312.1302 332.1409

0.4 0.6 0.1 0.9 0.9 0.2 0.6 0.8 2.4 0.1 0.1 1.2

1.11 2.28 2.39 3.65 3.44 0.93 4.44 3.94 0.58 1.07 1.81 4.22

C10H19NO4 C11H20N2O3 C11H22N2O4 C12H24N2O3 C14H20N2O3 C10H12N4O5 C15H22N2O3 C16H21N3O3 C14H10O8 C15H23N3O4 C12H17N5O5 C17H18FN3O3

Propionylcarnitine Dipeptide Dipeptide Dipeptide Dipeptide Inosine Dipeptide Dipeptide Unknown Dipeptide Dimethylguanosine Ciprofloxacin

342.1456 360.1723

0.6 1.4

6.98 4.53

C13H25N3O4S (+Na) C19H22FN3O3

Tripeptide Enrofloxacin

360.2130

0.2

4.85

C16H29N3O6

Tripeptide

6.82

[M + 2H+]2+

Unknown

841.9228 a

RT (min)

Non-medicated. Medicated. Post-treatment.

38

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39

fact, together with the clearly observed separation between groups, reveal that levels of endogenous metabolites altered as a consequence of drug administration by up and down regulations, were modified 4 days after the end of the veterinary treatment, but they did not returned yet to their original levels (non-medicated samples). This finding suggests that a withdrawal period of 4 days was enough to decrease ENR + CIP levels under the permitted MRLs (Section 3.2), which means that in principle those chicken tissues could be marketed for human consumption, but was not sufficient to remove any sign of metabolic alterations as a consequence of the antibiotic administration. Detection of the most contributing and discriminatory features to the complete separation between groups was carried out from the corresponding PLS-DA loading plots (figure not shown) and the VIP values. In this work, only features with VIP >3 were included for further analyses. In order to select only the most statistically significant features, a one way-ANOVA & Post-hoc test (p < 0.01) was then performed. Following the detection, markers were tentatively identified following the same procedure accomplished by Cajka, Danhelova, Zachariasova, Riddellova, and Hajslova (2013). Taking advantage of the accurate exact mass data provided by the robust LC–MS analysis, some parameters such as deviation from the calculated mass (4 ppm) and the isotopic pattern score were used to evaluate the accuracy of possible molecular formulas. The presumed molecular formula for each selected feature were then searched and compared with metabolites reported in different on-line databases, such as Human Metabolome DataBase (HMDB), Madison Metabolomics Consortium Database (MMCD), Metlin, LipidMaps and many others. The tentative identification of these markers, regarded as the differentiating metabolites between non-medicated, medicated and post-treatment samples, as well as their trends (up and down regulation) for liver, kidney and muscle chicken tissues are summarised in Table 3. As results collected in Table 3 show, most of the endogenous metabolites influenced by the antibiotic administration for the three tissues were tentatively identified as oligopeptides (di- and tripeptides). On the other hand, it was also noticed that trends (up and down regulation) were in general common according to the studied tissue, except for ENR and CIP. The most significant markers tentatively identified for liver tissue (dipeptides) were moderately increased as a consequence of the antibiotic administration and were strongly increased after medication with ENR stopped, except for ENR and CIP, which increased their level during the medication and decreased their levels after the medication stopped, as a consequence of the animal metabolism to eliminate the drug and the related metabolites of ENR. Otherwise, markers tentatively identified for kidney tissue were down regulated during the medication with the antibiotic and increased their levels once the pharmacological treatment ended, except for ENR, CIP and the compound with m/z 360.2130 tentatively identified as a tripeptide, which showed just the opposite behaviour. In contrast, the most contributing features detected in the muscle tissues were up regulated during the veterinary treatment and down regulated after medication, including ENR, CIP and the compound with m/z 360.2130.

4. Conclusions The combination of high-resolution mass spectrometry with different data processing techniques permitted to identify 31 metabolites as residues of ENR in the animal tissues, which 12 of them have not been described before. Besides the identification of new metabolites of the antibiotic, mass spectrometry analyses were also successfully employed to estimate the distribution (accumulation and elimination) of these metabolites in the

different chicken tissues. On the other hand, MS data combined with powerful pattern recognition techniques (PCA and PLS-DA), showed the existence of differences in the metabolome of the animal tissues due to the therapeutic treatment with the antibiotic. A withdrawal period of 4 days was enough to decrease the levels of ENR and its major metabolite (CIP) under the permitted MRLs, but considering that levels of endogenous metabolites altered as a consequence of drug administration did not returned to their original levels, was not sufficient to eliminate the signs of metabolic alterations. In addition, 22 markers as the most contributing features for the separation between non-medicated, medicated and post-treatment samples were tentatively identified. Acknowledgements The authors are gratefully acknowledged for the financial support of the Spanish Ministry of Spanish Government (Project CTQ2010-19044/BQU). We also wish to acknowledge M. Rey and A. Villa, from PONDEX S.A. poultry farm, Juneda (Lleida), for kind donation of medicated chicken samples. F.J. Morales-Gutiérrez would like to thank the APIF fellowship from the University of Barcelona. Appendix A. Supplementary data Intensity showed by the LC–MS analyses for each identified metabolite. PLS-DA loading plots for chicken muscle (A), liver (B) and kidney (B) tissues. Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodchem.2014. 09.025. References Anadón, A., Martínez-Larrañaga, M. R., Iturbe, J., Martínez, M. A., Díaz, M. J., Frejo, M. T., et al. (2001). Pharmacokinetics and residues of ciprofloxacin and its metabolites in broiler chickens. Research in Veterinary Science, 71, 101–109. Blasco, C., Torres, C. M., & Picó, Y. (2007). Progress in analysis of residual antibacterials in food. Trends in Analytical Chemistry, 26, 895–913. Bogialli, S., & Di Corcia, A. (2009). Recent applications of liquid chromatography mass spectrometry to residue analysis of antimicrobials in food of animal origin. Analytical and Bioanalytical Chemistry, 395, 947–966. Cajka, T., Danhelova, H., Zachariasova, M., Riddellova, K., & Hajslova, J. (2013). Application of direct analysis in real time ionization-mass spectrometry (DARTMS) in chicken meat metabolomics aiming at the retrospective control of feed fraud. Metabolomics. http://dx.doi.org/10.1007/s11306-013-0495-z. Clemente, M., Hermo, M. P., Barron, D., & Barbosa, J. (2006). Confirmatory and quantitative analysis using experimental design for the extraction and liquid chromatography–UV, liquid chromatography–mass spectrometry and liquid chromatography–mass spectrometry/mass spectrometry determination of quinolones in turkey muscle. Journal of Chromatography A, 1135, 170–178. Commission Regulation (EU) No. 37/2010 (2010). Annexes I to IV to Council Regulation (EEC) No. 2377/90 on pharmacologically active substances and their classification regarding maximum residue limits in foodstuffs of animal origin. Official Journal of the European Union 18.02.2014, L 15/1. (http://europa.eu.int). Drexler, D. M., Reily, M. D., & Shipkova, P. A. (2011). Advances in mass spectrometry applied to pharmaceutical metabolomics. Analytical and Bioanalytical Chemistry, 399, 2645–2653. European Medicine Agency (EMA). Quinolones/fluoroquinolones – Article 35 referral – Annex I, II, III, 2010 URL: http://www.ema.europa.eu/docs/en_GB/ document_library/Referrals_document/quinolones_35/WC500094631.pdf. Accessed 07.02.2014. European Medicine Agency (EMA). Baytril 10% oral solution and associated names: Annex I, II, III, IV, 2012. URL: http://www.ema.europa.eu/docs/en_GB/ document_library/Referrals_document/Baytril_34/WC500134883.pdf. Accessed 07.02.2014. Fàbrega, A., Sánchez-Céspedes, J., Soto, S., & Vila, J. (2008). Quinolone resistance in the food chain. International Journal of Antimicrobial Agents, 31, 307–315. García-Reyes, J. F., Hernando, M. D., Molina-Díaz, A., & Fernández-Alba, A. R. (2007). Comprehensive screening of target, non-target and unknown pesticides in food by LC-TOF–MS. Trends in Analytical Chemistry, 26, 828–841. Hermo, M. P., Barron, D., & Barbosa, J. (2008). Determination of multiresidue quinolones regulated by the European Union in pig liver samples. Journal of Chromatography A, 1201, 1–14.

F.J. Morales-Gutiérrez et al. / Food Chemistry 172 (2015) 30–39 Hermo, M. P., Gómez-Rodríguez, P., Barbosa, J., & Barrón, D. (2013). Metabolomic assays of amoxicillin, cephapirin and ceftiofur in chicken muscle: Application to treated chicken samples by liquid chromatography quadrupole time-of-flight mass spectrometry. Journal of Pharmaceutical and Biomedical Analysis, 85, 169–178. Hermo, M. P., Nemutlu, E., Barbosa, J., & Barron, D. (2011). Multiresidue determination of quinolones regulated by the European Union in bovine and porcine plasma: Application of chromatographic and capillary electrophoretic methodologies. Biomedical Chromatography, 25, 555–569. Hermo, M. P., Saurina, J., Barbosa, J., & Barron, D. (2014). High-resolution mass spectrometry applied to the study of metabolome modifications in various chicken tissues after amoxicillin administration. Food Chemistry, 153, 405–413. Junza, A., Barbosa, S., Codony, M. R., Jubert, A., Barbosa, J., & Barron, D. (2014). Identification of metabolites and thermal transformation products of quinolones in raw cow’s milk by liquid chromatography coupled to highresolution mass spectrometry. Journal of Agricultural and Food Chemistry, 62, 2008–2021. Karl, W., Schneider, J., & Wetzstein, H. G. (2006). Outlines of an ‘‘exploding’’ network of metabolites generated from the fluoroquinolone enrofloxacin by the brown rot fungus Gloeophyllum striatum. Applied Microbiology and Biotechnology, 71, 101–113. Lykkeberg, A., Halling-Sorensen, B., & Jensen, L. (2007). Susceptibility of bacteria isolated from pigs to tiamulin and enrofloxacin metabolites. Veterinary Microbiology, 121, 116–124. Macarov, C. A., Tong, L., Martínez-Huélamo, M., Hermo, M. P., Chirila, E., Wang, Y. X., et al. (2012). Multi residue determination of the penicillins regulated by the European Union, in bovine, porcine and chicken muscle, by LC–MS/MS. Food Chemistry, 135, 2612–2621. Marazuela, M. D., & Bogialli, S. (2009). A review of novel strategies of sample preparation for the determination of antibacterial residues in foodstuffs using liquid chromatography-based analytical methods. Analytical Chimica Acta, 645, 5–17. Martínez-Huélamo, M., Jiménez-Gámez, E., Hermo, M. P., Barrón, D., & Barbosa, J. (2009). Determination of penicillins in milk using LCUV, LC–MS and LC–MS/MS. Journal of Chromatography A, 32, 2385–2393. Morales-Gutiérrez, F. J., Hermo, M. P., Barbosa, J., & Barrón, D. (2014). Highresolution mass spectrometry applied to the identification of transformation products of quinolones from stability studies and new metabolites of enrofloxacin in chicken muscle tissues. Journal of Pharmaceutical and Biomedical Analysis, 92, 165–176. Moreno-Bondi, M. C., Marazuela, M. D., Herranz, S., & Rodríguez, E. (2009). An overview of sample preparation procedures for LC–MS multiclass antibiotic determination in environmental and food samples. Analytical and Bioanalytical Chemistry, 395, 921–946. Parshikov, I. A., Freeman, J. P., Lay, J. O., Beger, R. D., Williams, A. J., & Sutherland, J. B. (2000). Microbiological transformation of enrofloxacin by the fungus Mucor ramannianus. Applied and Environmental Microbiology, 66, 2664–2667. Prieto, A., Moeder, M., Rodil, R., Adrian, L., & Marco-Urrea, E. (2011). Degradation of the antibiotics norfloxacin and ciprofloxacin by a white-rot fungus and

39

identification of degradation products. Bioresource Technology, 102, 10987–10995. Romero-González, R., Aguilera-Luiz, M. M., Plaza-Bolaños, P., Garrido Frenich, A., & Martínez Vidal, J. L. (2011). Food contaminant analysis at high resolution mass spectrometry: Application for the determination of veterinary drugs in milk. Journal of Chromatography A, 1218, 9353–9365. Smith, C. A., Want, E. J., O’Maille, G., Abagyan, R., & Siuzdak, G. (2006). XCMS: Processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Analytical Chemistry, 78, 779–787. Stolker, A. A. M., & Brunkman, U. A. Th. (2005). Analytical strategies for residue analysis of veterinary drugs and growth-promoting agents in food-producing animals-a review. Journal of Chromatography A, 1067, 15–53. Strohalm, M., Hassman, M., Košata, B., & Kodícˇek, M. (2008). mMass data miner: An open source alternative for mass spectrometric data analysis. Rapid Communications in Mass Spectrometry, 22, 905–908. Strohalm, M., Kavan, D., Novák, P., Volny´, M., & Havlícˇek, V. (2010). mMass 3: A cross-platform software environment for precise analysis of mass spectrometric data. Analytical Chemistry, 82, 4648–4651. Sun, J., Schnackenberg, L. K., Khare, S., Yang, X., Greenhaw, J., Salminen, W., et al. (2013). Evaluating effects of penicillin treatment on the metabolome of rats. Journal of Chromatography B, 932, 134–143. Tautenhahn, R., Boettcher, C., & Neumann, S. (2008). Highly sensitive feature detection for high resolution LC/MS. BMC Bioinformatics, 9 (No. pp. given). Tautenhahn, R., Patti, G. J., Kalisiak, E., Miyamoto, T., et al. (2011). MetaXCMS: Second-order analysis of untargeted metabolomics data. Analytical Chemistry, 83, 696–700. Tchoumtchoua, J., Njamen, D., Mbanya, J. C., Skaltsounis, A., & Halabalaki, M. (2013). Structure-oriented UHPLC-LTQ Orbitrap-based approach as a dereplication strategy for the identification of isoflavonoids from Amphimas pterocarpoides crude extract. Journal of Mass Spectrometry, 48, 561–575. Turnipseed, S. B., Storey, J. M., Clark, S. B., & Miller, K. E. (2011). Analysis of veterinary drugs and metabolites in milk using quadrupole time-of-flight liquid chromatography–mass spectrometry. Journal of Agricultural and Food Chemistry, 59, 7569–7581. Wetzstein, H. G., Schneider, J., & Karl, W. (2006). Patterns of metabolites produced from the fluoroquinolone enrofloxacin by Basidiomycetes indigenous to agricultural sites. Applied Microbiology and Biotechnology, 71, 90–100. Xia, J., Mandal, R., Sinelnikov, I., Broadhurst, D., & Wishart, D. S. (2012). MetaboAnalyst 2.0 – a comprehensive server for metabolomic data analysis. Nucleic Acids Research, 40, W127–W133. Zhang, H., Zhang, D., Ray, K., & Zhu, M. (2009). Mass defect filter technique and its applications to drug metabolite identification by high-resolution mass spectrometry. Journal Mass Spectrometry, 44, 999–1016. Zhu, M., Zhang, D., Ray, K., Zhao, W., Humphreys, W. G., et al. (2006). Detection and characterization of metabolites in biological matrices using mass defect filtering of liquid chromatography/high resolution mass spectrometry data. Drug Metabolism and Disposition, 34, 1722–1733.