Multi-residue Determination of Polar Veterinary Drugs in Livestock and Fishery Products by

Transcription

1 Multi-residue Determination of Polar Veterinary Drugs in Livestock and Fishery Products by Liquid Chromatography/Tandem Mass Spectrometry Maki Kanda, Takayuki Nakajima, Hiroshi Hayashi, Tsuneo Hashimoto, Setsuko Kanai, Chieko Nagano, Yoko Matsushima, Yukinari Tateishi, Soichi Yoshikawa, Yumi Tsuruoka, Takeo Sasamoto, and Ichiro Takano Tokyo Metropolitan Institute of Public Health, , Hyakunin-cho, Shinjuku-ku, Tokyo , Japan Abstract Residues of 37 polar veterinary drugs belonging to six families (quinolones, tetracyclines, macrolides, lincosamides, sulfonamides, and others) in livestock and fishery products were determined using a validated LC-MS/MS method. There were two key points in sample preparation. First, extraction was performed with two solutions of different polarity. Highly polar compounds were initially extracted with Na2EDTA-McIlvaine s buffer (ph 7.0). Medium polar compounds were then extracted from the same samples with acetonitrile containing 0.1% formic acid. Secondly, cleanup was performed using a single SPE polymer cartridge. The first extracted solution was applied to the cartridge. Highly polar compounds were retained on the cartridge. Then, the second extracted solution was applied to the same cartridge. Both highly and medium polar compounds were eluted from the cartridge. This method satisfied the guideline criteria for 37 out of 37 drugs in swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, milk, and

2 honey; 35 out of 37 in egg; and 34 out of 37 in flounder. The LOQ ranged from 0.1 to 5 µg/kg. Residues were detected in 24 out of 110 samples and analyzed using the validated method. Introduction Veterinary drugs are widely used on farms to treat and prevent diseases. However, over-dosing and noncompliance with the withdrawal period may cause drug residues to remain in animal tissues (1, 2). Drug-contaminated livestock and fishery products may have a potential risk for the consumer s health because they can provoke drug-resistant pathogenic strains of bacteria, allergic reactions, and toxicity (3, 4). Therefore, it is necessary to monitor livestock and fishery products for the residual veterinary drugs using accurate analysis. We have used two major analytical strategies to measure residual substances, namely, microbiological screening (5 7) and screening using LC- MS/MS (8, 9). However, the sensitivity of microbiological screening was insufficient to detect residual levels of multi-class veterinary drugs. Moreover, when positive results were found with microbiological methods, specific chromatographic analyses were needed to identify the antibiotics. The identifying process was so complicated that it was difficult to identify each residual drug. The accuracy of analysis for the residual drugs has been required worldwide in recent years. In Japan, the analytic methodologies used by inspection institutes had to be validated until December 13, 2013 according to the notice issued by the Japanese Ministry of Health, Labour, and Welfare (10, 11). On the other hand, the simultaneous analysis methodologies for multi-class veterinary drug residues using LC-MS/MS have already been reported (8, 9, 12 29). However, the trueness and precision of reported analysis using LC-MS/MS (8, 9, 12 23, 25, 27) for fluoroquinolones (FQs),

3 tetracyclines (TCs), penicillins (PCs), 5-hydroxythiabendazole, and clopidol did not achieve acceptable values according to the Guidelines for the Validation of Analytical Methods for Residual Agricultural Chemicals in Food. Furthermore, the sensitivity of some analysis was insufficient to detect residual levels of multi-class veterinary drugs (12, 17 19, 22, 23, 25, 27). On the Japanese positive list system, veterinary drugs of which no established maximum residues limits (MRLs) were given the default regulatory limit (uniform limit of level) at 10 µg/kg. Therefore, the analysis of multi-class drugs needs the LOQ for each drug to be less than 10 µg/kg. Residues of TCs and FQs have been reported frequently in analyses performed by national institutions in Japan or in the European Union (EU; 30, 31). TC residues were found in swine muscle, fish, and honey. The residues of enrofloxacin were found in shrimp from Asia. Therefore, we need analytical methods to accurately measure the residue concentrations of these drugs. The aim of this study was to determine residues of 37 polar veterinary drugs belonging to six families [quinolones (QLs), TCs, macrolides (MLs), lincosamides, sulfonamides (SDs), and others] in livestock and fishery products using a validated LC-MS/MS method. By addressing the following five points, we improved pretreatment procedures and LC-MS/MS conditions: (1) Simple and rapid analysis is desirable to speed up large amounts of sample inspections. (2) Polar veterinary drugs must be simultaneously extracted from livestock and fishery products. We attempted to use aqueous solvent on the first extraction and then organic solvent on the second extraction. Different pretreatment procedures, such as quick, easy, cheap, effective, rugged, and safe (QuEChERS) methods (8, , 26) or pressurized liquid extraction (PLE) were used recently. By using acetonitrile in the QuEChERS method, extraction of TCs, MLs, and FQs was insufficient

4 (8, 9, 12, 14, 15, 18, 26). By using other extraction solutions i.e., acidified acetonitrile (13 16, 26), methanol (12, 18), or methanol-acetonitrile (10, 17), extraction of these drugs was insufficient as well. As shown in Table 1, log P of these drugs was negative, which means that these drugs were soluble in the aqueous phase. Actually, a mixture of water and organic solvent was used (19, 20, 21, 23, 32). Using water at PLE was significantly more effective for the extraction of QLs, PC V, and SDs (25 29). (3) During the measurement by LC-MS/MS, the matrix interferes with the ionization of the target compounds, which precludes the quantification. The matrix interference from livestock and fishery products is removed by a cleanup using the SPE polymer cartridge. (4) To increase the sensitivity, LC conditions (mobile phase, column, and injection volume) and MS/MS parameters were modified. (5) The analytical method developed in this study was validated in 10 livestock and fishery products: swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, flounder, milk, egg, and honey in accordance with the Japanese guidelines. Experimental Samples Livestock and fishery products (swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, flounder, milk, egg, and honey) were purchased from local supermarkets in Japan and were confirmed to be free of the targeted analytes in this study. The tissues were minced with an electric household food processor and stored at 20 C.

5 Apparatus (a) LC system. LC-20A series (Shimadzu Corp., Kyoto, Japan). (b) MS system. API 5500 Qtrap mass spectrometer with an electrospray ionization (ESI) interface and Analyst (Version 1.4.2) software (AB Sciex, Framingham, MA). (c) LC column. Triart C18 column ( mm, 5 µm particle size) (YMC Co. Ltd, Kyoto, Japan). (d) Mixer. Vortex-Genie 2 (Scientific Industries Inc., Bohemia, NY). (e) Ultrasonic machine. B5510J-DTH (Branson, Danbury, CT). (f) Centrifuge. AX-320 (Tomy Seiko Co., Tokyo, Japan). (g) Microcentrifuge. 5415R (Eppendorf Co. Ltd, Hamburg, Germany). (h) Polypropylene centrifuge tubes. 15 ml and 50 ml (Corning Inc., Corning, NY). (i) Glass volumetric flasks. 50 and 100 ml (SIBATA Scientific Technology Ltd, Saitama, Japan). (j) Polymethylpentene and opaque volumetric flasks. 10 ml (VITLAB GmbH, Grossostheim, Germany). (k) SPE manifold system. Vacuum manifold system (GL Sciences Inc., Tokyo, Japan). (l) SPE polymer cartridges for the cleanup procedure. InertSep TM PLS-3 cartridge, 20 cc/200 mg (GL Sciences Inc.). Before use, the PLS-3 cartridges were conditioned with 5 ml acetonitrile, and then 5 ml Na 2 EDTA-McIlvaine s buffer solution (ph 7.0). (m) Microtubes. 1.5 ml (Eppendorf Co. Ltd). (n) Polypropylene and amber vial tubes. 300 µl (GL Sciences Inc.).

6 Reagents (a) Water. Obtained using a Milli-Q system (Millipore Corp., Billerica, MA). (b) Solvent. Acetonitrile (LC grade), hexane (for pesticide residue and polychlorinated biphenyl analysis grade) and methanol (LC grade; Wako Pure Chemical Industries Ltd, Osaka, Japan). (c) Formic acid (99%). LC-MS grade (Wako Pure Chemical Industries Ltd). (d) Citric acid monohydrate, Na 2 EDTA, sodium chloride, and anhydrous magnesium sulfate. Analytical grade (Wako Pure Chemical Industries Ltd). (e) Disodium hydrogen phosphate dihydrate. Analytical grade (Merck KGaA, Darmstadt, Germany). (f) Polar extraction solution 1; Na 2 EDTA-McIlvaine s buffer solution (ph 7.0). Prepared by dissolving g disodium hydrogen phosphate dihydrate, 2.73 g citric acid monohydrate, and g Na 2 EDTA in water and diluting to 1 L. (g) Polar extraction solution 2; Acetonitrile containing 0.1% formic acid. Freshly prepared by mixing 0.1 ml of formic acid with 100 ml of acetonitrile. (h) Standard (purity grade). Marbofloxacin (98.0%), norfloxacin (98.0%), ciprofloxacin (98.0%), difloxacin (98.0%), flumequine (98.0%), oxytetracycline (99.0%), erythromycin A (98.0%), sulfadiazine (99.0%), sulfathiazole (98.0%), sulfamonomethoxine (99.0%), sulfamethoxazole (99.0%), sulfadimethoxine (99.0%), 5-hydroxythiabendazole (98.0%), clopidol (98.0%), and thiabendazole (99.0%) were purchased from Wako Pure Chemical Industries Ltd Ofloxacin (97.7%), orbifloxacin (99.6%), and lincomycin A (98.0%) were from Hayashi Pure Medical Industry (Osaka, Japan). Danofloxacin (100.0%), enrofloxacin (99.8%), oxolinic acid (98.8%), nalidixic acid (99.8%), oleandomycin (96.5%), josamycin (86.8%), sulfamerazine (99.5%),

7 sulfadimidine (99.4%), and sulfaquinoxaline (99.6%) were from Kanto Chemical Co. (Tokyo, Japan). Sarafloxacin (97.3%), tetracycline (97.7%), chlortetracycline (99.1%), doxycycline (98.2%), and tiamulin (99.9%) were from Sigma-Aldrich (St. Louis, MO). Spiramycin (96.0%), tilmicosin (98.5%), and tylosin (98.0%) were from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Pirlimycin (86.6%) was from Pfizer Japan Inc. (Tokyo, Japan). Mirosamicin (97.7%) was from Kyoritsu Pharmaceutical Co. (Tokyo, Japan). (i) Internal standard (IS). Demeclocycline (92.3%) was from Hayashi Pure Medical Industry. Preparation of Standard Solutions and Calibration Standards (a) Stock standard solutions of 33 individual compounds except TCs (100 µg/ml). Stock standard solutions were prepared individually. The suitable quantity of standard taking into account the substance purity was weighed in a 50 ml glass volumetric flask. Clopidol was dissolved in 5 ml acetonitrile, and made up to 50 ml with methanol. Sulfadimidine and oxolinic acid were dissolved in acetonitrile, and made up to 50 ml with acetonitrile. The rest of compounds were dissolved in methanol, and made up to 50 ml with methanol. Stock standard solutions were kept in amber glass vials in the dark at 4 C, under which conditions, they were stable for one year. (b) Mixed standard solutions except TCs (1 µg/ml). An aliquot (500 µl) of each stock standard solution shown in (a) was transferred and mixed together in a 50 ml glass volumetric flask, and made up to 50 ml with methanol. This mixed standard solution was kept in an amber glass vial in the dark at 4 C, under which conditions this was stable for 3 months. (c) Stock standard solutions of 4 TCs (1000 µg/ml). Stock standard solutions of TCs (oxytetracycline, tetracycline, chlortetracycline and doxycycline) were prepared individually. The

8 suitable quantity of standard taking into account the substance purity was weighed in a 10 ml opaque polymethylpentene volumetric flask (light-shielding). TCs were dissolved in methanol and made up to 10 ml with methanol. The stock standard solutions were kept in polypropylene vials in the dark at 20 C, under which conditions they were stable for 1 month. (d) Mixed oxytetracycline and tetracycline standard solution (1 µg/ml). An aliquot (100 µl) of each stock standard solution of oxytetracycline and tetracycline shown in (c) was transferred and mixed together in a 10 ml opaque polymethylpentene volumetric flask, and made up to 10 ml with acetonitrile containing 0.1% formic acid (ACN/FA) immediately before use. This solution was diluted 10 times with ACN/FA. (e) Working standard solutions for 35 veterinary drugs (except for chlortetracycline and doxycycline) (from to 0.1 µg/ml). Working standard solutions were prepared immediately before use by serial dilution of each mixed standard solution shown in (b) and (d) with ACN/FA. (f) Matrix-matched standard solutions for 35 veterinary drugs (from to 50 ng/ml). Calibration curves for 35 veterinary drugs (except chlortetracycline and doxycycline) were obtained from matrix-matched calibration samples. Blank samples were prepared as described in the Sample Preparation section. Matrix-matched standard solutions were prepared by mixing an aliquot (500 µl) of blank solution and the appropriate volume of working standard solutions shown in (e), and then made up to 1 ml with ACN/FA, e.g., a ng/ml solution was made by mixing an aliquot (500 µl) of blank solution and the working standard solution (0.001 µg/ml, 25 µl), and then made up to 1 ml. (g) Mixed chlortetracycline and doxycycline standard solution (1 µg/ml). An aliquot (100 µl) of each stock standard solution of chlortetracycline and doxycycline shown in (c) was transferred

9 and mixed together in a 10 ml opaque polymethylpentene volumetric flask, and made up to 10 ml with ACN/FA immediately before use. This solution was diluted 10 times with ACN/FA. (h) Working standard solutions for chlortetracycline and doxycycline (from to 0.1 µg/ml). Working standard solutions were prepared immediately before use by serial dilution of the mixed standard solution shown in (g) with ACN/FA. (i) IS. Demeclocycline was the IS for the quantification of chlortetracycline and doxycycline. Demeclocycline (10.9 mg) was accurately weighed in a 10 ml opaque polymethylpentene volumetric flask, dissolved in methanol, and made up to 10 ml with methanol. The stock IS solution (1000 µg/ml) was kept in polypropylene vials in the dark at 20 C, under which conditions the solution was stable for 1 month. Working IS solutions (from 0.01 to 1 µg/ml) were prepared immediately before use by serial dilution of the stock IS solution with ACN/FA. (j) IS calibration standard solutions for chlortetracycline and doxycycline (from to 50 ng/ml). Calibration curves for chlortetracycline and doxycycline were obtained from IS calibration samples. IS calibration standard solutions were prepared by mixing the working IS solution shown in (i) (0.01 µg/ml, 100 µl) and the appropriate volume of working solutions shown in (h), and made up to 1 ml with ACN/FA, e.g., a ng/ml solution was made by mixing the working IS standard solution (0.01 µg/ml, 100 µl) together with the working standard solution for chlortetracycline and doxycycline (0.001 µg/ml, 25 µl), and then brought to 1 ml volume. LC Separation Conditions (a) Mobile phase. The 0.05% formic acid solution was prepared by mixing 0.5 ml of formic acid with 1 L water. (A) The 0.05% formic acid solution and (B) acetonitrile were mixed using the

10 pump in gradient mode as follows: 5% B (3 min); 5 90% B (12 min); 90% B (5 min); 90-5% B (0.1 min); and 5% B (5 min). (b) Flow rate. 0.3 ml/min. (c) Column temperature. 40 C. (d) Injection volume. 2 µl. MS/MS Conditions (a) Ionization mode. Positive-ion ESI. (b) Ion spray voltage V. (c) Source temperature. 650 C. (d) Entrance potential. 10 V. (e) Curtain gas pressure. 20 psi (nitrogen). (f) Collision gas pressure. 7 psi (nitrogen). (g) Ion source gas pressure psi (nitrogen). (h) Ion source gas pressure psi (nitrogen). (i) Acquisition function. Selected reaction monitoring (SRM); the SRM program is shown in Table 2. Sample Preparation The schematic procedure of sample preparation is shown in Figure 1. For the sample preparation, glass vessels were not used, because silica in the glass could make an interference signal during LC- MS/MS analysis of TCs.

11 Thoroughly minced sample (5.0 g) was poured in 50 ml polypropylene centrifuge tubes (A). IS was spiked at a level of 10 µg/kg. Na 2 EDTA McIlvaine s buffer (ph 7.0, 20 ml) was added. The tube (A) was vortexed for 1 min. A 5 ml amount of hexane was added. The tube (A) was vortexed again for 1 min, ultrasonicated for 10 min, and then centrifuged at g for 20 min at 4 C. The hexane layer was discarded by pipetting. Hexane washing was used at all sample types to ease the operations. As shown in First extraction of Figure 1, the Na 2 EDTA McIlvaine s buffer layer was transferred into new 50 ml polypropylene centrifuge tubes containing 1 ml of 25% NaCl solution (B). The tube (B) was vortexed for 1 min., and then centrifuged at 9600 g for 10 min at 4 C. The supernatant was loaded to the conditioned PLS-3 cartridge at approximately 1 ml/min. The target compounds were retained on the cartridge, while the solution containing the matrix of food was passed through the cartridge. The cartridge was washed with 5 ml of water, and then vacuum-dried for 3 min at a pressure of 10 mm Hg. In addition, the second extraction from the remaining substance in tube (A) was performed as shown in the Second extraction stage of Figure 1. The characteristics of the remaining matrixes were varied and depended on the different type of samples, as well as the pellets or the insoluble matrix floating on the top of the hexane layer. The following procedure was used for all sample types. Water (2 ml) was added to the tube (A) and then (A) was vortexed. Subsequently, 10 ml ACN/FA was added. The tube (A) was vortexed again for 1 min, and ultrasonicated for 10 min. Magnesium sulfate was added for dehydration (3 g each for bovine muscle, swine muscle, chicken muscle, prawn, milk, and honey). A 4 g amount of magnesium sulfate was added for salmon, red sea bream, and flounder; 5g was added for egg. Then tube (A) was vigorously shaken for 1 min, and centrifuged at 1800 g for 10 at 4 C.

12 As shown as the black arrow in Figure 1, the organic phase was used as the elution solution for the PLS-3 cartridge previously loaded with Na 2 EDTA McIlvaine s buffer layer. The eluate from the cartridge was collected into an opaque polymethylpentene volumetric flask. The eluate was made up to 10 ml with ACN/FA. An aliquot (1 ml) was transferred to a microtube, diluted to 2-fold with ACN/FA, and centrifuged at g for 5 min at 4 C. The supernatant was transferred into an amber polypropylene vial tube. The resultant solution was analyzed by LC-MS/MS twice on the same day. Each quantitative value was taken as a mean of two measurements. Single-Laboratory Validation Tests with Spiked Samples The method was validated according to the guidelines of the Japanese Ministry of Health, Labour, and Welfare. Selectivity was confirmed by analyzing blank samples. Trueness, repeatability (RSDr), and within-run reproducibility (RSD WR ) were determined by means of the recoveries using samples spiked with 37 veterinary drugs and demeclocycline at levels of 10 or 100 µg/kg, performed with two samples per day over five different days. LOQs and LODs were estimated from the repeatability data of the blank samples spiked with 0.1, 0.25, 0.5, 1, 2.5, and 5 µg/kg for each of the 37 veterinary drugs examined. LOQs were calculated as 10 times the SD, and LODs were calculated as 3 times the SD using the Analyst software (AB Sciex). Results and Discussion LC-MS/MS Parameters The MS scans of the 37 veterinary drugs revealed that the most abundant molecular ion was the protonated molecule [M+H] +. As each [M+H] + is a precursor ion, a further MS/MS scan was

13 performed after the collision energy was increased. Two fragment ions (corresponding to quantitative and confirmative ions) were monitored for each of the 37 veterinary drugs (Table 2). Several MS parameters including ion-spray voltage, source temperature, declustering potential, entrance potential, and four gas pressures were systematically varied according to the manual of flow injection analysis, and we selected the conditions that yielded the best sensitivity, as listed in the Experimental section. In particular, we noted the curtain gas, ion source gas 1 and 2 conditions that measured macrolides with high sensitivity. Because MS scans of some of penicillins showed that the most abundant molecular ion was the deprotonated molecule [M-H], it was excluded from the analytes in this study. LC Conditions LC conditions to determine multi-class veterinary drugs in livestock and fishery products were previously reported by our laboratory (9), in which a gradient mixture of 0.1% formic acid in 10 mm ammonium acetate and acetonitrile as the mobile phase and a C18 column were used. However, the sensitivities of TCs and QLs were low under these conditions. Because the ionization mode of these drugs was positive-ion ESI, the ammonium ion which lowered the sensitivity of [M+H] + was excluded from the mobile phase. The peak shapes of FQs and thiabendazole were split. The peak shapes of TCs, sulfathiazole, sulfamerazine, clopidol, and oxolinic acid were poor. The tailing factors of these drugs were TCs and QLs which are strong metal chelating compounds interact with metal ion impurities remaining in the C18 column, which made their peaks broad. The novel organic hybrid silica base column (YMC-Triart) has been reported to reduce metal ion impurities and achieve good chromatographic retention and separation of metal chelating and

14 hydrophilic compounds. Using the column, the peak shapes of TCs and QLs were better, and the sensitivities were improved by a factor of 3. Thiabendazole, sulfathiazole, sulfamerazine, and clopidol diluted with an organic solvent were poorly retained on the column because the organic solvent may act as a part of the mobile phase. We minimized the drug injection volume to 2 µl, which resulted in the peak widths at half height ranging from 0.1 to 0.44 min and the tailing factors ranging from 0.85 to Extraction and Cleanup Procedure The extraction and cleanup procedure was developed using 11 veterinary drugs, norfloxacin, ciprofloxacin, chlortetracycline, doxycycline, 5-hydroxythiabendazole, clopidol, erythromycin A, spiramycin, lincomycin A, oxolinic acid, and sulfadimidine. Among these drugs, norfloxacin, ciprofloxacin, chlortetracycline, doxycycline, 5-hydroxythiabendazole, and clopidol did not achieve acceptable values following the guidelines of Japanese Ministry of Health, Labour, and Welfare (10, 11) by using our reported QuEChERS methods (8, 9), because these compounds were soluble in the aqueous phase. Erythromycin A and spiramycin represent the macrolides class. Lincomycin A represents the lincosamides. Oxolinic acid and sulfadimidine had higher accuracy than other drugs on LC-MS/MS. These compounds served as indicators, showing that the LC-MS/MS measurements are stable. After spiking 50 µl of a 1 µg/ml standard solution of these drugs into a minced swine muscle, the following studies were performed. At this time, the drugs were quantified by using matrix-matched calibration standard curves. Veterinary drugs were extracted from the sample using an ultrasonic machine (33 35). This procedure allowed the simultaneous handling of many samples and lowered the risk of

15 contamination. The sufficient extraction ability was confirmed using the incurred swine muscle containing chlortetracycline. As extraction solvents, we compared ACN/FA used on our modified QuEChERS method (9) and Na 2 EDTA-McIlvaine s buffer used on our antibiotic extraction (5 7). The extracted rates of 11 drugs by Na 2 EDTA-McIlvaine s buffer were calculated as follows. Eleven drugs spiked into a swine muscle were extracted with Na 2 EDTA-McIlvaine s buffer. The extraction solution was loaded onto the PLS-3 cartridge, and was eluted with ACN/FA. This eluate was analyzed by LC-MS/MS. The recovery rates (a) were calculated. Na 2 EDTA-McIlvaine s buffer spiked with 11 drugs was loaded onto the PLS-3 cartridge. The recovery rates from the PLS-3 cartridge (b) were calculated. The extraction rates by Na 2 EDTA-McIlvaine s buffer were corrected (a) using (b). The ph of the buffer was set as 7.0 because the retention of drugs was better than at ph 4. Na 2 EDTA was added to the buffer because the extraction of TCs and QLs were better with buffer containing Na 2 EDTA which had the ability to chelate divalent cations (8, 13, 16, 19, 22). As shown in Figure 2a, the extraction rate of each drug, i.e., norfloxacin, ciprofloxacin, chlortetracycline, doxycycline, and lincomycin A was better with Na 2 EDTA-McIlvaine s buffer than with ACN/FA. The second extraction using Na 2 EDTA-McIlvaine s buffer did not improve recovery rates. The first extraction using Na 2 EDTA- McIlvaine s buffer and second extraction step using ACN/FA improved recovery rates to over 70%, except for chlortetracycline and doxycycline, which were unstable in solution. Therefore, polar veterinary drugs were extracted with two different polar solvents, Na 2 EDTA-McIlvaine s buffer (ph 7.0) and ACN/FA. Subsequently, we evaluated the two conditions to elute the compounds from the SPE polymer cartridge which retained the compounds first-extracted by Na 2 EDTA-McIlvaine s buffer.

16 (1) A new ACN/FA (10 ml) was used as the elution solution. The resultant eluate and the second extracted solution were mixed and analyzed by LC-MS/MS. (2) The second extracted solution was re-used as the elution solution. The eluate was diluted by 2- fold with ACN/FA, and analyzed by LC-MS/MS. On (1) and (2) conditions, the recovery rates of 11 veterinary drugs were the same (Figure 3a). However, the matrix effects were dramatically different. The matrix effect was defined as the ratio of the slope of the matrix-matched calibration curve and the standard solution calibration curve. On the condition of (1), strong matrix enhancements were found for norfloxacin, ciprofloxacin, chlortetracycline, doxycycline, spiramycin, and lincomycin A. In contrast, the matrix enhancements were not observed under (2) conditions. Because the pork fatty acids and phospholipids were reported to be retained by the SPE polymer cartridge (36, 37), the interfering matrix was considered to be cleaned-up when the second extraction solution was passed through the SPE polymer cartridge (Figure 3b). Finally, we chose the extraction and cleanup procedure shown in Figure 1. Instrument Performance Figure 4a shows the SRM chromatograms obtained from swine muscle spiked with 10 µg/kg of 37 veterinary drugs and demeclocycline. No matrix effect was observed on peak shape in all samples. The retention time determined for the spiked samples was not significantly different from that determined for the standards. The relative ion abundance ratios of the selected product ions for each compound are shown in Table 2 together with those of the standard solutions. All of the relative ion abundance ratios of the spiked samples were within 20% of those of the standard solutions, which satisfied the permitted tolerance required in the EU guidelines (38). These results indicated that the

17 matrix did not significantly affect the fragmentation patterns of each precursor ion of 37 veterinary drugs to two product ions. Linearity of Calibration The matrix-matched calibration curves of 35 veterinary drugs, except for chlortetracycline and doxycycline, were obtained for a series of standard solutions containing each matrix at five concentrations by plotting the peak area against the concentration. Chlortetracycline and doxycycline were unstable in the resultant solution. Therefore demeclocycline was used as an IS to more accurately measure the concentrations of both chlortetracycline and doxycycline. Chlortetracycline and doxycycline calibration curves were obtained for a series of standard solutions at five concentrations by plotting the peak area against the concentration, corrected by 0.01 µg/ml demeclocycline. All of the correlation coefficient (r) values were over 0.999, and deviations in individual points from the calibration curves were lower than 20%. Accordingly, satisfactory linearity was obtained in the range examined for each compound. To evaluate the matrix effect, slopes derived from the standard solution and matrix-matched calibration curves derived from each livestock and fishery product were compared at the same range as described above. The slope ratios of the matrix-matched/standard solution calibration curves were obtained for each of the 37 veterinary drugs (Figure 5). The slope ratios ranging from 0.8 to 1.2 were considered to be tolerable, whereas the ratio higher than 1.2 or lower than 0.8 implied a strong matrix effect (39). A significant matrix effect was noticed on marbofloxacin in bovine muscle and prawn; danofloxacin in chicken muscle, prawn, flounder, and honey; orbifloxacin in prawn, difloxacin in bovine muscle; tetracycline in swine muscle and prawn; chlortetracycline in prawn;

18 demeclocycline in prawn; spiramycin in swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, flounder, and honey; tilmicosin in swine muscle, chicken muscle, bovine muscle, red sea bream, flounder, and honey; pirlimycin in honey; sulfadiazine in bovine muscle and honey; sulfathiazole in bovine muscle, prawn, flounder, and honey; sulfamerazine in honey; sulfadimidine in honey; sulfamonomethoxine in bovine muscle and honey; sulfaquinoxaline in salmon trout, sulfadimethoxine in salmon trout; 5-hydroxythiabendazole in prawn; thiabendazole in bovine muscle and prawn. Therefore, we consider that the matrix-matched standard calibration curves were adequate for the quantification of each of the 37 veterinary drugs in livestock and fishery products. Method Validation Validation was carried out following the guidelines of the Japanese Ministry of Health, Labour, and Welfare (5, 6). The developed method in this study was validated by means of recovery tests using swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, flounder, milk, egg, and honey samples which were spiked with 50 µl of working standard solutions (1 µg/ml or 10 µg/ml) in two replicates for 5 separate days. As shown in Table 3, the overall recovery of the 37 drugs ranged from 25 to 118%. The RSDr ranged from 1 to 26%. The RSD WR ranged from 2 to 36%. In this method, the numbers of analytes that satisfied the guidelines criteria were 37 out of 37 in swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, milk, and honey samples, 35 out of 37 in egg, and 34 out of 37 in flounder. Only two analytes (sulfadiazine and clopidol) in egg and three analytes (norfloxacin, danofloxacin, and sulfaquinoxaline) in flounder were not sufficiently recovered. Selectivity was confirmed by analyzing blank samples, and no

19 interfering peaks were observed at the same retention times of the target analytes. Figure 4b shows the SRM chromatograms obtained from the blank swine muscle. LODs and LOQs As shown in Table 3, LOQs for the 37 veterinary drugs ranged from 0.1 to 5 µg/kg, which was less than the 10 µg/kg default regulatory limit, set by the positive list system for agricultural chemical residues in foods in Japan. LODs ranged from 0.03 to 2 µg/kg. Survey of Livestock and Fishery Products To demonstrate the applicability of the developed method in this study for the determination of 37 veterinary drug residues, 110 samples (20 swine muscle, 15 chicken muscle, 13 bovine muscle, 10 prawn, 10 salmon trout, 7 red sea bream, 10 flounder, 5 milk, 10 egg, and 10 honey), purchased from retail outlets in Japan, were tested. When the peak was detected, the ion ratios were compared with those of the standard solutions at comparable concentrations. Because the relative ion abundance ratios were within 20% recommended by EU guidelines (38), the identity of residual drugs was accurate. No analyte was detected in bovine muscle, prawn, red sea bream, milk, or egg. In swine muscle, TCs were detected in ten samples. In four samples, more than one TC was found; oxytetracycline (2.5 µg/kg) and chlortetracycline (3.9 µg/kg), oxytetracycline (1.3 µg/kg) and doxycycline (1.4 µg/kg), tetracycline (1.7 µg/kg) and chlortetracycline (10.9 µg/kg), chlortetracycline (18.7 µg/kg) and doxycycline (4.0 µg/kg). On the other hand, oxytetracycline was found in four samples (1.3, 3.6, 4.5, and 9.5 µg/kg), tetracycline in one sample (6.4 µg/kg), and doxycycline in one sample (0.8 µg/kg). In chicken muscle, three veterinary drugs were detected in four samples. Clopidol (3.4 µg/kg), oxytetracycline (8.5 µg/kg), and enrofloxacin (20.9 µg/kg) were contained in one sample. Oxytetracycline was contained in

20 one sample (10.1 µg/kg). Enrofloxacin was found in two samples (1.0, and 1.0 µg/kg). In salmon trout, the oxytetracycline residue was detected in two samples (4.5 and 27.0 µg/kg). In flounder, the oxytetracycline residue was detected in three samples (9.1, 20.5, and 23.7 µg/kg). In honey, the norfloxacin residue was detected in two samples (1.1 and 1.5 µg/kg), mirosamycin residue in two samples (3.0 and 36.3 µg/kg), and tylosin residue in one sample (4.4 µg/kg). All values were lower than the MRLs or regulatory default limits of 10 µg/kg for livestock and fishery products. The residues of veterinary drugs were found in 24 of 110 samples (22%). Conclusions We developed a novel method to determine 37 polar veterinary drugs in 10 livestock and fishery products using LC-MS/MS. The sample preparation for 10 samples takes only 2 h and no evaporation step is needed. Polar veterinary drugs were efficiently extracted from livestock and fishery products with two different polar solvents, Na 2 EDTA-McIlvaine s buffer (ph 7.0) and ACN/FA. Among the compounds examined, highly polar veterinary drugs were initially extracted from samples with Na 2 EDTA-McIlvaine s buffer, and then medium polar veterinary drugs were extracted by a second extraction step with ACN/FA. We re-used the second extracted solution as an elution solution from the SPE polymer cartridge which retained the compounds first-extracted by Na 2 EDTA-McIlvaine s buffer. The matrix extracted with Na 2 EDTA-McIlvaine s buffer was reduced by being passed through the SPE polymer cartridge. The matrix extracted by the second extraction step was retained on the same SPE polymer cartridge. Strong matrix effects were reduced by this cleanup procedure.

21 TCs and QLs were measured with good sensitivity and excellent peak shapes using the novel hybrid column, a mobile phase consisting of a mixture of 0.05% formic acid and acetonitrile, and a minimum injection volume. By preparing the gas pressure on MS/MS parameters, macrolides were measured with high sensitivity. LOQs of the 37 veterinary drugs were lower than the MRLs. Using this method, the numbers of analytes that were validated in accordance with the Japanese Ministry of Health, Labour, and Welfare guideline were 37 analytes out of 37 in swine muscle, chicken muscle, bovine muscle, prawn, salmon trout, red sea bream, milk, and honey, 35 in egg, and 34 in flounder. FQs, TCs, and 5-hydroxythiabendazole, which could not be determined using previously reported methods, were successfully analyzed using our novel method. This method was successfully applied on 110 commercially available livestock and fishery products. Veterinary drug residues were found in 24 samples. It is necessary to continue monitoring for the residues of 37 veterinary drugs in livestock and fishery products using this method. The method developed in this study provides high-quality performance and ease of implementation for the routine monitoring of 37 polar veterinary drugs in livestock and fishery products. References (1) Turnidge, J. (2004) Antimicrob. Chemother. 53, (2) Stolker, A.A.M., Zuidema, T., & Nielen, M.W.F. (2007) Trends Anal. Chem. 26, (3) Paige, J., Tollefson, L., & Miller, M. (1997) Vet. Hum. Toxicol. 39, (4) Tillotson, G.S., Doern, G.V., & Blondeau, J.M. (2006) Expert Optional on Investigational Drugs. 15,

22 (5) Jinbo, K., Monma, C., Maruyama, T., & Matsumoto, M. (1991) J. Food Hyg. Soc. Japan 32, (6) Kusano, T., Kanda, M., Kamata, K., & Miyazaki, Y. (2004) J. Food Hyg. Soc. Japan 45, (7) Kanda, M., Kusano, T., Kanai, S., Hayashi, H., Matsushima, Y., Nakajima, T., Takeba, K., Sasamoto, T., & Nagayama, T. (2010) J. AOAC Int. 93, (8) Nakajima, T., Sasamoto, T., Hayashi, H., Kanda, M., Takeba, K., Kanai, S., Kusano, T., Matsushima, Y., & Takano, I. (2012) J. Food Hyg. Soc. Japan, 53, (9) Nakajima, T., Nagano, C., Sasamoto, T., Hayashi, H., Kanda, M., Kanai, S., Takeba, K., Matsushima, Y., & Takano, I. (2012) J. Food Hyg. Soc. Japan. 53, (10) Direction Notification Syoku-An No (Nov. 11, 2007) Ministry of Health, Labour, and Welfare, Tokyo, Japan (11) Direction Notification Syoku-An 1224 No.1 (Dec. 24, 2010) Ministry of Health, Labour, and Welfare, Tokyo, Japan (12) Mol, H.G.J., Plaza-Bolaňos, P., Zomer, P., De Rijk, T.C., Stolker, A.A.M., & Mulder, P.P.J. (2008) Anal. Chem. 80, (13) Aguilera-Luiz, M.M., Vidal, J.L.M., Romero-González, R., & Frenich, A.G. (2008) J. Chromatogr. A 1205, (14) Stubbings, G., & Bigwood, T. (2009) Anal. Chim. Acta 637,

23 (15) Frenich, A.G., Aguilera-Luiz, M. del M., Vidal, J.L.M., & Romero-González, R. (2010) Anal. Chim. Acta 661, (16) Vidal, J.L.M., Frenich, A.G., Aguilera-Luiz, M.M., Romero-González, R. (2010) Anal. Bioanal. Chem. 397, (17) Lopes, R.P., Reyes, R.C., Romero-González, R., Vidal, J.L.M., & Frenich A.G. (2012) J. Chromatogr. B , (18) Capriotti, A.L., Cavaliere, C., Piovesana, S., Samperi, R., & Laganà, A. (2012) J. Chromatogr. A 1268, (19) Granelli, K., Elgerud, C., Lundstrom, A., Ohlsson, A., & Sjoberg, P. (2009) Anal. Chim. Acta 637, (20) Martos, P.A., Jayasundara, F., Dolbeer, J., Jin, W., Spilsbury, L., Mitchell, M., Varilla, C., & Shurmer, B. (2010) J. Agric. Food Chem. 58, (21) Chiaochan, C., Koesukwiwat, U., Yudthavorasit, S., & Leepipatpiboon, N. (2010) Anal. Chim. Acta 682, (22) Bittencourt, M.S., Martins, M.T., De Albuquerque, F.G.S., Barreto, F., & Hoff, R. (2012) Food Addit. Contam. 29, (23) Rezenda, C.P., Almeida, M.P., Brito, R.B., Nonaka, C.K., & Leite, M.O. (2012) Food Addit. Contam. 29, (24) Robert, C., Gillard, N., Brasseur, P.Y., Pierret, G., Ralet, N., Dubois, M., & Delahaut, Ph. (2013) Food Addit. Contam. 30, (25) Jiménez, V., Rubies, A., Centrich, F., Companyó, R., & Guiteras, J. (2011) J. Chromatogr. A 1218,

24 (26) Wang, L., Yang, H., Zhang, C., Mo, Y., & Lu, X. (2008) Anal. Chim. Acta 619, (27) Blasco, C., Masia, A., Morillas, F.G., & Picó, Y. (2011) J. AOAC Int. 94, (28) Kantiani, L., Farre, M., Freixiedas, J.M.G.I., & Barcelo, D. (2010) J. Chromatogr. A 1217, (29) Bogialli, S., Curini, R., Di Corcia, A., Lagana, A., & Rizzuti, G. (2006) J. Agric. Food Chem. 54, (30) Imported Foods Inspection Services ( ) Department of Food Safety, Ministry of Health, Labour, and Welfare, Tokyo, Japan. foods/index.html (31) Rapid Alert System for Foods and Feed ( ) European Commissioner for Health and Consumer, EU Search=1 (32) Ho, Y.B., Zakaria, M.P., Latif, P.A., & Sarri, N. (2012) J. Chromatogr. A 1262, (33) Marazuela, M.D., & Bigualli, S. (2009) Anal. Chim. Acta 645, (34) Hermo, M.P., Barron, D., & Barbosa, J. (2005) Anal. Chim. Acta 539, (35) Xu, H., Chen, L., Sun, L., Sun, X., Du, X., Wang, J., Wang, T., Zeng, Q., Wang, H., Xu, Y., Zhang, X., & Ding, L. (2011) J. Sep Sci. 34, (36) Smith, R.D. (2006) Biotechniques 41,

25 (37) Lehotay, S.J., & Maštovská, K. (2005) J. AOAC Int. 88, (38) Method Validation and Quality Control Procedures for Pesticide Residues Analysis in Food and Feed. (2011) European Union, Brussels, Belgium, Document No. SANCO/2011/12495 (39) Kollipara, S., Bende, G., Agarwal, N., Varshney, B., & Paliwal, J. (2011) Chromatographia 73, Figure Captions Figure 1. Schematic representation of the sample preparation procedure for the analysis of 37 veterinary drugs in livestock and fishery products. Figure 2. Effect on the extracted ratios of 11 veterinary drugs from swine muscle, twice extraction by the same solvent (a), by the different solvents (b). Mean of 5 replications. Figure 3. Effect of the two conditions eluting from the SPE polymer cartridge on the recovery rate of 11 veterinary drugs (a), the rate of matrix effect (b). Mean of 5 replications. Note: Not all Figures and Tables have been inserted in the Sample Document.

26 Table 1. log P values of veterinary drugs Analytes logp Quinolones Marbofloxacin 0.5 Norfloxacin 1.0 Ofloxacin 0.4 Enrofloxacin 0.2 Ciprofloxacin 1.1 Danofloxacin 0.3 Orbifloxacin 0.9 Sarafloxacin 0.3 Difloxacin 1.6 Oxolinic acid 1.7 Nalidixic acid 1.4 Flumequine 2.9 Tetracyclines Oxytetracycline 1.6 Tetracycline 2.0 Chlortetracycline 1.3 Doxycycline 0.7 Demeclocycline a 0.7 Macrolides Spiramycin 2.1 Tilmicosin 3.6 Mirosamicin 2.0 Oleandomycin 2.6 Erythromycin A 2.7 Tylosin 1.0 Josamycin 2.9 Lincosamides Lincomycin A 0.2 Pirlimycin 1.7 Sulfonamides Sulfadiazine 0.1 Sulfathiazole 0.1 Sulfamerazine 0.1 Sulfadimidine 0.3 Sulfamonomethoxine 0.8 Sulfamethoxazole 0.9 Sulfaquinoxaline 1.7 Sulfadimethoxine 1.6 Others Thiabendazole hydroxythiabendazole 2.1 Clopidol 2.6 Tiamulin 5.6 a The internal standard material for the quantification of chlortetracycline and doxycycline.

27 Table 2. SRM parameters Analytes Transition, m/z Retention time, min Declustering potential, V Collision energy, ev Collision cell exit potential, V Ion ratio, % c Quinolones Marbofloxacin a Norfloxacin a Ofloxacin a Ciprofloxacin a Danofloxacin a Enrofloxacin a Orbifloxacin a Sarafloxacin a Difloxacin a Oxolinic acid a Nalidixic acid a Flumequine a Tetracyclines Oxytetracycline a Tetracycline a Chlortetracycline a Doxycycline a Demeclocycline b a Macrolides Spiramycin a Tilmicosin a Mirosamicin a

28 Oleandomycin a Erythromycin A a Tylosin a Josamycin a Lincomycins Lincomycin A a Pirlimycin a Sulfonamides Sulfadiazine a Sulfathiazole a Sulfamerazine a Sulfadimidine a Sulfamonomethoxine a Sulfamethoxazole a Sulfaquinoxaline a Sulfadimethoxine a Others 5- hydroxythiabendazole a Clopidol a Thiabendazole a Tiamulin a a Ion used for quantification. b The internal standard material for the quantification of chlortetracycline and doxycycline. c The relative ion abundance ratio of the selected product ions for the standard solution, 10 ng/ml of each compound.

29 Figure 1

30 Figure 2

31 Figure 3

32