Identification of a Photooxygenation Product of Chlortetracycline in

Publication Date (Web): September 9, 2004. Copyright © 2004 ... Sandra Pérez, Peter Eichhorn, Mary Dawn Celiz, and Diana S. Aga. Analytical Chemistr...
0 downloads 0 Views 175KB Size
Download PDF
Anal. Chem. 2004, 76, 6002-6011

Identification of a Photooxygenation Product of Chlortetracycline in Hog Lagoons Using LC/ESI-Ion Trap-MS and LC/ESI-Time-of-Flight-MS Peter Eichhorn and Diana S. Aga*

Department of Chemistry, The State University of New York at Buffalo, 611 Natural Sciences Complex, Buffalo, New York 14260

In the present work, a photoproduct of chlortetracycline (CTC) was identified for the first time in hog lagoon samples from confined animal feeding operations. Screening of several samples by LC/ESI-MS indicated the presence of a potential photooxygenation product of CTC with a nominal mass that was 32 Da higher than the parent drug. Generation of this assumed photoproduct (designated M510) was achieved in a 24-h irradiation experiment of an oxygenated alkaline medium containing 50 mg L-1 CTC. Accurate mass measurements with an ESI-TOFMS of the protonated isomerization product of CTC (iCTC) and the postulated photooxygenation product, bearing two oxygen atoms more than iCTC, were m/z 479.1229 and 511.1109, respectively. These corresponded to errors of -2.8 ppm for iCTC and +1.0 ppm for M510 relative to the theoretical masses. The generation of second- and third-stage mass spectra in an ESI-ion trap-MS showed similar characteristic fragmentation patterns for iCTC and the photoproduct M510, leading to the conclusion that the M510 structure consisted of an iCTC-like skeleton bearing two additional hydroxy groups. The site and configuration of one hydroxylation was confidently assigned, while the position of the other hydroxy group was tentatively assigned. Comparison of the (+)-ESI-MS3 spectrum and the retention time of M510 in the sample from the irradiation experiment with those from the hog lagoon samples yielded excellent agreement, suggesting that the compounds were identical. Quantitative analysis of seven hog lagoon samples showed iCTC concentrations of 1910-15 800 µg L-1, while the levels of M510 were found to be between 46 and 303 µg L-1. Modern animal husbandry practices encountered in large-scale confined animal feeding operations (CAFOs) rely on the extensive use of antibiotics for prevention and treatment of diseases as well as for growth-promoting purposes and to increase feed efficiency. Of the total amount of 9900 metric tons of antibiotic compounds sold to U.S. farmers in the year 2001, about one-third belonged to the class of tetracycline antibiotics.1 The tetracyclines are broadspectrum antibiotics approved for a wide variety of farm animals. * Correponding author. E-mail: [email protected]. Phone: 1-716-645-6800 X2226. Fax: 1-716-645-6963. (1) Beef Mag. 2002, (August); http://beef-mag.com.

6002 Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

These compounds are only partially metabolized and are mainly excreted via urine and feces in its original form. Therefore, the manure of treated animals contains appreciable levels of tetracyclines. Analytical measurements of tetracyclines in hog lagoon samples, employing liquid chromatography/electrospray ionization-mass spectrometry (LC/ESI-MS), showed residues typically in microgram per liter levels with some maximum concentrations in the low milligram per liter range.2-5 Introduction of these nutrient-rich animal wastes into the terrestrial environment as organic fertilizers for agricultural crops raises the question on the whereabouts of the antibiotics.6,7 Due to the antibacterial activity of tetracyclines, biodegradation brought about by bacteria can be anticipated to play a negligible role in the elimination of the antibiotic. To date, no proof for the microbial degradability of tetracyclines in the environment has been reported. Chemical transformation processes, in turn, such as isomerization and epimerization have been reported, giving rise to structurally related compounds likewise exhibiting resistance to breakdown. For instance, chlortetracycline (CTC) is converted to isochlortetracycline (iCTC) under alkaline conditions (see Figure 1), while the epimerization has been found to be catalyzed in acidic solutions in a pH range from 2 to 6.8 The photochemical degradation of tetracyclines could be the major elimination pathway in the environment because tetracyclines absorb light in the UV-A range and are known to be susceptible to photolysis. A series of laboratory-scale studies have been conducted to identify photoproducts generated during light exposure of tetracycline solutions under various conditions.9-12 Structural elucidation of (2) Aga, D. S.; Goldfish, R.; Kulshrestha, P. Analyst 2003, 128, 658-662. (3) Campagnolo, E. R.; Johnson, K. R.; Karpati, A.; Rubin, C. S.; Kolpin, D. W.; Meyer, M. T.; Esteban, J. E.; Currier, R. W.; Smith, K.; Thu, K. M.; McGeehin, M. Sci. Total Environ. 2002, 299, 89-95. (4) Haller, M. Y.; Muller, S. R.; McArdell, C. S.; Alder, A. C.; Suter, M. J. F. J. Chromatogr., A 2002, 952, 111-120. (5) Meyer, M. T.; Bumgarner, J. E.; Varns, J. L.; Daughtridge, J. V.; Thurman, E. M.; Hostetler, K. A. Sci. Total Environ. 2000, 248, 181-187. (6) Hamscher, G.; Sczesny, S.; Hoeper, H.; Nau, H. Anal. Chem. 2002, 74, 1509-1518. (7) De Liguoro, M.; Cibin, V.; Capolongo, F.; Halling-Sørensen, B.; Montesissa, C. Chemosphere 2003, 52, 203-212. (8) Mitscher, L. A. The chemistry of tetracycline antibiotics; Medicinal Research Series 9; Marcel Dekker: New York, 1978. (9) Oka, H.; Ikai, Y.; Kawamura, N.; Yamada, M.; Harada, K.; Ito, S.; Suzuki, M. J. Agric. Food Chem. 1989, 37, 226-231. (10) Drexel, R. E.; Olack, G.; Jones, C.; Chmurny, G. N.; Santini, R.; Morrison, H. J. Org. Chem. 1990, 55, 2471-2478. (11) Morrison, H.; Olack, G.; Xiao, C. J. Am. Chem. Soc. 1991, 113, 8110-8118. 10.1021/ac0494127 CCC: $27.50

© 2004 American Chemical Society Published on Web 09/09/2004

Figure 1. Chemical structure of chlortetracycline and its pHdependent epimerization and isomerization reactions.

the products formed was primarily achieved by employing proton and carbon nuclear magnetic resonance (NMR) with some additional information gathered from UV-visible spectroscopy. Major reported photoproducts of tetracycline were lumitetracycline, formed with long-wavelength UV light in oxygen-free organic or aqueous media,10 anhydrotetracycline generated in the presence of a hydrogen donor likewise using long-wavelength UV light,13 de(dimethylamino)tetracycline formed in deoxygenated organic media,14 and a quinoid product formed in basic aqueous solutions in the presence of oxygen.15 The combination of NMR, infrared spectroscopy, mass spectrometry, X-ray, and elemental analysis after a complex fractionation process allowed Oka et al.9 to identify several photodecomposition products formed upon irradiation of aerated tetracycline solutions. In this study, hydroxy carboxylic acids and lactones with an intact C- and D-ring (see Figure 1) and molecular weights between 232 and 332 were identified. In contrast to the many laboratory simulations, only one work performed under field conditions has been referred in a recent review on the photodegradation of pharmaceuticals in the environment.16 In that study, the photodecomposition of oxytetracycline, one of the most frequently used antibiotics for treatment of fish raised in aquaculture, was examined in marine water.10 It was demonstrated that oxytetracycline degraded completely at both sea surface and submerged to a depth of 1 m within 21 days. However, no attempts were made to identify photoproducts. The light-induced breakdown of tetracyclines in soil under field conditions is thought to play a negligible role in their dissipation (12) Lunestad, B. T.; Samuelsen, O. B.; Fjelde, S.; Ervik, A. Aquaculture 1995, 134, 217-225. (13) Hasan, T.; Allen, M.; Cooperman, B. S. J. Org. Chem. 1985, 50, 17551757. (14) Hlavka, J. J.; Bitha, P. Tetrahedron Lett. 1966, 3843-3846. (15) Davies, A. K.; McKellar, J. F.; Phillips, G. O.; Reid, A. G. J. Chem. Soc., Perkin Trans. 2: Phys. Org. Chem. (1972-1999) 1979, 369-375. (16) Boreen, A. L.; Arnold, W. A.; McNeill, K. Aquat. Sci. 2003, 65, 320-341.

as a consequence of strong absorption into interlayers of clay minerals. Plowing of manure-amended soils further adds to the macroscopic incorporation into the soil matrix. Both effects are believed to severely attenuate the light exposure of the antibiotics. A far greater potential for photochemical reactions, however, can be anticipated to occur in hog lagoons as these are uncovered basins with large air-liquid interfaces. Hog lagoons are used for temporary storage of animal wastes from CAFOs until they are used for soil amendment. The present study aimed to detect and identify the structure of chlortetracycline photoproducts in hog lagoon samples. The findings obtained should provide evidence of the relevance of photochemical reactions as mechanisms of chlortetracycline transformations in the environment. This should substantially contribute to the understanding of the fate and behavior of this important veterinary medicine and allow extension of the database required for a comprehensive environmental risk assessment. A tiered approach was applied encompassing (a) screening of hog lagoon samples for chlortetracycline residues and possible photodegradates employing liquid chromatography/electrospray ionization-mass spectrometry (LC/ESI-MS), (b) generation of tentatively identified manure-borne photoproducts in well-defined irradiation experiments under laboratory conditions, (c) accurate mass measurements of photoproducts by liquid chromatography/ electrospray ionization-time-of-flight-mass spectrometry (LC/ESITOF-MS) for confirmation of proposed chemical composition, and (d) comprehensive structural elucidation of an environmentally relevant photoproduct by liquid chromatography/electrospray ionization-ion trap-mass spectrometry (LC/ESI-IT-MS). EXPERIMENTAL SECTION Chemical Standards. Chlortetracycline hydrochloride (CAS No. 64-72-2) was purchased from Riedel de Haen (Seelze, Germany); isochlortetracycline hydrochloride (514-53-4) was obtained from Janssen (Geel, Belgium). The organic solvents acetonitrile and methanol were ACS grade (Burdick & Jackson, Muskegon, MI). Water was prepared with a Nanopure Diamond water purifier (Barnstead, Dubuque, IA). All chemicals used for preparation of buffers were of reagent grade. Irradiation Experiments. Air-saturated or nitrogen-purged solutions of CTC at a concentration of 50 mg L-1 (McIlvaine buffer, pH 8.0; 194.5 mM Na2HPO4/2.75 mM citric acid) were irradiated in a quartz tube (d ) 1.5 cm, L ) 25 cm, V ) 100 mL) using a 30-W fluorescent bulb with emission wavelengths from approximately 350 to 750 nm (see Figure S-1 for wavelength distribution). The 400-µL aliquots were withdrawn at 1-3-h intervals during the irradiation and analyzed without delay by LC-diode array detection (DAD)-ESI-MS to monitor the disappearance of the parent compound and the formation of (photo)degradates. Dark control samples, prepared at the same time as the solutions used for irradiation, were submitted to analysis between the chromatographic runs of the samples from the irradiated solution. The control samples served to identify and follow up chemical conversion processes such as isomerization, epimerization, or both that occur instantly after preparation of the CTC solution in the absence of light.8 All experiments were carried out at 19 °C. LC/DAD-ESI-MS Analysis. The liquid chromatograph used in this study was an Agilent Series 1100 comprising the modular components: quaternary pump, a vacuum solvent microdegasser, Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

6003

an autosampler with 100-well tray, and an on-line DAD. Separations were achieved on a Thermo Hypersil-Keystone BetaBasic-18 100 × 2.1 mm (3 µm) column equipped with a 10 × 2.1 mm guard column of the same packing material. The mobile phases were (A) water acidified with 0.3% formic acid and (B) acetonitrile. The gradient program started from 90% A-10% B. After 1 min, the portion of A was linearly decreased to 45% within 11.6 min and further to 5% within 0.1 min. These conditions were held for 3.5 min. The initial mobile-phase composition was restored within 0.1 min and maintained for column regeneration for another 6.7 min, resulting in a total run time of 23 min. The flow rate was 250 µL min-1, and the injection volume was 20 µL. During the first 2 min and the last 6.7 min of each chromatographic run, the LC stream exiting the DAD was directed to the waste via a programmable switching valve integrated in the mass spectrometer. The DAD was programmed to acquire the UV spectra from 220 to 500 nm (step size 4 nm). In addition, the UV traces at 270 and 355 nm, the two adsorption maximums of CTC, were recorded. The mass spectrometric analysis was performed on an Agilent Series 1100 SL single-quadrupole instrument equipped with an ESI source. All acquisitions were performed under positive ionization mode with a capillary voltage of +4000 V. Nitrogen was used as nebulizer gas (35 psi) as well as as drying gas at 350 °C (10 L min-1). For the identification of (photo)degradation products, the mass spectrometer was operated in full-scan mode over a mass range of m/z 250-550. The value of the fragmentor, which determined the extent of in-source collision-induced dissociation and thus the quality of the mass spectra, was set to 200, enabling the production of intense fragment ions, while still providing an abundant signal of the molecular ions. For quantitative analysis of iCTC and the photolysis product in the manure samples, the data were acquired in selected ion monitoring (SIM) mode detecting the protonated molecular ions [M + H]+ with m/z 479 and 511, respectively, using fragmentor values of 140. The fragment ions with m/z 462 and 494 were used as qualifier ions (fragmentor, 230) for iCTC and M510, respectively. Data acquisition and processing was done with the software Chemstation Rev. A.09.03. LC/ESI-IT-MS Analysis. The ThermoFinnigan system used for multiple-stage MS experiments comprised the following components: Surveyor pump, Surveyor autosampler, and LCQ Advantage ion trap-mass spectrometer, equipped with an ESI interface. The analytical column, mobile-phase composition, and gradient elution were the same as described above for the singlequadrupole MS. The flow rate was 200 µL min-1, and the injection volume was 10 µL (no waste injection). The needle voltage of the ESI interface was set to +5000 V (-5000 V for (-)-ESI), the sheath gas flow was 30, the temperature of the ion transfer tube was 300 °C, and the capillary voltage was 10 V (-10 V for (-)-ESI). For MS2 and MS3 experiments, precursor ions were selected with an isolation width of 1.5 Da. LC/ESI-TOF-MS Analysis. Accurate mass measurements were performed on an Agilent TOF-MS instrument configured with an ESI source. Liquid chromatographic separations were achieved using a quaternary solvent delivery system and an autosampler (Series 1100, Hewlett-Packard, Palo Alto, CA). The separation column was packed with C18 material (Zorbax Eclipse XDB-C18 column, 150 × 2.1 mm i.d., 5 µm; Agilent, Wilmington, DE). Mobile phases were (A) 0.1% formic acid in water and (B) 6004

Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

0.1% formic acid in acetonitrile. A solvent composition of 95% A-5% B was held for 1 min and then linearly programmed to 5% A within 14 min. After a 5-min isocratic run, the starting conditions were reestablished and the column was equilibrated for 5 min. The flow rate was 250 µL min-1, and the injection volume was 2 µL. Source conditions were as follows: fragmentor 160 for [M + H]+ and 240 for the fragment ion [M + H - NH3]+, nebulizer pressure 25 psi, drying gas flow 10 L min-1, and drying gas temperature 350 °C. Mass spectra were acquired in full-scan mode in a range from m/z 100 to 1000 with a cycle time of 1 s. An external lock mass calibrant with the reference masses m/z 118.086 25, 121.050 87, and 922.009 79 was used. Sampling and Sample Preparation. Grab samples of swine manure were collected between June and November 2003 from various hog lagoons in CAFOs located in Iowa. Samples were stored in amber bottles at -40 °C until analysis. For the screening of potential photochemical transformation products using the single-quadrupole MS, the samples were preconcentrated by solidphase extraction (SPE). Manure slurry was centrifuged for 4 min at 12000g. From the supernatant, 15-mL aliquots were loaded onto 6 mL/500 mg Oasis hydrophilic-lipophilic (HLB) SPE cartridges (Waters, Milford, MA) preconditioned with 3 × 2 mL of methanol, followed by 3 × 2 mL of Nanopure water. Although acidification of aqueous samples to pH values of 2.5-4.7 has been frequently used to enhance extraction recoveries for tetracycline antibiotics, no pH adjustment of the slightly alkaline manure samples was performed in order to prevent pH-dependent chemical reactions, such as epimerization or dehydration, which are known to be favored under acidic conditions. The dried cartridges were eluted with 3 × 2 mL of methanol; the eluates were evaporated to dryness under a gentle stream of nitrogen. The dried residues were reconstituted in 1.5 mL of 10% acetonitrile-90% McIlvaine buffer (pH 8.0). Precautions were taken during sample preparation to minimize sample exposure to light. Quantification. Residual concentrations of iCTC and its epimer (e-iCTC) in the supernatant of the centrifuged manure samples were determined using the standard addition method, which allowed correcting for ionization suppression or enhancement effects caused by matrix components. To this end, the concentrations in the undiluted samples were first estimated using external calibration. Next, calibration curves were constructed for each sample by adding five different volumes of an aqueous iCTC solution to 100-µL aliquots of sample, followed by addition of McIlvaine buffer (pH 8.0) to adjust the final volume to 1000 µL. The volumes added were such that with each step the iCTC concentration increased by ∼30%. The concentration of the novel photolysis product M510 in the irradiated solution was estimated using the absorption intensity at 270 nm with iCTC used as the reference compound. This quantification approach was based on the assumption that both compounds have comparable molar extinction coefficients at 270 nm because they share the same chromophoric group. For the quantitative analysis of the identified photoproduct M510 in the manure samples, an irradiated CTC sample containing a UV-estimated amount of M510 was used to prepare a calibration series. RESULTS AND DISCUSSION Screening of Hog Lagoon Samples for Photoproducts and Generation of Identified Product under Defined Laboratory

Table 1. Concentrations (µg L-1) of Isochlortetracycline (iCTC) and Photooxygenation Product (epi-M510) in Hog Lagoons sample 1 2 3 4 5 6 7 a

pH 8.3 8.2 8.0 7.5 7.7 7.1 7.3

iCTCa (30b)

9810 10700 (35) 4180 (54) 2950 (49) 15800 (50) 3110 (54) 1910 (54)

epi-M510 138 163 71 61 303 47 46

Sum of iCTC and e-iCTC. b Percentage of e-iCTC.

Conditions. In a first step, SPE-concentrated hog lagoon samples were monitored by LC/ESI-DAD-MS in (+)-full-scan mode from m/z 250 to 550 for the presence of CTC, which was the tetracycline antibiotic that had been administered to the livestock along with the feed. In all seven samples examined, very strong signal intensities were obtained for both epimers of iCTC with [M + H]+ of m/z 479 and the diagnostic fragment ion of m/z 462. Under the applied chromatographic conditions, the regular form (iCTC) eluted at 9.4 min, while the epi form (e-iCTC) had a retention time of 8.3 min. No traces of the isobaric CTC were detected in any of the samples because of the alkaline character of the manure, containing high ammonia levels. The pH of the manure, ranging from 7.1 to 8.3 (Table 1), had brought about the complete isomerization of CTC (see also next section). The (+)-ion mode full-scan chromatograms were examined for compounds exhibiting mass spectra characterized by (a) a molecular ion cluster with an isotope pattern corresponding to one chlorine atom and (b) a fragment ion with a m/z of 17 Da less than the protonated molecular ion corresponding to the neutral loss of NH3, which is a fragmentation pathway common to all tetracycline derivatives.17,18 At this point, the assumption was made that the carboxy amide group attached to C-2 of the A-ring was not affected by photoreactions. This assumption was based on previously identified photodegradates of tetracyclines, all of which contained this group.10,13-15 Furthermore, the low-intensity solar radiation in the UV-B range that only slightly overlaps with the absorption band of the tricarbonyl chromophore of the A-ring (λmax ) 270 nm) makes this moiety less susceptible to light-induced reactions. Several candidates meeting the two requirements defined above were observed in the hog lagoon extracts. One monochlorinated compound with a protonated molecular ion of m/z 511 and a fragment ion at m/z 494 was particularly interesting because it occurred in all of the samples at significant amounts (for quantification, see below). Increasing the fragmentor voltage to 300 did not lead to the generation of any other fragment ion for this compound. The putative CTC-derived compound with a nominal mass of 510 (denoted as M510) provided two clearly distinct chromatographic peaks with identical mass spectral pattern: a major signal with a retention time of 7.8 min and a peak of markedly lower abundance at 9.0 min. This information suggested the presence of two isomers of tetracycline derivatives (17) Kamel, A. M.; Brown, P. R.; Munson, B. Anal. Chem. 1999, 71, 968-977. (18) Oka, H.; Ikai, Y.; Ito, Y.; Hayakawa, J.; Harada, K.-I.; Suzuki, M.; Odani, H.; Maeda, K. J. Chromatogr., B. 1997, 693, 337-344.

differing solely in the configuration of the dimethylamino group on the C-4 atom, characteristic of a set of the epi and the regular form (refer to Figure 1 for CTC and iCTC). Under the chromatographic conditions applied, the difference in retention times between the two signals is comparable to the difference observed for the configurational isomers of other tetracycline derivatives, such as tetracycline, demeclocycline, or anhydrochlortetracycline (the gradient profile employed had been developed for the separation of various tetracyclines and their configurational isomers). At this point of investigation we could only speculate on the identity of the chlortetracycline derivative, which was assumed to be a photoproduct. The increase in molecular mass by 32 Da relative to iCTC might have been due to two additional oxygen atoms, which rendered the molecule more polar (for assignment of elemental composition using TOF-MS, see next section); this is consistent with weaker retention of M510 observed on the reversed-phase column as compared to iCTC (the retention times of iCTC and CTC using the gradient elution were 9.4 and 10.7 min, respectively). Taking this hypothesis into account, irradiation experiments with CTC were set up to generate the assumed photoproduct M510. To this end, irradiation of buffered aqueous solutions (pH 8.0) of the antibiotic was carried out under both oxygen-saturated and oxygen-free conditions. Since CTC is known to undergo rapid isomerization to iCTC at elevated pH, samples withdrawn from the irradiated solutions were instantly analyzed by (+)-LC/ESIDAD-MS in order to enable tracking of both chemical and photochemical reactions. As a key outcome, it could be demonstrated that M510 was formed upon irradiation exclusively in the oxygen-containing solution, strongly suggesting that oxygen was involved in the photoreaction. The LC/ESI-DAD-MS analysis of an irradiated sample from the oxygenated CTC solution yielded the chromatogram shown in Figure 2. As can be clearly seen in both the UV and the MS trace, M510 was one of the major photoproducts formed. The corresponding concentration profile, as depicted in Figure 3, indicated that this compound was generated out of CTC but not out of its chemical transformation product iCTC. This conclusion was based on the observation that depletion of CTC in solution resulted in a steady-state concentration of M510 and iCTC. This was further corroborated by the fact that no M510 was built up upon irradiation of an oxygen-saturated iCTC solution in a separate experiment. In addition to the information on the circumstances of its formation, further useful structural data on M510 were gathered from its UV-visible absorption spectrum acquired at 220-500nm range. Whereas the spectrum of CTC exhibited two relative absorption maximums at 270 nm (tricarbonyl system in A-ring) and at 355 nm (aromatic/conjugated system in BCD-ring), respectively, the UV spectra of both iCTC and M510 were similar in that they lacked the absorption band centered around 355 nm (inset in Figure 2). Apparently, M510 had an iso-like structure where the bond between C-11 and C-11a atom of the C-ring had been disrupted analogous to iCTC as a consequence of a hydroxycatalyzed intramolecular cyclization that yielded an ester. Time-of-Flight-Mass Spectrometry. As demonstrated in the irradiation experiments of CTC, the generation of the photoproduct M510 required the presence of dissolved oxygen in the buffered solution, indicating the involvement of oxygen in the Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

6005

Figure 2. (+)-LC/ESI-DAD-MS chromatogram of a sample from the irradiation experiment of CTC in the presence of oxygen. Inset shows normalized UV-visible spectra of CTC, iCTC, and M510.

Figure 3. Concentration profile of chlortetracycline, isochlortetracycline, and photoproduct upon irradiation of oxygen-saturated CTC solution (initial concentration, 50 mg L-1). UV trace at 270 nm was used for quantification.

photoreaction. Assuming the simplest case that the increase in molecular mass by 32 Da was due to the addition of two oxygen atoms (CTC, C22H23N2O8Cl), the molecular formula of the photooxygenation product M510 was C22H23N2O10Cl. Firm evidence for this proposed formula was obtained through accurate mass measurements employing ESI-TOF-MS analysis after chromato6006

Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

graphic separation of an irradiated CTC solution. The calculated and experimental m/z values of the protonated molecular ions of iCTC and M510 are compiled in Table 2. The mass error of M510 was only +1.0 ppm, allowing us to strongly suggest the addition of two oxygen atoms. Two alternative molecular formulas of possible photooxygenation products, named A and B, with a nominal mass of 510 Da are proposed in Table 2. Both the deaminated species A (C22H21NO11Cl) and the demethylated compound B (C21H19N2O11Cl) yield far higher mass errors ruling out these options. Table 2 also lists accurate masses of iCTC and the characteristic fragment ions. Strong agreement in the accurate mass measurements of the target compounds found in hog lagoons with that of the irradiated compounds validated the identical composition of the two samples. Ion Trap-Mass Spectrometry. At this point of the structure elucidation of M510, the ESI-IT-MS played a critical role as it provided a unique means of performing multiple-stage experiments. This feature was of fundamental value for achieving the goal because the (+)-MS2 scan on the parent ion m/z 511 did not yield any fragment ion other than m/z 494 even at collision energies as high as 60%. For this reason, a photolysis sample containing M510 was analyzed in (+)-MS3 mode by isolating and subsequently fragmenting the first-generation fragment ion m/z 494. The scan of the product spectrum is displayed in Figure 4b along with the (+)-MS3 spectrum of iCTC (m/z 479 f 462) (Figure 4a). As can be seen, both compounds yielded very informative, fragment-rich mass spectra which, despite their

Table 2. Accurate Mass Measurement of Isochlortetracycline and Photooxygenation Product M510 Performed with ESI-TOF-MSa change rel to CTC

theor value (m/z)

exp value (m/z)

difference (mDa)

mass error (ppm)

DBE

479.1229 462.0965 511.1109 494.0854

-1.3 -1.5 +0.5 -0.5

-2.8 -3.2 +1.0 -1.1

11.5 12.5 11.5 12.5

compd

molecular formula

iCTC iCTC M510 M510

C22H23N2O8Cl C22H23N2O8Cl C22H23N2O10Cl C22H23N2O10Cl

+O2 +O2

Irradiated CTC Sample [M + H]+ 479.1216 [M + H - NH3]+ 462.0950 + [M + H] 511.1114 [M + H - NH3]+ 494.0849

A B

C22H21NO11Cl C21H19N2O11Cl

+O3 -NH2 +O3 -CH4

Alternative Compositions for M510 [M + H]+ 511.0881 [M + H]+ 511.0755

511.1109 511.1109

-23.8 -36.4

-46.6 -71.2

12.5 13.0

iCTC M510

C22H23N2O8Cl C22H23N2O10Cl

+O2

Hog Lagoon Sample (5) [M + H]+ 479.1216 [M + H]+ 511.1114

479.1214 511.1114

+0.2 (0.0

+0.4 (0.0

11.5 11.5

species

a The data suggest two options of elemental compositions for the photooxygenation product of CTC (designated A and B) with a nominal mass of 510 Da (DBE, double bond equivalents).

Figure 4. (a) (+)-ESI-MS3 spectrum of protonated isochlortetracycline (m/z 479 f 462 f 180-467 at CE ) 35 and 40%, respectively); (b) (+)-ESI-MS3 spectrum of protonated photolysis product (m/z 511 f 494 f 180-499 at CE ) 35 and 40%, respectively).

differences, revealed important details on their structural similarity. On one hand, both iCTC and M510 produced a series of characteristic fragment ions attributable to [M + H - H2O]+, [M + H - CO]+, and [M + H - H2O - CO]+ 19 with corresponding ion masses of m/z 444, 434, and 416 for the tetracycline and m/z 476, 466, and 448 for the photoproduct. The loss of CO originated

from the carboxy amide group on the C-2 atom.19 This substituent had already lost the NH2 group in the form of the neutral ammonia molecule during the first stage of fragmentation when isolating m/z 462 and 494 for iCTC and M510, respectively. On the other (19) Kamel, A. M.; Fouda, H. G.; Brown, P. R.; Munson, B. J. Am. Soc Mass Spectrom. 2002, 13, 543-557.

Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

6007

Figure 5. (a) (-)-ESI-MS2 spectrum of deprotonated isochlortetracycline (m/z 477 f 130-482 at CE ) 35%); (b) (-)-ESI-MS2 spectrum of deprotonated photolysis product (m/z 509 f 140-514 at CE ) 35%).

hand, the (+)-MS3 scans generated a common fragment ion with m/z 197 corresponding to the cleavage of the bond C-5a-C-6 that yielded a positively charged bicyclic structure encompassing the D-ring and the cyclic ester (see inset in Figure 4a). An additional evidence of this fragmentation for M510 was the formation of a fragment ion with m/z 298 comprising the other side of the molecule with the A- and B-rings (see inset in Figure 4b). That the fragment ion m/z 298 did not contain a chlorine atom while the ion m/z 197 did was confirmed by the (+)-MS3 scan of the 37Cl-isotope precursor ion [M + H]+, i.e., the sequence m/z 513 f 496 (in all MSn experiments, the precursor ions were isolated with a width of 1.5 Da). Whereas m/z 298 was still detectable, the fragment ion of the moiety containing the 37Cl isotope was now detected to m/z 199 (not shown). These results obtained in the (+)-ESI mode proved that M510 had an iCTC-like structure with the oxygenation sites somewhere in the A- and B-rings. The sites of the photochemical modification within this latter moiety could be further localized by performing acquisitions in the negative ionization mode. Though this mode had only been marginally applied in the study of tetracyclines owing to a lower ionization efficiency relative to positive ion mode,17 the low 6008 Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

milligram per liter concentrations of iCTC and M510 in the photolysis sample generated sufficiently intense deprotonated molecular ions for isolation and subsequent fragmentation in the ion trap. The (-)-ESI-MS2 spectra of iCTC (m/z 477) and M510 (m/z 509) shared several analogous fragmentation pathways (Figure 5): the ions detected at m/z 460, 434, and 416 for iCTC and at m/z 492, 466, and 448 for M510 were assigned as [M - H - NH3]-, [M - H - CONH]-, and [M - H - H2O - N(CH3)2]‚in accordance with the mass spectral interpretation by Kamel et al.19 who used H/D exchange for the elucidation of the fragmentation mechanisms of tetracyclines in an ESI-IT-MS. The fragmentation pattern in the spectra of iCTC and M510 was evidence that the left side of the A-ring comprising the moiety from C-1 to C-4 was still intact after irradiation. As has already been pointed out for the fragmentation under (+)-MS3 conditions where fragment ions arising from the cleavage of the C-5a-C6-bond were identified, the iso-like structure of M510 was further confirmed by the spectra generated in negative ion mode. The (-)-MS2 spectra of the deprotonated iCTC (m/z 477) yielded the fragment ion m/z 197, a possible structure of which is shown in the inset in Figure 5a (stabilization of the positive charge was accomplished through

delocalization over the aromatic ring). Although this species was not observed in the analogous spectrum of M510 (Figure 5b), its formation was achieved by performing a (-)-MS3 experiment on m/z 509 f 492. Taking into consideration the pieces of the puzzle collected and assembled so far, we concluded that M510 possessed an iCTClike structure with a chemical modification on any of the C-atoms of the B-ring. On the assumption that the proposed photooxidation of CTC resulted in the conversion of two C-H bonds into C-OH groups, possible sites of such oxidation were on the carbon atoms at the positions C-4a, C-5, C-5a, and C-11a (refer to Figure 1). The hypothesis put forward now was that one of the two hydroxy groups was located at C-5, which was derived from the fact that in the LC/ESI-IT-MS chromatogram of the photolysis sample (Figure 6a-2) two signals with a nominal mass of 510 were detected at 7.3 and 8.3 min (in the LC/ESI-DAD-MS at 7.8 and 9.0 min, respectively). This retention time pattern characteristic of the entire tetracycline group was assigned to an epi form and a regular form of M510, respectively. The outstanding prevalence of the supposed epi-M510, i.e., the isomer with the dimethylamino group above the plane, over the regular M510 could only be put down to a strong intramolecular interaction preventing the epi-M510 from equilibrating with the regular form. This in fact was achievable by hydrogen bonding between the nitrogen of the tertiary amine and a hydroxy group at C-5, which needed to stand on the same side of the plane (for hydrogen bonding see structure in Figure 6a-2). Such a stabilizing effect was reported in the literature 20 for oxytetracycline bearing a hydroxy group on the C-5 on the same side of the plane as the dimethylamino group and has also been observed in our own studies on this tetracycline derivative. The stability of epi-M510 toward epimerization was further proved by reanalyzing an irradiated CTC sample after storage over a 20-day period, which revealed that the ratio of the two epimers had not been changed to a notable extent. In contrast to this behavior, the ratio of iCTC versus e-iCTC had dropped from initially 28 to 4.3 as a result of equilibration of the two isomers. Regarding the position of the second hydroxy group on C-4a, C-5a, or C-11a, we could only speculate, but on a basis of steric accessibility of molecular oxygen during the photooxygenation, the position on C-11a might have been favored giving rise to the structure shown in the inset of Figure 6a-2, i.e., the tentatively identified structure of M510. Comparison of the (+)-MS3 scan of m/z 511 f 494 generated out of CTC during irradiation under defined laboratory conditions with that from the hog lagoon samples showed an excellent agreement in the quality of the spectra, suggesting that they were identical compounds. Additional evidence for the identity of M510 was provided by LC/ESI-TOF-MS measurements determining the accurate mass of the photooxygenation product in a hog lagoon sample (Table 2). The mass error for M510 was (0.0 ppm, while for iCTC the error was only +0.4 ppm. Though the mechanism of the photooxidation of CTC is beyond the scope of the present work, it appears reasonable to assume that a concerted reaction is involved, in which CTC is oxidized by molecular oxygen and at the same time isomerized to the iCTC-like structure. Such a pathway was strengthened by the fact that during the irradiation (20) Hussar, D. A.; Neibergall, P. J.; Sugita, E. T.; Doluisio, J. T. J. Pharm. Pharmacol. 1968, 20, 539-546.

experiment no transient species having both a nominal mass of 510 Da and a CTC-like structure with an intact four-ring skeleton were detected. Quantitative Analysis of the Photooxygenation Product in Hog Lagoons. To assess the relevance of the photooxygenation of CTC that leads to the formation of M510, residual concentrations of iCTC and its assumed dihydroxylated derivate were determined in seven real hog lagoon samples. Owing to high levels of iCTC, no sample preconcentration was necessary. The sum concentration of iCTC and e-iCTC determined by LC/ESI-(DAD)MS in SIM mode are summarized in Table 1, revealing significant contamination of the hog lagoons with tetracycline residues at levels from 1910 to 15 800 µg L-1. Previous studies conducted on animal wastes from CAFOs in various regions of the United States showed somewhat lower residue levels. Campagnolo3 reported CTC concentrations in samples from seven different swine lagoons of 70-1000 µg L-1. Similar amounts of this antibiotic, likewise determined by employing LC/ESI-MS, were detected in hog lagoon samples as part of a monitoring survey.5 The concentration range of CTC was from 5 to 870 µg L-1. Regarding the occurrence of the photooxidation product M510 in the samples examined in this study, concentrations were calculated by external calibration using a CTC-photolysis sample in which M510 was estimated based on the 270-nm UV response of iCTC. By applying this approach, the concentrations of epi-M510 were determined to be between 46 and 303 µg L-1 (Table 1). These values can be considered rather conservative in view of the fact that coeluting compounds from the centrifuged hog lagoon samples were likely to adversely affect the ionization of M510, thus reducing the ESI response. Moreover, the regular M510 (nonepimer), which slightly added to the total concentration of the photooxidation product, was not taken into account. What the less selective SIM mode on the single-quadrupole instrument did not allow us to observe in the chromatograms was that several samples contained structurally related compounds also with nominal masses of 510. As it turned out in (+)-MS3 experiments in the IT-MS generating the second-generation mass spectra of m/z 511 f 494, several closely eluting peaks could be distinguished. An example of this is shown for the hog lagoon sample 5 in Figure 6b-2 (Figure 6b-1 shows the signal corresponding to e-iCTC and iCTC). While the signals of e-M510 and of the regular M510 elute at 7.3 and 8.4 min, respectively, at least three further signals, marked with an asterisk, could be detected. Only minor differences in the fragmentation pattern relative to M510 were observed, suggesting that these species are possibly isomers of dihydroxylated photoproducts differing in the position of the hydroxy substituents. As far as the environmental whereabouts of M510 after application of the liquor from the lagoons onto agricultural croplands are concerned, four pathways, namely, sorption onto soil, further photochemical reactions, microbial degradation, and leaching into deeper soil layers, need to be addressed. Based on various studies dealing with these questions, sorption may be expected to be the chief parameter governing the fate of M510 in soils. A comparison to the behavior of the structurally similar compound iCTC would likely provide the most accurate assessment, but this species in particularsdespite the great deal of studies dedicated to the understanding of soil componenttetracycline interactionsshas not been subject to thorough Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

6009

Figure 6. (+)-LC/ESI-MS3 chromatograms of photolysis sample (a) and manure sample 5 (b) monitoring m/z 479 f 462 f 180-467 at CE ) 35 and 40%, respectively, for isochlortetracycline (1) and m/z 511 f 494 f 180-499 at CE ) 35 and 40%, respectively, for photolysis product (2). The analysis of the photolysis sample as shown in (a-1) was performed 20 days after the experiment; i.e., the ratio of e-iCTC to iCTC does not correspond to the actual value immediately after sampling (see text).

investigations. With regard to tetracycline antibiotics, the high water solubilities (230-52 000 mg L-1) and low water/1-octanol partition coefficients (log Kow, -1.3 to 0.05) 21,22 do not translate into weak sorption and, hence, high leaching potentials in soil,

because of pronounced nonhydrophobic interactions. In fact, sorption coefficients (Kd) ranged from 420 to 1620 L kg-1 for tetracycline and oxytetracycline in various soil types differing in pH, organic matter content, and clay content.23,24 Three major

(21) Thiele-Bruhn, S. J. Plant Nutr. Soil Sci. 2003, 166, 546. (22) Tolls, J. Environ. Sci. Technol. 2001, 35, 3397-3406.

(23) Rabolle, M.; Spliid, N. H. Chemosphere 2000, 40, 715-722. (24) Sithole, B. B.; Guy, R. D. Water, Air, Soil Pollut. 1987, 32, 315-321.

6010 Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

mechanisms had been proposed for the sorption of tetracycline including ion exchange, hydrogen bonding from acidic groups of humic acids to polar groups, and complexation by divalent cations.24,25 While the first two mechanisms are likely to play a crucial role in the sorption of M510, the extent of the latter is anticipated to be impaired as the chelation of metal cations is believed to involve preferably the tautomeric C-11-C-12 β-diketone system (refer to CTC structure in Figure 1), which is disrupted in M510. However, the two additional hydroxy groups in the photooxygenation product might give rise to new complexation sites within the molecule. Regarding further photochemical reactions of M510 in the soil, these can be assumed to have no significant effect once the manure has been spread onto the field and integrated into the soil matrix by subsequent ploughing. Also, the concentration of M510 did not decline in irradiation solutions, suggesting photostability. Microbial degradation of M510, in turn, could be an important elimination pathway because the antibacterial activity of tetracycline derivates is reported to be lost upon breaking of one of the rings of the perhydronaphthacene skeleton,26,27 thus rendering the molecule more amenable to bacterial attack. As for the transport of M510 into deeper soil horizons and ultimately into groundwater, the increase in polarity due to the (25) Sithole, B. B.; Guy, R. D. Water, Air, Soil Pollut. 1987, 32, 303-314. (26) Boothe, J. H.; Morton, J., II.; Petisi, J. P.; Wilkinson, R. G.; Williams, J. H. Antibiotics Ann. 1953-54, Proc. Symp. Antibiotics (Washington, D. C.) 1953, 46-48. (27) Blackwood, R. K.; English, A. R. Structure-activity relationship in the tetracycline series. In Structure-activity relationships among the semisynthetic antibiotics; Perlman, D., Ed.; Academic Press: New York, 1977.

introduction of two hydroxy groups is evident in the decrease in retention time in reversed-phase LC and this should make the molecule more vulnerable to leaching. According to the outcomes of soil core analyses of field samples fertilized with liquid CTCcontaining manure, antibiotic residues were found mainly in the upper 30 cm of soil. In addition, no CTC residues were detected in groundwater samples collected 2 m beneath the surface.6 Irrespective of these considerations, it remains an important task in future studies to investigate the fate of M510 in manureimpacted soil environments. ACKNOWLEDGMENT This study was supported by grants from the Interdisciplinary Research and Creative Activities Fund of the University at Buffalo and the National Science Foundation (CAREER grant CHE9984895). We express our sincere gratitude to Kasia Janota from Agilent Technologies Inc. for performing accurate mass measurements. We thank Mike Meyer from the U.S. Geological Survey for providing the hog lagoon samples. SUPPORTING INFORMATION AVAILABLE Emission spectrum of fluorescent bulb used for irradiation experiments. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review April 19, 2004. Accepted July 27, 2004. AC0494127

Analytical Chemistry, Vol. 76, No. 20, October 15, 2004

6011