First Evidence for Occurrence of Hydroxylated ... - ACS Publications

The two nonsteroidal anti-inflammatory drugs diclofenac (DCF) and aceclofenac (ACF) were monitored for the first time together with their major human ...
0 downloads 0 Views 261KB Size
Anal. Chem. 2008, 80, 8135–8145

First Evidence for Occurrence of Hydroxylated Human Metabolites of Diclofenac and Aceclofenac in Wastewater Using QqLIT-MS and QqTOF-MS Sandra Pe´rez*,† and Damia` Barcelo´†,‡ IDAEA-CSIC, Department of Environmental Chemistry, Jordi Girona 18-26, Barcelona 08034, Spain, and Catalan Institute for Water Research (ICRA), Parc Cientı´fic i Tecnolo`gic de la Universitat de Girona, Edifici Jaume Casademont, Porta A, Planta 1, Despatx 13C/ Pic de Peguera, 15E-17003 Girona, Spain The two nonsteroidal anti-inflammatory drugs diclofenac (DCF) and aceclofenac (ACF) were monitored for the first time together with their major human phase-I metabolites, namely, 4′-hydroxydiclofenac (4′-OH-DCF) and 4′-hydroxyaceclofenac (4′-OH-ACF), in untreated and treated sewage samples, collected from a municipal wastewater treatment plant which operated a continuous activated sludge (CAS) treatment in parallel with membrane bioreactor (MBR) technology. Mean concentrations of DCF and 4′-OH-DCF in the influent samples amounted to 349 and 237 ng/L, respectively, whereas levels of 4′-OH-ACF (average, 59 ng/L) exceeded those of its parent drug ∼2fold (31 ng/L). Removal rates of 26 and 56% were achieved for 4′-OH-DCF following CAS and MBR treatment, respectively. The most efficient elimination was observed for 4′-OH-ACF in the MBR with only 4% of the influent concentration remaining in the treated sewage. Biodegradation experiments in batch reactors loaded with mixed liquor demonstrated that ACF underwent rapid ester cleavage to liberate DCF, thus constituting a possible source of DCF release during biological sewage treatment. Studies on the microbial metabolism of DCF (295 Da) in controlled laboratory settings allowed us to identify three novel aerobic biotransformation products. Structure elucidation by means of ultraperformance liquid chromatography-electrospray ionization-hybrid quadrupole-time-offlight-mass spectrometry in conjunction with H/D-exchange experiments unequivocally identified them as deriving from nitrosation of the hydroxyl group in the carboxylic acid moiety (324 Da) and from nitration of one of the aromatic ring systems (340 Da). A third microbial metabolite emerging in the test medium was assigned as dichlorobenzoic acid (190 Da), possibly formed by Ndealkylation of DCF and subsequent carboxylation. Taken together, this work constitutes the first report on the occurrence of ACF and the human metabolites 4′-OH-DCF and 4′-OH-ACF in wastewater, underpinning the need of incorporating metabolites excreted by humans in moni* To whom correspondence should be addressed. E-mail: [email protected]. Phone: ++34-93 400 6100 ext 423. Fax: + +34-93 204. 5904 † IDAEA-CSIC. ‡ Parc Cientı´fic i Tecnolo`gic de la Universitat de Girona. 10.1021/ac801167w CCC: $40.75  2008 American Chemical Society Published on Web 09/26/2008

toring surveys as part of a risk evaluation for environmentally relevant pharmaceuticals. The nonsteroidal anti-inflammatory drug diclofenac (DCF: 2-(2(2,6-dichlorophenylamino)phenyl)acetic acid; Figure 3A) constitutes one of the environmentally relevant pharmaceuticals having received considerable attention by the scientific community in the past couple of years due to frequent detection in monitoring surveys on sewage-impacted surface waters, associated with high consumption rates and low removal efficiencies during conventional activated sludge treatment in wastewater treatment plants (WWTP).1,2 Despite the great interest in the fate of DCF in engineered and environmental systems, two crucial components in the study on the overall fate of this polar chlorinated compound (logD7.4, 1.11;3 solubilityNa-DCF, 1113 mg/L4) have as of today been considered only marginally: On the one hand, the occurrence of human metabolites of DCF in environmental samples has not been documented at all; on the other hand, very little evidence has been published as regards metabolic pathways in complex microbial communities like those encountered in the aeration tank of the activated sludge treatment. As far as the metabolism in humans is concerned,5,6 as much as 50% of the total dose of DCF is excreted in urine and bile as 4′-hydroxy metabolite (4′-OH-DCF; Figure 3B) and a glucuronide conjugate thereof ssuggesting this hydroxylated metabolite to be possibly more abundant in untreated sewage than the parent drugswhile 10-20% of the oral dose correspond to unaltered DCF and its acylglucuronide. Other minor oxidative metabolites excreted by humans are the 5-hydroxy, 4′,5-dihydroxy, 3-hydroxy, and 3′-hydroxy-4′-methoxy derivatives of DCF either as such or conjugated with glucuronic acid (see Figure SI-1 in Supporting Information for comprehensive metabolic pathway). Only recently, a novel urinary metabolite of DCF was identified by liquid chromatography-nuclear magnetic resonance-mass spectrometry (LC-NMR-MS) and high-resolution MS as the lactam dehydrate (1) Hilton, M. J.; Thomas, K. V. J. Chromatogr., A 2003, 1015, 129–141. ¨ llers, S.; Singer, H. P.; Fa¨ssler, P.; Mu (2) O ¨ ller, S. R. J. Chromatogr., A 2001, 911, 225–234. (3) Menasse, R.; Hedwall, P. R.; Kraetz, J.; Pericin, C.; Riesterer, L.; Sallmann, A.; Ziel, R.; Jaques, R Scand. J. Rheumatol. 1978, 22, 5–16. (4) Fini, A.; Laus, M.; Orienti, I.; Zecchi, V. J. Pharm. Sci. 1986, 75, 23–25. (5) Stierlin, H.; Faigle, J. W; Sallmann, A Xenobiotica 1979, 9, 601–610. (6) Stierlin, H.; Faigle, J. W. Xenobiotica 1979, 9, 611–621.

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8135

of 4′-OH-DCF.7 With respect to microbial transformation pathways of DCF in the aquatic environment, one investigation carried out in a fixed-bed column bioreactor filled with riverine sediment under aerobic conditions revealed the transient formation of the p-benzoquinone imine of 5-hydroxy-DCF, which was characterized by LC-electrospray (ESI)-MS and NMR spectroscopy.8 Neither the presumed intermediate 5-hydroxy-DCF nor the formation of 4′-OH-DCF, constituting the main human metabolite, was observed. In fact, studies on DCF conducted in a pilot WWTP,9 gas chromatographic separation of the silylated sample extracts indicated the formation of three degradation products, which were tentatively identified through their electron impact-mass spectra as the intramolecular lactam, an alcohol corresponding to the reduction of the carboxylic acid, and a methoxy derivative thereof. In full-scale WWTP relying on continuous activated sludge (CAS) treatment for the degradation of organic compounds, the biotransformation of the acidic DCF has generally been estimated to be low based on comparisons of influent and effluent concentrations. Despite the consensus viewing DCF as fairly recalcitrant to microbial attack, for an unbiased interpretation of differences in influent and effluent levels of the WWTP, it needs to be taken into consideration that conjugated DCF can be liberated during the biological wastewater treatment process. As mentioned above, a fraction of DCF is excreted from the human body as acylglucuronide, which is potentially cleavable by β-glucuronidase enzymes, thus releasing the parent drug in its native form. In addition to this, a second possible source for DCF formation in the activated sludge tank is the ester hydrolysis of aceclofenac (ACF: 2-(2-(2-(2,6-dichlorophenylamino)phenyl)acetoxy)acetic acid; Figure 3C), which is a potent anti-inflammatory and analgesic drug with efficacy similar to DCF but with improved gastrointestinal tolerance.10 In analogy to DCF, the major human metabolite of ACF derives from hydroxylation at the 4′-position of the halogenated aromatic ring, which is then partly excreted as the glucuronide conjugate.11,12 Minor amounts of DCF and 4′-OHDCF, predominantly conjugated, have also been identified in urine. Given the lack of field data on 4′-OH-DCF and the fact that ACF and 4′-OH-ACF are potential precursors of DCF and its major hydroxylated metabolite, respectively, the present study aimed at investigating still uncovered aspects in the environmental life cycle of DCF. The two principal objectives were (a) to determine for the first time concentrations of the four target analytes in influent and effluent samples from a WWTP operating in parallel a CAS process and a pilot-scale membrane bioreactor (MBR), and (b) to conduct biodegradation experiments under controlled laboratory settings in order to gain further insight into the biodegradability and metabolic pathways of ACF, DCF, and their hydroxylated metabolites. Due to the difficulty of obtaining commercially available reference standards of 4′-OH-DCF and 4′(7) Stu ¨ lten, D.; Lamsho ¨ft, M.; Zu ¨ hlke, S.; Spiteller, M. J. Pharm. Biomed. Anal. 2008, 47, 371–376. (8) Groning, J.; Held, C.; Garten, C.; Clauβnitzer, U.; Kaschabek, S. R.; Schlo ¨mann, M. Chemosphere 2007, 69, 509–516. (9) Kosjek, T.; Heath, E.; Kompare, B. Anal. Bioanal. Chem. 2007, 387, 1379– 1387. (10) Pasero, G.; Marcolongo, R.; Serni, U.; Parnham, M. J.; Ferrer, F. Curr. Med. Res. Opin. 1995, 13, 305–315. (11) Bort, R.; Ponsoda, X.; Carrasco, E.; Go´mez-Lecho´n, M. J.; Castell, J. V. Drug Metab. Dispos. 1996, 24, 969–975. (12) Bort, R.; Ponsoda, X.; Carrasco, E.; Go´mez-Lecho´n, M. J.; Castell, J. V. Drug Metab. Dispos. 1996, 24, 834–841.

8136

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

OH-ACF, these metabolites were synthesized from DCF and ACF, respectively, by selective biocatalysis using recombinant human CYP2C9 enzyme. For the trace determination of DCF and the related compounds in wastewater, a quantitative analytical methodology was developed based on solid-phase extraction (SPE) followed by LC/ESI-MS/MS analysis on a hydrid quadrupolelinear ion trap (QqLIT) instrument. Moreover, samples from the biodegradation studies were screened for the presence of stable intermediates and these were characterized by hybrid quadrupoletime-of-flight (QqTOF)-MS in combination with H/D-exchange experiments leading to the discovery of unusual microbial transformation products. EXPERIMENTAL SECTION ChemicalandBiologicalReagents.DCF(CASNo.15307-79-6) was purchased from Jescuder (Rubı´, Spain) and ACF (CAS No. 89796-99-6) was provided by Laboratorios Almirall S.A. (Barcelona, Spain). All HPLC organic solvents were Chromasol LC grade. Deuterium oxide (g99.9%) was obtained from Euriso-top (Gif-Sur-Yvette, France). Ultrapure water, acetic acid-d4, glucose6-phosphate dehydrogenase from baker’s yeast 1 KU (Saccharomycer cerevisiae), D-glucose 6-phosphate disodium salt hydrate (98-100%), and β-glucuronidase (Escherichia coli) were purchased from Sigma Aldrich (Munich, Germany). Acetonitrile and methanol were from Riedel de Ha¨en (Steinheim, Germany). Dichloromethane, hydrochloric acid (25%), ammonium acetate, sodium bicarbonate, disodium hydrogen phosphate, magnesium chloride, and disodium EDTA were obtained from Merck (Darmstadt, Germany). All reagents were of ACS grade. Microsomes containing recombinant human CYP2C9, expressed in baculovirusinfected insect cells, and recombinant rabbit NADPH-P450 reductase were obtained from Sigma Aldrich. CYP2C9-Mediated Synthesis of 4′-OH-DCF and 4′-OHACF. Given the difficulty in finding commercially available reference compounds of the hydroxy metabolites of DCF and ACF, they were produced by CYP2C9-mediated oxidation of the parent compounds at the C-4′ position of the dichlorophenyl ring. The regioselective biotransformation13 of DCF and ACF was accomplished by using recombinant human CYP2C9, affording the two desired metabolites. To this end, a 25 µM solution of each substrate was prepared in 50 mM phosphate buffer (pH 7.4) at a final volume of 900 µL containing 1% acetonitrile. After addition of a 50-µL aliquot of recombinant hCYP2C9 (80 pmol/mg of protein) and a preincubation period of 3 min, the reaction was initiated by adding 50 µL of a 10 mM NADPH-generating system. The addition of four volumes of ice-cold acetonitrile terminated the reaction after incubating 3 h at 37 °C, the precipitated proteins were then separated by centrifugation for 10 min at 12000g, and the recovered supernatants were evaporated to dryness under a gentle stream of nitrogen. After reconstitution of the samples in 1 mL of water/acetonitrile (1:1), purity and concentration of the synthesized standard were verified by LC-diode array detector (DAD)-(+)ESI-MS/MS (Agilent 1100 Series coupled to Applied Biosystems API QTRAP 4000, see below). Inspection of the UV trace at the absorption maximum of 270 nm confirmed the complete conversion of the substrates into their 4′-hydroxy derivatives with a purity of >95%. The compound identities were corroborated by recording the (+)ESI product ion mass spectra (13) Leeman, T; Transon, C; Dayer, P. Life Sci. 1993, 52, 29–34.

Figure 1. Concentrations of DCF, 4′-OH-ACF, ACF, and 4′-OH-ACF in sewage samples collected at Rubi WWTP. Boxes indicate lowest, mean, and highest concentration; percentage given on effluent boxes corresponds to removal efficiency.

of m/z 312 and 370, respectively (Figure 3B and D). The fragmentation patterns were in agreement with analytical data reported in the literature,14 exhibiting a mass shift of +16 Da for all fragment ions relative to the parent drug (see Table A in Supporting Information for high-resolution MS data). The concentrations of 4′-OH-DCF and 4′-OH-ACF in the standard solutions were estimated by comparison of the UV signal intensities with DCF and ACF standard solutions, respectively, assuming identical extinction coefficients of the respective metabolites. Sampling and Sample Preparation. Design and operational parameters of the WWTP Rubi including the MBR are described elsewhere.15 From the CAS treatment and the pilot MBR, a total of six 24-h composite samples were collected over a period of 2 weeks in June 2007 at the following locations in the treatment line: effluent of the primary sedimentation tank (referred to as influent in Figure 1) and the final effluents after CAS and MBR treatment. The samples were kept on ice during transportation and processed upon arrival in the laboratory. In a first step, samples were filtered through a 1.2-µm glass fiber filter Whatman GF/C (Maidstone, England) and 0.45-µm nylon membrane filter (Whatman). Then, 200-mL aliquots of the influent and effluent samples were adjusted to pH 3.0 with HCl and concentrated by tandem SPE using Oasis HLB cartridges (200 mg; Waters (Milford, MA)) and Isolute ENV+ cartridges (200 mg; Biotage (Uppsala, Sweden)). The cartridges had been conditioned with 2 × 3 mL of methanol followed by 3 mL of water (pH 3.0). Samples were aspirated through the cartridges at a rate of ∼10 mL/min. After drying of the cartridges under vacuum for 15 min, the analytes were eluted separately into the same tube using 2 × 3 mL of methanol followed by 2 mL of ethyl acetate. The organic solvents were evaporated to dryness under a gentle stream of nitrogen gas, and the residues were reconstituted with 1 mL of water/acetonitrile (8:2). Isolation of ACF Metabolites from Human Urine. In order to gain ready access to milligram amounts of the 4′-hydroxy metabolites of ACF and DCF in the most cost-effective fashion, for the biodegradation studies, one healthy male volunteer was given a single oral dose of the over-the-counter drug Airtal (14) Kang, W.; Kim, E. Y. J. Pharm. Biomed. Anal. 2007, 46, 587–591. (15) Radjenovic, J.; Petrovic, M.; Barcelo, D. Anal. Bioanal. Chem. 2007, 387, 1365–1377.

(Laboratorios Almirall), corresponding to 100 mg of aceclofenac. Urine was collected over a time period of 12 h. After filtration of the sample through a 1.6-µm glass fiber filter (Whatman GF/CA), the target analytes were enriched from a 250-mL aliquot using the same SPE protocol as applied in the analysis of the wastewater samples (see above). The final SPE eluate was evaporated to dryness under a gentle stream of nitrogen and reconstituted in 1 mL of 50 mM phosphate buffer (pH 6.3). Given the complex metabolic profile in urine, characterized by the presence of ACF metabolites originating from both phase I (oxidation and hydrolysis) and phase II reactions (glucuronidation), this extract was treated with 100 µL of β-glucuronidase (type IX-A from E. coli, 1 000 000-5 000 000 units/g protein, Sigma-Aldrich) in order to cleave the (acyl)glucuronides formed in vivo. Comparison of the metabolic profiles before and after treatment by LC-(+)ESI-MS in full-scan mode (m/z 200-800) confirmed the complete hydrolysis of the glucuronide metabolites. The concentrations of the four major components, namely, ACF, DCF, 4′-OH-ACF, and 4′OH-DCF, were determined in the treated sample after 100-fold dilution using LC-(+)ESI-MS/MS in MRM mode. Finally, the treated extract was divided into equal volumes and transferred into the biologically active batch reactor and the control batch reactor. Biodegradation in Laboratory Batch Reactors. Amber 1-L glass bottles were loaded with 1000 mL of mixed liquor freshly collected from the aeration tanks of the WWTP in Rubi. Bubbling of air through Teflon tubing into the test medium provided continuous aeration of the system and ensured suspension of the sludge particulate matter (5 g/L). Three different scenarios were set up in the batch reactors: (A) ACF alone, spiked at 10 mg/L, (B) DCF alone, spiked at 10 mg/L, and (C) urine extract containing a mixture of ACF and its three principal metabolites. Biologically inactive control batch reactors were run in parallel under identical substrate conditions. The inhibition of the biochemical activity of the microbial community of the sludge through addition of formaldehyde (5%) was allowed to account for possible abiotic removal mechanisms (mixed liquor was stirred but not aerated in order to not to strip out the volatile formaldehyde). In all three instances, concentration profiles without any marked decline of the spiked test compound(s) were observed for the control reactors. For the monitored period, the pH of the mixed liquor in the batch reactors was maintained at 7.4 ± 0.3, while the ambient temperature was 20-22 °C. H/D-Exchange Experiments. In order to obtain additional evidence for the identities of the proposed structures of the DCF metabolites, the shifts in the m/z values of the molecular ions generated in the negative ion mode were measured on the QqTOF-MS instrument following H/D exchange.16,17 To this end, the bioreactor sample from experiment B, day 7, was separated on the UPLC column with the mobile phase made of (A) D2O + 0.1% CD3COOD and (B) acetonitrile (see below). HPLC/ESI-QqLIT-MS Analysis. The quantitative analysis of ACF and its three human metabolites (DCF, 4′-OH-ACF, and 4′OH-DCF), as well as the semiquantitative determination of the novel microbial DCF metabolites, was carried out on an Agilent Series 1100 liquid chromatograph coupled to an API 4000 QTRAP (16) Liu, D. Q.; Hop, C.E.C.A. J. Pharm. Biomed. Anal. 2005, 37, 1–18. (17) Eichhorn, P; Ferguson, P. L.; Pe´rez, S.; Aga, D. S. Anal. Chem. 2005, 77, 4176–4184.

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8137

mass spectrometer (Applied Biosystems/MSD Sciex, Foster City, CA). The chromatographic separations were achieved on a Phenomenex Synergy Polar-RP (100 × 3.0 mm, 2.5-µm particle size). The mobile phases were (A) 10 mM aqueous ammonium acetate/acetic acid (pH 5.8) and (B) acetonitrile. The gradient was as follows: isocratic for 1 min at 85% A, linear decrease to 5% A within 6 min, hold for 2 min, return to initial conditions in 1 min, and equilibration for 3 min. The flow rate was 600 µL/min and the injection volume was 10 µL. For the analysis of the four target compounds, the Turbo Ion Spray source was operated in the positive ion mode using the following settings for the ion source and mass spectrometer: curtain gas 30 psi, spraying gas 50 psi, drying gas 50 psi, drying gas temperature 700 °C, and ion spray voltage 5500 V. The transitions for multiple reaction monitoring (q1, quantifier ion; q2, qualifier ion), declustering potential (DP), collision energy (CE), and collision cell exit potential (CXP) were optimized as follows. for ACF: q1 m/z 354 f 215 (DP 56 V, CE 31 eV, CXP 16 V) and q2 m/z 354 f 250 (56 V, 21 eV, 6 V). for DCF: q1 m/z 296f215 (46 V, 27 eV, 18 V) and q2 m/z 296 f 250 (46 V, 21 eV, 6 V). for 4′-OH-ACF: q1 m/z 370 f 294 (176 V, 19 eV, 8 V) and q2 m/z 370f266 (161 V, 19 eV, 14 V). For 4′-OH-DCF: q1 m/z 312f231 (66 V, 39 eV, 12 V) and q2 m/z 312f266 (76 V, 35 eV, 20 V). UPLC/ESI-QqTOF-MS Analysis. Accurate mass measurements of ACF, DCF, 4′-OH-ACF, and 4′-OH-DCF as well as of the DCF biotransformation products formed in the batch reactors were carried out in full-scan and product ion scan mode using a Micromass QqTOF-system interfaced with a Waters Acquity UPLC system (Micromass, Manchester, UK). Samples from the biodegradation experiments were separated on a Waters Acquity BEH C18 column (50 × 2.1 mm, 1.7-µm particle size) equipped with precolumn (5 × 2.1 mm) of the same packing material. The mobile phases were (A) 10 mM aqueous ammonium acetate/ acetic acid (pH 5.8) and (B) acetonitrile. For the H/D-exchange experiments, mobile phase (A) was replaced by D2O/0.1% CD3COOD. After 1 min of isocratic conditions at 95% A, the portion of A was linearly decreased to 5% within 6 min. This condition was held for 2 min and then the initial mobile-phase composition was restored within 0.1 min and maintained for column regeneration for another 1.9 min. The flow rate was 400 µL/min. The injection volume was 10 µL. The MS analysis was performed with an ESI interface in the positive or negative ion mode applying a capillary voltage of +3500 and -3000 V, respectively. The nebulizer gas flow was set to 50 L/h and the drying gas flow to 600 L/h with a temperature of 350 °C. In order to minimize thermal degradation of the molecular ion of metabolite M324 in the ion source, the temperature of the drying gas was lowered to 75 °C. The TOF analyzer was operated at a resolution of 5000 (fwhm) and ESI mass spectra were recorded in 1-s intervals with automatic switching of the dual sprayer every 10 s for infusion of the internal calibrant for a duration of 1 s. Tyrosine-valine-tyrosine served as internal lock mass with [M + H]+ ) m/z 380.2185 or [M - H]- ) m/z 378.2029. External mass calibration for positive and negative ESI modes was conducted prior to analysis for the mass range of m/z 80-500 infusing a solution of acetonitrile/0.1 M NaOH/10% HCOOH (98:1:1) at a flow rate of 10 µL/min. All MS data acquisition and processing were done using the software package MassLynx V4.0. 8138

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

Quantitation and Method Performance Parameters. All quantitative analyses of wastewater and batch reactor samples were carried out on the QqLIT-MS. Recovery studies included spiked samples from sedimentation tank effluent (0.2 and 10 µg/ L) and final CAS effluent (0.05 and 2 µg/L). Acidified 200-mL samples (pH 3) were extracted using the SPE protocol described above. Nine-point calibration curves (r > 0.991) were constructed for the four target compounds in the range from 0.1 to 2500 µg/L (0.1, 0.5, 2, 10, 25, 100, 250, 1000, 2500 µg/L) either as matrixmatched standards or in pure water, adding the calculated concentrations of the target analytes to the SPE extract. For each matrix-matched standard of the calibration curve, the residual concentration of the target analytes in the wastewater matrix were subtracted. The recoveries ranged from 58 to 92% for all analytes with method detection limits (signal-to-noise ratio >5) between 0.2 and 5 ng/L. Safety Considerations. Recognizing the potential hazards associated with the sampling and handling of human wastes, a series of precautions were taken in order to minimize any health risks. Personnel involved in sample collection and preparation were trained and experienced, wore personal safety equipment for protection of skin and eyes, and had vaccinations against wastewater-borne pathogens (hepatitis A and B and tetanus). RESULTS AND DISCUSSION Presence of DCF, ACF, and Their Hydroxylated Human Metabolites in CAS and MBR Samples. Whereas the determination of DCF residues in wastewater samples has been widely reported in monitoring surveys at treatment facilities in many different countries, as of today no field data have been published as regards the occurrence and fate of its principal human metabolite, 4′-OH-DCF. Neither is there any information available with respect to the wastewater relevance of ACF, considered a potential precursor of DCF formation in biological wastewater treatment. Against this background, a first field study was initiated with the objective of profiling the presence of ACF, DCF, and their two hydroxylated metabolites, 4′-OH-ACF and 4-OH-DCF, in influent and effluent samples from a municipal WWTP. In order to assess the impact of the treatment technology on the removal efficiencies of the four target analytes, two treatment lines receiving sewage from the primary sedimentation tank were compared, namely, a conventional CAS system and a pilot-plant MBR installed at the WWTP under examination. In Figure 1, the average concentrations, corresponding to the 2-week time period studied, are represented for the influent sample and the two types of effluent sample. As far as the determination of DCF is concerned, the mean influent level amounting to 349 ng/L is within the lower range reported for municipal WWTPs.16-19 No significant removal is observed during CAS treatment with an effluent concentration of 338 ng/mL. In contrast, 85% of the initially present DCF is eliminated following passage through the MBR (effluent, 9.6 ng/mL), confirming the findings from other comparative studies on the extent of DCF removal in CAS systems versus MBRs.20-23 (18) Tauxe-Wuersch, A.; De Alencastro, L. F.; Grandjean, D.; Tarradellas, J. Water Res. 2005, 39, 1761–1772. (19) Gros, M.; Petrovic, M.; Barcelo´, D. Talanta 2006, 678, 690. (20) Bernhard, M.; Muller, J.; Knepper, T. P. Water Res. 2006, 40, 3419–3428.

On the other hand, the metabolite 4′-OH-DCF is almost as prominent in the untreated sewage as the parent compound with a mean concentration of 237 ng/mL. The positive detection of 4′-OH-DCF in the sewage, however, is no surprise given the metabolic pathway of DCF and its excretion profile in urine. In a group of subjects with normal renal function, 6.2 ± 1.8% of the oral dose was excreted in urine as intact parent drug, whereas 4′-OH-DCF made up 16 ± 3% of the dose (sample processing encompassed alkaline hydrolysis to release glucuronic acid-bound analytes).24 A similar ratio of DCF to its 4′-hydroxy metabolite in urine was reported in a clinical study involving six healthy volunteers.25 Following alkaline treatment of the samples, DCF constituted 13.6 ± 6.5% of the 100-mg rectal dose while 4′-OHDCF accounted for 27.2 ± 12.8% of the administered amount. At this point, it is imperative to take into consideration the presence of conjugates in human urine since in vivo both DCF and its hydroxylated metabolite are known to be transformed in UGTmediated metabolic reactions to yield the corresponding glucuronides that are subsequently excreted into urine.5 For estrogenic hormones, the occurrence of phase II metabolites in untreated sewage has been recognized by Brownawell and colleagues,26 who measured the distribution and fate of free estrone and β-estradiol along with their glucuronides and sulfates in influent and effluent samples from a municipal WWTP. That estrogen glucuronides are susceptible to undergo cleavage in aerobic activated sludge systems upon liberation of the free estrogen was demonstrated by incubating glucuronide conjugates in batch reactors containing a suspension of activated sludge.27 In the case of DCF, enzymatic hydrolysis of the acylglucuronide conjugate during the biological sludge treatment was postulated by Lishman et al.,28 in particular to account for negative removal efficiencies of DCF reported in various WWTP surveys.29 But as yet, no experimental data have been published providing evidence for this entirely plausible hypothesis. As far as the quantitative determination of ACF and its 4′-OH metabolite in the present study is concerned, the levels of this hydroxyacetic acid ester of DCF in the influent samples did not exceed 33 ng/mL (mean, 31 ng/mL) and therefore were ∼10 times lower than the residue concentrations of DCF. This is possibly a reflection of differences in consumption rates. After biological treatment of the wastewater in the CAS system and the MBR, the concentrations of ACF were reduced on average by 58 and 53%, respectively. The substantially higher removal rate of ACF in the CAS treatment as compared to DCF (17%) is believed (21) Radjenovic, J.; Petrovic, M.; Barcelo´, D. Anal. Bioanal. Chem. 2007, 387, 1365–1367. (22) Quintana, J. B.; Weiss, S.; Reemtsma, T. Water Res. 2005, 39, 2654–2664. (23) Kimura, K.; Hara, H.; Watanabe, Y. Environ. Sci. Technol. 2007, 41, 3708– 3714. (24) Sawchuk, R. J.; Maloney, J. A.; Cartier, L. L.; Rackley, R. J.; Chan, K. K. H.; Lau, H. S. L. Pharm. Res. 1995, 12, 756–762. (25) Landsdorp, D.; Vree, T. B.; Janssen, T. J.; Guelen, P. J. M. Int. J. Clin. Pharmacol. Ther. Toxicol 1990, 28, 298–302. (26) Reddy, S.; Iden, C. R.; Brownawell, B. J. Anal. Chem. 2005, 77, 7032– 7038. (27) Ternes, T. A.; Kreckel, P.; Mueller, J. Sci. Total Environ. 1999, 225, 91– 99. (28) Lishman, L.; Smyth, S. A.; Sarafin, K.; Kleywegt, S.; Toito, J.; Peart, T.; Lee, B.; Servos, M.; Beland, M.; Seto, P. Sci. Total Environ. 2006, 367, 544–558. (29) De Wever, H.; Weiss, S.; Reemtsma, T.; Vereecken, J.; Mu ¨ ller, J.; Knepper, T.; Ro ¨rden, O.; Gonzalez, S.; Barcelo, D.; Hernando, M. D. Water Res. 2007, 41, 935–945.

to be due to esterase-mediated enzymatic cleavage of the former, giving rise to the latter (see next section on degradation experiments). This process, which is also a relevant in vivo biotransformation pathway, involving the conversion of ACF into DCF and 4′-OH-ACF into 4′-OH-DCF,11 can be attributed to esterase activity in the activated sludge, which contains structurally diverse esterase enzymes with broad, and partially overlapping, substrate specificity.30 With respect to 4′-OH-ACF, its concentrations in the untreated sewage ranged from 43 to 82 ng/L and were thus throughout higher than those of ACF. Comparison of the removal efficiencies between ACF and its hydroxy metabolite reveal a noticeably larger difference between CAS and MBR in the case of 4′-OH-ACF, which is almost completely eliminated in the MBR (96%). The higher abundance of 4′-OH-ACF relative to ACF in the influent is consistent with their excretion pattern in urine. The ratio between the two compounds, recovered from urine following an oral 100-mg dose, ranged from 13 to 61 for the four healthy individuals examined.11 In the cited study, though, the ratio corresponded to enzymatically deconjugated samples following treatment with β-glucuronidase. Taken together, the study on the occurrence and fate of ACF and DCF in the WWTP illustrates that human drug metabolites can (a) be at least as prominent in untreated sewage as the parent compound and (b) display a distinct behavior in the biological wastewater treatment. Biodegradation Experiments in Batch Reactors. With the goal of gaining further insight into the biodegradation processes taking place in the activated sludge treatment, 1-L batch reactors amended with undiluted mixed liquor from the CAS aeration tank of the Rubi WWTP were spiked at elevated concentrations with individual compounds or mixtures. Three distinct conditions were examined: (a) ACF alone in order to collect evidence for a rapid biotransformation generating DCF; (b) DCF alone aiming to demonstrate that this environmentally relevant contaminant is fairly recalcitrant to microbial attack; and (c) a β-glucuronidasetreated urine extract, obtained from a healthy subject following an oral dose of ACF in order to elaborate differences in the biodegradability of the four analytes targeted previously in the WWTP survey. The plot in Figure 2A corresponds to the biodegradation profile of ACF at 10 mg/L. After one day, ∼90% of the initially present amount of the substrate is converted to DCF while a further decline in concentration is observed by day 2 resulting in a total removal of 99.5%. The formed DCF, in turn, is far less amenable to biodegradation and by day 6 ∼50% of the maximum DCF concentration is still remaining in the mixed liquor. These outcomes are consistent with the field data insofar as DCF shows little tendency in the CAS treatment, operated with a hydraulic retention time of 12 h, to be eliminated. As the influent concentration of ACF to the studied WWTP is ∼1 order of magnitude below the one of DCF, an increase in the DCF level due to ACF hydrolysis cannot be discerned. Yet the biodegradation experiment suggests that the enzymatic hydrolysis of ACF constitutes a source of DCF formation in the biological wastewater treatment and, thus, along with the deconjugation of its acylglucuronide, (30) Boczar, B. A.; Forney, L. J.; Begley, W. M.; Larson, R. J.; Federle, T. W. Water Res. 2001, 35, 4208–4216.

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8139

Figure 2. Concentration-time profiles corresponding to the biodegradation of (A) ACF in activated sludge batch reactor spiked at 10 mg/L, of (B) in activated sludge batch reactor spiked at 10 mg/L, and of (C) preconcentrated urine extract, containing ACF, DCF, 4′-OHACF, and 4′-OH-DCF, spiked into activated sludge batch reactor. Concentration of metabolites in experiment B are estimates derived from using DCF calibration standards.

may eventually contribute to the observation of poor or negative removal efficiencies of DCF. 8140

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

Amending the batch reactor with DCF at 10 mg/L resulted in the degradation profile displayed in Figure 2B, which is characterized by a slow, though steady, decline in concentration of the parent drug, lacking an apparent lag phase. By day 7, the total elimination amounts to 60%. Screening of the reactor samples allowed us to track down three unknown biodegradation products of which the metabolite referred to as M324 was detectable from day 3 on, while the other two metabolites, termed M190 and M340, emerged on day 4. Their structural elucidation and possible formation routes are described in detail in the following section. In a third experiment, a mixture of the four target analytes proceeded from a urine sample collected from a healthy volunteer following a 100-mg oral dose of ACF. In this instance, an aliquot of the 12-h sample was preconcentrated using the same SPE protocol as applied to the wastewater. Analysis of a 100-fold-diluted urine extract by LC-DAD-MS (QqLIT) allowed us to confirm the presence of several glucuronide conjugates through their (+)ESI product ion spectra (data not shown). The mass spectra of the four putative glucuronides of DCF (5.87 min), ACF (5.91 min), 4′-OH-DCF (5.26 min), and 4′-OH-ACF (5.48 min) with calculated ion masses of m/z 472, 530, 488, and 546, respectively, displayed the characteristic neutral loss of 176 Da31 originating from dissociation of the ether (hydroxy metabolites) or ester bond (carboxylic acids) with the glucuronic acid. The resulting major fragment ion in each spectrum was identical in the m/z value to the protonated molecule of the nonconjugated compound. Precursor ion fragmentation of the glucuronides at a higher collision energy of 35 eV yielded product ion profiles showing the same ions as the free compound (data not shown). In order to eliminate the influence of glucuronide cleavage on the biodegradation profiles in the batch reactor, the urine extract was incubated with bacterial β-glucuronidase for 3 h. The analysis of the obtained sample by LC-DAD-MS confirmed the completeness of the enzymatic cleavage of the glucuronides; however, the predominance of 4′-OH-DCF over the other three target compounds suggested that partial ester hydrolysis had taken place, brought about by residual esterase activity in the E. coli-derived glucuronidase preparation. Yet, this apparent change in composition was deemed irrelevant to the objective of the study (in fact, it is reasonable to assume that esterase enzymes produced by E. coli and other intestinal microorganisms, and being excreted with feces, may bring about in-sewer hydrolysis of (hydroxylated) ACF prior to arrival at the WWTP). As to the biodegradation profile in the batch reactor (Figure 2C), negligible removal characterizes the behavior of DCF, whereas the concentration of its 4′-OH metabolite decreases from 6 to 0.06 mg/L within the 5-day period monitored. For the levels of ACF and 4′-OH-ACF, in turn, a drop within the first two days is noticed followed by relatively constant concentrations of 2-3 µg/L during the remaining time. Identification of Microbial DCF Metabolites by UPLC-ESIQqTOF-MS. The key components for the successful structure elucidation of the three novel metabolites were mass spectral comparison with the parent compound in positive and negative ESI ion modes, accurate mass measurements of the fragment ions in the MS2 mode aiding in proposing elemental compositions, and H/D-exchange experiments for determining the number of (31) Levsen, K.; Schiebel, H.-M.; Behnke, B.; Dotzer, R.; Dreher, W.; Elend, M.; Thiele, H. J. Chromatogr., A 2005, 1067, 55–72.

Figure 3. (+)ESI-QqTOF-product ion spectra of (A) DCF, [M + H]+ ) m/z 296; (B) 4′-OH-DCF, [M + H]+ ) m/z 312; (C) ACF, [M + H]+ ) m/z 354; and (D) 4′-OH-ACF, [M + H]+ ) m/z 370. All spectra were acquired at collision energy of 10 eV. Underlined ion masses denote odd-electron ions.

Figure 4. (-)ESI-QqTOF-product ion spectra of (A) DCF, [M - H]- ) m/z 294 at collision energy of 25 eV; (B) metabolite M340, [M - H]- ) m/z 339 at collision energy of 25 eV, the position of the nitro group is hypothetical; and (C) metabolite M190, [M - H]- ) m/z 189 at collision energy of 5 eV. Underlined ion masses denote odd-electron ions.

exchangeable protons. As far as the behavior of DCF is concerned, it afforded good ionization efficiencies in both the positive and negative ion modes and yielded characteristic spectra rich in fragment ions (Figure 4A; Table SI-1 in Supporting Information). In the (+)ESI mode, the product ion profile of the protonated molecule, m/z 296, derives from the sequential loss of H2O (m/z 278) and CO (m/z 250) followed by cleavage of the chlorine atoms as radicals (m/z 215 and 180), which was corroborated through pseudo-MS3 experiments by selecting the fragment ions m/z 278, 250, and 215, generated by in-source CID, in the Q1 analyzer

followed by further fragmentation in the collision cell (spectra not shown). As regards the fragmentation of the deprotonated molecule, m/z 294 (Figure 4A), decarboxylation of the anion initiates the sequence affording the fragment ion at m/z 250. Unlike the positive ion mode, where the chlorine atoms are cleaved off as radicals, the 2-fold elimination of HCl (36 Da) is observed for m/z 250 producing the even-electron species m/z 214 and 178. To circumvent postulating C-C triple bonds, stable tricyclic fragment ion structures can be proposed instead (insets in Figure 4A). Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8141

Figure 5. UPLC-ESI-QqTOF chromatograms corresponding to biodegradation experiment of DCF in activated sludge bioreactor operated at a test concentration of 10 mg/L: (A) (-)-XIC of DCF (m/z 294); (B) (-)-XIC of metabolite M340 (m/z 339); (C) (-)-XIC of metabolite M324 (m/z 339); (D) (-)-XIC of metabolite M190 (m/z 189); (E) (-)-TIC (m/z 100-500); (F) (+)-TIC (m/z 100-500).

As regards the identity of the three novel metabolites, which were detected in the activated sludge batch reactor spiked with DCF (see previous section), it turned out that only the second metabolite eluting at 4.13 min could be ionized in both ion source polarities (Figure 5), whereas the other two biotransformation products emerging at 3.90 and 4.30 produced measurable signal intensities of the molecular ions exclusively in the (-)ESI mode. Apparently, their structures were not sufficiently basic as to allow for attachment of a proton, originating from the slightly acidic mobile phase (pH 5.8), during the ionization process in the ESI interface. Metabolite M324. Examination of the (+)ESI mass spectrum of the major metabolite suggested a molecular ion at m/z 295, i.e., with a nominal mass of only 1 Da less than the parent compound. Recording the product ion spectrum of this species in the (+)ESI mode resulted in a fragmentation pattern similar to DCF as depicted in Figure 6B, namely, the sequential loss of H2O, Cl•, and CO. Analysis of the m/z values of the precursor ion and the fragment ion m/z 277 for possible elemental compositions (Table 1) indicated that both were radicals differing from the DCF ions m/z 296 and 278 by a single hydrogen atom (see Table SI-1 in Supporting Information for DCF data). As metabolic intermediates with radical character are known to be highly reactive under standard conditionssin this instance, the mixed liquor would contain a vast number of possible reaction partnerssthe only plausible explanation was the erroneous assignment of the molecular ion at m/z 295. In order test the hypothesis of thermal conditions impeding the generation of the molecular ion of this metabolite in the ion source, the desolvation gas temperature was lowered from 350 to 75 °C, while the temperature of the drying gas, through which the ions were sampled into the analyzer, was maintained at 120 °C. Under these very gentle ionization condi8142

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

tions, two additional species appeared in the (+)ESI full-scan spectrum along with m/z 295: namely, m/z 342 and 383 (Figure 6A). The measured difference between the two ions was 41 Da and was attributed to an acetonitrile molecule (Table 1). The presence of a solvent molecule in this adduct ion was not surprising, as the reduced desolvation gas temperature was now insufficient to decluster the electrosprayed ion completely before being sampled into the mass analyzer. Determination of the exact mass of m/z 342 indicated an elemental composition of C14H14N3O3Cl2 (Table 1), i.e., an increment by H3N2O relative to m/z 295. The ion m/z 342 was proposed to correspond to the ammonium adduct of the metabolite, [M + NH4]+. Further evidence for this assumption was provided by recording the (-)ESI spectrum under identical source conditions. An ion at m/z 323 corresponding to the deprotonated molecule was detected (see below). Based on this information, the mass shift of the metabolite relative to DCF had to be +29 Da, with a change in elemental composition corresponding to the replacement of an H-atom by a nitroso group. That the introduction of the NO group into the molecule went along with the loss of either of the exchangeable protons, NH or COOH, was corroborated by H/D-exchange experiments in the negative ion mode. The chromatographic run carried out in deuterium oxide (+CD3COOD) and acetonitrile as the mobile-phase components demonstrated no shift in the m/z value of the [MD - D]-; hence, the deprotonated molecule of the metabolite was devoid of any exchangeable hydrogen. That the structural modification concerned the carboxylic acid but not the N-atom of the dichlorophenylaniline moiety was further evidenced by the product ion spectrum of the deprotonated molecule at m/z 323 (Figure 6C). No fragment ion corresponding to decarboxylation, as observed for the parent compound, was detectable (expected at m/z 279).

Figure 6. QqTOF-mass spectra of metabolite M324 acquired at a desolvation gas temperature of 75 °C: (A) (+)ESI full-scan spectrum; (B) (+)ESI product ion spectrum of m/z 295 at collision energy of 20 eV; (C) (-)ESI product ion spectrum of m/z 323 at collision energy of 10 eV. Underlined ion masses denote odd-electron ions. Table 1. Accurate Mass Measurements of the Biodegradation Products of DCF As Determined by UPLC-(()ESI-QqTOF-MS in Product Ion Scan Mode ion

a

elemental composition

calculated mass (m/z)

measured mass (m/z)

relative error (SD)a (ppm)

DBEb

[M-H][M-H-CO2]-

C7H3O2Cl2 C6H3Cl2

Metabolite M190, (-)ESI 188.9510 144.9612

188.9515 144.9609

+2.9(1.8) -1.7(1.0)

5.5 4.5

[M+NH4+CH3CN]+ [M+NH4]+ m/z 295 m/z 277 m/z 242 m/z 214

C16H17N4O3Cl2 C14H14N3O3Cl2 C14H11NO2Cl2 C14H9NOCl2 C14H9NOCl C13H9NCl

Metabolite M324, (+)ESI 383.0678c 342.0412c 295.0167 277.0061 242.0373 214.0424

383.0654 342.0426 295.0181 277.0071 242.0382 214.0322

-6.2(2.5) +4.1(1.7) 5.0(0.5) 3.4(1.1) 4.3(1.1) -4.7(1.5)

9.0 10.0 10.5 9.5

[M-H]m/z 249

C14H9N2O3Cl2 C13H9NCl2

Metabolite M324, (-)ESI 322.9900 249.0112

322.9982 249.0119

-2.5(0.3) +2.9(1.8)

10.5 9.0

[M-H]m/z 295 m/z 259 m/z 229 m/z 213 m/z 193

C14H9N2O4Cl2 C13H9N2O2Cl2 C13H8N2O2Cl C13H8NOCl C13H8NCl C13H8NO

Metabolite M340, (-)ESI 338.9939 295.0041 259.0274 229.0294 213.0345 193.0528

338.933 295.0046 259.0258 229.0284 213.0324 193.0539

-1.8(2.8) +1.7(0.6) -5.8(1.2) -4.7(1.4) -10(2.5) +5.9(1.8)

10.5 9.5 10.5 10.0 10.0 11.0

N ) 3. b Double bond equivalents. c Measured in scan mode.

Taking together all the pieces of evidence, the formation of the metabolite was rationalized as O-nitrosation of the hydroxyl group of the carboxylic acid moiety in DCF, undoubtedly being an untypical metabolic pathway brought about by microorganisms (see below for further discussion). The reason why no protonated molecule of the metabolite, denoted as M324, could be obtained even under the low temperature settings, can be attributed to two circumstances: on one hand, the basicity of the N-atom is substantially diminished by the two chlorine atoms exerting a strong negative inductive effect on the aromatic ring and the adjacent N-atom, which consequently also holds true for DCF and its derivatives where the site of protonation is possibly an O-atom

rather than the N-atom (protonation of the secondary amine would have inactivated the molecule for the intramolecular amide formation according to a nucleophilic attack mechanism). On the other hand, the highly electronegative nitroso group reduces the basicity of the oxygen atoms to such an extent that no stable proton adduct is formed. The ammonium adduct of M324, in turn, is suggested to owe its relative stability to the chelate-like hydrogen-bonding mechanism between one of the ammonium H-atoms and the two O-atoms of the side chain as illustrated in the inset in Figure 6A. Isolation of the [M + NH4]+ in the quadrupole analyzer and subsequent collision-induced dissociation in the collision cell yielded exclusively the radical cation m/z 295, Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8143

possibly through proton transfer from the ammonium cation to the carbonyl oxygen and concommitantly homolytic cleavage of the CO-NO bond (loss of NO•). The resulting carbocation is further subject to nucleophilic attack from the N-atom upon dehydration (see inset in Figure 6A and product ion spectrum in Figure 6B). In contrast to DCF, which undergoes ring contraction upon extrusion of CO prior to the loss of the Cl radical, this order is reversed in the case of M324. Reestablishing an even-electron species is apparently favored over the loss of CO. With respect to the site of nitrosation in the DCF molecule, the microbial N-nitrosation of secondary amines in the presence of nitrate as nitrite source has been reported to occur in various environments, including lake water, soil, and wastewater.32 The substrates ranged from simple amines such as dimethylamine and diethanolamine to the fluoroquinoline antibiotics norfloxacin33 and ciprofloxacin, the latter two being N-nitrosylated on the piperazine ring by two strains of Mycobacterium.34 In a study on the in vitro N-nitrosation of medicinal drugs mediated by bacteria present in human saliva,35 the formation of the nitroso derivatives was proven indirectly by photolytical or acid-catalyzed hydrolysis with subsequent derivatization and colorimetrical detection of the azo dye formed. Of the 23 drugs assayed, 11 yielded detectable amounts of the nitroso compounds with yields differing by a factor of 800. Interestingly, the assayed DCF ranked second in terms of nitrosated product yield but it was the probe substance with the least basic secondary amine, supposed to be the reaction site for the nitrosation. Unfortunately, the analytical method employed did not enable distinction between possible nitroso regioisomers. For comparative purposes, the N-nitroso compound was studied here following chemical synthesis by classical nitrosation reaction with sodium nitrite in acidic solution.36 The product ion spectrum of the deprotonated N-nitroso derivative of DCF (m/z 323) was very similar to DCF characterized by stepwise elimination of CO2 (m/z 279), NO (m/z 249), and two HCl molecules (m/z 213 and 177) (Figure S1 in the Supporting Information). Metabolite M340. As far as the DCF metabolite eluting at 4.30 min is concerned (Figure 4B), its molecular ion [M - H]- was confirmed to be m/z 339 at both standard and low-desolvation gas temperature. The accurate mass measurements indicated a substitution of an H-atom by a nitro group (C14H9N2O4Cl2 for the metabolite versus C14H10NO2Cl2 for DCF; see Table 1). In this instance, the carboxylic acid moiety was not altered during the metabolization as evident by the initial loss of CO2 in the (-)ESI product ion spectrum (Figure 4B). The H/D-exchange experiment revealed that the structure of the unknown contained two exchangeable H-atoms; the [MD - D]- was shifted by one mass unit with respect to the [M - H]- ion (spectrum not shown). With the carboxylic acid being deprotonated (dedeuteronated) during the ionization process, the second exchangeable hydrogen remained therefore on the secondary amine bridging the two aromatic rings. After the decarboxylation of m/z 339 to 295, the (32) Alexander M. Biodegradation and Bioremediation; Academic Press, San Diego, CA, 1994; pp 53-55. (33) Adjei, M. D.; Heinze, T. M.; Deck, J.; Freeman, J. P.; Williams, A. J.; Sutherland, J. B. Appl. Environ. Microbiol. 2006, 72, 5790–5793. (34) Adjei, M. D.; Heinze, T. M.; Deck, J.; Freeman, J. P.; Williams, A. J.; Sutherland, J. B. Can. J. Microbiol. 2007, 53, 144–147. (35) Ziebarth, D.; Spiegelhalder, B.; Bartsch, H. Carcinogenesis 1997, 18, 383– 389. (36) Vogel, A. I. Practical Organic Chemistry; Longman Group: London, 1948.

8144

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

further fragmentation pathway was consistent with the loss of HCl, giving rise to m/z 259, the elimination of NO• (m/z 229), and a second loss of HCl (m/z 193) (Table 1). As regards the loss of the nitric oxide radical from m/z 259, this mechanism is commonly observed for aromatic nitro compounds.37,38 Assembling the available information on the structure elucidation of M340, the metabolite was obviously a product of microbial nitration at one of the aromatic rings. As compared to the series of studies dealing with bacterial nitrosation, the body of literature mentioning microbial nitration of environmentally relevant compounds is very scarce: From activated sludge, Sylvestre et al.39 isolated a strictly aerobic, Gramnegative bacterial strain that was capable of converting pchlorobiphenyl into a hydroxnitro derivative in the presence of nitrate. Mechanistic studies suggested a first-step monooxygenasemediated formation of a reactive arene oxide intermediate, which subsequently was subject to nucleophilic attack by nitrate or nitrite. In the present work, however, no concomitant ring hydroxylation was observed. In the case of the metabolite M340, the fragmentation pattern observed in the (-)ESI-MS/MS mode did not allow for confident assignment of the position of the nitro group. Nonetheless, two opposing scenarios can be considered in dependence on the possible reaction intermediate involved in the nitration of DCF. Supposing that the transfer of the nitro group involves an attack by an electron-deficient group, such as a nitronium ion, the aromatic ring of the phenylacetic acid would be the preferred target. Conversely, an electron-rich molecule, such as a nitrite anion, would be attracted to the other phenyl ring, which is activated for a nucleophilic attack by the presence of the two chlorine atoms. Metabolite M190. Besides the two metabolites M324 and M340, originating from O-nitrosation and aromatic nitration of DCF, respectively, a third biodegradation product emerged at a retention time of 3.90 min (Figure 5). The exact mass measurements of the deprotonated molecule at m/z 189 led to proposing the elemental composition C7H3O2Cl2, being in line with the observation of the chlorine isotope cluster in the full-scan mode. Acquisition of the product ion spectrum of the [M - H]- ion resulted in a fragment ion with m/z 145, which was attributed to the loss of CO2 out of a carboxyl group. (Figure 4C; Table 1). Chromatographic separation of this metabolite in the H/D-exchange mode using D2O in the mobile phase in place of H2O yielded an [MD D]- ion of m/z 189, confirming that the uncharged molecule contained a single exchangeable proton. Given these facts, the most probable structure corresponded to dichlorobenzoic acid. Since the dichlorophenyl ring in DCF does not bear any alkyl group, the metabolic conversion of the parent compound is proposed to have proceeded via N-dealkylation and carboxylation of the chlorine-containing phenyl ring. Environmental Implications. Regarding the circumstances leading to the generation of the microbial metabolites M324 and M340 in the batch reactor spiked with DCF, it seems reasonable (37) Schmidt, A.-C.; Herzschuh, R.; Matysik, F.-M.; Engewald, W. Rapid Commun. Mass Spectrom. 2006, 20, 2293–2302. (38) Levsen, K.; Schiebel, H.-M.; Terlouw, J. K.; Jobst, K. J.; Elend, M.; Preiβ, A.; Thiele, H.; Ingendoh, A. J. Mass Spectrom. 2007, 42, 1024–1044. (39) Sylvestre, M.; Masse´, R.; Messier, F.; Fauteux, J.; Bisaillon, J.-G.; Beaudet, R. Appl. Environ. Microbiol. 1982, 44, 871–877.

to postulate a link to nitrifying wastewater bacteria, which in a first stage of the microbiological process bring about oxidation of ammonia to nitrite (bacteria of the genera Nitrosomonos, Nitroscoccus, and Nitrososprira), followed by further oxidation of the nitrite to nitrate (involves Nitrobacter, Nitrococcus, Nitrospira, and Nitrospina). Although the contribution of nitrifying bacteria to the biomass in the mixed microbial community of the activated sludge tank in WWTPs is less than 5%, the operational conditions of the laboratory-scale reactors were favorable for the growth of nitrifiers in terms of oxygen supply and temperature and pH of the mixed liquor. The potential of nitrifying activated sludge bacteria for breaking down the antimicrobial trimethoprim was highlighted in study by Eichhorn et al.17 comparing its biodegradability in batch reactors amended with activated sludge and nitrifying activated sludge. The degradation studies revealed that nitrifying sludge bacteria were capable of facilitating an oxidation of trimethoprim, a pharmaceutical that is not amenable to biological degradation in a conventional activated sludge. Irrespective of mechanistic aspects and the interest in identifying the responsible bacteria for DCF conversion in the present work, it needs to be stressed that both M324 and M340 are the result of biotransformation rather than biodegradation as the chemical modification, unfortunately, does not imply a breakup of the core

structure. Last, but not least, it is worth noting that metabolite M324 was observed in independent fate studies40 of DCF conducted in flow-through bioreactors simulating the biological treatment of a WWTP. The O-nitrosation product was detectable at DCF concentrations spiked into the influent of as low as 5 µg/ L.

(40) Kosjek, T.; Heath, E.; Pe´rez, S.; Petroviæ, M.; Barcelo´, D., in preparation.

AC801167W

ACKNOWLEDGMENT The work presented in this article was supported by the Spanish Ministerio de Educacio´n y Ciencia, CEMAGUA (CGL200764551/HID). This work reflects only the author′s views and the European Community is not liable for any use that may be made of the information contained therein. S.P. acknowledges a postdoctoral contract from I3P Program (Itinerario Integrado de Insercio´n Profesional), cofinanced by CSIC and European Social Funds. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review June 9, 2008. Accepted August 13, 2008.

Analytical Chemistry, Vol. 80, No. 21, November 1, 2008

8145