Transformation of Oxcarbazepine and Human Metabolites of

Article pubs.acs.org/est

Transformation of Oxcarbazepine and Human Metabolites of Carbamazepine and Oxcarbazepine in Wastewater Treatment and Sand Filters Elena Kaiser,† Carsten Prasse,† Manfred Wagner,‡ Kathrin Bröder,† and Thomas A. Ternes*,† †

Federal Institute of Hydrology (BfG), Am Mainzer Tor 1, 56068 Koblenz, Germany Max Planck Institute for Polymer Research (MPI), Ackermannweg 10, Mainz, Germany

‡

S Supporting Information *

ABSTRACT: Carbamazepine (CBZ) and oxcarbazepine (OXC) are widely used anticonvulsants that are extensively metabolized in the human body. The pharmaceuticals and their human metabolites are present in influents and effluents of wastewater treatment plants (WWTPs), in rivers and streams, and in drinking water. In this study, the biodegradation of OXC and its main human metabolite, 10-hydroxyCBZ (10OHCBZ), and the main human metabolite of CBZ, 10,11-dihydro-10,11dihydroxy-CBZ (DiOHCBZ), was investigated in contact with activated sludge from a wastewater treatment plant (WWTP) and sand filter material from a waterworks. The transformation of DiOHCBZ, 10OHCBZ, and OXC led to the formation of the following main TPs: 1-(2-benzoic acid)-(1H,3H)-quinazoline-2,4-dione (BaQD), 1-(2benzoic acid)-(1H,3H)-quinazoline-2-one (BaQM), 9-aldehyde-acridine, 9-carboxylic acid-acridine (9-CA-ADIN), hydroxyl 9-CA-ADIN, acridone (ADON), 11-keto-OXC, and 2,2′-(carbamoylazanediyl)dibenzoic acid. TP formation could be explained by three proposed transformation pathways, including reactions such as oxidation, α-ketol rearrangement, or benzylic acid rearrangement. The results highlight the fact that the TP abundances strongly depend on the concentrations spiked in the labscale experiments. BaQD, 9-CA-ADIN, and ADON were detected in WWTP effluents, rivers, and streams. 9-CA-ADIN was found at maximum concentrations in WWTP effluent and rivers up to 920 ± 50 ng L−1 and 304 ± 26 ng L−1, respectively. Even in drinking water, BaQD and 9-CA-ADIN were present at concentrations of 26 ± 2 ng L−1 and 189 ± 3 ng L−1, respectively.

■

and is oxidized to a small extent to DiOHCBZ.10−12 Recent studies have shown that DiOHCBZ, 10OHCBZ, and OXC are present in the influent of wastewater treatment plants (WWTPs) at median concentrations of 4, 0.49, and 0.30 μg L−1, respectively.9,13 For OXC, a decrease in concentration was observed during conventional wastewater treatment,13 whereas DiOHCBZ and 10OHCBZ were discharged into receiving waters with nearly the same loads at which they are entering the WWTPs.8 Biological treatment is an important process for both WWTPs and waterworks. Nitrification and denitrification in WWTPs are utilized to limit the discharge of nutrients into surface waters and play an important role in the removal of organic compounds, including organic micropollutants. In waterworks, rapid and slow sand filtration are frequently applied to eliminate pathogens, to reduce turbidity, and to control taste and odor problems. Sand filtration has also been shown to remove selected organic micropollutants to some extent.14

INTRODUCTION The usage of pharmaceuticals has led to their ubiquitous presence in the aquatic ecosystem, with so far widely unknown consequences for the environment and human health.1 Although a large number of studies has been published dealing with the fate of pharmaceuticals, only little is known about the fate of their human metabolites. Carbamazepine (CBZ) is one of the most frequently detected pharmaceuticals in the aquatic environment and is present in treated wastewater, surface water, and drinking water at concentrations of up to 6.3, 1.1, and 0.03 μg L−1, respectively. 2−6 However, in the human body, CBZ is metabolized to approximately 70%, leading to the formation of more than 30 metabolites6 that are mainly excreted via urine.7 The predominant human metabolites of CBZ are 10,11-dihydro10,11-dihydroxy-CBZ (DiOHCBZ) and the N-glucuronide of CBZ that can be retransformed to CBZ during wastewater treatment.8 Oxcarbazepine (OXC), the 10-keto analogue of CBZ, exhibits similar chemical and therapeutic properties but fewer side effects.9,10 In 2012, CBZ was prescribed in Germany in an amount of 47.3 t, while OXC was prescribed to a minor extent with only 12.8 t.9 Similar to CBZ, OXC undergoes extensive metabolism leaving only 2−4% of OXC unaltered,11 forming as a main metabolite 10-hydroxy-CBZ (70% of the initial OXC concentration, 10OHCBZ), which also has anti-epileptic effects © XXXX American Chemical Society

Received: May 19, 2014 Revised: July 22, 2014 Accepted: July 24, 2014

A

dx.doi.org/10.1021/es5024493 | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

Article

(OXC-d4), and 13C,15N-carbamazepine (13C15N-CBZ) were added as internal standards (final concentration of 400 ng L−1). In addition to the parent compounds, transformation kinetics were also determined for those TPs that could be isolated in sufficient quantities. To distinguish between biotic and abiotic transformation processes, sterile control experiments (autoclaved for 15 min at 121 °C) were conducted. All experiments were conducted at least in duplicates. The samples from the transformation kinetic experiments were analyzed using a tandem mass spectrometer (AB Sciex QTrap 6500) equipped with an electrospray ionization interface (ESI), coupled with a high-performance liquid chromatography (HPLC) system (Agilent 1260 Series, Agilent Technologies, Waldbronn, Germany). Analysis was conducted in positive ion mode for all substances, using multiple-reaction monitoring (MRM). For the chromatographic separation, a Synergi HydroRP column (150 mm × 3 mm, 4 μm) with a Security Guard column (AQ C18, 4 mm × 2 mm inside diameter (i.d.); both from Phenomenex, Aschaffenburg, Germany) was used. For determining recoveries and matrix effects, water samples were spiked with 0.1 and 2 μg L−1 of target compounds and TPs isolated before. A 10-point calibration curve (0−10000 ng L−1) of the target compounds, their surrogate standards, and selected TPs (with available standards) was used for quantification. Limits of quantification (LOQs) in groundwater were defined as the second lowest calibration point with a signal-to-noise ratio (S/N) of >10 for the first transition (MRM 1) and >3 for the second transition (MRM 2). For the other matrices, LOQs were estimated by calculating a S/N ratio of 10 considering the S/N ratios obtained from spiked water samples in comparison to those of nonspiked samples (0.1 and 2 μg L−1). Further information about the analytical methods (e.g., recoveries and LOQs) is reported in the Supporting Information. Generation and Isolation of Transformation Products. For the isolation of TPs, the same experimental setup as described above for the elucidation of degradation kinetics was used, but individual compounds were spiked at elevated concentrations (50 mg L−1). Transformation of the parent compounds was monitored with an LTQ-Orbitrap mass spectrometer (LTQOrbitrap Velos) coupled to an Accela HPLC system (all from Thermo Scientific, Bremen, Germany). After incubation, the supernatant was decanted and filtered through 0.45 μm filters. In addition, the sand filter material was washed once using ultrapure water, combined with the filtrate, immediately frozen (−25 °C), and lyophilized. For the isolation of individual TPs, a HPLC system (Agilent 1100 and 1200 Series, Agilent Technologies, Waldbronn, Germany) consisting of a G2260 autosampler, a G1311A quaternary HPLC pump, a G1322A degasser, a G1316A column oven, and a G1215B DAD (scan range of 210−366 nm) was coupled to an automated fraction collector (Super Fraction Collector, SF2120, Techlab, Braunschweig, Germany). Chromatographic separation of TPs was achieved using a semipreparative HPLC column (Hydro-RP column (250 mm × 10 mm, 4 μm) with a Security Guard column (AQ-C18, 10 x 10 mm i.d.), both from Phenomenex, Aschaffenburg, Germany) and 0.2% formic acid (A) and methanol + 0.1% formic acid (B) as mobile phases. The gradient program was adapted for each parent compound to ensure optimal chromatographic separation of formed TPs. Residual methanol of individual fractions was removed using a rotary evaporator (Rotavapor, R-215, Büchi Labortechnik GmbH, Essen, Germany). The remaining aqueous phases were

The aim of this study was to elucidate the biotransformation of DiOHCBZ, 10OHCBZ, and OXC in selected biological processes in both WWTPs and waterworks. The substances have been chosen because of their widespread presence in raw wastewater as well as in drinking water resources. To the best of our knowledge, the biotransformation of DiOHCBZ, 10OHCBZ, and OXC has not been studied in detail. The biotransformation of the compounds was investigated in labscale systems in contact with both activated sludge from a nitrification tank of a WWTP and sand filter material from a waterworks. A main goal was to identify the transformation products (TPs) formed under aerobic conditions, to elucidate the (bio)transformation pathways, and to compare the formation of TPs in both biological wastewater treatment and sand filters. In addition, the TPs and their precursors were screened for potential toxicity using a TLC−AMD assay with Vibrio f ischeri. The environmental relevance of the identified TPs was shown by the analysis of influents and effluents of WWTPs, riverwater as well as feed water and finished drinking water from waterworks.

■

CHEMICALS AND METHODS Chemicals. Carbamazepine, 10,11-dihydro-10-hydroxycarbamazepine, 10,11-dihydro-10,11-dihydroxycarbamazepine, and oxcarbazepine as well as internal standards 10,11-dihydro10-hydroxycarbamazepine-d3 (10OHCBZ-d3) and oxcarbazepine-d4 (OXC-d4) were purchased from Toronto Research Chemicals (North York, ON). 9-Carboxylic acidacridine (9-CA-ADIN) was purchased from Santa Cruz Biotechnology (Dallas, TX) and acridone (ADON) from Sigma-Aldrich (St. Louis, MO). 13 C 15 N-carbamazepine (13C15N-CBZ) was obtained from Campro Scientific (Berlin, Germany). Stock solutions (1 mg mL−1) and working solutions (10 μg mL−1) were prepared in methanol (picograde, VWR) and stored at −24 and 4 °C, respectively. Dimethyl sulfoxide-d6 (DMSO-d6) (99.96%) used for NMR analysis was purchased from Deutero GmbH (Kastellaun, Germany). All other chemicals were purchased from Carl Roth (Karlsruhe, Germany) if not stated otherwise. Methods. Transformation Kinetics. Laboratory batch experiments were used for the elucidation of transformation kinetics of the target compounds. For lab-scale experiments with sand filter material, amber glass bottles were filled with 50 g of sand filter material taken from a German waterworks (groundwater, aerated and filtered with a rapid sand filter, height of 1.8 m, flow rate of 2.5 m h−1, backwashing every 10th day for 30 min) and 200 mL of pristine groundwater from a well in KoblenzArenberg, Germany. Activated sludge was sampled from a nitrification zone of a conventional WWTP (Koblenz-Wallersheim, 320000 population equivalents, daily flow rate of 61000 m3, hydraulic retention time of 24 h, sludge retention time of 12 days). The sludge (total suspended solids, 4 gss L−1; total organic carbon, 0.3 g gss−1) was diluted 1:20 with WWTP effluent to minimize sorption effects. The experiments were initiated within 3 h of sampling of the activated sludge. Bottles were aerated with a defined ratio of air and CO2 to maintain aerobic conditions and a constant pH between 7.0 and 7.5. After an equilibration period of 2−3 h, the target compounds were spiked individually at a concentration of 5 μg L−1. At defined sampling times, 2 mL were taken, filtered through a syringe filter (0.45 μm, cellulose acetate, Sartorius stedim biotech, Sartorius Biolab Products), and immediately frozen (−25 °C). Prior to analysis, 10,11-dihydro-10-hydroxycarbamazepine-d 3 (10OHCBZ-d 3 ), oxcarbazepine-d 4 B



Article

Table 1. Chemical Formulas, (proposed) Structures, Fragmentation Patterns (MSn−CID and/or −HCD experiments, ESI-pos. if not stated otherwise), and 1H NMR Chemical Shifts (if available) of Parent Compounds DiOHCBZ, 10OHCBZ, and OXC and Their TPs

C



Article

Figure 1. Proposed transformation pathway of DiOHCBZ, 10OHCBZ, and OXC. Compounds in brackets are proposed intermediates. The characters in brackets and their explanation can be found in the text. Reactions are (a) dehydrogenation for DiOHCBZ, hydroxylation for 10OHCBZ with subsequent dehydrogenation, and hydroxylation for OXC, (b) C10−C11 bond cleavage, (c) oxidation, (d) intramolecular ring closure and loss of water, (e) oxidation, (f) loss of the carbamoyl group, (g) α-ketol rearrangement, (h) ring contraction, (i) elimination of the hydroxyl group, (j) oxidation, (k) loss of the carbamoyl group followed by oxidation (DiOHCBZ) and hydroxylation followed by an oxidation (OXC and 10OHCBZ), (l) benzylic acid rearrangement, (m) loss of water, (n) decarboxylation, hydroxylation, and oxidation, and (o) hydroxylation.

the proposed structures of the TPs were verified. In addition, available reference standards were used for final confirmation. Further information about the NMR experiments can be found in the Supporting Information. Bioassay with V. f ischeri. To screen for potential toxic effects of CBZ, OXC, DiOHCBZ, and 10OHCBZ as well as their TPs on microorganisms, a HPTLC bioassay with V. f ischeri was used. To this end, different quantities of the target compounds and the TPs (0−1000 ng) were sprayed on HPTLC plates to obtain dose−response curves and the lowest observed effect concentrations (LOECs). For detailed information, see the Supporting Information. Detection of TPs in Environmental Samples. Grab samples of influents and effluents of WWTPs, five German rivers (Rhine River, Lippe River, a small canal, and two small streams) with different proportions of wastewater, as well as feed water and drinking water of waterworks were taken, to evaluate the occurrence of TPs. The samples (triplicates) were immediately filtered, spiked with internal standards (10OHCBZ-d3, OXC-d4,

analyzed via HPLC−UV to check the purity of isolated TPs prior to lyophilization. The received purified reference standards (purity >95%) were used for TP identification and for the preparation of a stock solution for quantification in the batch systems as well as environmental samples. Identification of Transformation Products. High-resolution mass spectrometry (LTQ OrbitrapVelos, Thermo Scientific) in positive and negative ESI mode was used to obtain exact masses and fragmentation patterns of formed TPs (tandem mass spectrometry (MS2) and triple mass spectrometry (MS3) experiments; details of analytical method as well as mass fragments are given in the Supporting Information). On the basis of these data, tentative structures of TPs were proposed. Confirmation of proposed chemical structures of the isolated TPs was achieved using nuclear magnetic resonance spectroscopy (NMR) (e.g., 1H NMR, 13C NMR, 1H, 1H COSY, 1 H, 13C HSQC, and 1H, 15N HSQC). For this, isolated TPs (0.3−2 mg) were dissolved in DMSO-d6. On the basis of the number of protons, their multiplicity, and their chemical shifts, D



Article

Figure 2. Ratios of molar concentrations of DiOHCBZ (a and b), 10OHCBZ (c and d), OXC (e and f), and OXC autoclaved (g and h) and their TPs in contact with sand filter material and activated sludge. Experiments were conducted at concentrations of 5 μg L−1.

and 13C15N-CBZ), and analyzed within 72 h via direct injection liquid chromatography with tandem mass spectrometry (using the same method applied for samples from the batch experiments).

■

general procedure applied for the identification of TPs, the elucidation of the chemical structures of TP282 is discussed in detail. From HRMS (ESI-pos.), an exact mass of 283.0712 Da and a chemical formula of C15H11O4N2 (Δppm −0.47) were obtained for TP282. Fragmentation experiments (MS2 and MS3) conducted in positive ion mode (CID 35% and HCD 80%) revealed the formation of several fragments (see also Figure S14 of the Supporting Information): m/z 265.0606 (−H2O), 240.0654 (−CHON), 222.0547 (−H2O, −CHON), 196.0756 (−CHON, −CO2), and 194.0559 (−H2O, −CHON, −CO).

RESULTS AND DISCUSSION

Identification of TPs. The formation of several TPs was observed during the incubation of DiOHCBZ, 10OHCBZ, and OXC with sand filter material from a waterworks as well as with activated sludge from a municipal WWTP. As an example for the E



Article

Transformation Pathway II. In the second transformation pathway, the formation of TP268 and the loss of a carbamoyl group (f) are followed by an α-ketol rearrangement (g) leading to a ring contraction (h).21 The subsequent elimination of a hydroxyl group (i) forms 9-aldehyde-acridine (9-CHO-ADIN, TP207).22,23 The oxidation of the aldehyde group leads to the formation of 9-CA-ADIN (j). Transformation Pathway III. In the third pathway, TP223A was formed by the loss of the carbamoyl group,20 an oxidation (for DiOHCBZ), and a hydroxylation followed by an oxidation (for OXC and 10OHCBZ) at position C10 and/or C11 (k). The formation of 9-CA-ADIN can be explained by a ring contraction caused by a benzylic acid rearrangement of TP223B (l), as previously shown for the biosynthesis of B-norviridiol lactone from viridiol.24 The consecutive loss of water resulted in the formation of 9-CA-ADIN (m). Decarboxylation, hydroxylation, and oxidation of the newly formed hydroxyl group to the respective ketone lead to the formation of ADON (n). In addition, hydroxylation of one of the aromatic ring of 9-CAADIN yielded the formation of hydroxy-9CA-ADIN (o). Also ADON was further transformed (see Figure S3 of the Supporting Information) without additional TPs being detected. Mass Balances and Sequence of Transformation Product Formation. The mass balances of the transformation observed in the lab-scale systems were calculated by considering the TPs BaQD, 9-CA-ADIN, and ADON (Figure 2). The other TPs such as BaQM, TP300 and OH-9CA-ADIN could not be isolated in sufficient quantities. Hence, they are not included in the mass balances since an accurate quantification was impossible due to the lack of reference standards. For those TPs, the peak area is given in Figure S4 of the Supporting Information. DiOHCBZ. For DiOHCBZ in the degradation experiments with sand filter material and activated sludge, half-lives of 12.0 ± 0.4 and 37.1 ± 4 days, respectively, were obtained (both reactions first-order, Figure S1a of the Supporting Information). Until 21 days, a closed mass balance was observed with sand filter material (Figure 2a). Afterwards, the main TP BaQD was transformed. The amount of isolated BaQD standard (∼2 mg) was sufficient to conduct NMR experiments as well as to determine its concentration in environmental samples. However, the quantity was not sufficient to isolate its TPs from lab-scale experiments. In contact with activated sludge, BaQD was also formed, but to a much smaller extent (approximately 0.1% of the initial DiOHCBZ concentration). Instead, the formation of 9-CAADIN and TP207 was enhanced, and the mass balance was largely unclosed (Figure 2b). Thus, TP207 which could not be quantified due to the lack of a reference standard, might account for at least a part of the gap in the mass balance (see Figure S4b.1 of the Supporting Information). 10OHCBZ. With sand filter material, a lag phase of 14 days followed by zero-order degradation (half-life of 28.4 ± 3.0 days) was observed. In contrast, transformation with activated sludge followed first-order kinetics with a half-life of 5.9 ± 3.0 days (Figure S1b of the Supporting Information). In the experiments with sand filter material, BaQD, 9-CA-ADIN, and ADON contributed only to a small extent to the overall mass balance in the transformation of 10OHCBZ (spiked at 5 μg L−1), while at elevated concentrations (50 mg L−1) used for the isolation of TPs, these TPs were formed in much larger amounts (Figure S5 of the Supporting Information). As a consequence, an increasing gap in the mass balance with an increasing level of removal of 10OHCBZ was observed. Even further TPs for which no reference standard was available might not totally explain the gap

The cleavage of carbon dioxide indicates the presence of a carboxylic acid moiety, while the loss of a CHNO group showed that the amide group remained unaltered. The loss of CO and H2O is not representative of one single structural element but indicates the presence of an aldehyde, an alcohol group, or a carboxylic moiety. The 1H NMR experiments confirmed the presence of a carboxylic acid moiety due to a characteristic broad signal at a chemical shift of 13.02 ppm (Figure S16a,b of the Supporting Information). The 1H,15N HSQC NMR spectrum (Figure S16f of the Supporting Information) underlined the presence of one hydrogen (11.66 ppm) bound to a nitrogen atom, whereas the large downfield shift is characteristic for an amide group. Furthermore, eight aromatic hydrogens (6.32−8.14 ppm) were observed in the 1H NMR spectrum. Together with results from 1 H−1H COSY and 1H−13C HSQC experiments (Figure S16d,e of the Supporting Information), this confirmed the presence of two aromatic ring systems with four hydrogens each, arranged in A−B−C−D systems (each A−B−C−D system consists of two duplets and two triplets, whereas signal H-7 is overlaid with H10). The information obtained from both HRMSn and NMR experiments allowed for the identification of TP282 as 1-(2benzoic acid)-(1H,3H)-quinazoline-2,4-dione (BaQD (see Table 1)). Interestingly, the same TP has been identified by McDowell et al. for the transformation of CBZ during ozonation.16 An additional ozonation experiment with CBZ confirmed the formation of BaQD (for details, see the Supporting Information). The chemical formula, proposed fragmentation pattern, and chemical structures of all observed TPs as well as results from 1 H NMR are listed in Table 1. Further information, MS spectra, 1 H NMR as well as 13C NMR, and two-dimensional NMR data, if available, are provided in the Supporting Information. Elucidation of Biotransformation Pathways. The degradation of DiOHCBZ, 10OHCBZ, and OXC in contact with sand filter material and activated sludge led to the formation of the same TPs. However, differences in abundances of individual TPs were observed. Three main transformation pathways for DiOHCBZ, 10OHCBZ, and OXC were proposed (Figure 1) on the basis of the sequence of TPs formed as well as the results of additional transformation experiments spiked with isolated TPs (9-carboxylic acid acridine (9-CA-ADIN) and acridone (ADON)). Transformation Pathway I. The formation of TP268 (10,11ketol-carbamazepine) is caused by (i) the dehydrogenation of DiOHCBZ, (ii) the hydroxylation of 10OHCBZ with subsequent dehydrogenation, and (iii) the hydroxylation of OXC (a). The formation of TP268 is most likely succeeded by the cleavage of the C10−C11 bond, leading to intermediate INTP268 (b). A similar reaction has been observed for the cleavage of the acyloin linkage of benozoin to benzaldehyde by thiamine diphosphate (ThDP)-dependent benzaldehyde lyase.17 INTP268 can be oxidized,18 leading to the formation of INTP284 and 2,2′-(carbamoylazanediyl)dibenzoic acid (TP300) (c). Afterwards, a ring closure via intramolecular reaction of the carbamoyl nitrogen with the carboxylic moiety (TP300) or the aldehyde moiety (IN-TP284), including the loss of water, leads to the formation of BaQD or 1-(2-benzoic acid)-(1H,3H)quinazoline-2-one (BaQM), respectively (d). Finally, BaQD can also be formed by the oxidation of BaQM (e). Both TPs are also being formed via ozonation of CBZ.16,19 BaQD and BaQM are further transformed, but no additional TPs were observed. F



Article

Table 2. Concentrations (ng L−1) of CBZ, OXC, DiOHCBZ, 10OHCBZ, and Their TPs in WWTP Influent and Effluent, Surface Water (approximate percentage of wastewater in parentheses), and a Waterworksa WWTP influent WWTP effluent surface water Rhine River Lippe River small canal small stream (90%) small stream (>90%) waterworks feed water finished drinking water

nb

CBZ

DiOHCBZ

10OHCBZ

OXC

BaQD

9-CA-ADIN

ADON

2 2

1500 ± 23 1500 ± 20

4960 ± 286 3380 ± 20

1590 ± 40 1430 ± 40

43 ± 16 27 ± 14

162 ± 6 135 ± 15

700 ± 50 920 ± 50

Transformation of Oxcarbazepine and Human Metabolites of

Recommend Documents