High-Throughput Identification of Microbial Transformation Products of

Jul 27, 2010 - The objective of this work was to develop an efficient procedure to allow for high-throughput elucidation of TP structures for a broad ...
1 downloads 10 Views 555KB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 6621–6627

High-Throughput Identification of Microbial Transformation Products of Organic Micropollutants D A M I A N E . H E L B L I N G , * ,† JULIANE HOLLENDER,† HANS-PETER E. KOHLER,† HEINZ SINGER,† AND K A T H R I N F E N N E R †,‡ Eawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 Du ¨ bendorf, Switzerland, and Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich, 8092 Zurich, Switzerland

Received March 26, 2010. Revised manuscript received July 2, 2010. Accepted July 7, 2010.

During wastewater treatment, many organic micropollutants undergo microbially mediated reactions resulting in the formation of transformation products (TPs). Little is known on the reaction pathways that govern these transformations or on the occurrence of microbial TPs in surface waters. Large sets of biotransformation data for organic micropollutants would be useful for assessing the exposure potential of these TPs and for enabling the development of structure-based biotransformation prediction tools. The objective of this work was to develop an efficient procedure to allow for high-throughput elucidation of TP structures for a broad and diverse set of xenobiotics undergoing microbially mediated transformation reactions. Six pharmaceuticals and six pesticides were spiked individually into batch reactors seeded with activated sludge. Samples from the reactors were separated with HPLC and analyzed by linear ion trap-orbitrap mass spectrometry. Candidate TPs were preliminarily identified with an innovative post-acquisition dataprocessingmethodbasedontargetandnon-targetscreenings of the full-scan MS data. Structures were proposed following interpretation of MS spectra and MS/MS fragments. Previously unreported microbial TPs were identified for the pharmaceuticals bezafibrate, diazepam, levetiracetam, oseltamivir, and valsartan. A variety of previously reported and unreported TPs were identified for the pesticides. The results showed that the complementary use of the target and non-target screening methods allowed for a more comprehensive interpretation of the TPs generated than either would have provided individually.

Introduction The occurrence of pharmaceuticals, personal care products, and other xenobiotic compounds in surface waters has been well documented over the past decade (1, 2). Many of these compounds reach the aquatic environment after passing through wastewater treatment plants (WWTPs) (3), within which they may undergo microbial transformations resulting in the formation of transformation products (TPs). Whereas * Corresponding author: [email protected], phone: +41 44 823 50 71, fax: +41 44 823 50 28. † Eawag, Swiss Federal Institute of Aquatic Science and Technology. ‡ Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich. 10.1021/es100970m

 2010 American Chemical Society

Published on Web 07/27/2010

specific microbial TPs formed in WWTPs have not yet been identified for many compounds, work with nonionic detergents (4) and a few pharmaceuticals (5, 6) have shown that some TPs can be as toxic as or more toxic than their parent compounds. Therefore, identifying TPs formed in WWTPs and quantifying the extent of their formation are critical to assess a compound’s total impact on the aquatic environment. The challenge in identifying TPs in WWTP effluents is a lack of knowledge on how compounds will transform, making in situ studies aimed at quantifying and assessing the risk of TPs rarely possible. Consequently, identification of TPs is typically initiated by spiking compounds of interest into batch reactors seeded with activated sludge. Transformation products generated within the reactors are often identified by means of advanced chromatography and mass spectrometry (MS) (7). While this general procedure has resulted in the successful identification of TPs of several xenobiotics of concern (5, 8-14), it is limited by labor-intensive and timeconsuming experimental and analytical steps, which usually allow for only a few compounds to be investigated in a single study. Further, TPs identified by means of advanced MS techniques must be confirmed with either a complementary analytical technique or authentic standards. Nuclear magnetic resonance (NMR) is another valuable analytical tool that allows for unequivocal identification of TP structures without additional confirmation steps; but the need for high compound mass recoveries and purified samples limits its use in activated sludge experiments (13). Larger sets of biotransformation data for organic micropollutants would not only be useful for assessing the potential for exposure to these TPs, but also in the development of structure-based biodegradation prediction tools designed to predict TPs formed in specific environmental compartments. As such, methods are needed that can increase the efficiency and throughput of biotransformation assays. High-resolution mass spectrometers offer the potential to increase the throughput of assays aimed at microbial TP elucidation by combining high-resolution (HR) and accurate-mass, full-scan MS with rapid, data-dependent MS/ MS acquisition (15). Recently, high-resolution mass spectrometry has been used for the high-throughput identification of mammalian metabolites based on full-scan MS and the application of a variety of post-acquisition data processing techniques that are uniquely afforded by high-resolution mass spectrometers (16, 17). Two such post-acquisition data processing techniques are expected to have particular value in the identification of microbial TPs in batch systems: target and non-target screenings. A target screening is one in which single ion chromatograms are extracted from full-scan MS data at the exact masses of known or plausible TPs and inspected for peak formation over a time course of samples (17). The University of Minnesota Pathway Prediction System (UM-PPS) (18) has recently been used to propose structures of plausible microbial TPs for xenobiotics forming in the environment (19), and is also expected to be useful for proposing plausible TP structures formed within batch reactors seeded with activated sludge. A non-target screening is one in which full-scan MS data from two samples are compared to identify compound masses uniquely present in one of the samples. Non-target screenings have previously been used in the field of metabolomics for the identification of unknown pharmaceutical metabolites (17) and would appear to be particularly promising when studying the formation of TPs in batch systems; subtraction of MS spectra from two samples from a single reactor should eliminate the majority of redundant background matrix ions. To our VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6621

knowledge, no systematically applicable non-target screening method has been previously developed for and applied to the identification of microbial TPs formed in sludge-seeded bioreactors. The objective of this work was to develop an integrated experimental, analytical, and post-acquisition data processing procedure to enable high-throughput elucidation of transformation product structures for a broad and diverse set of xenobiotics undergoing microbially mediated transformation reactions. To meet this goal, we (i) developed a low-volume biotransformation test system seeded with activated sludge into which a diverse set of xenobiotics were individually spiked; (ii) employed an LC-HR-MS/MS method developed for the general screening of xenobiotics in aqueous samples; and (iii) developed, applied, and compared the efficacy of target and non-target screening methods as postacquisition data processing techniques for identifying candidate TPs. The procedure was tested with six pharmaceuticals selected for their known occurrence in WWTPs; six pesticides for which TPs have been previously described were selected as positive controls. Additionally, most of the compounds contained an amide functional group for their inclusion in a companion study that applies the method presented herein to elucidate the preferred biotransformation pathways of compounds containing amide functional groups (20).

Experimental Procedures Compound Selection. Compounds selected included the pharmaceuticals atenolol, bezafibrate, diazepam, levetiracetam, oseltamivir, and valsartan; and the pesticides carbetamide, clomazone, DEET, napropamide, propachlor, and tebutam (see the Supporting Information (SI) for full structures and relevant compound data). Compounds were supplied by Dr. Ehrenstorfer GmbH (Ausburg, Germany), Roche (Basel, Switzerland), Sigma (Buchs, Switzerland), and Toronto Research Chemicals (North York, Canada). Biotransformation Test System. Sludge to seed the bioreactors was sampled from a pilot-scale wastewater treatment plant (ps-WWTP) receiving wastewater from the sanitary sewer system of Du ¨bendorf, Switzerland (21). Sludge was sampled in amber glass bottles, and total suspended solids were measured per Standard Method 2540B (22). Sampled sludge was diluted with nanofiltered ps-WWTP effluent to a final total suspended solids concentration of 3 g/L to reduce matrix interferences. Diluted sludge was added to 100-mL amber glass bottles to a volume of 49.95 mL. Each compound was spiked individually into each bioreactor to provide an initial compound concentration of 100 µg/L. Reactors were loosely capped with nongasketed plastic caps that allowed for free transfer of oxygen into the reactor and placed on a shaker table to keep the sludge aerated. Dissolved oxygen, temperature, and pH were continuously monitored throughout the experiments and each parameter remained relatively constant at 7.9 ( 0.4 mg/L, 20 ( 1 °C, and 6.2 ( 0.6, respectively. Samples (2 mL) were withdrawn from each reactor with a 10-mL borosilicate glass syringe (Macherey-Nagel, Du ¨ren, Germany) and immediately filtered through 25-mm glass fiber syringe filters with a pore size of 1.0 µm (Acrodisc, East Hills, New York) into 2-mL amber vials. Vials were sealed with an 11-mm crimp cap with PTFE septa and stored for between 1 h and 14 days at 4 °C in the dark until analysis (no degradation was observed during storage in quality control samples). The first samples were withdrawn immediately after spiking the compound of interest (t ) 0 sample). Subsequent samples were withdrawn from the reactors at time points based on either known or assumed reaction kinetics. Each reactor was sampled at least six times over 2-, 4-, or 17-day test periods. Additional details on the biotransformation test system and results of control 6622

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

FIGURE 1. Overall workflow of developed procedure. experiments designed to estimate the extent of abiotic transformations and losses due to sorption to sludge are provided in the SI. Analytical Method. We adopted the analytical method from one previously reported for ultratrace level screening of xenobiotics using high-pressure liquid chromatography coupled with a linear ion trap-orbitrap mass spectrometer (Orbitrap, Thermo, Waltham, MA) (19). Briefly, the mobile phase consisted of HPLC-grade water (Acros Organics, Geel, Belgium) and HPLC-grade methanol (Acros Organics, Geel, Belgium), each amended with 0.1% (volume) formic acid (98-100%, Acros Organics, Geel, Belgium). The mobile phase was pumped to an XBridge (Waters, Milford, MA) C-18 column (2.1 mm ×50 mm, particle size 3.5 mm) at 200 µL/ min. Samples were injected into the column at 20 µL with an initial mobile phase of 90:10 water/methanol and elution from the column was achieved with a final mobile phase of 5:95 water/methanol (full gradient provided in the SI). The Orbitrap was used with electrospray ionization in either positive or negative mode. We performed mass calibrations (compounds included caffeine, the tetrapeptide MRFA, and Ultramark 1621) and mass accuracy checks prior to every sample run; resolution was always greater than 60,000 and mass accuracy was always within (5 ppm. The Orbitrap acquired full-scan MS data within a mass-to-charge (m/z, hereafter referred to as “mass”) range of 115-1000 for each sample. We used XCalibur v2.0.7 (Thermo, Waltham, MA) software for chromatogram analysis and interpretation. Parent compounds were identified with authentic standards and quantification was facilitated with a matrix-matched external calibration. Parent compound concentrations were typically below the limit of quantification in blank matrix samples (see details in the SI). Data-dependent MS/MS acquisition was triggered at masses of plausible TPs predicted by the UM-PPS as described below. Transformation Product Identification. Candidate TPs formed within the biotransformation test system were preliminarily identified with two independent post-acquisition data processing techniques: a target screening of single ion chromatograms extracted at the exact masses of plausible TPs proposed by the UM-PPS; and a non-target screening of high-resolution, full-scan MS data for compound masses that formed during the biodegradation experiment based on

background subtraction and mass filtering. The workflow for the entire method is summarized in Figure 1 and described below. Target Screening. The UM-PPS was used to propose two generations of plausible TP structures for each of the 12 parent compounds. We used the ChemBioOffice (CambridgeSoft, Cambridge, MA) software to convert SMILES strings (output from the UM-PPS) of each plausible TP structure to molecular formulas and exact masses. Following high-resolution, fullscan MS acquisition, single ion chromatograms were extracted at the exact masses of each UM-PPS proposed TP. Chromatographic peaks forming over the time course of samples were considered to belong to masses of candidate TPs and were selected for further analytical inspection (as described below). Additional details on the UM-PPS and the target screening approach are available in the SI. Non-target Screening. High-resolution, full-scan MS data from two samples from a single bioreactor were compared to identify compound masses that formed during the biotransformation experiment. For each sample from the biotransformation test system, a two-dimensional matrix of masses (m/z) found within the full-scan mass range of 115-1000 and retention time (RT) range of 0-25 min and their corresponding intensities was extracted from the fullscan MS acquisition and imported into a spreadsheet. We wrote a script in Visual Basic that compared the massintensity matrices between the t ) 0 and each t > 0 sample from a single bioreactor and used a series of mass filters to reduce the number of extracted masses to a list of masses of candidate TPs that formed during the biotransformation experiment. The filters included a mass and retention time domain constraint, a background subtraction algorithm, a constrained molecular formula fit, and a plausibility check based on the presence of 13C monoisotopic masses. Details of each of these filters are discussed in the SI. Analytical Confirmation of Candidate TPs. Candidate TP masses identified with either the target or non-target postacquisition data processing methods were further confirmed or rejected as actual TPs following manual inspection of MS spectra for the relative abundance of 13C and/or 37Cl monoisotopic masses and/or adduct masses. Additionally, data-dependent MS/MS acquisitions were triggered when peaks were detected in the full-scan at the exact masses of any parent compound or candidate TP. Comparison of MS/ MS fragments between each parent compound and TP further confirmed or rejected each candidate mass as an actual TP. Authentic standards were used to confirm the structure of three of the TPs identified.

Results and Discussion Results from the control experiments showed no significant losses attributable to abiotic transformation processes or sorption to the sludge (see SI); therefore the losses and transformations observed in the bioassays are attributed solely to microbial processes. For the 12 compounds investigated, we identified 26 microbial TPs: 13 each for the set of pharmaceuticals and pesticides. For the pharmaceuticals, 3 were previously reported as microbial TPs but 10 are believed to be reported here for the first time. Because microbial TPs of pesticides have been more thoroughly investigated as part of the pesticide registration process, only 3 of 13 are believed to be newly identified in this work; however the formation of 10 previously described microbial TPs of the pesticides shows that the procedure is capable of generating and allowing for identification of major environmentally relevant microbial TPs. The UM-PPS (target screening) successfully predicted the structures of 21 plausible TPs while the non-target screening resulted in the identification of all 26 TPs found in this work. The time

FIGURE 2. Analytical data for bezafibrate and the amine transformation product of amide hydrolysis including (a) chromatogram of the parent compound; (b) MS spectrum of the parent compound showing the presence and relative abundance of the 13 C and 37Cl isotope content; (c) MS/MS fragments of the parent compound; (d) chromatogram of the transformation product; (e) MS spectrum of the transformation product; and (f) MS/MS fragment of the transformation product. In (b) and (e), mass spectra and molecular formulas are shown for the [M + H]+ ion. courses of parent compound degradation and TP formation are shown in the SI. Interpretation of Analytical Data. As an example of how we identified TPs in this work, the analytical details used to propose the structure of one of the TPs of bezafibrate are presented in Figure 2. The extracted ion chromatogram (XIC), the MS spectrum, and the MS/MS fragmentation for bezafibrate in the t ) 0 sample are provided in Figure 2a-c. The XIC shows a peak at a retention time of 10.5 min and the MS spectrum contains peaks at the exact mass of bezafibrate along with those corresponding to its 13C and 37 Cl monoisotopic masses. Additionally, the relative abundances of the 13C and 37Cl monoisotopic masses match the theoretical abundances for a molecule containing nineteen carbon atoms and one chlorine atom. Four distinct MS/MS fragments were identified for bezafibrate at nominal masses of 121, 138, 276, and 316 and the proposed structures of each fragment are shown in Figure 2c. The same type of analytical data from the t ) 2 h sample is provided in Figure 2d-f. Here, the XIC shows the formation of a peak (no peak was detected in the t ) 0 sample) at the exact mass of BEZ 224 and at a retention time of 4.4 min. The MS spectrum shows peaks at the exact masses of BEZ 224 and the corresponding 13C monoisotopic mass with a relative abundance matching the theoretical abundance of a twelve carbon molecule. No mass was identified corresponding to the 37Cl monoisotopic mass, indicating that the compound contained no chlorine atom. A single MS/MS fragment was identified for the TP at a nominal mass of 121, which matched an observed fragment of bezafibrate from a conserved location of the molecule and was the only fragment detected in all samples in which a chromatographic peak formed at VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6623

FIGURE 3. Observed biotransformation products for each of the 6 pharmaceuticals. ATE 268, BEZ 155, and BEZ 137 were confirmed with authentic standards. See SI for extent of formation and analytical details supporting the proposed structures. the exact mass of BEZ 224. This combination of the exact mass chromatographic data, MS spectra, and the matching MS/MS fragment acquired with high-resolution mass spectrometry all support the proposed structure of BEZ 224. The analytical details for the other parent compounds and TPs are provided in the SI. The extent of the analytical evidence supporting each proposed TP structure was variable, with some having extensive MS/MS fragmentation patterns and others yielding no interpretable MS/MS data (BEZ 256 and DIA 301). For those TPs for which limited MS/MS data were acquired, the proposed structures must be considered more uncertain. Alternately, the structures of three TPs were confirmed with authentic standards (ATE 268, BEZ 155, BEZ 137). Unequivocal identification of all other TPs would require the use of authentic standards or a complementary analytical technique. Pharmaceuticals. The pharmaceuticals selected for this work have all been detected either in the influent or effluent of WWTPs or in surface waters (3, 23-25). Atenolol and bezafibrate have been shown to be eliminated at levels greater than 50% during wastewater treatment (11, 26) while little to no elimination has been reported for diazepam and oseltamivir (24, 27). Less is known about the formation of microbial TPs of these pharmaceuticals. The proposed TP structures for each pharmaceutical investigated are provided in Figure 3 and discussed below. Quintana et al. previously identified 4-chlorobenzoic acid as a microbial TP of bezafibrate formed in sludge-seeded bioreactors; this TP was hypothesized to be the product of enzyme-catalyzed amide hydrolysis (11). A total of five TPs of bezafibrate were identified by means of the method described in this work. Among these were 4-chlorobenzoic acid and the previously unreported amine product of amide hydrolysis identified as BEZ 224. Both products of the bezafibrate hydrolysis reaction were observed to degrade further: 4-chlorobenzoic acid to 4-hydroxybenzoic acid, which is a known microbial transformation reaction (28); and BEZ 224 to a dihydroxylated product (BEZ 256), whose hydroxy substituents could not be unequivocally positioned 6624

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

based on the MS/MS data (see SI). A fifth TP was identified with an exact mass corresponding to a dehydrogenation product. While the position of the double bond proposed in Figure 3a is at the only apparent location in the structure of bezafibrate at which such a reaction can take place, the MS/ MS data can confirm neither the location of the double bond nor its cis-trans isomeric configuration. Radjenovic et al. identified atenolol acid as a microbial TP of atenolol (12). To our knowledge, no studies have investigated the microbial transformation of levetiracetam in the environment. The results here indicate that both atenolol and levetiracetam degraded by primary amide hydrolysis to their corresponding carboxylic acids. The proposed structures of these TPs were additionally supported by the observed formation of sodium adducts; carboxylic acids are known to form adducts with sodium during electrospray ionization (29). Diazepam and oseltamivir degraded slowly (see Figure S3 of the SI); nevertheless, two TPs were identified for diazepam and one for oseltamivir. Diazepam degraded to either an N-demethylated product or a monohydroxylated product. The N-demethylated product (nordiazepam) is itself a prescribed sedative and is an active human metabolite of diazepam (30), though its formation through microbially mediated reactions has not been previously reported. Although we extracted a clear chromatographic peak at the exact mass of the proposed monohydroxylated product, we obtained poor MS data and no MS/MS data. Therefore, the proposed structure for this TP remains uncertain. Oseltamivir degraded by ester hydrolysis to form another known active human metabolite (oseltamivir carboxylate; 24). The formation of this product through microbially mediated reactions has also not been previously reported. Three TPs were identified for valsartan and are proposed to form in a sequence of transformation steps. As detailed in Figure 3f, the first reaction in the proposed biotransformation pathway was an N-dealkylation reaction, yielding VAL 336. This TP was observed to be subsequently transformed to VAL 252 by an amide hydrolysis reaction. This was

FIGURE 4. Observed biotransformation products for each of the 6 pesticides. See SI for extent of formation and analytical details supporting the proposed structures. also a transient product and was subsequently transformed to VAL 267, the product of hydrolysis and oxidation of VAL 252. The peak intensity at the mass of VAL 267 increased throughout the bioassay (see Figure S3 of the SI), suggesting it may be the most relevant TP of valsartan in terms of its environmental persistence. Pesticides. Propachlor is an acetanilide-derived herbicide for which many microbial TPs have been previously reported (31). As shown in Figure 4a, a total of five TPs were identified for propachlor in this work, all of which have been previously identified in soil. Among these were the products of dehalogenation (PRO 178), hydrolytic chlorine substitution (PRO 194), and sulfur addition (e.g., glutathione conjugation) and subsequent oxidation (PRO 258), which were all observed as initial biotransformation reactions. The other TPs formed following a series of apparent oxidation reactions from PRO 194 to the aldehyde (PRO 192) and carboxylic acid (PRO 208) products. DEET (N,N-diethyl-metatoluamide) is a domestically used insect repellant that has been detected in surface waters in the µg/L level (1). Previous microbial degradation studies conducted with pure cultures have reported the product of direct amide hydrolysis (32). In this work, two TPs were observed: the product of amide N-deethylation and the product of oxidation reactions at the tolyl group. The N-deethylation product rapidly degraded to an undetected product, whereas the oxidation product was more stable (see Figure S4 of the SI). Carbetamide and napropamide are herbicides whose microbial transformation processes have been investigated with soil bacteria as part of the European Union registration process. For each compound, biotransformation reactions at the amide group have been reported: amide hydrolysis of carbetamide and N-dealkylation and amide hydrolysis for napropamide (33). Additionally, microbial degradation of carbetamide in soils has been reported with aniline being a commonly identified TP (34). For both compounds, only products of amide N-dealkylation and hydrolysis were identified. Soil bacteria have been shown to degrade clomazone to a variety of mono- and dihydroxylated TPs (35) but, to our knowledge, no TPs of tebutam have been previously reported. Here, mono- and dihydroxylated TPs were identified for

clomazone. Interpretation of the MS/MS data did not allow exact elucidation of the position of the hydroxy groups and they are only exemplarily shown to be attached to the phenyl group in Figure 4e. For tebutam, a single TP was identified as shown in Figure 4f. For both clomazone and tebutam, the intensity of the chromatographic peaks of their proposed TPs increased throughout the bioassay, suggesting they may be persistent TPs. Method Assessment. While the UM-PPS successfully predicted the formation of 21 TPs, it is interesting to consider the five TPs that were not predicted and explore why the corresponding reactions are under-represented within the UM-PPS. These five TPs were the products of dehydrogenation of bezafibrate (BEZ 360); oxidation of the tolyl group of DEET (DEET 222); oxidation (to the carboxylic acid) of propachlor (PRO 208); dehalogenation of propachlor (PRO 178); and sulfur addition and oxidation of propachlor (PRO 258). Dehydrogenation and sulfur additions (particularly glutathione conjugation) reactions are well-known reactions for which no transformation rules existed within UM-PPS at the time of our study. Although only a single parent compound degraded by either dehydrogenation or sulfur addition reactions, consideration should be given to adding transformation rules for these reactions to the UM-PPS (the addition of glutathione conjugation reactions has since been added). In the case of PRO 178, the UM-PPS does contain a transformation rule resulting in the prediction of dehalogenation products, but this rule is only triggered when anaerobic conditions are specified, thus its omission from the candidate list. Finally, the oxidation products of DEET and propachlor are both predicted by the UM-PPS following three consecutive oxidation steps. Expanding the prediction space for the target screening method from two generations to three generations would thus result in the prediction of these two TPs, but would simultaneously introduce a larger number of additional TP masses into the candidate list. It is therefore considered to be more efficient to predict only two generations of TPs and to supplement the results of the target screening method with those from the non-target screening method, as demonstrated in this work. Whereas the non-target screening method allowed for a more comprehensive identification of TPs formed within the biotransformation test system as compared to the target VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6625

method, our analysis showed that advantages specific to each method make their use as complementary approaches more valuable than either would be individually. First, both methods result in a list of candidate TP masses that must be further analyzed through manual inspection of the XICs, MS spectra, and MS/MS spectra. Candidate TP lists generated in each method contain false positives defined either as UMPPS predicted TPs that are not observed (target method) or as masses that formed during the bioassay that cannot be specifically attributed to a product of the parent compound (non-target method). A large percentage of the latter are attributed to extracellular polymeric substances and other cellular breakdown products that form during the course of the biodegradation experiment (see SI for additional information on false positives). With the exception of propachlor, the non-target method generated a higher number of false positive candidate TP masses than the target method. Thus, the increased ability of the non-target method to identify TPs comes at the cost of increased numbers of false positives which can only be rejected following manual analysis of the analytical data. Further, while the non-target method is purely driven by extracting mass spectra from the full-scan MS data, the UM-PPS predictions in the target screening are driven by metabolic logic and plausible aerobic biodegradation reactions. This logic can be highly useful in assigning structures to candidate TP masses identified in the non-target method and thus facilitate MS/MS interpretation. Although 26 TPs were identified in this work, the extent of TP formation remains unknown without mass balances and quantification with authentic standards. Therefore, the possibility remains that additional TPs may have formed but went undetected. In the experiments, bioreactors were spiked with each parent compound at a concentration of 100 µg/L. This concentration was selected to be high enough that major TPs could be identified without enrichment, while remaining low enough to be environmentally relevant. As a result, some minor TPs that may have formed to low levels may not have been detected. However, such TPs that eluded detection may not be relevant for environmental risk assessment if also present in the environment at comparatively low levels. Also, some TPs might elude detection as several technical issues may constrain the detection space of the analytical method. First, there are potential compound-specific properties that could limit ionization efficiency during electrospray ionization. Additionally, biotransformation could result in compounds with masses below the low mass (m/z) range of the full-scan (115). Finally, nonspecific interfering ions with identical exact masses of TPs could mask the formation of a chromatographic peak and result in failed detection. While these potential analytical difficulties must be noted, the analytical method detected all of the parent compounds and 26 TPs that formed during the biodegradation experiment, so we assume it is unlikely that significant TPs were lost in the analytical phase. A further analytical limitation is the inability of highresolution MS/MS data to resolve structural isomers, as discussed in the preceding for the mono- and dihydroxylated TPs. Application of advanced MSn experiments and a more detailed interpretation of the resulting fragmentation pathways could be beneficial to this end, but was outside the scope of this work. As a result, the presented method should be considered as a tool for relatively rapid and certain TP structure identification, rather than as a replacement for previously reported analytical techniques for unequivocal TP identification (7). The primary advantages of the procedure described in this work are the efficiency of the experimental system, the ability to acquire high-resolution, accurate-mass, full-scan MS data, and the innovative post-acquisition data processing methods developed and applied herein. The UM-PPS es6626

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010

tablished a limited prediction space of plausible TPs based on metabolic logic while the non-target screening method was used to extract all compound masses that formed during the biodegradation experiment under defined constraints. These independent and complementary screening methods allowed for a more comprehensive interpretation of the TPs that were generated in this work than either would have provided individually. The result was a high-throughput method that allowed for generation and identification of microbial TPs for a broad and diverse set of xenobiotic compounds.

Acknowledgments We thank Dr. Adriano Joss and Mark Boehler for maintaining the ps-WWTP and supplying the sludge used in the biodegradation experiments. We additionally thank three anonymous reviewers whose comments strengthened the quality and clarity of this manuscript. This work was funded internally by Eawag Discretionary Funds for Research - Category: Action Field Projects.

Supporting Information Available A full list of compounds studied, including structures; the results of control and recovery experiments; the HPLC gradient; details on the target and non-target post-acquisition screening methods; accounting of candidate mass lists and false positives; time course plots of parent compound degradation and TP formation; and the analytical details for each transformation product including extracted ion chromatograms, MS spectra, and MS/MS spectra. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, hormones, and other organic wastewater contaminants in US streams, 1999-2000: A national reconnaissance. Environ. Sci. Technol. 2002, 36 (6), 1202–1211. (2) Ku ¨ mmerer, K. Drugs in the environment: Emission of drugs, diagnostic aids and disinfectants into wastewater by hospitals in relation to other sources - A review. Chemosphere 2001, 45 (6-7), 957–969. (3) Onesios, K. M.; Yu, J. T.; Bouwer, E. J. Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review. Biodegradation 2009, 20 (4), 441–466. (4) Giger, W.; Gabriel, F. L. P.; Jonkers, N.; Wettstein, F. E.; Kohler, H. P. E. Environmental fate of phenolic endocrine disruptors: Field and laboratory studies. Philos. Trans. R. Soc., A 2009, 367 (1904), 3941–3963. (5) Kosjek, T.; Heath, E.; Perez, S.; Petrovic, M.; Barcelo, D. Metabolism studies of diclofenac and clofibric acid in activated sludge bioreactors using liquid chromatography with quadrupole - time-of-flight mass spectrometry. J. Hydrol. 2009, 372 (1-4), 109–117. (6) Nalecz-Jawecki, G.; Wojcik, T.; Sawicki, J. Evaluation of in vitro biotransformation of propranolol with HPLC, MS/MS, and two bioassays. Environ. Toxicol. 2008, 23 (1), 52–58. (7) Perez, S.; Barcelo, D. Application of advanced MS techniques to analysis and identification of human and microbial metabolites of pharmaceuticals in the aquatic environment. Trac.Trend. Anal. Chem. 2007, 26 (6), 494–514. (8) Eichhorn, P.; Ferguson, P. L.; Perez, S.; Aga, D. S. Application of ion trap-MS with QqTOF-MS in the identification H/D exchange and of microbial degradates of trimethoprim in nitrifying activated sludge. Anal. Chem. 2005, 77 (13), 4176– 4184. (9) Haiss, A.; Kummerer, K. Biodegradability of the X-ray contrast compound diatrizoic acid, identification of aerobic degradation products and effects against sewage sludge micro-organisms. Chemosphere 2006, 62 (2), 294–302. (10) Langin, A.; Alexy, R.; Konig, A.; Kummerer, K. Deactivation and transformation products in biodegradability testing of betalactams amoxicillin and piperacillin. Chemosphere 2009, 75 (3), 347–354.

(11) Quintana, J. B.; Weiss, S.; Reemtsma, T. Pathways and metabolites of microbial degradation of selected acidic pharmaceutical and their occurrence in municipal wastewater treated by a membrane bioreactor. Water Res. 2005, 39 (12), 2654– 2664. (12) Radjenovic, J.; Perez, S.; Petrovic, M.; Barcelo, D. Identification and structural characterization of biodegradation products of atenolol and glibenclamide by liquid chromatography coupled to hybrid quadrupole time-of-flight and quadrupole ion trap mass spectrometry. J. Chromatogr., A 2008, 1210 (2), 142–153. (13) Schulz, M.; Lo¨ffler, D.; Wagner, M.; Ternes, T. A. Transformation of the X-ray contrast medium lopromide in soil and biological wastewater treatment. Environ. Sci. Technol. 2008, 42 (19), 7207– 7217. (14) Zwiener, C.; Seeger, S.; Glauner, T.; Frimmel, F. H. Metabolites from the biodegradation of pharmaceutical residues of ibuprofen in biofilm reactors and batch experiments. Anal. Bioanal. Chem. 2002, 372 (4), 569–575. (15) Krauss, M.; Hollender, J.; Singer, H. P. LC-high resolution MS in environmental analysis: from target screening to the identification of unknowns. Anal. Bioanal. Chem. 2010, 397 (3), 943-951. (16) Peterman, S. M.; Duczak Jr, N.; Kalgutkar, A. S.; Lame, M. E.; Soglia, J. R. Application of a linear ion trap/Orbitrap mass spectrometer in metabolite characterization studies: Examination of the human liver microsomal metabolism of the nontricyclic anti-depressant nefazodone using data-dependent accurate mass measurements. J. Am. Soc. Mass Spectrom. 2006, 17 (3), 363–375. (17) Ruan, Q.; Peterman, S.; Szewc, M. A.; Ma, L.; Cui, D.; Humphreys, W. G.; Zhu, M. S. An integrated method for metabolite detection and identification using a linear ion trap/Orbitrap mass spectrometer and multiple data processing techniques: application to indinavir metabolite detection. J. Mass Spectrom. 2008, 43 (2), 251–261. (18) Gao, J.; Ellis, L. B. M.; Wackett, L. P. The University of Minnesota Biocatalysis/Biodegradation Database: improving public access. Nucleic Acids Res. 2010, 38, D488–D491. (19) Kern, S.; Fenner, K.; Singer, H. P.; Schwarzenbach, R. P.; Hollender, J. Identification of transformation products of organic contaminants in natural waters by computer-aided prediction and high-resolution mass spectrometry. Environ. Sci. Technol. 2009, 43 (18), 7039–7046. (20) Helbling, D. E.; Hollender, J.; Kohler, H. P. E.; Fenner, K. A structure-based interpretation of biotransformation pathways of amide-containing compounds in sludge-seeded bioreactors. Environ. Sci. Technol. 2010 (doi: 10.1021/es101035b). (21) Joss, A.; Boehler, M.; Wedi, D.; Siegrist, H. Proposing a method for online permeability monitoring in membrane bioreactors. Water Sci. Technol. 2009, 60 (2), 497–506. (22) Clesceri, L. S., Greenberg, A. E., Eaton, A. D., Eds. Standard Methods for the Examination of Water and Wastewater, 20th

(23)

(24)

(25)

(26) (27) (28) (29)

(30)

(31) (32)

(33) (34) (35)

ed.; American Public Health Association/American Water Works Association/Water Environment Federation: Washington, DC, 1998. Hollender, J.; Zimmermann, S. G.; Koepke, S.; Krauss, M.; McArdell, C. S.; Ort, C.; Singer, H.; Von Gunten, U.; Siegrist, H. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale post-ozonation followed by sand filtration. Environ. Sci. Technol. 2009, 43 (20), 7862–7869. Fick, J.; Lindberg, R. H.; Tysklind, M.; Haemig, P. D.; Waldenstrom, J.; Wallensten, A.; Olsen, B. Antiviral oseltamivir is not removed or degraded in normal sewage water treatment: Implications for development of resistance by influenza A virus. PLoS ONE 2007, 2 (10), 10.1371/journal.pone.0000986. Kasprzyk-Hordern, B.; Dinsdale, R. M.; Guwy, A. J. The occurrence of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs in surface water in South Wales, UK. Water Res. 2008, 42 (13), 3498–3518. Maurer, M.; Escher, B. I.; Richle, P.; Schaffner, C.; Alder, A. C. Elimination of beta-blockers in sewage treatment plants. Water Res. 2007, 41 (7), 1614–1622. Lo¨ffler, D.; Ro¨mbke, J.; Meller, M.; Ternes, T. A. Environmental fate of pharmaceuticals in water/sediment systems. Environ. Sci. Technol. 2005, 39 (14), 5209–5218. Field, J. A.; Sierra-Alvarez, R. Microbial transformation of chlorinated benzoates. Rev. Environ. Sci. Bio/Technol. 2008, 7 (3), 191–210. Habicht, S. C.; Vinueza, N. R.; Archibold, E. F.; Duan, P.; Kentta¨maa, H. I. Identification of the carboxylic acid functionality by using electrospray ionization and ion-molecule reactions in a modified linear quadrupole ion trap mass spectrometer. Anal. Chem. 2008, 80 (9), 3416–3421. Gonza´lez Alonso, S.; Catala´, M.; Maroto, R. R.; Gil, J. L. R.; de ´ . G.; Valca´rcel, Y. Pollution by psychoactive pharmaMiguel, A ceuticals in the Rivers of Madrid metropolitan area (Spain). Environ. Int. 2009, 36 (2), 195–201. Stamper, D. M.; Tuovinen, O. H. Biodegradation of the acetanilide herbicides alachlor, metolachlor, and propachlor. Crit. Rev. Microbiol. 1998, 24 (1), 1–22. Rivera-Cancel, G.; Bocioaga, D.; Hay, A. G. Bacterial degradation of N, N-diethyl-m-toluarnide (DEET): Cloning and heterologous expression of DEET hydrolase. Appl. Environ. Microbiol. 2007, 73 (9), 3105–3108. Draft Assessment Report; European Commission EU Review Programme, 2005 (http://dar.efsa.europa.eu/dar-web/ provision). Hole, S. J.; McClure, N. C.; Powles, S. B. Rapid degradation of carbetamide upon repeated application to Australian soils. Soil Biol. Biochem. 2001, 33 (6), 739–745. Liu, S. Y.; Shocken, M.; Rosazza, J. P. N. Microbial Transformations of Clomazone. J. Agric. Food Chem. 1996, 44 (1), 313–319.

ES100970M

VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6627