Environ. Sci. Technol. 2010, 44, 6628–6635
Structure-Based Interpretation of Biotransformation Pathways of Amide-Containing Compounds in Sludge-Seeded Bioreactors DAMIAN E. HELBLING,† JULIANE HOLLENDER,† HANS-PETER E. KOHLER,† 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 April 1, 2010. Revised manuscript received July 16, 2010. Accepted July 19, 2010.
Partial microbial degradation of xenobiotic compounds in wastewater treatment plants (WWTPs) results in the formation of transformation products, which have been shown to be released and detectable in surface waters. Rule-based systems to predict the structures of microbial transformation products often fail to discriminate between alternate transformation pathways because structural influences on enzyme-catalyzed reactions in complex environmental systems are not well understood. The amide functional group is one such common substructure of xenobiotic compounds that may be transformed through alternate transformation pathways. The objective of this work was to generate a self-consistent set of biotransformation data for amide-containing compounds and to develop a metabolic logic that describes the preferred biotransformation pathways of these compounds as a function of structural and electronic descriptors. We generated transformation products of 30 amide-containing compounds in sludge-seeded bioreactors and identified them by means of HPLC-linear ion traporbitrap mass spectrometry. Observed biotransformation reactions included amide hydrolysis and N-dealkylation, hydroxylation, oxidation, ester hydrolysis, dehalogenation, nitro reduction, and glutathione conjugation. Structure-based interpretationoftheresultsallowedforidentificationofpreferences in biotransformation pathways of amides: primary amides hydrolyzed rapidly; secondary amides hydrolyzed at rates influenced by steric effects; tertiary amides were N-dealkylated unless specific structural moieties were present that supported other more readily enzyme-catalyzed reactions. The results allowed for the derivation of a metabolic logic that could be used to refine rule-based biotransformation pathway prediction systems to more specifically predict biotransformations of amidecontaining compounds.
Introduction The potential environmental relevance of transformation products (TPs) of xenobiotic compounds forming from abiotic * Corresponding author phone: +41 44 823 50 85; fax: +41 44 823 53 11; e-mail:
[email protected]. † Eawag, Swiss Federal Institute of Aquatic Science and Technology. ‡ Institute of Biogeochemistry and Pollutant Dynamics (IBP). 6628
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010
or biotic degradation processes has long been an issue of concern (1). Recent studies have shown that some TPs of xenobiotics can be as toxic as or more toxic than their parent compounds (2–4). However, pesticides are the only use-class of chemicals for which the formation, fate, and effects of their TPs have been systematically investigated (5). Additionally, except for pesticides, the concern over TPs has not led to concrete guidance for their inclusion in chemical risk assessments (6). In silico systems that predict plausible intermediate structures of biological and chemical transformations of xenobiotics could play an important role in obtaining a more comprehensive picture of exposure to TPs. Predicted product structures can be used as target lists for screening of environmental samples (7) or for model-based prioritization of plausible TPs. This is an important prerequisite for targeted investigation of TPs that bare the potential to significantly add to the risk posed by their parent compounds. Microbial transformation occurring within wastewater treatment plants (WWTPs) is often the dominant transformation process for compounds such as pharmaceuticals, personal care products, and biocides that reach the aquatic environment through urban sewer systems. Several tools are currently available to predict microbial transformation pathways of xenobiotics and the structures of likely TPs. These tools include META (8), CATABOL (9), and the University of Minnesota Pathway Prediction System (UM-PPS) (10). Each of these tools contains sets of transformation rules that predict likely microbial transformations based on compound substructure recognition. Transformation rules in the latter two systems are based on data contained in the University on Minnesota Biocatalysis/Biodegradation Database (UM-BBD), which is a manually maintained collection of literature-reported, microbially mediated metabolic pathways and enzymecatalyzed reactions (11). However, most data aggregated within the UM-BBD have been generated in pure culture systems within which the compound of concern serves as the sole carbon source. As a result, transformations observed may not be representative of the transformation pathways occurring in the environment. In environmental systems such as WWTPs, mixed bacterial consortia transform chemicals present in trace concentrations relative to high carbon fluxes of easily degradable material. Therefore the product spectrum is expected to be determined by the microbial diversity of the system and the available pool of enzymes (12). Additionally, enzymes from fungus species relevant to WWTPs have been shown to catalyze unique biotransformation reactions (13). It therefore remains unknown how well predictive systems such as UM-PPS or CATABOL are suited to predict microbial transformations of xenobiotics under environmentally relevant conditions. While knowledge on the general removal of micropollutants in WWTPs is growing (see ref 14 for a review), relatively few studies have looked at the formation of TPs in WWTPs (15–17). Studies aimed at TP structure elucidation are expensive and time-consuming and therefore cannot be expected to completely fill the knowledge gap on TP formation in the environment. As such, predictive tools such as UM-PPS could be a highly beneficial and cost-effective approach to provide a priori knowledge on how compounds might transform within WWTPs. However, before applying such tools to predict structures of specific TPs formed within WWTPs, a self-consistent set of data derived from experiments that more closely approximate 10.1021/es101035b
2010 American Chemical Society
Published on Web 08/09/2010
the conditions of activated sludge processes would be needed to test and refine the transformation rules. Many pharmaceuticals and other xenobiotics whose fate is through WWTPs contain amide groups (see ref 14 for a review of xenobiotics that have been identified in WWTPs). Data in the literature further suggest that amides may either be hydrolyzed (18) or N-dealkylated (19) during biotransformation, but the structural features that may promote or inhibit either biotransformation reaction remain unknown. Accordingly, the UM-PPS contains transformation rules for hydrolysis and N-dealkylation which are both triggered by secondary and tertiary amide substructures. Therefore, amides are an ideal compound class to show how an improved understanding of the preferences for different transformation pathways, and how these preferences depend on specific steric and/or electronic factors within the substrate structure, carries the potential to considerably improve the accuracy of biotransformation predictions. The primary objectives of this study were therefore as follows: (i) to generate a self-consistent set of biotransformation data for amide-containing compounds under conditions that are more representative of WWTPs than those described in the UM-BBD, and (ii) to analyze these data to develop a metabolic logic that more closely reflects the preferred biotransformation pathways observed. Nineteen general amides were selected to systematically test how specific structural properties governed their preferred biotransformation pathways. This set of compounds was supplemented with data on the biotransformation pathways of 11 amide-containing pharmaceuticals and pesticides previously investigated under the same experimental conditions (20). A structure-based interpretation of the transformation pathways of the resulting set of 30 amide-containing compounds was used to identify preferred biotransformation pathways.
Experimental Procedures Compound Selection. The 19 general amides selected for this work were supplied from Ambinter SARL (Paris, France) and AKos GmbH (Steinen, Germany). Compound IDs, names, and structures are provided in Table 1 (see the Supporting Information (SI) for relevant compound data). Also included are the names and structures of the 11 pharmaceuticals and pesticides with previously reported transformations (20). Experimental Methods. All experimental, analytical, and post-acquisition data processing methods are described elsewhere (20). Briefly, sludge from a pilot-scale membrane bioreactor (ps-MBR) was added to 50 mL amber glass bioreactors and diluted to a final solids concentration of 3 g/L with nanofiltered effluent from the ps-MBR. Each compound was individually spiked into the bioreactors to an initial concentration of 100 µg/L. Bioreactors were loosely capped, and dissolved oxygen concentrations, temperature, and pH were monitored throughout the experiment. Sampling schedules were developed based on known or assumed reaction kinetics. A minimum of 5 samples were taken from each reactor over a period of between 2 and 17 days. Samples were filtered through 1.0 µm glass fiber syringe filters (Arcodisc, East Hills, New York) and added to 2 mL amber vials and sealed. Controls established to account for losses attributable to sorption to sludge, other abiotic processes, and sample filtration are described in the SI. Separation and detection was achieved with high-pressure liquid chromatography (HPLC) coupled to a linear ion trap-orbitrap mass spectrometer (Orbitrap, Thermo, Waltham, MA) using a previously described method (7). Data-dependent MS/MS acquisition was achieved using a mass list containing the exact masses of the parent compounds and plausible TPs. Two complementary post-acquisition data processing methods were
employed to identify candidate TP masses within the fullscan MS data; MS and MS/MS spectra data were used to identify TP structures (20). Kinetic Analysis. Biotransformation rate constants were estimated from the experimental data for each compound to allow for relative comparisons of degradation rates among the 30 compounds studied and to support interpretation of the observed transformation reactions. The control experiments described in the SI were used to independently estimate rates of compound losses attributable to sorption to sludge and other abiotic processes (e.g., abiotic hydrolysis, volatilization) as has been described previously for batch biotransformation experiments (21). Biotransformation rate constants were then estimated using eq 1, which is derived in detail in the SI. (-kbioXss - ka) dCaq ) Caq dt (1 + KdXss)
(1)
In eq 1, Caq is the aqueous phase concentration of the parent compound in the biotransformation reactor, [µg/L]; t is time, [days]; kbio is the biotransformation rate constant of the dissolved fraction, [L/gss-day]; Xss is the suspended solids concentration, [gss/L]; ka is the abiotic transformation rate constant (estimated values provided in Table S2), [1/day]; and Kd is the biomass-water partitioning coefficient (estimated values provided in Table S2), [L/gss]. The analytical solution of eq 1 was used to estimate the biotransformation rate constant (kbio) by nonlinear regression (Solver, Excel). Correlation coefficients were calculated between the measured and fitted values of Caq to assess the quality of the fit of eq 1. To account for the increased uncertainty in estimated rate constants for compounds that degraded very rapidly or slowly, biotransformation rate constants are given as greater than 3 L/gss-day for compounds that were degraded by more than 90% in fewer than 6 h and as less than 0.002 L/gss-day for compounds that were degraded by less than 10% over the maximal test period of 17 days.
Results and Discussion A total of 53 TPs were identified over a maximum of three consecutive transformation steps and at least one TP was found for all except two of the 30 amide-containing compounds. The enzyme-catalyzed reactions observed during biotransformation included amide hydrolysis, amide Ndealkylation, mono- and dihydroxylation, alcohol and aldehyde oxidation, ester hydrolysis, reductive and hydrolytic dehalogenation, nitro reduction, and glutathione conjugation and subsequent oxidation. The transformation pathways and analytical details describing the identification of each TP are provided in the SI. All compounds degraded according to apparent pseudofirst-order kinetics without any observable lag-phase. Biotransformation rate constants estimated with eq 1 varied significantly among the 30 compounds as shown in Table 2 along with an overview of observed transformation reactions. The correlation coefficients (r2) calculated for fitting eq 1 were always greater than 0.88 and averaged 0.97. While the majority of the compounds degraded through pathways initiated by a single transformation, diazepam, DEET, propachlor, EEclB, and EEclB2 degraded through multiple pathways initiated by different transformation steps. Consequently, the biotransformation rate constants estimated for these five compounds cannot be clearly attributed to any single reaction type. The range of estimated biotransformation rate constants for the primary, secondary, and tertiary amides and the major types of transformations observed are compared in Figure VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6629
TABLE 1. Compound ID, Compound Names, and Structures
6630
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 17, 2010
FIGURE 1. Box and whisker plot comparing the estimated rates of biotransformation for (a) all compounds, (b) the primary amides, (c) the secondary amides, (d) the tertiary amides, (e) those compounds observed to undergo amide hydrolysis as the sole initial transformation step, (f) those compounds observed to undergo amide N-dealkylation as the sole initial transformation step, and (g) those compounds observed to undergo other reactions. The dashed line is the mean value, the solid line in the box is the median, the box edges are the 25th and 75th percentiles, and the whiskers are the fifth and 95th percentiles. A minimum of nine data points were required to plot the whiskers. 1. For the structural classes, the primary and secondary amides transformed faster than the tertiary amides, which is a trend further evidenced by comparing the biotransformation rate constants for specific pairs of tertiary amides (EmP2 and aPMclB) and their analogous N-dealkylated secondary amides (mPH2 and aPHclB). For the reaction classes, hydrolysis reactions were generally faster than N-dealkylation reactions, while the full compliment of other reactions exhibited a wide range of rate constants comparable to that of the full set of compounds. Additionally, hydrolysis and N-dealkylation reactions exhibited biotransformation rate constants ranging over more than an order of magnitude. Because it is assumed that each bioreactor has the same metabolic capabilities, the observed differences in hydrolysis or N-dealkylation reaction rates are attributed to structural features of the substrate. Therefore, the following discussion employs a structure-based analysis to attempt a more detailed interpretation of the observed transformation reactions and rates. Primary and Secondary Amides. Two of the 30 compounds are primary amides (i.e., both N-substituents are hydrogen atoms), namely the pharmaceuticals atenolol and levetiracetam. Levetiracetam additionally contains a tertiary amide (i.e., both N-substituents are alkyl/aryl groups) embedded within a ring structure. As previously described for both primary amides (20), the only TPs identified were the products of primary amide hydrolysis and were identified as atenolol acid and levetiracetam acid. No transformation was observed at any non-amide fragment of either compound or at the tertiary amide group of levetiracetam. Seven of the 30 compounds were secondary amides (i.e., one N-substituent is a hydrogen atom and the other is an alkyl/aryl group). These seven compounds included the following: the lipid regulator bezafibrate; the herbicide carbetamide; the antiviral oseltamivir; and four general amides (iPHclB, PHclB, aPHclB, and mPH2) selected to
investigate the effects of various R1 and R2 substituents. The dominant transformation pathway observed for the secondary amides was hydrolysis to primary amine and carboxylic acid products. The lone exception was oseltamivir, which transformed by ester hydrolysis indicating that its amide hydrolysis rate was negligible. Amidases and proteolytic enzymes are ubiquitous and constitutive, are found in different varieties and quantities within many genera of bacteria, have a wide substrate specificity, and are capable of hydrolyzing primary and secondary amides into their corresponding acids (see refs 22 and 23 for reviews on the abundance, classification, and function of amide hydrolyzing enzymes). The mechanism of enzyme-catalyzed amide hydrolysis is believed to involve the following: nucleophilic enzymatic attack of thecarbonylcarbon;formationofatetrahedralenzyme-substrate complex; release of the amine to form an acyl-enzyme complex; and reaction with water to release the carboxylic acid product (as shown in Figure S2a of the SI). Structural features of the substrate have been shown to hinder the formation of the enzyme-substrate complex and affect the rates of enzyme-catalyzed hydrolysis of secondary amides (24). Additionally, the proposed mechanism is analogous to the well characterized mechanism for basecatalyzed abiotic amide hydrolysis, where it has been shown that the rate of hydrolysis is influenced by the ease of protonation of the amine leaving group and steric hindrance from N-substituents (25). In an effort to identify structural features that influenced the observed biotransformation rates in this work, we focused on several molecular descriptors that could potentially influence the rate of amide hydrolysis based on this proposed mechanism. These included the pKb and partial charge of the amide nitrogen (which could influence the ease of protonation of the amine leaving group) and the van der Waals volume (determined with MarvinSketch from Chemaxon (26)) of the compounds’ N-substituents. We additionally looked at other molecular descriptors that would be expected to influence various rate limiting steps of biotransformation processes in general, such as substrate uptake and binding with the active site of the enzyme (e.g., log Kow, molecular volume, polarizability (27)). None of the molecular descriptors correlated well with the observed rates of enzyme-catalyzed hydrolysis of primary and secondary amides (see the SI for details and data for all molecular descriptors investigated). Whereas no quantitative relationship with the molecular descriptors could be determined, specific molecular substructures were found to influence the observed rates. First, the primary amide-containing compounds hydrolyzed rapidly, indicating that the lack of a substituent in the R2 position allowed these compounds to biotransform readily. Next, the three anilides (R2 is a mono- or unsubstituted aniline; aPHclB, PHclB, and mPH2) reacted significantly faster than all other compounds, which may indicate the presence of enzymes such as aryl-acylamidases (EC 3.5.1.13 (28)) with strong specificity for mono- or unsubstituted anilide substructures. Finally, structural features surrounding the reaction center of the substrate seemed to affect the hydrolysis rates of the secondary amides, presumably by blocking the formation of the enzyme-substrate complex. For example, oseltamivir contains an o,o-disubstituted cyclohexyl group in the R2 position, and no amide hydrolysis was observed for this substrate. The same reasoning could also explain why no hydrolysis of secondary amide groups has been observed in the biotransformation of iodinated contrast media (29, 30) which contain an o,o-disubstituted aromatic ring in either the R1 or R2 positions. The importance of the structural influence of the R2 substituent on the rate of VOL. 44, NO. 17, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6631
TABLE 2. Numbers of TPs Identified and Reactions Observed for Each Compound reactions observed ID
kbio, [L/gss-day]
r2
number of TPs observed
amide hydrolysis
ATE LEV BEZ CAR OSE iPHclB PHclB aPHclB mPH2 DEET NAP PRO TEB VAL DIA EP5 BiP4 BEB PPB EmP2 EEclB EEclB2 EEaB EEaB2 EEnB EEp2 MMclB BEclB aPMclB hPMclB
1.5 1.1 1.8 0.03 0.007 0.09 >3 >3 >3 0.3 0.03 >3 0.006 0.7
