Environ. Sci. Technol. 2009, 43, 7039–7046
Identification of Transformation Products of Organic Contaminants in Natural Waters by Computer-Aided Prediction and High-Resolution Mass Spectrometry S U S A N N E K E R N , †,‡ K A T H R I N F E N N E R , * ,†,‡ HEINZ P. SINGER,† ´ P. SCHWARZENBACH,‡ AND RENE JULIANE HOLLENDER† Eawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 Du ¨ bendorf, Switzerland, and Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich, 8092 Zu ¨ rich, Switzerland
Received December 19, 2008. Revised manuscript received July 20, 2009. Accepted July 22, 2009.
Transformation products (TPs) of organic contaminants in aquatic environments are still rarely considered in water quality and chemical risk assessment, although they have been found in concentrations that are of concern. Since many different TPs can potentially be formed in the environment and analytical standards are typically lacking for these compounds, knowledge on the prevalence of TPs in aquatic environments is fragmentary. In this study, an efficient procedure was therefore developed to comprehensively screen for large numbers of potential TPs in environmental samples. It is based on a target list of plausible TPs that has been assembled using the University of Minnesota Pathway Prediction System (UM-PPS) for the computer-aided prediction of products of microbial metabolism and an extensive search for TPs reported in the scientific literature. The analytical procedure for screening of the compounds on the target list has been developed to allow for the detection of a broad range of compounds in complex environmental samples in the absence of commercially available reference standards. It includes solid phase extraction with broad enrichment efficiency, followedbyliquidchromatographyandtandemmassspectrometry with high mass resolution and accuracy. The identification of target TPs consisted of extracting the exact mass from the chromatogram, selecting peaks of sufficient intensity, checking the plausibility of the retention time, and interpreting mass spectra. The procedure was used to screen for TPs of 52 pesticides, biocides, and pharmaceuticals in seven representative surface water samples from different regions in Switzerland. Altogether, 19 TPs were identified, including both some wellknown and commonly detected TPs, and some rarely reported ones (e.g., biotransformation products of the pharmaceuticals venlafaxine and verapamil, or of the pesticide azoxystrobin). Overall, the rather low number of TPs detected suggests that TPs may not pose a problem of unexpected magnitude for aquatic resources. * Corresponding author phone: +41 44 823 50 85; fax: +41 44 823 54 71; email:
[email protected]. † Eawag. ‡ ETH Zurich. 10.1021/es901979h CCC: $40.75
Published on Web 08/11/2009
2009 American Chemical Society
Introduction Transformation products (TPs) of organic contaminants, particularly of the structurally more complex compounds of current concern such as pesticides, biocides, and pharmaceuticals, have been shown to contribute to the overall chemical burden in the environment (1). For some pesticide classes, e.g., triazines and chloroacetanilides, TPs can be present in higher concentrations and occur more frequently than their parent compounds (2). Accordingly, identification of relevant TPs formed upon application in soils and plants is required in environmental risk assessment of pesticides (e.g., ref 3). Also, current European regulations on drinking water and groundwater quality have responded to these findings by including specific TPs into the group of compounds to be monitored. For the majority of chemicals, however, including industrial chemicals and pharmaceuticals, chemical legislation requiring the identification of TPs has emerged only recently (e.g.. ref 4). Furthermore, TPs for these chemicals typically have to be identified for high tonnage compounds only and actual guidance on how to include them into risk assessment is weak. Additional limitations to our understanding of how much TPs contribute to the overall presence of chemicals in the environment are the lack of analytical reference standards for most potential TPs and the fact that the results of laboratory degradation studies may not be representative of actual environmental conditions. Therefore, the goal of this study was to develop a method for obtaining a more comprehensive picture of the presence of TPs in the environment. To reach this goal, we developed a systematic and efficient procedure to screen for large numbers of potential TPs in environmental water samples. The procedure applies highresolution mass spectrometry (HR-MS), a rather recent analytical technique that allows overcoming the lack of analytical reference standards, to a target list that contains as comprehensive a compilation of potential TPs as possible. The target list was assembled to include both TPs and metabolites already reported in registration dossiers or the scientific literature, as well as proposed products of aerobic, microbial metabolism. These microbial TPs might be formed in soil or activated sludge, which are two of the major sources of water-relevant contaminants. For their prediction, the University of Minnesota Pathway Prediction System (UMPPS) (5), a rule-based system to predict products of microbial metabolism, was used. Regarding HR-MS techniques for screening, quadrupole time-of-flight tandem mass spectrometry or ion trap orbitrap tandem mass spectrometry (LTQ orbitrap MS/MS) are two techniques that have been shown to enable fast, sensitive, and reliable detection and identification of low molecular weight substances, even in the absence of reference standards (e.g., refs 6-8). They allow recording full-scan chromatograms with high mass accuracy and resolution that make it possible to selectively search for a given TP based on its exact mass (9). To unequivocally identify the molecular structure of a compound, further analysis by nuclear magnetic resonance would be most preferable, but often is not yet sensitive enough for environmental samples. The interpretation of fragmentation patterns of HR-MS/MS spectra is a second option that can lead to successful structure elucidation even for the low concentrations present in environmental samples (10, 11), although it may fall short of discriminating between isomeric structures. Several studies report on the use of HRMS to search for TPs of organic chemicals in biodegradation VOL. 43, NO. 18, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7039
experiments or photolysis studies carried out in the laboratory (12-15). In these studies, concentrations of TPs are typically rather high, and the samples only contain the parent compound and a few TPs. Here, in contrast, the goal was to detect and identify potential TPs in actual environmental water samples, which contain a broad variety of organic compounds at concentrations in the ng/L range. The HRMS analytical procedure was therefore optimized by combining LTQ orbitrap LC-MS/MS analysis with solid phase extraction exhibiting broad enrichment efficiency. The overall screening procedure was applied exemplarily to examine the presence of potential TPs of 52 highly used and structurally diverse pesticides, biocides, and pharmaceuticals in representative surface water samples from various locations in Switzerland.
Materials and Methods All details about sampling, materials and reference standards, as well as LC-MS/MS analysis can be found in Supporting Information (SI) pp. S2-S4, Table S1, and Figure S1. Sampling Scheme. Surface water grab samples from seven different catchments in four agricultural areas in Switzerland were taken between May and September 2007 during high discharge following rain events to detect pesticide pollution during preferential flow events. The first area was near Lake Greifensee including the sampling sites Aabach Moenchaltorf, Gossauer Bach, and Chindsmuehlebach (winter wheat, maize, barley). The second was near the city of Zurich with the sampling site Furtbach (maize, sugar beets), the third was near Lake Murtensee including the sites Hauptkanal (various vegetables) and La Petite Glane (maize, potatoes, cereals, sugar beets), and the fourth was near Lake Geneva with samples from Le Boiron (viticulture). All of the rivers sampled, with the exception of Chindsmuehlebach, also received effluents from sewage treatment plants (discharge percentage from about 4% for the catchment of La Petite Glane up to 65% in the case of Furtbach). Glass bottles of 4 L and 2.5 L were filled with 2 L and 1.5 L grab samples, respectively. All samples were stored at -20 °C before preparation. Sample Preparation and Extraction. The applied analytical method was developed by Singer et al. (16). Briefly summarized, sample preparation, extraction and analysis by LC-MS/MS were as follows. Thawed surface water samples and a blank nanopure water sample (all 500 mL) were adjusted to pH 6.7 before filtration (cellulose filter, 0.45 µm, Satorius). To achieve sufficient enrichment for a broad range of compounds, four different SPE cartridge materials were simultaneously used. Manually filled SPE cartridges contained 100 mg Strata-X-AW (Phenomenex, U.S.), 100 mg Strata-X-CW (Phenomenex, U.S.), 150 mg Isolute ENV+ (Separtis GmbH, Germany) and 200 mg Oasis HLB (Waters AG, U.S.). All samples were passed through the preconditioned SPE-cartridges at a flow rate of 10 mL/min. After drying of the solid phase material, the analytes were sequentially eluted with 4 mL of methanol/ethylacetate (v:v 50:50) containing 2% ammonia followed by 2 mL of methanol/ ethylacetate (v:v 50:50) containing 1.7% formic acid, and then recombined to yield a neutral extract. The extracts were evaporated under a gentle nitrogen stream to a final volume of 100 µL and finally reconstituted with water to a volume of 1000 µL with a pH of about 7. LC-MS/MS Analysis. All LC-MS/MS measurements were performed using an LTQ (Linear Trap Quadrupole) orbitrap mass spectrometer from Thermo Fisher Scientific Corporation with electrospray ionization (ESI). For liquid chromatography separation an XBridge C-18 column from Waters (2.1 × 50 mm with particle size of 3.5 µm) was used. The flow rate was 200 µL/min. A two-step linear gradient was run 7040
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 18, 2009
TABLE 1. Number of TPs for 52 Parent Compounds (Pesticides, Biocides, Pharmaceuticals) That Are Predicted (UM-PPS List) or Described in Literature (Literature List)a parent compound class pesticides (n ) 24) biocides (n ) 7) pharmaceuticals (n ) 21) total (n ) 52)
UM-PPS literature target list list overlap list 615 148 749 1512
194 54 141 389
63 16 28 107
746 186 862 1794
a Combination of the two lists, with some TPs appearing on both lists (overlap), yields the target list.
from 90% H2O and 10% methanol, both containing 0.1% formic acid, to 5% H2O and 95% methanol in 20 min. This resulted in an elution pH of about 3. HR-MS spectra with a resolution of 60 000 and a mass accuracy of 80% in spiked surface water samples. With the instrumental parameters used in this study, method detection limits for the set of reference standards between 0.1 and 10 ng/L were found for more than 75% of the compounds in those same samples. About 90% of our target TPs overlap with the standard set with respect to polarity, size, and speciation (see SI p. S9 and Figure S2). However, the weight
FIGURE 1. Six-step funneling procedure to identify target TPs using high mass resolution, plausibility of retention time, and interpretation of MS/MS fragments. of their distribution is shifted toward more polar and ionic compounds at pH 7. It was assumed that, due to their higher polarity, the percentage of proposed TPs with problems during workup due to sorption, hydrolysis, or volatilization would be smaller than for the standard set. This might be offset by the increased presence of phenolic compounds in the TP set, which show a poor ionization efficiency (23). Based on these considerations and given the fact that no significant deterioration in enrichment efficiency for ionic compounds was observed in the standard set, it is estimated that the majority (about 60%) of the TPs on the target list have detection limits between 1 and 100 ng/L. Identification Procedure. We developed an identification procedure to efficiently examine a large number of proposed TPs for positive identification without reference standards. The procedure is based on a series of steps as shown in Figure 1, with each step reducing the number of potential false positive findings. In the following, the single identification steps are described in more detail. (i) Exact Mass. The mass error of our measurements was
