Aqueous Photochemical Reaction Kinetics and ... - ACS Publications

Fluoxetine (FLX) was shown to be photoreactive in sunlit surface waters. FLX degraded in deionized water when exposed to simulated sunlight with a hal...
0 downloads 0 Views 215KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 513-522

Aqueous Photochemical Reaction Kinetics and Transformations of Fluoxetine MONICA W. LAM, CORA J. YOUNG, AND SCOTT A. MABURY* Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada, M5S 3H6

FIGURE 1. Structure of fluoxetine, fluometuron, and flutalanil.

Fluoxetine (FLX) was shown to be photoreactive in sunlit surface waters. FLX degraded in deionized water when exposed to simulated sunlight with a half-life of 55.2 ( 3.6 h-1. Photodegradation products were identified using high performance liquid chromatography-UV (HPLC-UV) and liquid chromatography tandem mass spectrometry (LC-MS-MS) using electrospray (ES) ionization. Defluorination of the trifluoromethyl group in FLX and in fluometuron and flutalanil, two other compounds containing this functional group, is suggested to be a common direct photolysis pathway for trifluoromethylated compounds. Products resulting from O-dealkylation of FLX were also observed. The rate of degradation was faster in synthetic field water where •OH was the likely dominant oxidant in the system. The bimolecular rate constant for the reaction between FLX and •OH was measured as (8.4 ( 0.5) × 109 and (9.6 ( 0.8) × 109 M-1 s-1 using two different methods of competition kinetics. Indirect photodegradation reactions could lead to the production of hydroxylated and O-dealkylated compounds. Although direct photolysis could potentially limit the persistence of FLX in surface waters, its degradation by indirect photolysis would proceed faster. Thus, this latter process could be important in the elimination of FLX in surface waters.

Introduction There is currently much scientific and public interest in pharmaceuticals and personal care products (PPCPs) as environmental pollutants. These chemicals are produced on a large scale, similar to that of agrochemicals, yet their environmental fate has until recently received comparatively little attention (1). PPCPs are thought to enter the environment through sewage treatment plants (STPs) following human excretion or by way of landfill after direct disposal (2). Other potential routes of entry into the environment are via land application of wastewater sludge and discharge or runoff of untreated sewage from animal feed operations (2). Many PPCPs have been identified at ng/L to µg/L levels in STP effluents and surface waters around densely populated areas (3-5). However, pharmaceutical use is not seasonal, and chronic infusion could lead to their constant environmental presence (1). Because many pharmaceuticals and personal care products have been detected in aqueous environmental compartments, understanding their fate under natural water conditions is important. Selective serotonin reuptake inhibitors (SSRI) are a large family of drugs commonly used to combat depression and * Corresponding author phone: (416)978-1780; fax: (416) 9783596; e-mail: [email protected]. 10.1021/es0494757 CCC: $30.25 Published on Web 12/08/2004

 2005 American Chemical Society

anxiety. Fluoxetine (FLX), typically marketed under the trade name Prozac, is one of the most frequently prescribed pharmaceuticals of this class (Figure 1). It has been observed at µg/L levels in streams susceptible to contamination (e.g., downstream from large urban centers) in the United States by Kolpin et al. (5), but little is currently known about the environmental fate of this pharmaceutical. Fluoxetine has also been shown to negatively affect nontarget organisms (6-8). A change in seratonin concentration or metabolism can affect the reproductive output of some species. One study has shown that SSRIs induce spawning in some crustaceans and bivalves (7). Recent studies have demonstrated that pharmaceuticals are susceptible to photodegradation in environmental settings. Therefore, if they are not removed by other chemical means, these chemicals could be degraded by photochemical reactions in sunlit surface water (9, 10). Direct and indirect photolysis could play an important role in the degradation of FLX. Direct photolysis occurs when a compound absorbs light, becomes unstable, and subsequently decomposes. Indirect photolysis occurs through reactions with reactive intermediates generated by another light-absorbing molecule. These types of reactions could be of importance in the environment when the UV-vis absorption spectrum of a compound does not overlap with the actinic spectrum. The nonselective and reactive hydroxyl radical, •OH, has been shown to limit the persistence of many compounds that degrade relatively slowly by direct photolytic means (11, 12). This photooxidant is produced by the photolysis of nitrate and dissolved organic matter (DOM) (13, 14), which are natural constituents of field water, and is present at concentrations ranging from 10-14 to 10-18 M (15, 16). It was hypothesized that FLX would be susceptible to both direct and indirect photolysis, which likely occur through •OH-meditated pathways due to the high reactivity of this photooxidant. The direct photodegradation pathway of FLX will be compared to fluometuron and flutalanil (Figure 1), two other trifluoromethylated compounds, to determine if common elements exist in the phototransformation of these compounds. The objective of this study was to examine the kinetics of the degradation of FLX with respect to direct and indirect photolysis in aqueous systems, and to identify transformation mechanisms of fluorinated organic pollutants.

Materials and Methods Chemicals. All chemicals were reagent grade and were used as received. Fluoxetine hydrochloride (99%) was obtained from Interchem (Paramus, NJ); p-trifluoromethylcresol (99%), 2-(methylamino)ethyl benzyl alcohol (99%), p-hydroxybenzoic acid (99%), R,R,R-trifluoro-o-toluic acid (98%), phthalic acid (99%), N,N-dimethylaniline (99%), sodium bicarbonate, potassium nitrate, p-chlorobenzoic acid (99%), p-nitroaniVOL. 39, NO. 2, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

513

sole, and humic acid were obtained from Aldrich (Mississauga, ON). Potassium monophosphate, 30% hydrogen peroxide, and pyridine were purchased from ACP (Montreal, PQ). Formic acid, o-phosphoric acid, aniline, and di-sodium tetraborate were all obtained from BDH (Toronto, ON). Fluometuron (98%) and flutalanil (99%) were obtained from Chem Service (West Chester, PA), and levofloxacin (90.8%) was purchased from Zhejiang Wonderful Pharma and Chem Co. (Zhejiang, China). Labeled water (H218O, 95% pure) was purchased from Cambridge Isotope Labs (Andover, MA). Optima grade acetonitrile and methanol purchased from Fisher (Nepean, ON) was used for LC-MS-MS analyses; HPLC grade acetonitrile (>99%) from Caledon (Georgetown, ON) was used for HPLC-UV analyses. Stock solutions of FLX, fluometuron, flutalanil, levofloxacin, and their proposed degradation products were prepared in Millipore 18 MΩ water for photolysis experiments or in methanol for instrument calibration standards and were stored in the dark at 4 °C. Analytical Methods. UV-vis absorption spectra were recorded using a Hewlett-Packard 8452A diode array spectrophotometer to determine which wavelength of detection would be used in later chromatographic analyses. The HPLC-UV analyses were performed at room temperature using a Waters 600S chromatograph fitted with a reverse phase column (Alltima C-18 5 µm 250 × 4.6 mm; Alltech). A 40 µL injection of each sample was done in duplicate using a Waters 717 autosampler. Detection was performed with a Waters 996 photodiode array detector. The detection of FLX was performed at 230 nm, and all analyses were conducted using isocratic conditions. The mobile phase used for analyses consisted of 50% ACN:50% 10 mM potassium monophosphate adjusted to pH 3 with o-phosphoric acid. A flow rate of 1 mL/min was used for all analyses. Ion chromatography (IC) analyses were done at room temperature using a Dionex GP50 Gradient pump with a GD20 detector. An anion self-regenerating suppressor was used in autosuppressor external water mode. The column used was an IonPac AS14 fitted with an IonPac AG14 guard column. The sample loop was 25 µL, and fluoride was detected using an isocratic program with a flow rate of 1.2 mL/min. The mobile phase was a 96% water, 4% 7 mM borate buffer solution. The expected background conductivity was 3-5 µS. The LC-MS-MS system consisted of a Waters 616 pump and a 600 controller (Waters Ltd., Milford, MA) connected to a Micromass Quattro Micro LC tandem mass spectrometer with Z Spra ES source operating in positive MS-MS mode (Micromass UK Ltd., Cheshire, UK). Chromatographic separations were performed using an Allure C18 column (50 mm × 2.1 mm, 5 µm) (Chromatographic Specialties, Brockville, ON). The mobile phase eluents were acetonitrile (ACN) and ammonium formate buffer (10 mM adjusted to pH 3 with formic acid). The initial conditions for the gradient method of 15% ACN:85% buffer were held for 3 min, and a linear ramp to 85% ACN was made over 1 min and this final composition was held for 10 min. A flow rate of 200 µL/min was used. Sample injections were made using a Waters 717 autosampler. Tuning was performed for each analyte of interest by direct infusion of a 1 ng/µL standard, or samples of photolysate from duplicate photolysis experiments, at a flow rate of 20 µL/min using a Hamilton syringe (500 µL, Reno, NV). The MS-MS parameters used for the analysis of all compounds are as follows: capillary voltage, 3.00 kV; source temperature, 110 °C; desolvation temperature, 250 °C; gas flow, 250 L/h; resolution (LM1, HM1, LM2, HM2) 13.5; ion energy 1 and 2, 0.6; entrance 3; exit, 3; multiplier, 650 V; dwell time, 0.2 s. The optimal cone voltage and collision energy, corresponding to close to 100% fragmentation of the molecular ions, differed for each chemical and suspected photoproduct examined. The instrument was operated in 514

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 2, 2005

multiple reaction monitoring (MRM) mode, using for each analyte one precursor ion > product ion transition. Photodegradation Kinetics Experiments. Photochemical experiments were conducted in a Suntest CPS photosimulator (Atlas, Chicago, IL) equipped with a Xe lamp as the UV radiation source and special glass filters restricting the transmission of wavelengths below 290 nm. However, even with the presence of these filters, there was a slight bleed of wavelengths below 290 nm, thus introducing an inconsistent spectral match with actinic radiation. The lamp was set to maximum intensity (765 W/m2). Cold water flowed through the bottom of the photosimulator to maintain the internal temperature at approximately 27 °C. For all degradation kinetics studies, each experiment was conducted in triplicate and was carried out until degradation to at least one half-life was achieved. Five time points were sampled over the irradiation period. For all direct photolysis experiments, 10 µM solutions were prepared in deionized water. To simulate natural water conditions in indirect photolysis kinetics experiments for FLX, three different synthetic field water (SFW) solutions containing FLX and different concentrations of nitrate, DOM, and bicarbonate were exposed to radiation in the sunlight simulator (17). No organic solvent was used in the preparation of these solutions. The preparation of the humic material used for photolysis experiments has been described elsewhere (17). The concentrations of the components in the SFW solutions were as follows: 8 µM NO3-, 0.7 mg C/L DOM, 0.8 mM HCO3- in SFW A (pH 8); 81µM NO3-, 7 mg C/L DOM, 0.8 mM HCO3in SFW D (pH 8); and 81 µM NO3-, 7 mg C/L DOM, 5 mM HCO3- in SFW E (pH 10), where the letters A, D, and E correspond to the nomenclature used in PhotoFate (17). The test system PhotoFate is an experimental system used to evaluate the contributions of various reactive transient species to the kinetics of indirect photolysis reactions (17). The solutions were placed in large quartz vessels, with approximately 60 mL of solution per vessel for direct photolysis experiments, and in smaller 10 mL vessels for the indirect photolysis experiments. The samples were quantitatively analyzed by HPLC immediately for the amount of compound of interest remaining in the solution after irradiation based on external calibration. Pseudo-first-order rate constants, kobs, for the model compounds were obtained by linear regression of logarithmic concentration values determined as a function of time. The sample aliquots were also analyzed via IC for the production of fluoride. Control experiments included the direct photolysis of the test compounds in buffer solutions at pH 4, 7, and 9 to investigate the pH dependence of direct photolysis. Dark control experiments in DI water, in pH 4, 7, and 9 buffers, in the presence of H2O2, and the SFW solutions were also conducted concurrently with irradiation experiments. Losses from these experiments were considered negligible ( 268, [M - H2O]). Although the proposed photoproduct has a carboxylic acid functionality and thus the potential to form a [M - H]- ion, no molecular ion for this proposed photoproduct was observed in ES(-) mode. The fluoroquinolone, levofloxacin, like the proposed photoproduct contains an amine and carboxylic acid functional group, and thus has the potential to form a molecular ion by ES in both positive and negative modes. This compound was infused to the mass spectrometer to determine whether the inability to form a molecular ion in negative mode was unusual, despite the presence of an acidic functionality. However, only a [M + H]+ ion for levofloxacin was formed by ES ionization, indicating that the presence of a carboxylic acid group does not predicate the formation of a negative ion by ES. The buildup of the proposed photoproduct III resulting from a photonucleophilic displacement reaction over the course of the irradiation was monitored by LC-MS-MS using MRM. The peak area for this compound using the 286 > 268 transition increased over the duration of the experiment. Although the daughter ion obtained from the fragmentation of this photometabolite may not be sufficiently characteristic to definitively identify the modification in the -CF3 functional group, the loss of water from semiaromatic quinone-type carboxyl groups is a characteristic fragmentation commonly used to confirm the identity of fluoroquinolone antibiotics (25). In addition, the molecular mass of the ion observed does correspond to that of the proposed photometabolite. Further studies described below were conducted to support the production of the proposed photometabolite. A stoichiometric ratio of 3:1 fluoride:FLX should also be observed if defluorination of the -CF3 group to -COOH occurs. Using IC, the concentration of fluoride was observed to increase over the course of the experiment. For every mole of FLX that degrades into the carboxylic acid product, three moles of fluoride ions should be produced; thus, the production of fluoride would support the proposed reaction mechanism. However, in addition to this pathway, fluoride is released in the production of the semiquinone product (Product I). The further hydrolysis of this compound has previously been reported by Ellis et al. (23) to occur at a much slower rate leading to the production of p-hydroxybenzoic acid and two more fluoride ions. Therefore, this reaction is not a likely additional source of fluoride in this experiment as samples were analyzed immediately after each time point. Furthermore, with the use of an authentic standard to compare retention time and UV-vis spectra, the presence of phydroxybenzoic acid was never observed in the chromatograms of sample aliquots taken during the experiment.

FIGURE 3. Comparison of (a) UV-vis spectra and HPLC-UV retention time of a standard of 4-(difluoromethylene)-2,5-cyclohexadiene-1-one (product I) and a sample taken 24 h after irradiation, and (b) LC-MS-MS parent-daughter transitions and retention time for a standard of 2-(methylamino)ethyl benzyl alcohol (product II) and a direct photolysis sample taken 3 h after irradiation. Fluoride can be produced by defluorination of -CF3 on FLX, and by the formation of the semiquinone. If degradation by O-dealkylation is the only degradation pathway, the expected stoichiometry would be 1:1 fluoride:FLX. If this stoichiometric ratio is greater than 1, it suggests that there could be another pathway leading to fluoride production. The number of moles of fluoride produced over the course of the experiment was determined by external calibration. The ratio was determined to be greater than 1 (1.31 ( 0.05, n ) 2), suggesting that the formation of the semiquinone, which produces only one molar equivalent of fluoride, was not the only pathway occurring. This observation provides further support of the hypothesized hydrolysis of the trifluoromethyl group to a carboxylic acid. Direct photodegradation experiments for two other trifluoromethylated compounds were conducted to support the proposed defluorination mechanism that is hypothesized to occur in the direct phototransformation of FLX. Fluometuron is a herbicide applied for weed control in sugarcane and cotton plants (Figure 1). A recent study reported that this compound is susceptible to direct photolysis in sunlit surface waters, although photodegradation products were not reported (17). Following the method described above for direct photolysis experiments, the degradation of fluometuron was monitored by HPLC-UV, and photolysates taken during the experiment were infused to the MS-MS to determine whether a mass corresponding to the hypothesized defluorinated photoproduct could be observed. A molecular ion in the photolysates, not originally present prior to irradiation but corresponding to the carboxylic acid version

of fluometuron, was observed at 209 m/z in ES(+) mode. A parent-daughter transition of 209 > 120 for this proposed photoproduct corresponded to its molecular ion and the cleavage of +NH3-CO-N(CH3)2 [M - 89] (Figure 4). In ES(-) mode, a molecular ion with a m/z ratio of 207 was observed, and the loss of CO2 [M - 44] resulted in a 207 > 163 transition (Figure 4). These fragmentation patterns strongly support the proposed mechanism whereby the -CF3 group on the molecule is hydrolyzed to -COOH. The last trifluoromethylated compound studied, flutalanil, is a fungicide that belongs to the class of aryl carboxanilides and is commonly used in the control of rice sheath blight in paddy fields (26) (Figure 1). The photolysis of this compound has been studied in aqueous ethanolic solution (27), but the results of these experiments cannot be extrapolated to environmental settings. However, other aryl carboxanilide fungicides have been demonstrated to be susceptible to direct photochemical reactions in aqueous systems (28). Earlier unpublished work completed in our laboratory suggested that the direct photodegradation mechanism of flutalanil involved cleavage of the molecule into 3-isopropoxy-aniline and R,R,R-trifluoro-o-toluic acid. In addition to the observation of a peak for R,R,R-trifluoro-o-toluic acid (189 m/z), an ion with a mass that corresponded to phthalic acid (165 m/z) was observed in ES(-) mode. The mass spectra obtained from these ions in the photolysate samples were in agreement with those for purchased authentic standards. The observation of phthalic acid suggests the initial flutalanil cleavage product, R,R,R-trifluoro-o-toluic acid, further photodegrades by defluorination of the -CF3 group to form phthalic acid VOL. 39, NO. 2, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

517

FIGURE 4. (a) Proposed direct photodegradation reaction for fluometuron leading to production of the carboxylic acid analogue. (b) LC-MS-MS chromatograms in electrospray positive and negative modes using characteristic transitions for the carboxylic acid product.

FIGURE 5. Proposed direct photodegradation reaction for flutalanil to produce r,r,r-trifluoro-o-toluic acid which defluorinates to form phthalic acid. (Figure 5). Furthermore, LC-MS-MS analysis using characteristic parent > daughter ion transitions for these photoproducts demonstrated that the retention times for the proposed degradation compounds correspond to those for the authentic standards. There was no mass spectral evidence to suggest that flutalanil itself underwent a defluorination reaction to form its carboxylic analogue. However, the observation of an ion for phthalic acid suggests this reaction does happen for R,R,R-trifluoro-o-toluic acid. The N-dealkylation of FLX to produce norfluoxetine was not expected to occur in this study because the absorption of light below 185 nm for an n f σ* transition to occur and lead to bond breakage would be needed. An authentic standard of norfluoxetine was not available for this study, but two known parent-daughter transitions (296 > 134 and 296 > 30 m/z) (29) for this compound were included in the 518

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 2, 2005

FIGURE 6. Mass spectrum of proposed hydroxylated fluoxetine (IV) formed by indirect photolysis obtained in positive electrospray mode. LC-MS-MS method so that its potential production could be monitored. Indirect Photolysis Kinetics. FLX degraded considerably faster in SFW than in DI water, and R2 correlation values greater than 0.99 indicated the rates of degradation followed pseudo-first-order kinetics. While the half-life (t1/2) in DI water was 55.2 h, the t1/2 ranged from 5.5 ( 0.3 to 22 ( 0.9 h (N ) 3) in the SFW solutions. A number of radicals, such as •OH and 1O2, could be generated by the photolysis of the SFW components nitrate and DOM. In addition, DOM could be involved in photosensitization reactions, and carbonate radicals, CO3•-, could form from the reaction between •OH and HCO3-. FLX exhibited the shortest half-life in SFW E (5.5 ( 0.3 h, N ) 3), while the longest half-life was observed in SFW A (22 ( 0.9 h, N ) 3). Given that NO3-, DOM, and HCO3levels were lowest in SFW A, it was expected that FLX would exhibit the slowest degradation rate in this matrix. In SFW D, the concentrations of nitrate and DOM were higher and

FIGURE 7. (a) Overlay of LC-MS-MS chromatograms obtained in positive electrospray mode for 2-(methylamino)ethyl benzyl alcohol (product II) and the hydroxylated product (IV) in samples taken from an indirect photolysis experiment conducted in the presence of O2 and under reduced O2 levels. the observed rate constant increased as well. FLX degraded the fastest in SFW E, which had the same concentration of nitrate and DOM as SFW D but a greater concentration of bicarbonate. Because bicarbonate is the source of CO3•-, the increased rate constant in SFW E could be due to the higher steady-state concentration of CO3•-, which is thought to react selectively with N-containing compounds (30). However, increasing the concentration of bicarbonate from 0.8 to 5 mM also caused the pH to increase from 8 to 10. Consequently, the chemistry of the SFW system may be significantly changed by the pH, and comparison with the other SFW solutions is not possible. Competition kinetics was used to determine the secondorder rate constant for reaction of FLX with •OH. The secondorder rate constant of hydroxyl radical reaction with p-CBA is well documented as 5 × 109 M-1 s-1 (20). The bimolecular rate constant determined for the reaction of FLX with •OH using the method employed by Mabury et al. (11) was (8.4 ( 0.5) × 109 M-1 s-1. Using the second method which has been described previously by Haag and Yao (18) and Huber et al. (19), the second-order rate constant was measured to be (9.6 ( 0.8) × 109 M-1 s-1. The two methods yielded experimental values that were in general agreement, with a 15% relative difference between the two measurements. This second-order rate constant is within the observed range for rate constants of other organic compounds, including pharmaceuticals (10, 18, 19). The measured rate constant is high and approaches diffusion-controlled bimolecular rate constants; thus indirect photochemical reactions through •OH-mediated pathways could be an important process in the photochemical fate of FLX in surface waters. Indirect Photodegradation Products Identification. Samples were infused into the LC-MS-MS, and mass spectra were obtained for molecular ions corresponding to hypothesized photoproducts formed from known •OH-mediated reactions, such as hydroxylations and dealkylations. A possible photoproduct (IV) resulting from ring addition of •OH to the parent compound FLX was observed (Figure 6). Because the molecular ion observed in ES(+) mode at 326 m/z is 16 amu greater than the parent FLX (310 m/z), this suggests that an oxygenated photoproduct is formed. The molecular ion for this oxygenated degradation compound was not observed during the time course of direct photolysis experiments. The molecular ion for the proposed product IV has the same daughter ion (44 m/z) as that observed for FLX and product II (Figure 6). This daughter ion appears to be characteristic of molecules that retain the alkylamine side

FIGURE 8. Proposed ipso-mechanism leading to the formation of 4-(difluoromethylene)-2,5-cyclohexadiene-1-one (I) and 2-(methylamino)ethyl benzyl alcohol (product II) by •OH-mediated indirect photolysis. chain. The product ion spectrum also revealed a M - 18 ion (M - H2O), which is a characteristic loss for compounds containing alcohol functional groups (Figure 6). Furthermore, the 164 m/z daughter ion corresponds to HO-C6H4-+CH(CH2)2NHCH3 (Figure 6). Although the formation of an N-oxide could also lead to the observation of an ion with the same molecular weight as IV (326 m/z), it is unlikely this N-oxygenated product was formed. If an N-oxide was produced, oxygenation would have to occur on the same amino group that is found in the characteristic 44 m/z daughter fragment (H2Cd+NH-CH3). That is, this fragment would have also been oxygenated, and this was not observed in the product ion spectrum of the product IV. Using MRM monitoring, parent-daughter transitions for IV were monitored over the course of the experiment, showing that peak areas increased as the reaction progressed. This photoproduct was also observed in the SFW solutions, which also suggests that •OH-mediated reactions were occurring in these systems. The generation of product I previously observed by HPLC-UV in the direct photodegradation experiments also appeared to take place by indirect photolysis, although its production appeared to occur at a faster rate than by direct photolysis. That is, product I was observed at earlier time points in solutions containing 1 mM H2O2 as compared to in solutions in pure water exposed to simulated sunlight for the same time period. The identification of this compound was confirmed with an authentic standard. Furthermore, product II was observed by LC-MS-MS after 10 min of light exposure in solutions containing 1 mM H2O2. This same transition was not observed until after 3 h of irradiation of FLX in DI water. After the same duration of light exposure, a greater amount of product II was observed in the SFW solutions, suggesting this compound could be produced by direct and indirect photolysis in these solutions. Hydroxyl radicals are thought to be involved in Odealkylation (31) and N-dealkylations (32), and thus these degradation reactions are possible pathways for FLX in the presence of this photooxidant. Although O-dealkylation has been identified as one of the pathways for direct photolysis of FLX, this type of reaction was also observed to occur in the indirect photolysis experiments for this compound. Hydroxyl radical-induced N-dealkylation reactions have been shown to be O2-dependent (32). Following the abstraction of an H by •OH from the R-carbon to the amine, O2 adds to the carbon-centered radical to form a peroxy radical. A proton on the amine is then removed by a hydroxide ion VOL. 39, NO. 2, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

519

FIGURE 9. Comparison of LC-MS-MS chromatograms obtained in positive electrospray mode of samples grouped by column: (A) In unlabeled water 6 h after irradiation, (B) 18O-labeled water after 6 h of light exposure, and (C) in 18O-labeled water with 1 mM H2O2 10 min after irradiation. The monitored transitions and structures for FLX (310 > 44), the labeled product II (168 > 44), and the unlabeled product II (166 > 44) are shown in the figure. followed by the loss of O2- leaving an imine, which in turn hydrolyzes to produce the N-dealkylated product and an aldehyde. If the •OH-induced O-dealkylation of FLX proceeded by a similar mechanism, then the products of this reaction would be influenced by the O2 levels in the reaction system. The production of product II from FLX in the presence of H2O2 under normal and depleted O2 conditions was compared by LC-MS-MS. Results from this O2-dependence experiment indicated that the formation of product II was not influenced by the O2 levels (see Figure 7). The parentdaughter transitions for product IV were also monitored in this O2-dependence experiment, and its production was significantly reduced under depleted O2 conditions (see Figure 7). The observed effect of O2 was expected because in the addition of •OH to aromatic rings, a cyclohexadienyl radical is formed once •OH adds to the ring and aromaticity is re-established as a result of H-abstraction by O2. Another mechanism that may have led to the production of the benzylic alcohol (II) has been described previously by Mill et al. (33). However, a ketone version of product II would also be expected if this mechanism were followed and there was no spectral evidence from the infusion of photolysates and photolysates containing dinitophenylhydrazine, a known derivatizing agent for ketones and aldehydes, to support the formation of a ketone. The generation of both products I and II initiated by •OH could be explained by an ipso-substitution mechanism, which has been observed previously for other alkyl aryl ethers (31). If the •OH-induced O-dealkylation of FLX proceeds by this mechanism, then •OH adds to the ring with the -CF3 substituent and forms a hydroxycylohexadienyl radical, which is subsequently converted into a phenolic species via a phenoxyl radical with the release of an alcohol (Figure 8). The phenolic compound in this case would hydrolyze instantaneously to form product I (23), and the alcohol that is released would correspond to product II. This mechanism is not influenced by O2 levels in the reaction system, as illustrated in Figure 7 where the peaks for product II under normal and reduced O2 levels are approximately the same 520

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 2, 2005

size. This mechanism could be tested by using isotopically labeled H2O2 where the production of a labeled product I would support the ipso-mechanism. However, labeled H2O2 is not commercially available. In the SFW solutions, photosensitization reactions involving the transfer of energy from light-absorbing humic material to FLX could also lead to the formation of the photoproducts at earlier time points. Although •OH and CO3•- could also carry out the Ndemethylation of FLX to produce norfluoxetine, this degradation compound was not observed using known parentdaughter transitions from literature. Product Yield Determination. Using the purchased authentic standards, it was determined that the yield of FLX cleavage product 4-(difluoromethylene)-2,5-cyclohexdiene (product I) from direct and indirect photoinduced Odealkylation was determined to be 60% and 45%, respectively. The same yield was observed for the other cleavage product, 2-(methylamino)-ethyl benzyl alcohol (product II). Because standards for the carboxylic acid analogue (product III) and the hydroxylated analogue (product IV) were not available, their peak areas relative to FLX were compared to determine approximate yields of 5% and 25%, respectively. Thus, there are likely a number of other minor products of direct and indirect photolysis of FLX that remain unidentified. Confirmation of Hypothesized Photodegradation Mechanisms. The postulated mechanism for the formation of the product II from the direct photolysis of FLX was tested by performing this experiment in H218O. If a water molecule adds to the benzylic carbocation, then the molecular ion for II should be 2 amu greater (Figure 2). Therefore, a peak corresponding to the transition 168 > 44 should be observed in the experiment carried out in labeled water. Likewise, a peak corresponding to the usual transition of 166 > 44 should be absent or significantly reduced in intensity. From previous results, it was observed that the production of II reached a maximum after 6 h of irradiation; thus the experiment using 18O-labeled water was carried out for the same amount of time. To demonstrate that water did not have the same role

in indirect photochemical reactions, an indirect photolysis experiment was also conducted in labeled water with the addition of 1 mM H2O2. A comparison of LC-MS-MS chromatograms of photolysates taken 6 h after irradiation in regular water, after 6 h irradiation in 18O-labeled water, and after 10 min irradiation in 18O-labeled water containing 1 mM H2O2 is shown in Figure 9. As illustrated in the figure, the production of a [M + 1 + 2]+ molecular ion and the corresponding daughter fragment ion version of II (168 > 44) was not observed in the photolysate taken from unlabeled water. However, a peak corresponding to this transition was observed in the photolysate taken after the same light exposure period from labeled water. A small peak corresponding to the unlabeled product II was also observed, but this is explained by the labeled water being 95% pure and by the addition of a small volume of FLX from a stock solution that was made up entirely in unlabeled water. Labeled water experiments also showed that unlabeled Product II predominated in indirect photolysis experiments, indicating that radical reactions were primarily responsible for the generation of this degradate in SFW solutions. FLX was demonstrated to be susceptible to photodegradation with O-dealkylation identified as one primary pathway by which transformation occurs by direct and indirect means. Hydroxyl radicals appeared to dominate radical reaction pathways, and these transients could be important in limiting the persistence of this pharmaceutical in surface waters. Future work comparing the possible ecotoxicological effect of the identified photoproducts from this study with the parent compound is planned, because it remains unknown whether these photoproducts retain the biological activity associated with FLX.

Acknowledgments We gratefully acknowledge Michael Jaskolka and Nettie Che for their preliminary work on fluometuron and flutalanil, respectively, Dr. Andy Dicks for his research advice, Dr. David Ellis for his invaluable assistance in the mechanistic aspect of this research, and finally the three anonymous reviewers who provided helpful comments. We would also like to thank NSERC and the Canadian Network of Toxicology Centres for funding this project.

Literature Cited (1) Daughton, C. G.; Ternes, T. A. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect. 1999, 107, 907-938. (2) Halling-Sorensen, B.; Nielsen, N.; Lansky, P. F.; Ingerslev, F.; Hansen Lutzhoft, H. C.; Jorgensen, S. E. Occurrence, fate and effects of pharmaceutical substances in the environment. A review. Chemosphere 1998, 36, 357-393. (3) Buser, H. R.; Muller, M. D. Occurrence of the pharmaceutical drug clofibric acid and the herbicide mecoprop in various Swiss lakes and in the North Sea. Environ. Sci. Technol. 1998, 32, 188-192. (4) Stumpf, M.; Ternes, T. A.; Wilken, R.; Todrigues, S. V.; Baumann, W. Polar drug residues in sewage and natural waters in the state of Rio de Janeiro, Brazil. Sci. Total Environ. 1999, 225, 135-141. (5) Kolpin, D. W.; Furlong, E. T.; Meyer, M. T.; Thurman, E. M.; Zaugg, S. D.; Barber, L. B.; Buxton, H. T. Pharmaceuticals, and organic wastewater contaminants in US streams: a national reconnaissance. Environ. Sci. Technol. 2002, 36, 1202-1211. (6) Brooks, B. W.; Foran, C. M.; Richards, S. M.; Weston, J.; Turner, P. K.; Sanley, J. K. Aquatic ecotoxicity of fluoxetine. Toxicol. Lett. 2003, 142, 169-183. (7) Brooks, B. W.; Turner, P. K.; Stanley, J. K.; Weston, J. J.; Glidewell, E. A.; Foran, C. M.; Slattery, M.; La Point, T. W.; Huggett, D. B. Waterborne and sediment toxicity of fluoxetine to select organisms. Chemosphere 2003, 52, 135-142.

(8) Richards, S. M.; Wilson, C. J.; Johnson, D. J.; Castle, D. M.; Lam, M. W.; Mabury, S. A.; Sibley, P. K.; Solomon, K. R. Effects of pharmaceutical mixtures in aquatic microcosms. Environ. Toxicol. Chem. 2003, 23, 1035-1042. (9) Lam, M. W.; Young, C. J.; Brain, R. A.; Hanson, M. L.; Johnson, D. J.; Wilson, C. J.; Richards, S. M.; Solomon, K. R.; Mabury, S. A. Aquatic persistence of eight pharmaceuticals in a microcosm study. Environ. Toxicol. Chem. 2003, 23, 1431-1440. (10) Latch, D. E.; Stender, B. L.; Packer, J. L.; Arnold, W. A.; McNeill, K. Photochemical fate of pharmaceuticals in the environment: cimetidine and ranitidine. Environ. Sci. Technol. 2003, 37, 33423350. (11) Mabury, S. A.; Crosby, D. G. Pesticide reactivity toward hydroxyl and its relationship to field persistence. J. Agric. Food Chem. 1996, 44, 1920-1924. (12) Armbrust, K. L. Pesticide hydroxyl radical rate constants: measurements and estimates of their importance in aquatic systems. Environ. Toxicol. Chem. 1999, 19, 2175-2180. (13) Zepp, R. G.; Hoigne, J.; Bader, H. Nitrate-induced photooxidation of trace organic chemicals in water. Environ. Sci. Technol. 1987, 21, 443-450. (14) Vaughan, P. P.; Blough, N. V. Photochemical formation of hydroxyl radical by constituents of natural waters. Environ. Sci. Technol. 1998, 32, 2947-2953. (15) Brezonik, P. L.; Fulkerson-Brekken. Nitrate-induced photolysis in natural waters: controls on concentrations of hydroxyl radical photointermediates by natural scavenging agents. Environ. Sci. Technol. 1998, 32, 3004-3010. (16) Russi, H.; Kotzias, D.; Korte, F. Photoinduce hydroxylation reactions of organic chemicals in natuual waters- nitrates as potenital sources for OH-radicals. Chemosphere 1982, 11, 10411048. (17) Lam, M. W.; Tantuco, K.; Mabury, S. A. PhotoFate: a new approach in accounting for the contribution of indirect photolysis of pesticides and pharmaceuticals in surface waters. Environ. Sci. Technol. 2003, 37, 899-907. (18) Haag, W. R.; Yao, C. C. D. Rate constants for reaction of hydroxyl radicals with several drinking water contaminants. Environ. Sci. Technol. 1992, 26, 1005-1013. (19) Huber, M. M.; Canonica, S.; Park, G.-Y.; Gunten, U. V. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 2003, 37, 1016-1024. (20) Buxton, B. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms, and hydroxyl radicals in aqueous solution. J. Phys. Chem. Ref. Data 1988, 17, 513-886. (21) Risley, D. S.; Bopp, R. J. Analytical Profiles of Drug Substances; Academic Press Inc.: San Diego, CA, 1990; Vol. 19. (22) Souter, R. W.; Dinner, A. GLC determination of degradation of two amine uptake-inhibitors. J. Pharm. Sci. 1976, 65, 457-459. (23) Ellis, D. A.; Mabury, S. A. The aqueous photolysis of TFM and related trifluoromethylphenols. An alternate source of trifluoroacetic acid in the environment. Environ. Sci. Technol. 2000, 34, 632-637. (24) Bosca, F.; Cuquerella, M. C.; Marin, M. L.; Miranda, M. A. Photochemistry of 2-hydroxy-4-trifluoromethylbenzoic acid, major metabolite of the photosensitizing platelet antiaggregant drug triflusal. Photochem. Photobiol. 2001, 73, 463-468. (25) Volmer, D. A.; Mansoori, B.; Locke, S. J. Study of 4-quinolone antibiotics in biological samples by short column liquid chromatography coupled with electrospray ionization tandem mass spectrometry. Anal. Chem. 1997, 69, 4143-4155. (26) Hutson, D.; Miyamoto, J. Fungicidal activity: chemical and biological approaches to plant protection; John Wiley and Sons: New York, 1998. (27) Tsao, R.; Eto, M. Photolysis of flutalanil fungicide and the effect of some photosensitizers. Agric. Biol. Chem. 1991, 55, 763-768. (28) Hustert, K.; Moza, P. N.; Kettrup, A. Photochemical degradation of carboxin and oxycarboxin in the presence of humic substances and soil. Chemosphere 1999, 38, 3423-3429. (29) Sutherland, F. C. W.; Badenhorst, D.; de Jager, A. D.; Scanes, T.; Hundt, H. K. L.; Swart, K. J.; Hundt, A. F. Sensitive liquid chromatographic-tandem mass spectrometric method for the determination of fluoxetine and its primary active metabolite norfluoxetine in human plasma. J. Chromatogr., A 2001, 914, 45-51. (30) Larson, R. A.; Zepp, R. G. Reactivity of the carbonate radical with aniline derivatives. Environ. Toxicol. Chem. 1988, 7, 265274. VOL. 39, NO. 2, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

521

(31) Urano, Y.; Higuchi, T.; Hirobe, M. Substrate-dependent changes of the oxidative O-dealkylation mechanisms of several chemical and biological oxidizing systems. J. Chem. Soc., Perkin Trans. 2 1996, 6, 1169-1173. (32) Tauber, A.; von Sonntag, C. Products and kinetics of the OHinduced dealkylation of atrazine. Acta Hydrochim. Hydrobiol. 2000, 28, 15-23.

522

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 2, 2005

(33) Mill, T.; Hendry, D. G.; Richardson, H. Free-radical oxidants in natural waters. Science 1980, 207, 886-887.

Received for review April 6, 2004. Revised manuscript received August 27, 2004. Accepted October 20, 2004. ES0494757