direct photolysis of human metabolites of the antibiotic sulfamethoxazole: evidence for abiotic...

Direct Photolysis of Human Metabolites of the AntibioticSulfamethoxazole: Evidence for Abiotic Back-TransformationFlorence Bonvin,† Julien Omlin,† Rebecca Rutler,† W. Bernd Schweizer,‡ Peter J. Alaimo,§

Timothy J. Strathmann,∥ Kristopher McNeill,⊥ and Tamar Kohn*,†

†Ecole Polytechnique Federale de Lausanne (EPFL), Environmental Chemistry Laboratory, School of Architecture, Civil andEnvironmental Engineering (ENAC), Lausanne, Switzerland‡Organic Chemistry Laboratory, ETH Zurich, 8093-Zurich, Switzerland§Seattle University, Department of Chemistry, Seattle, Washington 98122, United States∥University of Illinois, Department of Civil and Environmental Engineering, Urbana-Champaign, IL, United States⊥Institute of Biogeochemistry and Pollutant Dynamics (IBP), ETH Zurich, 8092-Zurich, Switzerland

*S Supporting Information

ABSTRACT: The presence of potentially persistent andbioactive human metabolites in surface waters gives rise toconcern; yet little is known to date about the environmentalfate of these compounds. This work investigates the directphotolysis of human metabolites of the antibiotic sulfamethox-azole (SMX). In particular, we determined photolysis kineticsand products, as well as their concentrations in lake water.SMX, N-acetyl sulfamethoxazole, sulfamethoxazole β-D-glucur-onide, 4-nitroso sulfamethoxazole, and 4-nitro sulfamethox-azole were irradiated under various light sources and pHconditions. All investigated metabolites, except sulfamethoxazole β-D-glucuronide were found to be more photostable than SMXunder environmentally relevant conditions. Between two and nine confirmed photoproducts were identified for SMX-metabolitesthrough ultraperformance liquid chromatography/high-resolution mass spectrometry. Interestingly, photolytic back-trans-formation to SMX was observed for 4-nitroso-SMX, indicating that this metabolite may serve as an environmental source ofSMX. Moreover, two human metabolites along with SMX were regularly detected in Lake Geneva. The knowledge that somemetabolites retain biological activity, combined with their presence in the environment and their potential to retransform to theparent compound, underlines the importance of including human metabolites when assessing the effects of pharmaceuticals inthe environment.

! INTRODUCTIONAntibiotics are now well-acknowledged contaminants of naturalaquatic systems. The chronic exposure of bacteria and otheraquatic organisms to trace concentrations of antibiotics raisesconcerns regarding their ecotoxicological effects, but also theirpotential to induce bacterial resistance. Among the targetcompounds measured, the antibiotic sulfamethoxazole (SMX)has regularly been detected in wastewaters and natural aquaticenvironments with median concentrations between 60 and 150ng/L.1,2 Moreover, the PNEC (predicted no effect concen-tration) of SMX, a level which should not be surpassed toensure an acceptable risk to the environment, has beenrepeatedly exceeded. The frequent detection of this antibioticin surface waters can be explained by its extensive use in bothhuman and veterinary medicine,3 its poor elimination inconventional wastewater treatment plants4−7 and its relativepersistence in the environment.8 Much effort has been spent oninvestigating the presence and fate of SMX in the aquaticenvironment.1,9 Photodegradation was identified as the majordegradation pathway for SMX in surface waters,10 thus direct

and indirect photodegradation kinetics of SMX have beenextensively studied.11−15

More recently, research interests have shifted towardidentifying photolysis products of pharmaceuticals, as well asthe presence and fate of their human metabolites, as both maypresent a risk to the aquatic ecosystem. Selected photoproductshave been shown to be more persistent than the correspondingparent compound and to retain biological activity.16,17 Similarly,though human metabolites are generally more polar than theparent compounds, they are not always less toxic.18,19

Only 14% of ingested SMX is excreted in its original form,yielding a large fraction of metabolites19 (Figure 1, right panel).The most prominent metabolite is N-acetyl sulfamethoxazole

Special Issue: Rene Schwarzenbach Tribute

Received: September 18, 2012Revised: November 19, 2012Accepted: November 27, 2012Published: November 27, 2012

Article

pubs.acs.org/est

© 2012 American Chemical Society 6746 dx.doi.org/10.1021/es303777k | Environ. Sci. Technol. 2013, 47, 6746−6755

pubs.acs.org/est

(Ac-SMX), which represents 50% of the excreted administereddose. Other metabolites include sulfamethoxazole β-D-glucur-onide (SMX-glucuronide) (9%), N-hydroxy sulfamethoxazole(OH-SMX) (2.2%), and minor fractions of 4-nitrososulfamethoxazole (NO-SMX) and 4-nitro sulfamethoxazole(NO2-SMX).19

The presence and fate of these metabolites in wastewater andreceiving waters has received little scrutiny to date. NO-SMXwas determined to be even more cytotoxic than SMX itself20

but has never been reported in environmental studies. Acetyl-SMX has been detected in wastewater effluent21 as well assurface waters at concentrations comparable to its parentcompound.22 In addition, there is evidence that Ac-SMX maybe transformed back to the parent compound duringwastewater treatment.3 Biological back-transformation wasalso reported for both Ac-SMX and SMX-glucuronide inwater sediment tests23 Moreover, a potential abiotic back-transformation of NO2-SMX, was recently revealed understrongly reducing conditions.24 Photolytic back-transformation,however, has not been investigated to our knowledge.Understanding the photochemical fate of SMX metabolites

and their photoproducts is essential to fully evaluate the riskassociated with SMX in the environment. Therefore, thepresent study aims to investigate the photolysis kinetics andproducts of a selection of SMX metabolites, namely the onesmost abundantly found in urine,19 Ac-SMX and SMX-glucuronide, as well as OH-SMX, NO-SMX, and NO2-SMX.The possibility of photolytic back-transformation to SMX wasassessed, and the influence of the small structural differencesbetween SMX and its metabolites on photolysis rates andpathways was investigated. Finally, the influence of thephotolability of metabolites on their fate and expecteddistribution in surface waters is discussed and compared toactual concentrations measured in Lake Geneva near awastewater treatment plant effluent discharge point.

! MATERIALS AND METHODSChemicals. SMX, Ac-SMX, OH-SMX, NO-SMX (>90%),

NO2-SMX, SMX-glucuronide were obtained from TorontoResearch Chemicals. 4-(Acetylamino)benzensulfonic acid, 3-amino-5-methylisoxazole, 4-nitrobenzenesulfonamide, N-phe-nylacetamide, aniline, and sulfanilic acid were all analyticalgrade from Sigma-Aldrich. (5-Methylisoxazol-3-yl)sulfamatewas synthesized via sodium (5-methylisoxazol-3-yl)sulfamateusing an adaptation of a procedure by Spillane et al.25 Detailsregarding this synthesis and confirmation of the product areavailable in the Supporting Information (SI). All solutions weremade using Nanopure water from either a Millipore SynergyUV or a Barnstead NANOpure Diamond Water PurificationSystem (resistivity >18.2 MΩ cm). Stock solutions were madein HPLC grade methanol and other solvents used were allanalytical grade quality. Further chemicals were used in buffersolutions and eluents, namely ammonium acetate (Sigma-Aldrich), acetic acid (Fluka), disodium tetra borate decahydrate(Merck), sodium bicarbonate (Fluka), pyridine (Acros), and p-nitroanisole (Sigma-Aldrich).

Spectrophotometric Titration for pKa,2 Determina-tion. The absorbance spectra of aqueous solutions (Nanopurewater) of each individual metabolite (5 mg·L−1) were measuredwith a 2550 Spectrophotometer (Shimadzu Scientific Instru-ments) in the pH range from 2 to 11. pH was adjusted usingHCl and NaOH. pKa,2 values were determined by a least-squares fit of absorbance versus pH data at a single wavelength,with the inflection point representing the pKa,2 value (FigureS2, SI). pKa,1-values of sulfa-drugs are generally below pH 2 andthus not relevant to natural water systems.

Direct Photolysis Experiments. The direct photolysiskinetics of SMX and metabolites were measured utilizing twodifferent setups.

Solar Simulator Setup. In the solar simulator setup,solutions of individual target compounds (1 mg·L−1) inbuffered Nanopure water were irradiated between 1 and 24 h

Figure 1. Left panel: Structure of parent compound sulfamethoxazole (SMX), protonation states and main cleavage sites of SMX photolysis(adapted from ref 11). Right panel: Major human metabolites of sulfamethoxazole.

Environmental Science & Technology Article

dx.doi.org/10.1021/es303777k | Environ. Sci. Technol. 2013, 47, 6746−67556747

from above by a Sun 2000 Solar Simulator (ABETTechnologies, Milford, Connecticut) equipped with a 1000W Xe lamp and an AM1.5 filter to mimic solar radiation. Theirradiance for this setup was determined spectroradiometrically(Model ILT-900-R, International Light) before and after eachexperiment. The absolute irradiance was calibrated usingchemical actinometry (p-nitroanisole (pNA)) and was 76 W/m2 between 265 and 430 nm and showed no day-to-dayvariation (lamp spectrum in SI, Figure S1). The irradiatedsolutions (400 mL) in amber glass beakers were kepthomogeneous by stirring, and their temperature was main-tained at ∼19 °C using a water-filled tray, connected to arecirculation cooler (F240 Recirculating Cooler, Julabo).Identical solutions were left in the dark during each experimentto serve as dark controls. Samples (1.2 mL) were collected atselected time points (min. 10) to monitor the parentcompound concentration decrease over irradiation time. Thesamples in amber glass vials were maintained at 4 °C aftercollection and analyzed within 48 h by ultraperformance liquidchromatography (Acquity UPLC system, Waters) coupled to atandem mass spectrometer (MS/MS, Acquity TQD, Waters).Each experiment was conducted at least twice. Data acquisitionand processing was performed using Masslynx. In parallel,absorbance spectra for all solutions were collected with a UV−vis 2550 Spectrophotometer (Shimadzu Scientific Instruments)at various irradiation time points to assess the change inabsorbance over time and wavelength. Averaged values wereused to correct for light screening. As pH can affect theprotonation state and thus the photolysis kinetics of organiccompounds,11 direct photolysis experiments were conductedunder acidic (pH 3.2) and mildly basic (pH 8.4) conditions.The various buffers, chromatographic conditions and analyticalmethods are described in the SI (S6).Rayonet Setup. The second setup used a turn-table

apparatus inside a photochemical reactor (Rayonet) equippedwith six 300 nm bulbs (Southern New England Ultraviolet Co.RPR-3500 Å). The bulb spectrum is shown in the SI (FigureS1). The irradiation intensity, also calibrated with a chemicalactinometer (pNA) from 265 to 430 nm, was 44.4 W/m2. Inthis setup, solutions containing 12−30 mg/L of the targetcompound in borate-buffered (pH 9) Nanopure water wereirradiated in borosilicate glass test tubes for 1 to 24 h. Higherconcentrations were used to facilitate and enhance photo-product identification. Two 150 μL aliquots were taken atvarious time points, and analyzed in parallel via twocomplementary methods: the first used a HPLC (Dionex)equipped with a UV detector to obtain quantitative kineticdata; the second allowed for high resolution mass determi-nation of photoproducts by UPLC separation of compounds(NanoAcquity, Waters) and detection on a high-resolutionmass spectrometer (HR-MS, Thermo Exactive Orbitrap)(details below). Dark controls of the experimental solutionswere monitored for parent compound decay over the durationof the photolysis experiments. Data acquisition and processingwas done using Chromeleon (HPLC-UV) or Excalibur andToxid (UPLC-HRMS). Again, to account for light screening,the absorbance (200−600 nm) of the solutions was monitoredover time using a Cary 100 Bio (Varian) UV/Visiblespectrophotometer. Solutions of a chemical actinometer pNAwith pyridine26 were photolyzed concurrently with SMX andmetabolites as a measure of the effective light seen by thecompounds in solution. The analytical methods are describedthoroughly in the SI (S6).

Product Identification. Samples were directly analyzed byUPLC-HRMS and MS data was recorded scanning from m/z90 to m/z 600 in both positive and negative ESI mode. Fullmass spectra in both modes were explored for appearance ofpeaks with increasing irradiation time. Chemical formulas withtheoretical exact masses within 5 ppm of the detected masswere considered as possible products. Thereafter, formulaswhich made the most chemical sense with respect to thestarting material were investigated further. In absence of MS/MS fragmentation, no structural elucidation per se was possible.However the presence of fragments appearing at the sameretention time due to in-source fragmentation aided thestructural elucidation. The identity of a selection of photo-products could be fully confirmed by matching exact mass andretention time to authentic standards (commercially availableor synthesized). Identification of HPLC-UV peaks was achievedthrough fractionation of concentrated photolysate. Explicitly,after reduction of the photolyzed experimental solution to ∼1mL via evaporation, the concentrated solution was fractionatedby HPLC and the fractions directly injected on the HRMS todetermine the exact mass of the peak captured in the fraction.

Direct Photolysis Calculations. The direct photolysis rateconstants were determined under various conditions (pH, lightsource and concentration). The degradation of all metabolitesfollowed first-order kinetics (Figure S4a, SI) and degradationrate constants kobs,i were calculated using the slope of ln(At/A0)plotted against time, where At refers to the peak area ofchromatograms at time t and A0 refers to the initial peak area.For experiments with relatively high concentrations of targetcompound and/or large volumes, a light-screening factor wasused to account for the fact that the average light intensity inthe tube was lower than in optically dilute solutions, because ofself-screening by the compound itself or transformationproducts forming. The observed degradation rate constant,kobs,i, for each compound i was corrected to yield a photolysisrate constant representative of an optically dilute system, kobs,i

0 .The direct photolysis quantum yield (Φi) was determined for

each compound (i). It represents the efficiency of directphotolysis and was calculated as follows:

Φ =kki

i

i

obs,0

abs,0

(1)

where kabs,i0 is the specific rate of light absorption (kabs,i

0 ), ameasure of the spectral overlap of light irradiance andcompound absorbance (see SI, S7). Details regarding lightscreening corrections, quantum yield calculations, and calibra-tion of light intensity by chemical actinometry and errorcalculations can be found in the SI.Finally, knowledge of the quantum yield and solar irradiance

allows for the estimation of environmental half-lives. Accord-ingly, these were calculated by multiplying the quantum yieldby the specific rate of light absorption of each compound undera theoretical solar irradiation (SMARTS), at 47° N latitude, onthe 21st of June (summer scenario) and 21st of December(winter scenario), assuming continuous exposure, half-lightattenuation, and omitting indirect photolysis processes.

Surface Water Concentrations. The presence of Ac-SMXand SMX-glucuronide along with the parent compound SMXwas investigated in water samples collected on a monthly basisat various depths of two sampling sites (reference sites) in theVidy Bay of Lake Geneva between April 2010 and January



Table

1.SpecificRateof

LightAbsorbtion(k

abs,i

0),DirectPh

otolysisDegradatio

nRateCon

stants

(kabs,i

0),ExperimentalHalf-L

ives,Q

uantum

YieldandCom

puted

Environm

entalHalf-L

ives

(!)forSM

XandItsHum

anMetabolitesObservedin

theSolarSimulator

Setup(sol.sim

.)andin

theRayon

etSetupa

compound

pHirradiancesetup

concentration(m

g/L)

k abs,i

0(!

10−3

h−1 )

k abs,i

0(h

−1)b

experim

entalh

alf-life

(h)c

ndquantum

yield

τ sum

mer(h)e

τ(h)e

SMX

3.2

sol.sim.

16.45

±1.48

22.3±

1.86

0.031±

0.002

20.959±

0.235

10.4±

0.4

44±

28.4

sol.sim.

0.5−

35.36

±0.87

1.42

±0.08

0.494±

0.018

40.074±

0.013

9.2

Rayonet

2216.95±

2.88

1.36

±0.10

0.440±

0.018

40.028±

0.005

Ac-SM

X3.2

sol.sim.

12.32

±0.38

6.19

±0.49

0.111±

0.006

20.543±

0.132

61.5±

2.5

246±

108.4

sol.sim.

0.5−

33.17

±0.72

0.207±

0.017

3.43

±0.22

40.025±

0.005

9.2

Rayonet

3111.24±

1.66

0.096±

0.008

6.14

±0.353

30.003±

0.0005

NO-SMX

3.2

sol.sim.

135.42±

8.15

0.303±

0.034

2.29

±0.19

20.002±

0.0006

10.7±

0.3

47±

18.4

sol.sim.

0.5−

379.64±

12.95

0.184±

0.010

3.87

±0.15

40.0006

±0.0001

9.2

Rayonet

12−2

4212.65

±25.60

0.540±

0.056

1.10

±0.05

50.0009

±0.0001

NO

2-SM

X3.2

sol.sim.

111.49±

2.64

0.062±

0.020

11.3

±3.25

20.0015

±0.0006

337±

301515

±135

8.4

sol.sim.

129.40±

4.78

0.0029

±0.0005

79.3

±3.15

10.00003±

0.00001

9.2

Rayonet

n.d.e

n.d.

n.d

n.d

0n.d

SMX-glucuronide

3.2

sol.sim.

112.84±

2.95

16.27±

1.43

0.042±

0.003

20.352±

0.087

1.3±

0.1

±0.7

8.4

sol.sim.

111.23±

1.83

12.41±

1.03

0.043±

0.002

30.307±

0.064

9.2

Rayonet

1234.91±

4.01

15.57±

0.71

0.045±

0.001

20.124±

0.015

aErrorsrepresent95%

confidenceintervals(detailsin

theSI,S

8).bObserveddirect

photolysisreactio

nrate

constant

correctedforlight

screening.c Corrected

forlight

screening.dNum

berof

replicas.

e Calculatedusingthedirect

photolysisquantum

yielddeterm

ined

intheRayonet

setupandSM

ARTSirradiancevalues

forsolarirradiatio

non

asunnysummer

day(21stJune)/winterday(21st

Decem

ber)

at47°N.L

ight

attenuationby

thewater

columnwas

incorporated

byassuminghalf-light

attenuationcorrelates

with

ahalvingof

thedirect

photolysisreactio

nrate

constant.11

e nd:

Not

determ

ined

becauseof

high

photostabilityobserved

insolarsim

ulator

setup.



2011. Sampling strategy, sample preparation, extraction, andanalytical methods have been described previously.2,27

! RESULTS AND DISCUSSIONAcid−Base Speciation of SMX Metabolites. SMX is a

diprotonic acid (Figure 1, left panel), yet only the transitionfrom the neutral to the anionic form through deprotonation ofthe sulfonamide NH (pKa,2) is relevant for natural waters. ForSMX and its human metabolites, Ac-SMX, NO-SMX, and NO2-SMX, pKa,2 values of 5.89 ± 0.07, 5.07 ± 0.08, 4.71 ± 0.1 and3.66 ± 0.01, respectively, were obtained (Figure S2, SI). Boreenet al. found a comparable pKa,2 for SMX of 5.7.11 No valueswere determined herein for SMX-glucuronide or OH-SMX.SMX-glucuronide is missing this acidic functionality and thusremains in its neutral form over the entire tested pH range (2−10). OH-SMX was rapidly converted to NO-SMX in neutraland basic aerated solutions, and because of this instability inexperimental conditions its pKa,2 was not determined nor was itconsidered for further experiments.Direct Photolysis Kinetics, Quantum Yields, and

Environmental Half-Lives. The protonation state of organiccompounds affects their absorbance (Figure S3, SI) resulting indifferences in direct photolysis kinetics.11 For this reason, directphotolysis experiments were performed at different pH valuesand the resulting photolysis rate constants of the predominantlyneutral form (pH 3.2) and anionic form (8.4 and 9.2) are listedin Table 1. The pH dependence of the photolysis kinetics wasevident, with faster reaction kinetics in acidic conditions formost compounds. In accordance with previous research,11,13

the neutral form of SMX (prevalent at pH 3.2) was found to bemore photolabile than the anionic form. SMX metabolites alsodegraded faster under acidic conditions, showing up to 30-timesfaster degradation kinetics of the neutral form. The accelerateddirect photolysis in acidic solution was not observed for SMX-glucuronide. This result was expected as SMX-glucuronide ismissing the sulfonamide NH acidic functionality; therefore, incontrast to SMX and the other metabolites, it remains in itsneutral and more photoreactive form at environmental pH. Thespecific rate of light absorption of each compound (Table 1) inits neutral or anionic form was compared to its correspondingreaction rate constant. Both SMX and Ac-SMX showed apositive correlation between absorbed light and rate constant,with higher absorbance of the neutral compounds leading tofaster kinetics. In contrast, despite greater light absorption ofNO-SMX and NO2-SMX in their anionic form, the neutralform degraded more readily.Among the target compounds at pH 3.2, the fastest

degradation was observed for SMX. The degradation ofSMX-glucuronide and Ac-SMX were nearly as fast as SMX,whereas NO-SMX and NO2-SMX were orders of magnitudeslower. At pH 8.4 and 9.2, SMX-glucuronide showed the fastestdegradation kinetics of all investigated compounds followed bythe parent compound SMX. On the other hand, NO-SMX, Ac-SMX, and NO2-SMX showed significantly lower rate constants(Table 1). Finally, self-sensitization by the compound itself wasruled out, as experiments performed at different concentrationsshowed no significant differences in kinetics (Figure S4b, SI).The majority of investigated compounds differ only in their

aromatic-ring substituent, yet large differences were observedwithin direct photolysis rate constants. Hence, the different ringsubstituents influence the photochemical behavior of thesecompounds, as discussed in detail below. Similarly, previousresearch found different direct photolysis behavior among five

sulfa drugs with varying five-membered heterocyclic substitu-ents.10 Direct photolysis is thus clearly sensitive to both ringfunctionalities of the molecule, though differences in five-membered ring substituents led to relatively small variationscompared to the aromatic-ring substituents investigated here.Similar trends as for rate constants could be observed for

quantum yields (Table 1). The quantum yields of the neutralcomponents (pH 3.2) were greater than of the anionic form ofthe compound, except for SMX-glucuronide, which remained inits neutral form in all experimental conditions. Furthermore,both SMX-glucuronide and SMX showed relatively highquantum yields, compared to values 2−3 orders of magnitudelower for Ac-SMX, NO-SMX, and NO2-SMX.Quantum yields determined in the solar simulator setup were

systematically higher than those determined using the Rayonetsetup, with the exception of NO-SMX. A recent reviewreported a wavelength specific quantum yield for SMX of Φ254= 0.038 ± 0.002,28 which is comparable to the one measured inthe Rayonet setup (λmax = 300 nm), ΦRayonet = 0.028 ± 0.005 inthe present study but lower than the quantum yield determinedby Zhou and Moore, Φ257 = 0.084 ± 0.016 at pH 9.13 The latterfound values closer to the quantum yields determined undernatural sunlight Φsunlight = 0.09 ± 0.01,11 which in turn matchthe ones determined here using the solar simulator setupΦsol.sim = 0.074 ± 0.012. The sensitive nature of quantum yieldcalculations has been reported29 and may explain varianceamong different experimental setups. The molar absorptivity ofall compounds at pH 9.2 and 8.4 are identical and accordinglydo not affect the quantum yields in the present case. However,owing to the small spectral overlap of molar absorptivity andsolar simulator irradiance for certain compounds (all but NO-SMX and NO2-SMX) (Figure S5, SI), slight errors in theirdetermination may largely influence the rate of light absorption,kabs,i0 , and hence lead to differences in quantum yields. Incontrast to the solar simulator irradiance, which begins at 280nm, the Rayonet irradiance spans a wavelength range from 250to 400 nm, allowing for a larger spectral overlap with the targetcompounds (Figure S5, SI) and leading to a likely more robustquantum yield. On the other hand, NO-SMX shows a largespectral overlap in both experimental setups because of itsabsorption range up to 430 nm; it is thus more robust withrespect to small inaccuracies in the absorbance and irradiancemeasurements. Correspondingly, it is the only compound withcomparable quantum yields in both setups. Furthermore,quantum yields may be wavelength-dependent, which mayalso account for the observed differences with various lightsources. Environmental half-lives were computed using theRayonet quantum yields, which we consider to be more robust.The minimum environmental half-lives were computed for a

sunny summer day (τsummer) at 47° N latitude (Geneva,Switzerland) and ranged from 1.3 h for SMX-glucuronide to 14days for NO2−SMX. In the winter, the calculated half-lives(τwinter) increased to range from 6 h to 63 days (Table 1). In alake environment, such as Lake Geneva, summer stratificationconstrains the mixing of the water column and consequentlyprevent inflowing micropollutants entering below the thermo-cline to reach the surface and undergo photolysis. Additionally,they may accumulate below the thermocline.2 In wintermonths, photodegradation is greatly reduced; however, theenhanced mixing of the water column ensures a morehomogeneous sunlight exposure throughout the water column.Therefore, the reduced winter photolysis may nevertheless have



a significant impact on concentrations of SMX and itsmetabolites in the lake.Product Identification. Various studies have examined

photoproducts arising from direct photolysis of SMX. Themajority of reported photoproducts arise from cleavage of themolecule at various positions (Figure 1, left panel). Severalauthors have identified δ-cleavage, yielding sulfanilic acid, as theprevailing mechanism,11,30 though at least one study reportedphotoisomerization of the five-membered isoxazole as the mainmechanism.13

The percent conversion was used to quantify the relativeimportance of confirmed photoproducts of SMX and itsmetabolites. Explicitly, the conversion corresponds to thepercentage of irradiated target compound converted to a givenfirst-generation product; it is calculated from the ratio of theinitial rates of product concentration growth and targetcompound concentration loss. The conversion of the targetcompounds to first generation products could only becalculated for products with available standards.

SMX. Similarly to previous research, sulfanilic acid (m/z 172)was found to be the most important quantifiable product of

Figure 2. Photolysis (at pH 9.2) of SMX (a), Ac-SMX (b) and NO-SMX (c) over time with appearance of quantifiable photoproducts and carbonmass balance. Left axis shows concentration of irradiated substance (mg·L−1), first right-axis the concentration of each quantifiable photoproduct(mg·L−1) and second right-axis the carbon-mass balance (%). (d) Enlargement of (c) with concentration of SMX and Ac-SMX (1st or 2ndgeneration products of NO-SMX) on the left axis (mg·L−1) and their corresponding photoproducts on the right axis (mg·L−1). +, MS positiveionization mode; −, MS negative ionization mode.



SMX photolysis, with a conversion of 20%. SMX also produced(5-methylisoxazol-3-yl)sulfamate (m/z 176), a product whichhas not been reported to date and could be confirmed withmatching exact mass, retention time and UV spectra with thesynthesized standard (validation of standard (NMR, crystalstructure, IR, HRMS) available in the SI). Generated throughgamma-cleavage, it accounts for 11% of degraded SMX. Furtherproducts that could be confirmed with authentic standardsinclude aniline (5%, m/z 94) and 3-amino-5-methylisoxazole(2%, m/z 99). Both have previously been reported asphotoproducts of SMX, and result respectively from γ- and δ-cleavage.13,16

The concentrations of quantifiable photoproducts of SMXover time are depicted in Figure 2a. As mentioned above, Zhouand co-workers reported an isomer of SMX, generated throughphotorearrangement of the isoxazole ring, as the major productwith a yield of 30%.13 They also identified a hydrated productof 2H-azirine, with mass [m − H]− = 270.0550.13 Thoughthese have not systematically been observed in more recentstudies,11 the exact masses of both an SMX isomer andhydrated 2H-azirine were observed in the present study. These,however, could not be confirmed, because of a lack of authenticstandards or/and mass spectral fragmentation data. A numberof additional products were formed during SMX photolysis.Their exact masses in combination with mass spectral data,when available, were used to propose structures (Table S10,SI).Among the proposed structures are hydroxylated products,

such as isomers of OH-SMX [m + H]+ = 270.0533 (at severalretention times), the hydroxylated sulfamate (4-hydroxy-5-methylisoxazole-3-yl)sulfamic acid ([m − H]− = 192.9915),hydroxyaminobenzene sulfonic acid ([m − H]− = 188.0013),hydroxysulfonic acid ([m − H]− = 172.9906), and benzenesulfonic acid ([m − H]− = 156.9954). β-cleavage may also beoccurring, with removal of the NH2 group, yielding [m + H]+ =239.0478. A recent study on photolysis of other sulfonamidesand their acetylated-metabolites found desulfonated productsto be the most relevant photodegradation products identified.31

In our work, the exact mass signal corresponding to the SO2-extrusion product of SMX was also weakly visible.Ac-SMX. The degradation of Ac-SMX and the growth of

confirmed products over time are shown in Figure 2b. Totalconversion was highest for Ac-SMX, which was mainlytransformed to 4-acetamidobenzenesulfonic acid (88%, m/z214), the acetylated equivalent of sulfanilic acid, generatedthrough δ-cleavage. Other first-generation products from directirradiation of Ac-SMX include N-phenylacetamide (13%, m/z136) and (5-methylisoxazol-3-yl)sulfamate (2%, m/z 176),both produced through γ-cleavage. The production of 3-amino-5-methylisoxazole (m/z 99), already observed during SMXphotolysis, was also confirmed with an authentic standard. Thehighest concentrations of most photoproducts appeared wereattained after 8h of photolysis.A further selection of products can be proposed on the basis

of exact mass information. As observed for SMX photolysis, [m+ H]+ = 239.0485 is the proposed product of a β-cleavage. [m+ H]+ = 270.0536 could be observed at various retention timesand may represent isomers of hydroxy-SMX. In contrast toSMX, a mass likely corresponding to the SO2-extrusion productof Ac-SMX ([m + H]+ = 232.1073) appeared after 1 hirradiation with a high intensity at a retention time of 10.52min. The exact masses of all confirmed and proposedphotoproducts are within 5 ppm of calculated masses of the

proposed elemental composition and can be found Table S11,SI.

SMX-Glucuronide. SMX-glucuronide underwent δ-cleavage,yielding sulfanilic acid (m/z 172), which was, with aniline (m/z94) the only confirmed photoproduct for this compound.Though exact masses corresponding to glucopyranuric acid ([m+ H]+ = 194.0659) and glucuronic acid ([M − H]− =193.0354) appeared with increasing irradiation-time, support-ing cleavage of the glucuronide, no production of the parentcompound SMX was detected. The accurate mass correspond-ing to the desulfonated product of SMX-glucuronide ([M +H]+ = 366.1294) was also observed. Proposed structures arepresented in Table S12, SI.

NO-SMX. NO-SMX showed the highest number of identifiedproducts, mainly due to the production of SMX and Ac-SMX,which subsequently photolyzed to some of the above-mentioned products. This notable back-transformation ofNO-SMX to SMX (m/z 254 in Figure 2c) is the first evidencefor photolytic back-transformation of a metabolite to its parentcompound. However, the relative magnitude of this pathwayremains small: the initial 18.6 mg·L−1 NO-SMX yielded nomore than 0.5 mg·L−1 of SMX, which rapidly degraded to formsulfanilic acid (m/z 172) and aniline (m/z 94). Similarly, Ac-SMX (m/z 296) was observed right from the first sampled timepoint (0.3 h), but its maximum concentration remained below0.5 mg·L−1. In accordance with the slower photolysis kinetics ofAc-SMX, its respective photoproducts appeared with a slighttime shift relative to SMX photoproducts.Direct reduction of NO-SMX to SMX and Ac-SMX is

chemically improbable, therefore both are likely at least secondgeneration products of NO-SMX photolysis. The contributionof buffer components to the unexpected observation of Ac-SMX could be ruled out, as Ac-SMX was systematicallyobserved with various types of buffers, only one containing anacetylated derivative (acetic acid). Moreover, a second orderreaction involving NO-SMX itself could be excluded asdoubling the initial concentration of NO-SMX did no morethan double the production of Ac-SMX.The largest fraction of NO-SMX was transformed to the

most stable metabolite, NO2−SMX (m/z 282 in Figure 2c),with a conversion of 55%. It seems that the NO-substituentfavors modifications of the substituent (NO2, NH2 andNHCOCH3) rather than cleavage reactions, which predomi-nated in SMX and Ac-SMX photolysis. As for SMX and Ac-SMX, (5-methylisoxazol-3-yl)sulfamate (m/z 99) was detectedafter 0.66 h of irradiation. Its emergence before SMX and Ac-SMX products (after 2 and 3h respectively), yet with a slightdelay on NO-SMX degradation, lead us to believe it is not afirst generation product; thus there is no evidence that γ-cleavage occurs directly on NO-SMX.Apart from the new products discussed above and previously

mentioned products from SMX and Ac-SMX, three newstructures can be proposed, and supported by in-sourcefragmentation information. The nitro-derived products, nitro-benzenesulfonic acid ([m − H]− = 201.9812) and hydroxylatednitro-SMX ([m − H]− = 298.0143) elute at 12.0 and 22.2 min,respectively. Furthermore, a product likely corresponding toaddition of a methyl group on the aromatic ring of NO-SMX([m − H]− = 280.0401) was observed at 14.7 min. Allconfirmed products, as well as proposed structures, are shownin Table S13, SI.A comparison of the conversion rates illustrates that the

different ring substituents on the metabolites not only affect the



direct photolysis reaction rate, but also the principal photolysispathways. Among the confirmed products, cleavage reactionswere predominant for SMX and Ac-SMX, whereas the nitro-group was very resistant to phototransformation and thenitroso group mainly underwent modifications of thesubstituent (Figure 3). Despite the nitro-group being a verypotent chromophore, nitroaromatics are known to have veryshort triplet lifetimes;32 this may explain the low reactivity ofNO2−SMX. Irradiation experiments with nitrobenzene effec-tively found it to be photostable under natural sunlight.33 Thephotochemistry of nitroso-aromatics has also received muchattention, and some have been shown to oxidize photochemi-cally to nitro-compounds.34,35 Several reaction schemes,involving either C−N bond dissociation or excitation to thefirst singlet state, followed by intersystem crossing wereproposed, but none could be ascertained.35 Whatever thepathway, these observations are in line with the photooxidationof NO-SMX to NO2-SMX observed here. The transformationof NO-SMX to SMX constitutes a photoreduction, and thusrequires the presence of an electron donor. In our experimentalsystem, trace amounts of methanol may have played this role.In environmental systems, potential electron donors includeorganic matter or chloride, though their effect on thephotoreduction of NO-SMX was not specifically addressedherein.The main products of SMX and Ac-SMX resulted from δ-

and γ-cleavage. The acetyl-group may have a stabilizing effecton the C−S bond as δ-cleavage was predominant, while bothN−S and C−S bonds seemed equally susceptible to cleavageduring SMX-photolysis. β-cleavage, involving cleavage of thesubstituent, was observed for all target compounds, but was not

quantified due to the lack of analytical standard. Accuratemasses corresponding to the desulfonation-products of SMXand its metabolites were observed for every compound.Extrusion of SO2 was already identified as an important directphotolysis pathway for similar sulfa-drugs (sulfapyridine andsulfmethazine) and their acetylated metabolites.31 The relativeimportance of this pathway could not be quantified here due tolack of standards.Finally it may be worth recalling that nitroso- and nitro-

substituted compounds already showed opposite behavior toSMX and Ac-SMX regarding the correlation between spectraloverlap and reaction rate constant. Thus it seems that thenitro-/nitroso- group affects the photoexcitation and resultingpathways of sulfonamides. Evidently, further research usingmodel compounds would be required to fully elucidate themechanisms.

Contribution of Direct Photolysis to the Degradationof SMX Metabolites in Lake Geneva. Water samples fromLake Geneva were investigated to confirm the presence ofhuman metabolites of SMX in the environment, and to analyzeobserved concentrations with respect to their excretionfractions and relative photolability. Over 70 lake water samplesfrom 2 sampling sites and various depths (0−30 m) in LakeGeneva were analyzed for the presence of the two mostabundantly excreted metabolites of SMX, Ac-SMX, and SMX-glucuronide. Both metabolites were detected, but theirfrequency of detection varied with compound, sampling site,and season. SMX-glucuronide was only detected at a lowfrequency (9% of samples). The observed concentrations weresystematically and significantly lower than SMX (Figure 4),despite their comparable excretion fractions. This is consistent

Figure 3. Photolyzed substances and structures of confirmed photoproducts. Formulas are given for proposed products which were not validatedwith authentic standards; proposed structures for these formulas are given in SI Tables S10, S11, S12, and S13.



with our finding that SMX-glucuronide is the most photolabileof all compounds targeted in this study. Given the rapidphotolysis, combined with the unstable nature of theglucuronide bond, which can also be microbially cleaved,23 alow frequency of detection and low environmental concen-trations are expected.Ac-SMX, on the other hand, is excreted in higher amounts

than SMX but nevertheless was detected at a lower frequency(Ac-SMX = 43%, SMX = 94%). Furthermore, the concen-trations of Ac-SMX detected in the lake were significantly lowerthan SMX. The present study found Ac-SMX to be slightlymore resistant to direct photolysis than SMX; hence, the lowerconcentrations observed for Ac-SMX in lake water may indicatethat other processes, such as indirect photolysis processes andbiodegradation may also contribute to the total degradation inthe investigated environment. In fact, previous research showedthat Ac-SMX biodegrades twice as fast as SMX in water-sediment tests.23 Considering lake parameters and pastresearch,11 we expect a minor contribution of indirectphotolysis processes to the total degradation of SMX. However,the susceptibility of human metabolites of SMX to indirectprocesses is still to be determined.Environmental Relevance. This study observed an abiotic

production of SMX from NO-SMX. In the present case, owingto the low fraction SMX excreted as NO-SMX, the lowconversion back to the parent compound and the relativelyhigh photolability of SMX, the back-transformation of NO-SMX to SMX may not be of high environmental relevance,though actual environmental concentrations of NO-SMX inlake water would be necessary for a better assessment.Nevertheless, this is the first evidence that photolytic back-transformation of a human metabolite to its parent compoundcan occur, which in essence presents an environmental sourceof SMX. Likewise, it is pertinent to recall that NO2−SMX, themain photoproduct of NO-SMX photolysis, was also found toretransform to SMX under reducing conditions,24 thusrepresenting an additional abiotic source of SMX. As such,further studies should be pursued to investigate the back-transformation of NO-SMX yield under different environ-mental conditions, where the process may be more significant.Moreover, the back-transformation of other nitroso-metabolitesshould also be considered. Furthermore, a recent studyeffectively found a potential increase of the ecotoxicological

risk due to the presence of, among others, SMX metabolites insurface waters.27 The knowledge that some metabolites retaincertain biological activity, combined with their potential toretransform to the parent compound, underlines theimportance of considering human metabolites in environmentalfield work and including product identification in degradationstudies.

! ASSOCIATED CONTENT*S Supporting InformationUV−vis spectrum of solar simulator, 300 nm bulb (Rayonet)and simulated solar spectrum; absorbance versus pH for pKadetermination; molar absorptivity at experimental pH; directphotolysis kinetics; spectral overlap of compound absorbanceand irradiance; analytical methods; direct photolysis calcu-lations; error calculations; synthesis procedure of sodium (5-methylisoxazol-3-yl)sulfamate, X-ray analysis and confirmation;confirmed photoproducts and proposed structures for SMXand metabolites. This material is available free of charge via theInternet at http://pubs.acs.org.

! AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] authors declare no competing financial interest.

! ACKNOWLEDGMENTSWe thank Paul Erickson, Rachel Lundeen, Dr. Sarah Page, andDr. Sarah Kliegman (all from ETH Zurich) for help withlaboratory equipment and experimental procedures. Prof.Charles Sharpless (University of Mary Washington) isgratefully acknowledged for helpful discussions. This projectwas funded by the Swiss National Science Foundation (Projectno. PDFMP2-123028/1) and is part of an interdisciplinaryproject (www.leman21.ch).

! REFERENCES(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 U.S. streams, 1999−2000: A national reconnaissance. Environ. Sci. Technol. 2002, 36 (6),1202−1211.(2) Bonvin, F.; Rutler, R.; Chevre, N.; Halder, J.; Kohn, T. Spatialand temporal presence of a wastewater-derived micropollutant plumein Lake Geneva. Environ. Sci. Technol. 2011, 45 (11), 4702−4709.(3) Gobel, A.; Thomsen, A.; McArdell, C. S.; Joss, A.; Giger, W.Occurrence and sorption behavior of sulfonamides, macrolides, andtrimethoprim in activated sludge treatment. Environ. Sci. Technol. 2005,39 (11), 3981−3989.(4) Garcia-Galan, M. J.; Diaz-Cruz, M. S.; Barcelo, D. Occurrence ofsulfonamide residues along the Ebro river basin: Removal inwastewater treatment plants and environmental impact assessment.Environ. Int 2011, 37 (2), 462−473.(5) Morasch, B.; Bonvin, F.; Reiser, H.; Grandjean, D.; de Alencastro,L. F.; Perazzolo, C.; Chevre, N.; Kohn, T. Occurence and fate ofmicropollutants in the Vidy Bay of Lake Geneva, Switzerland. Part II:Micropollutant removal between wastwater and raw drinking water.Environ. Toxicol. Chem. 2010, 29 (8), 1658−1668.(6) Miao, X.-S.; Bishay, F.; Chen, M.; Metcalfe, C. D. Occurrence ofAntimicrobials in the Final Effluents of Wastewater Treatment Plantsin Canada. Environ. Sci. Technol. 2004, 38 (13), 3533−3541.(7) Oulton, R. L.; Kohn, T.; Cwiertny, D. M. Pharmaceuticals andpersonal care products in effluent matrices: A survey of transformation

Figure 4. Measured concentrations (ng·L−1) of SMX and its humanmetabolites SMX-glucuronide and Ac-SMX in Lake Geneva. f =frequency of detection. Boxes represent 25th percentile, median and75th percentile. The whiskers are determined by the 5th and 95thpercentiles and the stars show the mean value.



http://pubs.acs.org

mailto:[email protected]

www.leman21.ch

and removal during wastewater treatment and implications forwastewater management. J. Environ. Monit. 2010, 12 (11), 1956−1978.(8) Mompelat, S.; Le Bot, B.; Thomas, O. Occurrence and fate ofpharmaceutical products and by-products, from resource to drinkingwater. Environ. Int. 2009, 35 (5), 803−814.(9) Hirsch, R.; Ternes, T.; Haberer, K.; Kratz, K.-L. Occurrence ofantibiotics in the aquatic environment. Sci. Total Environ. 1999, 225(1−2), 109−118.(10) Boreen, A. L.; Arnold, W. A.; McNeill, K. Photodegradation ofpharmaceuticals in the aquatic environment: A review. Aquat. Sci.2003, 65 (4), 320−341.(11) Boreen, A. L.; Arnold, W. A.; McNeill, K. Photochemical fate ofsulfa drugs in the aquatic environment: Sulfa drugs containing five-membered heterocyclic groups. Environ. Sci. Technol. 2004, 38 (14),3933−3940.(12) Ryan, C. C.; Tan, D. T.; Arnold, W. A. Direct and indirectphotolysis of sulfamethoxazole and trimethoprim in wastewatertreatment plant effluent. Water Res. 2011, 45 (3), 1280−1286.(13) Zhou, W.; Moore, D. E. Photochemical decomposition ofsulfamethoxazole. Int. J. Pharm. 1994, 110 (1), 55−63.(14) Andreozzi, R.; Raffaele, M.; Nicklas, P. Pharmaceuticals in STPeffluents and their solar photodegradation in aquatic environment.Chemosphere 2003, 50 (10), 1319−1330.(15) Lam, M. W.; Mabury, S. A. Photodegradation of thepharmaceuticals atorvastatin, carbamazepine, levofloxacin, and sulfa-methoxazole in natural waters. Aquat. Sci. 2005, 67 (2), 177−188.(16) Trovo, A. G.; Nogueira, R. F. P.; Aguera, A.; Fernandez-Alba, A.R.; Sirtori, C.; Malato, S. Degradation of sulfamethoxazole in water bysolar photo-Fenton. Chemical and toxicological evaluation. Water Res.2009, 43 (16), 3922−3931.(17) Gomez, M. J.; Sirtori, C.; Mezcua, M.; Fernandez-Alba, A. R.;Aguera, A. Photodegradation study of three dipyrone metabolites invarious water systems: Identification and toxicity of their photo-degradation products. Water Res. 2008, 42 (10−11), 2698−2706.(18) Celiz, M. D.; Tso, J.; Aga, D. S. Pharmaceutical metabolites inthe environement: Analytical challenges and ecological risks. Environ.Toxicol. Chem. 2009, 28 (12), 2473−2484.(19) Vanderven, A.; Vree, T. B.; Kolmer, E.; Koopmans, P. P.;Vandermeer, J. W. M. Urinary recovery and kinetics of sulfamethox-azole and its metabolites in HIV-seropositive patients and healthy-volunteers after a single oral dose of sulfamethoxazole. Br. J. Clin.Pharmacol. 1995, 39 (6), 621−625.(20) Naisbitt, D. J.; Hough, S. J.; Gill, H. J.; Pirmohamed, M.;Kitteringham, N. R.; Park, B. K. Cellular disposition of sulphamethox-azole and its metabolites: implications for hypersensitivity. Br. J.Pharmacol. 1999, 126 (6), 1393−1407.(21) Gobel, A.; McArdell, C. S.; Suter, M. J. F.; Giger, W. Tracedetermination of macrolide and sulfonamide antimicrobials, a humansulfonamide metabolite, and trimethoprim in wastewater using liquidchromatography coupled to electrospray tandem mass spectrometry.Anal. Chem. 2004, 76 (16), 4756−4764.(22) Ashton, D.; Hilton, M.; Thomas, K. V. Investigating theenvironmental transport of human pharmaceuticals to streams in theUnited Kingdom. Sci. Total Environ. 2004, 333 (1−3), 167−184.(23) Radke, M.; Lauwigi, C.; Heinkele, G.; Murdter, T. E.; Letzel, M.Fate of the Antibiotic Sulfamethoxazole and Its Two Major HumanMetabolites in a Water Sediment Test. Environ. Sci. Technol. 2009, 43(9), 3135−3141.(24) Noedler, K.; Licha, T.; Barbieri, M.; Perez, S. Evidence for themicrobially mediated abiotic formation of reversible and non-reversiblesulfamethoxazole transformation products during denitrification.Water Res. 2012, 46 (7), 2131−2139.(25) Spillane, W. J.; Kelly, L. M.; Feeney, B. G.; Drew, M. G. B.;Hattotuwagama, C. K. Synthesis of heterosulfamates. Search forstructure-taste relationships. Arkivoc 2003, 297−309.(26) Leifer, A. The Kinetics of Environmental Aquatic Photochemistry:Theory and Practice: American Chemical Society: Washington, DC,1988.

(27) Bonvin, F.; Chevre, N.; Rutler, R.; Kohn, T. Pharmaceuticalsand their human metabolites in Lake Geneva: occurrence, fate andecotoxicological relevance. Arch. Sci. 2012, in press.(28) Wols, B. A.; Hofman-Caris, C. H. M. Review of photochemicalreaction constants of organic micropollutants required for UVadvanced oxidation processes in water. Water Res. 2012, 46 (9),2815−2827.(29) Fery-Forgues, S.; Lavabre, D. Are fluorescence quantum yieldsso tricky to measure? A demonstration using familiar stationeryproducts. J. Chem. Educ. 1999, 76 (9), 1260.(30) Trovo, A. G.; Nogueira, R. F. P.; Aguera, A.; Sirtori, C.;Fernandez-Alba, A. R. Photodegradation of sulfamethoxazole invarious aqueous media: Persistence, toxicity and photoproductsassessment. Chemosphere 2009, 77 (10), 1292−1298.(31) Jesus Garcia-Galan, M.; Silvia Diaz-Cruz, M.; Barcelo, D. Kineticstudies and characterization of photolytic products of sulfamethazine,sulfapyridine and their acetylated metabolites in water under simulatedsolar irradiation. Water Res. 2012, 46 (3), 711−722.(32) Dopp, D. Reactions of aromatic nitro compounds via excitedtriplet states. Top. Curr. Chem. 1975, 55, 49−85.(33) Vialaton, D.; Richard, C. Phototransformation of aromaticpollutants in solar light: Photolysis versus photosensitized reactionsunder natural water conditions. Aquat. Sci. 2002, 64 (2), 207−215.(34) Tanikaga, R. Photolysis of nitrosobenzene. Bull. Chem. Soc. Jpn.1969, 42 (1), 210−&.(35) Zelentsov, S. V.; Logunov, A. A. A study on the photochemicaltransformations of p- and m-nitrosonitrobenzenes. High Energy Chem.2005, 39 (4), 244−249.



direct photolysis of human metabolites of the antibiotic sulfamethoxazole: evidence for abiotic...

Documents