直接光化学降解.pdf

Direct photochemistry of three fluoroquinolone antibacterials:Norfloxacin,ofloxacin,and enrofloxacinKristine H.Wammera,*,Andrew R.Kortea,Rachel A.Lundeena,Jacob E.Sundberga,Kristopher McNeillb,William A.ArnoldcaDepartment of Chemistry,University of St.Thomas,2115 Summit Ave.,St.Paul,MN 55105,USAbInstitute for Biogeochemistry and Pollutant Dynamics,ETH Zurich,8092 Zurich,SwitzerlandcDepartment of Civil Engineering,University of Minnesota,500 Pillsbury Dr.SE,Minneapolis,MN 55455,USAa r t i c l e i n f oArticle history:Received 8 June 2012Received in revised form12 October 2012Accepted 13 October 2012Available online 22 October 2012Keywords:FluoroquinolonesQuantum yieldPhotolysisAntibacterial activitya b s t r a c tFluoroquinolone(FQ)antibacterial compounds are frequently detected in the aquaticenvironment,and photodegradation is expected to play an important role in FQ fate insome sunlit surface waters.This study investigated the direct aquatic photochemistry ofthree FQs:norfloxacin,ofloxacin,and enrofloxacin.The direct photolysis rate of each drugexhibited strong pH dependence when exposed to simulated sunlight.For each FQ,directphotolysis rates and total light absorbance were used to calculate quantum yields for eachof three environmentally relevant protonation states:a cationic,a zwitterionic,and ananionic form.In each case,quantum yields of the species varied significantly.Thequantum yield for the zwitterionic form was 2e3 times higher than that of the anionic formand over an order of magnitude higher than that of the cationic form.Antibacterial activityassays were used to determine whether the loss of parent FQ due to photolysis led to loss ofactivity.Norfloxacin and ofloxacin photoproducts were found to be inactive,whereasenrofloxacin photoproducts were found to retain significant activity.These results areimportant for aiding in predictions of the potential impacts of FQs in surface waters.2012 Elsevier Ltd.All rights reserved.1.IntroductionFluoroquinolones(FQs)are a class of antibacterial compoundsused extensively in both human and veterinary medicine.FQshave been found in hospital wastewaters(at concentrationsranging from approximately 60e120,000 ng/L),in wastewatertreatment plant(WWTP)effluents(w2e580 ng/L)and insurfacewaters(w5e1300ng/L)throughouttheworld,including in the United States(Kolpin et al.,2002;Renew andHuang,2004),Italy(Andreozzi et al.,2003),Switzerland(Giger et al.,2003;Golet et al.,2003),Finland(Vieno et al.,2007),Sweden(Lindberg et al.,2006),Germany(Hartmann et al.,1999),China(Xu et al.,2009),and Australia(Watkinson et al.,2009).Research about which processes are most likely to controlthe environmental fate of FQs in aquatic environments isneededto assesspotential long-termimpacts.The presence ofFQs and other antibacterial compounds in the environment isof concern primarily due to the possibility that long-termexposure to sub-therapeutic doses may provide selectivepressure for antibiotic-resistant organisms(Ku mmerer,2009).There is also concern that FQs may inhibit photosynthesis inplants(Aristilde et al.,2010).Previous work has shown thatphotodegradation is expected to play an important role in FQfate in some sunlit surface waters(Schmitt-Kopplin et al.,1999;Andreozzi et al.,2003;Knapp et al.,2005;Lam andMabury,2005;Sturini et al.,2010;Li et al.,2011),and inparticularmaybesignificantrelativetootherabiotic*Corresponding author.Tel.:[phone];fax:[phone].E-mail address:khwammerstthomas.edu(K.H.Wammer).Available online at journal homepage: research 47(2013)439e4480043-1354/$e see front matter 2012 Elsevier Ltd.All rights reserved.http:/dx.doi.org/10.1016/j.watres.2012.10.025processes,such as sorption,in waters with low particulateorganic carbon(POC)(Cardoza et al.,2005).A recent studyfocused on the photochemical fate of eight FQs in differentnatural waters(Ge et al.,2010).In the study,photolysis rates ofthe two drugs studied in detail(sarafloxacin and gatifloxacin)exhibited a strong pH dependence,which must be due tovariations in absorption spectra and/or quantum yields forspecies with different protonation states.Quantifying thesevariations would aid in the prediction of FQ degradation ratesin a range of natural waters.In this work,the direct photochemical behavior of threeFQs was investigated:norfloxacin(NOR),ofloxacin(OFL),andenrofloxacin(ENR).NOR and OFL are used primarily in humanmedicine and ENR is used in veterinary medicine.All of theseFQs have been detected in WWTP effluents and/or naturalwaters(e.g.Kolpin et al.,2002;Andreozzi et al.,2003;Gigeret al.,2003;Renew and Huang,2004;Watkinson et al.,2009).Direct photolysis degradation rates and total light absorbancewere used to calculate quantum yields for each of threeenvironmentally relevant protonation states for each drug.Thisanalysisisnecessaryforaccurateprediction ofphotolysisrates over the range of pH values(typically between 6 and 9)found in natural waters.In addition,bacterial assays wereused to determine whether loss of parent FQ due to directphotolysiscorrelateswithlossofantibacterialactivitybecause any FQ intermediates or photoproducts retaining theability to inhibit bacterial growth could potentially have long-term effects on antibiotic resistance.Selected photoproductswere isolated,characterized,and assessed for activity.2.Material and methods2.1.Reagents and materialsENR(98%),NOR(98%),and ciprofloxacin(CIP,98%)werepurchased from SigmaeAldrich.OFL(99%)was purchasedfrom MP Biomedicals Inc.Pyridine(99%),p-nitroanisole(PNA;99%),and Iso-Sensitest bacterial growth medium weresupplied by Thermo Fisher Scientific.Tryptone and yeastextractwerepurchasedfromBecton,DickinsonandCompany.All chemicals were used as received.Solvents wereofHPLC grade.Deionizedwaterwas purifiedusinga BarnsteadNanopure system.2.2.Natural water photolysisStock solutions of each FQ in deionized water(100 mM)werediluted to 10 mM using deionized water or filtered Lake Jose-phine water(St.Paul,MN,DOC w6 mg/L,filtered through0.2 mm filters).Lake Josephine water solutions of each drugwere photolyzed alongside deionized water solutions.Thedeionized water samples were adjusted to a pH similar to thatof the natural water samples(7.8e7.9)using small amounts(approximately 1e10 mL)of NaOH or HCl.Solutions forphotolysis were placed in 13 mm?100 mm quartz test tubesin a rack at an angle of approximately 30?to the horizontal.Photolyses were conducted in an Atlas Suntest CPS solarsimulator equipped with a xenon lamp and an Atlas UVSuntest filter,to provide an emission spectrum accuratelysimulatingthatof naturalsunlightand anirradianceof 765 W/m2(see Fig.S1).At pre-determined time intervals,aliquotswere removed from the tubes and stored in the dark untilanalysis.The pH of each sample was measured before andafter each photolysis experiment to ensure no significantchanges over the course of the experiment.2.3.HPLC analysesHigh performanceliquid chromatography(HPLC)analysis wasconducted using an Agilent 1100 Series chromatograph withan ultravioletevisible(UVeVis)diode array detector.Allsamples were adjusted to similar pH by addition of a smallamount of strong base(6 M NaOH)prior to analysis.ThecolumnusedwasaSupelcoDiscoveryRPAmideC16(100 mm?4.6 mm,5 mm particle size).The mobile phase wasan isocratic 85:15 mixture of pH 3 phosphate buffer(8.7 mM):acetonitrile at a 1.0 ml per minute flow rate.NOR and ENRwere detected at 271 nm while OFL was detected at 286 nm.2.4.pKaand quantum yield determinationThe pH of 50 mM solutions of individual FQs in deionized waterwas adjusted to a range of values(approximately 3e11)byaddition of small amounts(approximately 1e10 mL)of NaOHor HCl.UVeVis absorbance spectra were obtained of eachsolution on a Hewlett Packard 8452A spectrophotometer touse for determination of pKavalues.Similar pH-adjustedsolutions of each FQ(10 mM)were used for quantum yielddetermination.Approximately 2 mL of each solution wasremoved and used to obtain UVeVis spectra,which were thendeconvoluted according to the method of Boreen et al.(2004)to obtain molar absorptivities for the three relevant proton-ation species of each drug.The remaining sample was thenirradiated in the solar simulator and aliquots were removed atpre-determined time intervals and analyzed by HPLC asdescribed above.Photolyses were conducted alongside a PNA/pyridine actinometer solution prepared according to Leiferwith a quantum yield of 0.00465(Leifer,1988).2.5.Antibacterial activity screeningThe ability of each FQ to inhibit growth of Escherichia coli DH5awas determined as described previously(Wammer et al.,2006).E.coli were grown overnight on Iso-Sensitest broth(ISB,made by adding 23.4 g of ISB powder per liter of pH 7phosphate buffer(4.9 g KH2PO4and 9.7 g Na2HPO4in deionizedwater)and sterilized by autoclaving(20 min;121?C;15 psig).Aliquots of E.coli(100 mL)and the appropriate volume ofunirradiated 100 mM FQ solution were added to test tubescontaining 9 mL of sterile ISB to obtain a range of FQconcentrations varying from approximately 0.01e10 mM.Testtubes were incubated in the dark for six(ENR)or eight(NOR,OFL)hours(37?C,200 rpm).Bacterial growth was assessed bymeasuring optical density at 600 nm and comparing to initialoptical density of each solution(final OD600einitial OD600).To assess the ability of FQ photoproduct mixtures to inhibitE.coli growth,FQ solutions(100 mM)were photolyzed asdescribed above and samples were collected over time.Pho-tolysate(photoproductmixtureplusremainingparentwater research 47(2013)439e448440fluoroquinolone,1 ml)from each time point was added to thetest tubes in place of the 1 ml of unirradiated FQ.The photo-lysate samples corresponding to each time point wereanalyzed by HPLC to determine the amount of parent FQremaining.The antibacterial activity of selected isolatedphotoproductsorgroupsofphotoproducts(isolatedbypreparative HPLC,see below)was assessed in a similarmanner with the exception that,without an authentic stan-dard forunknown photoproducts,concentration ofthephotoproducts could not be quantified by HPLC.Only a quali-tative assessment of whether activity was observed could bemade.All bioassays were performed in triplicate.2.6.Selected Photoproduct characterizationSolutions of ENR(500 mM)in deionized water were irradiatedoutdoors(under summer sunlight in St.Paul,MN,45?N lati-tude)or in the solar simulator for a long enough period of time(approximately one half life)to generate significant quantitiesof major photoproducts.Because this set of experiments wasnot used to generate data about photolysis rates,it was not ofconcern that these solutions were not optically dilute.Following photolysis,the resultant mixtures of products andresidual parent compound were concentrated via rotaryevaporation.Selected photoproducts(see Section 3.5 fordetails)were isolated by preparative HPLC,using a SupelcoDiscovery RP Amide C16column(100 mm?21.2 mm,5 mmparticle size),a 5 mL per minute flow rate,an isocratic 85:15mixture of pH 3 formic acid buffer:acetonitrile,and the samedetectionasdescribedabove.Fractionswerecollected,concentrated again,and stored in the dark to prevent degra-dation of photolabile products.High-resolution mass spectraof each isolated product were obtained on a Bruker BioTOF IIinstrument equipped with an ESI source using polyethyleneglycol(PEG)as an internal reference standard.Spectra wereobtained under positive ionization mode.A JEOL 400 MHznuclear magnetic resonance(NMR)spectrometer was used toobtain a fluorine NMR spectrum of one isolated ENR product.3.Results and discussion3.1.Natural water photolysisPrevious work has shown that direct photolysis,rather thanindirect photolysis involving interactions with natural waterconstituents,islikelytobethedominantphotolysisprocessforFQs in most natural waters(Schmitt-Kopplin et al.,1999;Cardoza et al.,2005;Lam and Mabury,2005;Prabhakaran et al.,2009;Li et al.,2011)with the exception of one study in whichenhanced OFL degradation was observed in the presence ofhumic acids and nitrate(Andreozzi et al.,2003).Most compre-hensively,Ge et al.(2010)used a central composite design todetermineindividualeffectsofvariousfreshwaterandseawaterconstituentsandfoundthatseveral(i.e.,humicacids,Fe3,and NO3?)inhibited rather than enhanced FQ photolysisrates.Lietal.(2011)observedself-sensitizedphotolysisforENR,but this was at significantly higher concentrations(10 mg/L)than would be expected in environmentally relevant samples;self-sensitization is unlikely at the concentrations typicallyobserved in natural waters.In the present study,the three FQs of interest were irradi-ated in a local natural water sample(Lake Josephine)simulta-neously alongside deionized water samples adjusted to thesame pH(all at pH 7.8e7.9).Photolysis was rapid for all threedrugs(t1/2 1e4 min)and followed apparent first-orderkinetics(Fig.1).Any significant indirect photochemical path-ways(e.g.reaction with excited state dissolved organic matter(DOM)or with photochemically produced reactive intermedi-ates such as singlet oxygen(1O2)would have been seen asenhanced photodegradation in the Lake Josephine water.Thiswas not observed.The slightly slower degradation rate in thenaturalwatersamplesmaybeattributabletolightscreeningbyDOM,or the DOM may be preventing oxidation by quenchinganexcitedtripletstate(Geetal.,2010;Lietal.,2011;Wenketal.,2011).Because direct photolysis will clearly be an importantprocess,if not the primary photodegradation pathway in mostnatural waters,all subsequent experiments focused on directphotolysis and were conducted in deionized water.3.2.pKadeterminationTable 1 shows the dominant protonation state of each of thethreeFQsatneutralpH,whichisazwitterionicform.Therearetwo relevant acidebase processes for FQs within the range ofpH values encountered in natural waters.The first(pKa1)involves the proton on the carboxylic acid group while thesecond(pKa2)applies to the external amino group on thepiperazine ring(Lizondo et al.,1997;Ross and Riley,1992).Atmore acidic pH values,the most prevalent form of thesemolecules will be cationic and at more basic pH values ananionic form will be dominant.Apparent macroscopic disso-ciation constants corresponding to pKa1and pKa2in the liter-ature are somewhat variable.For example,reported pKa1valuesforENRrangefromapproximately5.8to 6.3(Goletetal.,2003;Schmitt-Kopplinetal.,1999;Lizondoetal.,1997;Jime nez-Lozano et al.,2002;Qiang and Adams,2004).The pKavalues,therefore,were measured directly by spectrophotometrictitration rather than attempting to select appropriate valuesfrom the literature.The absorbance spectra are shown in the SI(Figs.S2eS4).The absorbance at a single wavelength(282 nm for NOR,292nmforOFL,and336nmforENR)wasplottedagainstpHandpKa values were determined by fitting the data to Equation(1):?cFH2?AFH2l?cFHAFHl cF?AF?lhAtot(1)wherecreferstothefractionofthecationic(FH2),zwitterionic(FH),and anionic(F?)species present,and A is absorbance ataselectedwavelength.ThisfitisalsoshowninFigs.S2eS4.Dueto lac