Ion trapping of amines in protozoa – a novel removal ... · 1 1 Ion trapping of amines in protozoa – a novel removal 2 mechanism for micropollutants in activated sludge 3 4 Rebekka

1

Ion trapping of amines in protozoa – a novel removal1

mechanismformicropollutantsinactivatedsludge2

3

RebekkaGuldea,SabineAnlikera,b,Hans-PeterKohlera,b,KathrinFenner*a,b,c4

5aEawag,SwissFederalInstituteofAquaticScienceandTechnology,8600Dübendorf,Switzerland6bInstituteofBiogeochemistryandPollutantDynamics,ETHZürich,8092Zürich,Switzerland7cDepartmentofChemistry,UniversityofZürich,8057Zürich,Switzerland8

9

*Correspondingauthor:[email protected],Tel.:+4158765508510

11

Wordcount:601312

Tables(2small)+Figures(2small,1large):180013

Totalwordcount:781314

15

16

Abstract17

Tooptimizeremovaloforganicmicropollutants fromthewatercycle,understandingtheprocesses18

during activated sludge treatment is essential. In this study,wehypothesize that aliphatic amines,19

which are highly abundant amongst organic micropollutants, are partly removed from the water20

This document is the accepted manuscript version of the following article:Gulde, R., Anliker, S., Kohler, H. -P. E., & Fenner, K. (2018). Ion trapping of amines in protozoa: a novel removal mechanism for micropollutants in activated sludge. Environmental Science and Technology, 52(1), 52-60. http://doi.org/10.1021/acs.est.7b03556

2

phase in activated sludge through ion trapping in protozoa. In ion trapping, which has been21

extensively investigated in medical research, the neutral species of amine-containing compounds22

diffusethroughthecellmembraneandfurtherintoacidicvesiclespresentineukaryoticcellssuchas23

protozoa.Theretheybecometrappedbecausediffusionofthepositivelychargedspeciesformedin24

the acidic vesicles is strongly hindered.We testedour hypothesiswith two experiments. First,we25

studied the distribution of the fluorescent amine acridine orange in activated sludge by confocal26

fluorescence imaging.Weobserved intensefluorescence indistinctcompartmentsoftheprotozoa,27

butnotinthebacterialbiomass.Second,weinvestigatedthedistributionoftwelveamine-containing28

and eight controlmicropollutants in both regular activated sludge and sludgewhere theprotozoa29

hadbeeninactivated.Incontrasttomostcontrolcompounds,theamine-containingmicropollutants30

displayedadistinctlydifferentbehaviorinthenon-inhibitedsludgecomparedtotheinhibitedone:i)31

moreremovalfromtheliquidphase;ii)deviationfromfirst-orderkineticsfortheremovalfromthe32

liquidphase;and iii)higheramounts inthesolidphase.Theseresultsprovidestrongevidencethat33

iontrappinginprotozoaoccursandthatitisanimportantremovalmechanismforamine-containing34

micropollutants in batch experiments with activated sludge that has so far gone unnoticed. We35

expect thatour findingswill trigger further investigationsonthe importanceof thisprocess in full-36

scalewastewatertreatmentsystems,includingitsrelevanceforaccumulationofammonium.37

38

Introduction39

Manycompoundsineverydayuse,suchaspharmaceuticals,personalcareproducts,surfactantsand40

biocides,areconveyedbysanitarysewerstowastewatertreatmentplants(WWTPs),wheretheyare41

removedtodifferentextents.1Thetreatmentstagemainlyresponsibleforremovaloftheseso-called42

organicmicropollutants (MPs) inWWTPs is activated sludge treatment, duringwhichMPsmaybe43

removedbydifferentprocesses, includingabiotic transformation, sorption to the sludge flocs, and44

microbial biotransformation.1-4 Different classes of wastewater-relevant MPs contain structural45

motifs that include a basic functional group. Particularly among the active ingredients of46

pharmaceuticals, basic functional groups have been reported to be present in about 40% of the47

structuresandtobedistributedacrossmanytherapeuticclasses.5, 6Amongthese,aliphaticamines48

withaciddissociationconstants (pKa) in the rangeof7 to10aremostabundant.5Their speciation49

changesatenvironmentallyrelevantpHvalues,suchthatunderratheracidicpHconditionsaliphatic50

amines are typically protonated and hence cationic, while they become deprotonated and hence51

neutral at higherpH. In aprevious study,we showed that, due to their basicity, amine-containing52

MPsexperiencepH-dependentbiotransformationinactivatedsludgecommunities.7Specifically,we53

observedthatremovalofaminesfromtheliquidphaseproceededfasterathigherpHvalues,which54

wasexplainedbythefactthattheneutralspeciescanpassivelydiffusethroughthecellmembranes,55

4

byenergy-dependentprotonpumps.11Asaconsequence,theamine-containingcompoundsbecome82

positively charged, and, since the charged form is hindered from diffusing back into the cytosol,83

becomeeffectivelytrappedinsidethevesicles.Diffusionintothevesiclescontinuesuntilequilibrium84

of the neutral species between the extracellular environment, the cytosols and the vesicles is85

reached. Due to the considerable pH difference between the extracellular environment and the86

vesicles, this can lead to several orders of magnitude difference in total amine concentrations87

between the extracellular environment and the vesicles.12 Consistent with this mechanism, ion88

trapping has been described to increase with increasing pKa and lipophilicity of the drugs and at89

elevatedextracellularpH.12,15Iontrappinginacidicvesicleshasbeendescribedacrossawiderange90

ofeukaryoticcells,fromyeasttoanimalcells.12,13Inmedicalresearch,iontrappingisevenusedfor91

staining acidic vesicles with fluorescent amines such as LysoTracker red, quinacrine, and acridine92

orange16, which can subsequently be detected by means of confocal fluorescence imaging. 6, 1693

Especiallyacridineorangeisahighlyversatiledye,whichevenallowsdifferentiatingbetweenacidic94

vesicleswithdifferentpH.Uponexcitation,itemitsgreenlightatlowconcentrationsand,duetothe95

formation of stacks of molecules, red light at high concentrations. 16 It is noteworthy that ion96

trapping goes beyond the effect of extracellular pH on bacterial toxicity that has previously been97

described for speciating chemicals, including ammonia. 17-19 Themagnitude of this latter effect is98

typicallysufficientlyexplainedbythepH-differencesbetweenextra-and intracellularpH17without99

theneedtoinvoketrappinginacidicvesicles,whichareabsentfrommostbacteriaanyway.100

At this point, we hypothesized that protozoa, whichmake up about 10% of the activated sludge101

biomass,20 could be important for the fate of amine-containingMPs in sewage sludge treatment.102

Since protozoa are eukaryotes, they posses acidic vesicles which can trap amine-containing103

compounds. Indeed, it has previously been demonstrated through staining experiments that the104

protozoic ciliateTetrahymena thermophilawas able to accumulate the fluorescent amine acridine105

orange.21Protozoainactivatedsludgemainlybelongtooneoffivegroups,namelyamoeba,ciliates106

(free-swimmingandstalked),flagellates,suctoreansandrotifers,whicharemulti-celledorganisms.20107

Therefore, the goal of this study was to test the hypothesis that ionizable amine-containing108

compoundsare trapped in theacidicvesiclesof theprotozoiccommunityofactivatedsludge.This109

would constitute a previously unknown, additional removal process for amine-containing MPs in110

activated sludge. To test this hypothesis, we conducted two kinds of experiments. First, we111

investigated the accumulation of the fluorescent amine acridine orange in an activated sludge112

community by confocal fluorescence imaging. Second, we examined the distribution of non-113

fluorescent MPs in the liquid and solid phase of activated sludge and compared it to conditions114

wheretheprotozoawereinactivatedbytheadditionoftheinhibitordigitonin.20Theseexperiments115

5

wereconductedwith12amine-containingtargetMPsandeightcontrolMPswitheitherneutralor116

neutral-anionicspeciation.117

118

MaterialsandMethods119

Thefollowingisacompendiouspresentationofthematerialsandmethods;fulldetailsaregivenin120

theSupportingInformation(SI).121

MicropollutantSelection122

Altogether, experimentswere carriedoutwith20environmentally relevantMPs. The following12123

amine-containingMPsthatundergocationic-neutralspeciationwereselectedastargetcompounds124

(atenolol, ranitidine, venlafaxine, lidocaine, tramadol, levamisole, mexiletine, fenfluramine,125

citalopram,propranolol,mianserin, ticlopidine).ThreeMPsthatundergoneutral-anionicspeciation126

(sulfathiazole, naproxen, trinexapac-ethyl) and five MPs that remain predominately neutral127

(diethyltoluamide, alachlor, azoxystrobin, isoproturon, chlortoluron) were selected as control128

compounds.Abbreviations(ID),chemicalstructures,andpredictedpKavaluesarepresentedinTable129

1. Additionally, separate experimentswere conductedwith the fluorescent amine acridine orange130

(AO)alsolistedinTable1.131

132

Table1:CompoundID,CompoundName,Structure,ChargeStateofIonizedSpeciesintheRelevantpHRange133

(i.e.,pH4-8),andPredictedpKaValues.134

ID Name Structure ChargeStatestate

pKaa

SUL Sulfathiazole

anionic 6.9

NAP Naproxen

anionic 4.2

TRI Trinexapac-ethyl anionic 3.4

DET Diethyltoluamide

neutral

ALA Alachlor

neutral

AZO Azoxystrobin

neutral

ISO Isoproturon

neutral

CLT Chlortoluron neutral

ATE Atenolol

cationic 9.7

6

RAN Ranitidine

cationic 7.8

VEN Venlafaxine cationic 8.9

LID Lidocaine

cationic 7.8

TRA Tramadol cationic 9.2

LEV Levamisole

cationic 7.0

MEX Mexiletine

cationic 9.5

FEN Fenfluramine

cationic 10.2

CIT Citalopram

cationic 9.8

PRO Propranolol cationic 9.7

MIA Mianserin

cationic 6.9

TIC Ticlopidine

cationic 6.7

AO AcridineOrange

cationic 8.2

apKavaluesaspredictedby22135136

Experimentswithacridineorange137

Reactors (100mLamber Schott bottles)were filledwith50mLactivated sludge sourced from the138

nitrificationbasinof a full-scale SwissWWTP (diluted to a total suspended solids concentrationof139

approximately 1 g/L) and shaken at 160 rpm on a circulating shaker table to ensure continuous140

mixing and aeration. Triplicate reactors were spiked with 60 µL AO solution (50 mg/L in141

methanol:water1:9),resultinginafinalconcentrationofabout60µg/L.Afteratimeperiodof26to142

29.5 h, samples were investigated with a Leica SP5 Laser Scanning Confocal Microscope (Leica,143

Heerbrugg, Switzerland)with an excitationwavelength of 458 nm and an emissionwavelength of144

480-560nmformonomersemittinggreenlightatlowconcentrations,and590-660nmforstacksof145

AOemittingredlightathighconcentrations.146

147

ExperimentswithselectedMPs148

In activated sludge, removal of MPs from the liquid phase can in principle be caused by abiotic149

transformation, sorption to sludge, ion trapping, andmicrobialbiotransformation. Ingeneral,both150

7

ion trapping and microbial biotransformation might involve bacterial and protozoic parts of the151

community.Totest thehypothesis thatprotozoic iontrapping isan importantremovalprocess for152

amines, the following experimental setupwas chosen. (i) Control experimentswere conducted to153

assess the magnitude of abiotic transformation and sorption processes (see below for details on154

these experiments). (ii) Experiments were carried out with the addition of digitonin. Digitonin155

inactivatestheprotozoaoftheactivatedsludgecommunitybyformingcomplexeswithcholesterol156

that are largeenough to inducepermanentholes ineukaryoticmembranes 23. Since cholesterol is157

present ineukaryoticbutnotbacterialcellmembranes,additionofdigitonin isexpectedto leadto158

selectivefunctionalinactivationorevendestructionofprotozoiccellsinactivatedsludge.Therefore,159

throughcomparisonoftheexperimentscarriedoutunderinhibitingandnon-inhibitingconditionsfor160

protozoa, the contribution of protozoic and bacterial processes, which could include both ion161

trappingandbiotransformation,couldbedifferentiated.Toverifythattheadditionofdigitonindid162

not affect other processes than the protozoic ones, experimental conditions (i.e., pH, total163

suspended solids concentration, oxygen uptake rate, ammonia uptake rate, and nitrate formation164

rate)wereclosely followedandcomparedbetween inhibitingandnon-inhibitingconditions. (iii)To165

differentiate ion trapping from biotransformation, the fate of the 12 target amines and the eight166

controlMPswasassessedaccordingtothreespecificcriteriaindicativeofiontrappingasfollows:167

• Concentration inthe liquidphase:Since, inthecaseof iontrapping,concentration levels in168

the liquid phase are expected to be lower under non-inhibiting than under inhibiting169

conditions, the differences between themean concentrations under these two conditions170

were calculated for each time point. The time point with the maximal concentration171

differencewasselectedforevaluation.Differencesof>8.7µg/L(givenaspikeconcentration172

of60µg/L,fordetailsseebelow)wereconsideredsignificantsincethisvaluecorresponded173

to the average maximal difference between concentrations of replicate samples for all174

compoundsanalyzed.175

• Removalkineticsintheliquidphase:PlotsoftheconcentrationofMPsasafunctionoftime176

fortheexperimentscarriedoutundernon-inhibitingconditionsareexpectedtodeviatefrom177

first-orderkinetics.Thisisbecausetheiontrappingprocess,whichshouldeventuallyleadto178

the establishment of equilibrium across the different membranes, needs some time to179

establish,butalsocoincideswithbiotransformation.Thepresenceorabsenceoffirst-order180

kinetics was evaluated by means of visual inspection and R2 of the linear fit to the181

logarithmizeddata.However, since theevaluationof removalkinetics isonlymeaningful if182

significantremovaltakesplaceatall,thisevaluationwasonlyconductedforcompoundswith183

a removal rate constant > 0.1 d-1. Under protozoa-inhibiting conditions, where first-order184

kineticsareexpected,allcompoundsmeetingtheaboveremovalratecriteriondisplayedR2185

8

values > 0.88 (see Table S8). Therefore, this value was chosen as a threshold to judge186

whetherthefitdeviatedstronglyfromfirst-orderkineticsundernon-inhibitingconditions.187

• Extracted amounts from the solid phase: Since the trapped MPs are expected to be188

extractable from the solid phase along with the physically sorbed MPs, the extracted189

amounts should be higher under non-inhibiting than under inhibiting conditions. The time190

point with the maximal ratio between the extracted amounts from non-inhibiting and191

inhibiting conditions (after correction for differing total suspended solids concentrations192

underthetwoconditions)wasusedtoevaluatethiscriterion.Aratioof>1.9wasselectedas193

being indicative of significant accumulation in sludge under non-inhibiting conditions. This194

value was chosen because all compounds exhibited ratios of <1.9 in the recovery195

experiments(seeSIsectionS6.4andTableS6therein)anditisthereforeconsideredacrude196

estimateofthemaximaluncertaintyofthisratioduetouncertaintyinthesludgeextraction197

method.198

MPexperimentswereconductedinbioreactors(100and250mLSchottbottles)filledwithactivated199

sludge(50and100mL,respectively)sourcedfromthenitrificationbasinofafull-scaleSwissWWTP200

(moredetailsontheexperimentalset-uparegiveninChapterS2.2).Toachieveinhibitingconditions201

for protozoa, a digitonin solution (100mg/ml)20was added to selectedbioreactors to yield a final202

concentrationof600mg/L.After2hof incubation,experimentswerestartedbyspiking60µLofa203

MPsolution (50mg/L foreachMP) resulting ina startingconcentrationofabout60µg/L foreach204

MP.TomeasuretheliquidphaseconcentrationsoftheMPs,sampleswerewithdrawnfromtriplicate205

reactorsforeachconditionwithin35minutes(timezerosample)andatapproximately2h,4h,7h,206

12h,24h,30h,52h,and71hafterthestartoftheexperiment.TodeterminetheamountofMPsin207

thesolidphase,non-inhibitedandinhibitedactivatedsludgesampleswereextracted24at4h,24h,208

and 72 h. For this, filtered sludge samples from the reactorswere freeze-dried. After addition of209

internalstandards,thefollowingextractionprocedurewasrepeatedthreetimes.Extractionsolution210

(nanopure water:methanol:formic acid 200:200:1) was added, the mixture was vortexed,211

ultrasonicated (15 minutes at 50°C), and centrifuged (10 minutes, 4000 rpm, Megafuge 1.0 R,212

Heraeus).Thethusresultingsupernatantswerecombined,evaporatedtodrynessandreconstituted213

in nanopure water. Additional blank and recovery tests were conducted for the solid phase214

measurements. Additionally, sorption control experiments with autoclaved activated sludge and215

abioticcontrolexperimentswithautoclavedsludgefiltratewereconducted.MPswereanalysedby216

meansof reversed-phase liquid chromatography coupled to ahigh-resolutionquadrupoleOrbitrap217

massspectrometer(Qexactive,ThermoScientific)(adetaileddescriptionoftheanalyticalmethodis218

given in Chapter S3).7 Finally, another set of biotic reactors were run to measure specific219

9

experimentalconditions,includingpH,oxygenuptakerate,ammoniauptakerate,nitrateformation220

rate,andtotalsuspendedsolidsconcentration.221

222

ResultsandDiscussion223

Distributionofacridineorangeinactivatedsludge224

Figure2andFiguresS1-S7intheSIshowexamplesofthedistributionofacridineorange(AO)inthe225

activatedsludgeflocs. Inthesefigures,regionsemittinggreenorredlightarerenderedingreenor226

redcolor, respectively,whereasareasemittingbothgreenandred lightare rendered inyellow. In227

sludge samples without AO addition, no autofluorescence could be detected under the imaging228

conditionsused(FigureS8intheSI),confirmingthatanydetectedfluorescenceintreatedsamplesis229

associatedwiththepresenceofAO.230

In Figure 2, individual protozoa (identified as amoeboidsVahlkampfia) can be recognized and are231

clearlydistinguishablefromsludgeflocs(Figure2a).Theyshowedintensegreenandredfluorescence232

inclearlydelineatedcellularcompartmentsindicatingthatAOaccumulatedinthesestructures(note233

thatyellowareasinFigure2indicatesimultaneousemissionofbothgreenandredlight).Incontrast,234

sludgeflocsemittedashadedgreenishlightonly,mostlikelycausedbysorptionofAOtothesludge235

flocs or by binding of AO to bacterial DNA. Similar non-localized and less intense green light236

emissionscouldalsobeobservedinthecytosoloftheprotozoa.Totheextentthattheprotozoain237

Figure2andFigures S1-S7 couldbe identifiedbasedonvisual inspectionunder themicroscope, it238

was found that all amoeboids, some of the ciliates and none of the rotifers showed distinct239

fluorescentcompartments.Onlygreenfluorescencewasobservedwhensludgewasincubatedwith240

digitonin prior to adding AO (Figure S9 in the SI). Also, no protozoa could be recognized in these241

cases, confirming complete destruction of protozoa by digitonin.20 Together, these observations242

providestrongevidencethatAOwashighlyselectivelyaccumulatedinspecificcellularcompartments243

of certain groups of live protozoa in activated sludge. These observations not only confirmed the244

existenceofacidicvesicles inthoseprotozoa,butalsovisuallydemonstratedtheaccumulationofa245

specificamine-containingcompoundinthosevesicles.246

247

10

248

Figure 2: Confocal laser-scanningmicroscope images of activated sludge stainedwith acridine orange (AO).249

EmissionsfromAOmonomers,i.e.lowconcentrationsofAO,aredetectedatwavelengthsof480-560nmand250

showningreen.EmissionsfromAOstacks,i.e.highconcentrationsofAO,aredetectedatwavelengthsof590-251

660nmandshowninred.Yellowareas indicateemissionofbothgreenandred light. (a)Sludgeflocswitha252

protozoa(amoeboid,Vahlkampfia)attached;(b)Isolatedprotozoan(amoeboid,Vahlkampfia).253

254

b) a)

11

Fateofmicropollutantsundernon-inhibitingandinhibitingconditions255

Experimental parameters of the sludge communities in the bioreactor experiments (i.e., pH, total256

suspended solids concentration, oxygen uptake rate, ammonia uptake rate, and nitrate formation257

rate)undernon-inhibitingandinhibitingconditionsaregivenanddiscussedinSectionS6.1oftheSI.258

Briefly,observedtrendsinthetotalsuspendedsolidsconcentrationandtheoxygenuptakeratewere259

consistentwiththeabsenceofprotozoagrazingonbacteria inthebioreactorsrununder inhibiting260

conditions,whilenitrificationdidnotseemtobeaffectedbydigitonintreatment.261

ThemeasuredamountsofallinvestigatedMPsandtheirquantifiedTPsintheliquidandsolidphase262

ofactivatedsludgeincubatedunderbothinhibitingandnon-inhibitingconditionsaregiveninFigures263

S11-S30.Exemplaryresultsforfourcompoundsrepresentingdistinctlydifferentbehaviors(TRI,AZO,264

FEN,MIA)aregiveninFigure3.CalculatedrateconstantsandsorptioncoefficientsaregiveninTable265

2,TableS7,andTableS8.Accordingtotheresultsofthecontrolexperiments,sorptionandabiotic266

transformation were not affected by the addition of digitonin. The data also showed that abiotic267

transformationwas ofminor relevance and could be neglected. Results on the three criteria that268

wereassessedtocomparethebehavioroftheMPsundernon-inhibitingandinhibitingconditionsare269

presented in Table 2, namely concentration levels in liquid phase (as difference between mean270

concentrations), removal kinetics in the liquid phase (as R2 of the linear fit to the logarithmized271

concentrations), and extracted amounts from the solid phase (as maximal ratio of extracted272

amounts). Additionally, plots of the logarithmized concentrations against time, including linear fits273

andresidualplots,aregiveninFiguresS31-S50.274

Basedontheirbehaviorinthebioreactorexperiments,wecouldclassifytheMPsintothreegroups,275

wherebytwocompoundswerejudgedexceptions.ControlgroupI includedSUL,NAP,TRI,DET,and276

ALA,controlgroupIIincludedAZO,ISO,andCLT,andthetargetgroupincludedallamines(RAN,VEN,277

LID, TRA, LEV, MEX, FEN, CIT, PRO, andMIA) except for the two exceptions ATE and TIC. In the278

following,thesegroupswillbediscussedseparately.InFigure3,wepresentplotsofconcentrationas279

functionoftimeforatleastonerepresentativeofeachgroup.280

ControlgroupI.ThefivecompoundsofcontrolgroupIcouldnotbetrappedinacidicvesiclesdueto281

theirspeciationcharacteristics,whichareneutral-anionicforSUL,NAP,andTRI,andneutralforthe282

twocompoundsDETandALA.Therefore,onlybiotransformationandsorptionremainedaspossible283

removalprocesses,wherebythedatashowedthatthelatterwasofminorrelevance.Ascanbeseen284

in Figure 3a for TRI and in Table 2 for all five compounds, the deactivation of the protozoic285

community did not affect the removal of these compounds, since no significant difference in286

concentration-time courses was observed under non-inhibiting and inhibiting conditions, and287

removalfromtheliquidphasefollowedfirst-orderkineticsunderbothconditions.Thisindicatesthat288

12

protozoic biotransformation was not relevant for theseMPs and that the main removal process,289

namelybacterialbiotransformation,wasnotaffectedbytheadditionofdigitonin.290

Control group II. For the three fully neutral compoundsAZO, ISO, and CLT of control group II, ion291

trappingcouldalsobeexcludedasremovalprocess,andsorption,too,wasshowntobenegligible.292

Therefore, only bacterial and protozoic biotransformation remained as potentially relevant293

processes.Thismatchedtheobservationthattheirremovalfollowedfirst-orderkineticsunderboth294

conditions and that the extracted amounts from the solid phase were not significantly different295

betweenconditions.However,ascanbeseenfromtheconcentrationlevelsintheliquidphase(for296

AZO see Figure 3b; for ISO and CLT see Figures S17 and S18, respectively) biotransformationwas297

affectedbytheadditionofdigitonin.ThiswassupportedbytheconcentrationsofthetwoTPsAZOA298

and NISO, which were both formed in higher amounts under non-inhibiting conditions. The299

differenceinTPformationbetweenthetwoconditionswasmostpronouncedforAZOA,whichwas300

formedquantitatively fromAZOundernon-inhibiting conditions, yetwasnot formed in significant301

amounts under inhibiting conditions. Since it can be assumed that digitonin did not affect the302

bacterialactivity,20whichissupportedbyourmeasurementsoftheexperimentalparametersandthe303

resultsforcontrolgroupI,weconcludethatAZO,ISO,andCLTwereremovedfromtheliquidphase304

through biotransformation by protozoa. Interestingly, a common characteristic of all three305

compoundsisthattheypossessahydrolysablemoiety(i.e.,carboxylicacidesterandureagroups).At306

the same time, lysosomes, which are one type of acidic vesicles present in eukaryotic cells, are307

knowntoharbordifferentdegradativeenzymesthatbelongtotheacidhydrolasefamily.25Basedon308

ourresults,wethusspeculatethatAZO, ISO,andCLTarebiotransformedtoasignificantextentby309

hydrolasescontainedinprotozoiclysosomes.310

Target group. For all amine-containing MPs, except for ATE and TIC, which will be discussed311

separately,measuredamountsintheliquidphaseweresignificantlylowerundernon-inhibitingthan312

under inhibiting conditions,which indicates thatprotozoawere very important for the removal of313

amines from the liquid phase. Furthermore, the clear deviation from first-order kinetics and the314

higherextractedamountsfromthesolidphaseinthenon-inhibitedsludgecomparedtotheinhibited315

sludge(Table2)clearlypointedtowardsiontrappingbeingtheimportantprotozoicremovalprocess.316

MPs that were also removed under inhibiting conditions followed a first-order removal process317

(TableS8andFiguresS31-S50),indicatingthatbacterialbiotransformationwastherelevantremoval318

processforthemunderconditionswhereprotozoawereinhibited.319

320

13

Table 2: Sorption Coefficients, Removal Rate Constants, and Evaluation of Target and Control Compound321

Behavior(summarizedas“Levelofagreement”inthelastcolumn).322

Sorptioncoefficient,Kd,1 (non-inhibited)[L/kg]

Removalrateconstant,(inhibited)[1/day]

Maximaldifferencebetweenmeanliquid phaseconcentrations[timepoint]

R2 of thelinear fit tothelogarithmized liquidphaseconcentrations (non-inhibited)

Maximal ratiobetween theextractedamounts fromthe solid phase[timepoint]

Level ofagreementwith thethreecriteria toassesstargetbehavior

SUL -17(±21) 1.56(±0.03) 4.2[24h] 0.98 -

NAP -1(±11) 2.10(±0.15) 0.9[31h] 0.94 1.2[24h]

TRI -53(±16) 3.22(±0.19) 1.4[7h] 0.94 1.2[4h]

DET -14(±13) 1.17(±0.07) 0.5[7h] 0.90 1.2[24h]

ALA 45(±16) 1.91(±0.03) 7.2[12h] 0.99 0.9[4h]

AZO 19(±24) 0.03(±0.01) 42.4[52h] 0.99 0.8[4h] +

ISO 1(±18) 0.15(±0.01) 14.2[71h] 0.95 1.0[4h] +

CLT 6(±13) 0.36(±0.01) 22.4[52h] 0.88 0.9[4h] +

ATE -17(±20) 3.56(±0.19) 5.8[7h] 0.96 0.8[4h]

RAN 52(±19) 0.99(±0.05) 21.5[7h] 0.50 - ++

VEN 2(±14) 0.04(±0.01) 18.8[71h] 0.65 3.0[72h] +++

LID -5(±18) 0.08(±0.01) 26.1[31h] 0.84 6.8[24h] +++

TRA -2(±16) 0.05(±0.01) 18.1[31h] 0.63 4.0[24h] +++

LEV 227(±36) 0.23(±0.01) 18.4[31h] 0.83 3.1[24h] +++

MEX 190(±34) 0.48(±0.03) 23.0[7h] 0.70 2.7[4h] +++

FEN 36(±23) 0.62(±0.02) 23.0[7h] 0.63 7.2[72h] +++

CIT 199(±29) 0.62(±0.02) 10.7[7h] 0.75 2.9[72h] +++

PRO 231(±34) 0.58(±0.02) 16.5[7h] 0.76 1.4[72h] ++

MIA 570(±76) 0.53(±0.01) 11.8[7h] 0.77 3.9[72h] +++

TIC 5274(±906) 1.16(±0.05) 1.1[24h] 0.98 1.0[4h]

Data are given asmean and standard deviation. Sorption coefficients, Kd,1, calculated from the liquid phase323

concentrations in the abiotic and sorption control experiments as given in Equation S4 in the SI (for more324

details on calculation of sorption coefficients, including alternative methods and explanation of negative325

values, seesectionS6.5 in theSI).Removal rateconstantscalculated fromthe linear fit to the logarithmized326

14

data.Valuesjudgedindicativeofexpectedtargetcompoundbehaviorarehighlightedinbold(i.e.>8.7µg/Lfor327

the concentration difference in the liquid phase, <0.88 for R2, and >1.9 for themaximal ratio between the328

extractedamounts).329

330

331Figure3:Plotsofmeasuredamounts(innmol)asafunctionoftimefora)trinexapac-ethyl(controlgroupI),b)332

azoxystrobin(controlgroupII),c)fenfluramine,andd)mianserin(bothtargetgroup)undernon-inhibiting(left333

graph) and inhibiting (right graph) conditions. The following TPs are shown: b) azoxystrobin acid, c)334

fenfluramineN-desethyl,andd)mianserinN-oxideasTP1andmianserinformamideasTP2.Amountsinliquid335

and solid phase were calculated for a typical bioreactor with 100mL sludge and are shown asmeans and336

standard deviation of replicate measurements (n≥3) (If the error bar is not visible, it is smaller than the337

symbol).Totalamounts intheliquidphaseandparentaswellasTPamounts inthesolidphasearegivenfor338

thetimepointswherebothliquidandsolidphaseamountsweredetermined(4h,24h,72h).Fortheparent339

compounds, first-order fits are indicated as solid lines. TP amounts in the liquid phase are connected with340

dotted lines. TP amounts in the liquid phase are only shown if the amounts are higher than 1% of the341

theoreticalamountofparentspiked.342

343

Withinthetargetgroup,FEN,LEV,LID,MEX,TRAandVENbehaveveryconsistentlyasrepresented344

byFENinFigure3c.Ofthose,thebehaviorofVENandTRAismosteasilyinterpreted(seeFiguresS21345

andS23,respectively).Inbothcases,theamountsofTPsformedwereminorunderbothconditions346

andhardly any removalof theparent compoundswasobservedunder conditionswhereprotozoa347

wereinhibited.Itcanbeconcludedthatthesetwocompoundswerehardlybiotransformedatalland348

that the observed disappearance of the parent compounds from the liquid phase of the non-349

inhibited sludge was almost exclusively due to ion trapping. This is also supported by the larger350

amountsofparentcompoundsextractedfromthesolidphaseofthenon-inhibitedsludgerelativeto351

theinhibitedsludge(i.e.,ratiosof4.0at24handof3.0at72hforTRAandVEN,respectively),and352

0

5

10

15

20

25 a) Trinexapac-ethyl non-inhibiting

parent, liquid phase exponential fitTP1, liquid phaseTP2, liquid phasetotal, liquid phaseTPs, solid phaseparent, solid phase

inhibiting

0

5

10

15

20

25 b) Azoxystrobin non-inhibiting inhibiting

0

5

10

15

20

25 c) Fenfluramine non-inhibiting

0 10 20 30 40 50 60 70

inhibiting

0 10 20 30 40 50 60 70

0

5

10

15

20

25 d) Mianserin non-inhibiting

0 10 20 30 40 50 60 70

inhibiting

0 10 20 30 40 50 60 70Time [h]

Am

ount

[nm

ol]

15

thefactthattheparentcompoundsandTPsinthesolidandliquidphaseofthenon-inhibitedreactor353

summeduptowithin± 20%ofthetheoreticallyspikedamountatalltimepoints.354

Findings for LID, FEN, LEV and MEX are similar to those for VEN and TRA, yet more bacterial355

biotransformationwasobservedforthesecompoundsastheywerestillremovedininhibitedsludge356

and did so following first-order kinetics. While, for LID, consideration of the two quantified TPs,357

namelyNLIDandLINO,yieldedaclosedmassbalance(seeFigureS22),thiswasnotthecaseforFEN,358

LEVandMEX(seeFigures3c,S24,andS25,respectively).Forthelatterthreecompounds,thesumof359

all species in the bioreactors decreasedover time,whichwasmost likely due to the fact that the360

quantifiedTPswerenot stable,but transformed furtherduring thecourseof theexperiment.This361

alsoseemedtobethecaseforthemorestronglysorbingcompoundsCIT,PRO,andMIA.362

TP analysis additionally provided some evidence that biotransformation proceeded faster under363

inhibiting conditions than under non-inhibiting conditions for some of the compounds that364

underwentbacterialbiotransformation.ForFENandMIA,forinstance,theirrespectiveTPsNFENand365

MINO are formed in considerably larger amounts in the inhibited case (see Figures 3c and 3d,366

respectively).Thisseemsconsistentwithourhypothesisthatunderinhibitingconditionsmoreparent367

compound is available for biotransformation than under non-inhibiting conditions where a large368

fractionoftheparentistrappedinprotozoa.However,therewerealsocompounds,e.g.,LIDandCIT369

(seeFiguresS22andS27,respectively),forwhichformationofTPswaslessaffectedbytheinhibition370

oftheprotozoa.Finally,forRAN,forwhichtheamountinsludgecouldnotbequantifiedbutwhose371

behaviorwasconsistentwithiontrappingwithrespecttotheothertwocriteria,theformationofthe372

majorTPRASOwasevenfasterinthenon-inhibitedsludge.ThissuggeststhatinthecaseofRAN(see373

FigureS20), similarly to thecontrolcompounds incontrolgroup II,biotransformationwas tosome374

extentdirectlyaffectedbytheadditionofdigitonin,likelybecauseprotozoicbiotransformationwas375

alsorelevantforthisMP.376

For the strongly sorbing amines CIT, PRO,MIA (see Figures S27, S28, and 3d, respectively) and to377

someextentalsoforLEVandMEX,plotsofmeasuredamountsasafunctionoftimedeviatedfrom378

thoseof thepreviouslydiscussedamines.Theextractedamounts fromthesolidphaseof thenon-379

inhibitedand the inhibited sludgewerealmost the same.No significantdifferenceof themaximal380

ratiosofextractedamountswasobservedforPROatalltimepoints,andforCITandMIAatthefirst381

twotimepoints(i.e.,ratiosof1.4forCITandMIAat4h,and1.5forCITand1.7forMIAat24h).Yet,382

the data clearly show that the extracted amounts of CIT andMIA from the non-inhibited sludge383

remainedfairlyconstantoverthecourseoftheexperiment,whilethoseextractedfromtheinhibited384

sludgedecreasedwithtime(ratiosof2.9forCITand3.9forMIAat72h).Thissuggeststhatunder385

inhibitingconditionsmostofthecompoundsextractedfromthesolidphasewerereversiblysorbed386

and consequently equilibrated with the dissolved fraction, which was available for387

16

biotransformation. In contrast, under non-inhibiting conditions,most of the extracted compounds388

camefromthetrappedcompoundpool,whichwasnotavailableforbiotransformation.ForMIA,this389

explanation is consistent with the observed increased formation of its TP MINO under inhibiting390

conditions. Overall, these observations suggest that for the strongly sorbing amines, ion trapping,391

reversiblesorption,andbiotransformationwerehappeningveryreadilyandonsimilartimescales.392

ThetwomajorexceptionswithrespecttothetypicalpatternsobservedforaminesareATEandTIC393

(seeFiguresS19undS30,respectively).WhileATEshowedverylittlesorptiontosludge,TICwasthe394

most strongly sorbing amine studied here. Neither of them showed a clear difference in the395

concentration-timeplots betweennon-inhibiting and inhibiting conditions (see Table 2), indicating396

that iontrappingdidnotoccurtoanyrelevantextent.Whilethereasonsforthesefindingsremain397

largelyelusive,itisnoteworthythatthetwocompoundsexhibitedthefastestbiotransformationrate398

constantsunder inhibitingconditionsofallaminesstudied(Table2).ForATE, itsTPATAC,which is399

known to be formed through enzyme-catalyzed hydrolysis,26 was formed nearly quantitatively,400

confirming ATE removal to be due to biotransformation. Based on these findings, it could be401

hypothesized thatATEandTICwere transformed so readily in the cells’ cytosol thatno significant402

trappingoccurred.403

Allinall,thedatafromimagingexperimentswithacridineorangeaswellasfromexperimentswith404

amine-containingMPs in activated sludge with and without inhibition of protozoa provide strong405

evidence that ion trapping of amine-containing MPs occurs in the acidic vesicles of protozoic406

eukaryotes present in activated sludge. Furthermore, the experiments with the control MPs also407

indicatedthatprotozoicbiotransformationisrelevantforsomeMPs,suchasAZO,ISO,CLT,andRAN,408

whereasbacterialiontrappingseemedtobeofminorimportanceintheactivatedsludgeusedinour409

experiments.410

411

Additionalexperimentalevidenceforiontrapping412

Beyond theexperimentsdescribed indetail here, results fromourprevious researchonamines in413

activated sludge provide evidence for other characteristics of the ion trapping process. First, the414

extentof trappinghasbeendescribed to increasewithelevatedextracellularpH levels.12Wehave415

observed and described an increase in removal rate constants of amine-containing MPs with416

increasing pH in previous experiments conducted at pH 6, 7 and 8.7 While the observation was417

correctly interpretedasbeingdue to thedifference inmembranepermeabilityof thecationicand418

neutralspeciesoftheamines,wemostlikelyincorrectlyattributedtheobservedlossoftheamines419

to biotransformation only. Based on the findings presented here and the fact that, also in the420

previousstudy,theremovalkineticsdidnotalwaysfollowfirst-orderkinetics(e.g.,forPAR,MIA,ORP421

17

and PYR in 7), we now assume that at least part of the observed, pH-dependent removal in the422

previousstudywasalsoduetoiontrapping.423

Second,itisknownthatthevitalityofthecellsinfluencestheirtrappingcapacity.9,27,28Thisisrelated424

to the fact that themaintenanceof the lowpH level in theacidic vesicles is anactivemechanism425

consumingenergy.Instressedcells,energysupplyislimited,whichreducesthetrappingcapacityof426

theacidicvesicles.Inourpreviouswork,wehaveobserveddecreaseddisappearanceofthestudied427

amines from the liquid phase under conditions where the reactors were stirred with magnetic428

stirrers7ascomparedtowhensludgesuspensionwasmaintainedbyshaking10only.Stirringofthe429

sludgeaffectstheintactnessofthesludgeflocsbyexertingmechanicalstress.Thismightnegatively430

affecttheprotozoa,whichneedintactsludgeflocstoattachto.However,sinceourdataallowsthis431

comparisonforamine-containingMPsonly,itremainsunclearwhicheffectstirringorshakinghason432

otherremovalprocessessuchasbacterialbiotransformation.433

Finally,ithasbeenshowninmammaliansystemsthatthetrappingcapacityoftheacidicvesiclescan434

be saturated.29 Thus, ifmixtures of several positively ionizable compounds are present, their total435

concentrationmightexceedasaturationthreshold,leadingtolesstrappingforeachofthemthanif436

presentindividually.Inourpreviousstudy10,amineswerespikedindividuallyintobioreactorsorina437

mixture of ten. As can be seen from the measured concentrations given in the supporting438

informationofthatpaper,concentrationlevelsintheliquidphasewerehigherwhenthemixturewas439

spikedthanwhenthecompoundswerespikedindividually(e.g.,differencesaremostpronouncefor440

VEN, LID and PHE). This confirms that the saturation of the acidic vesicles also occurs within the441

protozoiccellsoftheactivatedsludgecommunity.442

443

Implications444

Taken together, the results of this study and previous experimental observations provide strong445

evidencethationtrappingisanadditional,importantremovalmechanisminbatchexperimentswith446

activated sludge for a highly prevalent class of MPs, i.e., aliphatic amines, that has so far gone447

unnoticed.Asaconsequence,ithasoftenbeenmisinterpretedasbiotransformationbyresearchers448

studying the fate of amines in activated sludge, including ourselves.7, 10, 30, 31 Here, we provide a449

mechanisticunderstandingof the ion trappingprocess,whichweobserved tooccur fora rangeof450

aliphaticamineswithpKavaluesbetween7to10andlipophilicitiesrangingbetweenlogPof0.4and451

4.2.Wealsobrieflydiscusstheinfluenceofdifferentexperimentalfactorsontheobservationofion452

trapping during batch experiments,which should enable recognition and correct interpretation of453

thisremovalprocessinfuturestudies.454

Beyonddemonstratinganovelremovalmechanisminbatchexperimentswithactivatedsludge,our455

observationstriggeranumberof follow-upquestionsthatarerelevantfordifferentscientific fields456

18

andpractical applications. First, how relevant is ion trappingof amine-containingMPs in full-scale457

systems? Since we worked with sludge sourced from a full-scale WWTP and still observed ion458

trapping,weconcluded that thecapacityof theacidicvesicles for takingupadditionalcompounds459

was not exhausted under the conditions of the experiments. However, more quantitative460

investigationson the typesand relativeamountsofprotozoicbiomass involved inaccumulationof461

aminesinactivatedsludgewouldbeneededtoestimatethepotentialmagnitudeoftheeffectinfull-462

scale systems. Second and potentially evenmore importantly, one could ask if not only aliphatic463

amines, but also ammonium itself accumulates in the acidic vesicles of protozoa. This thought is464

backed up by the fact that NH4Cl is used in cellular biology to alkalinize acidic vesicles.8, 32-34465

Additionally, own preliminary experiments with different sludge communities indeed pointed466

towards a short-term increase in ammonium concentrations upon destruction of eukaryotic cells467

throughdigitoninaddition(datanotshown).Storageofammoniuminprotozoa inactivatedsludge468

could potentially have far-reaching implications for our current view of the nitrogen cycle during469

wastewater treatment and might explain some of the irregular behaviors (i.e., sudden onsets of470

incompletenitrogenremoval)observedinfull-scaletreatmentplants.Additionally,sinceammonium471

concentrations during activated sludge treatment are several orders of magnitude higher than472

micropollutant concentrations, trapping of ammonium would potentially outcompete trapping of473

amine-containing compounds in full-scale systems. Third, a specific type of acidic vesicles, the so-474

called acidocalcisomes, have also been described to occur in bacteria, in particular in phosphate-475

accumulatingbacteria.35,36Thisraisesthequestionwhetherbacterialiontrappingalsocontributesto476

theextentof trappingofamine-containingMPsand/orammonium inactivatedsludge fromplants477

with enhanced biological phosphorus removal. Fourth, amine-containing compounds and/or478

ammonium trapped in acidic vesiclesmight be re-released from sludge upon cell disruption, e.g.,479

during anaerobic sludge digestion. Since the majority of micropollutants is expected to be480

recalcitrant to biodegradation under anaerobic conditions,37 we expect amine-containing481

micropollutants thus released to be transported out of the WWTP together with the biosolids.482

Finally, ion trapping may be a relevant process not only in microbial communities, but also in483

eukaryoticorganismsrelevant inenvironmentalsciences ingeneral,suchastestorganismsused in484

ecotoxicological studies. Taking this into consideration would require re-interpretation of the485

toxicokineticbehaviorofpositivelyionizablecompoundsinsuchsystems.486

487

Acknowledgments488

TheauthorsthankthestafffromtheWWTPNeugut,Dübendorf,Switzerland,forprovisionofsludge489

and help with sludge sampling. Martin Ackermann, Eawag and ETH Zürich, Switzerland, is490

acknowledged for letting us use the Laser Scanning Confocal Microscope and Joachim Hehl from491

19

ScopeM, ETH Zürich, Switzerland, for providing support with the instrument. We further greatly492

thank André Wullschleger from the WWTP Werdhölzli, Zürich, Switzerland for annotation of the493

protozoa. Financial support for this project has been provided by the European Research Council494

under the European Union's Seventh Framework Programme (ERC grant agreement no. 614768,495

PROduCTS)andbytheSwissNationalScienceFoundation(projectno.200021_134677).496

SupportingInformation497

Details on materials and methods, including MP selection, batch experiments, measurement498

methods, and data analysis, and details on results, including microscope images, experimental499

parameters,MPamounts in liquid and solidphasewithevaluationanddiscussion. Thismaterial is500

availablefreeofchargevietheInternetathttp://pubs.acs.org.501

502

References503

1. Verlicchi,P.;AlAukidy,M.;Zambello,E.,Occurrenceofpharmaceuticalcompoundsinurban504wastewater:removal,massloadandenvironmentalriskafterasecondarytreatment-areview.Sci.505TotalEnviron.2012,429,123-55.5062. Luo,Y.;Guo,W.;Ngo,H.H.;Nghiem,L.D.;Hai,F.I.;Zhang,J.;Liang,S.;Wang,X.C.,Areview507ontheoccurrenceofmicropollutantsintheaquaticenvironmentandtheirfateandremovalduring508wastewatertreatment.Sci.TotalEnviron.2014,473-474,619-41.5093. Joss,A.;Zabczynski,S.;Gobel,A.;Hoffmann,B.;Loffler,D.;McArdell,C.S.;Ternes,T.A.;510Thomsen,A.;Siegrist,H.,Biologicaldegradationofpharmaceuticalsinmunicipalwastewater511treatment:proposingaclassificationscheme.WaterRes.2006,40,(8),1686-96.5124. Oulton,R.L.;Kohn,T.;Cwiertny,D.M.,Pharmaceuticalsandpersonalcareproductsin513effluentmatrices:Asurveyoftransformationandremovalduringwastewatertreatmentand514implicationsforwastewatermanagement.J.Environ.Monit.2010,12,(11),1956-78.5155. Manallack,D.T.,Theacid-baseprofileofacontemporarysetofdrugs:implicationsfordrug516discovery.SARQSAREnviron.Res.2009,20,(7-8),611-55.5176. Marceau,F.;Roy,C.;Bouthillier,J.,Assessmentofcationtrappingbycellularacidic518compartments.MethodsEnzymol.2014,534,119-31.5197. Gulde,R.;Helbling,D.E.;Scheidegger,A.;Fenner,K.,pH-DependentBiotransformationof520IonizableOrganicMicropollutantsinActivatedSludge.Environ.Sci.Technol.2014,48,(23),13760-52113768.5228. Yang,W.C.T.;F.,S.F.;Pomerat,C.M.,Mechanismofdrug-inducedvacuolizationintissue523culture.Exp.Cell.Res.1965,38,195-506.5249. Morissette,G.;Lodge,R.;Marceau,F.,Intensepseudotransportofacationicdrugmediated525byvacuolarATPase:procainamide-inducedautophagiccellvacuolization.Toxicol.Appl.Pharmacol.5262008,228,(3),364-77.52710. Gulde,R.;Meier,U.;Schymanski,E.L.;Kohler,H.-P.E.;Helbling,D.E.;Derrer,S.;Rentsch,D.;528Fenner,K.,SystematicExplorationofBiotransformationReactionsofAmine-Containing529MicropollutantsinActivatedSludge.Environ.Sci.Technol.2016,50,(6),2908-2920.53011. Goldman,S.D.B.;Funk,R.S.;Rajewski,R.A.;Krise,J.P.,Mechanismsofamineaccumulation531in,andegressfrom,lysosomes.Bioanalysis2009,1,(8),1445-1459.53212. Marceau,F.;Bawolak,M.T.;Lodge,R.;Bouthillier,J.;Gagne-Henley,A.;Gaudreault,R.C.;533Morissette,G.,Cationtrappingbycellularacidiccompartments:beyondtheconceptof534lysosomotropicdrugs.Toxicol.Appl.Pharmacol.2012,259,(1),1-12.53513. DeDuve,C.;DeBarsy,T.;Poole,B.;Trouet,A.;Tulkens,P.;VanHoof,F.,Lysosomotropic536Agents.Biochem.Pharmacol.1974,23,2495-2531.537

21

34. Docampo,R.;Scott,D.A.;Vercesi,A.E.;Moreno,S.N.J.,IntracellularCa2+storagein590acidocalcisomesofTrypanosomacruzi.Biochem.J.1995,310,(3),1005-1012.59135. Docampo,R.,AcidocalcisomesandPolyphosphateGranules.InInclusionsinProkaryotes,592Shively,J.M.,Ed.SpringerBerlinHeidelberg:Berlin,Heidelberg,2006;pp53-70.59336. Docampo,R.;Moreno,S.N.J.,Acidocalcisomes.Cell.calcium2011,50,(2),113-119.59437. Falås,P.;Wick,A.;Castronovo,S.;Habermacher,J.;Ternes,T.A.;Joss,A.,Tracingthelimitsof595organicmicropollutantremovalinbiologicalwastewatertreatment.WaterResearch2016,95,596(SupplementC),240-249.597598