Received 22April 2003 Accepted 25 June 2003 Publishedonline 5September 2003

Antimicrobialegg cleaningby the fringed darter (: : Etheostomacrossopterum ): implicationsof anovel component of parentalcare in fishes Jason H.Knouft 1* ,Lawrence M.Page 1 and Michael J.Plewa 2 1Centerfor Biodiversity, Illinois NaturalHistory Survey, 607East Peabody Drive, Champaign, IL 61820,USA 2Departmentof Crop Sciences, University ofIllinois atUrbana-Champaign, 1101 West Peabody Drive, Urbana, IL 61801, USA Broad-spectrumantimicrobial compoundshave recentlybeen identified in theepidermal mucusof fishes andprobably serveas a first line ofdefenceagainst microbial pathogens.Because of the ubiquitous nature offungi andbacteria in aquatic systems,defence against thesepathogens should be required throughout thelifespan offishes, including theegg stage.We conducted experiments on Etheostomacrossopterum (Percidae: Catonotus),thefringed darter, todetermine if thepresence of a guarding male inhibits microbial colonization ofeggs. Based on results from acombination ofin-stream experiments, in vitro microbial assays,and morphological characteristics andbehaviour ofbreeding males,we proposethat antimicrobial egg cleaning by theguarding male isan effective component of parental care in thesefish. Although innate antimicrobial compoundshave beenidentified in avariety oforganisms ranging from insectsto vertebrates, integration ofthese compounds into a species’s reproductivelife history has beenidentified only in asmall numberof insect species. The resultsfrom this studynot only indicatethat E.crossopterum males provide anovel form ofvertebrate parental care,but also have implications regarding theevolution of parental care in fishesand transitional evolutionary stagesfrom noparental care tomale parental care. Keywords: nestguarding; parental care; Etheostomacrossopterum ;egg cleaning; antimicrobial compounds

1. INTRODUCTION Although aconsiderableamount of effort has been directedtowards understanding reproductive behaviour in Fishesdisplay aremarkable variety ofparental care,with fishes(Breder &Rosen1966; Perrone& Zaret 1979; formsof parental investmentquite different from other Baylis 1981; Gross& Sargent 1985), almost noattention vertebrate groups (Breder &Rosen1966; Perrone& Zaret has beengiven tothe relationship betweenparental care, 1979; Gross& Sargent 1985). Nestguarding is themost microbial infectionand egg viability. This is surprising commonparental behaviour exhibited among families of consideringthat all organic surfacesin aquatic systemsare fishes,with males more likely than femalesto be the coveredby biofilms that are composedprimarily ofhetero- guarding sex(Blumer 1979). This bias may bedue to the trophic bacteria andfungi (Lock et al. 1984). Localized greater netfitness advantage tomales resulting from growth inhibition ofthese microbes shouldbe essential to guarding (Gross& Sargent 1985). In any case,this par- egg survival. ental presencepresumably inhibits predation and Recentstudies have identifiedthe presence of anti- decreasesthe amount of debrison the eggs, provides oxy- microbial compoundsin theepidermal mucusof avariety genand assists in theremoval ofwaste (Moyle & Cech offishes (e.g. Oncorhynchusmykiss (Austin& McIntosh 1996). 1988); Cyprinus carpio (Cole et al. 1997); Pleuronectes Male guarding behaviour andthe evolution of parental americanus (LeMaitre et al. 1997); Moronesaxatilis ´ M. care in fishesare well studied(Balshine-Earn &Earn chrysops hybrid (Silphaduang &Noga 2001)). These 1998; Goodwin et al. 1998; Lindstrom2000; StMary et speciesare members ofdifferent orders (Salmoniformes, al. 2001). Male guarding in externally fertilizing species Cypriniformes, Pleuronectiformesand Perciformes, hasbeen proposed to be the result of an evolutionary tran- respectively), suggesting that thepresence of antimicrobial sitionfrom noparental care tomale parental care compoundsin fishmucus is not phylogenetically con- (Gross& Sargent 1985). However,this transition doesnot strainedand such compounds may beubiquitous among automatically ensureparental care tothe offspring. In fishes.These compounds have beenproposed to serve as somecases, an intermediate stage is evidentin which the afirst line ofdefence against pathogens(Boman 1995). male gains somebenefit from remaining at thenest while Becausenest-guarding fishes are in contactor closeprox- theoffspring realize nobenefit from themale’ s presence imity totheir eggs,antimicrobial compoundsin fish (Gross& Sargent 1985; Clutton-Brock1991). mucusmay servean adaptive purposeby preventing microbial colonization ofeggs. (Perciformes:Percidae), containing approxi- *Authorand address for correspondence: Department of Biology, CampusBox 1137,Washington University, StLouis, MO 63130-4899, mately 140 species,is themost diverse genus of freshwater USA([email protected]). fishesin North America (Page 2000). Catonotus,

Proc.R. Soc.Lond. B (2003) 270, 2405–2411 2405 Ó 2003 TheRoyal Society DOI10.1098/ rspb.2003.2501 2406J. H.Knouft andothers Antimicrobial egg cleaning bythe fringed darter containing 18 species,is asubgenusof Etheostoma in opaque and/orfungus-covered eggs. Wealso noted whether or whichall speciesexhibit egg-clustering behaviour charac- not the malewas stillguarding the eggs. No eggs hatched during terizedby male guarding (Page 1983, 1985; Porterfield et the experiment.A comparisonof the percentageof infectedeggs al. 1999). Within Catonotus,adultsare sexually dimorphic versusuninfected eggs among treatments was madeusing data during thebreeding season.Males are larger andmore collectedon day 4(paired t-test, a = 0.05).Percentage data were boldly patterned.During thereproductive period arcsinetransformed before statistical analyses to achievenor- (typically March –May),a male establishesa territory mality.Based on dailyegg counts,a minimalnumber of eggs underthe cavity ofa flat stone.Multiple femaleswill wereconsumed by guarding males(less than 5%)and ouranaly- sequentially attach eggs totheunderside of the stone that sesof the percentageof infectedeggs assumesthat maleswere are simultaneouslyfertilized by themale. This process not selectiveregarding which eggs they consumed.Males and resultsin asingle layer ofup to 1500 eggs ( ca. 2 mm in nest stones werereturned to the streamafter terminationof diameter) depositedon thenest stone, i.e. on the ‘ceiling’ the experiment. ofthe nest cavity directly above themale. The male then Weperformeda colony-formingunit (CFU) assay to demon- remains at thenest until theeggs hatch.Time tohatching strate that opaque eggs containeda greater numberof bacteria is dependenton water temperature andcan range from 5 than cleareggs. Fiveclear and fiveopaque eggs wereselected to10 daysamong nestsacross the spawning season (Page froma nest and placedin separate 1.5ml microcentrifugetubes 1983), while time tohatching within aspecificnest is simi- and 250 m l of Luria–Bonnerbroth (LB)was addedto both lar for all eggs. tubes. Eggs werepulverized with asteriletoothpick and vortexed Although several studieshave investigated facetsof for 1min;100 m lquantities of serialdilutions (10 2 1 to 102 6) Catonotus reproductive behaviour (Knapp& Sargent fromeach egg type werethen incubatedon LB plates for 24h 1989; Page &Bart 1989; Lindstrom& Sargent 1997; at 37 °C.Afterincubation, CFUs were counted and averaged Page &Knouft2000; Porter et al. 2002), whether Caton- across serialdilutions to determinethe numberof CFUsper five otus males actually provide parental care is unclear. eggs. This experimentwas replicatedthree times. Becausefemales prefer todeposit eggs in nestsalready containing eggs (Knapp& Sargent 1989), males may (b) Collection ofepidermal mucus remain at thenest simply toobtain spawningswith To obtain samplesof epidermalmucus, we collected5 –12 females.The benefitof male guarding toeggs is evenmore E.crossopterum breedingmales from a stream.Epidermal mucus uncertainconsidering that Lindstrom& Sargent (1997) was acquiredby gently drawing asterilespatula across the pre- foundhigh levels offilial cannibalism by guarding males. dorsalnape of eachmale. We then combinedmucus from all The objectiveof this studywas to determineexperimen- malesfrom a particularstream and collectiondate ina 1.5ml tally whethermucus from male E.crossopterum confers an microcentrifugetube containing300 m lof 0.1M potassium antimicrobial benefitto guarded eggs. Observations indi- phosphate buffer (PPB)pH 7.4.Pooling of samplesfrom mul- catethat nestsabandoned by theguarding male oftencon- tiplemales was necessarybecause of the smallamount ofmucus tain non-viable eggs with fungal ( Saprolegnia spp.) and that canbe obtained fromeach male. Samples were either used bacterial infection.We predict that becauseof the behav- immediatelyor frozen at 280 °Cuntilrequired. iour andproximity tothe eggs ofthe breeding male, anti- microbial compoundscontained in themale ’s mucus will (c) Microbial assays inhibit microbial growth oneggs, increase egg viability, Previousstudies of vertebrate antimicrobialcompounds indi- andthus support the hypothesis that speciesof Catonotus catethat these substances arenot specific,with inhibitoryeffects provide male parental care. demonstratedon abroad range of microbes(Zasloff 1987,2002; Boman1995; Cole et al. 1997;LeMaitre et al. 1997;Silphadu- ang &Noga 2001).Because a limitedamount of mucuscould 2. MATERIALAND METHODS beacquired from each E.crossopterum male,and relativelysmall (a) In-stream study population sizesin eachstream limited the numberof malesthat Wecollected12 nest stones fromFerguson Creek,Livingston couldbe captured, we directedantimicrobial assays at bacterial County, KY,USA.At the timeof collection,each of the 12nest and fungal strains that representedcommon characteristics of stones was guarded by abreedingmale and containedfrom 67 the microbialcommunity. to 1440eggs. None ofthe eggs displayedobvious signs offungal orbacterial infection (i.e. opaque eggs and/orvisible hyphae). (d) Cytotoxicityassay Wecollectedguarding maleswith six of the nest stones and each Basedon previous studies (Cole et al. 1997;LeMaitre et al. maleremained with theirrespective nest stone throughout the 1997),the epidermalmucus of fishesappears to serveas a experiment.Males from the remainingsix stones werereturned mediumfor the concentrationand, hypothetically, deliveryof to the stream. an antimicrobialagent. Weused three separate samplesof E. Foreach experimental replicate, an in-streamcage crossopterum mucusto determinethe effectof mucuson bacterial (0.91 m ´ 0.61 m ´ 0.46m) enclosed with 1mmmesh screen growth. Sampleswere collected as describedabove and con- was dividedlaterally into two equal-sizedchambers with a1mm tainedmucus from five to sevenbreeding males. Sample A was meshscreen insert. We placeda nest stone with aguarding male collectedfrom individuals from Ferguson Creek,Livingston intoone chamber, while a nest stone without the guarding male County,KY, USA,on23 April 2000. Sample B, collectedon was addedto the adjoiningchamber. We paired nest stones in 16April 2000, and sampleC, collectedon 17April 2000 were eachcage to minimizethe differencesin egg numberbetween acquiredfrom individuals from Big Creek, Union County, IL, nests. USA. Wechecked eggs dailyfor 4days. During eachcheck, we To conductcytotoxicity assays, we first determinedprotein countedthe total numberof eggs and the total numberof concentrationsof the threeseparate samples.Fish mucusis

Proc.R. Soc.Lond. B (2003) Antimicrobial egg cleaning bythe fringed darter J.H.Knouft andothers 2407 primarilycomposed of mucin,which isaglycoprotein(Helfman 50 et al. 1997).Because the antimicrobialagent in E.crossopterum fish mucusis unknown, we quantifiedthe treatment doseof the 40 antimicrobialagent inthe cytotoxicity assay as microgramsof )

proteinin the formof mucin.We determined protein concen- % (

y tration ineachsample of mucus 1 PPBusing aBio-Radprotein t 30 i l a concentrationassay (microassayprocedure). A proteinstandard t r (1.36 mg ml2 1 albumin)was usedto generatea standard curve o m

20 by determiningabsorbance at 595nm of serialdilutions of the g g proteinstandard. Wethen comparedabsorbances of dilutionsof e mucus 1 PPBto the proteinstandard curveto calculateprotein 10 concentrationof the fish mucussamples. Weemployed a modifiedmicroplate-based in vitro bacterial cytotoxicity assay to quantitatively measurethe effectof E. cros- 0 1 2 3 4 sopterum mucuson the growth of Salmonellatyphimurium . This day number assay was modifiedfrom that publishedby Kargalioglu et al. (2000,2002). Salmonellatyphimurium was assayed becauseit Figure 1. Relationship between presence and absence of a was the standard microbeused to developthe cytotoxicity assay guarding male Etheostoma crossopterum and non-viability of (Kargalioglu et al. 2000,2002); 500 m loflog-phase S.typhimur- eggsin thenest. Open squares: male present; filled squares: ium cells(strain TA100) were grown in5 mlof LBfor 2hat male removed. 37 °Cwhileshaking. The celltitre was adjusted to an optical density(OD) of 0.30at 595nm. The cellswere exposed to a Usingthe methods describedabove, we collecteda pooled concentrationseries of the fish mucusin atotal volumeof 100 m l mucussample from 12 E.crossopterum malesfrom Ferguson perwell. A seriesof wellswere prepared with 100 m l of PPB. Creek,Livingston County, KY,USA.Wealso collected three The negativecontrol wells contained 30 m lofthe titredbacterial nest stones with apparently uninfectedeggs and a1lsampleof cellsuspension plus 70 m lof PPB.The treatment wellscon- unfilteredstream water fromFerguson Creek.Three treatments, tained 30 m lof the titredbacterial cell suspension and 70 m l of eachconsisting of fivereplicates, were applied to eggs fromeach known concentrationsof the fish mucus 1 PPB,and the con- stone.A replicatein treatment Aconsistedof a0.5ml centrifuge centrationwas determinedbased on the proteinconcentration tube containingone egg 1 120 m lPPB,areplicatein treatment of eachsample. The microplatewas incubatedfor 1hat 37 °C Bconsistedof a0.5ml centrifuge tube containingone egg whileshaking. Afterthe treatment time,100 m l of 2´ LB was 1 20 m l PPB 1 100 m lstreamwater, and areplicatein treatment addedto eachwell. At time0 the initialOD of eachwell was Cconsistedof a0.5ml centrifuge tube containingone egg measuredwith aBio-RadModel 550 Microplate Reader at 1 20 m l (mucus 1 PPB) 1 100 m lstreamwater. 595nm. This provideda time0 blank readingfor each Wescored each replicate with arank based onfungal infes- microplatewell. The microplatewas then placedin a padded tation at 0,8, 16 and 24h. Ascoreof 1indicatedno apparent holder,incubated at 37 °C,and shaken at 200r.p.m. for fungal infestation.In this case,the egg was nearlytransparent. 210min. The finalOD ofeachwell was determinedat 595nm Ascoreof 2indicatedthe presenceof fungus onthe external with the microplatereader. surfaceof the egg. Atthis level,embryos arestill living. A score Foreach well, we subtracted the blank ODvalue(time 0 of3indicatedopaqueness of the egg, suggesting that fungus had reading)from the ODreadingof eachspecific well after 210min invadedthe egg, and ascoreof 4indicatedthe presenceof both of incubation.The blank-correcteddata for the wellsof each externaland internalfungal infestation.Scores of 3and 4indi- concentrationof the fish mucuswere averaged. The concurrent catecases when the embryois no longer viable. Comparisons negative controlconsisted of bacteriawith nofish mucus among treatments weremade separately using data collectedat exposure.The blank-correcteddata for the negativecontrol eachtime point. Becausethree tests aremade on the samedata- wereset at 100%.The blank-correcteddata for eachfish mucus set,a Bonferroni-corrected a (a9)was usedto determinestatisti- concentrationwere converted into a percentageof the negative calsignificance of data fromeach time-point ( a = 0.050,three control,which isthe measureof bacterialcytotoxicity. tests, a9 = 0.017).

3. RESULTS (e) Saprolegnia spp.inhibition Weconductedan experimentto assess the inhibitoryeffect of (a) In-stream study mucuson the growth of Saprolegnia spp. (Oomycota). Saproleg- The presenceof a guarding male significantly reduced nia spp. was chosenfor this experimentbecause it is an rates offungal and/or bacterial infectionof eggs (d.f. = 5, extremelycommon freshwater fungus, acommonfish fungal tstat = 24.879, p = 0.0046; figure 1). The elevatedmor- pathogen, and alsofound to bethe primaryfungal colonizerof tality onday 4in the ‘male present ’ treatment is dueto a eggs inabandoned E.crossopterum nests (Wolke1975; J. H. high level ofinfection(53%) in onereplicate. During two Knouft and L.M.Page,personal observations). Identification outof four nest checks, this male wasfound away from to speciesin Saprolegnia relieson the morphology of the sexual his neststone. Consequently the high rate ofinfection may phase. Saprolegnia samplesacquired from fishes and other bedue to nest abandonment by this male. Percentageof aquatic animalstypically lack a sexualphase (Hughes 1994), infectedeggs wasstill low(less than 5%) at thetime of and species-levelidentification was not possiblein this study. abandonment,suggesting that thehigh rate ofinfection Consequently,samples were grouped intothe classificationof occurredafter abandonment.Infections in theother five Saprolegnia spp. replicates onday 4ranged from 1% to10%.

Proc.R. Soc.Lond. B (2003) 2408J. H.Knouft andothers Antimicrobial egg cleaning bythe fringed darter

Examination ofnon-viable eggs revealed thepresence (a) ofboth Gram-positive andGram-negative ( Aeromonas 100 hydrophila, Pseudomonas sp.;identified with Benton- ) DickinsonOxi/ Ferm TubeII) bacteria aswell ashigh lev- l o

r 80 t

elsof fungal ( Saprolegnia spp.)infestation. Opaque eggs n o c containeda muchgreater numberof CFUsthan clear eggs e 60 v

(meanclear-egg CFUsper five eggs = 1308.3, mean i t a opaque-egg CFUsper five eggs = 767 160.8). g e 40 n

% (

(b) Cytotoxicityassay

m 20

In all threetrials, S.typhimurium displayed adose- u i r dependentcytotoxic responseto E.crossopterum mucus u m (least squareslinear regression:sample A: y = 21.60x 1 i 0 h

2 p 0 15 30 45 60 81.573, = 11.949, = 0.749, = 0.026; sample B: y F1 ,5 r p t b 2 a ( ) y = 20.297x 1 102.634, F = 67.265, r = 0.871, p , l 1 ,11 l e

0.001; sample C: y = 20.410x 1 100.869, F1 ,1 0 = n

o 100 2 m

24.153, r = 0.729, p , 0.001; figure 2). Becausewe were l a S

unawareof the effect of the mucus from eachsample

n 90 o

beforetesting, we attempted to include a broad range of s doseswith amaximum numberof concentrations for each u c sample. Consequently,because different amounts of u m 80

mucuswere collected from eachsample group, different m u r

numbersof concentrations were used in eachtrial. e t

p 70 o s s

(c) Saprolegnia spp.inhibition o r c Becausewe usedeggs from threeseparate stonesin the 60 a

Saprolegnia spp. inhibition study,we conducted an initial m o t testusing data collectedat 24 htodetermineif therewas s (c) o e

astoneeffect across treatments. Neither treatment A h t

E 100 (Kruskal–Wallis: H = 0.000, d.f. = 2, p . 0.999), treat-

stat f o

ment B (KW: Hstat = 2.333, d.f.= 2, p = 0.311), or treat- t

c 90 e ment C (Hstat = 0.000, d.f. = 2, p . 0.999) displayed any f f e differencesin responseamong stones.Therefore, we 80 y r o

groupedreplicates among stonesfor subsequentanalyses. t i

b 70 Kruskal–Wallis testsusing data collectedat eachtime- i h n point indicateda significant differencein fungal infes- i 60 tation among treatmentsat all time pointsafter 0h(8 h:

Hstat = 42.429, d.f. = 2, p , 0.001; 16 h: Hstat = 42.308, 50 d.f. = 2, p , 0.001; 24 h: Hstat = 42.857, d.f. = 2, p , 0.001), with only treatment Bexhibiting obviousfungal 40 0 30 60 90 120 150 infestation(figure 3). Saprolegnia spp.was the only appar- entfungus present on infected eggs at theconclusion of protein per well ( m g) theexperiment. Figure 2. Cytotoxic response of Salmonella typhimurium to Etheostoma crossopterum mucus. (a)SampleA from Ferguson 4. DISCUSSION Creek, Livingston County, KY,USA, collected on 23 April 2000. (b)SampleB from Big Creek, Union County, IL, Although previous researchhas examined male repro- USA,collected on 16 April 2000. ( c)SampleC from Big ductivebehaviour in speciesof Catonotus (e.g.Knapp & Creek, Union County, IL, USA, collected on 17 April 2000. Sargent 1989; Porter et al. 2002), actual identification of male parental care has beenelusive. The resultsfrom this studyconfirm that E.crossopterum males,and presumably for microbial inhibition. Wenoted the location ofinfected other speciesof Catonotus,provide parental care by redu- eggs in theguarded nests and observed, in subsequentnest cing theamount of microbial infestationof eggs and checks,that infectedeggs werenot removed by themale. consequentlyincreasing egg viability. This reductionin Single eggs wouldbecome infected in theguarded nests microbial infestationis apparently facilitated by antimicro- butinfection would not spread to adjacent, contiguously bial compoundspresent in themucus of the guarding placedeggs eventhough eggs werenot removed by the male. male. Localizedinfections in theunguarded nests would Whether theguarding male consumedinfected eggs quickly spreadto adjacent eggs. This suggeststhat during thein-stream experiment andthus reduced the microbial spreadis inhibited by theantimicrobial effectof spreadof microbes among theuninfected eggs deserves themucus and not by removal ofeggs by themale. consideration.Visual observationsof the eggs during the Consideringthe results from thecytotoxicity assay,the experiment, coupledwith theresults from themicrobial mucusappears tohave antiseptic effectson the bacteria. assayssuggest that this isprobably notthe major reason Weconclude that theeffect is cytotoxic asopposed to

Proc.R. Soc.Lond. B (2003) Antimicrobial egg cleaning bythe fringed darter J.H.Knouft andothers 2409 n o

i 4 t a t s e f n i

l

a 3 g n u f

f o

k 2 n a r

n a e

m 1

0 8 16 24 time (h) Figure 4. Breeding male Etheostoma crossopterum . Oval indicates mucus-cell-rich pre-dorsal pad. Figure 3. Meanrank of Saprolegnia spp.infestation of eggs among treatments over time. Black bars: treatment A,1 egg 1 0.1 Msterile PPB; light grey bars: treatment B,1 egg 1 stream water 1 0.1 Msterile PPB; dark grey bars: thepre-dorsal pad ofthe male with theeggs is extremely treatment C,1egg 1 stream water 1 (0.1 Msterile common.In fact,previous studieshave suggestedthat the PPB 1 mucus). male coatsthe eggs with mucusto reducehis tactile dam- age tothe eggs aswell asprovide aphysical (rather than chemical) barrier against pathogens(Bart &Page 1991). cytoinhibitory basedon results acquired during assay Thus,the delivery ofthe antimicrobial compounds developmentand calibration in whichtreatment wells appears tobe facilitated andenhanced by thepresence of exhibiting aneffect at 210 min had asmaller numberof themucus-cell-rich fleshy pre-dorsal pad in breeding CFUsthan treatment wellsat 0min (Kargalioglu et al. males. 2002). It is unclearwhether the mucus is fungicidal, Consideringthe results from thein-stream and inhibits fungal growth orpreventsfungal adhesionto the microbial experiments aswell asthe morphology and egg surface.Nevertheless, the effect can still beviewed as behaviour ofbreeding males,antimicrobial egg cleaning is antifungal given that thepresence of the guarding male probably aneffective form ofparental care in E. crossop- decreasedin-stream fungal growth oneggs andthe terum.This type ofparental care isnovel largely because addition ofmucus apparently eliminated fungal growth on ofthe associated chemical componentcoupled with the eggs in thecentrifuge tubes. apparently adaptive male pre-dorsal morphology, which The recentdiscovery ofantimicrobial compoundsin permits increasedproduction and application ofthe epi- fish epidermal mucushas revealed astriking line of dermal antimicrobial mucusto the eggs. Previous studies defencein responseto the microbial compositionof the have speculatedthat nest-guardingfishes can mechan- aquatic environment(Austin & McIntosh1988; Cole et ically remove debrisby fanning eggs andthus prevent al. 1997; LeMaitre et al. 1997; Ebran et al. 2000). If adult microbial colonization (e.g.Co ˆte´ &Gross1993); however, fishesbenefit from this defence,eggs depositedon the sucha mechanism has notreceived strong experimental biofilm-covered substratewill also presumably benefit support.Moreover, it seemsunlikely that theminimal flow from adefenceagainst microbial pathogens.Results from ofwater createdby fanning wouldbe sufficient to remove thebacterial cytotoxicity assay and Saprolegnia spp. inhi- attachedbacteria andfungal spores,or inhibit thehyphal bition studysuggest that acompoundin E.crossopterum spreading andgrowth offungi that have adheredto the mucusexhibits antimicrobial activity. Consequently,the egg surface. identification ofparticular morphological and/or behav- ioural characteristics presentin breeding males that facili- (a) Broaderimplications ofantimicrobial tate application ofmucus to eggs shouldsupport compoundsin parental care antimicrobial egg cleaning asa form ofparental care in Innateantimicrobial compoundswith abroad anti- fishes. microbial effecthave beenidentified in avariety ofmulti- Antimicrobial compoundshave beenfound associated cellular organisms (Zasloff2002), ranging from insects with anddispersed from theepithelial mucus-secreting (review in Bulet et al. 1999) toseveral groups ofver- cellsof fishes(Cole et al. 1997). Therefore,a higher con- tebrates(e.g. amphibians (Zasloff1987), fishes(Cole et centrationof mucus cells should increase the amount of al. 1997) andmammals (Harder et al. 1997)). To our antimicrobial compounddelivered tothe skin surface. In knowledge,the integration ofthesecompounds into a spe- ahistological study,Bart &Page (1991) notedthat the cies’sreproductive life history has only beenidentified in fleshy pre-dorsal pad displayed only in breeding males had insects. Drosophilamelanogaster males transferantibacterial ahigher concentrationof mucus-secreting cells than the proteinsto their mates,presumably protecting both the samearea in non-breedingmales aswell asother areas of female’sreproductive tract andthe eggs against bacterial thebreeding male (figure 4). The disproportionately large infection(Lung et al. 2001). Female reproductive glands amountof mucus secreted by breeding males from this in themedfly ( Ceratitiscapitata )producean antimicrobial area is clearly apparent during handling ofthese fishes secretionthat istransferred to the egg surface(Marchini (J.H.Knouft,personal observation).The area occupied et al. 1995, 1997). This transferis apparently theonly by themale underthe nest stone is small andcontact by previously documentedcase of a speciesapplying ananti-

Proc.R. Soc.Lond. B (2003) 2410J. H.Knouft andothers Antimicrobial egg cleaning bythe fringed darter microbial compoundto deposited eggs (Marchini et al. activityof proteins extracted from epidermal mucus of fish. 1997). Comp.Biochem. Physiol. 122, 181–189. The fewantimicrobial agentsstructurally identifiedin Ebran, N.,Julien, S., Orange, N., Auperin, B.&Molle, G. themucus of bony fishes(Osteichthyes) are proteins.It 2000 Isolation and characterization of novel glycoproteins has beenproposed that thesecompounds bind to and from fishepidermal mucus: correlation between their pore- forming properties and their antibacterial activities. Biochim. essentially dissolvecellular membranes (Pouny et al. 1992; Biophys. Acta 1467, 271–280. Gazit et al. 1995; Shai 1995; Ebran et al. 1999, 2000; Gazit, E., Boman, A., Boman, H.G.&Shai,Y. 1995 Interac- Zasloff2002). Evidencesuggests that an antimicrobial tion of themammalian antibacterial peptide cecropin p1 chemical componentto parental care may bepresent in withphospholipid-vesicles. Biochemistry 34, 11 479–11 488. both egg-guarding andegg-dispersing freshwaterfish Goodwin, N.B., Balshine-Earn, S.&Reynolds, J.D.1998 species(e.g. Pimephales spp.(Smith &Murphy 1979); Evolutionary transitions in parental care in cichlid fish. Proc. Percafluviatilis (Paxton &Willoughby 2000)). Consider- R.Soc. Lond. B 265, 2265–2272. (DOI10.1098/rspb. ing therelatively recentidentification ofantimicrobial 1998.0569.) compoundsin fish mucusand the apparent needfor a Gross, M.R.&Sargent, R.C.1985 Theevolution of male and defenceagainst microbes in aquatic environments,identi- femaleparental care in fishes. Am. Zool. 25, 807–822. Harder, J., Bartels, J., Christophers, E.&Schroder, J.M.1997 fication ofantimicrobial componentsto reproductive Apeptide antibiotic from human skin. Nature 387, 861. behaviour may eventually berevealed tobe a relatively Helfman, G.S., Collette, B.B.&Facey, D.E.1997 The diver- commonphenomenon. sity offishes .Boston, MA:Blackwell Science. Hughes, G.C.1994 Saprolegniasis, then and now: aretrospec- D.White and thestaff at theHancock Biological Station were tive. In Salmon saprolegniasis (ed. G.J.Mueller), pp. 3 –32. extremely helpful and gracious during our research attheir Portland, OR:Department of Energy, Bonneville Power facility. WethankW. Chen for isolating and identifying Sapro- Administration. legnia spp.from experiments during thisresearch, and E. Kargalioglu, Y., McMillan, B.J., Minear, R.A.&Plewa, M.J. Wagner and Y.Kargalioglu for assistanceduring thecytotoxic- 2000 Anew assessmentof thecytotoxicity and genotoxicity ity assays.We also thankP. Levin for assistancewith Gram- of drinking water disinfection by-products. In Natural staining and CFUprocedures. S.Phelpsprovided helpful organicmatter and disinfection by-products:characterization and suggestions regarding experimental design. We also appreciate control in drinking water (ed. S.E.Barrett, S.W.Krasner & thecomments from L.Harmon, J.Losos and three anonymous G.L.Amy), pp. 16 –27. Washington: American Chemical reviewers on aprevious version of thismanuscript. This research wassupported bythe Illinois Natural History Survey Society. and theIllinois Department of Transportation. Kargalioglu, Y., McMillan, B.J., Minear, R.A.&Plewa, M.J. 2002 Analysis of thecytotoxicity and mutagenicity of drink- ing water disinfection by-products in Salmonella typhimur- REFERENCES ium. Terat. Carcin.Mutagen. 22, 113–128. Knapp, R.A.&Sargent, R.C.1989 Egg-mimicry asa male Austin, B.&McIntosh, D.1988 Natural antibacterial com- mating strategyin thefantail darter, Etheostoma flabellare : pounds on thesurface of rainbow trout. J. Fish Dis. 11, femalesprefer maleswith eggs. Behav. Ecol. Sociobiol. 25, 275–277. 321–326. Balshine-Earn, S.&Earn, D.J.D.1998 On theevolutionary LeMaitre, C., Orange, N.,Saglio, P., Saint, N., Gagnon, J.& pathwayof parental care in mouth-brooding cichlid fish. Molle, G.1997 Characterization and ion channel activities Proc.R. Soc. Lond. B 265, 2217–2222. (DOI 10.1098/ of novel antibacterial proteins from theskin mucosa of carp rspb.1998.0562.) (Cyprinuscarpio ). Eur.J. Biochem. 240, 143–149. Bart Jr, H.L.&Page, L.M.1991 Morphology and adaptive Lindstrom, K.B.2000 Theevolution of filial cannibalism and significance of fin knobs in egg-clustering darters. Copeia femalemate choice strategies asresolutions tosexual conflict 1991, 80–86. in fishes. Evolution 54, 617–627. Baylis, J.R.1981 Theevolution of parental care in fishes, with Lindstrom, K.B.&Sargent, R.C.1997 Food access, brood reference toDarwin ’srule of male sexual selection. Environ. size and filial cannibalism in thefantail darter, Etheostoma Biol. Fish. 6, 223–251. flabellare. Behav. Ecol. Sociobiol. 40, 107–110. Blumer, L.S.1979 Maleparental care in thebony fishes. Q. Lock, M.A., Wallace, R.R., Costerton, J.W., Ventullo, Rev. Biol. 54, 149–161. R.M.&Charleton, S.E.1984 River epilithon: toward a Boman, H.G.1995 Peptide antibiotics and their role in innate structural-functional model. Oikos 42, 10–22. immunity. A.Rev. Immun. 13, 61–92. Lung, O.,Kuo, L.&Wolfner, M.F.2001 Drosophila males Breder, C.&Rosen, D.1966 Modesof reproduction in fishes . transfer antibacterial proteins from their accessory gland and Neptune City, NJ:TFHPublishing. ejaculatory duct to their mates. J.Insect Physiol. 47, 617– Bulet, P., Hetru, C.,Dimarcq, J.L.&Hoffmann, D.1999 622. Antimicrobial peptides in insects; structure and function. Marchini, D.,Manetti, A.G.O.,Rosetto, M.,Bernini, L.F., Devl Comp.Immun. 23, 329–344. Telford, J.L., Baldari, C.T.&Dallai, R.1995 cDNA Clutton-Brock, T.H.1991 The evolution ofparental care . Prin- sequence and expression of theceratotoxin gene encoding ceton University Press. anantibacterial sex-specificpeptide from themedfly Ceratitis Cole, A.M.,Weis, P.&Diamond, G.1997 Isolation and capitata. J.Biol. Chem. 270, 6199–6204. characterization of pleurocidin, an antimicrobial peptide in Marchini, D.,Marri, L., Rosetto, M.,Manetti, A.G.O.& theskin secretions of winter flounder. J.Biol. Chem. 272, Dallai, R.1997 Presence of antibacterial peptides on thelaid 12 008–12 013. egg chorion of themedfly Ceratitis capitata . Biochem. Biophys. Coˆte´,I.M.&Gross, M.R.1993 Reduced disease in offspring: Res. 240, 657–663. abenefit of coloniality in sunfish. Behav. Ecol. Sociobiol. 33, Moyle, P.B.&CechJr, J.J.1996 Fishes: an introduction to ich- 269–274. thyology.Upper Saddle River, NJ:Prentice Hall. Ebran, N.,Julien, S., Orange, N., Saglio, P., Lemaitre, C.& Page, L.M.1983 Handbook ofdarters .Neptune City, NJ: Molle, G.1999 Pore-forming properties and antibacterial TFHPublications.

Proc.R. Soc.Lond. B (2003) Antimicrobial egg cleaning bythe fringed darter J.H.Knouft andothers 2411

Page, L.M.1985 Evolution of reproductive behaviors in per- escently labeled analogs withphospholipid-membranes. Bio- cid fishes. Bull. Ill. Nat. Hist. Surv. 33, 275–295. chemistry 31, 12 416–12 423. Page, L.M.2000 . In Percid fishes: systematics, Shai,Y. 1995 Molecular recognition between membrane-span- ecology,and exploitation (ed. J.F.Craig), pp. 225 –253. ning polypeptides. TrendsBiochem. Sci. 20, 460–464. Oxford: Blackwell Science. Silphaduang, U.&Noga, E.J.2001 Peptide antibiotics in mast Page, L.M.&Bart, H.1989 Eggmimics in darters (Pisces: cells of fish. Nature 414, 268–269. Percidae). Copeia 1989, 514–517. Smith, R.J.F.&Murphy, B.D.1979 Functional morphology Page, L.M.&Knouft, J.H.2000 Variation in egg-mimic size of thedorsal padin fatheadminnows ( Pimephales promelas in Etheostoma oophylax ,theguardian darter (Percidae: Rafinesque). Trans.Am. Fish. Soc. 103, 65–72. Catonotus). Copeia 2000, 782–785. StMary, C.M.,Noureddine, C.G.&Lindstrom, K.2001 Paxton, C.G.M.&Willoughby, L.G.2000 Resistance of Environmental effectson male reproductive success and par- eggsto attackby aquatic fungi. J.Fish Biol. 57, ental care in theFlorida flagfish Jordanella floridae . Ethology 562–570. 107, 1035–1052. Perrone, M.&Zaret, T.M.1979 Parental care patterns of Wolke, R.E.1975 Pathology of bacterial and fungal diseases affecting fish. In (ed. W.E.Ribelin & fishes. Am. Nat. 113, 351–361. The pathology offishes G.Migaki), pp. 33 –116. Madison, WI: University of Wis- Porter, B.L.,Fiumera, A.C.&Avise, J.C.2002 Eggmimicry consin Press. and allopaternal care: two mate-attracting tactics bywhich Zasloff, M.1987 Magainins, aclassof antimicrobial peptides nesting striped darter ( Etheostoma virgatum )malesenhance from Xenopus skin. Isolation, characterization of two active reproductive success. Behav. Ecol. Sociobiol. 51, 350–359. forms, and partial cDNAsequence of aprecursor. Proc. Natl Porterfield, J.C., Page, L.M.&Near, T.J.1999 Phylogenetic Acad. Sci. USA 84, 5449–5453. relationships among fantail darters (Percidae: Etheostoma: Zasloff, M.2002 Antimicrobial peptides of multicellular organ- ):total evidence analysisof morphological and Catonotus isms. Nature 415, 389–395. molecular data. Copeia 1999, 551–564. Pouny, Y., Rapaport, D.,Mor, A., Nicolas, P.&Shai,Y. 1992 As this paper exceedsthe maximum lengthnormally permitted, the Interaction of antimicrobial dermaseptin and its fluor- authors have agreedto contributeto production costs.

Proc.R. Soc.Lond. B (2003)