Thephylogenetictreeconstructedusingtheaminoacidse-quencesofviperinhomologues,includingBjVip,from106differentspecies(SupplementaryS3)revealedtheuniqueclus-teringofrespectivetaxonomicgroups,wellre?ectingthephy-logenyofchosenorganisms(Fig.1B).AsearchoftherecentlycompleteddraftassemblyandautomatedannotationoftheFlor-idaamphioxusB.?oridaegenome(http://genome.jgi-psf.org/)revealedthepresenceofasingleviperinhomologuecDNAanditsgenomicDNAsequence(XM_002609092).ThesingleB.?oridaeviperinhomologuecDNAencodesaproteinwith~87.8%identitytoBjVipataminoacidlevel,indicatingthatBjVipishighlyconservedininterspecies.Thisalloweduspossibletoanalyzethesyntenyofviperinhomologuesofrepresentativespecies.AsshowninFig.1C,amphioxusviperin,RSAD2,waslocatedimme-diatelyupstreamofCMPK2,andallvertebrateRSAD2,exceptthatofplatypus,waslocatedimmediatelyeitherdownstreamorup-streamofCMPK2.Moreover,thedirectionofRSAD2andCMPK2geneswasoppositetoeachotherinallcases.TheseindicatedthattheRSAD2-CMPK2linkagewashighlyconservedthroughoutchordateevolution.

3.2.N-terminalpolymorphismandpositiveselection

DNApolymorphismanalysisbyDNAspver5.0indicatedthattheN-terminalsequences(~70aminoacids)ofviperinhomo-loguesfromdifferentspecies(SupplementaryS4)variedgreatly,whiletheircentralandC-terminalregionswerehighlyconserved(Fig.2A).Toanalyzewhethertherewaspotentialpositiveselec-tionintheancestrallineagesofviperin,site-speci?cmodelandbranch-speci?cmodelinPAMLwereusedtotestselectionpres-sureonviperinhomologuesofamphioxusand?shspecies(SupplementaryS4),andsite-speci?cmodelwasusedtotestforthepositiveselectionofspeci?ccodons.Insite-speci?cmodel,LRTsforM0eM3comparison(P<0.05)indicatedvariableuratioamongsites.LRTsforM7eM8comparison(P<0.05)clearlyshowedthatviperinhomologuesin?shandamphioxuswereevolvingunderpositiveselection(Table2).BEBanalysisinM8

M.Leietal./DevelopmentalandComparativeImmunology53(2015)293e302297

Fig.1.Molecularmodeling,phylogenetictreeandsynteny.(A)MolecularmodelofBjVippredictedbyfully-automatedproteinstructurehomologymodeling.(B)PhylogenetictreeofviperinproteinsconstructedintheJTTmodelofMEGA6.0usingtheaminoacid-basedmaximumlikelihood(ML)algorithm.Thereliabilityofeachnodewasestimatedbybootstrappingwith1000replications.AccessionnumbersforsequencesusedarelistedinSupplementaryS3.(C)Syntenicanalysisofviperinhomologuesofhuman,chimpanzee,mouse,platypus,chicken,zebra?nch,anolelizard,frog,zebra?sh,tilapia,elephantsharkandFloridaamphioxus.Abbrevation:PAQR3,progestinandadipoQreceptorfamilymember3-like;RSAD2,radicalS-adenosylmethioninedomaincontaining2,viperin;CMPK2,cytidinemonophosphate(UMP-CMP)kinase2;TMEM144,transmembraneprotein144;EIF5B,eukaryotictranslationinitiationfactor5B;SOX11,SRY(sexdeterminingregionY)-box11;SOX11a,SRY-boxcontaininggene11a;RNF144A,ring?ngerprotein144A;RNF144aa,ring?ngerprotein144aa;RNF144ab,ring?ngerprotein144ab;Kidins220,kinaseD-interactingsubstrate,220kDa;Kidins220a,kinaseD-interactingsubstrateof220a;ID2,inhibitorofDNAbinding2,dominantnegativehelix-loop-helixprotein,ID2a,inhibitorofDNAbinding2,dominantnegativehelix-loop-helixprotein,a;DCDC2C,doublecortindomaincontaining2C.Arrowsshowthedirectionoftranscriptionofgenes.

revealedthatsevencodonsites(sites14A,20T,25T,26I,28S,29Vand31P)weredetectedtoevolveunderpositiveselection(SupplementaryS5).ThesesevenpositiveselectivesiteswerealllocatedattheN-terminalamphipathica-helixdomain.Thesite28Swasdetectedtoevolveunderstrongpositiveselectionwithposteriorprobabilities!0.95.Totestwhethertherewaspositiveselectionamongdifferentbranches,branch-speci?cmodeltestwasperformed.LRTfromcomparisonofthefree-ratiomodelwiththeone-ratiomodel(P?4.734?10¨¤5)showedthaturatioswerevariableamonglineages.Totally,threebrancheswithu>1sug-gestedpositiveselectionduringevolution(Fig.2B),butthecom-parisonofthetwo-ratiomodelwiththeone-ratiomodelrevealedallbrancheswithP>0.05,indicatingthattherewerenopositiveselectionamongbranches.

3.3.ExpressionofBjvipinnormalandpolyI:C-challengedtissuesqRT-PCRwasusedtoexaminetherelativeexpressionlevelsofBjvipinthenormalandpolyI:C-challengedtissuesofamphioxus.Thedissociationcurveofampli?cationproductshowedasinglepeak,indicatingthattheampli?cationwasspeci?c(Datanotshown).AsshowninFig.3A,Bjvipwaspredominantlyexpressedinthehepaticcecum,hind-gut,gillandmuscle,andatalowerlevelintheovary,testisandnotochord,indicatingthatBjvipisexpressedinatissue-speci?cfashion.ChallengewithpolyI:Cresultedinasigni?cantlyenhancedexpressionofBjvipinallthetissuestested,includingthehepaticcecum,hind-gut,gillandmuscle(Fig.3BeE),implicatingthattheexpressionofBjvipissubjectedtothestimu-lationoftheviralmimicpolyI:C.

3.4.AntiviralactivityofBjVipincellculture

ToexaminethepotentialeffectofBjViponviralinfection,theFGcellsweretransfectedwiththeplasmidspcDNA3.1/GFP/BjvipandpcDNA3.1/GFP(asnegativecontrol).AsshowninFig.4A,BjVipwasproducedinthetransfectedcells.ThecellsweretheninfectedwithLCDV,andtherelativeamountofLCDVinthecellswasdeterminedbyqRT-PCRat24,48and72h.Comparedwithcontrolgroup,therelativeamountofLCDVinthecellstransfectedwithpcDNA3.1/GFP/Bjvipplasmidwassigni?cantlyloweredat48and72hpostinfection(Fig.4C),suggestingthatBjVipiscapableofeitherkillingLCDVorinhibitingitspropagation.3.5.AntiviralactivityofrBjVipinshrimp

RecombinantBjVip(rBjVip)withtheHistagexpressedinE.coliwaspuri?edbychromatographyonaNi-NTAresincolumn.TherBjVipobtainedwasanalyzedbySDS-PAGE,whichyieldedasinglebandofapproximately45kDa(Fig.5A),wellmatchingtheexpectedsizes(43.5kDacalculatedbyEditseq).WesternblottingshowedthatrBjVipreactedwithrabbitanti-Histagantibody,indicatingthatitwascorrectlyexpressed(Fig.5A).

PilotexperimentsshowedthatnoWSSVwasdetectedinthehepatopancreasandmusclesoftheshrimpsusedintheexperi-ments(Datanotshown).Theshrimpswereinjectedintraperito-neallywithTris-NaClaloneorWSSVinTris-NaClorWSSVthathadbeenco-incubatedwithrBjVip,andthentherelativequantitiesofWSSVinthehepatopancreasandmusclesweremeasuredbyqRT-PCR.Nopositivesignalsweredetectedinthehepatopancreas

and

298M.Leietal./DevelopmentalandComparativeImmunology53(2015)293e302

Fig.2.DNAPolymorphomismandpositiveselection.(A)DNApolymorphismofnucleotidesequencesofviperinhomologuesanalyzedbyDNAspver5.0withslidingwindowlengthof50bpandstepsizeof10bp.(B)Thepositiveselectionevaluatedwithfree-ratioofbranch-speci?cmodel.Thebrancheswithu>1werelabeledinredlines.(Forinterpretationofthereferencestocolorinthis?gurelegend,thereaderisreferredtothewebversionofthisarticle.)

musclesoftheshrimpsinjectedwithTris-NaClalone(Datanotshown),providingmoreauthenticitythattheshrimpsusedwerenotinfectedwithWSSV.Bycontrast,positivesignalsweredetectedinthehepatopancreasandmusclesoftheshrimpsinjectedWSSVinTris-NaClaswellwithWSSVco-incubatedwithrBjVipinTris-NaCl,buttherelativequantityofWSSVinthetissuesoftheshrimpsinjectedwithWSSVthathadbeenco-incubatedwithrBjVipwasmarkedlylowerthanthatinthetissuesoftheshrimpsinjectedwithWSSVat8he24hpostinjection(Fig.5BandC).Collectively,thesedatasuggestthatrBjViphasanantiviralactivityagainstWSSV.

4.Discussion

Viperin,oneofISGs-encodingproteinsinvertebrates,haspreviouslybeenreportedtoexhibitantiviralactivityagainsta

Table2

Testsforpositiveselectioninsitemodels.ModelM0M3M7M8

(oneratio)(discrete)(beta)

(beta&u)

Inl

¨¤9447.479596¨¤8917.613168¨¤8917.164988¨¤8914.069859

#p1524

2Dl

1059.73(M0vs.M3)6.19(M7vs.M8)

df42

LRTP<0.05P<0.05

Inl,loglikelihood;#p,numberofparametersintheudistribution;2Dl,twicethedifferencebetweenthelnl;df,degreesoffreedom,equaltothedifferenceinthenumberofparameters;LRT,likelihoodratiotest.

widerangeofDNAandRNAviruses,butfunctionalcharacteriza-tionofviperinislackingininvertebratesspecies.Thepresentstudydescribestheidenti?cation,expressionandfunctionalcharacterizationofviperinintheamphioxusB.japonicum.Toourknowledge,thisisthe?rstsuchdataininvertebrates.Thededuced357aminoacid-longprotein,BjVip,displaysfeaturesincommonwiththoseinhuman,chimpanzee,mouse,platypus,chicken,zebra?nch,anolelizard,frog,zebra?sh,tilapiaandelephantshark(Dangetal.,2010;Milicetal.,2015;Zhangetal.,2014);itisstructurallycharacterizedbythepresenceoftheSAMsuperfamilydomainwiththecharacteristicCNYKCGFCmotif,anditslinkageanddirectionrelativeCMPK2genearehighlyconservedthroughoutchordateevolution.Moreover,thepre-dicted3DstructureofBjVipishighlysimilartothatofhumanandzebra?shviperinhomologues.Additionally,wedemonstratethatviperinhasbeenevolvingunderpositiveselectionduringevolu-tionfromamphioxusto?shes.Inparticular,positiveselectiondrivesrapidevolutionofcertainaminoacidresiduesattheN-terminalamphipathica-helixdomainofamphioxusviperin,butnoneofaminoacidresidueatthecentralandC-terminaldomainsisunderpositiveselection.Theseresultsareconsistentwiththe?ndingbyPadhi(2013)thattheN-terminusofviperinin?shisunderpositiveselection.However,theyareincontrasttotheobservationsthattheC-terminalportionofviperinisunderpos-itiveselectionduringtheevolutionofprimates(Limetal.,2012).ManystudieshaveproventhatbothN-andC-terminiareimportantforantiviralactivityofviperin.TheN-terminalamphipathica-helicaldomain,whichtargetsviperintotheER,