Manuscript bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Title(working):AFacilitatedDiffusionMechanismEstablishestheDrosophilaDorsalGradient Authors:SophiaN.Carrell1,MichaelD.O’Connell1,AmyE.Allen1,2,StephanieM.Smith1andGregoryT. Reeves1,* 1 DepartmentofChemical&BiomolecularEngineering,NorthCarolinaStateUniversity,Raleigh,NC 27605,USA 2 Currentaddress:CurriculuminBioinformaticsandComputationalBiology,UniversityofNorthCarolina, ChapelHill,NC27599,USA *Correspondence:[email protected] Summary ThetranscriptionfactorNF-κBplaysanimportantroleintheimmunesystemasanapoptoticand inflammatoryfactor.IntheDrosophilamelanogasterembryo,ahomologofNF-κBcalledDorsal(dl) patternsthedorsal-ventral(DV)axisinaconcentration-dependentmanner.Duringearlydevelopment, dlissequesteredoutsidethenucleusbyCactus(Cact),homologoustoIκB.Tollsignalingattheventral midlinebreaksthedl/Cactcomplex,allowingdltoenterthenucleuswhereittranscribestargetgenes. Hereweshowthatdlaccumulatesontheventralsideoftheembryooverthelast5cleavagecyclesand thatthisaccumulationistheresultoffacilitateddiffusionofdl/Cactcomplex.Wespeculatethatthe predominantroleforCactinDVaxisspecificationistoshuttledltowardstheventralmidline.Giventhat thismechanismhasbeenfoundinother,independentsystems,wesuggestitmaybemoreprevalent thanpreviouslythought. Introduction Inadevelopingorganism,tissuesarepatternedbylong-rangesignalingenactedthroughmorphogen concentrationgradientsthatcarrythepositionalinformationnecessarytocontrolgeneexpressionina spatially-dependentfashion.IntheDrosophilablastoderm,thetranscriptionfactorDorsal(dl)actsasa morphogentoregulatethespatialpatterningofgreaterthan50genesalongthedorsal-ventral(DV)axis (MoussianandRoth,2005;ReevesandStathopoulos,2009).However,despiteitscentralityto development,themechanismthatcontrolstheformationofthedlnuclearconcentrationgradientatthe whole-embryoscaleisnotwellunderstood. Atthesinglenucleus/celllevel,dlissequesteredtothecytoplasminaninactivecomplexwiththeIκB homologCactus(Cact).Ontheventralsideoftheembryo,Tollsignalingresultsinthedegradationof Cact,freeingdltoenterthenucleuswhereitregulatestranscription.Thismechanism,combinedwitha ventral-to-dorsalgradientofTollsignaling,wouldseemsufficienttodevelopagradientofnucleardl concentration(Rothetal.,1989).However,takenalone,thismechanismwouldresultinalocal depletionofdlfromthecytoplasmsurroundingtheventralnucleitocreateacounter-gradientin cytoplasmicdl,whichisincontrasttotheobservationthatdlaccumulatesbothinthenucleiandinthe cytoplasmontheventralsideoftheembryoovertime(Reevesetal.,2012)(seeFigure1A-D). Inthisstudy,weinvestigatethemechanismofdlnuclearconcentrationgradientformationatthe whole-embryoscale.Weuseacomputationalmodeltoguidethedesignandinterprettheoutcomeof experimentsthatdistinguishbetweencompetinghypotheses.Ourexperimentalandcomputationaldata 1 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. supportthehypothesisthatamechanismoffacilitateddiffusionisresponsiblefortheglobalformation ofthedlnuclearconcentrationgradient.Underthis“shuttling”hypothesis,adorsal-to-ventral concentrationgradientofcytoplasmicdl/CactcomplexdevelopsastheresultofToll-mediated degradationofCact,whichdrivestheoverallfluxofdltowardstheventralsideoftheembryo.Besides ventrally-directeddlaccumulation,ourcomputationalmodelpredictsthattheshuttlingmechanismcan accountforseveralotherpreviouslyinadequatelyexplainedphenotypes,suchasduplicateddlgradient peaks,regionsofdepleteddl,dlheterozygousphenotypes,andwideneddl-GFPgradients(Huangetal., 1997;Libermanetal.,2009;Reevesetal.,2012;RothandSchüpbach,1994).Wehaveexperimentally verifiedthesepredictionsthatafacilitateddiffusionmechanismisresponsiblefortheformationofthe dlnuclearconcentrationgradientinthedevelopingDrosophilaembryo. Themechanismoffacilitateddiffusionhasbeenpreviouslydescribedandisresponsibleforgradient formationinothersystems,suchastheDpp/BMPsignalingpathway(Ben-zvietal.,2008;Eldaretal., 2002;Marquésetal.,1997;Mizutanietal.,2005;Shimmietal.,2005)andhasalsobeensuggestedfor theformationoftheSpätzlegradientupstreamofTollsignaling(Haskel-Ittahetal.,2012).Thedl/Cact systemhaseachofthefeaturesrequiredforfacilitateddiffusion(Shiloetal.,2013):(1)theprimary molecule(dl)bindstoa“carrier”molecule(Cact)thatprotectsitfromcapture,(2)theprimary/carrier complexisdiffusible(duetothesyncytialstateoftheearlyembryo),and(3)thecomplexisbrokenina spatially-dependentmanner(throughTollsignalingontheventralsideoftheembryo).Ifthesefeatures areinplace,theprimarymolecule(dl)accumulatesovertimeatthesiteofcomplexdegradation(the ventralsideoftheembryo).Therefore,weconcludethatCactperformsaroleindlgradientformation additionaltoregulatingdl’sentryintothenucleus:shuttlingdltotheventralsidetoformthemature gradient. Results Dorsalaccumulationontheventralsideoftheembryoresultsfrommovementofthedl/Cactcomplex Initially,DorsalisuniformlydistributedalongtheDVaxisofthedevelopingembryo(Figure1C);during nuclearcycles11-14,itaccumulatesontheventralside(Figure1B,D).Thisobservationisconsistent withpreviously-publishedfluorescentimagesofanti-dlimmunostainingsinfixedembryosandinimages ofdl-GFPfluorescenceinliveembryos(Kanodiaetal.,2009;Libermanetal.,2009;Reevesand Stathopoulos,2009;Reevesetal.,2012).Themechanismforthisoverallpolarizationoftotaldlinthe embryocouldbeasymmetricdegradationorproduction;however,ourmodelingworkandexperiments favoranoverallfluxofdltotheventralsideoftheembryo.UsingaphotoactivatableGFP(paGFP)tag (PattersonandLippincott-Schwartz,2002),wenoticedthatdlappearsinregionsoftheembryoover810nucleiawayfromthesiteofpaGFPactivationinthetimespanof90minutes.Givenacelldiameter (i.e.,theaveragedistancebetweenthecentroidsoftwoneighboringnuclei)ofroughly7μm,this translatestoaneffectivediffusivityof∼0.5-0.9μm2/sandatimescaletocrossonecelldiameterthatis ontheorderofminutes.Whenactivatedneartheventralmidline,dl-paGFPdiffusesfromitslocationof activationandfillsadjacentnuclei(Figure1E,MovieS1).Whenactivatedonthedorsalsideofthe embryo,dl-paGFPdiffusesandshowsatypicalpatternofexclusionfromnucleiinmoredorsalregions andmoderateuptakeintothenucleiinmorelateralregions(Figure1F,MovieS2).Asdiscussed previously,thisfluxofdlcouldbeduetodiffusionoffreedl,dl/Cactcomplex,orbothspecies(O’Connell andReeves,2015). 2 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Todeducewhichprocess(es)mayberesponsiblefortheoverallfluxofdltotheventralsideofthe embryo,weusedamodelofdl/Cactinteractions(Figure2A,B;seealsoSupplementaryExperimental Procedures)andexaminedtheequationfortotaldl(seeExperimentalProcedures).Fromthisequation, weseethatonlytwospeciesmaybecontributingtotheoverallmovementofdltotheventralsideof theembryo:freecytoplasmicdlandCact-boundcytoplasmicdl.Notethatthisisagenericpredictionof ourmodelequationsanddoesnotrelyonanyspecificmodelparametervalues. Modelanalysisandexperimentalpredictions Todeterminewhichspecies,freedlordl/Cactcomplex,isresponsiblefortheaccumulationofdlonthe ventralsideoftheembryo,weanalyzedthefullsetofmodelequations(Equations1-8inSupplemental ExperimentalProcedures)andmadethemodelconsistentwithspatiotemporaldataofdlgradient dynamics(Reevesetal.,2012)usingevolutionaryoptimization.Analysisoftheresultingparametersets revealsthat,tobeconsistentwiththedata,therateofdl/Cactcomplexdiffusionisgenerallymuch fastercomparedtothatoffreedl(Figure2C). Wethenanalyzedthedirectioninwhichdlanddl/Cactcomplexdiffuseinthemodel.AccordingtoFick’s law,diffusiveflux(typicallydenotedJ)referstothemovementofasolutefromareasofhigh concentrationtoareasoflowconcentrationatarateproportionaltoitsconcentrationgradient,𝑑𝐶 𝑑𝑥, yielding𝐽 ∝ − 𝑑𝐶 𝑑𝑥.Thenegativesignaccountsforthefactthatdiffusivefluxisinthedirectionof decreasingconcentration.Byconvention,wedefineourx-axistobeequaltozeroattheventralmidline andoneatthedorsalmidline(seeFigure2A,B);thus,anegativevalueforthefluxindicatesmovement towardtheventralmidline,andapositivevalueindicatesmovementtowardthedorsalmidline. Ourmodelingresultsindicatethatdl/Cactcomplexpreferentiallymovestowardstheventralmidline, whilefreedlmovesaway(Figure2D).Therefore,theventrally-directedfluxoftotaldlmustbedueto diffusionofdl/Cactcomplex,whichresultsfromTollsignalingactingasasinkattheventralmidline(see alsoFigure1A,B).Inthisway,Cactshuttlesdlagainstitsconcentrationgradient,therebyallowingan accumulationofnucleardlattheventralmidlinewhereitisrequiredforproperformationofthedl gradient. Themodelusedhereisanextensionofapreviouslypublishedmodel(O’ConnellandReeves,2015),in whichthecatalyticactivityofTollsignalingwasmodeledphenomenologically(Figure2A,B).Weupdated themodeltoexplicitlyaccountforbothactivatedTollreceptorsandTollreceptorsboundtodl/Cact complex(Equations7&8,respectively)toallowforthepossibilitythatTollactivityislimiting.This possibilitywasprecludedinthepreviousmodel,becausetherateofToll-mediatedCactusdegradation wasproportionaltotheconcentrationofdl/Cact,whichcreatedapotentiallyunlimitedsink. Furtheranalysisofourmodelshowedthatthemechanismoffacilitateddiffusioncanbetestedby alteringthreeexperimentally-tunablebiophysicalparameterstoslowthemechanismoffacilitated diffusion:(1)decreasingthedldosage,(2)loweringthediffusivityofdlorCact,and(3)increasingthe widthoftheactiveTolldomain(Figure2E-G).Themodelpredictsthattheseperturbationswilleach resultinahallmarkprogressionofphenotypes(dependingonthestrengthoftheperturbation):thedl gradientfirstwidens,thenbecomesflatatthetop,thensplitsintotwopeaks(Figure2H-J). Furthermore,combinationsoftheseperturbationshavesynergisticeffects.Weconductedexperiments toaddressallthreeparameters,andfoundthattheysupportedourhypothesis. 3 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Decreasingdldosagewidensandflattensthedlgradient Inembryosfrommothersheterozygousnullfordl(hereafterreferredtoasdl/+embryos),theshapeof thegradientbecomesslightlywider,flatter,andlowerinamplitude,withtheslopemaintainedinthe lateralregionsbutpeaklevelsofthegradientflattened(Libermanetal.,2009)(Figure3).Froma functionalityperspective,thisphenotypeisstrikingbecausetheportionofthedlgradientthatislost (thepeakconcentration)issuperfluousasanylevelabovethethresholdlevelforhigh-concentration targetgeneswillresultinexpression,whiletheportionofthegradientthatisnecessaryforpatterning geneexpressionboundaries—thegradedportionbetween20%and45%DVposition—ismaintained. Toexplainthiseffect,onesimplehypothesisisthattheventral-mostnucleiwouldhavereachednormal dllevels,butonlyiftherewereenoughdlpresent.Sincethetotalamountofdliscompromised,the ventralmostnucleireachalower-than-normalconcentrationofdl.However,modelsofdlgradient formationthatlacktheshuttlingmechanismpredictthelossof50%dlnuclearconcentrationgloballyin dl/+embryos,ratherthanaspatiallypatternedloss(FigureS1). Incontrast,amodelbasedontheshuttlingmechanismmakessenseofthisphenotype.Inwildtype embryos,thesmoothpeakofthedlgradientisobtainedduetoasaturationeffect.Thereisenough dl/CactcomplextosaturateactiveTollreceptorsintheventral-lateralregionsoftheembryo,which allowsasignificantfluxofdl/Cactcomplextoreachtheventralmidlineandresultsinasmoothlypeaked,Gaussian-likecurve.Inembryoswithhalfthenormaldldosage,activeTollreceptorsinthetails ofthegradientarenotsaturated;therefore,lessdl/Cactcomplexarrivesattheventralmidline,and theseembryosneveraccumulateasmoothpeakofthedlgradient(Figure3B,D). Decreaseddiffusionofdl/Cactcomplexwidensthedlgradient OtherresearchershavegeneratedseveralGFP-taggedversionsofdlinordertostudyitsdynamicsin livingembryos(Delottoetal.,2007;Reevesetal.,2012).Ineachcase,taggingdlwithanextraprotein moietycausedthedlgradienttoexpand.Ifthegradientdevelopsviaadiffusionmechanism,thisresult wouldseemcounterintuitive.Onewouldexpectthatincreasingthesizeofamoleculewoulddecrease itsdiffusivity,andthusrestrictthespatialrangeofthegradient.Ontheotherhand,ashuttling mechanismwouldpredicttheopposite:thatloweringthediffusivityofdl/Cactcomplexwouldwidenthe gradient.Totestthisprediction,weanalyzeddlandCactconstructstaggedwithproteinsthateither dimerizeortetramerizeinordertoformlargeclustersofdl/Cactcomplex. OtherstudieshaveshownthattaggingdlwithamonomericVenus(dl-mVenus)causesthedlgradientto widen(widthparameterσ=0.16±0.01;comparetoσ=0.14±0.01forwildtype),whileaGFPtag causesamuchgreaterwidening(σ=0.20±0.02)(Libermanetal.,2009;Reevesetal.,2012).We surmisedthatthedifferencebetweenthesescenariosisthatVenusisanobligatemonomerwhilethe GFPconstructweaklydimerizes(Zachariasetal.,2002).Tocarefullymeasuretheeffectofdimerization oftheproteintagonthewidthofthegradient,werepeatedtheseexperimentsinordertocontrolfor differencesinfluorescentproteintag,protocol,researcher,andequipmentsoastohaveadirect comparisonforourworkinnewflylines.First,weanalyzedthegradientinembryosthathadone endogenouscopyofdlreplacedwithdltaggedwiththeweakly-dimerizingGFP(dl-dGFP)(Reevesetal., 4 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. 2012).Wefoundthatthedlgradientintheseembryoswidenedsignificantly(σ=0.21±0.03)as comparedtowildtype(Figure4A,B). TheweakdimerizationofGFPcanbeabolishedbytheA206Kmutation(Zachariasetal.,2002); therefore,wedesigneddl-GFPconstructswiththeobligatemonomerversionofGFP(dl-mGFP)inorder toinvestigatetheeffectofincreasingthemassofdlwithoutcausingdimerization.Embryoscarrying theseconstructshavegradientsslightlywiderthanwildtype(σ=0.19±0.03),comparabletoour measurementsinembryoswiththemonomericVenustag(σ=0.18±0.03)(Figure4B,seealsoFigure S2A,B).Weexpectthatthewideningeffectinembryosthatcarryamonomerictagislikelyduetothe increasedmassofthedlproteinandnotthedimerizationofthetag.Wealsodesigneddl-Venus constructsthathadweakdimerizationrestoredduetoaK206Amutation(dl-dVenus);embryoscarrying thisdl-dVenustransgenehaddlgradientsthatweresignificantlywiderthanthoseinembryoscarrying thedl-mVenustransgene(σ=0.19±0.02),matchingthetrendseenindl-mGFPanddl-dGFPembryos (seeFigureS2A,B).Takentogether,thesedataindicatethattheweakdimerizationofthefluorescent tags,whichpresumablyslowsdiffusionofdl/Cactcomplex,isthekeyfactorinwideningthegradient. Itwasalsoobservedthatasinglecopyofdl-dGFPwasunabletocomplementnullmutationsinthe endogenousdlgene;twocopiesarenecessarytorescuethemutant(Libermanetal.,2009;Reeveset al.,2012).Thisresultcontrastsdl-mVenus,whichcomplementsatonecopy.Ourshuttlinghypothesis explainsthisphenotype,asembryosfromdlmotherscarryingonecopyofdl-dGFPhavetwodefectsin shuttling:alowercopynumberofdlandareduceddiffusivityofthedl/Cactcomplex.Theseembryos havehighlywidenedgradients(σ=0.22±0.04)(seeFigure4A,BandFigureS2C,D).Wesurmisedthat thesegradientsareperturbedenoughtoseverelydisruptgeneexpression,perhapsnotbecoming strongly-peakedenoughtoturnonhigh-thresholdgenes,suchassnail(sna).Therefore,weanalyzed snaexpressioninembryosfromdl;dl-dGFP/+mothers,andfoundthattheseembryoseitherlackedsna expressionorexpressedsnainaverynarrowdomain(Figure4C-E,seealsoFigureS2D),confirmingour hypothesisthatthedoubleperturbationresultsinabreakdownofthetypicallyrobustpatterning system. Toinvestigatetheeffectoflargercomplexesondl/Cactdiffusion,weevaluatedtheeffectofadl-lacZ transgeneonthedlgradient(Govindetal.,1992).AstheproteinproductoflacZ,β-galactosidase(β-gal), tetramerizes,weexpectthatthisfusionwillslowthediffusionofdltoagreaterextentthandGFP. However,wefoundthattwocopiesofthistransgeneinadl/+backgroundwereneededinordertosee aneffectonthewidthofthedlgradient,whichwidenedaspredictedbytheshuttlingmodel(SeeFigure S2B,E).Thisislikelyduetothefactthatdl-lacZactsasanantimorphicallele(Govindetal.,1992);such dl-βgalfusionproteinsaresuspectedtobeexpressedatverylowlevelsinsurvivingflylines(Govindet al.,1996). InordertostudytheeffectofdecreasedCactdiffusionondlgradientformation,weexaminedembryos frommotherscarryingasinglecopyofacact-lacZtransgeneinaheterozygouscactbackground (Fernandezetal.,2001).Weexpectthatthisfusionwillslowdiffusionofthedl/Cactcomplexwithout affectingthediffusionoffreedl.Intheseembryos,thedlgradientisexpanded(Figure4B,F),indicating aninabilityofCacttoproperlyshuttledltotheventralmidline.Inanaïvemodel,whereCactdoesnot shuttledlandthereisnofluxtowardtheventralmidline,wedonotexpectchangingthediffusionof Cacttohavesuchaneffectonthedlgradient.Itisimportanttonotethat,underthetypicalmorphogen gradientphenomenon,decreasingthediffusivityofthemorphogenwouldresultinanarrower(more 5 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. concentrated)gradient.Sinceourresultsconsistentlyshowthatthedlgradientwidensratherthan narrowswhenthediffusivityofdlorCactisreduced,weconcludethatdl/Cactcomplex,andnotfreedl, isthedominantdiffusivespecies. IncreasingthewidthoftheactiveTolldomainresultsinasplitpeakofdl TheshuttlinghypothesispredictsthatwideningtheactiveTolldomainresultsfirstinawidened,then flattened,thensplitdlgradient(Figure2H-J).AstheextentoftheTolldomainiscontrolledby Gurken/EGFRsignalingduringoogenesis(Schüpbach,1987;Senetal.,1998),weanalyzedembryosfrom motherscarryingahypomorphicEGFRallele(egfrt1)(RothandSchüpbach,1994).Wefoundthat embryosfrommothersheterozygousforthisallelehavesignificantlywideneddlgradients,andmost (10/12)embryosfromhomozygousmothershavegradientssowidethatasplitpeakforms(seeFigure 5).Thisresultisconsistentwithpreviousreportsthatvariousgurkenandegfrmutationsgeneratea duplicateddlgradientasmeasuredbydlstaining,Twiststaining,andsitesofventralfurrowformation (RothandSchüpbach,1994),whichisnotreadilyexplainedintheabsenceofashuttlingphenomenon (Meinhardt,2004;MoussianandRoth,2005). Ananteroposteriorgradientofdlsupportstheshuttlinghypothesis IthasbeensuggestedthatashuttlingphenomenonalsooccursthroughtheprocessingofSpätzle(Spz), theligandforTollsignaling(Haskel-Ittahetal.,2012),whichmayexplainsomeofthephenotypes describedhere.Todeterminewhetherthehallmarkphenotypesofshuttlingcouldoccurwithout assistancefromtheproteasecascadeorSpzprocessing,weexpressedaconstitutivelyactiveformofToll (Toll10b)attheanteriorpoleofthedevelopingembryousingthebicoid3’UTRandthebicoidpromoter (Huangetal.,1997).Embryosfrommotherscarryingthisconstruct(bcd>toll10b:bcd3'UTR)havean anterior-posterior(AP)dlgradientinadditiontothenativeDVgradient.Naïvely,onemayexpectthat theexistenceoftwogradientsinactiveTollsignalingwouldresultinhigherconcentrationsofnucleardl wherethesetwogradientsoverlap.Incontrast,theshuttlinghypothesispredictsthatdlnuclear concentrationwouldbelowerintheregionofoverlap,asthetwocompetingdl/CactsinkscauseCactto shuttledltowardboththeanteriorpoleandtheventralmidline(seeFigure6A-C).Thispredictionis borneoutinexperiment,astheseembryosshowadecreasedintensityofthedlgradientintheregionof overlap.Furthermore,64%(9/14)oftheseembryosshowavisiblenarrowingofthesnaexpression domainatroughly30%egglength,consistentwithpreviouslypublishednumbers(Huangetal.,1997). Thereisadipindlnuclearintensityatthislocationaswell(Figure6B,C). Ourmodelingresultsfurthersupportthattheshuttlingmechanismisresponsibleforthisphenomenon (Figure6C).Weexpandedourmodelsystemfromaone-dimensionalarrayofnuclearcompartmentsto atwo-dimensionalarrayandaddedasecondTollsignalingdomainperpendiculartothefirst.Our simulationresultsshowthattheoverlapbetweentheAPandDVTolldomainscreatesalowpointin nucleardlconcentration,similartoourexperimentalresults.Wethenapproximatedathresholdforsna expressionbasedontheDVdlgradient,andfoundthelocalminimumindlconcentrationresultsin reducedorabolishedsnaexpressioninthatregion(Figure6C’). Wealsoexaminedembryosfrommotherscarryingahomozygousmutationingastrulationdefective (gd7),whicheliminatestheendogenousventral-to-dorsalgradient.Swappingthebicoidpromoterfor thestrongerhsp83promoter(hsp83>toll10b:bcd3'UTRconstruct)(Huangetal.,1997),wewereable createawiderdlgradientattheanteriorpole.Halfoftheseembryos(7/16)showtwopeaksofdlboth 6 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. byeyeandwhenthenuclearintensityisplotted(Figure6D,E).Furthermore,theseembryoshavea doublepeakofsna(Trisnadietal.,2013).Inordertodeterminewhetherornotthisdoublepeak phenomenonwasaresultoftheembryo’sgeometryatthepole,weanalyzedembryoswithanAPdl gradientinitiatedbytheweakerbicoidpromoter.Theseembryosshowednosuchdoublepeakeffect (seeFigureS3).Theseresultsfurthersupportourshuttlinghypothesisasthedlgradientprogressesfrom narrow(weakpromoter)todoublepeak(strongpromoter)muchasitdoesinthenativeDVsystem whentheTolldomainexpands. Discussion Inthisstudy,weinvestigatedtheformationofthedlmorphogengradientintheearlyDrosophila embryo.Basedonourmodelandexperimentalverification,weconcludethatthedlgradientis establishedbyafacilitateddiffusion,or“shuttling”mechanism,inwhichdl/Cactcomplexdiffuses towardstheventralmidlinewhereCactisdegraded.Furthermore,ourhypothesizedmechanismmakes senseofseveralpreviouslyunexplainedphenomenaintheliterature. BesidesdlbindingtoCact,theshuttlingmechanismrequirestwophenomenatooccur.First,dl/Cact complexmustbeabletomovethroughouttheembryo.Ourexperimentswithdl-paGFPshowthatdl doesindeedmovethroughtheembryoinamannerthatappearstobeconsistentwithdiffusion.In contrast,apreviousreportshowedthateachnucleushasitsownwell-mixedcytoplasmicpoolofdlto drawfrom,andthatdldoesnotreadilydiffusefrompooltopool(Delottoetal.,2007).Thisobservation wasconfirmedbyastudywhichshowedthatbarrierstodiffusionexistintheearlyembryo,despitethe lackofcellmembranes(Mavrakisetal.,2009).However,theseresultsdonotcontradictour hypothesizedmechanism,whichreliesonmovementofdlfromonepooltoanother,becausethetime scalesaredifferent(secondsvs.minutes).Furthermore,observationalevidencesupportsadiffusionbasedshuttlingmechanism,asloweringthediffusivityofeitherdlorCactwidensthegradient(Figure 7A),ratherthanthenarrowingthatonemightexpectfromamorphogengradientestablishedby(nonfacilitated)diffusion.TheshuttlingmechanismexplainswhydltaggedwithaweaklydimerizingGFP widensthegradientmorethanonetaggedwithmonomericVenus(Reevesetal.,2012),andalso explainstherelatedobservationthatonecopyofdl-mVenuscomplementslossofendogenousdlwhile onecopyofdl-dGFPdoesnot(Libermanetal.,2009;Reevesetal.,2012).Similarly,thismechanism makessenseoftheobservationthatdltaggedwithβ-galactosidase,whichformstetramers,isantimorphic(Govindetal.,1992),asthedlmoietiesintetramersofdl-βgaldimerizewithendogenousdlto disrupttheformationoftheendogenousdlgradient. Second,theshuttlingmechanismrequirestheventralmidlinetobeasinkforthedl/Cactcomplex.This becomesespeciallyclearinembryosthathaveanectopicAPdlgradientinadditiontotheendogenous DVgradient(Huangetal.,1997).Intheseembryos,asecondsinkforthedl/Cactcomplexisestablished, andthetwocompetingsinksformalocalminimuminthedlgradientwherethetwooverlap,rather thantheglobalmaximumonemightexpectifthetwogradientswereadditiveastheywouldbeina modelwithoutfacilitateddiffusion(Figure6B,C).Thisresulthasbeenpreviouslyobservedasa"gap"in thesnadomainwherethetwogradientsoverlap(Huangetal.,1997),similartoour2Dsimulation results.TheseresultsarguethattheTolldomaindoesactasasink,anecessaryconditionforshuttling. Inthissystem,itappearsthatactiveTollreceptorsaresomewhatlimitingastheyareeasilysaturated withdl/Cactcomplex.Thissaturationisnotessentialtotheshuttlingmechanismperse,butitis necessaryforthemechanismtoexplainseveralobservationsofthedl/Cactsystem.Underwildtype 7 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. conditions,asignificantfluxofdl/CactcomplexcanbypassthesaturatedactiveTollreceptorsinthetails ofthegradient,whichresultsintheaccumulationofasmooth,intensepeakofdlsignalingattheventral midlineoftheembryo.However,inembryosfromdl/+mothers,thereisnotenoughdl/Cactcomplexto saturateactiveTollreceptorsinthetailsofthegradient,andthusthereislessaccumulationofdlatthe ventralmidline(Libermanetal.,2009)(seeFigure7B).Similarly,whentheTollgradientisgreatly expanded,asinembryosfromegfrt1mothers(RothandSchüpbach,1994),thedomainsofsaturatedToll receptorsmovefurtherfromtheventralmidline,resultinginasplitpeak(Figure7C).Additionally,the split-peakphenomenonhasalsobeenobservedinthedlgradientinabnormallylargeembryos(Garcia etal.,2013).ThisspecificphenotypecouldbeexplainedbytheshuttlingoftheTollligandSpz(HaskelIttahetal.,2012);however,weobservedthesamephenomenoninembryoswithanectopicAPdl gradientestablishedbyconstitutivelyactiveToll.Inbothcases,ventrally-(anteriorly-)diffusingdl/Cact complexdoesnotmakeittotheventralmidline(anteriorpole)beforebeingdissociated,leavingthe ventral-(anterior-)mostnucleisomewhatdevoidofdl.Asimilarmechanism,inwhichtheremovalrate ofBMPligandssurpassestherateofBMPfluxtothedorsalmidline,hasbeensuggestedtoexplainthe computationally-predictedsplit-peakphenotypefortheBMPsystemintheearlyembryo(Umulisetal., 2006). Theshuttlingmechanismhasbeenproposedinseveralothercontextsandinmultipleorganisms(Benzvietal.,2008;Eldaretal.,2002;Haskel-Ittahetal.,2012;Marquésetal.,1997;Mizutanietal.,2005; Shimmietal.,2005).Forexample,duringthesamestageofDrosophiladevelopment,shuttlingofthe BMPligandsDppandScwthroughtheactionofSogandTsgisresponsiblefortheconcentrationofBMP signalingtoanarrow,intensestripeatthedorsalmidlineoftheembryo.Asimilarmechanismisalso presentinBMPsignalinginvertebrateembryostospecifytheDVaxis.Inlightoftheseexamples,we speculatethatthismaybeamoreprevalentmechanismofmorphogengradientformationthan previouslyappreciated. ExperimentalProcedures Flylines ywflieswereusedaswildtype.dl-paGFP,dl-mGFP,anddl-dVenuswerecreatedbyBACrecombineering. Forliveimaging,fliescarryingdl-paGFPwerecrossedtofliescarryingH2A-RFPonthesecond chromosome(BS#23651).dl/+flieswerecreatedbycleaningupdl1viatwohomologousrecombinations withywtogeneratedl1.2.5.cact/+;cact-fulllacZ25flieswereobtainedfromDavidStein(Fernandezet al.,2001).dl-lacZflieswereobtainedfromShubhaGovind(Govindetal.,1992).dl-dGFPanddl-mVenus fliesandtheoriginalBACsusedtocreatethemwereobtainedfromAngelaStathopolous.dl-mGFP,dl1.2.5 flieswerecreatedbyhomologousrecombination.Presenceofdl-mGFPwasconfirmedbyw+.Presence ofdl1.2.5wasconfirmedviasequencingbyGENEWIZ,ResearchTrianglePark,NC.egfrt1/CyOflieswere obtainedfromtheBloomingtonStockCenter(#2079).Fliescarryingthebcd>toll10b:bcd3’UTRconstruct wereobtainedfromAngelaStathopolous.TheplasmidcarryingFRT-stop-FRThsp83>toll10b:bcd3’UTR wasalsoobtainedfromAngelaStathopoulos.ToremovetheFRT-stop-FRTcassette,wecrossedflies carryingthisconstructintoalinecarryinghsFLPontheXchromosome(BS#8862).Toremovethenative DVdlgradient,fliescarryingthetoll10b:bcd3’UTRconstructwerecrossedintoagd7background.See SupplementalExperimentalProceduresformoredetails. 8 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. BACRecombineering WefollowedProtocol3oftheNCIatFrederickRecombineeringwebsite (http://ncifrederick.cancer.gov/research/brb/recombineeringinformation.aspx)togeneratedl-paGFP, dl-mGFP,anddl-dVenusinpACMAN(Venkenetal.,2006).NEB’sproofreadingQ5DNApolymerasewas usedtoamplifysequencesatahighlevelofauthenticity.UsingaGalKselectionprotocol(Warminget al.,2005),singleaminoacidmutationswereintroducedatresidue206inpreviouslyestablishedBACs (Reevesetal.,2012)togeneratedl-mGFP(A206K)anddl-dVenus(K206A).dl-paGFPwascreatedby addingtheopenreadingframeofpaGFP(Addgeneplasmid#11911),(PattersonandLippincott-Schwartz, 2002)inframetothe3’endofthedlopenreadingframeinadlrescueconstruct(Reevesetal.,2012)in pACMAN(Venkenetal.,2006)usinga6x-Glylinker.SeeTableS1foralistofprimersused. FluorescentinsituHybridization Allembryoswereagedto2-4hours,exceptfor“young”ywembryos,whichwereagedto0-2hours, thenfixedin37%formaldehydeaccordingtostandardprotocols(Kosmanetal.,2004).Acombination fluorescentinsituhybridization/fluorescentimmnuostainingwasperformedaccordingtostandard protocols(Kosmanetal.,2004).Briefly,fixedembryoswerewashedinTween/PBSbuffer,then hybridizedwithanti-senseprobesat55oCovernight.Theembryoswerethenwashedandincubated withprimaryantibodiesat4oCovernight.Next,theembryoswerewashedandincubatedfor1-2hrs withfluorescentsecondaryantibodiesatroomtemperature.Theembryoswerethenwashedand storedin70%glycerolat-20oC.Embryoswereimagedwithinonemonthofcompletingtheprotocol. SeeSupplementalExperimentalProceduresformoredetails. MountingandImagingofFixedSamples Embryoswerecrosssectionedandmountedin70%glycerolasdescribedpreviously(CarrellandReeves, 2015).Briefly,arazorbladewasusedtoremovetheanteriorandposteriorthirdsoftheembryo,leaving acrosssectionroughly200µmlongby200µmindiameter.Thesesectionswerethenorientedsuchthat thecutsidesbecamethetopandbottom.Theywerethenimagedat20xonaZeissLSM710 microscope.15z-slices1.5µmapartwereanalyzed.EmbryoswithanAPDorsalgradientweremounted laterallyin70%glycerolusingonepieceofdouble-sidedtape(weakbicoidpromoter)ortwopiecesof double-sidedtape(stronghsp83promoter).Imagesweretakenat2.5µmintervalsfromjustabovethe topoftheembryotothedepthatwhichtheembryoreachedmaximalsizeinthexyplane,whichwas assumedtobethemidsagittalsection.Stacksrangedfrom15-25slices. DataAnalysis Imagesofembryocrosssectionswereanalyzedusingpreviouslyderivedcode(Trisnadietal.,2013). Briefly,theborderoftheembryowasfoundcomputationally,thenthenucleiweresegmentedusinga localthresholdingprotocol.Theintensityofdlineachsegmentednucleuswascalculatedastheratio betweentheintensityinthedlchanneldividedbytheintensityinthenuclearchannel.Theintensityof mRNAexpressionwascalculatedasaverageintensitywithinanannulusroughly18μmwidearoundthe perimeteroftheembryo.Adescriptionoftheimageanalysisofwholemountembryoscanbefoundin theSupplementalExperimentalProcedures. AlldlgradientswerefittoaGaussian,andthesefitswereusedtodeterminethewidthparameter,σ. Normalizedintensityplotsweregeneratedbyfittingeachembryo’sdatatoitsownGaussianby 9 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. subtractingtheBvalueand70%oftheMvalue,thendividingbytheAvalue.(X=(x–B–0.7M)/A)).The averagenormalizedintensityplotwasgeneratedbyaveragingtheplotsofallembryosinthespecified genotype. Statisticalsignificancewascalculatedusingtwo-tailedhomoscedastict-tests. ActivatingpaGFPinLiveEmbryos Embryosweredechorionatedbyhandorfor30sin100%bleach.Theywerethenmountedlaterallyona slidecoatedwithheptaneglue.Deionizedwaterwasusedasamountingmedium.Twopiecesof double-sidedtapewereusedtoattachthecoverslip.Imagesweretakenusinga40xwaterimmersion objectiveonanLSM710confocalmicroscope.Activationbox:9000pixels(300micronsx30microns), numberofactivationpasses:10,resttime:15s,numberofcycles:40,laserpower:3%forembryo activatednearventralmidline,2.5%forembryoactivatedneardorsalmidline.A405nmlaserwasused foractivationanda488nmlaserwasusedforexcitationoftheactivatedGFP. Modelofdl/Cactinteractions Weconstructedamodelofdl/Cactinteractionsconsistingofeightdifferentialequations,representing nuclearandcytoplasmicdl,Cactanddl/Cact,aswellasactivatedTollreceptorsandTollreceptorsbound todl/Cactcomplex(Toll:dl/Cact).Theequationswerethennon-dimensionalized,resultinginamodel with20freeparameters.Interactionsweresimulatedduringnuclearcycles10-14,aspreviously published(O’ConnellandReeves,2015).OptimizationwasperformedinMATLAB®usinganevolutionary optimizationalgorithmwithstochasticranking(RunarssonandYao,2000,2005),yielding114parameter setsthatweregenerallyconsistentwiththespatiotemporaldatapublishedbyReevesetal.(2012). FurtherdetailonmodelequationscanbefoundintheSupplementaryExperimentalProcedures. Addingtogetherthemodelequationsfordlanddl/Cact(Eqns1-4inSupplementalExperimental Procedures),weobtainthefollowingequationfortotaldl: 𝜕 1 𝑑𝑙 23* 𝜕 1 𝑑𝑙/𝐶𝑎𝑐𝑡 𝜕 𝑑𝑙 *+* = 𝐷/0 + 𝐷 /05672* 𝜕𝑡 𝜕𝑥 1 𝜕𝑥 1 23* where𝐷; representsthediffusioncoefficientforspecies𝑖. AuthorContributions SNC,AEA,andSMSconductedexperiments.MDOconductedmathematicalmodeling.SNC,MDO,and GTRwrotethemanuscript.SNCandMDOgeneratedfigures. Acknowledgments ThankstoAngelaStathopoulosandLeslieDunipaceforprovisionofdl-dGFPanddl-mVenusfliesand BACs,bcd>toll10b:bcd3'UTRflies,andtheDNAusedtogeneratehsp83>toll10b:bcd3'UTRflies.Thanks toImmunoReagentsforprovidingtheirgoatant-biotinantibody.WearegreatfultoAlexThomasfor generatingthesna-biotinprobe.ThankstoParvGondaliaforgeneratingthesna-fitcprobe.Thanksto DavidSteinforprovidingthecact-lacZflyline.WearegratefultoShubbaGovindforprovidingthedllacZflies.Duringthiswork,SNCandGTR(partially)weresupportedbyaCAREERawardfundedbythe UnitedStatesNationalScienceFoundation(www.nsf.gov),grantnumberCBET-1254344;MDOwas 10 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. supportedbyaGraduateAssistanceinAreasofNationalNeedFellowshipinComputationalScience fromtheUnitedStatesDepartmentofEducation(www.ed.gov),grantnumberP200A120047. 11 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. References Ben-zvi,D.,Shilo,B.,Fainsod,A.,andBarkai,N.(2008).ScalingoftheBMPactivationgradientin Xenopusembryos.Nature453. Carrell,S.N.,andReeves,G.T.(2015).Imagingthedorsal-ventralaxisofliveandfixedDrosophila melanogasterembryos.InTissueMorphogenesis,C.M.Nelson,ed.(NewYork,NY,NY:SpringerNew York),pp.63–78. Delotto,R.,Delotto,Y.,Steward,R.,andLippincott-schwartz,J.(2007).Nucleocytoplasmicshuttling mediatesthedynamicmaintenanceofnuclearDorsallevelsduringDrosophilaembryogenesis. Development4241,4233–4241. Eldar,A.,Dorfman,R.,Weiss,D.,Ashe,H.,Shilo,B.,andBarkai,N.(2002).RobustnessoftheBMP morphogengradientinDrosophilaembryonicpatterning.Nature419. Fernandez,N.Q.,Grosshans,J.,Goltz,J.S.,andStein,D.(2001).Separableandredundantregulatory determinantsinCactusmediateitsdorsalgroupdependentdegradation.Development128,2963–2974. Garcia,M.,Nahmad,M.,Reeves,G.T.,andStathopoulos,A.(2013).Size-dependentregulationofdorsal– ventralpatterningintheearlyDrosophilaembryo.Dev.Biol.381,286–299. Govind,S.,Whalen,A.M.,andSteward,R.(1992).Invivoself-associationoftheDrosophilarel-protein dorsal.Proc.Natl.Acad.Sci.U.S.A.89,7861–7865. Govind,S.,Drier,E.,Huang,L.H.,andSteward,R.(1996).RegulatednuclearimportoftheDrosophilarel proteindorsal:structure-functionanalysis.Mol.Cell.Biol.16,1103–1114. Haskel-Ittah,M.,Ben-Zvi,D.,Branski-Arieli,M.,Schejter,E.D.,Shilo,B.-Z.,andBarkai,N.(2012).SelfOrganizedShuttling:GeneratingSharpDorsoventralPolarityintheEarlyDrosophilaEmbryo.Cell150, 1016–1028. Huang,A.M.,Rusch,J.,andLevine,M.(1997).AnanteroposteriorDorsalgradientintheDrosophila embryo.GenesDev.11,1963–1973. Kanodia,J.S.,Rikhy,R.,Kim,Y.,Lund,V.K.,DeLotto,R.,Lippincott-Schwartz,J.,andShvartsman,S.Y. (2009).DynamicsoftheDorsalmorphogengradient.Proc.Natl.Acad.Sci.U.S.A.106,21707–21712. Kosman,D.,Mizutani,C.M.,Lemons,D.,Cox,W.G.,McGinnis,W.,andBier,E.(2004).Multiplex detectionofRNAexpressioninDrosophilaembryos.Science305,846. Liberman,L.M.,Reeves,G.T.,andStathopoulos,A.(2009).QuantitativeimagingoftheDorsalnuclear gradientrevealslimitationstothreshold-dependentpatterninginDrosophila.Proc.Natl.Acad.Sci.U.S. A.106,22317–22322. Marqués,G.,Musacchio,M.,Shimell,M.J.,Wünnenberg-Stapleton,K.,Cho,K.W.Y.,andO’Connor,M.B. (1997).ProductionofaDPPActivityGradientintheEarlyDrosophilaEmbryothroughtheOpposing ActionsoftheSOGandTLDProteins.Cell91,417–426. Mavrakis,M.,Rikhy,R.,andLippincott-schwartz,J.(2009).PlasmaMembranePolarityand CompartmentalizationAreEstablishedbeforeCellularizationintheFlyEmbryo.Dev.Cell16,93–104. Meinhardt,H.(2004).Differentstrategiesformidlineformationinbilaterians.Nat.Rev.Neurosci.5, 502–510. Mizutani,C.M.,Nie,Q.,Wan,F.Y.M.,Zhang,Y.,Vilmos,P.,Sousa-Neves,R.,Bier,E.,Marsh,J.L.,and 12 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Lander,A.D.(2005).FormationoftheBMPActivityGradientintheDrosophilaEmbryo.Dev.Cell8,915– 924. Moussian,B.,andRoth,S.(2005).DorsoventralaxisformationintheDrosophilaembryo--shapingand transducingamorphogengradient.Curr.Biol.15,R887–R899. O’Connell,M.D.,andReeves,G.T.(2015).ThePresenceofNuclearCactusintheEarlyDrosophilaEmbryo MayExtendtheDynamicRangeoftheDorsalGradient.PLOSComput.Biol.11,e1004159. Patterson,G.H.,andLippincott-Schwartz,J.(2002).AphotoactivatableGFPforselectivephotolabelingof proteinsandcells.Science297,1873–1877. Reeves,G.T.,andStathopoulos,A.(2009).Gradeddorsalanddifferentialgeneregulationinthe Drosophilaembryo.ColdSpringHarb.Perspect.Biol.1,a000836. Reeves,G.T.,Trisnadi,N.,Truong,T.V,Nahmad,M.,Katz,S.,andStathopoulos,A.(2012).Dorsal-ventral geneexpressionintheDrosophilaembryoreflectsthedynamicsandprecisionofthedorsalnuclear gradient.Dev.Cell22,544–557. Roth,S.,andSchüpbach,T.(1994).Therelationshipbetweenovarianandembryonicdorsoventral patterninginDrosophila.Development120,2245–2257. Roth,S.,Stein,D.,andNüsslein-Volhard,C.(1989).Agradientofnuclearlocalizationofthedorsal proteindeterminesdorsoventralpatternintheDrosophilaembryo.Cell59,1189–1202. Runarsson,T.P.,andYao,X.(2000).Stochasticrankingforconstrainedevolutionaryoptimization.IEEE Trans.Evol.Comput.4,284–294. Runarsson,T.P.,andYao,X.(2005).Searchbiasesinconstrainedevolutionaryoptimization.IEEETrans. Syst.ManCybern.PartCAppl.Rev.35,233–243. Schüpbach,T.(1987).Germlineandsomacooperateduringoogenesistoestablishthedorsoventral patternofeggshellandembryoinDrosophilamelanogaster.Cell49,699–707. Sen,J.,Goltz,J.S.,Stevens,L.,andStein,D.(1998).SpatiallyRestrictedExpressionofpipeinthe DrosophilaEggChamberDefinesEmbryonicDorsal–VentralPolarity.Cell95,471–481. Shilo,B.-Z.,Haskel-Ittah,M.,Ben-Zvi,D.,Schejter,E.D.,andBarkai,N.(2013).Creatinggradientsby morphogenshuttling.TrendsGenet.29,339–347. Shimmi,O.,Umulis,D.,Othmer,H.,andConnor,M.B.O.(2005).FacilitatedTransportofaDpp/Scw HeterodimerbySog/TsgLeadstoRobustPatterningoftheDrosophilaBlastodermEmbryo.120,873– 886. Trisnadi,N.,Altinok,A.,Stathopoulos,A.,andReeves,G.T.(2013).Imageanalysisandempirical modelingofgeneandproteinexpression.Methods62,68–78. Umulis,D.M.,Serpe,M.,O’Connor,M.B.,andOthmer,H.G.(2006).Robust,bistablepatterningofthe dorsalsurfaceoftheDrosophilaembryo.Proc.Natl.Acad.Sci.U.S.A.103,11613–11618. Venken,K.J.T.,He,Y.,Hoskins,R.A.,andBellen,H.J.(2006).P[acman]:aBACtransgenicplatformfor targetedinsertionoflargeDNAfragmentsinD.melanogaster.Science314,1747–1751. Warming,S.,Costantino,N.,Court,D.L.,Jenkins,N.A.,andCopeland,N.G.(2005).Simpleandhighly efficientBACrecombineeringusinggalKselection.NucleicAcidsRes.33,e36. 13 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Zacharias,D.a,Violin,J.D.,Newton,A.C.,andTsien,R.Y.(2002).Partitioningoflipid-modified monomericGFPsintomembranemicrodomainsoflivecells.Science296,913–916. 14 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. FigureLegends Figure1:DorsalAccumulatesontheVentralSideoftheEmbryo.(A)Currentresearchsuggeststhatdl doesnotrapidlydiffusefromnucleustonucleusandthereforeitislocallydepletedfromthearea aroundeachnucleus.(B)Weproposethattotaldlaccumulatesattheventralmidlineduetodiffusion. "TollDomain"indicatesthatactiveTollreceptorsarebeingproducedontheventralsideoftheembryo. (C)and(D)Cross-sectionsofwildtypeembryos,stainedfordlandnuclei.(C)dlisequallydistributed throughoutwhentheembryoisyoung.(D)Totaldlhasaccumulatedattheventralmidlinebylatenc14. Compareto(B).(E)and(F)dl-paGFPresultsindicatethatdldiffusesthroughouttheembryo.Activation areainthewhitebox.Anteriortotheleft.(EandE’)Activationneartheventralmidlineapprox.(E)3 minutesand(E’)90minutesafterfirstactivation.(FandF’)Activationnearthedorsalmidlineapprox.(F) 3minutesand(F’)90minutesafterfirstactivation.SeealsoMoviesS1andS2. 15 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure2:TheModelPredictsaShuttlingMechanismandMutantPhenotypes.(A)Thepreviously publishedmodelofdlgradientdynamicsinwhichtheactiveTollregionwasmodeled phenomenologically(catalyticactivitylabeled𝑓(𝑥)).(B)Expandedmodelusedinthecurrentstudy. ActiveTollreceptorsareexplicitlyaccountedfor(producedatarate𝑓(𝑥)).(C)Evolutionaryoptimization favorshighvaluesfordl/Cactdiffusion(dimensionlessparameterλdl/Cact)andlowvaluesforfreedl diffusion(λdl).(D)Themodelpredictsdl/Cactdiffusiontowardtheventralmidline(negativevalues)isfar greaterthanfreedldiffusionawayfromtheventralmidline(positivevalues).(E-G)Perturbingasetof representativeparameterstoreducerateofshuttlingresultsinahallmarkprogressionofphenotypes. Thedlgradientfirstwidens(widthparameterσincreases),thenflattensatthepeak,thenobtainsasplit peakbydecreasingthedosageofdl(E),increasingdl/Cactdiffusion(F),andincreasingthewidthofthe activeTolldomain(G).Arrowsindicatedirectionofchangeupondecreasingtherateofshuttling.(H-J) Illustrationsofthehallmarkprogressionofphenotypesassociatedwithshuttling:thepeakexpands(H), thenflattens(I),thensplitsintotwopeaks(J).Arrowsindicatethedirectionoffluxoftotaldl. 16 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure3:EmbryosfromMothersHeterozygousfordlHaveaDifferent-ShapeddlGradient.(A)CrosssectionofaNC14embryofromamotherheterozygousfordlstainedfordl.(B)Plottednuclearintensity ofembryoin(A)asafunctionofDVcoordinate(eachbluedotisonenucleus).ErrorbarsindicateSEM ofnuclearintensity.Pinkcurveisthesmootheddata;blackcurveisafitofthedatatoaGaussian.(C) Boxplotofgradientwidths(σ)forembryosfrommotherswith2(wt)and1(dl/+)copiesofdl.Hereand elsewhereinthepaper,boxesindicateinterquartilerange(IQR):fromthe25thtothe75thpercentileof thedata,whiskersextendamaximumof1.5timesthewidthoftheIQRfromthebox,andoutliers(plus signs)aredefinedaspointsthatlieoutsidethewhiskers.Numbersindicatesamplesize.Asterisk indicatesstatisticalsignificanceofp<10-4.(D)Normalizedaverageplotofdlgradientinembryosfrom (C).SeealsoFigureS1. 17 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure4:DecreasingDiffusionofdl/CactIncreasestheWidthofthedlGradient.(A)Normalizedaverage plotsofthedlgradientinembryoswith2copiesofwtdl(wt),1copyofwtdland1copyofdl-mGFP (mGFP),1copyofwtdland1copyofdl-dGFP(dGFP),and1copyofdl-dGFP(1xdGFP).(B)Boxplotof gradientwidths(σ)forembryosshownin(A)and(F).(C)and(C’)snadomaininwt(C)and1xdGFP(C’) embryos.Grayarrowheadsindicatethewtdomainofsna;whiteindicatethesnadomainin1xdGFP embryo.(D)Normalizedaverageplotsofthesnaintensityinwtand1xdGFPembryos.(E)Boxplotof snadomainwidthforembryosshowninD.(F)Normalizedaverageplotsofthedlgradientinembryos frommotherswith2copiesofwtcact(wt)and1copyofwtcactand1copyofcact-lacZ(cact-lacZ). Insetshowsmoreclearlythatcact-lacZisslightlywiderthanwt.Asterisksindicatestatisticalsignificance (*p<0.02;**p<10-3;***p<10-11)fromwtunlessotherwisenoted.Numbersindicatesamplesize.Seealso FigureS2. 18 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure5:TheHypomorphicAlleleegfrt1SignificantlyWidensthedlGradient.(A)Cross-sectionofan embryofromamotherhomozygousforegfrt1stainedfordl.Bracketsindicatepeaksofnucleardl.(B) Plottednuclearintensityofembryoin(A)asafunctionofDVcoordinate(eachpinkdotisonenucleus). Theshapehaschangedsignificantlyfromwildtype,astheGaussiancurve(black)doesnotrepresentthe gradientwell.ErrorbarsindicateSEMofnuclearintensity.(C)Boxplotofgradientwidths(σ)for embryosfrommotherswith0(wt),1(t1/+),and2(t1)copiesofegfrt1.Numbersindicatesamplesize. Asterisksindicatestatisticalsignificance(*p<0.01,**p<10-8)fromwt.(D)Normalizedaverageplotofdl gradientinembryosfrom(C). 19 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure6:AnEctopic,Anterior-PosteriorDorsalGradientExhibitsShuttlingPhenomena.(A)and(A’)dl andsnaexpressioninawildtypeembryo.(B)and(B’)dlandsnaexpressioninanembryowithan anteroposteriorgradientofdldrivenbythebcdpromoterinadditiontothewildtypeDVgradient. Whitearrowheadsindicateanarrowingofeachdomainat~30%EL.(C)and(C’)A2-Dversionofthe modelfeaturingAPandDVactiveTolldomainsexhibitsacompetingsinkphenotype,wherealowpoint inthedlgradientleadstoagapinsnaexpression.(D)and(D’)dlandsnaexpressioninanembryowith ananteroposteriorgradientofdldrivenbythehsp83promoterandtheDVgradientabolishedbya homozygousmutationingd.(E)Plotofdlandsnadomainsfromtheembryoin(D).Eachgreendotsis onenucleusandtheblackcurveisasmoothingofthedldata.Embryoimagesaremaximalintensity projections.SeealsoFigureS3. 20 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure7:TheShuttlingMechanismExplainstheFlatandDoublePeakPhenomena,IllustratedinThree Ways:(A)slowingdl/Cactdiffusion(coloredarrows,fromgreentored),(B)halvingthedosageofdl (bluedots),and(C)expandingtheTolldomain(fromgreentored).Cartoonrepresentativeofembryo crosssection.Arrowsindicatediffusionofdl/Cactcomplex,Tolldomainhighlightedinorange,nuclear concentrationindicatedbyblue-pinkgradient. 21 Figure 1 Click here to download Figure Figure bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. 1.tiff Figure 2 Click here to download Figure Figure bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. 2.tiff Click here to download Figure Figure 3.tiff Figure 3 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Click here to download Figure Figure 4.tiff Figure 4 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Click here to download Figure Figure 5.tiff Figure 5 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 6 Click here to download Figure Figure bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission. 6.tiff Click here to download Figure Figure 7.tiff Figure 7 bioRxiv preprint first posted online Jun. 3, 2016; doi: http://dx.doi.org/10.1101/057091. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.
© Copyright 2026 Paperzz