Contact-dependent cell-cell communications drive morphological … · Contact-dependent cell-cell communications drive morphological invariance during ... Integrated with the known

1

Contact-dependentcell-cellcommunicationsdrivemorphologicalinvarianceduringascidianembryogenesis

LéoGuignard1,2,3*,†,Ulla-MajFiuza1,4*,†,BrunoLeggio1,2,5,EmmanuelFaure1,2,5,6,JulienLaussu1,LarsHufnagel4,#,GrégoireMalandain7,#,ChristopheGodin2,5*,#,$andPatrickLemaire1,5#,$

1)CRBM,UniversitédeMontpellier,France

2)Inriaproject-teamVirtualPlants,CIRAD,INRA,UniversitédeMontpellier,France

3) JaneliaResearchCampus,HowardHughesMedical Institute,19700Helixdrive,Ashburn,VA,USA

4)EuropeanMolecularBiologyLaboratory,CellBiologyandBiophysicsUnit,Meyerhofstrasse1,69117,Heidelberg,Germany

5)InstitutdeBiologieComputationnelle,IBC,UniversitédeMontpellier,France

6)IRIT,CNRS,INPT,ENSEEIHT,UniversitésdeToulouseIetIII,France

7)UniversitéCôted'Azur,Inria,CNRS,I3S,France

†:Equalcontribution,#:Correspondingauthors,$:Equalcontribution.

*Currentaddresses:LG:JaneliaResearchCampus,HowardHughesMedicalInstitute,19700Helixdrive, Ashburn, VA, USA; UMF: EMBL, Meyerhofstrasse 1, D-69117 Heidelberg, Germany; EF:TeamVORTEX,InstitutdeRechercheenInformatiquedeToulouse(IRIT,UMR5505,CNRS-INPT-UniversitiesToulouseIandIII),ENSEEIHT,2,rueCamichel,BP7122,F-31071ToulouseCedex7,France;CG:MosaicInriateam,InriaRhône-AlpesandRDPResearchUnitLyon,France

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

2

ABSTRACT(150words):

Canalization of developmental processes ensures the reproducibility and robustness ofembryogenesis within each species. In its extreme form, found in ascidians, earlyembryonic cell lineages are invariant between embryos within and between species,despite rapid genomic divergence. To resolve this paradox, we used live light-sheetimagingtoquantifyindividualcellbehaviorsindigitalizedembryosandexploretheforcesthat canalize theirdevelopment.Thisquantitativeapproach revealed that individualcellgeometries and cell contacts are strongly constrained, and that these constraints aretightly linked to the control of fate specification by local cell inductions. While invertebratesligandconcentrationusuallycontrolscellinductions,wefoundthatthisroleisfulfilled in ascidians by the area of contacts between signaling and responding cells.Weproposethatthedualitybetweengeometricandgeneticcontrolofinductionscontributesto the counterintuitive inverse correlation between geometric and genetic variabilityduringembryogenesis.INTRODUCTIONWithin each animal species, embryonic development is highly reproducible, ensuring theproductionofacomplexorganismwithpreciselyarrangedandshapedorgansandtissues.Thisconstancy of embryogenesis against genetic polymorphism and fluctuating environmentalconditions is critical for the perpetuation of the species, and has been referred to asdevelopmentalcanalization(1).Although tissue-scale embryonic reproducibility results from the careful orchestration of cellbehaviors during development (2), it does not imply that individual cells behave reproduciblyfromembryotoembryo.Rather,tissue-levelinvarianceisinmostspeciesanemergentproperty,which results from the averaging of the variable (3) or even stochastic (4) behaviors ormechanicalproperties(5,6)ofindividualcells.Someinvertebratespecies,includingmostnematodes(7)andascidians(8),exemplifyanextremeformofcanalization.Theydevelopinsuchahighlystereotypedmannerthatthepositionandfateofindividualembryoniccells,aswellastheorientationandtimingoftheirdivisions,showverylittle variabilitybetween individuals, leading to essentially invariant embryonic cell lineages (9,10).Thiscellularreproducibilityofwildtypedevelopmentisrobusttotheunusuallyhighlevelofgeneticpolymorphismfoundinnematodesandascidians(11–13).Earlyascidiancelllineagesandembryogeometrieshaveevenremainedessentiallyunchangedsincetheemergenceofthegroup,around 500 million years ago, despite extensive genomic divergence (8, 14). Because of thestereotypy of their early development and remarkable ability to buffer genetic divergence,ascidians constitute an attractive system to understand the mechanistic origin of extremecanalization.CanalizationhasmostlybeenanalyzedthroughtheprismofthedevelopmentalGeneRegulatoryNetworks(GRNs)thatdriveandcoordinatedevelopment.Thecurrentviewisthatreducinggeneexpressionvariability isa key featureof canalization achieved through theuseof specificGRNarchitectures (15–18) or through the folding or stabilization of signal transductionproteins byspecificchaperones(19).Generegulatorynetworksare,however,muchlessevolutionaryrobustthanmorphologiesinbothascidians(14)andnematodes(20).Additionally,intheascidianCionaintestinalis,themajorityofgenesshowvariablematernalexpressionbetweenindividuals(21).As

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

3

extreme canalization of embryonicmorphologies is observed despite variable gene expressionandgeneregulatorynetwork,itisunlikelytobeexplainedsolelybythecanalizationofregulatorygeneexpression.Whatelsecouldexplaincell-scalegeometricinvariance?Thelarvaeofmostspecieswithinvariantcelllineagesaremadeofonlyafewhundredstothousandsofcells.Thisreducedcomplexitycouldfavor stereotypy, but it is not per se sufficient to explain it: the marine nematode, E. brevis,developswithsimilarcellnumbersasC.elegans,butwithvariable,indeterminatecelllineages(8).Stereotypydoes alsonotresult from theuseof qualitativelydifferentdevelopmental strategiesfromthoseofotheranimals:whiletheprecisepartitioningduringcelldivisionofautonomously-acting localizedmaternal developmentwas initially thought to drive stereotyped development(23,24),cellcommunicationisnowrecognizedasequallyimportantinembryoswithinvariantorvariablecelllineages(25).In this study, we test the hypothesis that the combination of short-range cell communicationevents involved in ascidian fate specification and reduced cell numbers, strongly constrainsembryonic geometries. For this, we first developed experimental and algorithmic methods toperform long-termmulti-view live imagingofdevelopingascidiansandtoautomaticallyextracttheprecisegeometriesofcellsandcell-cellcontactswithunsurpassedaccuracyandusedthesetoestimatetheirdegreeofvariabilitybetweenembryos.Wethencombinedcellcontactgeometrieswith an atlas of signaling gene expression to build a computational model of cell inductions,whoserobustnesstogeometricandgeneticvariationswestudied.Ourresultsrevealthatcellfatespecification by surface-dependent inductions imposes strong local constraints on the areas ofcontact between communicating cells. This strongly canalizes embryonic geometries, whileputtinglesserconstraintsontheevolutionofthegenome.Ahigh-resolutiongeometricatlasofembryoniccellshapesandinteractions.Using confocal multiview lightsheet microscopy (26), we imaged live Phallusia mammillataembryos with fluorescently-labeled plasma membranes every two minutes from 4 orthogonalviewingdirectionsandforseveralhours,withoutcompromisingtheirdevelopment.Highquality4D datasetswith isotropic spatial resolutionwere obtainedafter fusion of the four 3D imagescaptured at each time point (Figures S1, S2). They extend from the 64-cell stage to the initialtailbud (Embryo1) and late neurula (Embryo2) stages respectively, covering two majormorphogenetic processes, gastrulation and neurulation (Supp. datasets, Video S1, Figure 1),throughupto4celldivisions.Systematic segmentation and long-term tracking of allmembrane-labeled cells of adevelopingembryo is notoriously challenging (27). Building on our previous low-throughput MARS-ALTpipeline (28)(Figure S3-6),wedevelopedanovel algorithm,ASTEC, forAdaptive SegmentationandTrackingofEmbryonicCells(Figure2A),abletofaithfullysegmentwholecellsandtrackcelllineagesoverlongperiodsoftimewithhightemporalresolution.ASTECisasinglepassalgorithm,which simultaneouslysegments and tracks cell shapesbypropagating informationover time, astrategy pioneered for nuclear labels by Amat and colleagues (29). ASTEC is initiated with amanuallycuratedsegmentationofthefirsttimepoint,anditerativelyprojectssegmentationsfromonetimepointtothenext(FiguresS7-11andSupp.information),thendetectscelldivisioneventsbetween consecutive time points. It produces a segmentation of all embryonic cells present ateachtimepoint,andforeachcell"snapshot"(theimageofacellatagiventime-point)theidentity

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

4

of itsprogenyat thenext timepoint, fromwhichglobalcell lineagesand temporaldynamicsofindividualcellshapesandphysicalcell-cellcontactscanbereconstructed.The resulting segmentation (Figure 2B, Supp. Dataset, Video S2) and global lineage trees ofEmbryo1,ASTEC1,containatotalof58454digital3Dcell"snapshots",describingthebehaviorintimeof1342individualcellsgeneratedby639celldivisionevents.Integratedwiththeknownfateofearlyblastomeres(10),thisgeometricdescriptionofadevelopmentalprogramallowstotrackwithhightemporalandspatialresolutiontheposition,geometry,contactsandlineagehistoryofeveryembryoniccell(Figures2C,S12,13,VideoS3).QualityassessmentofthesegmentationandlineageofASTEC1indicatedthat99%ofvoxelswereassignedtotherightcell(FigureS14),thatcellvolumeswereconsistentbetweenmatchingcellsintheleftandrighthalvesoftheembryoandthatthesumofthevolumesofdaughtercellscloselycorrespondedtothevolumeoftheirmother(FigureS15).Thepatternofroundingupofcellsas theyapproachedmitosisindicated thatcelldivisionsweredetectedwithatemporalaccuracyof2minutes(FigureS16,S17).Noprogrammedcell death was detected, as described in other ascidian species (30). The segmentation andtracking accuracy of the second embryo, ASTEC2, obtained by running ASTEC with the sameparameters,was comparable (see Supp. information).TheASTEC1andASTEC2datasets (Supp.Data sets) constitute to our knowledge the first systematic quantitative descriptions of thedynamicsdevelopmentincludingcell lineages,andthegeometryandinteractionsofwholecells,ratherthancellnuclei,acrossalargefractionofametazoandevelopmentalprogram.Stereotypyofascidiandevelopment.Althoughascidiansareconsideredatextbookexampleofinvariantdevelopment,fewstudieshaveattempted to gobeyondqualitativeobservations toquantify inter-embryovariability.One suchstudyindicatesthatatleastthenotochordshowsevidenceofstochasticcellintercalationduringthetailbudstages(31).Ourgeometricatlasgaveustheopportunitytoquantitativelyassessthevariability of ascidian cell lineages, cell geometries and cell neighborhoods during gastrulationandneurulation.We first compared the temporal progression in cell numbers in our two ASTEC-reconstructedembryosandinanindependentlyimaged,manually-curated,Phallusiamammillataembryowhosenuclei rather than membranes were fluorescently labeled (BioEmergences embryo) (32). Cellnumbers were remarkably consistent between embryos until the late neurula stage (7 hpf at18°C),withonlyslightdifferencesbeyond(Figure3A).Consistently,thestructureofcell lineagetrees originating from matching cells of early gastrula progenitors of ASTEC1 and theBioEmergencesembryos,orfromequivalentleftandrightcellswithineachembryo,differedbylessthan10%whencomparedwithatreeeditmetric(33)(Figures3BandS18-19).Lifespansofmatchingcellsalsoonlydifferedmarginally,by less than10%onaverage,betweenembryosorbetweentherightandlefthalvesofeachembryo(Figures3C,S21).Temporal invariance of cell lineages does not necessarily imply geometric invariance (34).WethereforenextassessedgeometricvariabilityinPhallusia.Whilethedistributionofcellvolumesata given time pointwas broad, the volume difference ofmatching cellswas inferior to10% onaverage,bothbetweenASTEC1and2andbetweenthebilateralhalvesofeachembryo(Figures3DandS20).Cellorganizationwasalsoshared.Morethan80%ofmatchingcellsinASTEC1andASTEC2sharedatleast80%oftheirphysicalneighbors(Figure3E).Mostcell-cellcontactssharedbetweenembryospersistedforthewholelifeofthecellsandmostcellskeptthesameneighborsthroughout their life (Figure S22). As a consequence, we found no evidence of individual cell

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

5

migrationorcellintercalationuptotheinitialtailbudstage.Thistopologicalinvarianceextendedtotheinvarianceofthemeasureofareasofcontactsbetweencells.In80%ofmatchingcells,theaverage variation of surface of contact to sharedneighborswas smaller than 20%, again bothbetween bilateral halves of each ASTEC embryo, and between embryos (Figures3E, S23). Thisconservationoflocalneighborhoodstranslatedintoaconservationoftheglobalspatialpositionofeachcellwithrespecttoembryonicaxes(Figure3F).Thus, cell divisions, cell geometries, cell arrangements and positions are highly reproduciblebetweenindividualPhallusiaembryosuntilatleastthelateneurulastages.Thecomparableextentof inter-embryonic andbilateral intra-embryonic variability suggests thatdevelopmental noise,ratherthangeneticvariation,isanimportantdriverofthisvariability.Asymmetriccelldivisionscouplefatespecificationtoembryoniccellgeometries.ToidentifythecausesofPhallusiadevelopmentalinvariance,wefirstconsideredthecellularbasisof embryonic morphogenesis. Embryonic morphogenesis is built from few elementary cellbehaviors(2).Intheabsenceofcellmigration/intercalation,cellgrowthandcelldeath,Phallusiamorphogenesis must be primarily driven by oriented cell division, geometrically unequaldivisions and cell shape changes. Asymmetric cell divisions, defined as divisions generatingdifferently fated daughter cells, can couple fate specification to morphogenesis as someasymmetric divisions also control the relative position in space (orienteddivisions) or volume(unequalcleavages)ofdaughtercells(35).Usingourgeometricatlas,wesetouttosystematicallyidentifyasymmetriccelldivisionsandassessedtheirgeometricconsequences.We first scanned cell lineages for a signature of asymmetric cell divisions. Using our tree editdistance,wehierarchically clustered all 64-cellPhallusiaprogenitors inASTEC1on thebasis ofthesimilarityofthecell lineagetreestheyseeded(FiguresS24,S23).Themajorityofcellswithidenticalorsimilartissuefatesclusteredtogether, indicatingthatthemitotichistoryofascidianembryonic cells is strongly correlated to their larval fates. We therefore reasoned that acomparisonofthestructureofthesublineagetreesoriginatingfromtwosistercellsmayidentifysisterswithdistinctfates,andthusoriginatingfromcandidateasymmetriccelldivisions(Figure4A).Figure4Bshowsthedistributionofdistancesbetweenthecelllineagetreesoriginatingfrom108bilateralpairsofsistercellsbetweenthe64-cellandmid-gastrulastages(timepoint124forASTEC1).Only35/108divisionsgaverisetolineagestreesdifferingbymorethan10%,asubsetincluding 19/22 known fate restriction events (Table S1 and Figure S26). This approach alsoidentified16novelasymmetriccelldivisions.14ofthesedivisionswerefoundintissuesknowntogive rise to several larval or juvenilemesodermal tissues, such as themesenchyme and trunklateralcells(TLC)(36–38),ortoshowstrongcellularheterogeneity,suchasthecentralnervoussystem(39,40) or the tailepidermis(41).Wethusused lineage treestructureasymmetryasamarkerforasymmetriccellsdivisions.Wenextexploredthegeometricalimpactofasymmetricdivisions(Figure4CandTableS1).WhileunequalcleavagesareinfrequentinPhallusia(FigureS27A,B),23/35asymmetricdivisionswerealsogeometricallyunequal.Interestingly,ingeometricallyunequaldivisions,thesmallerdaughtergenerallyhadalongerlifespan(FigureS27C),inagreementwithageneralanti-correlationfoundbetweencellvolumesandlifespansinourdataset(FigureS27D),andinC.elegans(42).UnequallydividingcellsalsosignificantlydepartedfromthedefaultHertwigrule(43)fortheorientationofthe division (Figure S27E). Co-occurrence of asymmetric divisions, geometrically unequalcleavageanddifferentiallifespanswasparticularlystronginthemid-gastrulaneuralplate(Figure

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

6

4D-F).Thus, up to the mid-gastrula stage, around 30% of all cell divisions are asymmetric and themajorityofthesedivisionscouplethefatespecificationprocesstothegeometry,relativepositionandlifespanofdaughtercells.GeometriccontrolofdifferentialcellinductionsWe next analyzed whether, conversely, cell geometry and neighborhood relationships couldimpact fate specification. In ascidians,most studied asymmetric cell divisions are triggered bycontact-dependent cell communication events, which either polarize the mother cell ordifferentially induce its daughters (44). The frequency of asymmetric divisions we detectedsuggests that contact-dependent inductions may collectively impose a global constrain onembryonic geometries, therebypromoting stereotypy.To test this idea,webuilt computationalmodelsofdifferentialcell inductionintegratinggeometricalfeaturesextractedfromtheASTEC1dataset with the expression pattern of signaling genes, which in ascidians have beensystematically determined with a cellular resolution up to the onset of gastrulation (45, 46).Simulatedcellinductionswerethenconfrontedtoagroundtruthofexperimentallycharacterizeddifferentialinductioneventsanduninducedcells(TableS2).Extracellularligandsorantagonistsforonlysixmajorsignalingpathways(FGF,Ephrin,Wnt,Bmp,Nodal and Notch) show patterned expression during the cleavage and early gastrula stages(Figure S29). As receptors and intracellular components of these pathways are maternallyexpressed, we considered them ubiquitous. All cells were therefore considered competent torespondtoallpathways,andligandavailabilitywasthelimitingfactorcontrollinginductions.Wecomputedligandavailabilitybyintegratingthepatternofexpressionoftheligandgenewiththatofitsextracellularantagonists,thepresenceofanantagonistatacellinterfacefullyblockingtheactionoftheligandatthisinterface.First,we testedwhether direct physical contact of a competent cell to a cell emitting a ligandcould be sufficient to ensure its induction, irrespective of the area of contact (Figure 5A).Wedefined the signaling state of a cell as the combination of free ligands a cell is exposed to andcompared the signaling states of sister cells to experimentally determined fate specificationdecisions(Figure5B,CandS35).Cellswere foundtorespondonaverage to5.5ligandsandtheresulting25cell signalingstates fromthe64-cell to theearlygastrulastagescouldonlyexplain4/14knowndifferentialinductionevents.Thus,asimpleinductionmodelbasedonthetopologyofphysicalcellcontactsisinsufficienttoaccountforknownasymmetriccelldivisions.The analysis of FGF-mediated ascidian neural induction (47,48) and of Notch signalingduringhaircellspecificationinthechickbasilarpapilla(49)suggestthatcellscantakeintoaccounttheareaofcontactwithligand-expressingcellsintheinterpretationofinducingcues.Thehighspatialand temporal resolution of our digital embryo made it possible to computationally test thegenerality of such a hypothesis at the level of a whole embryo. For this, we developed aquantitativecontinuousmodelbasedonthelawofmassaction(FiguresS28,S30).Throughasetofcoupleddifferentialequationsmodelingligand-receptor-effectorcascadedynamics,thismodelcomputes foreachembryoniccell,ateachstage,andforeachpathway,the fractionofactivatedintracellulareffectorasafunctionofthesurfaceofthecell,orofthepartofitsmotheritinherited,exposed toaligand fromthispathway.Under thehypothesesof themodel(Supp. Information),

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

7

the fractionofactivated intracellular effectors computed for eachpathwaywasaquasi-linearlyincreasingfunctionoftheareaofexposuretofreeligands(FigureS31).Toidentifyrelevantinductionthresholds,wedesignedasetoflogicalconsistencyrules(FiguresS32, 33) transforming this continuous information into a binarymap of differentially inducedsistercells(FigureS28).Wethenexploredtheparameterspace(FigureS34)tomaximizethefitofthe model with a set of experimentally characterized differential inductions up to the earlygastrula stage (Table S2). The best sets of parameters identified (see Supp. Information andFigure S34)were consistentwithpublishedknowledge (e.g. induction timeof8minutes for allpathways, similar as in (50) for FGF). Compared to the topological cell contact model, thisquantitative model led to a reduction of the number of signaling pathways each cell wasrespondingto(1.7onaverage),whileincreasingthenumberofsignalingstates(38betweenthe64 and early gastrula stages), which were highly correlated to differential inductions events(Figures5D,S35).Themodelcorrectlypredictedinducers forallknowninductioneventswhileover-predictingdifferential inductionsforonly12outof56(21%)symmetricdivisions(Figures6A, S36andTable S3).The fractionsof activated intracellular effectors for eachpathway, eachsistercellpairandeachstagearegiveninFiguresS37-54.This remarkable resultwas stronglydependenton thehypotheses and initial conditionsof themodel. The performance was, as expected, much degradedwhenwemimicked the topologicalcontactmodelbyuncouplingthenumberofactivatedreceptorsfromtheareaofcell-cellcontactsorupon randomizationof thepatternsof expressionof ligandand inhibitor genes(Figures6A,S55,56andTableS3).Themodelwasalsoabletorecapitulateexperimentalperturbationswhichhadnotbeenusedinitstraining.VirtualinhibitionofEphrinsignalsinthemodelphenocopiedthebiological knock-down (51–53) (Figure6A, S57andTable S3). Finally, the results of themodelwere robust to naturally occurring evolutionary changes in embryo geometry. While eggdiametersareuniformwithineachspecies,theyrangeacrossascidianspeciesfromabout100µmto nearly 1 mm. Consistent with the idea that most molecular mechanisms are evolutionarilyconserved, evenbetweendistantly-related ascidian species (54),predictionsof themodelwererobusttoa4-folduniformscalingofthesurfacesofcell-cellcontact(FiguresS58,S59).Takentogether,ourmodelofsurface-dependentdifferentialcellinductions,despiteitssimplicity,accountswith remarkable explanatory ability for early ascidian fate specification events up togastrulation.Higherrobustnesstogeneticthangeometricperturbations.Ouroriginalaimwastoshedlightontheapparentparadoxofahighlycanalizedandevolutionaryconserved embryo morphogenesis in a context of rapidly divergent genomes, gene regulatorynetworks and gene expressions.We reasoned that if cells operate close to induction thresholdvalues,smallchangesinembryogeometrymightaltertheinductionorpolarizationstatusofsomecells, thereby constraining a significant fraction of cell-cell interfaces. Indeed 32% of cells sawtheirpolarizationorinductionstatusforatleastonepathwaychangeupona+/-20%changeinthe surface of contacts they establishedwith signaling cells. This local geometric sensitivity ofinductions imposes a global constraint on cell interfaces (Figures 6B and S60), indicating thatsurface-dependent cell inductions have a strong canalizing effect on embryo geometries. Bycontrast,similarchangesinthemagnitudeofligandgeneexpressionhadamuchsmallereffectoncellinductionsandpolarizations(FigureS61).

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

8

Therelianceofascidianembryonicfatespecificationontheareasofcontactswithsignalingcellstherefore ensures both high canalization of morphogenesis and more relaxed constraints ongenomicinformation.ConcludingremarksSystematicallyandquantitativelytrackingthetemporaldynamicsofmultiplecellularproperties,includingshape,andcell interactionsinallcellsofadevelopingorganismhasbeenadreamforgenerationsofdevelopmentalbiologists.Previousdynamicatlasesofdevelopmentwithcellularresolutionhavesofarmostlybeengeneratedfromembryoswithfluorescently-labelednuclei(29,55,56),asthefewercasesoflarge-scaleautomatedsegmentationoffluorescentmembrane-labelswerenotsufficientlyaccuratetoreliablytrackthepreciseshapeofeachindividualcell(27,57).We show here that, at least in the simple transparent ascidian embryo, a high-resolutionmorphodynamicatlasofcellgeometriescanbeproduced.This quantitative atlas confirmed the extreme nature ofmorphological canalization of ascidiandevelopment, until the late neurula stage at least, and revealed the quasi-absence of cellmigration/intercalation. It alsoprovided thenecessary information tobuildamodel to test thehypothesisthatlocalsurface-dependentcellinductionscollectivelyexertstrongglobalconstraintsonembryonicgeometries,whilechangesinthelevelofexpressionofinducingligandswerebettertolerated.Weproposethatthisdualitybetweengeometricandgeneticcontroloffatespecificationmaybeageneralphenomenoninanimaldevelopment.Invertebratesandotherembryosdevelopingwithhighcellnumber,variantsofthe"Frenchflag"modelofmorphogengradientshypothesizethat:"Acellisbelievedtoreaditspositioninaconcentrationgradientofanextracellularsignalfactor,andtodetermine itsdevelopmental fate accordingly" (58). In thismodel, individual cell geometriesare considered uniform and the main quantitative information is provided by the spatial andtemporal shape of the morphogen gradient (Figure 6C). Precisely shaping such gradient, andensuringtheirrobustness, involvesthecoordinatedrecruitmentofmultiplecellularfunctionstocontroltheproduction,degradation,transportsorendocytosisofligandsandtheirreceptors(59).Evenso,theshallownessofsuchgradientmeansthatinductivecuesreceivedbydirectneighborsareverysimilar.Thecoarseinformationprovidedbythegradientissubsequentlysharpenedoverseveralhourstoformclearboundariesthroughregulatorycross-talkbetweentargettranscriptionfactors (60) or cell migration. The precise response to morphogen gradients, thus, involvessophisticated layers of regulation, which we propose strongly constrain the architecture ofregulatory networks and the evolution of the genome.Modern vertebrate genomes are indeedslowlyevolving(61).Whileneighboringcellsmayreceiveverysimilarsignals inashallowmorphogengradient, theyusuallyshareaminorityofdirectphysicalneighbors.Thesurface-dependentreadoutofligandswe described here thus ensures that the magnitude of the signals they experience differssufficiently to alleviate the necessity for subsequent transcriptional refinements (Figure 6C).Consistently,fatespecificationinascidianoccursinafewminuteswithinasinglecellcycle(62),and known gene regulatory networks in ascidians are shallow (63). Surface-dependentinductions, inaddition tobeing tolerant to changes in fluctuations in ligand concentrations arethereforelikelytoinvolvemuchfewerlayersofregulation,therebyrelaxingfurtherconstraintsongenomeevolution.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

9

Acknowledgements:This work was funded by core support from CNRS to PL, by Inria (core support and IPLMorphogenetics)toCGandGM,bytheGeneshapeproject(ANR-SYSC-018-02)toPLandCGandbytheDig-Emproject(ANR-14-CE11-0013-01)toPL,CGandGM.Supportwasalsoreceivedfromthe European Molecular Biology Laboratory (L.H.) and the EMBL Interdisciplinary PostdocProgramme underMarie Curie Actions (U.MF.). L.H. acknowledges support from the Center ofModeling and Simulation in the Biosciences (BIOMS) of the University of Heidelberg. LG wassupportedbyadoctoralcontractfromtheCBS2doctoralschooloftheUniversityofMontpellier2,by the Fondation pour la Recherche Médicale (FRM) (FDT20140931061), and by theMorphoscope2Equipexproject.UMFwassupportedby theGeneshapeproject,andby theFRM(SPF20120523969).The InstitutdeBiologieComputationelleofMontpellier (IBC)supportedEFandBL.BLandJLweresupportedbytheDig-Emproject.WethankthemembersoftheLemaire,Godin and Hufnagel teams for discussion and advice throughout this project, G. Michelin foradvice on the segmentation process, and A. McDougall (Villefranche/mer, France) for thegenerousgiftofthePH-GFPexpressionconstruct.WearegratefultoP.KellerandA.Pavlopoulos(Janeliaresearchlabs,USA),andtoF.Fagotto(CRBM,Montpellier,France)forcriticalreadingofthemanuscript.Wealso thanktheSI2C2service for ITsupportandP.RichardandM.Plays forcarefulanimalhusbandry.Authorcontributionsarelistedinthesupplementarymaterials.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

10

Figure 1: High-resolution multiview lightsheet live imaging of Phallusia mammillataembryos. A) Vegetal/dorsal views of 3D rendering at the indicated time points of the imagedembryoafterfusionoftheimagestakenalongthe4anglesofviews.B)Vegetalviewoftheembryoatthe64-cellstage(t=1).C)SagittalsectionalongtheplaneshownonB.D)Dorsalviewoftheembryoattimet=180.E)SagittalsectionalongtheplaneshownonD.A-E,anterioristothetop.Scalebar:20µm

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

11

Figure 2:ASTEC segmentation of Phallusiamammillata embryos. A) The ASTEC pipeline.Three different cases are illustrated: a non-dividing cell (yellow), a dividing cell (white), anoversegmented non-dividing cell (blue) corrected during post-correction. Formore details seeFigureS7.B)SegmentationofASTEC1at t=76.Top:view fromthedorsalside.Bottom:sagittalsection.Colorsarearbitrary.C)EvolutionofthepositionofmesodermandneuralprogenitorsinASTEC1 at the early gastrula (t=35), late gastrula (t=80) and initial tailbud (t=170) stages.Mesoderm:yellow,B-linemesenchyme;purple,secondarynotochord;green,primarynotochord;cyan,TrunkLateralCells;muscleisnotrepresented.Neuralplate:green, lateraltailnervecord;blue,dorsalanteriorneuralplate;orange,ventralanteriorneuralplate.Anterioristothetop.FormoredetailsseeFigureS13.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

12

Figure3: Stereotypyof ascidiandevelopment.A)Comparisonoftheevolutionintimeofcellnumbersbetween three individual embryos after linear temporal rescaling(see Supp. data).B)Distributionsofpairwiselineagedistancesbetweentreesoriginatedatthe112-cellstagefromtheASTEC1 and BioEmergences embryos. Within Embryo, distribution of all pairwise distances;Within tissue (epi), pairwise distances between epidermis and endoderm; (Between tissues),pairwisedistancesbetweentreesfrommatchingbilateralcells(L/Rcells)andforbothASTECandBioEmergences lineage trees (All cells). Formore details, seeFigure S19.Boxes show the first,secondandthirdquartiles,whiskers therangeup to1.5 interquartile.C-D)Distributionsof therelativedifferencesinlifespans(C)andVolumes(D)betweenmatchingcellswithinandbetweenASTEC1andASTEC2embryos.SeealsofigsS20andS21.E)Percentageofcellswith80%or50%of common physical neighbors (contact >5% of the cell surface) between matching cells inASTEC1andASTEC2embryos.SeealsoFigureS23.F)ExampleofacellwithperfectlyconservedneighborhoodinthetwoASTECembryos.Notetheconservationofthespatialpositionofthecellinbothembryos.Left:ventralview,right:lateralview.Lightgreycellsaretranslucent.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

13

Figure4:Relationshipsbetweenfaterestriction,unequalcleavageandunequallifespans.A) Example of the asymmetric division of the A8.7 andA8.8posterior neural progenitors. Thetrees seeded by the two daughters of A8.7 and A8.8 are coloreddifferently. Percentage valuesindicatethetree-editdistancebetweenthelineagesseededfromthedaughtersofA8.7andA8.8cells.Note the similarityof the lineagesoriginating from the leftand rightA7.4progenitors.B)Distributionof thecell lineagedistancebetweensistercellsbornbetweentimepointst=1andt=124. C) Venn diagram showing the overlap of cell divisions leading to high sister lineagedistance,sisterlifespanratioandsistervolumeratioforcellsgeneratedbetweentimepointst=1andt=124.D,E,F)Dorsalandlateralviewofthea-andA-linederivedneuralplatecolor-codedforlineagetreedistance(D),Volumeratios(E)andlifespanratios(F).Whitebarslinksistercells.Theidentityof9thgenerationcellsisindicatedonpanelE.Lineagetreedistancesarefeaturesofacellpair,whileratiosarefeaturesofindividualcells.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

14

Figure5:Modelsofcell-cellcontact-dependentinductions.A)Twoscenariosofcontact-dependentcellinductions.Inthe"Topologicalcellcontact"scenario,anyrespondingcell(grey)contactingaligand-emittingcell(red)isinduced.Inthe"Quantitativecellcontactareas"scenario,acellisonlyinducedifitsareaofcontactwithligand-expressingcellsissufficientlylarge.B)Color-codedfatemapofcellsatthe64-cellstage.C)"Topologicalcellcontact"scenario.Top:color-codedsignalingstates;Bottom:distributionofthenumberofpathwaysaffectingacell'sfateforcellsbornbetweentimepoints1and124.D)"Quantitativecellcontactareas"scenarioresultingfromthecomputationalmodel.Top:color-codedsignalingstates.Bottom:distributionofthenumberofpathwaysaffectingacell'sfateforcellsbornbetweentimepoints1and124.InB-D,awhitebarlinkssiblingcells.ColorsinB,Darearbitrary.SeealsoFigureS35foranequivalentanalysisatthe112-cellstage.

A

Topological Induced Induced Induced UninducedUninducedUninducedUninducedInducedQuantitative

Notochord

PrimaryMuscle

Mesenchyme

Vent.Head NP

Not restrictedEndoderm

TLC

TVC #induction events: 1149#induced cells: 216

#induction events: 459#induced cells: 172

00

255075

100

2 4 6 8 10#inducers

#cel

ls

0255075

100

0 2 4 6 8 10#inducers

#cel

ls

B C D

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

15

Figure 6:Predictions of the quantitative cell contactareasmodel.A)Percentageofknown(blue)oroverpredicted(red)cellinductioneventspredictedbythemodelinWildtype(WT),andin situationsof ligandexpression randomization, constantavailable receptor situation andEphreceptor mutation. B) Visualization of the 112-cell stage cell-cell interfaces constrained bydifferential cell inductions. See also Figure S60 for equivalent data at the 64-cell stage. C)Conceptualmodelsofvertebrateandascidianembryoniccellinductions.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

16

Supplementaryvideos:VideoS1:3Drenderingofan intensity imageofdevelopingEmbryo1after fusion.Vegetalview.Anterioristothetop.VideoS2:VegetalviewoftheASTEC1segmentedembryo.Colorcodearbitrary.Notetheshapeofclonesproducedfromindividual64-cellprecursors.Video S3: Vegetal view (left) and side view through a sagittal section (right) of the ASTEC1segmented and fate colored Phallusia mammillata embryo. Anterior is to the top. Dark red :unrestrictedmesoderm, endoderm, andneuralprogenitors;Darkgrey: endoderm;Purple:TLC;Light beige: primary notochord; Light grey: secondary notochord; Dark beige: TVC; Pink: tailmuscle;White:mesenchyme;Darkblue:tailepidermis;Brightred:headepidermis;Salmon,lightgreen,lightblue,darkgreen:differentregionalizedneuralplateprogenitors;Goldenyellow:germline.

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

17

References:1. C.H.Waddington,canalizationofdevelopmentandtheinheritanceofacquiredcharacters.Nature.150,563–565(1942).2. T.Lecuit,L.L.Goff,Orchestratingsizeandshapeduringmorphogenesis.Nature.450,189(2007).3. L.Hongetal.,VariableCellGrowthYieldsReproducibleOrganDevelopmentthroughSpatiotemporalAveraging.Dev.Cell.38,15–32(2016).4. R.J.Johnston,C.Desplan,StochasticMechanismsofCellFateSpecificationthatYieldRandomorRobustOutcomes.Annu.Rev.CellDev.Biol.26,689–719(2010).5. M.Uyttewaaletal.,MechanicalStressActsviaKatanintoAmplifyDifferencesinGrowthRatebetweenAdjacentCellsinArabidopsis.Cell.149,439–451(2012).6. M.vonDassow,L.A.Davidson,Physicsandthecanalizationofmorphogenesis:agrandchallengeinorganismalbiology.Phys.Biol.8,045002(2011).7. B.Goldstein,OntheevolutionofearlydevelopmentintheNematoda.Philos.Trans.R.Soc.Lond.B.Biol.Sci.356,1521–1531(2001).8. P.Lemaire,Evolutionarycrossroadsindevelopmentalbiology:thetunicates.Dev.Camb.Engl.138,2143–2152(2011).9. J.E.Sulston,E.Schierenberg,J.G.White,J.N.Thomson,TheembryoniccelllineageofthenematodeCaenorhabditiselegans.Dev.Biol.100,64–119(1983).10. H.Nishida,Celllineageanalysisinascidianembryosbyintracellularinjectionofatracerenzyme.III.Uptothetissuerestrictedstage.Dev.Biol.121,526–41(1987).11. M.L.Nydam,R.G.Harrison,PolymorphismanddivergencewithintheascidiangenusCiona.Mol.Phylogenet.Evol.56,718–726(2010).12. A.Dey,C.K.W.Chan,C.G.Thomas,A.D.Cutter,MolecularhyperdiversitydefinespopulationsofthenematodeCaenorhabditisbrenneri.Proc.Natl.Acad.Sci.U.S.A.110,11056–11060(2013).13. E.M.Leffleretal.,RevisitinganOldRiddle:WhatDeterminesGeneticDiversityLevelswithinSpecies?PLOSBiol.10,e1001388(2012).14. A.Stolfietal.,Divergentmechanismsregulateconservedcardiopharyngealdevelopmentandgeneexpressionindistantlyrelatedascidians.eLife.3,e03728(2014).15. Manuetal.,CanalizationofGeneExpressionintheDrosophilaBlastodermbyGapGeneCrossRegulation.PLOSBiol.7,e1000049(2009).16. S.E.Lott,M.Kreitman,A.Palsson,E.Alekseeva,M.Z.Ludwig,CanalizationofsegmentationanditsevolutioninDrosophila.Proc.Natl.Acad.Sci.104,10926–10931(2007).17. A.Raj,S.A.Rifkin,E.Andersen,A.vanOudenaarden,Variabilityingeneexpressionunderliesincompletepenetrance.Nature.463,913(2010).18. M.L.Siegal,A.Bergman,Waddington’scanalizationrevisited:Developmentalstabilityandevolution.Proc.Natl.Acad.Sci.99,10528–10532(2002).19. S.Rutherford,Y.Hirate,B.J.Swalla,TheHsp90Capacitor,DevelopmentalRemodeling,andEvolution:TheRobustnessofGeneNetworksandtheCuriousEvolvabilityofMetamorphosis.Crit.Rev.Biochem.Mol.Biol.42,355–372(2007).20. M.-A.Félix,M.Barkoulas,Robustnessandflexibilityinnematodevulvadevelopment.TrendsGenet.28,185–195(2012).

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

18

21. T.Matsuoka,T.Ikeda,K.Fujimaki,Y.Satou,Transcriptomedynamicsinearlyembryosoftheascidian,Cionaintestinalis.Dev.Biol.384,375–385(2013).22. D.A.Voronov,Y.V.Panchin,CelllineageinmarinenematodeEnoplusbrevis.Dev.Camb.Engl.125,143–150(1998).23. J.S.Laufer,P.Bazzicalupo,W.B.Wood,Segregationofdevelopmentalpotentialinearlyembryosofcaenorhabditiselegans.Cell.19,569–577(1980).24. L.Chabry,EmbryologienormaleettératologiquedesAscidies(FelixAlcanEditeur,Paris,1887).25. M.Labouesse,S.E.Mango,PatterningtheC.elegansembryo:movingbeyondthecelllineage.TrendsGenet.15,307–313(1999).26. G.deMedeirosetal.,Confocalmultiviewlight-sheetmicroscopy.Nat.Commun.6,8881(2015).27. J.Stegmaieretal.,Real-TimeThree-DimensionalCellSegmentationinLarge-ScaleMicroscopyDataofDevelopingEmbryos.Dev.Cell.36,225–240(2016).28. R.Fernandezetal.,Imagingplantgrowthin4D:robusttissuereconstructionandlineagingatcellresolution.Nat.Methods.7,547–553(2010).29. F.Amatetal.,Fast,accuratereconstructionofcelllineagesfromlarge-scalefluorescencemicroscopydata.Nat.Methods.11,951–958(2014).30. W.R.Jeffery,Programmedcelldeathintheascidianembryo:modulationbyFoxA5andManxandrolesintheevolutionoflarvaldevelopment.Mech.Dev.118,111–124(2002).31. M.Carlson,W.Reeves,M.Veeman,StochasticityandstereotypyintheCionanotochord.Dev.Biol.397,248–256(2015).32. E.Faureetal.,Aworkflowtoprocess3D+timemicroscopyimagesofdevelopingorganismsandreconstructtheircelllineage.Nat.Commun.7,8674(2016).33. P.Ferraro,C.Godin,Adistancemeasurebetweenplantarchitectures.Ann.For.Sci.57,445–461(2000).34. J.Schulze,W.Houthoofd,J.Uenk,S.Vangestel,E.Schierenberg,Plectus-asteppingstoneinembryoniccelllineageevolutionofnematodes.EvoDevo.3,13(2012).35. C.Roubinet,C.Cabernard,Controlofasymmetriccelldivision.Curr.Opin.CellBiol.31,84–91(2014).36. T.Hirano,H.Nishida,DevelopmentalfatesoflarvaltissuesaftermetamorphosisinascidianHalocynthiaroretzi.I.Originofmesodermaltissuesofthejuvenile.Dev.Biol.192,199–210(1997).37. T.Hirano,H.Nishida,Developmentalfatesoflarvaltissuesaftermetamorphosisintheascidian,Halocynthiaroretzi.II.Originofendodermaltissuesofthejuvenile.Dev.GenesEvol.210,55–63(2000).38. M.Tokuoka,N.Satoh,Y.Satou,AbHLHtranscriptionfactorgene,Twist-like1,isessentialfortheformationofmesodermaltissuesofCionajuveniles.Dev.Biol.288,387–396(2005).39. D.Nicol,I.A.Meinertzhagen,CellcountsandmapsinthelarvalcentralnervoussystemoftheascidianCionaintestinalis(L.).J.Comp.Neurol.309,415–429(1991).40. A.G.Cole,I.A.Meinertzhagen,Thecentralnervoussystemoftheascidianlarva:mitotichistoryofcellsformingtheneuraltubeinlateembryonicCionaintestinalis.Dev.Biol.271,239–262(2004).

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

19

41. A.Pasinietal.,Formationoftheascidianepidermalsensoryneurons:insightsintotheoriginofthechordateperipheralnervoussystem.PLoSBiol.4,e225(2006).42. R.Fickentscher,P.Struntz,M.Weiss,SettingtheClockforFail-SafeEarlyEmbryogenesis.Phys.Rev.Lett.117,188101(2016).43. T.E.Gillies,C.Cabernard,CellDivisionOrientationinAnimals.Curr.Biol.21,R599–R609(2011).44. T.Negishi,H.Nishida,inAsymmetricCellDivisioninDevelopment,DifferentiationandCancer(Springer,Cham,2017;https://link.springer.com/chapter/10.1007/978-3-319-53150-2_12),ResultsandProblemsinCellDifferentiation,pp.261–284.45. K.S.Imai,K.Hino,K.Yagi,N.Satoh,Y.Satou,Geneexpressionprofilesoftranscriptionfactorsandsignalingmoleculesintheascidianembryo:towardsacomprehensiveunderstandingofgenenetworks.Dev.Camb.Engl.131,4047–58(2004).46. K.S.Imai,M.Levine,N.Satoh,Y.Satou,Regulatoryblueprintforachordateembryo.Science.312,1183–1187(2006).47. O.Tassy,F.Daian,C.Hudson,V.Bertrand,P.Lemaire,Aquantitativeapproachtothestudyofcellshapesandinteractionsduringearlychordateembryogenesis.Curr.Biol.CB.16,345–58(2006).48. N.Ohta,K.Waki,A.Mochizuki,Y.Satou,ABooleanFunctionforNeuralInductionRevealsaCriticalRoleofDirectIntercellularInteractionsinPatterningtheEctodermoftheAscidianEmbryo.PLoSComput.Biol.11,e1004687(2015).49. O.Shayaetal.,Cell-CellContactAreaAffectsNotchSignalingandNotch-DependentPatterning.Dev.Cell.40,505–511.e6(2017).50. G.J.Kim,A.Yamada,H.Nishida,AnFGFsignalfromendodermandlocalizedfactorsintheposterior-vegetaleggcytoplasmpatternthemesodermaltissuesintheascidianembryo.Dev.Camb.Engl.127,2853–62(2000).51. W.Shi,M.Levine,EphrinsignalingestablishesasymmetriccellfatesinanendomesodermlineageoftheCionaembryo.Dev.Camb.Engl.135,931–40(2008).52. V.Picco,C.Hudson,H.Yasuo,Ephrin-Ephsignallingdrivestheasymmetricdivisionofnotochord/neuralprecursorsinCionaembryos.Dev.Camb.Engl.134,1491–7(2007).53. N.Haupaixetal.,p120RasGAPmediatesephrin/Eph-dependentattenuationofFGF/ERKsignalsduringcellfatespecificationinascidianembryos.Development.140,4347–4352(2013).54. P.Lemaire,Unfoldingachordatedevelopmentalprogram,onecellatatime:invariantcelllineages,short-rangeinductionsandevolutionaryplasticityinascidians.Dev.Biol.332,48–60(2009).55. Z.Bao,Z.Zhao,T.J.Boyle,J.I.Murray,R.H.Waterston,ControlofCellCycleTimingduringC.elegansEmbryogenesis.Dev.Biol.318,65–72(2008).56. J.L.Richards,A.L.Zacharias,T.Walton,J.T.Burdick,J.I.Murray,AquantitativemodelofnormalC.elegansembryogenesisanditsdisruptionafterstress.Dev.Biol.374,12–23(2013).57. K.R.Mosaliganti,R.R.Noche,F.Xiong,I.A.Swinburne,S.G.Megason,ACME:automatedcellmorphologyextractorforcomprehensivereconstructionofcellmembranes.PLoSComput.Biol.8,e1002780(2012).58. J.B.Gurdon,P.-Y.Bourillot,Morphogengradientinterpretation.Nature.413,797(2001).

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint

20

59. A.D.Lander,HowCellsKnowWhereTheyAre.Science.339,923–927(2013).60. J.Briscoe,S.Small,Morphogenrules:designprinciplesofgradient-mediatedembryopatterning.Development.142,3996–4009(2015).61. S.Huangetal.,Deceleratedgenomeevolutioninmodernvertebratesrevealedbyanalysisofmultiplelanceletgenomes.Nat.Commun.5(2014),doi:10.1038/ncomms6896.62. Y.Nakatani,H.Nishida,InductionofNotochordduringAscidianEmbryogenesis.Dev.Biol.166,289–299(1994).63. L.Katikalaetal.,FunctionalBrachyuryBindingSitesEstablishaTemporalRead-outofGeneExpressionintheCionaNotochord.PLOSBiol.11,e1001697(2013).

.CC-BY-NC 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted February 22, 2018. ; https://doi.org/10.1101/238741doi: bioRxiv preprint