th

,b,ngUSAash372720

e Department of Pediatrics, Vanderbilt University School of Medicine, Nashville, TN 37203, USA

a r t i c l e i n f o

Article history:Received for publication 23 June 2009Revised 27 January 2010Accepted 19 February 2010

Developmental Biology 341 (2010) 196204

Contents lists available at ScienceDirect

Developmen

v ieIntroduction

In zebrash embryos, progenitor cells of the dorsal aorta and theposterior cardinal vein migrate sequentially to the midline andcoalesce to form the vascular cord. Subsequently, this endothelial cordundergoes differentiation andmorphogenesis to form the dorsal aortaand the posterior cardinal vein (Jin et al., 2005; Zhong et al., 2001;Herbert et al., 2009). Despite recent progress in dissecting theregulation of arterial and venous endothelial cell differentiation(Lawson, 2002; Swift, 2009; Zhong, 2005), the mechanisms thatcontrol specication and formation of arterial and venous endothelial

in neovascularization and angiogenesis. In mouse, loss of sonichedgehog (shh) causes lack of proper vascularization of the developinglungs (Rowitch et al., 1999), whereas shh overexpression in the dorsalneural tube results in hypervascularization (Pepicelli et al., 1998).In the developing heart, a wave of Hh activation is required forcoronary vascular development (Lavine et al., 2008, 2006). shhexpression upregulates vegf and angiopoietins to enhance myocardialneovascularization in ischemic hearts (Kusano et al., 2005; Pola et al.,2001). Inactivation of the smoothened (smo) gene, encoding the co-receptor for Shh, Indian hedgehog (Ihh) and Tiggy-winkle hedgehog(Twhh), causes severe angiogenesis defects in the yolk sac of mousecells are not well understood.Hedgehog (Hh) signaling plays importa

and formation of a variety of cell types andment (Ingham andMcMahon, 2001). It has be

Corresponding author. 358 PRB, 2220 Pierce AvenuE-mail address: tao.zhong@vanderbilt.edu (T.P. Zhon

0012-1606/$ see front matter 2010 Elsevier Inc. Adoi:10.1016/j.ydbio.2010.02.028during vasculogenesis. 2010 Elsevier Inc. All rights reserved.Available online 26 February 2010

Keywords:HedgehogEndothelial progenitor cellsArteryVeinSpecicationa b s t r a c t

In vertebrate embryos, the dorsal aorta and the posterior cardinal vein form in the trunk to comprise theoriginal circulatory loop. Previous studies implicate Hedgehog (Hh) signaling in the development of thedorsal aorta. However, the mechanism controlling specication of artery versus vein remains unclear. Here,we investigated the cell-autonomous mechanism of Hh signaling in angioblasts (endothelial progenitorcells) during arterialvenous specication utilizing zebrash mutations in Smoothened (Smo), a G protein-coupled receptor essential for Hh signaling. smomutants exhibit an absence of the dorsal aorta accompaniedby a reciprocal expansion of the posterior cardinal vein. The increased number of venous cells is equivalentto the loss of arterial cells in embryos with loss of Smo function. Activation of Hh signaling expands thearterial cell population at the expense of venous cell fate. Time-lapse imaging reveals two sequential wavesof migrating progenitor cells that contribute to the dorsal aorta and the posterior cardinal vein, respectively.Angioblasts decient in Hh signaling fail to contribute to the arterial wave; instead, they all migrate mediallyas a single population to form the venous wave. Cell transplantation analyses demonstrate that Smo plays acell-autonomous role in specifying angioblasts to become arterial cells, and Hh signaling-depletedangioblasts differentiate into venous cells instead. Collectively, these studies suggest that arterial endothelialcells are specied and formed via repressing venous cell fate at the lateral plate mesoderm by Hh signalingnt roles in specicationorgans during develop-en shown to be involved

embryos (Byrd eHh signaling

tiation of the dolack the dorsalexpression (Geriand you-too (yotstream effector glial differentiatio

e, Nashville, TN 37232, USA.g).

ll rights reserved.Hedgehog signaling induces arterial endocell fate

Charles Williams a,b, Seok-Hyung Kim b,c, Terri T. Ni a

Scott H. Baldwin b,e, Lila Solnica-Krezel b,c, Tao P. Zhoa Department of Medicine, Vanderbilt University School of Medicine, Nashville, TN 37203,b Department of Cell & Developmental Biology, Vanderbilt University School of Medicine, Nc Department of Biological Science, Vanderbilt University School of Medicine, Nashville, TNd Department of Ophthalmology, Vanderbilt University School of Medicine, Nashville, TN 3

j ourna l homepage: www.e lseelial cell formation by repressing venous

Lauren Mitchell a,b, Hyunju Ro a,b, John S. Penn b,d,a,b,

ville, TN 37203, USA03, USA3, USA

tal Biology

r.com/deve lopmenta lb io logyt al., 2002).has also been implicated in formation and differen-rsal aorta in zebrash. Zebrash smoothened mutantsaorta resulting in absence of all arterial gene

ng and Patient, 2005) . Mutations in sonic-you (syu)), which encode zebrash shh ortholog and its down-li2, respectively, display defects in arterial endothe-n (Brown et al., 2000; Chen et al., 1996; Lawson et al.,

2002). These data suggest that Hh signaling is required for formationof the dorsal aorta at early stages as well as for arterial endothelialdifferentiation during later developmental stages. Hh signaling hasbeen implicated to regulate arterial endothelial differentiation viamodulating Vegf (Lawson et al., 2002). Studies in avian and murineembryos, however, indicate that Hh signaling acts independently ofVegf to mediate vascular tubulogenesis and arterial assembly (Vokeset al., 2004). Although these studies reveal early roles for Hh signalingin artery development, it remains unclear when, where or how Hhsignaling controls formation of the dorsal aorta. In addition, little isknown about whether the absence of the dorsal aorta has any effecton development of the cardinal vein in Hh-decient embryos.

In this study, we have analyzed early roles of Hh signaling inarterialvenous specication and identied the cell-autonomousrequirement for reception of Hh signaling by endothelial progenitorcells to form the dorsal aorta. In Hh signaling-decient embryos,we observed a reciprocal expansion of the cardinal vein that isaccompanied by an absence of the dorsal aorta, whereas activated Hhsignaling by Smo agonist causes arterial cell expansion that is pro-portional to reduction of venous cells. These data suggest that arterialendothelial cells develop at the expense of venous cell fate. In vivotime-lapse imaging has revealed two waves of migrating endothelialprogenitor cells. The rst wave contributes, mainly, to the formationof the dorsal aorta, and the second wave gives rise to the posterior

cardinal vein. Angioblasts decient in Hh signaling all migrate in thesecond (venous) wave to form an expanded cardinal vein. Our celltransplantation analyses demonstrated that reception of Hh signalingby angioblasts is required for them to adopt an arterial cell fate andcontribute to formation of the dorsal aorta, whereas angioblasts withimpaired Hh signaling differentiate into venous cells instead.

Methods and materials

Zebrash maintenance and staging

Embryos were produced by pairwisematings and raised at 28.5 C,then staged according to hours post fertilization (hpf) and days postfertilization (dpf) (Kimmel et al., 1995). Smohi640 null allele (Chenet al., 2001) and transgenic sh included Tg(k:EGFP) (Jin et al., 2005),Tg(i:EGFP-nuc)(Roman et al., 2002) and Tg(k:DsRed) (Huang et al.,2005) were used in our studies.

Cyclopamine and purmorphamine treatment

Embryos were incubated in Embryo Medium (EM) containing50 M and 100 M of cyclopamine or purmorphamine respectively.Treatments were carried out at 50% epiboly unless stated otherwise.

orsaN)yped exowh

197C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 1. smomutants display expansion of the posterior cardinal vein and absence of the d(E, F), notch1b (G, H) and deltaC (I, J) in smomutants, compared to wild-type embryos. (Kexpression in the neural tube and the hypocord in smomutants (L, N) compared to wild-tcardinal vein in wild-type embryos [Tg(k:EGFP)] (O); and absence of the dorsal aorta an(P). Red arrow: dorsal aorta. Blue arrow: cardinal vein. Black arrow: oor plate. Black arr

DLAV: dorsal longitudinal anastomotic vessel.l aorta. (AJ) Lateral views displaying absent expression of efnb2 (A, B), grl (C, D), notch5Double in situ hybridization exhibiting the expansion of dab2 venous domain and col2aembryos (K, M). (O, P) Confocal microscopy depicting the dorsal aorta and the posteriorpansion of the posterior cardinal veins in smomutants [smohi640/smohi640; Tg(k:EGFP)]ead: hypocord. s: somite. n: neural tube. DA: dorsal aorta. PCV: posterior cardinal vein.

In situ hybridization and immunostaining

RNA in situ hybridization was carried out using ephrinb2 (Zhonget al., 2001), dab2 (Song et al., 2004), k1 (Zhong et al., 2001), t4(Zhong et al., 2001), grl (Zhong et al., 2001), notch1b, notch 5 (Lawsonet al., 2001), and col2a (Appel et al., 1999) markers.

Immunohistochemistry was performed as previously described(Trinh and Stainier, 2004). Briey, embryos were xed overnight with4% paraformaldehyde in sucrose buffer. Embryos were processed inPBDT (1% BSA, 1% DMSO, and 1% Triton X-100 in PBS). The followingantibodies were used at the following dilutions: goat anti-EphrinB2(R&D System) at 1:100, rabbit anti-goat IgG-Cy3 (Sigma Aldrich) at1:200. Processed samplesweremounted in 2% agarose and the imageswere acquired using a Leica dissecting microscope and a Zeiss LSM5confocal microscope.

Quantication of cell number

Heterozygous smo mutants [smohi640/+; Tg(i:EGFP-nuc)] werecrossed, where the nuclei of all endothelial cells were labeled withEGFP. The embryos were mounted in 1% agarose at 36 hpf or 72 hpfand subjected to confocal microscopy on a Zeiss LSM 510 confocalmicroscope using a 20/0.75 NA objective. 3040 confocal slices wereacquired as a z-stack for each embryo, focusing on four segmentlength of intersomitic vessels along the axial vessels above the yolkextension region in the middle trunk. Number of cells was countedin each confocal z-stack using Pickpointer embedded in Image J.Pickpointer permits a single user-dened mark to appear throughoutz-stacks of images, allowing the tracking of a single cell in overlappedz sections to avoid double counting.

Time-lapse imaging

At 14 hpf, embryos carrying the transgene (i:EGFP) were manuallydechorionated andmounted in 1% low-melting agarose in 35 mmglass-bottom Petri dishes. Transgenic embryos were treated with 50 Mcyclopamine from 50% epiboly for the duration of experiments. Time-lapse images were captured using a 20 dry (NA=0.75) objectivemounted on a motorized Zeiss Axiovert 200 microscope equippedwith a heated chamber to maintain embryos at 28.5 C using UltraviewERS (Perkin-Elmer Inc). Z image stacks were collected every 10 min.Three-dimensional data sets were compiled using Sorenson 3 videocompression and exported using QuickTime (Apple Inc.) to createmovies. Z stacks were cropped and computationally rotated using 3Dopacity function in Velocity Visualization Software (Improvision Inc.) toyield virtual transverse sections.

Transplantation

Blastula stage transplantations were conducted as previously de-scribed (Ho and Kane, 1990; Parker and Stainier, 1999). Donor embryosfrom crosses of smo mutant sh [smohi640 /+; Tg(i:EGFP-nuc)] orwild-typesh Tg(i:EGFP-nuc)wereused formosaic analysis. Between3and 4 hpf, 50100 donor blastomeres were transplanted from donorembryos and placed into themarginal region of wild-type Tg(k:DsRed)hosts. After transplantation, the donor and mosaic embryos were sub-sequently grown at 28.5 C in Ringer's solution (116 mM NaCl, 2.9 mMKCl, 1.8 mMCaCl2, 5 mMHEPES, pH 7.2) in correlatedwells until 72 hpf.Homozygous smo donors were distinguished from wild-type siblingdonors based on their phenotypes. Endothelial cell contribution of

s shC).answildanng trrow

198 C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 2. smomutation causes loss of the aortic bifurcation. (AC) Transverse section analyseembryos (A); single enlarged cardinal vein (B) and collapsed cardinal vein in smomutants (cardinal vein in wild-type embryos and smomutants. Lumen sizes were measured in 15 trError bars indicate standarddeviationandasterisks indicate statistical signicancebetweenmicroscopy in Tg(k1:EGFP) embryos at 48 hpf (G, H) exhibiting loss of the aortic bifurcationbifurcation and bilateral cardinal veins inwild-type embryos (E;G). (I, J) Schematics depictienlargedbilateral cardinal veins in smomutants. Red arrow: aorta. Bluearrow: vein. Yellowa

ccv: common cardinal vein. pcv: posterior cardinal vein. LDA', acv',ccv', pcv': denote the corresowing normal lumen sizes of the dorsal aorta and the posterior cardinal vein in wild-type(D) Bar chart depicting perimetermeasurements of lumen sizes of the dorsal aorta and theverse sections derived from three wild-type and three smomutant embryos, respectively.-typeand smodata (Pb0.01). (EH) Insituhybridizationofk1at24 hpf (E, F) and confocald expansion of bilateral cardinal veins in smomutants (F;H), compared to the normal aortiche aortic bifurcation anked by bilateral cardinal veins inwild-type embryos, compared to: intersomitic vessel.DA: dorsal aorta. LDA: lateral dorsal aorta. acv: anterior cardinal vein.

ponding vessel on the other side of the embryo.

transplanted cells was examined by EGFP expression inmosaic embryosusing a Zeiss LSM 510 Meta Confocal microscope.

Results

smo mutant embryos display an expansion of venous cells with areciprocal loss of arterial cells

Previous studies have shown that embryos decient in Hh sig-naling lack the dorsal aorta (Brown et al., 2000; Lawson et al., 2002;Gering and Patient, 2005). However, the developmental status of thecardinal vein has not been investigated. Thus, we examined venousdevelopment in smo mutant embryos using arterial and venousendothelial markers. In agreement with previous studies, expressionof arterial markers efnb2, grl, notch5 and notch1b were completelyabsent in smo mutants (Fig. 1AH) (Lawson et al., 2002). Absence ofdeltaC expression in the mutant aorta was observed, while deltaCexpression in somites was reduced in the mutants (Fig. 1I, J). Notably,expression of the venous marker disabled homolog 2 (dab2) wasexpanded dorsally into the domain of the dorsal aorta and reachedto the hypocord marked by col2a expression at 24 hpf and 30 hpf(Fig. 1KN) (Appel et al., 1999). Confocal microscopy analysis re-vealed expansion of the posterior cardinal vein and absence of thedorsal aorta in smomutants [smohi640/smohi640; Tg(k:EGFP)] (Fig. 1P),

when compared to the normal size of both the dorsal aorta and theposterior cardinal vein in wild-type embryos at 48 hpf (Fig. 1O).Transverse sections validated the enlarged lumen size of the posteriorcardinal vein in smo mutants (Fig. 2B), which was collapsed in someplaces (Fig. 2C), as compared to normal size of the dorsal aorta andthe posterior cardinal vein in wild-type embryos (Fig. 2A). Notably,the enlarged lumen size of the cardinal vein (201.213.7 m) insmo mutants is equal to the combined lumen size of the dorsal aorta(85.74.3 m) and the posterior cardinal vein (126.45.5 m) inwild-type embryos (Fig. 2D). Both in situ hybridization and confocalmicroscopy analyses revealed that, in the region of aortic bifurcationin wild-type embryos, bilateral cardinal veins ank the aortic bifur-cation, where bilateral dorsal aortaemerge into the single dorsal aortaat the midline in wild-type embryos (Fig. 2E;G;I). In smo mutants,bilateral dorsal aortae and the dorsal aorta at the aortic bifurcationwere completely absent, whereas bilateral cardinal veins were sig-nicantly expanded (Fig. 2F;H;J). These data suggest that Hh signalingdeciency results in the expansion of cardinal veins at the expense ofaortae.

To determine whether the venous expansion is due to increasedendothelial cell number, we quantied endothelial cells in thedorsal aorta and the posterior cardinal vein in transgenic embryos[Tg(i:EGFP-nuc)], in which green uorescent protein gene fused toa nuclear localization signal (EGFP-nuc) is under the control of the

. (Ain w640;TumT),terso

199C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 3. The increase of venous cells is equivalent to loss of arterial cells in smo mutants(A, B stained with efnb2 antibody, the posterior cardinal vein and intersomitic vesselsincreased endothelial nuclei in enlarged cardinal veins in smomutants [smohi1640/smohi1

and venous cells in wild-type embryos compared to smomutants. At 36 hpf, arterial cell n72 hpf, arterial cell number: 20.21.3 (WT), 0 (smo); Venous cell number: 541.6 (W6 wild-type embryos and 6 smomutants. Each section covers four segment lengths of in

above the yolk extension region. a: aorta. v: vein.D) Confocal optics displaying individual nuclei of endothelial cells in the dorsal aortaild-type embryos [Tg(i:EGFP-nuc)] at 36 hpf (A) and 72 hpf (C),as compared to theg(i:EGFP-nuc)] at 36 hpf (B) and 72 hpf (D). (E) Bar chart depicting number of arterialber: 19.21.7 (WT), 0 (smo); Venous cell number: 43.21.3 (WT), 60.32.4 (smo). At66.83.4 (smo). Endothelial cells were counted from 12 lateral sections derived frommitic vessels along the dorsal aorta and the posterior cardinal vein in the middle trunk

. (Athebryof d

200 C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 4. Temporal development stages when Hh signaling affects arteryvein specicationembryos (A, C) and smo mutants (B, D). (E, F) Confocal microscopy analysis revealingcontrol embryos (E), and an expanded posterior cardinal vein in cyclopamine-treated emarteryvein specication. Embryos were incubated in 50 M cyclopamine during a seriesstage. 10 s:10-somite stage.i promoter (Roman et al. 2002). In these transgenic embryos,individual endothelial cell nuclei are marked by green uorescence(Fig. 3), permitting quantitative assessment of endothelial cellnumber in the dorsal aorta and the posterior cardinal vein usingconfocal microscopy analysis. At 36 hpf, the number of cells in theexpanded cardinal vein in smomutants is equivalent to the combinedcell number of the dorsal aorta labeled by efnb2 and the cardinal veinin wild-type embryos (Fig. 3A,B;E). The same arterialvenous speci-cation defect persists in smomutants at 72 hpf, when the dorsal aortais clearly separated from the posterior cardinal vein (Fig. 3C,D;E).These data indicate that the increase in the venous cell number isdirectly proportional to the loss of arterial cells in smo mutants.

Hh signaling regulates arterialvenous specication at the lateral platemesoderm

To determine whether angioblast generation is affected in smomutants, we performed k1 and ptc1 double in situ hybridization. Weused expression of ptc1, a seven transmembrane receptor for Hh, todistinguish smomutant embryos fromsiblingwild-type embryosduringearly development. Previous studies indicate that Hh positivelyregulates ptc1 expression; and in smomutant embryos, ptc1 expression

Fig. 5. Activated Hh signaling causes an increase of arterial cells and loss of venous cells.(AD) In situ hybridization analysis displaying the increased efnb2 expression inpurmorphamine (PMA)-treated embryos (A) compared to controls (B); and the reduceddab2 expression in embryos treatedwith PMA (C) compared to controls (0.1% DMSO) (D).(E) Bar chart depicting numbers of arterial and venous cells in PMA-treated embryosfrom5 to16 hpf and 12 to16 hpf, compared to controls (0.1%DMSO). Arterial cell number:19.30.6 (control); 27.50.9 (516 hpf; PMA treated); 26.70.6 (1216 hpf; PMAtreated). Venous cell number: 41.61.5 (control); 32.71.2 (516 hpf; PMA treated);33.51.8 (1216 hpf; PMA treated). Endothelial cells were counted from 6 sectionsderived from 3 wild-type embryos or 3 PMA-treated embryos. Each section covers foursegment lengths of intersomitic vessels along the dorsal aorta and the posterior cardinalvein in the middle trunk above the yolk extension region. a: aorta. v: vein.D) Dorsal views displaying comparable levels of k1-positive angioblasts in wild-typenormal size of the dorsal aorta, the posterior cardinal vein and intersomitic vessels inos (F). (G) Schematic representation depicting the temporal activity of cyclopamine inevelopmental window. Red and blue line: artery and vein. Blue line: vein. 5 s:5-somiteis down regulated (Hooper and Scott, 2005; Ingham and McMahon,2001). At the 5- and 10-somite stages, formation of angioblastsvisualized by k1 expression at the lateral plate mesoderm was notaltered in smo mutants compared to wild-type embryos, while ptc1expression was drastically reduced (Fig. 4AD). We next determined

the time window when defective Hh signaling results in the absenceof the dorsal aorta and expansion of the cardinal vein. We usedcyclopamine to inhibit the Smo receptor and block Hh signaling in k1transgenic embryos [Tg(k1:EGFP)] during a series of developmentaltimestages. In therst set of pulse experiments, cyclopaminewasaddedat 3 hpf (1000-cell stage) andwashed away at progressively later stagesof development, including 5 hpf (50% epiboly), 10 hpf (tailbud stage),12 hpf (6-somite stage), 14 hpf (10-somite stage), 16 hpf (14-somitestage) and 17 hpf (16-somite stage) (Fig. 4G). We examined the de-velopmental status of the dorsal aorta and cardinal veins at 48 hpf usingconfocal microscope analysis (Fig. 3E, F). Cyclopamine treatment failedto cause expansion of the cardinal vein and absence of the dorsal aortauntil 12 hpf (Fig. 4G), suggesting that cyclopamine treatment is effectivein inhibiting artery specication only after the 6-somite stage. In asecond set of experiments, cyclopamine was added at progressivelylater stages of development with treatments starting at 5 hpf, 10 hpf,12 hpf, 16 hpf and washed away at 17 hpf (Fig. 4G). Cyclopaminetreatment prior to 16 hpf resulted in the expansion of the veins andelimination of the aorta (Fig. 4G). Together, these data indicate thatcyclopamine-mediated inhibition of Smo receptor causes expansion ofthe cardinal vein and elimination of thedorsal aorta between 12 hpf and16 hpf at the lateral plate mesoderm. Indeed, cyclopamine treatment ofembryos during a short time period (1216 hpf) caused venousexpansion and arterial elimination (Fig. 4G). This developmentalwindow correlates well with the stages when endothelial progenitorcells migrate at the posterior lateral mesoderm (Zhong, 2005; Zhonget al., 2001).

We next examined whether elevating Hh signaling causes anincrease in arterial cells and a reduction of venous cells during the

specic time window. We used purmorphamine (PMA) to activatethe Hh signaling pathway. PMA directly targets and activates Smoreceptor as an agonist (Sinha and Chen, 2006).When PMAwas addedto Tg(i:EGFP-nuc) embryos during both time windows (516 hpf;1216 hpf), efnb2 expression was increased in treated embryoscompared to controls (Fig. 5A,B), whereas dab2 expression wasreduced (Fig. 5C,D). We quantied endothelial cell number andfound that PMA treatment caused an increase in arterial cell numberand a reduction in venous cell number (Fig. 5E). Furthermore, theincrease in arterial cells is proportional to the decrease of venouscells after treatment during both developmental time windows(Fig. 5E). Together, these ndings suggest that activated Hh signalingcauses expansion of arterial cell formation via repressing venous cellfate.

Hh signaling deciency causes loss of the rst wave of angioblastmigration

Angioblasts migrate to the midline to form the dorsal aorta and theposterior cardinal vein. Previous immunostaining and histochemicalstudies suggest that there are two distinct waves of angioblast migra-tion: The rst migratory wave of angioblasts, the arterial wave, con-tributes largely to the formation of the dorsal aorta, while the secondwave of migration, the venous wave, gives rise to the posterior cardinalvein (Jin et al., 2005; Isogai et al., 2003). To determine whetherangioblast migration is affected in smo mutants, we performed time-lapse experiments in i1 transgenic embryos [Tg(i1:EGFP)] usingspinning disk confocal microscopy from 14 hpf, when angioblasts startto migrate, to 24 hpf, when the dorsal aorta and the posterior cardinal

deeateatedof tof t

201C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 6. The rst wave of angioblast migration is diminished in embryos with Hh signalingthe midline from 14.5 hpf to 20 hpf in wild-type embryos (A;C;E), and cyclopamine-trsequence viewing angioblast migration in wild-type embryos (G), and cyclopamine-trecaptured from 300-min time-lapse sequence. Images were obtained from the dorsal side14.5 hpf stage (G;G'). Times shown in right corners represent time elapsed after the start

in G and G' were derived from.ciency. (AF) Time-lapse imaging revealing sequential migration of angioblasts towardd embryos (B;D;F). (G and G') Crossover optical sections derived from the time-lapseembryos (G'). Red arrowhead: arterial cells. Blue bracket: venous cells. Frames were

ransgenic embryos Tg(i1:EGFP) (AF). The 0 min time point is from the approximatelyhe sequence (G;G'). Yellow dot lines (AF) indicate where the optical crossover sections

vein form.We observed that in wild-type embryos, angioblasts migratein streams towards the midline (Supplementary movie 1). Moreover,the initial vascular cord was observed at 16 hpf (Fig. 6C), most cells ofwhich is largely associated with formation of the dorsal aorta (Jin et al.,2005). In silico cross-sections, derived from the z-stack time sequence ofa single embryo, reveal sequentialmigration of angioblasts (Fig. 6G; 0 to300 min; Supplementarymovie 1). Therstmigratorywave formed theinitial vascular cord during 60-minute interval (Fig. 6G; 0 to 60 min).The second venous wave followed the rst wave to coalesce into theformedvascular cord (Fig. 6G; 0 to 300 min). Thesedata suggest that thearterial wavemoves much faster than the venouswave. In Hh-decientembryos treated with cyclopamine, the initial vascular cord failedto form (Fig. 6D; G'). Time-lapse sequence from 30 min to 120 mindisplayed the absence of the rst arterial wave of angioblasts (Fig. 6G';Supplementary movie 2). Notably, in Hh signaling-decient embryos,angioblasts moved in a concerted fashion (Fig. 6G' vs. G; 30 min to120 min; Supplementarymovie 2). These angioblastsmigrated towardsthemidline to form the vascular cord that is destined to be the posteriorcardinal vein (Fig. 6G'; Fig. 1L;N;P). Thus, Hh-decient progenitor cellsfail to contribute to the rst arterial wave; instead, they all migratemedially in the venouswave to contribute to the posterior cardinal vein.These data suggest that migratory ability of Hh-depleted angioblasts isnot affected.

Reception of Hh signaling is required cell-autonomously for the formationof arterial endothelial cells

We hypothesized that angioblasts that receive Hh signaling atearly stages adopt an arterial cell fate and contribute to formation

of the dorsal aorta, while angioblasts that are decient in Hhsignaling adopt a venous cell fate and give rise to the posteriorcardinal vein. Hence, we conducted transplantation experiments inzebrash to test whether receipt of Hh signaling is required forarterial versus venous cell specication. Blastomeres were trans-planted at blastula stages, using donor embryos from heterozygouscross carrying smo mutation and transgene Tg(i:EGFP-nuc) toassess arterial or venous contribution. The recipients were wild-type embryos carrying the transgene Tg(k:DsRed). Analyses of theresulting chimeric embryos at 72 hpf revealed that wild-typedonor cells were able to contribute to both arterial and venouscells at approximately equal frequency (Fig. 7A,B;E,F). In contrast,none of the transplanted cells from smo mutant donors contributedto the arterial endothelial system; instead, all donor cells from smoembryos contributed to the venous system, including the posteriorcardinal vein (Fig. 7C,D;E,F). One explanation for the lack ofarterial contribution is that smo donor cells might not survive inthe arterial endothelium, a general problem of cell transplantation.As a control, we performed transplantations using donor cellslabeled with a uorescein-dextran lineage and found that smodonor cells were present in many cell lineages in wild-type hosts.For example, smo cells were capable of contributing to fast muscleas the same frequency as wild-type cells (Fig. 7G,H;I). Thus,angioblasts that receive Hh signaling contribute to both arterialand venous endothelia, whereas Hh-depleted angioblasts only giverise to venous cells. These data suggest that reception of Hhsignaling is required cell-autonomously in angioblasts to bespecied into arterial endothelial cells at the expense of venouscell fate.

g vel aoics shosnorcs shing t

202 C. Williams et al. / Developmental Biology 341 (2010) 196204Fig. 7. Hh signaling acts cell-autonomously to induce arterial cell formation by inhibitinEGFP-positive donor cells derived from wild-type Tg(i:EGFP-nuc) embryos in the dorsawhen the dorsal aorta is completely separated from the cardinal vein. (C, D) Confocal optof smomutants [smo/+; Tg(i:EGFP-nuc)] in the posterior cardinal vein in wild-typeratio of hosts with donor-derived arterial (or venous) cells to all host embryos with do(F) Bar chart depicting arterial or venous contribution frequency. (G, H) Fluorescent optiBoth wild-type and smo donors give rise to skeletal muscle tissue. (I) Data table quantitat

derived muscle cells to all host embryos was used to determine muscle contribution frequenous cell development. (A, B) Confocal microscopic analysis revealing the presence ofrta and the posterior cardinal vein of wild-type host embryos Tg(k:DsRed) at 72 hpf,howing the presence of EGFP-positive smo-null cells derived from heterozygous crossest embryos Tg(k:DsRed) at 72 hpf. (E) Data table quantitating transplantation results. Acells in any tissue was used to determine arterial (or venous) contribution frequency.owing transplantation of uorescein-dextran injected donor cells into wild-type hosts.ransplantation results of uorescein-dextran injected hosts. A ratio of hosts with donor-

ncy.

203C. Williams et al. / Developmental Biology 341 (2010) 196204Discussion

In this study, we show that smomutants create a surplus of venouscells as an alternative fate to the lost arterial cells. Activation of Hhsignaling by Smo agonist causes an increase in the arterial cell numberand a reduction in venous cells. The arterial and venous cell speci-cation occurs at the lateral plate mesoderm during migration beforethe formation of the vascular cord. Hh signaling deciency causes theloss of arterial migration wave that is accompanied by reciprocalexpansion of the venous wave. We thus propose a model that recep-tion of Hh signaling is required cell-autonomously for angioblaststo differentiate into arterial cells at the expense of venous cell fate(Fig. 8).

Previous studies revealed absence of the dorsal aorta in smomutants and cyclopamine-treated embryos, concluding that thismight be due to defects in the medial migration of angioblasts con-tributing to the dorsal aorta (Gering and Patient, 2005). We haveextended these studies to show that loss of the dorsal aorta isassociated with the corresponding expansion of the posterior cardinalvein in Hh signaling-decient embryos. We have further used time-

Fig. 8. Shh signal instructs angioblasts to differentiate into arterial cells rather thanvenous cells.lapse imaging to directly visualize two distinct waves of angioblastmigration in wild-type embryos, with the rst wave contributingto the arterial and the second wave to the venous system. Previousstudies (Gering and Patient, 2005) had only compared the static k1expression at different stages using in situ hybridization, and thusfailed to observe dynamic migration of venous angioblasts to themidline. Moreover, we show that Hh signaling deciency causesall angioblasts to reach to the midline in the later venous wave ofmigration. Importantly, our studies provided evidence that loss of thedorsal aorta in Hh-depleted embryos is due to adoption of a venouscell fate by all angioblasts, leading to a surplus of venous cells. Ourtransplantation experiments reveal that smo-null angioblasts fail tocontribute to the dorsal aorta but rather contribute to the posteriorcardinal vein in wild-type hosts. Thus, lack of the arterial migrationwave in Hh signaling-depleted embryos could be an indirect con-sequence of the venous cell fate assignment. Hh signaling has beenshown to be involved in cell fate assignment in several other celllineages. For example, activation of Hh pathway species neuroblastsin Drosophila (Bhat, 1996; McDonald and Doe, 1997), determinesventral cell fate in the neural tube (Chiang et al., 1996; Gunhaga et al.,2000; Jessell, 2000) and promotes cardiomyocyte formation in thedeveloping heart (Thomas et al., 2008). Interestingly, mouse embryoslacking function of smo fail to form the anterior paired dorsal aortaeand some segments of the descending dorsal aorta. However, k1-positive endothelial cells were accumulated in these arterial regions(Vokes et al., 2004). Although this is attributed to tubulogenesisdefects, it remains unknown whether these accumulated endothelialcells have a venous cell identity.

Hh signaling has been shown to be important in arterial endo-thelium differentiation and maturation after formation of the dorsalaorta. This primarily reects absent expression of arterial geneephrinb2 in the formed aorta in syumutants as well as cyclopamine-treated embryos at late stages from the 15-somite stages (Geringand Patient, 2005; Lawson et al., 2002; Byrd et al., 2002). This Hh-dependent arterial differentiation and maturation is thought to actvia Vegf, which acts upstream of Notch pathway (Lawson et al., 2002;Byrd and Grabel, 2004). Likewise, interrupted Notch signaling inmibmutants or vegf knockdown in zebrash reduces expression of thearterial marker efnb2 but fails to ablate the dorsal aorta in themidline (Nasevicius et al., 2000; Lawson et al., 2001, 2002; Jin et al.,2005). This is not due to defective specication; rather, due to theirroles in regulating arterialvenous segregation at later stages(Herbert et al., 2009). After the vascular cord formation, Vegf andNotch act upstream of Efnb2EphB4 signaling to sort arterial andvenous angioblasts into the separate trunk vessels (the dorsal aortaand the cardinal vein) from the single precursor vessel (Herbertet al., 2009). Thus, the Hh-dependent arterial differentiation andmaturation could act at the levels of arterialvenous segregationthrough modulation of Vegf signaling. Our studies demonstratethat the early role of Hh signaling involved in arterialvenousspecication is via a cell-autonomous manner. In avian embryos,endoderm-derived shh has been shown to directly act on angioblastsand regulate vascular assembly (Vokes et al., 2004). In zebrash,three Hh ligands (shh, ihh and twhh) are expressed in the notochordand the oor plate but not the endoderm during the early seg-mentation stages (Roy et al. 2001). Most likely, the midline-derivedHh signals are required for action directly on migrating angioblasts(Byrd and Grabel, 2004). The diffusible range and concentrationgradient of Hh encountered by angioblasts, or the timing andduration of Hh exposure to angioblasts may be critical for them tomake an arterial versus a venous cell fate choice. Collectively, thesendings suggest that Hh signaling plays multiple roles duringvasculogenesis. Further elucidation of the Hh-dependent compo-nents and signaling mechanisms in regulating arterialvenousspecication may lead to novel therapeutic strategies for diversedisorders such as coronary heart disease and cancer.

Acknowledgments

We acknowledge Sarah Kucenas, Robert Taylor andWenbiao Chenfor their invaluable assistance during our experimental procedures.We thank excellent sh care by John Quan. We are grateful tomembers of Solnica-Krezel and Zhong laboratories for comments onthe manuscript and helpful discussions. This research was supportedby NIH grants (R01HL073348 to TPZ; RO1GM055101 to LSK).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ydbio.2010.02.028.

References

Appel, B., Fritz, A., Westereld, M., Grunwald, D.J., Eisen, J.S., Riley, B.B., 1999. Delta-mediated specication of midline cell fates in zebrash embryos. Curr. Biol. 9,247256.

Bhat, K.M., 1996. The patched signaling pathway mediates repression of gooseberryallowing neuroblast specication by wingless during Drosophila neurogenesis.Development 122, 29212932.

Brown, L., Rodaway, A., Schilling, T., Jowett, T., Ingham, P., Patient, R., Sharrocks, A.,

2000. Insights into vasculogenesis revealed by expression of ETS-domain

transcription factor Fli-1 in wild type and mutant zebrash embryos. Mech. Dev.237252.

Byrd, N., Grabel, L., 2004. Hedgehog signaling in murine vasculogenesis and angio-genesis. Trends Cardiovasc. Med. 14, 308313.

Byrd, N., Beck, S., Maye, P., Narasimhaiah, R., St-Jacques, B., Zhang, X., McMahon, J.,McMahon, A.P., Grabel, L., 2002. Hedgehog is required fo murine yolk sac angio-genesis. Development 129, 361372.

Chen, J.N.,Haffter, P., Odenthal, J., Vogelsang, E., Brand,M., vanEeden, F.J., Furutani-Seiki,M.,Granato, M., Hammerschmidt, M., Heisenberg, C.P., Jiang, Y.J., Kane, D.A., Kelsh, R.N.,Mullins, M.C., Nusslein-Volhard, C., 1996. Mutations affecting the cardiovascular sys-tem and other internal organs in zebrash. Development 123, 293302.

Chen, W., Burgess, S., Hopkins, N., 2001. Analysis of the zebrash smoothened mutantreveals conserved and divergent functions of hedgehog activity. Development 128,23852396.

Chiang, C., Litingtung, Y., Lee, E., Young, K.E., Corden, J.L., Westphal, H., Beachy, P.A.,1996. Cyclopia and defective axial patterning in mice lacking Sonic hedgehog genefunction. Nature 383, 407413.

Gering, M., Patient, R., 2005. Hedgehog signaling is required for adult blood stem cellformation in zebrash embryos. Dev. Cell 8, 389400.

Gunhaga, L., Jessell, T.M., Edlund, T., 2000. Sonic hedgehog signaling at gastrula stagesspecies ventral telencephalic cells in the chick embryo. Development 127,32833293.

Herbert, S.P., Huisken, J., Kim, T.N., Feldman,M.E., Houseman, B.T.,Wang, R.A., Shokat, K.M.,Stainier, D.Y., 2009. Arterial-venous segregation by selective cell sprouting: analternative mode of blood vessel formation. Science 326, 294298.

Ho, R.K., Kane, D.A., 1990. Cell-autonomous action of zebrash spt-1 mutation inspecic mesodermal precursors. Nature 348, 728730.

Hooper, J.E., Scott, M.P., 2005. Communicating with Hedgehogs. Nat. Rev. Mol. Cell Biol.6, 306317.

Huang, H., Zhang, B., Hartenstein, P.A., Chen, J.N., Lin, S., 2005. NXT2 is required forembryonic heart development in zebrash. BMC Dev. Biol. 5, 7.

Ingham, P.W., McMahon, A.P., 2001. Hedgehog signaling in animal development:

Lavine, K.J., Long, F., Choi, K., Smith, C., Ornitz, D.M., 2008. Hedgehog signaling todistinct cell types differentially regulates coronary artery and vein development.Development 135, 31613171.

Lawson, N.D., Weinstein, B.M., 2002. Arteries and veins: making a difference withzebrash. Nature Rev. Genet. 3, 674682.

Lawson, N.D., Scheer, N., Pham, V.N., Kim, C.H., Chitnis, A.B., Campos-Ortega, J.A.,Weinstein, B.M., 2001. Notch signaling is required for arterial-venous differenti-ation during embryonic vascular development. Development 128, 36753683.

Lawson, N.D., Vogel, A.M., Weinstein, B.M., 2002. sonic hedgehog and vascularendothelial growth factor act upstream of the Notch pathway during arterialendothelial differentiation. Dev. Cell 3, 127136.

McDonald, J.A., Doe, C.Q., 1997. Establishing neuroblast-specic gene expression in theDrosophila CNS: huckebein is activated by Wingless and Hedgehog and repressedby Engrailed and Gooseberry. Development 124, 10791087.

Nasevicius, A., Larson, J., Ekker, S.C., 2000. Distinct requirements for zebrashangiogenesis revealed by a VEGF-A morphant. Yeast 17, 294301.

Parker, L., Stainier, D.Y., 1999. Cell-autonomous and non-autonomous requirements forthe zebrash gene cloche in hematopoiesis. Development 126, 26432651.

Pepicelli, C.V., Lewis, P.M., McMahon, A.P., 1998. Sonic hedgehog regulates branchingmorphogenesis in the mammalian lung. Curr. Biol. 8, 10831086.

Pola, R., Ling, L.E., Silver, M., Corbley, M.J., Kearney, M., Blake Pepinsky, R., Shapiro, R.,Taylor, F.R., Baker, D.P., Asahara, T., Isner, J.M., 2001. The morphogen Sonichedgehog is an indirect angiogenic agent upregulating two families of angiogenicgrowth factors. Nat. Med. 7, 706711.

Roman, B.L., Pham, V.N., Lawson,N.D., Kulik,M., Childs, S., Lekven,A.C., Garrity,D.M.,Moon,R.T., Fishman, M.C., Lechleider, R.J., Weinstein, B.M., 2002. Disruption of acvrl1 increasesendothelial cell number in zebrash cranial vessels. Development 129, 30093019.

Rowitch, D.H., B, S. J., Lee, S.M., Flax, J.D., Snyder, E.Y., McMahon, A.P., 1999. Sonichedgehog regulates proliferation and inhibits differentiation of CNS precursor cells.J. Neurosci. 19, 89548965.

Roy, S., Qiao, T., Wolff, C., Ingham, P.W., 2001. Hedgehog signaling pathway is essentialfor pancreas specication in the zebrash embryo. Curr. Biol. 17, 13581363.

Sinha, S., Chen, J.K., 2006. Purmorphamine activates the Hedgehog pathway by

204 C. Williams et al. / Developmental Biology 341 (2010) 196204Isogai, S., Lawson, N.D., Torrealday, S., Horiguchi, M., Weinstein, B.M., 2003. Angiogenicnetwork formation in the developing vertebrate trunk. Development 130,52815290.

Jessell, T.M., 2000. Neuronal specication in the spinal cord: inductive signals andtranscriptional codes. Nat. Rev. Genet. 1, 2029.

Jin, S.W., Bels, D., Mitchell, T., Chen, J.N., Stainier, D.Y.R., 2005. Cellular and molecularanalyses of vascular tube and lumen formation in zebrash. Development 132,51995209.

Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., 1995. Stages ofembryonic development of the zebrash. Dev. Dyn. 203, 253310.

Kusano, K.F., Pola, R., Murayama, T., Curry, C., Kawamoto, A., Iwakura, A., Shintani, S., Ii, M.,Asai, J., Tkebuchava, T., Thorne, T., Takenaka, H., Aikawa, R., Goukassian, D., vonSamson, P., Hamada, H., Yoon, Y.S., Silver, M., Eaton, E., Ma, H., Heyd, L., Kearney, M.,Munger, W., Porter, J.A., Kishore, R., Losordo, D.W., 2005. Sonic hedgehog myocardialgene therapy: tissue repair through transient reconstitution of embryonic signaling.Nat. Med. 11, 11971204.

Lavine, K.J., White, A.C., Park, C., Smith, C.S., Choi, K., Long, F., Hui, C.C., Ornitz, D.M.,2006. Fibroblast growth factor signals regulate a wave of Hedgehog activation thatis essential for coronary vascular development. Genes Dev. 20, 16511666.targeting Smoothened. Nat. Chem. Biol. 2, 2930.Song, H.D., Sun, X.J., Deng,M., Zhang, G.W., Zhou, Y.,Wu, X.Y., Sheng, Y., Chen, Y., Ruan, Z.,

Jiang, C.L., Fan, H.Y., Zon, L.I., Kanki, J.P., Liu, T.X., Look, A.T., Chen, Z., 2004.Hematopoietic gene expression prole in zebrash kidneymarrow. Proc. Natl. Acad.Sci. U. S. A. 101, 1624016245.

Swift, M.R., Weinstein, B.M., 2009. Arterial-venous specication during development.Circ. Res. 104, 576588.

Thomas, N.A., Koudijs, M., van Eeden, F.J., Joyner, A.L., Yelon, D., 2008. Hedgehogsignaling plays a cell-autonompus role in maximizing cardiac developmentalpotential. Development 135, 37893799.

Trinh, L.A., Stainier, D.Y., 2004. Fibronectin regulates epithelial organization duringmyocardial migration in zebrash. Dev. Cell 6, 371382.

Vokes, S.A., Yatskievych, T.A., Heimark, R.L., McMahon, J., McMahon, A.P., Antin, P.B.,Krieg, P.A., 2004. Hedgehog signaling is essential for endothelial tube formationduring vasculogenesis. Development 131, 43714380.

Zhong, T.P., 2005. Zebrash genetics and formation of embryonic vasculature. Curr. Top.Dev. Biol. 71, 5381.

Zhong, T.P., Childs, S., Leu, J.P., Fishman, M.C., 2001. Gridlock signalling pathwayfashions the rst embryonic artery. Nature 414, 216220.paradigms and principles. Genes Dev. 15, 30593087.

Hedgehog signaling induces arterial endothelial cell formation by repressing venous cell fateIntroductionMethods and materialsZebrafish maintenance and stagingCyclopamine and purmorphamine treatmentIn situ hybridization and immunostainingQuantification of cell numberTime-lapse imagingTransplantation

Resultssmo mutant embryos display an expansion of venous cells with a reciprocal loss of arterial cell.....Hh signaling regulates arterialvenous specification at the lateral plate mesodermHh signaling deficiency causes loss of the first wave of angioblast migrationReception of Hh signaling is required cell-autonomously for the formation of arterial endotheli.....

DiscussionAcknowledgmentsSupplementary dataReferences