06 (2007) 160–169www.elsevier.com/locate/ydbio

Developmental Biology 3

The secreted EGF-Discoidin factor xDel1 is essential fordorsal development of the Xenopus embryo

Akiko Arakawa a,c,1, Mami Matsuo-Takasaki a,1, Akira Takai a,b, Hidehiko Inomata a,Michiru Matsumura a, Makoto Ikeya a, Kenzo Takahashi c, Yoshiki Miyachi c,

Noriaki Sasai a, Yoshiki Sasai a,b,⁎

a Organogenesis and Neurogenesis Group, Center for Developmental Biology, RIKEN, 2-2-3 Minatojima-minamimachi, Chuo, Kobe 650-0047, Japanb Department of Medical Embryology, Graduate School of Medicine, Kyoto University, Kyoto 606-8507, Japan

c Department of Dermatology, Graduate School of Medicine, Kyoto University, Kyoto 606-8507, Japan

Received for publication 20 October 2006; revised 6 March 2007; accepted 7 March 2007Available online 13 March 2007

Abstract

We show here that a secreted EGF-Discoidin-domain protein, Xenopus Del1 (xDel1), is an essential factor for dorsal development in the earlyXenopus embryo. Knockdown of the xDel1 function causes obvious ventralization of the embryo. Conversely, overexpression of xDel1 expandsdorsal-marker expression and suppresses ventral-marker expression in the gastrula embryo. Forced expression of xDel1 dorsalizes ventralmarginal zone explants, whereas it weakly induces neural differentiation but not mesodermal differentiation in animal caps. The dorsalizingactivity of xDel1 is dependent on the Discoidin domains and not on the RGD motif (which is implicated in its angiogenic activity) or EGF repeats.Luciferase assays show that xDel1 attenuates BMP-signaling reporter activity by interfering with the pathway downstream of the BMP receptor.Thus, xDel1 functions as a unique extracellular regulatory factor of DV patterning in early vertebrate embryogenesis.© 2007 Elsevier Inc. All rights reserved.

Keywords: DV axis; Secreted factor; Del1; Discoidin; BMP; Xenopus

Introduction

In early vertebrate development, the embryo is patternedalong two main axes: the dorsal–ventral (DV) and rostral–caudal (RC) axes. In Xenopus, the DV polarity is establishedfollowing sperm entry and cortical rotation by the dorsallydominant activation of canonical Wnt signaling (Moon andKimelman, 1998; Heasman, 2006). With regard to mesoendo-dermal patterning, the DV and RC axes are closely associateduntil the onset of gastrulation, when the two are separated by therostral migration of head mesoendodermal tissues from thedorsal marginal zone (DMZ) (Keller, 2005).

Dorsal determination in the early Xenopus embryo iscontrolled by at least three signaling pathways in a stage-

⁎ Corresponding author. Organogenesis and Neurogenesis Group, Center forDevelopmental Biology, RIKEN, 2-2-3 Minatojima-minamimachi, Chuo, Kobe650-0047, Japan. Fax: +81 78 306 1854.

E-mail address: sasaicdb@mub.biglobe.ne.jp (Y. Sasai).1 These authors contributed equally to this work.

0012-1606/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.ydbio.2007.03.008

specific manner. Canonical Wnt signaling acts during the earlycleavage stages as described above, and it also plays an inductiverole for dorsal gene expression (e.g., Siamois and Chordin)during the late blastula and early gastrula stages (Carnac et al.,1996; Heasman, 2006). Nodal/Activin signals promote dorsalmesodermal differentiation during the mid- to late-blastulastages in a dose-dependent fashion (Osada and Wright, 1999).Nodal/Activin signaling is unique in that it can inducemesodermal tissues directly from uncommitted animal capcells (Smith, 1995). BMP antagonists, such as Chordin andNoggin, are considered to function from the late blastula stagesonwards by inhibiting the ventralizing activity of the BMPsignals (Sasai et al., 1995; Sasai and De Robertis, 1997; Piccoloet al., 1996; Zimmerman et al., 1996). Although BMP inhibitionplays a role in DV patterning at relatively late stages (lateblastula∼early gastrula), a recent report using a combination ofMOs has demonstrated that BMP antagonists are indispensablefor the development of dorsal structures (Khokha et al., 2005).BMP signaling controls DV patterning via an intricate system

Fig. 1. Spatial and temporal pattern of xDel1 expression analyzed by whole-mount in situ hybridization. (A) The 8-cell stage, lateral view. (B) Stage 9, lateral view. (C)Stage 11, vegetal view. Arrowhead, dorsal lip. (D) Stage 11, sagittal section across the dorsal lip (white arrow). Black arrow, bottle cells. Open and closed arrowheads,dorsal ectoderm and mesoderm, respectively. (E) Mid-neurula stage, dorsal view. (F) Late-neurula stage, dorsal view.

161A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

(Hogan, 1996), which involves multiple ligands, receptors,antagonists, modulators of antagonists (De Robertis et al., 2000),promoting factors (Little and Mullins, 2004; Rentzsch et al.,2006) and extracellular matrix proteins (Ohta et al., 2004;Moreno et al., 2005). However, how these signals areorchestrated to form a signaling network leading to the clearembryonic DV pattern remains elusive.

In this study, we introduce an extracellular protein, Del1(Developmentally regulated endothelial cell locus 1), as a newsignaling factor for dorsal determination in Xenopus. Del1 wasfirst isolated as an embryonic endothelial-specific gene in amouse enhancer-trap study (Hidai et al., 1998). Del1 encodes asecreted protein with two or three Notch-like EGF repeats(depending on the splicing form), an RGD motif, and twoDiscoidin domains (Supplementary Fig. S1), and is structurallyrelated to SED1, which is involved in sperm–egg binding(Ensslin and Shur, 2003). Secreted Del1 protein associateswith extracellular matrices such as the basement membrane(Hidai et al., 1998). Although Del1 promotes angiogenesis andthe remodeling of blood vessels (Penta et al., 1999), little is

known about its role in early embryogenesis. In this report, weshow that xDel is essential for DV patterning in the Xenopusembryo. Both gain- and loss-of-function studies indicate thatxDel promotes dorsalization of the mesoderm in the gastrulaembryo. We also show that this dorsalizing effect does notdepend on the RGD motif, which is essential for theangiogenic activity, but on the Discoidin domains.

Materials and methods

Isolation of xDel1, plasmid construction and in vitro transcription

A partial chick Del1 cDNA was isolated in a signal-sequence-trapscreen using a retrovirus-mediated expression system, SST–REX (Kojimaand Kitamura, 1999) of an HH9-11 chick head cDNA library. For theisolation of the frog counterpart, we designed primers according to adatabase sequence (AY491055), amplified full-length xDel1 cDNAs fromXenopus neurula mRNA by RT–PCR, sequenced and subcloned theminto pCS2+ vector. During this process, we noticed that xDel1 cDNAshave two alternative splicing variants, which encode two and three EGFrepeats, respectively (xDel1-E2 and xDel1-E3; Supplementary Fig. S1; bothvariants were expressed during early development). Injection of xDel1-E2

162 A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

Fig. 3. Knockdown of xDel1 causes ventralization in DMZ explants. (A) Control DMZ explants (equivalent to the mid-neurula stage). (B) xDel1-MO-injected DMZexplants. (C) RT–PCR analysis of DMZ explants (equivalent to stage 11): lane 1, whole embryo; lane 2, uninjected DMZ; lane 3, xDel1-MO-injected DMZ; lane 4,DMZ injected with 5-base-mispaired MO.

163A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

and xDel1-E3 caused indistinguishable effects on DV patterning in Xe-nopus. Since xDel1-E2 was translated relatively more efficiently andtherefore required less amounts of RNA, the data in this study weremainly obtained with the xDel1-E2 isoform. For the deletion study, xDel1-RGE, xDel1-delRGD, xDel1-delC (amino acid residues 1–155) and xDel1-delN cDNA (amino acid residues 156–481 and the signal sequence) weregenerated by PCR and subcloned into pCS2+. For mRNA injection, theplasmids were linearized with NotI and transcribed with SP6 polymerase(mMessage mMachine, Ambion). pCS2-eGFP was used as control RNAwhen needed.

Embryonic manipulation and microinjection of RNA and MO

Embryos were staged according to the normal table of Nieukoop andFaber. The embryos were then transferred into 0.1× Barth's solution untilfurther manipulation or harvesting. For explant assays, animal cap and MZtissues were excised with a tungsten needle at stages 9 and 10.5,respectively, and cultured in 1× LCMR supplemented with 0.2% BSA untilthe indicated stage. Injection of synthetic RNAs and MOs was done with afine glass capillary and a pneumatic pressure injector (Narishige) in 1×Barth's solution. For the MO experiments, the following MOs weredesigned (MO-A for pseudoallele a and MO-B for pseudoallele b): MO-A:5′-GCTAGCAACAGATAAGCGGAGATCC-3′, MO-B: 5′-TTCTA-GGCGTGGCTGAGTGCTGTGC-3′. MO-A and MO-B were mixed andinjected (6.25 ng/cell each) at the four-cell stage. Chd-MO was designedand used as described previously (Oelgeschlager et al., 2003). Allexperiments were performed at least twice to confirm reproducibility.Rescue experiments were done with the coding sequence of xDel1 RNA(allele a), which does not overlap with either MO.

RT–PCR, whole-mount in situ hybridization and luciferase assay

RT–PCR and whole-mount in situ hybridization analyses were performedas described previously (Mizuseki et al., 1998; Sasai et al., 2001; Matsuo-Takasaki et al., 2005). Quantitative PCR was performed using the 7500 FastReal-Time PCR System (Applied Biosystems) according to the manufacturer's

Fig. 2. Knockdown of xDel1 causes ventralization of the Xenopus embryo. (A) ExDel1-MO-injected embryos (MO-A and -B, 6.25 ng/cell each at the 4-cell stage)embryos. (D, G, J, M) Embryos injected with 5-base-mispaired control MO. (N, O)by whole-mount in situ hybridization with a probe for Chd (B–D, N), Gsc (E–G)

instruction and data were normalized by histone H4 expression. For luciferaseassays using the animal cap, the BRE (Xvent2)-luc reporter vector (vonBubnoff et al., 2005) was injected into the animal blastomeres of 8-cellembryos together with a loading-control plasmid (phRL-null) and theexperimental mRNAs. The amount of DNA was made constant by addingempty expression plasmid. Animal caps were excised at stage 9, cultured untilthe siblings reached stage 11 and subjected to luciferase assays using theDual-Luciferase Reporter Assay System (Promega) as previously described(Onai et al., 2004).

Results

Expression pattern of xDel1

We first isolated a partial Del1 cDNA in a signal-sequence-trap screen (the SST–REX method, which uses a retrovirusexpression system; Kojima and Kitamura, 1999) for newsecreted factors involved in chick development (see Materialsand methods). We became interested in this gene because of itsexpression in the dorsal and anterior tissues of early embryos(data not shown).

To elucidate the role of Del1 during early embryogenesis, weisolated a Xenopus homologue (xDel1) and analyzed itsexpression and functions during amphibian development. Inwhole-mount in situ hybridization analysis, maternal xDel1expression was detected widely on the animal side of the 8-cellembryo (Fig. 1A). At stage 9, xDel1 was expressed in theanimal pole and in the equatorial region of both dorsal andventral sides (Fig. 1B). In early gastrulation (Figs. 1C, D),xDel1 expression was observed in the marginal zone (both onthe dorsal and ventral sides; quantitative PCR analysis is shownin Supplementary Fig. S1B) and ectodermal tissues (Fig. 1D,

xternal appearance at the late-tailbud stage. Top, control. Middle and bottom,. (B, E, H, K) Uninjected gastrula embryos. (C, F, I, L) xDel1-MO-injectedCo-injection of xDel1-MO and xDel1 RNA (coding). Embryos were analyzed

, Vent1 (H–J, O) or Szl (K–M).

164 A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

closed and open arrowheads, respectively) and also in theinvoluting bottle cells (Fig. 1D, arrow). After mid-gastrulation,xDel1 expression was expressed in the dorsal ectoderm andmesoderm (Figs. 1E, F and data not shown).

Knockdown of xDel1 causes ventralization in the Xenopusembryo

To study the role of xDel1 in Xenopus embryogenesis, weperformed loss-of-function experiments using xDel1-MO (two

Fig. 4. Overexpression of xDel1 induces dorsal markers and suppresses ventral ones(C, D), Vent1 (E, F) or Szl (G, H). (A, C, E, G) Control. (B, D, F, H) Injection of

MOs against 5′ UTR sequences of pseudoalleles a and bwere combined; Supplementary Fig. S2A–D; the frequency ofthe phenotypes caused by each MO alone is shown inSupplementary Fig. S2E). Radial injection of xDel1-MO atthe 4-cell stage caused strong ventralization of the injectedembryo at the tailbud stage (Fig. 2A; top, control). In earlygastrula embryos, MO injection suppressed expression of thedorsal genes Chd (suppression in 59%, n=64; Fig. 2C) andGoosecoid (Gsc; 64%, n=28; Fig. 2F), while it induced theventral marker genes Vent1 (expansion in 88%, n=25; Fig.

. Whole-mount in situ hybridization analysis with a probe for Chd (A, B), GscxDel1 RNA (400 pg/cell, 4-cell stage) into the right blastomeres.

165A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

2I) and Sizzled (Szl; 75%, n=44; Fig. 2L) on the dorsal side.In contrast, the control 5-base-mispaired MO did not affectDV marker expression (Figs. 2D, G, J, M). The suppressionof Chd and ectopic expression of Vent1 induced by xDel1-MO were reversed by co-injecting xDel1 mRNA consisting ofthe coding-sequence only (Figs. 2N, O; suppression in 13%,n=38 for Chd and expansion in 22%, n=27 for Vent1,respectively).

These findings show that the secreted signal xDel1 isindispensable for dorsal determination of the Xenopusgastrula embryo.

We next tested the effect of xDel1-MO on the in vitrodifferentiation of DMZ explants (Fig. 3A). Injection ofxDel1-MO inhibited the elongation of the DMZ explants(round morphology in 89%, n=38; Fig. 3B). Consistent withthe in vivo findings, RT–PCR analysis (Fig. 3C) showed thatinjection of xDel1-MO (lane 3), but not of 5-base-mispairedMO (lane 4), suppressed expression of the dorsal genes(Chd and Gsc) and induced expression of the ventral genes(Vent1 and Szl) in the DMZ explant.

Fig. 5. Dorsalization of MZ explants by xDel1. (A–C) External appearance of VMradially injected with control (A), xDel1 (B) or Chd (100 pg/cell; C) RNA, and VMZ(lateral–ventral) and Vent1 (ventral) expression in the VMZ explants (stage 11); lane 1injected VMZ. (E) Animal cap assay. RT–PCR analysis of the mesodermal markers Xinjected animal caps; lane 4, CA-Smad2-injected animal caps.

Overexpression of xDel1 promotes dorsal development inXenopus embryos

We next examined the activity of xDel1 using a gain-of-function assay. Injection of xDel1 RNA induced ectopicexpression of Chd (41%, n=27; Figs. 4A, B) and Gsc (40%,n=25; Figs. 4C, D). In contrast, the ventral genes Vent1 andSzlwere suppressed (64%, n=39, and 92%, n=37, respectively;Figs. 4E–H).

Injection of xDel1 caused the elongation of VMZ explants(Figs. 5A, B; panel C shows positive controls with Chdinjection). RT–PCR analysis showed that xDel1 suppressed Szland induced Chd and Gsc (Fig. 5D, lane 3). These findingsindicate that xDel1 promotes dorsal development at the cost ofventral development.

We next asked whether xDe1l directly induced dorsalmesoderm from the blastula animal cap (Fig. 5E). Unlike aconstitutively active Smad2 construct (CA-Smad2; lane 4),which mimics Nodal/Activin signals (Armes and Smith, 1997),xDel1 failed to induce the dorsal mesodermal marker Gsc or the

Z explants (equivalent to the mid-neurula stage). Embryos (4-cell stage) wereexplants were prepared at stage 10.5. (D) RT–PCR analysis of Gsc, Chd, Wnt8. whole embryo; lane 2, control VMZ; lane 3, xDel1-injected VMZ; lane 4, Chd-bra and Gsc: lane 1, whole embryo; lane 2, control animal caps; lane 3, xDel1-

166 A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

pan-mesodermal marker Xbra in the animal cap explant (lane3). Taken together, these observations show that xDel1functions as a dorsalizing factor of mesodermal tissues, ratherthan directly inducing dorsal mesoderm from uncommittedcells.

Dorsalizing activity of xDel1 does not depend on its RGD motif

Previous studies demonstrated that mammalian Del1 exertsits angiogenic activity via the RGD motif (Supplementary Fig.S1), which acts on alpha(V) beta(3)-type integrin (Hidai et al.,1998; Penta et al., 1999). Therefore, we next tested whether thedorsalizing activity of xDel1 depended on the RGD motif.Interestingly, injection of a mutant xDel1 encoding a proteinthat lacks the RGD motif (replaced with RGE, which does notbind to integrin; Fig. 6A) dorsalized VMZ. The RGE mutantinduced Chd, Kielin (notochord and floor plate),MyoD (somite)and N-tubulin (neuron) and suppressed alpha-globin (bloodcell) and Szl (Fig. 6B, lane 5), as did the wild-type xDel1 (lane4). This demonstrates that the dorsalizing activity of xDel1 doesnot require the RGD motif, suggesting that the dorsalizing

Fig. 6. The RGD motif is not required for the dorsalizing activity of xDel1. (A)Schematic of mutant xDel1 constructs (xDel1-E2 type; see Materials andmethods). EGF, EGF repeat; RGD, RGD motif; Discoidin, Discoidin domain.(B) RT–PCR analysis of RNA-injected VMZ explants for dorsalizaton assay(late neurula stage-equivalent): lane 1, whole embryo; lane 2, control DMZ; lane3, control VMZ; lane 4, wild-type xDel1-injected VMZ; lane 5, xDel1-RGEmutant-injected VMZ; lane 6, xDel1-delN mutant-injected VMZ; lane 7, xDel1-delC mutant-injected VMZ.

activity of this protein involves a mechanism distinct from thepreviously known mechanism for angiogenic activity.

Furthermore, an N-terminal deletion mutant (mainly con-sisting of the two Discoidin domains and lacking the RGDmotif; Fig. 6A, bottom) had a dorsalizing effect on the VMZ(Fig. 6B, lane 6). In contrast, the C-terminal deletion mutant(lacking the Discoidin domains) lacked such an activity (lane 7).These findings show that, unlike the angiogenic activity, thedorsalizing activity of xDel1 is conveyed in the C-terminal half,which contains the Discoidin domains.

Effect of xDel1 on BMP signaling

Dorsal determination of the early Xenopus embryo ispromoted by the attenuation of BMP signals and/or by theenhancement of Wnt and Nodal signals. We therefore askedwhether xDel1 acts directly through these already-identifiedDV-patterning signals. To evaluate the output of BMPsignaling, we used a luciferase assay with the BRE-reporter(the Xvent2 promoter; von Bubnoff et al., 2005), whichresponses to the level of BMP signaling (Fig. 7A, lanes 3 and4; see the decrease and increase by Chd and BMP4,respectively). Injection of xDel1 substantially reduced lucifer-ase activity (Fig. 7A, lane 2). Furthermore, xDel1 injectionsuppressed the BRE-luciferase activity enhanced by BMP4(Fig. 7B, lanes 2 and 3), indicating that xDel1 has an anti-BMPeffect in this system.

Next, to clarify the mode of xDel1's action, we testedwhether xDel1 inhibited BMP signaling that was activated by aconstitutively active BMP receptor (CA-ALK2; Shibuya et al.,1998). The BRE-luciferase activity enhanced by CA-ALK2 wassuppressed by co-injection of xDel1 (Fig. 7C, lanes 2 and 3;similarly to the effect of Smad6, which inhibits Smad1; lane 4)but not by co-injection of Chd (negative control; lane 5). Thesefindings suggest that xDel1 triggers intracellular events thatattenuate the BMP pathway, instead of inhibiting BMP signalsoutside the cell.

Consistent with the inhibitory effect of xDel1 on BMPsignaling, the animal cap assay showed that xDel1 injectioninduced neural differentiation in the ectodermal explants (Fig.7D, lane 3).

In contrast, xDel1 injection did not increase the luciferaseactivity of the TOPflash reporter for Wnt signaling or of theARE reporter (with the Mix.2 promoter; Yeo et al., 1999.) forNodal/Activin signaling (data not shown). The lack of AREactivation is consistent with the observation that xDel1 did notinduce mesodermal differentiation in the animal cap (Fig. 5E).

Discussion

Role of xDel1 in dorsal development of the Xenopus embryo

In this report, we demonstrated that xDel1 plays an essentialrole in dorsal development of the Xenopus embryo. This is thefirst report of a function in early embryogenesis for the secretedsignaling factor xDel1. A previous report (Tsabar et al., 2005)showed that overexpression of N-terminus-truncated xDel1

Fig. 7. BRE-reporter activity is attenuated by xDel1. Animal caps injected with reporters were excised at stage 9 and harvested for the luciferase assay at stage 11. (A)The inhibitory effect of xDel1 (200 pg/cell, radial injection into all animal cells at the 8-cell stage) on BRE-reporter activity in the animal cap (lane 2). BMP4 (2.5 pg/cell, lane 4) and Chd (50 pg/cell, lane 3) were used as positive controls for activation and inhibition, respectively. (B) Injection of xDel1 (lane 3) inhibited theactivation effect by BMP4 (lane 2). BMP4+Chd (lane 4), positive control. (C) Injection of xDel1 (lane 3; 200 pg/cell) inhibited the activation effect by CA-BMPR(lane 2; 5 pg/cell). CA-BMPR+Smad6 (800 pg/cell; lane 4), positive control; CA-BMPR+Chd (lane 5), negative control. (D) RT–PCR analysis of animal capexplants: lane 1, whole embryo at stage 15; lane 2, control animal caps; lane 3, xDel1-injected animal caps; lane 4, Chd-injected animal caps.

167A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

(lacking the signal sequence) in the Xenopus embryo results inabnormal aggregation of maternal pigment granules in theembryo. However, this may be due to an artifact caused byoverproduction of the mutant protein in an inappropriate space(intracellular instead of extracellular), since a number ofprevious reports as well as our preliminary study show thatxDel1 is a secreted extracellular protein. We did not observealtered pigment accumulation in embryos injected with the full-length xDel1 or xDel1-MO (our unpublished observations).

The loss-of-function analysis showed that xDel1-MO causedventralization, which was reversed by xDel1 RNA injection,demonstrating the essential function of xDel1 in dorsal axisspecification. As compared with this strong ventralizationphenotype caused by the loss of xDel1 function, the over-expression of xDel1 in the ventral domain resulted in relatively

moderate dorsalization. Although it suppressed expression ofthe ventral genes such as Vent1 and Szl (Figs. 4F, H), expansionof Chd and Gsc expression was limited to the dorsal region (notreaching the ventral side; Figs. 4B, D). Similarly, althoughxDel1 injection elongated VMZ explants and suppressed theirSzl expression (Fig. 5), injection of xDel1 into a ventral–vegetal blastomere of the 8-cell embryo rarely produced astrong secondary axis (our unpublished observations; the VMZexplant assay is generally more sensitive in examiningdorsalizing activity than the axis-induction assay in vivo;Niehrs et al., 1994). These observations imply that xDel1 playsan essential role for dorsalizing the mesoderm on the dorsalside, but that its full dorsalizing activity may require certaindorsally located cofactors or be counteracted by ventrallydominant antagonists. An important question for future inves-

168 A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

tigation is whether xDel1 mainly plays a permissive role fordorsal determination or also acts as an instructive signal.

One of the possible downstream genes that mediate a partof xDel1's dorsalizing activity is Chd. In the VMZ assay, Chdis induced by xDel1 and the dorsal marker induction by xDel1was substantially attenuated by co-injection of Chd-MO(Supplementary Fig. S3A–C). However, the induction ofChd does not seem to explain all xDel1-induced phenotypes.For instance, in the same VMZ assay, unlike the induction ofthe dorsal genes, the suppression of the ventral genes byxDel1 was not efficiently reversed by co-injected Chd-MO(Supplementary Fig. S3D, E). In addition, the induction of N-CAM by xDel1 in the animal cap is unlikely to be mediated byChd, which was not expressed in xDel1-injected animal cells(Fig. 7D, lane 3). These observations suggest that a part ofxDel1's activity (but presumably not all of it) requiressecondarily induced Chd.

Molecular mechanisms of the xDel1 function

The detailed molecular mechanism of how the secretedfactor xDel1 dorsalizes embryonic tissues remains to beclarified in future investigation. One intriguing finding fromthis study is that this dorsalizing activity is not mediated by theintegrin-binding RGD motif, which is implicated in Del1'sangiogenic effect (Penta et al., 1999). Thus, xDel1 appears to bea multi-functional protein that mediates signals in more than onepathway.

One candidate pathway for xDel1 function that mediates thedorsalizing activity is the BMP signaling pathway. The BRE-promoter-luciferase assay indicated that xDel1 suppresses thesignaling activity caused by BMP4 (Fig. 7A). The neuralizingeffect of xDel1 in the animal cap is consistent with this idea(Fig. 7D). We have so far failed to observe the co-immunoprecpitation of xDel1 and BMP4 or BMPR proteinsin vitro (our preliminary observations). We infer that xDel1functions in a more intricate manner than Chd and Noggin,which act simply as BMP-binding inhibitors (Piccolo et al.,1996; Zimmerman et al., 1996). The finding that xDel1inhibits CA-BMPR-induced BRE-reporter activity is in accor-dance with this idea of an indirect and intracellular interactionof the BMP and xDel1 pathways, rather than directextracellular binding of these ligands (as shown for Chd).The identification of xDel1-binding proteins, such as non-integrin receptors and co-factors, is the important andchallenging next topic of study, which may elucidate a novelmode of action (or signaling pathway) in the regulation ofembryonic DV patterning.

Acknowledgments

We are grateful to Nick Love for the critical reading of themanuscript, to Tetsuro Watabe for advice on the luciferasereporter assay, and to Masako Suzuki for excellent assistance inthe maintenance of the frog facility. This work was supported inpart by grants-in-aid (YS) from MEXT, the Kobe ClusterProject and the Leading Project.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.ydbio.2007.03.008.

References

Armes, N.A., Smith, J.C., 1997. The ALK-2 and ALK-4 activin receptorstransduce distinct mesoderm-inducing signals during early Xenopusdevelopment but do not co-operate to establish thresholds. Development124, 3797–3804.

Carnac, G., Kodjabachian, L., Gurdon, J.B., Lemaire, P., 1996. The homeoboxgene Siamois is a target of the Wnt dorsalisation pathway and triggersorganiser activity in the absence of mesoderm. Development 122,3055–3065.

De Robertis, E.M., Larrain, J., Oelgeschlager, M., Wessely, O., 2000. Theestablishment of Spemann's organizer and patterning of the vertebrateembryo. Nat. Rev. Genet. 1, 171–181.

Ensslin, M.A., Shur, B.D., 2003. Identification of mouse sperm SED1, a bimotifEGF repeat and discoidin-domain protein involved in sperm-egg binding.Cell 114, 405–417.

Heasman, J., 2006. Patterning the early Xenopus embryo. Development 133,1205–1217.

Hidai, C., Zupancic, T., Penta, K., Mikhail, A., Kawana, M., Quertermous, E.E.,Aoka, Y., Fukagawa, M., Matsui, Y., Platika, D., Auerbach, R., Hogan, B.L.,Snodgrass, R., Quertermous, T., 1998. Cloning and characterization ofdevelopmental endothelial locus-1: an embryonic endothelial cell proteinthat binds the alphavbeta3 integrin receptor. Genes Dev. 12, 21–33.

Hogan, B.L., 1996. Bone morphogenetic proteins: multifunctional regulators ofvertebrate development. Genes Dev. 10, 1580–1594.

Keller, R., 2005. Cell migration during gastrulation. Curr. Opin. Cell Biol. 17,533–541.

Khokha, M.K., Yeh, J., Grammer, T.C., Harland, R.M., 2005. Depletion of threeBMP antagonists from Spemann's organizer leads to a catastrophic loss ofdorsal structures. Dev. Cell 8, 401–411.

Kojima, T., Kitamura, T., 1999. A signal sequence trap based on a constitutivelyactive cytokine receptor. Nat. Biotechnol. 17, 487–490.

Little, S.C., Mullins, M.C., 2004. Twisted gastrulation promotes BMP signalingin zebrafish dorsal-ventral axial patterning. Development 131, 5825–5835.

Matsuo-Takasaki, M., Matsumura, M., Sasai, Y., 2005. An essential role ofXenopus Foxi1a for ventral specification of the cephalic ectoderm duringgastrulation. Development 132, 3885–3894.

Mizuseki, K., Kishi, M., Matsui, M., Nakanishi, S., Sasai, Y., 1998. Xeno-pus Zic-related-1 and Sox-2, two factors inducer by Chordin, havedistinct activities in the initiation of neural induction. Development 125,579–587.

Moon, R.T., Kimelman, D., 1998. From cortical rotation to organizer geneexpression: toward a molecular explanation of axis specification in Xenopus.BioEssays 20, 536–545.

Moreno, M., Munoz, R., Aroca, F., Labarca, M., Brandan, E., Larrain, J., 2005.Biglycan is a new extracellular component of the Chordin-BMP4 signalingpathway. EMBO J. 24, 1397–1405.

Niehrs, C., Steinbeisser, H., De Robertis, E.M., 1994. Mesodermal patterning bya gradient of the vertebrate homeobox gene goosecoid. Science 263,817–820.

Oelgeschlager, M., Kuroda, H., Reversade, B., De Robertis, E.M., 2003.Chordin is required for the Spemann organizer transplantation phenomenonin Xenopus embryos. Dev. Cell 4, 219–230.

Ohta, K., Lupo, G., Kuriyama, S., Keynes, R., Holt, C.E., Harris, W.A., Tanaka,H., Ohnuma, S., 2004. Tsukushi functions as an organizer inducer byinhibition of BMP activity in cooperation with chordin. Dev. Cell 7,347–358.

Onai, T., Sasai, N., Matsui, M., Sasai, Y., 2004. Xenopus XsalF: anteriorneuroectodermal specification by attenuating cellular responsiveness to Wntsignaling. Dev. Cell 7, 95–106.

Osada, S.I., Wright, C.V., 1999. Xenopus nodal-related signaling is essential for

169A. Arakawa et al. / Developmental Biology 306 (2007) 160–169

mesendodermal patterning during early embryogenesis. Development 126,3229–3240.

Penta, K., Varner, J.A., Liaw, L., Hidai, C., Schatzman, R., Quertermous, T.,1999. Del1 induces integrin signaling and angiogenesis by ligation ofalphaVbeta3. J. Biol. Chem. 274, 11101–11109.

Piccolo, S., Sasai, Y., Lu, B., De Robertis, E.M., 1996. Dorsoventral patterningin Xenopus: inhibition of ventral signals by direct binding of Chordin toBMP-4. Cell 86, 589–598.

Rentzsch, F., Zhang, J., Kramer, C., Sebald, W., Hammerschmidt, M., 2006.Crossveinless 2 is an essential positive feedback regulator of Bmp signalingduring zebrafish gastrulation. Development 133, 801–811.

Sasai, Y., De Robertis, E.M., 1997. Ectodermal patterning in vertebrateembryos. Dev. Biol. 182, 5–20.

Sasai, Y., Lu, B., Steinbeisser, H., De Robertis, E.M., 1995. Regulation of neuralinduction by the Chd and Bmp-4 antagonistic patterning signals in Xenopus.Nature 376, 333–336.

Sasai, N., Mizuseki, K., Sasai, Y., 2001. Requirement of FoxD3-classsignaling for neural crest determination in Xenopus. Development 128,2525–2536.

Shibuya, H., Iwata, H., Masuyama, N., Gotoh, Y., Yamaguchi, K., Irie, K.,Matsumoto, K., Nishida, E., Ueno, N., 1998. Role of TAK1 and TAB1in BMP signaling in early Xenopus development. EMBO J. 17,1019–1028.

Smith, J.C., 1995. Mesoderm-inducing factors and mesodermal patterning. Curr.Opin. Cell Biol. 7, 856–861.

Tsabar, N., Gefen, A., Elias, S., Frank, D., 2005. Aggregation of maternalpigment granules is induced by the cytosolic discoidin domain of theXenopus Del1 protein. Dev. Dyn. 233, 224–232.

von Bubnoff, A., Peiffer, D.A., Blitz, I.L., Hayata, T., Ogata, S., Zeng, Q.,Trunnell, M., Cho, K.W., 2005. Phylogenetic footprinting and genomescanning identify vertebrate BMP response elements and new target genes.Dev. Biol. 281, 210–226.

Yeo, C.Y., Chen, X., Whitman, M., 1999. The role of FAST-1 and Smads intranscriptional regulation by activin during early Xenopus embryogenesis.J. Biol. Chem. 274, 26584–26590.

Zimmerman, L.B., De Jesus-Escobar, J.M., Harland, R.M., 1996. The Spemannorganizer signal noggin binds and inactivates bone morphogenetic protein 4.Cell 86, 599–606.