The Wnt signaling regulator R-spondin 3 promotes angioblast and vascular development

In vertebrate embryos, blood and vascular endothelial cells are formed during gastrulation from the mesoderm, and both lineages develop in close association. The vasculature forms from vascular endothelial precursor cells,or angioblasts, which proliferate and build a tubular network by a process of coalescence and fusion. Following vasculogenesis, growth and arborization of this network lead to branching angiogenesis, which continues in the adult organism (Risau and Flamme,1995). Vasculogenesis proceeds in close association with hematopoiesis, and in mouse these processes occur both extra-embryonically, in the yolk sac, the allantois and the placenta, as well as intra-embryonically,in the aortic primordia (Moore and Metcalf, 1970; Gekas et al.,2005; Ottersbach and Dzierzak,2005; Zeigler et al.,2006). The homologous compartments in the Xenopus embryo are the mesoderm of the dorsal lateral plate near the pronephros and pronephric ducts (intra-embryonic equivalent) and the ventral blood islands(extra-embryonic equivalent) (Cleaver et al., 1997; Cleaver and Krieg,1998; Ciau-Uitz et al.,2000; Walmsley et al.,2002).

The close association of hematopoietic and angioblastic cell development is probably due to the fact that both cell types derive from a mesodermal hemangioblast precursor, for which there is good evidence in zebrafish, Xenopus, chick, mouse and humans(Eichmann et al., 1997; Kennedy et al., 1997; Choi et al., 1998; Jaffredo et al., 1998; Nishikawa et al., 1998; Jaffredo et al., 2000; Walmsley et al., 2002; Huber et al., 2004; Ferguson et al., 2005; Vogeli et al., 2006; Kennedy et al., 2007)(reviewed by Red-Horse et al.,2007; Xiong, 2008)[but see Ueno and Weissman (Ueno and Weissman, 2006)].

Little is known about the growth factors that control the cell fate decision between hematopoietic cells and angioblasts, with exception of vascular endothelial growth factor (VEGF). VEGF plays a central role in many aspects of vascular development and embryonic angiogenesis(Carmeliet et al., 1996; Ferrara et al., 1996; Coultas et al., 2005)(reviewed by Ferrara, 1999; Tammela et al., 2005; Olsson et al., 2006). In particular, VEGF signaling is necessary and sufficient for promoting early endothelial differentiation in vertebrates(Carmeliet et al., 1996; Ferrara et al., 1996; Eichmann et al., 1997; Shalaby et al., 1997; Koibuchi et al., 2006).

Unlike for angioblast specification, a plethora of growth factors are known to regulate angiogenesis, including (but not limited to) VEGF, angiopoietin,PDGF, ephrin, Delta-Notch, BMPs, FGF and EGF (reviewed by Rossant and Howard, 2002). More recently, Wnt/Frizzled signaling has been added to this list. Wnts are glycoproteins that can activate multiple receptors and pathways, the best characterized of which is the Wnt/β-catenin pathway, which involves Frizzled (Fz) and LRP5/6 receptors (Logan and Nusse, 2004). Gain- and loss-of-function experiments have implicated Wnt signaling in promoting growth and differentiation of endothelial cells in vitro as well as angiogenesis in vivo (reviewed by Zerlin et al., 2008). However,the role of Wnt/β-catenin signaling in angioblast specification and vasculogenesis is unclear. R-spondins (Rspo1-Rspo4) encode a novel family of secreted proteins in vertebrates, which activate Wnt/β-catenin signaling and interact with LRP6 (Kazanskaya et al.,2004; Nam et al.,2006; Wei et al.,2007; Kim et al.,2008). R-spondins are involved in embryonic patterning and differentiation in frogs, mice and humans(Kazanskaya et al., 2004; Kim et al., 2005; Aoki et al., 2006; Blaydon et al., 2006; Parma et al., 2006; Aoki et al., 2008; Bell et al., 2008).

Here, we have focused on Rspo3, which we found to be prominently expressed in hematopoietic organs. Analysis of Rspo3 by gain- and loss-of-function experiments in both mouse and Xenopus reveals that this novel growth factor plays an essential role during vertebrate vasculogenesis and angiogenesis. In Xenopus embryos, Rspo3 regulates the balance between hematopoietic and endothelial differentiation by promoting angioblast specification and inhibiting blood cell specification. We go on to show that Rspo3 mouse mutants die of angiogenesis defects in yolk sac and placenta. To determine whether these effects represent a direct action on endothelial cells, we demonstrate that recombinant R-spondin promotes proliferation and angiogenesis in endothelial cell lines in vitro. Finally, we show that Rspo3 triggers Wnt/β-catenin signaling to activate the immediate early target gene Vegf, which mediates the effects of Rspo3. The results shed light on the poorly understood growth factor regulation of early angioblast specification and vascular development.

In vitro fertilization, embryo and explant culture, whole-mount staining,microinjection and culture of Xenopus embryo explants were carried out essentially as described (Gawantka et al., 1995). Prospective ventral blood islands were dissected from stage 13, 16 or 20 embryos by cutting the ventral of the embryo from the posterior limit of the cement gland to the anus. In situ hybridization on cryosections was performed according to Yang et al.(Yang et al., 1999). O-dianisidine staining was carried out without fixation as described(Huber et al., 1998). VEGFR inhibitors were used at 1 μM KRN633[N-(2-Chloro-4-((6,7-dimethoxy-4-quinazolinyl)oxy)phenyl)-N′-propylurea, Calbiochem] and 2 μM MAZ51[3-(4-Dimethylamino-naphthalen-1-ylmethylene)-1,3-dihydroindol-2-one,Calbiochem], respectively. For bead implantations, Affigel beads (BioRad) were rinsed in PBS and incubated with affinity-purified recombinant xRspo2 or VEGF(Promocell) at 1 μg/μl overnight at 4°C. Beads soaked in BSA were used as controls. BSA-, Rspo2- or VEGF-soaked beads were transplanted into the prospective ventral blood islands of neurulae (stage 15) and embryos fixed for analysis at tadpole stage (stage 25). The antisense morpholino oligonucleotide targeting X. tropicalis Rspo3 (gi: 114149217) and X. tropicalis VEGFA (gi: 77626411) were 5′atgcaattgcgactgctttctctgt and 5′cggtgaggctacagtgtaacagtgt.

RT-PCR assays were carried out with primers described previously (see Dosch et al., 1997; Glinka et al,. 1997);additional primers were XSCL (forward, actcaccctccagacaagaa; reverse,atttaatcaccgctgcccac), Xα-globin (forward,tccctcagaccaaaacctac; reverse, cccctcaattttatgctggac), Xmsr (forward,aacttcgctctcgctcctccatac; reverse, gccagcagatagcaaacaccac), XVEGF A(forward, aggcgagggagaccataaac; reverse, tctgctgcattcacactgac) and mα-globin (forward, actttgatgtaagccacggc; reverse,tagccaaggtcaccagcag).

qPCR assays were performed using an LC480 light cycler (Roche). Details of PCR primers used for amplification of mouse transcripts can be provided on request.

Mice carrying a targeted disruption of Rspo3 (gi:94388197) were obtained commercially from Artemis (Cologne). In brief, targeted mutagenesis of murine Rspo3 was carried out in mouse embryonic stem cells following standard procedures, using the targeting vector shown in Fig. S3A in the supplementary material. Transgenic mice were generated on a C57Bl/6 background via standard diploid injection. Homozygous mutant embryos were generated by heterozygote intercrosses. C57Bl/6 heterozygotes were then backcrossed to CD1 females for at least six generations. No significant phenotypic differences were detected between homozygous embryos in C57BL/6 and CD1 background. Mouse tail tips or portions of yolk sacs or embryos were used for genotyping by PCR. Genotyping was performed by PCR analysis using three primers: 5′ATGCTTTGAGGCTTGTGACC; 5′TGCACC GACTCCAGTACTGG; and 5′TACATTCTGGTTTCTCATCTGG.

Mice were mated overnight and the morning of vaginal plug detection was defined as embryonic day (E) 0.5. For routine histological analysis, tissues were fixed in 4% paraformaldehyde overnight and embedded in paraffin wax for sectioning. Sections (4 μm) were stained with Hematoxylin/Eosin. For whole-mount in situ hybridization, the embryos were fixed and processed as described previously (del Barco Barrantes et al., 2003).

BrdU labeling of embryos was performed in vitro to avoid the effects of placental-embryo deficiency. Whole concepti (E8.5 and E9.5) were dissected from the decidua and incubated at 37°C with 100 μM BrdU for 4 hours. Embryos and extra-embryonic tissues were fixed in paraformaldehyde and embedded in paraffin. Immunochemistry for BrdU was performed as previously described (Schorpp-Kistner et al.,1999). The percentage of BrdU incorporating cells was counted for three independent groups of siblings, using three wild-type and three mutant samples (allantois and chorionic plate) each.

BATGAL^+/-/Rspo3^-/- and BATGAL^+/-/Rspo3^+/+ mice were fixed at different stages and stained for β-galactosidase activity using a standard protocol. Then, allantois and chorionic plates were removed, embedded in Mowiol, photographed and the number of stained cells within a fixed area was counted. The number of β-galactosidase positive cells was counted for three independent groups of siblings, using three wild-type and three mutant samples (allantois and chorionic plate) each.

For quantification of nuclear β-catenin staining, embryos were photographed in a Zeiss confocal microscope (LSM510). The average intensity of Hoechst staining and background β-catenin staining in the cytoplasm of cells was determined using ImageJ(http://rsb.info.nih.gov/ij)in brightness-normalized images. The intensity of β-catenin staining in the nuclei was measured, background staining was subtracted, and the staining ratio between nuclear β-catenin and nuclear Hoechst was calculated for each nucleus.

Transfection of HEK293T cells with X. laevis Rspo2 (gi:54145367)and harvest of conditioned medium were as described(Kazanskaya et al., 2004). Anti-FLAG M2-Agarose beads (Sigma) were incubated overnight with X. laevis Rspo2-conditioned medium or control medium from untransfected HEK 293T cells. After washing, xRspo2FLAG recombinant protein was eluted from beads using FLAG peptide (Sigma).

Human umbilical vein endothelial cells (HUVEC; PromoCell) were cultured in endothelial cell growth medium (Promocell) supplemented with 10% fetal bovine serum (FBS). For proliferation studies, cells were plated at 50% confluence in a 96-well plate; the next day they were supplemented with VEGF and X. laevis Rspo2 proteins for 48 hours, after which BrdU (10 μM) was added to each well for 4 hours. BrdU analysis of cell proliferation was carried out using Cell Proliferation ELISA BrdU chemiluminescent from Roche Applied Science.

For chicken chorioallantoic membrane (CAM) assay, chicken eggs were incubated at 37°C in a humidified chamber. On day 3 of development, a window was made in the outer shell and on 6 day of development a 20 ml of xRspo2-FLAG or control beads or filter disk carrying recombinant VEGF(Sigma-Aldrich,100 ng/filter) was placed onto the surface of the CAM. The beads (anti-FLAG M2-Agarose, Sigma) were incubated overnight with X. laevis Rspo2 conditioned medium or control medium from untransfected HEK 293T cells and washed three times in PBS. After 5 days of incubation, the filter disks and the attached CAM were excised, washed with PBS and processed for histology using Hematoxylin/Eosin staining.

Endothelial cell spheroids of defined cell number were generated as described previously (Korff and Augustin,1998). In brief, 12 hours after transfection, HUVEC were suspended in culture medium containing 0.2% (w/v) carboxymethylcellulose (Sigma) and seeded in nonadherent round-bottom 96-well plates (Greiner). Under these conditions, all suspended cells contribute to the formation of a single spheroid per well of defined size and cell number (400 cells/spheroid). Spheroids were generated overnight, after which they were embedded into collagen gels. The spheroid containing gel was rapidly transferred into prewarmed 24-well plates and allowed to polymerize (30 minutes), then 100μl endothelial basal medium with or without the indicated growth factors was added on top of the gel. After 24 hours, in vitro capillary sprouting was quantified by measuring the cumulative length using a digital imaging software(Axioplan, Zeiss). In order to obtain a measure of the cumulative sprout length per spheroid, every sprout from 10-15 spheroids was assessed, and from these data the mean cumulative sprout length per spheroid was calculated.

HUVECs were seeded in six-well dishes at a density of 1.2×10⁵ cells per well 24 hours prior to experimental use. Predesigned annealed small interfering RNA (siRNA) (400 pmoles) control and directed to human VEGFR2 (Control siRNA and KDR ID#220 Ambion; http://www.ambion.com/)were transfected using oligofectamine (Invitrogen; http://www.invitrogen.com/)according to the manufacturer's instructions. The transfection was carried out in 1 ml OptiMEM (Invitrogen) for 4 hours. The medium was changed to ml ECGM. HUVEC spheroids were generated 24 hours after transfection. Downregulation of VEGF-R2 mRNA was analyzed using reverse-transcription polymerase chain reaction (RT-PCR) after 48 hours.

Following our finding that R-spondins are novel Wnt regulators and that Rspo2 is involved in early frog development(Kazanskaya et al., 2004), we wished to extend our analysis to other Rspo family members. We specifically focused on Rspo3, because it showed an early embryonic expression, suggesting an essential role during early patterning. The expression pattern of mouse Rspo3 was meanwhile reported(Aoki et al., 2006) and we confirm that expression of the gene shows a complex pattern: the brain, neural tube, tail, heart, somites and limbs (see Fig. S1A-C in the supplementary material; data not shown). In addition, we found that mouse Rspo3 was prominently expressed in sites of vasculogenesis and angiogenesis. Already in E8.0 embryos, Rspo3 is expressed in the posterior primitive streak(Huber et al., 2004) and the allantois, which are prominent sources of hematopoietic cells(Fig. 1A). Rspo3expression persists in the allantois at later stages(Fig. 1B) and also becomes expressed in the chorionic plate of the placenta at E9.0(Fig. 1B,C; see Fig. S1D,F in the supplementary material), and in heart and umbilical cord starting from E9.5 (see Fig. S1A-C in the supplementary material). Rspo3 is also expressed in the blood vessels of the embryo proper(Fig. 1D).

Similar to mouse, Xenopus Rspo3 is expressed in the CNS, and in a variety of precursors of the vasculature, such as dorsal lateral plate and ventral blood islands (Fig. 1E-K), which give rise to adult and embryonic blood, respectively. In particular, Rspo3 shows prominent co-expression with or nearby the angioblast marker Msr (Devic et al., 1996) in the vitelline vein precursors in the periphery of the vbi, surrounding and partially overlapping the more centrally expressed Scl, an early marker of blood forming cells(Mead et al., 1998)(Fig. 1I-K; see Fig. S2 in the supplementary material). Rspo3 is also co-expressed with Vegf in the hypochord (Fig. 1G,H), a tissue that is important for formation of the dorsal aorta (Cleaver et al., 2000). We conclude that Rspo3 is prominently expressed in or close by endothelial cells, and their precursors in Xenopus and mouse.

To investigate a possible role in Xenopus development, we inhibited Rspo3 by injection of antisense morpholino oligonucleotides. The resulting tadpoles displayed ventral edema, as they are characteristically seen following vascular defects(Fig. 2A,B). Staining of erythrocytes with o-dianisidine confirmed that the experimental embryos develop fewer vessels and accumulate erythrocytes in the ventral side(Fig. 2C,D). Molecular marker analysis revealed an expansion of the hematopoietic markers α-globin and Scl in the ventral blood islands (vbi). By contrast, Msrexpression was reduced, which was most pronounced in the lateral plate region,where vbi-derived angioblasts normally develop into the vitelline network(Fig. 2E-L). Rescue experiments carried out in ventral marginal zone explants (VMZ), which contain the vitelline vein and vbi precursors, validated the specificity of the morpholino: α-globin expression was expanded in Rspo3morpholino-injected VMZs, whereas expression was blocked by co-injection of morpholino and Rspo2 mRNA (Fig. 2M-O). These results suggest a requirement of Rspo3 in promoting angioblast fate at the expense of the hematopoietic lineage in the vbi.

To confirm this conclusion, we misexpressed Rspo2 (which is better produced than Rspo3). To rule out the possibility that early embryonic misexpression of Rspo2 may act indirectly by affecting early patterning rather than by directly regulating hemangiogenic fates, we implanted beads soaked with recombinant Rspo2 or VEGF into the vbi precursors of late neurulae. We thereby avoided global overexpression, and targeted R-spondin signaling to the hematopoietic lineage. Rspo2 and VEGF beads both repressed α-globin, and instead induced patches of Msr expression(Fig. 2P-U). Furthermore, Rspo2 induced robust Vegf expression around the bead(Fig. 2V,W). This confirms that R-spondin signaling regulates the decision between blood and endothelial cell fates in the vbi and suggests that its effects are mediated by VEGF.

To further molecularly dissect R-spondin signaling in Xenopus, we used VMZ explants of microinjected embryos followed by RT-PCR analysis of marker genes. RT-PCR analysis showed that Scl, α-globin, Msr and Vegf are expressed in VMZ(Fig. 3A-D). Rspo3morpholino injection upregulated the expression of Scl andα-globin at the expense of Msr as well as of Vegf(Fig. 3A). Conversely, Rspo2 and Rspo3 mRNA injection, which generally behave indistinguishably in overexpression, induced Msr and Vegf,and blocked expression of Scl and α-globin(Fig. 3A,B). The effects of Rspo3 gain- and loss-of-function were phenocopied by up- and downregulation of Wnt ligand, following microinjection of wild-type or dominant-negative Wnt8 (Hoppler et al., 1996), respectively(Fig. 3A). Similar to dnWnt8, mRNA microinjection of Dkk1, a specific inhibitor of the Wnt receptor LRP5/6 (Glinka et al.,1998), upregulated Scl and downregulated angioblast markers (Fig. 3B). Furthermore Dkk1 blocked the ability of Rspo3 to upregulate Msrand Vegf (Fig. 3B,lanes 7-8). These results confirm that during vasculogenesis Rspo3 acts as promoting Wnt/β-catenin signaling.

As VEGF is known to control cell fate decision between hematopoietic cells and angioblast in chick (Eichmann et al.,1997), we tested two pharmacological VEGF receptor inhibitors (MAZ 51 and KRN 633) (Kirkin et al.,2001; Nakamura et al.,2004). Treatment of VMZs with both drugs at low μM concentration mildly phenocopied the effects of Rspo3 inhibition(Fig. 3B, lanes 5-6) and partially inhibited the effect of Rspo3 mRNA overexpression(Fig. 3B, lanes 10-11). As an alternative to VEGF receptor inhibitors, we used Vegf Morpholino injection, which, like Rspo3 Mo, upregulated Scl expression(Fig. 3C). These results indicate that VEGF promotes endothelial at the expense of blood cell differentiation in Xenopus. To corroborate the VMZ explant data, we used explanted prospective vbi from different embryonic stages for marker gene analysis (Fig. 3D). This confirmed Vegf expression in the vbi by qPCR, which was reduced by Rspo3 Mo. This was accompanied by downregulation of Msr and upregulation of Scl.

In gain-of-function, Vegf mRNA injection upregulated Msrand blocked Scl expression (Fig. 4A, lane 3), similar to Rspo2/3 mRNAs. Importantly, Vegfoverexpression rescued Msr expression in Rspo3-depleted embryos (Fig. 4A, lane 5). Conversely, Vegf morpholino injection blocked the ability of Rspo2-soaked beads to inhibit α-globin expression(Fig. 4B), confirming the effects seen with VEGF receptor inhibitors. This indicates that R-spondin signaling is mediated by VEGF.

To test whether Vegf is an immediate early R-spondin target gene,we carried out induction experiments in the presence of cycloheximide. Xenopus animal caps (responders) were transiently conjugated with Wnt3a- or Rspo2-expressing animal caps (inducers) and then separated again to analyze the responders in isolation (see Fig. 4C for experimental set up). In these conjugates, responder caps showed Vegf induction, even when they were cycloheximide treated (Fig. 4D). This immediate early Vegf induction by Rspo2 and Wnt3a is consistent with the fact that Vegf is a directβ-catenin target gene, which harbors seven Tcf/Lef-binding sites(Easwaran et al., 2003).

We conclude that: (1) Rspo3 regulates the balance between hematopoietic and angioblast cell fate in the vbi; (2) it does so by promoting Wnt/β-catenin signaling, which (3) is required for expression of Vegf, that (4) acts as distal regulator of both cell fates in this cascade, promoting angioblast and inhibiting hematopoietic cell fate.

To extend the analysis to mammals, we generated Rspo3 mutant mice by targeted gene disruption (see Fig. S3 in the supplementary material). Homozygous Rspo3 mutant embryos have been previously described(Aoki et al., 2006). The mice were reported to arrest development around E10 because of placental defects,the nature of which remained unknown. We confirmed that Rspo3 mutant embryos arrest development at around day 10 and show placental defects (see below). In addition, we found that mutant mice show hemorrhages and have pale yolk sacs and placentas (Fig. 5A-C). These are characteristic features of mice that display embryonic vascular deficiencies. PECAM staining and histological analysis confirmed that the yolk sac of mutant embryos had an underdeveloped vasculature (Fig. 5D,E),indicative of an angiogenesis defect. Similar defects were seen in the vasculature of the embryo proper (Fig. 5F). Histological inspection of the placenta at E10.5 showed that embryonic blood vessels failed to invade the labyrinth layer in Rspo3mutants (Fig. 5G). Vascular staining for PECAM indicated that the primary capillary plexus was formed and vessels appeared normal at E9.5 (not shown). However, these vessels failed to undergo proliferation and remodeling, as is characteristic for later angiogenesis (Fig. 5H). These results indicate that Rspo3 signaling in the placenta is required to promote endothelial cell growth and remodeling, rather than initial differentiation and formation of the primary plexus. Consistent with this interpretation, BrdU labeling experiments showed a marked decrease in proliferating endothelial cells in the mutants by E9.5 (see below, Fig. 7E).

We next asked whether as in Xenopus, mouse Vegf may be regulated by Rspo3. Similar to Rspo3, VEGF is required for placental development (Carmeliet et al.,1996; Ferrara et al.,1996; Coultas et al.,2005) and Rspo3 is co-expressed with Vegf and the endothelial marker Vegfr2 (Flk1) in the placenta (see Fig. S1D-F in the supplementary material). Whereas Vegfr2 expression in endothelia was mostly unaffected (Fig. 6A), expression of Vegf was strongly reduced when analyzed by both in situ hybridization(Fig. 6B) and RT-PCR (see Fig. S4 in the supplementary material). By contrast, expression of Tpbpaand Csh1, markers of giant trophoblast and spongiotrophoblast cells that produce VEGF (Coultas et al.,2005), was mostly unaffected(Fig. 6C,D). We conclude that Rspo3 is required for Vegf expression and for endothelial cell proliferation to promote proper vascularization of the mouse placenta.

Interestingly, two Wnt/β-catenin pathway components, Wnt2 and frizzled5, are also required for the vascularization of the mouse placenta(Monkley et al., 1996; Ishikawa et al., 2001) and evidence for a general role of Wnt/β-catenin signaling in angiogenesis is accumulating (Zerlin et al.,2008). To corroborate therefore that the vascular defects in Rspo3 mutants were also due to reduced Wnt/β-catenin signaling,we carried out immunofluorescence staining to monitor nuclear β-catenin. A marked reduction of nuclear β-catenin staining was observed in the mutant allantois (Fig. 6E,G). To confirm this downregulation of Wnt signaling, we bred Rspo3heterozygotes with the Wnt-reporter mouse line BATGAL(Maretto et al., 2003), which drives lacZ expression at sites of Wnt signaling. Staining forβ-galactosidase showed a lacZ signal in the chorionic plate of wild-type BATGAL mice, which was reduced in Rspo3^-/-/BATGAL allantoises and placentas(Fig. 6F,G). Taken together,these results suggest that Rspo3 functions in placental vascular remodeling by promoting Wnt/β-catenin signaling to activate Vegfexpression.

As the results indicated that Rspo3 is required for vascular development,we determined whether R-spondin signaling may directly promote angiogenesis by activating endothelial cells. As recombinant protein we used both Rspo2 and Rspo3, because the former is better produced, whereas in our experience both are qualitatively interchangeable. Characteristic properties of angiogenic factors are their ability to induce endothelial proliferation and capillary morphogenesis. In chicken chorionic membrane (CAM) assays(Fig. 7A-C), implantation of Rspo3-soaked beads induced a strong proliferative response in the CAM,accompanied by intense angiogenesis. A similar induction was seen with VEGF-soaked beads. Likewise, in human umbilical vein endothelial cells(HUVEC), recombinant Rspo2 induced cell proliferation, as did VEGF(Fig. 7D). Correspondingly, the proliferation of endothelial cells labeled in situ with BrdU in pregnant mice was reduced in Rspo3-null mutants(Fig. 7E).

To test whether R-spondin signaling can also induce the morphogenesis that is characteristic of angiogenic factors, we employed the human dermal microvascular endothelial cells (HDMEC) capillary tube formation assay, where Rspo2 readily induced capillary-like morphology(Fig. 7F,G).

Another angiogenic morphogenesis test is the spheroidal sprouting assay, in which collagen gel-embedded endothelial cells give rise to a three-dimensional capillary-like network upon angiogenic stimulation(Korff and Augustin, 1998). Similar to VEGF, Rspo2 induced endothelial sprouting in a dose-dependent manner (Fig. 7H-L). The ability of Rspo2 to induce sprouting was blocked by addition of the Wnt inhibitor Dkk1 but not by the related Dkk3, which does not inhibit Wnt signaling(Mao et al., 2001)(Fig. 7K). In contrast to Rspo2, the sprouting effect of VEGF was not affected by Dkk1. Once again the effects were dependent on VEGF signaling, as siRNA knockdown of Vegfr2 (>70% by RT-PCR, not shown) blocked the ability of Rspo2 to induce sprouting (Fig. 7L). We conclude that R-spondin signaling can act directly on endothelial cells to promote angiogenesis by activating Wnt/β-catenin and VEGF signaling.

An important issue in developmental biology relates to the mechanisms whereby multipotent precursors give rise to distinct differentiated progeny. One such fate decision, which is poorly understood, concerns the choice between hematopoietic and blood cell differentiation in hematopoietic organs where these cells develop in close association. The co-specification of these lineages is supported by a large body of evidence in all vertebrates,including chick, mouse, zebrafish and Xenopus (reviewed by Ferguson et al., 2005; Red-Horse et al., 2007; Xiong, 2008). In the hematopoietic regions of Xenopus embryos, for example, a combination of lineage tracing and gene expression experiments showed that both lineages pass through a progenitor state in which they co-express blood and endothelial genes (Ciau-Uitz et al., 2000; Walmsley et al., 2002). The co-development of both lineages and their requirement for the same genes begs the question of which cues regulate their separate specification? Although a number of growth factor signaling pathways, such as VEGF, TGFβ, BMP4 and hedgehog, have been identified that are required directly or indirectly for formation of both lineages (Dickson et al., 1995; Winnier et al.,1995; Carmeliet et al.,1996; Ferrara et al.,1996; Shalaby et al.,1997; Dyer et al.,2001; Vokes et al.,2004; Gering and Patient,2005), the extrinsic factors that regulate the fate separation between these lineages remain largely unknown. A main conclusion of our study is that Rspo3-Wnt/β-catenin signaling plays a key role in this process,by promoting angioblast and inhibiting hematopoietic cell fate in Xenopus ventral blood island (Fig. 8). The early embryonic expression of Rspo3 in the gastrula ventrolateral mesoderm in Xenopus is consistent with this proposition.

In support of our model, constitutive activation of Wnt/β-catenin signaling has been shown to block hematopoietic cell differentiation in transgenic mice (Kirstetter et al.,2006; Scheller et al.,2006). Furthermore, experiments with Wnt2^-/- embryoid bodies suggest that Wnt/β-catenin signaling is required for normal endothelial maturation and vascular plexus formation, and that Wnt2 suppresses hematopoietic differentiation in vitro(Wang et al., 2007).

The molecular analysis indicates that an essential mediator of Rspo3 is its immediate downstream target gene Vegf. In support of its crucial role, VEGF in chicken progenitor cells reduces hematopoietic and induces endothelial cell differentiation (Eichmann et al., 1997). A role of VEGF in regulating angioblastic versus hematopoietic cell fate decision is also corroborated by recent experiments in Xenopus (Koibuchi et al.,2006). As it acts upstream of the VEGF pathway, Rspo3 signaling may be the earliest step regulating this hematopoietic lineage bifurcation in the Xenopus vbi.

Our results indicating a negative role of VEGF in Scl expression and blood cell specification are in contrast to a body of studies specifically in mouse. For example, VegfR2-deficient mice lack both endothelial and hematopoietic cells, suggesting that VEGF signaling is essential in a common progenitor (Shalaby et al.,1997). Other mouse data also indicate a requirement for VEGF signaling for Scl expression and blood cell differentiation (reviewed by Ferrara, 1999; Tammela et al., 2005; Olsson et al., 2006). The reason for this discrepancy may be differences in the role of Scl and VEGF in regions of embryonic and definitive hematopoiesis, respectively. In mouse, Scl is a general hematopoietic marker expressed in common precursors of blood and endothelial cells, and required for the development of both lineages (Shivdasani et al.,1995; Visvader et al.,1998). Also in the Xenopus dorsal lateral plate and anterior vbi, which are responsible for definitive hematopoiesis, Scland the endothelial marker xfli are co-expressed, suggestive of a common precursor (Walmsley et al.,2002). By contrast, in the Xenopus vbi, which give rise to embryonic blood and which is the focus of our study, Scl is a red blood lineage marker, the expression of which is mutually exclusive with the endothelial marker xfli (Walmsley et al., 2002) (compare Fig. 1I with Fig. 1K). As our study focuses on Xenopus primitive hematopoiesis, the role of Vegf and Scl may be different in organs of definitive hematopoiesis. We note, however, that VEGF signaling in Sclinduction, primitive hematopoiesis and vasculogenesis is also not conserved between mouse and either zebrafish or chicken(Eichmann et al., 1997; Habeck et al., 2002; Patterson et al., 2005).

The defects in mutant mice reveal that embryonic vasculature regulation is a conserved function of Rspo3-Wnt/β-catenin signaling, albeit at a different stage, namely in angiogenesis of the yolk sac and the placenta,where it is required for endothelial cell proliferation and remodeling. This is also supported by the pro-angiogenic effect of R-spondin signaling in adult endothelial cells. Taken together with its endothelial expression, this suggests an autocrine mode of Rspo3 signaling. Recently, it was shown that VEGF also functions in an autocrine fashion in endothelial cells during vascular development (Lee et al.,2007). Indeed, the rather broad effect of Rspo3 on early specification as well as later angiogenesis is reminiscent of VEGF, which likewise plays a role not only during early hematopoietic cell specification,but also during subsequent cell proliferation, vascular morphogenesis and remodeling, both in the embryo and in the adult(Ferrara 1999; Coultas et al., 2005).

There is large evidence that Wnt signaling plays an important role in angiogenesis. Wnt2, Wnt4 and Wnt7b mutant mice have reduced embryonic or adult vasculature (Monkley et al., 1996; Shu et al.,2002; Jeays-Ward et al.,2003). Likewise, Fz4 and Fz5 mutant mice show defective angiogenesis (Ishikawa et al.,2001; Xu et al.,2004). The Fz4 mutation is also linked to human familial exudative vitreoretinopathy (FEVR), which is caused by a failure of peripheral retinal vascularization (Robitaille et al., 2002). Mutations in norrin, which encoding a high affinity Fz4 ligand, are also linked to FEVR(Xu et al., 2004). However, in most cases it remains unclear whether the angiogenic role revealed by these Wnt/Fz mutants involves Wnt/β-catenin signaling. At least in vitro it is the Wnt/β-catenin signaling pathway, which promotes endothelial cell proliferation (reviewed by Zerlin et al.,2008). Furthermore, Wnt/β-catenin signaling promotes expression of Vegf (Zhang et al.,2001), which is a direct β-catenin target gene(Easwaran et al., 2003).

Consistent with this, we find that R-spondins also promote endothelial cell proliferation and in vitro angiogenesis by activating Wnt/β-catenin signaling, as it is inhibited by Dkk1. This effect is blocked by siVEGFR2,suggesting that, as in Xenopus and mouse embryos, R-spondins impact endothelial cells ultimately through VEGF signaling. However, in HUVEC, we were unable to observe upregulation of Vegf by Rspo2 treatment (data not shown), whereas in Xenopus and mouse embryos, Rspo3 was required for Vegf expression. This suggests that there may be different mechanisms underlying the promotion of VEGF signaling, endothelial proliferation and morphogenesis by Rspo3.

Finally, from a medical perspective, the identification of Rspo3 as a secreted angiogenic factor may provide novel opportunities for pharmaceutical intervention in angiogenic and antiangiogenic therapies (reviewed by Ferrara, 1999; Dor et al., 2003).

We thank M. Schorpp-Kistner and P. Angel for advice and reagents. We thank Y. Audigier, M. Gessler, P. Gruss and S. Piccolo for providing materials, and K. Ellwanger, Y.-L. Huang and H. Weber for technical help. Supported by DFG(Ni 286/9-2 and Au83/9-1).