Deletion of Vascular Endothelial Growth Factor C (VEGF-C) and VEGF-D Is Not Equivalent to VEGF Receptor 3 Deletion in Mouse Embryos

Abstract

Lymphatic vessels play an important role in the regulation of tissue fluid balance, immune responses, and fat adsorption and are involved in diseases including lymphedema and tumor metastasis. Vascular endothelial growth factor (VEGF) receptor 3 (VEGFR-3) is necessary for development of the blood vasculature during early embryogenesis, but later, VEGFR-3 expression becomes restricted to the lymphatic vasculature. We analyzed mice deficient in both of the known VEGFR-3 ligands, VEGF-C and VEGF-D. Unlike the Vegfr3^−/− embryos, the Vegfc^−/−; Vegfd^−/− embryos displayed normal blood vasculature after embryonic day 9.5. Deletion of Vegfr3 in the epiblast, using keratin 19 (K19) Cre, resulted in a phenotype identical to that of the Vegfr3^−/− embryos, suggesting that this phenotype is due to defects in the embryo proper and not in placental development. Interestingly, the Vegfr3^neo hypomorphic mutant mice carrying the neomycin cassette between exons 1 and 2 showed defective lymphatic development. Overexpression of human or mouse VEGF-D in the skin, under the K14 promoter, rescued the lymphatic hypoplasia of the Vegfc^+/− mice in the K14-VEGF-D; Vegfc^+/− compound mice, suggesting that VEGF-D is functionally redundant with VEGF-C in the stimulation of developmental lymphangiogenesis. Our results suggest VEGF-C- and VEGF-D-independent functions for VEGFR-3 in the early embryo.

The lymphatic vascular system plays crucial roles in the regulation of tissue fluid homeostasis, immune surveillance, and fat adsorption and is involved in the pathogenesis of diseases such as lymphedema and tumor metastasis (4). In lymphedema, the formation or function of lymphatic vessels is defective, leading to fluid collection in tissues, chronic swelling, tissue fibrosis, adipose degeneration, and susceptibility to infections (26). Tumor metastasis to lymph nodes involves cell spread via lymphatic vessels that is enhanced by vascular endothelial growth factor C (VEGF-C) and VEGF-D (2).

The blood vascular system consists of a hierarchy of vessels formed by angiogenesis from a primitive vascular network. The lymphatic vascular system comprises a separate network of capillaries and collecting vessels that permeate most organs of the body. In embryos, lymphatic vessel formation starts when a subset of endothelial cells in the cardinal vein commits to the lymphatic lineage expressing the homeobox transcription factor Prox1, and cells sprout to form the primary lymph sacs (31, 36). In mice, the lymphatic system starts to develop at embryonic day (E) 10.5, when the cardiovascular system is already functional (33). Peripheral lymphatic vessels form by centrifugal sprouting from the lymph sacs and form a network, followed by maturation of large collecting lymphatic vessels.

VEGF receptor 3 (VEGFR-3) is expressed almost exclusively in the lymphatic endothelium in adults (15), whereas in mid-gestation embryos, this receptor-type tyrosine kinase is expressed on blood vessels and is required for remodeling of the blood vascular network (9). Mice that lack a functional Vegfr3 gene die at E10.5, before the emergence of the lymphatic vessels (9). Heterozygous missense mutations in the Vegfr3 gene that inactivate the tyrosine kinase in the encoded protein do not lead to the arrest of embryonic angiogenesis; instead, they have been linked to lymphedema in humans and in Chy mice, which carry such mutations in their germ line (16, 18). Growth of the compound Chy; Vegfr3^+/− mice appears to be retarded at approximately the same stage at which the Vegfr3^−/− embryos die, suggesting a lack of signaling through the VEGFR-3(I1053F) mutant (18). Adenoviral or transgenic expression of a soluble VEGFR-3 that competes with the endogenous receptor for VEGF-C and VEGF-D binding induces lymphatic vessel regression in late-gestation embryos or during the first 2 postnatal weeks but not thereafter (19, 23).

VEGF-C and VEGF-D are the only ligands known to activate VEGFR-3 (1, 14). Both human VEGF-C and human VEGF-D bind to VEGFR-2, whereas mouse VEGF-D is a VEGFR-3-specific ligand (5). We have previously shown that VEGF-C is essential for the formation of lymph sacs from embryonic veins. In the Vegfc^−/− mice, endothelial cells commit to the lymphatic lineage but do not sprout to form lymphatic vessels, which leads to death before birth due to fluid accumulation in tissues (17). Furthermore, VEGF-C haploinsufficiency results in hypoplasia of the lymphatic vessels in the newborn Vegfc^+/− mice and lymphedema in adult mice, indicating that both Vegfc alleles are required for normal lymphatic development (17). Surprisingly, the Vegfd knockout mice have only a subtle lymphatic phenotype involving a decrease in the abundance of lymphatic vessels in the lungs, suggesting that VEGF-D is dispensable for lymphatic vessel development (6). Lymphangiogenesis is, however, stimulated in the skin of transgenic mice overexpressing human VEGF-D, human VEGF-C, or a VEGFR-3-specific mutant form of human VEGF-C (VEGF[C156S]) in basal skin keratinocytes under the K14 promoter (13, 35).

To clarify the molecular mechanisms underlying the role of the VEGFR-3 signaling pathway in embryonic angiogenesis, we produced Vegfc; Vegfd double-knockout mice. Surprisingly, the double-knockout mice survive without problems of angiogenesis and reproduce essentially the Vegfc knockout phenotype. The Vegfr3^neo hypomorphic mutant mice also lacked a blood vascular phenotype but showed defective lymphatic vessel development. Epiblast-restricted ablation confirmed that cardiovascular defects in the Vegfr3^−/− embryos are due to the lack of Vegfr3 in the embryo proper and excluded the placental contribution to the phenotype. We also demonstrate that the K14-VEGF-D transgene can rescue the lymphatic vessel hypoplasia in the skin of the Vegfc^+/− mice. These results suggest that other ligands might signal via the VEGFR-3 pathway during embryonic angiogenesis or that ligand-independent signaling mechanisms could play a role.

MATERIALS AND METHODS

Transgenic and knockout mice.

The Vegfr3^+/lz (9), Vegfc^+/lz (17), Vegfd^+/lz (6), K14-hVEGF-D (35), β-actinFLPe (27), keratin 19 (K19) Cre (24), PGKCre (22), and ROSA26Cre mice (30) have been described previously. The Vegfr3 and Vegfc gene-targeted mice were genotyped as previously described (9, 17). The Vegfd-targeted mice were genotyped by PCR from tail or yolk sac DNA, using the forward primer 5′-CTTTCTCCCATACTAAGATTG and the reverse primers 5′-CCAATAAAGAGAAATATTCAAGC (wild-type allele) and 5′-AAGTTGGGTAACGCCAGG (targeted allele), which amplified a 341-nucleotide (nt) fragment for the wild-type allele and a 286-nt fragment for the targeted allele.

The K14-mouseVEGF-D (K14-mVEGF-D) transgenic mice were generated by cloning a DNA fragment consisting of base pairs 282 to 1360 of mouse VEGF-D cDNA (GenBank accession number X99572) into the K14 promoter expression cassette (34). The linearized construct was injected into fertilized FVB mouse oocytes, and the transgenic mice were genotyped by PCR from tail DNA, using the forward primer 5′-GAAAGCCCAAAACACTCCAAAC and the reverse primer 5′-CTGAGCGTGAGTCCATACTG, which amplified a 460-nt fragment from the transgenic mice.

For construction of the Vegfr3^neo allele, the first exon and part of the first intron of the Vegfr3 mouse gene were flanked with loxP sites. The 3.0-kb 5′ arm was cloned as a KpnI-NotI fragment and the 4.8-kb 3′ arm as a SacII-HindIII fragment into the pKO vector (Lexicon Genetics) containing a thymidine kinase cassette. The NotI-SacII fragment containing exon 1 of the Vegfr3 gene was cloned into a blunted BsaI site in the NotI-AscI cassette of the plasmid, which contained loxP-BsaI-loxP-frt-neomycin-loxP-frt. Subsequently, the NotI-AscI fragment was cloned between the 5′ and 3′ arms to generate the final targeting construct, in which the first exon was flanked with loxP sites and the neomycin resistance gene was flanked with frt sites in the first intron. The construct was electroporated into R1 (progeny of 129/Sv × 129/SvJ) embryonic stem cells. Positive clones identified by Southern blot analyses were aggregated with ICR strain morulas to obtain chimeric mice, which were bred with ICR mice. The mice were genotyped by PCR using the forward primer 5′-TCACTCCCAGCCTAGAGCTGC (the Vegfr3 promoter upstream of the NotI site) and the reverse primer 5′-CGAGGCAGAGCCACAGGCGC (exon 1), which amplified a 95-nt fragment for the wild-type and a 165-nt fragment for the targeted, floxed allele. The recombination status of the floxed Vegfr3 allele was detected by PCR using the primers 5′-TCACTCCCAGCCTAGAGCTGC and 5′-CCTCGAGGTCGACGGTATC, which amplified a 445-nt fragment for the floxed and a 115-nt fragment for the recombined allele.

All animal experiments were conducted in accordance with the guidelines set by the Committee for Animal Experiments of the District of Southern Finland.

Immunostaining.

For whole-mount analyses, tissues were fixed in 4% paraformaldehyde and used for immunoperoxidase staining, using a Vectastain ABC kit (Vector Laboratories), or for immunofluorescence staining. Alternatively, tissues were dehydrated in graded concentrations of ethanol, embedded in paraffin, and cut into 6-μm sections. Paraffin sections were used for immunoperoxidase staining with a tyramide signal amplification kit (Perkin Elmer Life Sciences). Peroxidase activity was developed with 3-amino-9-ethyl carbazole (Sigma) or 3,3′-diaminobenzidine (Sigma).

Antibodies.

The antibodies used were biotinylated goat anti-mouse VEGFR-3 (R&D Systems), goat anti-mouse VEGFR-3 (R&D Systems), rat anti-mouse VEGFR-3 (20), rabbit anti-mouse LYVE-1 (17), and rat anti-mouse PECAM-1 (Pharmingen). Secondary antibodies were Alexa-conjugated (Molecular Probes) or biotinylated (Vector Laboratories).

LacZ staining.

Vegfr3 expression in Vegfr3^neo/lz and Vegfr3^+/lz control embryos was detected, using whole-mount LacZ staining. Embryo tissues were fixed in 0.2% glutaraldehyde and stained with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal; Sigma) at 37°C for β-galactosidase activity.

RESULTS

Blood vascular development is normal, but lymphatic vascular development fails in the Vegfc; Vegfd double-knockout embryos.

Previous analyses of the Vegfr3^−/− embryos showed a severe blood vascular defect which resulted in death at E10.5 (9). While the Vegfc^−/− mice lack lymphatic vasculature, the Vegfd^−/− mice have essentially normal lymphatic vessels (6, 17). Since VEGF-C and VEGF-D are the only ligands known to activate VEGFR-3, we hypothesized that the Vegfc^−/−; Vegfd^−/− double-knockout embryos would also display blood vascular defects, thus phenocopying the Vegfr3^−/− embryos. We therefore mated the Vegfc and Vegfd gene-targeted mice in order to study the combination phenotype. Surprisingly, embryos deficient in both Vegfc and Vegfd appeared to be similar to the wild-type embryos at E11.5 (Fig. 1A and B), while the Vegfr3^−/− embryos were already dead at this time point (Fig. 1C). Analysis of the double-knockout embryos by whole-mount staining for PECAM-1 revealed a blood vasculature comparable to that of the wild-type embryos (Fig. 1D to G). Analysis of the lymph sac formation in the E13.5 Vegfc^−/−; Vegfd^−/− double-knockout embryos and the Vegfc^−/−, Vegfd^−/−, and Vegfc^+/− control embryos was performed by immunostaining of serial sections from the jugular area for the lymphatic markers LYVE-1 (Fig. 1H to K) and VEGFR-3 (Fig. 1L to O). This analysis showed arrested sprouting of lymphatic endothelial cells from the jugular vein and an absence of lymph sacs in the Vegfc^−/−; Vegfd^−/− embryos (Fig. 1H and L), as well as in the Vegfc^−/− embryos (Fig. 1I and M). PECAM-1 staining revealed normal jugular veins and dorsal aorta in the E13.5 Vegfc^−/−; Vegfd^−/− embryos (Fig. 1P to S).

FIG. 1.

Vegfc^−/−; Vegfd^−/− double-knockout phenotype. The Vegfc^−/−; Vegfd^−/− double-knockout embryo at E11.5 (A) is compared to the E11.5 wild-type embryo (B) and the E10.5 Vegfr3^−/− embryo (C). Blood vasculature in the head (D and E) and intersomitic area (F and G) was visualized by whole-mount immunostaining for PECAM-1 in the E11.5 Vegfc^−/−; Vegfd^−/− double-knockout (D and F) and wild-type (E and G) embryos. Serial sections are shown from the jugular areas of the E13.5 Vegfc^−/−; Vegfd^−/−, Vegfc^−/−, Vegfd^−/−, and Vegfc^+/− embryos stained for LYVE-1 (H to K), VEGFR-3 (L to O), and PECAM-1 (P to S). Arrowheads point to the dorsal aorta (da) in panels P to S. Note the absence of lymph sacs in the Vegfc^−/−; Vegfd^−/− and Vegfc^−/− embryos. ls, lymph sac; jv, jugular vein.

The Vegfr3^neo/neo and Vegfr3^+/neo hypomorphic mutant mice show defective lymphatic vessel formation and function.

To further understand the role of VEGFR-3 signaling during development, we generated conditional Vegfr3 knockout mice. Analyses of the gene-targeted mice revealed that the neomycin cassette inserted between the first two Vegfr3 exons (Fig. 2A) produced a hypomorphic phenotype. The Vegfr3^neo/neo and, interestingly, also the Vegfr3^+/neo embryos appeared swollen at E14.5 compared to the wild-type embryos (Fig. 2B to D), suggesting defects in lymphatic vessel formation. While the blood vasculature appeared normal in the Vegfr3^neo/neo and Vegfr3^+/neo embryos (data not shown), analysis of the lymphatic vasculature in the skin at E17.5, as shown by staining with VEGFR-3 (Fig. 2E to G) and LYVE-1 (Fig. 2H to J), showed that the Vegfr3^neo/neo embryos lacked lymphatic vasculature and that the Vegfr3^+/neo embryos showed only some remnants of lymphatic vessels. LYVE-1-positive single nonendothelial cells that presumably were macrophages were abundant in all skin preparations (Fig. 2H to J).

FIG. 2.

Vegfr3^neo targeting strategy and mouse phenotype. (A) The first exon of Vegfr3 is flanked by loxP sites (red triangles). The first intron of Vegfr3 contains a neomycin resistance gene (neo) that is flanked by frt sites (green triangles). Construction of the targeting vector is described in Materials and Methods. The figure represents the targeted Vegfr3^neo allele. Edema shown in the Vegfr3^neo/neo (B, arrow) and Vegfr3^+/neo (C, arrow) embryos is compared to that in the wild-type embryos (D) at E14.5. Cutaneous lymphatic vessels were stained for VEGFR-3 (E to G) and LYVE-1 (H to J) at E17.5 in the Vegfr3^neo/neo, Vegfr3^+/neo, and wild-type embryos. Note the absence of lymphatic vessels in the Vegfr3^neo/neo embryos. Lymphatic vessels are indicated by arrows in panels F, G, I, and J.

We crossed the Vegfr3^+/neo mice with the Vegfr3^+/lz mice to visualize lymphatic vessels in compound embryos, using a genetic β-galactosidase marker. LacZ staining of the Vegfr3^neo/lz compound embryos at E18.5 revealed the absence of lymphatic vessels in the skin and mesenterium, unlike Vegfr3^+/lz control embryos (Fig. 3A to D).

FIG. 3.

Defective lymphangiogenesis is shown in the Vegfr3^neo/lz embryos and Vegfr3^+/neo pups. LacZ-stained skin (A and B) and mesenterium (C and D) are shown in the E18.5 Vegfr3^neo/lz and Vegfr3^+/lz embryos. Note the absence of blue-stained lymphatic vessels in the Vegfr3^neo/lz embryos. (E and F) Comparison of chylous ascites in a P5 Vegfr3^+/neo pup and a wild-type pup. Leaky mesenteric lymphatic vessels (arrows) in a P1 Vegfr3^+/neo pup (G) compared to those in a wild-type pup (H).

The Vegfr3^neo/neo mice died perinatally, whereas the Vegfr3^+/neo mice survived to adulthood, although they showed transient accumulation of chylous ascites in the abdomen after birth (Fig. 3E and F). The diffuse appearance of the mesenteric lymphatic vessels in the Vegfr3^+/neo pups indicated that these vessels are leaky and dysfunctional (Fig. 3G and H).

Lack of Vegfr3 in the embryo proper produces the Vegfr3^−/− phenotype.

To determine whether defective placental morphogenesis contributes to the Vegfr3^−/− phenotype, we used the conditional mice to specifically delete the Vegfr3 gene in the epiblast. To do this, we used K19Cre, which drives Cre expression in early postimplantation embryos in cells giving rise to all embryonic tissues, while the extraembryonic tissues are not targeted (12, 24). We first crossed the Vegfr3^+/neo mice with mice expressing FLPe recombinase under the β-actin promoter (27) to remove the neomycin cassette. The resulting Vegfr3^lx/+ and Vegfr3^lx/lx mice were born in the expected Mendelian ratio and were phenotypically normal (data not shown). To confirm that the conditional allele was functional, we crossed the Vegfr3^lx/lx mice with PGKCre mice (22). Germ line deletion of Vegfr3 resulted in the Vegfr3^−/− phenotype (Fig. 4A), and no VEGFR-3 protein expression was detected in the Cre-positive embryos with Western blot analysis of proteins extracted from E9.5 embryos (Fig. 4B). Deletion of Vegfr3 in the epiblast, using K19Cre, produced an identical phenotype (Fig. 4C and E), suggesting that the blood vessel defects detected in the Vegfr3^−/− embryos are not caused by defects in placental development but that cardiovascular development is dependent on VEGFR-3 signaling in the embryo proper. A small percentage of the K19Cre; Vegfr3^lx/lz embryos and mice survived without an apparent phenotype (data not shown). PCR analysis of DNA isolated from these animals, using primers to detect the recombined allele, showed only partial recombination (Fig. 4D), while all embryos displaying the null phenotype showed 100% recombination (Fig. 4D). In agreement with previously reported data, our analysis of Cre activity using ROSA26 reporter mice (30) showed a mosaic pattern, with some embryos showing lower levels of recombination (data not shown) (24).

FIG. 4.

Testing of the conditional Vegfr3^lx allele. (A) The PGKCre; Vegfr3^lx/lz embryo at E10.5 is compared to a control embryo. (B) Western blot analysis of protein extracted from E9.5 PGKCre; Vegfr3^lx/lx and control embryos. The VEGFR-3 protein is not detected in the Cre-positive embryo. (C) Deletion of Vegfr3 in the epiblast produces a null phenotype. The K19Cre; Vegfr3^lx/lz embryo at E10.5 is compared to a littermate control. (D) PCR genotyping analysis of embryos by using primers that detect the floxed (lx) and recombined allele (lx*) is shown. A majority of the K19Cre; Vegfr3^lx/lz embryos showed 100% recombination and displayed a null phenotype, while a minority of the embryos and the few surviving mice showed only partial recombination (band *) and no phenotype. (E) Adjacent sections of the E10 Vegfr3^lx/lz control and the K19Cre; Vegfr3^lx/lz embryos stained for PECAM-1 (upper panels) and VEGFR-3 (lower panels).

VEGF-D is able to rescue lymphatic hypoplasia in the Vegfc^+/− mice.

To determine if VEGF-D is able to rescue defective VEGF-C function in embryos, we analyzed the hypoplastic lymphatic phenotype in the skin of the Vegfc^+/− mice mated with the K14-hVEGF-D or K14-mVEGF-D transgenic mice overexpressing VEGF-D in the basal skin layer of keratinocytes under the control of the K14 promoter. Analysis of the cutaneous lymphatic vessels in the K14-hVEGF-D; Vegfc^+/− compound mice by whole-mount staining for LYVE-1 revealed a hyperplastic lymphatic vessel network (Fig. 5A) that was almost as dense as that in the K14-hVEGF-D mice (Fig. 5B). Hypoplastic lymphatic vessels in the skin of the Vegfc^+/− mice and the normal lymphatic vasculature in wild-type mice are shown for comparison in Fig. 5C and D, respectively. In the K14-mVEGF-D; Vegfc^+/− compound mice, the hypoplasia of the cutaneous lymphatic vessels in the Vegfc^+/− mutant was also rescued but not as strongly as that in the K14-hVEGF-D; Vegfc^+/− mice (Fig. 5E). The skin of the K14-mVEGF-D mice with hyperplastic lymphatic vessels (Fig. 5F) is shown in comparison with that of the Vegfc^+/− mice (Fig. 5G) and the wild-type mice (Fig. 5H).

FIG. 5.

Cutaneous lymphatic vessel whole mount stained for LYVE-1 in the skin of K14-hVEGF-D; Vegfc^+/− (A), K14-hVEGF-D (B), Vegfc^+/− (C), wild-type (D), K14-mVEGF-D; Vegfc^+/− (E), K14-mVEGF-D (F), Vegfc^+/− (G), and wild-type (H) mice.

DISCUSSION

Our results show that the blood vasculature develops normally in the Vegfc^−/−; Vegfd^−/− double-knockout embryos, indicating that they do not phenocopy the cardiovascular failure of the Vegfr3^−/− knockout embryos. These data, together with previous results describing the Vegfd gene-targeted mice (6), raise the possibility that another ligand for VEGFR-3 exists that is important for embryonic angiogenesis starting at E9.5. Alternatively, VEGFR-3 may be able to act by an as-yet-unknown mechanism, perhaps involving heterodimers with VEGFR-2 (3, 8). Interestingly, VEGFR-3 is induced by the inhibition of Notch signals and is highly expressed in endothelial tip cells of angiogenic sprouts, where it positively regulates angiogenic sprouting (32). In contrast, the knockdown of VEGF-C in Xenopus laevis tadpoles induced aberrant blood vessel formation in addition to lymphatic vessel defects (25). Recently, lymphatic vessels were also described in zebrafish, and the vascular formation was shown to depend on VEGF-C/Flt4(VEGFR-3) signaling, suggesting that this pathway is conserved in evolution (21, 38). Interestingly, the zebrafish Flt4 gene was shown to affect artery morphogenesis in cooperation with the Vegfr2 homolog Kdr (7), and overexpression of zebrafish VEGF-D via the injection of mRNA was shown to affect embryonic blood vessel formation (29).

Mice carrying one hypomorphic Vegfr3^neo allele showed normal blood vessel development but impaired lymphatic vessel development and function, similar to that of the Vegfc^+/− and Chy mice that display transient lymphatic hypoplasia and the accumulation of chylous fluid in the abdomen after birth (17, 18). Mice homozygous for the Vegfr3^neo allele showed even more severe defects in embryonic lymphangiogenesis and embryonic death but normal development of blood vessels. These data suggest that higher levels of VEGFR-3 signaling are required to sustain embryonic lymphangiogenesis than are required for angiogenesis. The Vegfr3^neo allele caused more severe perturbation of lymphatic development than the Vegfr3^lz allele, bred to another strain of mice; such alleles will be useful tools for further analyzing the effects of decreased levels of Vegfr3 expression on development (Fig. 6).

FIG. 6.

A schematic comparison of the viability of the Vegfc, Vegfd, and Vegfr3 mutant mice.

In normal mouse skin, lymphatic vessel growth can be stimulated by using recombinant adenovirus or adeno-associated virus vectors encoding VEGF-C or VEGF(C156S) (10, 28). Moreover, lymphatic vessel growth can be stimulated and lymphatic vessel hypoplasia rescued in Chy lymphedema mice, which have an inactivating missense mutation in one Vegfr3 allele, by using these vectors or by crossing Chy mice with K14-VEGF-C156 mice (18). According to our results, transgenic overexpression of VEGF-D in the skin keratinocytes was similarly able to rescue the lymphatic hypoplasia in the Vegfc^+/− mice, suggesting that VEGF-D can compensate for the loss of VEGF-C when it is expressed at sufficient levels. The more severe hyperplasia of the cutaneous lymphatic vessels in the K14-hVEGF-D mice than in the K14-mVEGF-D mice may be explained by the fact that mouse VEGF-D binds only to VEGFR-3, whereas human VEGF-D binds to both VEGFR-2 and VEGFR-3 (5). Also, VEGFR-2 can contribute to lymphangiogenic signaling, and VEGFR-2 and VEGFR-3 can form heterodimers that could play a role in lymphangiogenesis (3, 8, 11, 37).

In summary, our results indicate that sufficient VEGFR-3 signaling is required in the embryo proper for embryonic angiogenesis and in a dosage-sensitive manner for embryonic lymphangiogenesis. Furthermore, although mouse VEGF-D rescued defective lymphangiogenesis, it did not seem to compensate for the loss of VEGF-C in embryonic lymphangiogenesis, suggesting the existence of additional ligands that act via VEGFR-3 or the possibility of ligand-independent signaling mechanisms.