Lymphangiogenesis is a dynamic process that involves the sprouting of lymphatic endothelial cells (LECs) from veins to form lymphatic vessels. Vegfr3 signalling, through its ligand Vegfc and the extracellular protein Ccbe1, is essential for the sprouting of LECs to form the trunk lymphatic network. In this study we determined whether Vegfr3, Vegfc and Ccbe1 are also required for development of the facial and intestinal lymphatic networks in the zebrafish embryo. Whereas Vegfr3 and Ccbe1 are required for the development of all lymphatic vessels, Vegfc is dispensable for facial lymphatic sprouting but not for the complete development of the facial lymphatic network. We show that zebrafish vegfd is expressed in the head, genetically interacts with ccbe1 and can rescue the lymphatic defects observed following the loss of vegfc. Finally, whereas knockdown of vegfd has no phenotype, double knockdown of both vegfc and vegfd is required to prevent facial lymphatic sprouting, suggesting that Vegfc is not essential for all lymphatic sprouting and that Vegfd can compensate for loss of Vegfc during lymphatic development in the zebrafish head.
The lymphatic vasculature is a set of bloodless vessels crucial for maintaining fluid homeostasis in the body, with additional important roles in immune cell trafficking and lipid absorption (Tammela and Alitalo, 2010). The growth of new lymphatic vessels, which is termed lymphangiogenesis, can occur aberrantly in tumours and in areas of chronic inflammation (Stacker et al., 2014).
It is now well established from experiments in mice and in zebrafish that the lymphatic vasculature arises from the veins. In mice the sprouting of lymphatic endothelial cells (LECs) from the cardinal vein requires vascular endothelial growth factor receptor 3 (VEGFR3; FLT4 – Mouse Genome Informatics) signalling through its ligand VEGFC (Karkkainen et al., 2004; Veikkola et al., 2001). Vegfr3 is expressed in blood endothelial cells but becomes upregulated in LECs during their specification from the venous endothelial cells (Kaipainen et al., 1995). Heterozygote inactivation of Vegfr3 in mice causes severe lymphatic defects (Karkkainen et al., 2001), and Vegfr3−/− mice die during early embryogenesis (E9.5) owing to failure in the development of the cardiovasculature (Dumont et al., 1998). Vegfr3-expressing LECs respond to VEGFC secreted in the mesenchyme, causing LECs to migrate from the cardinal vein during the early steps of lymphatic vessel development. Accordingly, Vegfc−/− mice lack a lymphatic vasculature because LEC progenitors fail to sprout from the cardinal vein (Karkkainen et al., 2004), whereas overexpression of Vegfc causes lymphatic hyperplasia (Jeltsch et al., 1997). The importance of the VEGFR3/VEGFC signalling pathway in human lymphatic vessel development is highlighted by the observation that patients with mutations in either of these genes develop lymphedema (Gordon et al., 2013; Irrthum et al., 2000; Karkkainen et al., 2000). The other ligand for VEGFR3 is VEGFD (FIGF – HUGO Gene Nomenclature Committee; Mouse Genome Informatics). A mouse knockout of Vegfd has confirmed that it is not essential for mammalian lymphatic vessel development (Baldwin et al., 2005); however, Vegfd can rescue the lymphatic hypoplasia in Vegfc−/− mice (Haiko et al., 2008) and is able to induce both lymphangiogenesis and angiogenesis in a variety of assays (Achen and Stacker, 2012).
Recently, zebrafish (Danio rerio) have emerged as an excellent model for the study of lymphangiogenesis; they have the advantage that developing lymphatic vessels are easily observed in transparent embryos (Yaniv et al., 2006). The majority of zebrafish studies have focused on development of the trunk lymphatic network, which by 5 days post fertilisation (dpf) consists of the thoracic duct (TD) that runs between the dorsal aorta and the posterior cardinal vein (PCV), the dorsal longitudinal lymphatic vessel (DLLV) and the intersegmental lymphatic vessels (ISLVs) that connect the TD to the dorsal longitudinal lymphatic vessel (Koltowska et al., 2013). The trunk lymphatic network is derived from lymphatic precursors that sprout from the PCV at around 1.5 dpf and migrate to the horizontal myoseptum, where they are termed parachordal lymphangioblasts. These lymphangioblasts then migrate both dorsally and ventrally using arterial intersegmental blood vessels as guides to remodel into the TD, ISLVs and the DLLV (Bussmann et al., 2010). Importantly, the formation of this network requires Vegfr3 (known as Flt4 in zebrafish) signalling through its ligand Vegfc, as mutation or morpholino (MO)-mediated knockdown of these genes prevents the sprouting of lymphatic cells from the cardinal vein (Hogan et al., 2009b; Kuchler et al., 2006; Villefranc et al., 2013). In addition, the chemokine receptor Cxcr4a and its ligands Cxcl12a and Cxcl12b have been shown to be required for the correct migration and remodelling of lymphangioblasts to form the TD (Cha et al., 2012). Studies using zebrafish have identified the collagen and calcium binding EGF domains 1 (ccbe1) gene as also being required for both zebrafish and mammalian lymphatic development (Alders et al., 2009; Bos et al., 2011; Hogan et al., 2009a). Recently, Ccbe1 has been identified as a crucial component of the Vegfc/Vegfr3 pathway; it has been shown to genetically interact with vegfc and flt4 in zebrafish (Le Guen et al., 2014) and also to indirectly promote the proteolytic cleavage of mammalian VEGFC into its active form (Jeltsch et al., 2014).
We recently generated a map of lymphatic development in zebrafish and characterised the development of two lymphatic networks, namely the facial lymphatics and the intestinal lymphatics, that are distinct from the previously characterised trunk lymphatic network (Okuda et al., 2012). We were able to show that the facial lymphatics develop initially from a single facial lymphatic sprout (FLS) that forms from the common cardinal vein (CCV) near the primary head sinus (PHS) at 1.5 dpf. Following initial migration, lymphangioblasts from the PHS, as well as other veins, are recruited to the tip of the FLS to drive the migration and development of the lateral facial lymphatic vessel (LFL) at 3 dpf, and, by 5 dpf, the otolithic lymphatic vessel (OLV), the medial facial lymphatic (MFL) and the lymphatic branchial arches (LAAs) have formed.
In this study, we investigated whether formation of the intestinal and facial lymphatic networks requires the same signalling pathways as those required for formation of the trunk lymphatic network. We find that flt4 and ccbe1 are required for the development of all three lymphatic networks; however, vegfc is only essential for the formation of the trunk and intestinal lymphatics and for the complete development of the facial lymphatic network, but is not required for initial facial lymphatic sprouting. We go on to show that expression of the other Flt4 ligand, Vegfd, in the head allows the FLS to form in the absence of Vegfc. In addition, we show that chemokine signalling is only required for the formation of the trunk lymphatics. These findings show that although there are common signalling pathways that drive lymphatic vessel development, there are also differences between the growth factors required to form different lymphatic networks.
vegfc is not essential for initial development of the facial lymphatic network
We investigated whether Flt4, Vegfc and Ccbe1 are required for development of the facial lymphatic network. Facial lymphatic vessels were identified by the expression of the lymphatic and venous marker lyve1 and by the absence of the blood vessel-specific marker kdrl (Okuda et al., 2012) (supplementary material Fig. S1). We confirmed that flt4 is essential for both facial and trunk lymphatic formation by: injecting a MO against flt4 into lyve1:dsred or lyve1:egfp embryos; examining lymphatic vessel formation in a previously characterised hypomorphic mutant of Flt4 in which signalling is impaired (flt4hu4602) (Hogan et al., 2009b); or injection of an mRNA encoding the Ig domain of human FLT4 (sFLT4) that has been shown to act as a dominant inhibitor of zebrafish Flt4 signalling (Hogan et al., 2009b; Ober et al., 2004).
Following injection with either flt4 MO or sFLT4 mRNA and in flt4hu4602 mutant embryos, we observed a decrease in the length of the LFL at 3 dpf (Fig. 1A,D,K) and also a failure in the formation of the TD (Fig. 1F,I,L) due to a defect in secondary sprouting from the PCV (supplementary material Fig. S2). A similar result was observed in embryos injected with a MO against ccbe1, with an inhibition of both LFL and TD development (Fig. 1E,J,K,L; supplementary material Fig. S2). There are a number of lyve1-positive vessels that remain in the head following knockdown of either flt4 or ccbe1 in 3 dpf embryos, but these were confirmed to be veins (PHS, CCV and PCeV) as they retain kdrl expression and have blood flow (supplementary material Figs S1, S3).
We used two different MOs designed to prevent the splicing of vegfc: vegfc MO-1 was designed to bind to the intron 1/exon 2 boundary, whereas vegfc MO-2 was designed against the exon 3/intron 3 boundary. We confirmed by RT-PCR that both vegfc MO-1 and vegfc MO-2 were able to prevent the correct expression of vegfc mRNA at 1 dpf (supplementary material Fig. S4). We also found that embryos injected with either of these vegfc MOs did not form the TD (Fig. 1G,L) due to a reduction in the number of secondary sprouts from the PCV (supplementary material Fig. S2), confirming that vegfc is essential for trunk lymphatic development. However, vegfc morphant embryos displayed normal LFL formation, with the length of the LFL being the same as in controls (Fig. 1B,K). Similar results were observed in vegfchu5055 embryos that contain a missense mutation, that removes a highly conserved cysteine residue in the C-terminal propeptide of Vegfc (C365R), and that has been previously shown to have a defect in the formation of the TD (Fig. 1C,H,K,L) (Le Guen et al., 2014).
Next, we examined the formation of the FLS by live imaging lyve1:egfp embryos from 1.5 dpf when the FLS forms from the CCV. In three embryos injected with a control MO we found that the FLS consistently formed by 1.6 dpf and had migrated towards the PHS by 1.9 dpf (Fig. 2A-E; supplementary material Movie 1). Time-lapse imaging of three flt4 morphant embryos showed that no embryos developed an FLS by 1.9 dpf, despite all three embryos displaying normal blood vessel development, providing evidence that flt4 is essential for FLS formation and explaining the inhibition of LFL vessel length at 3 dpf in flt4 morphant embryos (Fig. 2F-J; supplementary material Movie 2). However, in three embryos injected with vegfc MO-1 (the MO with the best efficacy at silencing vegfc), we observed normal FLS formation between 1.6 and 1.9 dpf, despite the absence of lymphatic development in the trunk in the same morphant embryos (Fig. 2K-O; supplementary material Movie 3). Taken together, our results show that whereas flt4 and ccbe1 are required for development of the facial lymphatics, vegfc is not essential for the formation of the FLS and embryos with impaired Vegfc are still able to develop an LFL vessel at 3 dpf.
Next, we examined the effect of the Flt4/Vegfc/Ccbe1 pathway on development of the facial lymphatic network after 3 dpf, when the OLV, MFL and LAA begin to form from the developing LFL (supplementary material Fig. S1). Imaging of 5 dpf lyve1 embryos showed that injection of either ccbe1 MO or sFLT4 mRNA inhibited the formation of these facial lymphatic vessels (Fig. 1M-R); however, the phenotype was less severe in flt4 morphants or in flt4hu4602 mutant embryos (Fig. 1R). Despite vegfc not being essential for the formation of the FLS and early LFL development, it is required for the formation of the OLV, MFL and LAA, as the development of all of these lymphatic vessels was severely impaired in vegfchu5055 embryos and to a lesser extent in vegfc morphant embryos (Fig. 1M-R; supplementary material Fig. S3). In addition, although the LFL had extended anteroventrally from the level of the PHS in the majority of vegfchu5055 embryos, it was still shorter than the LFL in control embryos that had extended below the eye, suggesting that vegfc is required for the later development of the LFL.
vegfc is required for development of the intestinal lymphatic network
The intestinal lymphatic network in zebrafish has been shown to be dependent on both Flt4 and Ccbe1 signalling (Okuda et al., 2012). By 6 dpf the intestinal lymphatics consist of the upper right intestinal lymphatic (UR-IL) that runs alongside the subintestinal artery and the of lower right intestinal lymphatic (LR-IL) that runs alongside the subintestinal vein. Analysis of flt4 and ccbe1 morphants confirmed that these genes are essential for the formation of the intestinal lymphatics. We also show that vegfc is required, as vegfc morphants consistently displayed either complete loss or fragmentation of the intestinal lymphatic network at 6 dpf (supplementary material Fig. S5).
cxcr4a and cxcl12a are only necessary for development of the trunk lymphatic network
Recently, the chemokine receptor Cxcr4a and its ligands Cxcl12a and Cxcl12b were identified as being required for early lymphatic vessel development in the zebrafish trunk (Cha et al., 2012). We therefore examined the role of chemokine signalling in facial and intestinal lymphatic development. Although both cxcr4a and cxcrl12a were expressed in a spatiotemporal manner consistent with a possible role in facial lymphatic development, MO knockdown of cxcr4a or cxcrl12a did not result in any defect in facial lymphatic development at either 3 or 5 dpf, despite causing a modest inhibition of formation of the TD (supplementary material Fig. S6). Formation of the intestinal lymphatics also appeared normal in cxcr4a or cxcrl12a morphant embryos. Taken together, our results suggest that, with regard to lymphatic development, cxcr4a and cxcrl12a are only required for the growth of trunk vessels.
vegfd is expressed in the head and can compensate for loss of vegfc
Given that Flt4 signalling is required for the development of the FLS but its ligand Vegfc appears to be dispensable, we examined whether the other Flt4 ligand, Vegfd, has a role in the development of the facial lymphatics. Previous studies have shown that zebrafish vegfd is expressed in the tail bud but the expression of vegfd in the head was not determined (Hogan et al., 2009a). Using whole-mount in situ hybridization, we examined the expression of flt4, vegfc, vegfd and ccbe1 in the head and trunk during early lymphatic vessel development. We found that mRNA of all four genes was present in the head region at 1.5-2 dpf (Fig. 3; supplementary material Fig. S7) and we observed expression of vegfd in the vicinity of the developing FLS at 1.5 dpf (Fig. 3F′). Taken together, these data suggest that vegfd is expressed in a spatiotemporal manner consistent with a role in the development of facial lymphatics. Quantitative PCR (qPCR) analysis of vegfd and vegfc mRNA levels in isolated heads of 1.5, 3 and 5 dpf embryos showed that mRNA levels of both genes are maintained from 1.5 to 5 dpf (Fig. 3K).
We were also able to show that flt4, vegfc and ccbe1 are all expressed in the trunk, matching the spatiotemporal patterns observed in previous studies and consistent with the essential role of these genes in trunk lymphatic development (Hogan et al., 2009a,,b; Kuchler et al., 2006; Villefranc et al., 2013) (Fig. 3; supplementary material Fig. S7). We did not observe expression of vegfd in regions of trunk lymphatic development, suggesting that it might not be involved in this process (Fig. 3I), but did observe expression in the pectoral fin buds, which, along with previous reports of vegfd expression in the tail bud (Hogan et al., 2009a), suggests a possible role for vegfd in fin development.
Previous studies have shown that zebrafish Vegfd, like Vegfc, has angiogenic properties (Hogan et al., 2009b; Song et al., 2007). We were able to confirm this by injecting either vegfc or vegfd mRNA into embryos expressing GFP in the blood vasculature (kdrl:egfp), and we observed that both genes, when expressed ectopically, can induce misguidance of the arterial intersegmental blood vessels (ISVs) along the horizontal myoseptum (Fig. 4A-C). Recently, it has been shown that ccbe1 can genetically interact with vegfc; silencing ccbe1 can inhibit the ability of vegfc to induce branching of the ISVs (Le Guen et al., 2014). We were able to confirm this observation and further showed that vegfd and ccbe1 also interact genetically, as injection of vegfd mRNA into ccbe1 morphants had reduced levels of ISV branching compared to embryos injected with vegfd mRNA and control MO (Fig. 4D-G). To determine whether zebrafish vegfd can compensate for vegfc, we injected vegfd mRNA into either vegfc morphant embryos or progeny from an incross of vegfchu5055 heterozygous carriers and observed that vegfd mRNA was able to partially rescue TD development in these embryos with a similar efficacy to that of rescue by vegfc mRNA (Fig. 4H-L).
To determine the ability of Vegfd to induce sprouting of lyve1-positive vessels, we utilised a xenograft model in which human breast cancer cell lines, transfected with expression vectors containing zebrafish vegfc or vegfd, were implanted into the perivitelline space of 2 dpf lyve1:dsred embryos and then examined at 5 dpf for the formation of ectopic lyve1-positive vessels that form from the CCV around the xenograft. We called these ectopic vessels, ‘lyve1-positive vessels’ as they also express the blood vessel marker kdrl (data not shown) and thus we were unable to distinguish whether these vessels are veins or immature lymphatic vessels that still retain kdrl expression. Both Vegfc- and Vegfd-expressing cancer cells were able to induce more lyve1-positive vessels than cancer cells expressing the vector alone, suggesting that both Vegfc and Vegfd have the potential to induce lyve1 vessel growth in zebrafish (Fig. 5A-E).
Double knockdown of both vegfc and vegfd is required to prevent initial development of the facial lymphatics
Given that vegfd can compensate for vegfc and can induce similar phenotypes to vegfc when expressed ectopically, we examined whether vegfd had a function in lymphatic vessel development. We used an MO targeting the exon 4/intron 4 boundary of vegfd, and injection of 1 pmol of this MO resulted in a 65 bp deletion of the 819 bp vegfd transcript (supplementary material Fig. S8). Full-length vegfd mRNA encodes a 273 amino acid protein. Sequencing the morphant vegfd mRNA showed that the deletion causes a frameshift from position 186 and a premature stop at position 200, resulting in loss of the C-terminal end of the VEGF homology domain that is essential for the function of VEGF ligands. From these data we predict that the vegfd MO is disrupting any potential Vegfd signalling in morphant embryos. Injection of vegfd MO into either lyve1:egfp or lyve1:dsred embryos showed no lymphatic defects in either the trunk, intestinal or facial lymphatic networks, suggesting that vegfd is dispensable for zebrafish lymphatic vessel development (Fig. 6K; supplementary material Fig. S8). However, given that vegfd is expressed in the head during the formation of the facial lymphatics, we examined whether it could compensate for disrupted vegfc levels.
To test this hypothesis, we injected a mixture of vegfc MO-1 and vegfd MO and examined the formation of the FLS by live imaging lyve1:egfp embryos from 1.5 dpf. In three embryos injected with control MO and vegfc MO-1 we found that the FLS consistently formed by 1.9 dpf (45 hpf) (Fig. 6A-D; supplementary material Movie 4). In three embryos injected with vegfc MO-1 and vegfd MO we found that the FLS failed to form by 1.9 dpf (45 hpf) (Fig. 6E-H; supplementary material Movie 5). We also observed a reduction in LFL length in vegfc/vegfd double morphants at 3 dpf that could be rescued by injection of either vegfd or vegfc mRNA (Fig. 6I,M-O,Q). A reduction in LFL length was also observed when vegfd MO was injected into vegfchu5055 embryos (Fig. 6L,P,Q). Examination of OLV, MFL and LAA vessel formation at 5 dpf revealed that both vegfc/vegfd double morphants and vegfchu5055 mutants injected with vegfd MO lacked all facial lymphatic vessels and phenocopied embryos injected with sFLT4 mRNA (Fig. 6R-V). These results show that although vegfd is not essential for lymphatic vessel development, it can compensate for loss of vegfc in the formation of the facial lymphatics. Furthermore, knockdown of both vegfc and vegfd is required to prevent the formation of the FLS and the initial development of the LFL in zebrafish embryos.
Our results show that different lymphatic networks in the zebrafish have distinct requirements for lymphatic growth factors (Fig. 7). We have analysed the requirement for Flt4, Vegfc and Ccbe1 in development of the intestinal and the facial lymphatics and have found that they are all essential for development of the intestinal lymphatics and also for maturation of the facial lymphatics. The LAA, MFL and OLV did not form in embryos with knockdowns or mutations in these genes.
Recently, it has been shown that chemokine signalling is required for the development of the trunk lymphatic network (Cha et al., 2012). Although we were able to reproduce the requirement for Cxcr4a and Cxcl12a in the development of the trunk lymphatics, MO knockdown of cxcr4a or cxcl12a had no effect on the development of the facial or the intestinal lymphatics, despite being highly expressed in the head. Thus, chemokine signalling appears to be a specific requirement for the trunk lymphatic network and highlights how different lymphatic networks may utilise different growth factors (Fig. 7).
The facial lymphatic network is derived from a single sprout, the FLS, which forms from the CCV. Surprisingly, whereas flt4 and ccbe1 are essential for the formation of the FLS and the early LFL, we found that knockdown of vegfc by MO injection did not inhibit LFL formation in 3 dpf embryos. Further analysis showed that the FLS formed normally in vegfc morphants despite these embryos showing a defect in secondary sprouting from the PCV. One possible explanation for this result is that MO knockdown of vegfc might be less efficacious in the zebrafish head compared with the trunk, possibly due to the higher levels of vegfc expression in the head. To account for this we also examined lymphatic vessel formation in a vegfc mutant (vegfchu5055) that has lost a conserved cysteine residue in the C-terminal propeptide (C365R) (Le Guen et al., 2014). It is possible that this mutant allele of vegfc is not completely null, as other mutations in vegfc have been shown to have more penetrant defects in trunk lymphatic development (Le Guen et al., 2014). Nevertheless, we have shown that the vegfchu5055 embryos have a significant defect in TD development but, like vegfc morphant embryos, these mutant embryos display normal facial lymphatic development up to 3 dpf. Taken together, these observations suggest that the formation of the FLS does not require vegfc or at least is more resistant to loss of vegfc than the formation of lymphatic sprouts in the trunk. As far as we are aware, this is the first report showing that Vegfc is not essential for all lymphatic sprouting. Whereas early facial lymphatic development (up to 3 dpf) was normal in embryos with impaired vegfc, we noted that the OLV, MLV and LAA did not form in the majority of vegfchu5055 embryos at 5 dpf, suggesting that vegfc is essential for the development of these vessels. We also noted that the later development of the LFL under the eye also required vegfc. The presence of the OLV and MFL in a small percentage of 5 dpf vegfchu5055 embryos might be due to the incomplete penetrance of this allele. Therefore, there appears to be a differential requirement for Vegfc in early versus late lymphatic vessel development in the zebrafish head.
There are two possible explanations for the observation that Vegfr3/Flt4 is required for FLS formation but its ligand Vegfc is not essential: Flt4 is signalling through ligand-independent mechanisms, or another Flt4 ligand is involved in the formation of the FLS. Whereas VEGFR3 is known to have ligand-independent roles, to date the lymphangiogenic function of VEGFR3 has primarily been demonstrated to be via ligand signalling (Haiko et al., 2008; Veikkola et al., 2001). In support of this, we were able to robustly inhibit all facial lymphatic development by the injection of sFLT4 mRNA, which functions as a dominant inhibitor by binding endogenous Flt4 ligands (Hogan et al., 2009b; Ober et al., 2004). We therefore examined the role of the other Flt4 ligand, Vegfd, in facial lymphatic development.
In mammals, it has been shown that VEGFD is closely related in structure to VEGFC (Leppanen et al., 2011) and, like VEGFC, can activate both VEGFR3/FLT4 and VEGFR2, thereby promoting both angiogenesis and lymphangiogenesis (Achen and Stacker, 2012). Unlike VEGFC, VEGFD does not appear to be essential for mammalian lymphatic vessel development (Baldwin et al., 2005), but it has been shown in Xenopus embryos to subtly modify lymphangioblast migration (Ny et al., 2008). Mammalian VEGFD is known to be involved in the growth of lymphatic vessels in tumours and also in areas of chronic inflammation (Huggenberger et al., 2010; Kataru et al., 2009). It has also been shown to compensate for loss of VEGFC (Haiko et al., 2008; Karkkainen et al., 2004). In this study, we show that, like mammalian VEGFD, zebrafish Vegfd has both angiogenic and lymphangiogenic potential and can compensate for loss of Vegfc. Furthermore, we show that at 1.5 dpf vegfd is expressed in a region near the intersection of the CCV and the PHS, where the FLS develops, consistent with vegfd having a role in early facial lymphatic development.
We show that, as in mammals, vegfd is not essential for lymphatic development, as the trunk, intestinal and facial lymphatics form normally in vegfd morphant embryos. However, we found that double knockdown of both vegfc and vegfd inhibits FLS formation and the subsequent formation of the facial lymphatics, which phenocopies knockdown of either ccbe1 or embryos with impaired Flt4 signalling. It has been shown that Ccbe1 is required for the activity of Vegfc (Jeltsch et al., 2014; Le Guen et al., 2014) and our data suggest that Ccbe1 is also required for Vegfd activity. In support of this we found that ccbe1 and vegfd genetically interact, as silencing ccbe1 reduced the levels of ISV branching induced by ectopic expression of vegfd.
We believe that Vegfd can compensate for Vegfc in early facial lymphatic formation due to the restricted expression of vegfd in the head region. However, why vegfd is expressed in this region and not in other areas of lymphatic vessel development, and whether vegfd plays an active role in facial lymphatic development, are not clear. In mammalian embryos, Vegfd is expressed in many tissues, including the lung, skin, kidney, limb buds and the mandibular arches (Avantaggiato et al., 1998; Stacker et al., 1999). We observed expression of zebrafish vegfd in some of the tissue equivalents to counterparts in mice, such as the pectoral fin buds and the mandibular arches, suggesting some conservation of Vegfd function between teleosts and mammals. It is possible that vegfd has other developmental roles that necessitate its expression in the head region and it would therefore be serendipitously available for facial lymphatic development when Vegfc levels are impaired. It is intriguing that endogenous vegfd can rescue the formation of the FLS and the early development of the LFL in embryos depleted of vegfc, but not the formation of lymphatic vessels after 3 dpf (OLV, MLV and LAA). We were unable to determine if this is due to differences in the spatial expression of vegfd. However, we do show that vegfd expression is maintained in the head throughout lymphatic development, so it is possible that the OLV, MLV and the LAA have an essential requirement for Vegfc signalling compared with the FLS and the early LFL.
To date, no characterisation of mouse ‘facial lymphatics’ has been conducted and it is unclear if equivalent lymphatic structures exist in mammals. This might explain why our results appear to contradict mouse studies, which have shown that double mutants of Vegfc and Vegfd had similar lymphatic defects to the single Vegfc mutant mouse (Haiko et al., 2008; Karkkainen et al., 2004). In addition, these studies would not have been able to evaluate possible delays in lymphatic sprouting due to technical limitations in the imaging of early lymphatic vessel development in the mouse.
Our findings have three important implications for lymphatic vessel development. First, we show that vegfc is not essential for all lymphatic sprouting in the zebrafish. Second, we demonstrate that although vegfd may not be required for lymphatic vessel development, it nevertheless can have a role in lymphatic development by compensating for loss of vegfc. Third, and more importantly, this study provides evidence that the development of different lymphatic networks can involve different sets of lymphatic growth factors.
MATERIALS AND METHODS
All zebrafish strains were maintained under standard husbandry conditions and animal studies were approved by the University of Auckland Animal Ethics Committee. The lines used in this study were: wild type (AB), TG(lyve1:egfp)nz150, TG(lyve1:dsred)nz101 (Okuda et al., 2012), TG(kdrl:egfp)s843 (Jin et al., 2005), TG(kdrl:rfp)la4 (Huang et al., 2005), flt4hu4602 (Hogan et al., 2009b) and vegfchu5055 (Le Guen et al., 2014).
Whole-mount in situ hybridisation
In situ hybridisation was performed as described previously (Thisse and Thisse, 2008). Antisense RNA probes for ccbe1 (Hogan et al., 2009a), flt4, vegfc (Hogan et al., 2009b), cxcr4a, cxcl12a and cxcl12b (Cha et al., 2012) were designed as described previously. The antisense probe for vegfd was based on the full-length cDNA. Primers used to generate the vegfd probe are given in supplementary material Table S1.
MO and RNA injections
MO sequences and doses are given in supplementary material Table S2. Full-length vegfc and vegfd cDNAs were cloned (primers given in supplementary material Table S1) and ligated into the pCS2+ vector. The sFLT4 construct was described previously (Hogan et al., 2009b; Ober et al., 2004). mRNA was synthesised using the mMessage in vitro transcription kit (Ambion) from an Sp6 primer and 200 pg (sFLT4, vegfd) or 400 pg (vegfc) mRNA was injected into 1-cell embryos.
Confocal live imaging of zebrafish
Embryos were imaged as described (Hall et al., 2009) with a Nikon D-Eclipse C1 confocal microscope or with an Olympus FV1000 confocal microscope for time-lapse microscopy. Still images were taken using z-stacks 5 μm apart. For time-lapse microscopy, z-stacks 5 μm apart were taken at 10 min intervals. Confocal images in this study are maximum projections of z-series stacks. Images were processed using ImageJ (NIH), Photoshop CS5 (Adobe) and Volocity 5.4 image analysis software (Improvision/PerkinElmer Life and Analytical Sciences). Microangiography was performed as described previously (Isogai et al., 2001). Lymphangiography was performed by subcutaneous injection of 2.5 mg/ml 500 kDa fluorescein dextran (Invitrogen) diluted in water as described (Okuda et al., 2012).
Image analysis and statistics
TD formation at 5 dpf was scored by counting the presence of a TD in the first 15 somites using either lyve1:egfp or lyve1:dsred and was represented as percentage. The length of the LFL was measured at 3 dpf as described previously (Okuda et al., 2012). For 5 dpf embryos the LFL was considered as present if it had developed anteroventrally from the PHS (towards the eye). The LAA, MFL and OLV were scored as present at 5 dpf if they could be distinguished from the LFL. Secondary sprout formation was scored using the lyve1:egfp transgenic at 36 hpf. Only the numbers of secondary sprouts from the anterior portion of the PCV (from the first somite to the cloaca) were counted. Statistical analysis was performed using Prism 5.0 software (GraphPad Software). Significance was determined by Mann–Whitney tests.
Full-length zebrafish vegfc and vegfd cDNAs were cloned into the mammalian expression vector pIRES-P (Hobbs et al., 1998) that was transfected into the human breast cancer line MDA-MB-231 using Lipofectamine 2000 reagent (Invitrogen). Transfected cells were selected and maintained in 2 μg/ml puromycin, grown in MEM-Alpha medium supplemented with 10% FBS. Cells were trypsinised, labelled with 2 μM Cell Tracker Green (Invitrogen) and injected into the ventral part of the perivitelline space of 2 dpf lyve1:dsred embryos as described previously (Nicoli and Presta, 2007). Xenografted embryos were imaged at 5 dpf by confocal imaging through the tumour volume. To normalise for differences in xenograft size, the volume of ectopic lyve1:dsred-expressing vessels around and in the tumour was determined by Volocity 5.4 image analysis software and expressed as a percentage of the tumour volume.
RNA extraction and RT-PCR
We thank Alhad Mahagaonkar for expert management of our zebrafish facility and the Biomedical Imaging Research Unit, School of Medical Sciences, University of Auckland, for assistance in confocal microscopy.
J.W.A., B.M.H., K.E.C. and P.S.C. conceived and planned the experiments and analysed the data. J.W.A., M.J.L.H., K.S.O., L.L., J.P.M. and A.T. performed the experiments. The manuscript was written by J.W.A., with assistance from L.L., B.M.H., K.E.C. and P.S.C.
This work was supported by a grant from the New Zealand Ministry of Business, Innovation and Employment [UOAX0813 to P.S.C.], by a grant from the Auckland Medical Research Foundation [1111015 to J.W.A.] and by an Australian Research Council Future Fellowship [FT100100165 to B.M.H.).
The authors declare no competing financial interests.